The Pyramid Web Application Development Framework

Pyramid is a small, fast, down-to-earth Python web application development framework. It is developed as part of the Pylons Project. It is licensed under a BSD-like license.

Here is one of the simplest Pyramid applications you can make:

from paste.httpserver import serve
from pyramid.config import Configurator
from pyramid.response import Response


def hello_world(request):
    return Response('Hello %(name)s!' % request.matchdict)

if __name__ == '__main__':
    config = Configurator()
    config.add_route('hello', '/hello/{name}')
    config.add_view(hello_world, route_name='hello')
    app = config.make_wsgi_app()
    serve(app, host='0.0.0.0')

When saved to helloworld.py, the above application can be run via:

$ easy_install pyramid
$ python helloworld.py
serving on 0.0.0.0:8080 view at http://127.0.0.1:8080

And when you visit http://localhost:8080/hello/world in a browser, you will see the text Hello, world!.

See Creating Your First Pyramid Application for a full explanation of how this application works. Read the Narrative documentation to understand how Pyramid is designed to scale from simple applications like this to very large web applications.

Front Matter

Typographical Conventions

Literals, filenames and function arguments are presented using the following style:

argument1

Warnings, which represent limitations and need-to-know information related to a topic or concept are presented in the following style:

Warning

This is a warning.

Notes, which represent additional information related to a topic or concept are presented in the following style:

Note

This is a note.

We present Python method names using the following style:

pyramid.config.Configurator.add_view()

We present Python class names, module names, attributes and global variables using the following style:

pyramid.config.Configurator.registry

References to glossary terms are presented using the following style:

URLs are presented using the following style:

References to sections and chapters are presented using the following style:

Code and configuration file blocks are presented in the following style:

1
2
def foo(abc):
    pass

When a command that should be typed on one line is too long to fit on a page, the backslash \ is used to indicate that the following printed line should actually be part of the command:

c:\bigfntut\tutorial> ..\Scripts\nosetests --cover-package=tutorial \
      --cover-erase --with-coverage

A sidebar, which presents a concept tangentially related to content discussed on a page, is rendered like so:

This is a sidebar

Sidebar information.

“What’s New” Documents

What’s New In Pyramid 1.2

This article explains the new features in Pyramid version 1.2 as compared to its predecessor, Pyramid 1.1. It also documents backwards incompatibilities between the two versions and deprecations added to Pyramid 1.2, as well as software dependency changes and notable documentation additions.

Major Feature Additions

The major feature additions in Pyramid 1.2 follow.

Debug Toolbar

The scaffolding packages that come with Pyramid now include a debug toolbar component which can be used to interactively debug an application. See The Debug Toolbar for more information.

route_prefix Argument to include

The pyramid.config.Configurator.include() method now accepts a route_prefix argument. This argument allows you to compose URL dispatch applications together from disparate packages. See Using a Route Prefix to Compose Applications for more information.

Tweens

A tween is used to wrap the Pyramid router’s primary request handling function. This is a feature that can be used by Pyramid framework extensions, to provide, for example, view timing support and can provide a convenient place to hang bookkeeping code. Tweens are is a little like WSGI middleware, but have access to Pyramid functionality such as renderers and a full-featured request object.

To support this feature, a new configurator directive exists named pyramid.config.Configurator.add_tween(). This directive adds a “tween”.

Tweens are further described in Registering “Tweens”.

A new paster command now exists: paster ptweens. This command prints the current tween configuration for an application. See the section entitled Displaying “Tweens” for more info.

Scaffolding Changes
  • All scaffolds now use the pyramid_tm package rather than the repoze.tm2 middleware to manage transaction management.
  • The ZODB scaffold now uses the pyramid_zodbconn package rather than the repoze.zodbconn package to provide ZODB integration.
  • All scaffolds now use the pyramid_debugtoolbar package rather than the WebError package to provide interactive debugging features.
  • Projects created via a scaffold no longer depend on the WebError package at all; configuration in the production.ini file which used to require its error_catcher middleware has been removed. Configuring error catching / email sending is now the domain of the pyramid_exclog package (see http://docs.pylonsproject.org/projects/pyramid_exclog/dev/).
  • All scaffolds now send the cache_max_age parameter to the add_static_view method.

Minor Feature Additions

  • The [pshell] section in an ini configuration file now treats a setup key as a dotted name that points to a callable that is passed the bootstrap environment. It can mutate the environment as necessary during a paster pshell session. This feature is described in Writing a Script.
  • A new configuration setting named pyramid.includes is now available. It is described in Including Packages.
  • Added a pyramid.security.NO_PERMISSION_REQUIRED constant for use in permission= statements to view configuration. This constant has a value of the string __no_permission_required__. This string value was previously referred to in documentation; now the documentation uses the constant.
  • Added a decorator-based way to configure a response adapter: pyramid.response.response_adapter. This decorator has the same use as pyramid.config.Configurator.add_response_adapter() but it’s declarative.
  • The pyramid.events.BeforeRender event now has an attribute named rendering_val. This can be used to introspect the value returned by a view in a BeforeRender subscriber.
  • The Pyramid debug logger now uses the standard logging configuration (usually set up by Paste as part of startup). This means that output from e.g. debug_notfound, debug_authorization, etc. will go to the normal logging channels. The logger name of the debug logger will be the package name of the caller of the Configurator’s constructor.
  • A new attribute is available on request objects: exc_info. Its value will be None until an exception is caught by the Pyramid router, after which it will be the result of sys.exc_info().
  • pyramid.testing.DummyRequest now implements the add_finished_callback and add_response_callback methods implemented by pyramid.request.Request.
  • New methods of the pyramid.config.Configurator class: set_authentication_policy() and set_authorization_policy(). These are meant to be consumed mostly by add-on authors who wish to offer packages which register security policies.
  • New Configurator method: pyramid.config.Configurator.set_root_factory(), which can set the root factory after the Configurator has been constructed.
  • Pyramid no longer eagerly commits some default configuration statements at Configurator construction time, which permits values passed in as constructor arguments (e.g. authentication_policy and authorization_policy) to override the same settings obtained via the pyramid.config.Configurator.include() method.
  • Better Mako rendering exceptions; the template line which caused the error is now shown when a Mako rendering raises an exception.
  • New request methods: current_route_url(), current_route_path(), and static_path().
  • New functions in the pyramid.url module: current_route_path() and static_path().
  • The pyramid.request.Request.static_url() API (and its brethren pyramid.request.Request.static_path(), pyramid.url.static_url(), and pyramid.url.static_path()) now accept an absolute filename as a “path” argument. This will generate a URL to an asset as long as the filename is in a directory which was previously registered as a static view. Previously, trying to generate a URL to an asset using an absolute file path would raise a ValueError.
  • The RemoteUserAuthenticationPolicy, AuthTktAuthenticationPolicy, and SessionAuthenticationPolicy constructors now accept an additional keyword argument named debug. By default, this keyword argument is False. When it is True, debug information will be sent to the Pyramid debug logger (usually on stderr) when the authenticated_userid or effective_principals method is called on any of these policies. The output produced can be useful when trying to diagnose authentication-related problems.
  • New view predicate: match_param. Example: a view added via config.add_view(aview, match_param='action=edit') will be called only when the request.matchdict has a value inside it named action with a value of edit.
  • Support an onerror keyword argument to pyramid.config.Configurator.scan`(). This argument is passed to venusian.Scanner.scan() to influence error behavior when an exception is raised during scanning.
  • The request_method predicate argument to pyramid.config.Configurator.add_view() and pyramid.config.Configurator.add_route() is now permitted to be a tuple of HTTP method names. Previously it was restricted to being a string representing a single HTTP method name.
  • Undeprecated pyramid.traversal.find_model, pyramid.traversal.model_path, pyramid.traversal.model_path_tuple, and pyramid.url.model_url, which were all deprecated in Pyramid 1.0. There’s just not much cost to keeping them around forever as aliases to their renamed resource_* prefixed functions.
  • Undeprecated pyramid.view.bfg_view, which was deprecated in Pyramid 1.0. This is a low-cost alias to pyramid.view.view_config which we’ll just keep around forever.
  • Route pattern replacement marker names can now begin with an underscore. See https://github.com/Pylons/pyramid/issues/276.

Deprecations

  • All Pyramid-related deployment settings (e.g. debug_all, debug_notfound) are now meant to be prefixed with the prefix pyramid.. For example: debug_all -> pyramid.debug_all. The old non-prefixed settings will continue to work indefinitely but supplying them may print a deprecation warning. All scaffolds and tutorials have been changed to use prefixed settings.
  • The deployment settings dictionary now raises a deprecation warning when you attempt to access its values via __getattr__ instead of via __getitem__.

Backwards Incompatibilities

  • If a string is passed as the debug_logger parameter to a Configurator, that string is considered to be the name of a global Python logger rather than a dotted name to an instance of a logger.

  • The pyramid.config.Configurator.include() method now accepts only a single callable argument. A sequence of callables used to be permitted. If you are passing more than one callable to pyramid.config.Configurator.include(), it will break. You now must now instead make a separate call to the method for each callable.

  • It may be necessary to more strictly order configuration route and view statements when using an “autocommitting” Configurator. In the past, it was possible to add a view which named a route name before adding a route with that name when you used an autocommitting configurator. For example:

    config = Configurator(autocommit=True)
config.add_view('my.pkg.someview', route_name='foo')
config.add_route('foo', '/foo')

    The above will raise an exception when the view attempts to add itself. Now you must add the route before adding the view:

    config = Configurator(autocommit=True)
config.add_route('foo', '/foo')
config.add_view('my.pkg.someview', route_name='foo')

    This won’t effect “normal” users, only people who have legacy BFG codebases that used an autommitting configurator and possibly tests that use the configurator API (the configurator returned by pyramid.testing.setUp() is an autocommitting configurator). The right way to get around this is to use a default non-autocommitting configurator, which does not have these directive ordering requirements:

      config = Configurator()
  config.add_view('my.pkg.someview', route_name='foo')
  config.add_route('foo', '/foo')

The above will work fine.

  • The pyramid.config.Configurator.add_route() directive no longer returns a route object. This change was required to make route vs. view configuration processing work properly.

Behavior Differences

  • An ETag header is no longer set when serving a static file. A Last-Modified header is set instead.
  • Static file serving no longer supports the wsgi.file_wrapper extension.
  • Instead of returning a 403 Forbidden error when a static file is served that cannot be accessed by the Pyramid process’ user due to file permissions, an IOError (or similar) will be raised.

Documentation Enhancements

  • Narrative and API documentation which used the route_url, route_path, resource_url, static_url, and current_route_url functions in the pyramid.url package have now been changed to use eponymous methods of the request instead.
  • Added a section entitled Using a Route Prefix to Compose Applications to the “URL Dispatch” narrative documentation chapter.
  • Added a new module to the API docs: pyramid.tweens.
  • Added a Registering “Tweens” section to the “Hooks” narrative chapter.
  • Added a Displaying “Tweens” section to the “Command-Line Pyramid” narrative chapter.
  • Added documentation for Explicit Tween Configuration and Including Packages to the “Environment Variables and .ini Files Settings” chapter.
  • Added a Logging chapter to the narrative docs.
  • All tutorials now use - The route_url, route_path, resource_url, static_url, and current_route_url methods of the pyramid.request.Request rather than the function variants imported from pyramid.url.
  • The ZODB wiki tutorial now uses the pyramid_zodbconn package rather than the repoze.zodbconn package to provide ZODB integration.
  • Added What Makes Pyramid Unique to the Introduction narrative chapter.

Dependency Changes

  • Pyramid now relies on PasteScript >= 1.7.4. This version contains a feature important for allowing flexible logging configuration.
  • Pyramid now requires Venusian 1.0a1 or better to support the onerror keyword argument to pyramid.config.Configurator.scan().
  • The zope.configuration package is no longer a dependency.

What’s New In Pyramid 1.1

This article explains the new features in Pyramid version 1.1 as compared to its predecessor, Pyramid 1.0. It also documents backwards incompatibilities between the two versions and deprecations added to Pyramid 1.1, as well as software dependency changes and notable documentation additions.

Terminology Changes

The term “template” used by the Pyramid documentation used to refer to both “paster templates” and “rendered templates” (templates created by a rendering engine. i.e. Mako, Chameleon, Jinja, etc.). “Paster templates” will now be refered to as “scaffolds”, whereas the name for “rendered templates” will remain as “templates.”

Major Feature Additions

The major feature additions in Pyramid 1.1 are:

  • Support for the request.response attribute.
  • New views introspection feature: paster pviews.
  • Support for “static” routes.
  • Default HTTP exception view.
  • http_cache view configuration parameter causes Pyramid to set HTTP caching headers.
  • Features that make it easier to write scripts that work in a Pyramid environment.
request.response

  • Instances of the pyramid.request.Request class now have a response attribute.

    The object passed to a view callable as request is an instance of pyramid.request.Request. request.response is an instance of the class pyramid.request.Response. View callables that are configured with a renderer will return this response object to the Pyramid router. Therefore, code in a renderer-using view callable can set response attributes such as request.response.content_type (before they return, e.g. a dictionary to the renderer) and this will influence the HTTP return value of the view callable.

    request.response can also be used in view callable code that is not configured to use a renderer. For example, a view callable might do request.response.body = '123'; return request.response. However, the response object that is produced by request.response must be returned when a renderer is not in play in order to have any effect on the HTTP response (it is not a “global” response, and modifications to it are not somehow merged into a separately returned response object).

    The request.response object is lazily created, so its introduction does not negatively impact performance.

paster pviews
  • A new paster command named paster pviews was added. This command prints a summary of potentially matching views for a given path. See the section entitled Displaying Matching Views for a Given URL for more information.
Static Routes
  • The add_route method of the Configurator now accepts a static argument. If this argument is True, the added route will never be considered for matching when a request is handled. Instead, it will only be useful for URL generation via route_url and route_path. See the section entitled Static Routes for more information.
Default HTTP Exception View

  • A default exception view for the interface pyramid.interfaces.IExceptionResponse is now registered by default. This means that an instance of any exception class imported from pyramid.httpexceptions (such as HTTPFound) can now be raised from within view code; when raised, this exception view will render the exception to a response.

    To allow for configuration of this feature, the Configurator now accepts an additional keyword argument named exceptionresponse_view. By default, this argument is populated with a default exception view function that will be used when an HTTP exception is raised. When None is passed for this value, an exception view for HTTP exceptions will not be registered. Passing None returns the behavior of raising an HTTP exception to that of Pyramid 1.0 (the exception will propagate to middleware and to the WSGI server).

http_cache

A new value http_cache can be used as a view configuration parameter.

When you supply an http_cache value to a view configuration, the Expires and Cache-Control headers of a response generated by the associated view callable are modified. The value for http_cache may be one of the following:

  • A nonzero integer. If it’s a nonzero integer, it’s treated as a number of seconds. This number of seconds will be used to compute the Expires header and the Cache-Control: max-age parameter of responses to requests which call this view. For example: http_cache=3600 instructs the requesting browser to ‘cache this response for an hour, please’.
  • A datetime.timedelta instance. If it’s a datetime.timedelta instance, it will be converted into a number of seconds, and that number of seconds will be used to compute the Expires header and the Cache-Control: max-age parameter of responses to requests which call this view. For example: http_cache=datetime.timedelta(days=1) instructs the requesting browser to ‘cache this response for a day, please’.
  • Zero (0). If the value is zero, the Cache-Control and Expires headers present in all responses from this view will be composed such that client browser cache (and any intermediate caches) are instructed to never cache the response.
  • A two-tuple. If it’s a two tuple (e.g. http_cache=(1, {'public':True})), the first value in the tuple may be a nonzero integer or a datetime.timedelta instance; in either case this value will be used as the number of seconds to cache the response. The second value in the tuple must be a dictionary. The values present in the dictionary will be used as input to the Cache-Control response header. For example: http_cache=(3600, {'public':True}) means ‘cache for an hour, and add public to the Cache-Control header of the response’. All keys and values supported by the webob.cachecontrol.CacheControl interface may be added to the dictionary. Supplying {'public':True} is equivalent to calling response.cache_control.public = True.

Providing a non-tuple value as http_cache is equivalent to calling response.cache_expires(value) within your view’s body.

Providing a two-tuple value as http_cache is equivalent to calling response.cache_expires(value[0], **value[1]) within your view’s body.

If you wish to avoid influencing, the Expires header, and instead wish to only influence Cache-Control headers, pass a tuple as http_cache with the first element of None, e.g.: (None, {'public':True}).

The environment setting PYRAMID_PREVENT_HTTP_CACHE and configuration file value prevent_http_cache are synonymous and allow you to prevent HTTP cache headers from being set by Pyramid’s http_cache machinery globally in a process. see Influencing HTTP Caching and Preventing HTTP Caching.

Easier Scripting Writing

A new API function pyramid.paster.bootstrap() has been added to make writing scripts that need to work under Pyramid environment easier, e.g.:

from pyramid.paster import bootstrap
info = bootstrap('/path/to/my/development.ini')
request = info['request']
print request.route_url('myroute')

See Writing a Script for more details.

Minor Feature Additions

  • It is now possible to invoke paster pshell even if the paste ini file section name pointed to in its argument is not actually a Pyramid WSGI application. The shell will work in a degraded mode, and will warn the user. See “The Interactive Shell” in the “Creating a Pyramid Project” narrative documentation section.

  • The paster pshell, paster pviews, and paster proutes commands each now under the hood uses pyramid.paster.bootstrap(), which makes it possible to supply an .ini file without naming the “right” section in the file that points at the actual Pyramid application. Instead, you can generally just run paster {pshell|proutes|pviews} development.ini and it will do mostly the right thing.

  • It is now possible to add a [pshell] section to your application’s .ini configuration file, which influences the global names available to a pshell session. See Extending the Shell.

  • The pyramid.config.Configurator.scan() method has grown a **kw argument. kw argument represents a set of keyword arguments to pass to the Venusian Scanner object created by Pyramid. (See the Venusian documentation for more information about Scanner).

  • New request property: json_body. This property will return the JSON-decoded variant of the request body. If the request body is not well-formed JSON, this property will raise an exception.

  • A JSONP renderer. See JSONP Renderer for more details.

  • New authentication policy: pyramid.authentication.SessionAuthenticationPolicy, which uses a session to store credentials.

  • A function named pyramid.httpexceptions.exception_response() is a shortcut that can be used to create HTTP exception response objects using an HTTP integer status code.

  • Integers and longs passed as elements to pyramid.url.resource_url() or pyramid.request.Request.resource_url() e.g. resource_url(context, request, 1, 2) (1 and 2 are the elements) will now be converted implicitly to strings in the result. Previously passing integers or longs as elements would cause a TypeError.

  • pyramid_alchemy scaffold now uses query.get rather than query.filter_by to take better advantage of identity map caching.

  • pyramid_alchemy scaffold now has unit tests.

  • Added a pyramid.i18n.make_localizer() API.

  • An exception raised by a pyramid.events.NewRequest event subscriber can now be caught by an exception view.

  • It is now possible to get information about why Pyramid raised a Forbidden exception from within an exception view. The ACLDenied object returned by the permits method of each stock authorization policy (pyramid.interfaces.IAuthorizationPolicy.permits()) is now attached to the Forbidden exception as its result attribute. Therefore, if you’ve created a Forbidden exception view, you can see the ACE, ACL, permission, and principals involved in the request as eg. context.result.permission, context.result.acl, etc within the logic of the Forbidden exception view.

  • Don’t explicitly prevent the timeout from being lower than the reissue_time when setting up an pyramid.authentication.AuthTktAuthenticationPolicy (previously such a configuration would raise a ValueError, now it’s allowed, although typically nonsensical). Allowing the nonsensical configuration made the code more understandable and required fewer tests.

  • The pyramid.request.Request class now has a ResponseClass attribute which points at pyramid.response.Response.

  • The pyramid.response.Response class now has a RequestClass interface which points at pyramid.request.Request.

  • It is now possible to return an arbitrary object from a Pyramid view callable even if a renderer is not used, as long as a suitable adapter to pyramid.interfaces.IResponse is registered for the type of the returned object by using the new pyramid.config.Configurator.add_response_adapter() API. See the section in the Hooks chapter of the documentation entitled Changing How Pyramid Treats View Responses.

  • The Pyramid router will now, by default, call the __call__ method of response objects when returning a WSGI response. This means that, among other things, the conditional_response feature response objects inherited from WebOb will now behave properly.

  • New method named pyramid.request.Request.is_response(). This method should be used instead of the pyramid.view.is_response() function, which has been deprecated.

  • pyramid.exceptions.NotFound is now just an alias for pyramid.httpexceptions.HTTPNotFound.

  • pyramid.exceptions.Forbidden is now just an alias for pyramid.httpexceptions.HTTPForbidden.

  • Added mako.preprocessor config file parameter; allows for a Mako preprocessor to be specified as a Python callable or Python dotted name. See https://github.com/Pylons/pyramid/pull/183 for rationale.

  • New API class: pyramid.static.static_view. This supersedes the (now deprecated) pyramid.view.static class. pyramid.static.static_view, by default, serves up documents as the result of the request’s path_info, attribute rather than it’s subpath attribute (the inverse was true of pyramid.view.static, and still is). pyramid.static.static_view exposes a use_subpath flag for use when you want the static view to behave like the older deprecated version.

  • A new api function pyramid.scripting.prepare() has been added. It is a lower-level analogue of pyramid.paster.boostrap() that accepts a request and a registry instead of a config file argument, and is used for the same purpose:

    from pyramid.scripting import prepare
info = prepare(registry=myregistry)
request = info['request']
print request.route_url('myroute')

  • A new API function pyramid.scripting.make_request() has been added. The resulting request will have a registry attribute. It is meant to be used in conjunction with pyramid.scripting.prepare() and/or pyramid.paster.bootstrap() (both of which accept a request as an argument):

    from pyramid.scripting import make_request
request = make_request('/')

  • New API attribute pyramid.config.global_registries is an iterable object that contains references to every Pyramid registry loaded into the current process via pyramid.config.Configurator.make_app(). It also has a last attribute containing the last registry loaded. This is used by the scripting machinery, and is available for introspection.

  • Added the pyramid.renderers.null_renderer object as an API. The null renderer is an object that can be used in advanced integration cases as input to the view configuration renderer= argument. When the null renderer is used as a view renderer argument, Pyramid avoids converting the view callable result into a Response object. This is useful if you want to reuse the view configuration and lookup machinery outside the context of its use by the Pyramid router. (This feature was added for consumption by the pyramid_rpc package, which uses view configuration and lookup outside the context of a router in exactly this way.)

Backwards Incompatibilities

  • Pyramid no longer supports Python 2.4. Python 2.5 or better is required to run Pyramid 1.1+. Pyramid, however, does not work under any version of Python 3 yet.

  • The Pyramid router now, by default, expects response objects returned from view callables to implement the pyramid.interfaces.IResponse interface. Unlike the Pyramid 1.0 version of this interface, objects which implement IResponse now must define a __call__ method that accepts environ and start_response, and which returns an app_iter iterable, among other things. Previously, it was possible to return any object which had the three WebOb app_iter, headerlist, and status attributes as a response, so this is a backwards incompatibility. It is possible to get backwards compatibility back by registering an adapter to IResponse from the type of object you’re now returning from view callables. See the section in the Hooks chapter of the documentation entitled Changing How Pyramid Treats View Responses.

  • The pyramid.interfaces.IResponse interface is now much more extensive. Previously it defined only app_iter, status and headerlist; now it is basically intended to directly mirror the webob.Response API, which has many methods and attributes.

  • The pyramid.httpexceptions classes named HTTPFound, HTTPMultipleChoices, HTTPMovedPermanently, HTTPSeeOther, HTTPUseProxy, and HTTPTemporaryRedirect now accept location as their first positional argument rather than detail. This means that you can do, e.g. return pyramid.httpexceptions.HTTPFound('http://foo') rather than return pyramid.httpexceptions.HTTPFound(location='http//foo') (the latter will of course continue to work).

  • The pyramid Router attempted to set a value into the key environ['repoze.bfg.message'] when it caught a view-related exception for backwards compatibility with applications written for repoze.bfg during error handling. It did this by using code that looked like so:

    # "why" is an exception object
try:
    msg = why[0]
except:
    msg = ''

environ['repoze.bfg.message'] = msg

    Use of the value environ['repoze.bfg.message'] was docs-deprecated in Pyramid 1.0. Our standing policy is to not remove features after a deprecation for two full major releases, so this code was originally slated to be removed in Pyramid 1.2. However, computing the repoze.bfg.message value was the source of at least one bug found in the wild (https://github.com/Pylons/pyramid/issues/199), and there isn’t a foolproof way to both preserve backwards compatibility and to fix the bug. Therefore, the code which sets the value has been removed in this release. Code in exception views which relies on this value’s presence in the environment should now use the exception attribute of the request (e.g. request.exception[0]) to retrieve the message instead of relying on request.environ['repoze.bfg.message'].

Deprecations and Behavior Differences

Note

Under Python 2.7+, it’s necessary to pass the Python interpreter the correct warning flags to see deprecation warnings emitted by Pyramid when porting your application from an older version of Pyramid. Use the PYTHONWARNINGS environment variable with the value all in the shell you use to invoke paster serve to see these warnings, e.g. on UNIX, PYTHONWARNINGS=all bin/paster serve development.ini. Python 2.5 and 2.6 show deprecation warnings by default, so this is unecessary there. All deprecation warnings are emitted to the console.

  • The pyramid.view.static class has been deprecated in favor of the newer pyramid.static.static_view class. A deprecation warning is raised when it is used. You should replace it with a reference to pyramid.static.static_view with the use_subpath=True argument.

  • The paster pshell, paster proutes, and paster pviews commands now take a single argument in the form /path/to/config.ini#sectionname rather than the previous 2-argument spelling /path/to/config.ini sectionname. #sectionname may be omitted, in which case #main is assumed.

  • The default Mako renderer is now configured to escape all HTML in expression tags. This is intended to help prevent XSS attacks caused by rendering unsanitized input from users. To revert this behavior in user’s templates, they need to filter the expression through the ‘n’ filter:

    ${ myhtml | n }.

    See https://github.com/Pylons/pyramid/issues/193.

  • Deprecated all assignments to request.response_* attributes (for example request.response_content_type = 'foo' is now deprecated). Assignments and mutations of assignable request attributes that were considered by the framework for response influence are now deprecated: response_content_type, response_headerlist, response_status, response_charset, and response_cache_for. Instead of assigning these to the request object for later detection by the rendering machinery, users should use the appropriate API of the Response object created by accessing request.response (e.g. code which does request.response_content_type = 'abc' should be changed to request.response.content_type = 'abc').

  • Passing view-related parameters to pyramid.config.Configurator.add_route() is now deprecated. Previously, a view was permitted to be connected to a route using a set of view* parameters passed to the add_route method of the Configurator. This was a shorthand which replaced the need to perform a subsequent call to add_view. For example, it was valid (and often recommended) to do:

    config.add_route('home', '/', view='mypackage.views.myview',
                  view_renderer='some/renderer.pt')

    Passing view* arguments to add_route is now deprecated in favor of connecting a view to a predefined route via pyramid.config.Configurator.add_view() using the route’s route_name parameter. As a result, the above example should now be spelled:

    config.add_route('home', '/')
config.add_view('mypackage.views.myview', route_name='home',
                renderer='some/renderer.pt')

    This deprecation was done to reduce confusion observed in IRC, as well as to (eventually) reduce documentation burden (see also https://github.com/Pylons/pyramid/issues/164). A deprecation warning is now issued when any view-related parameter is passed to add_route.

  • Passing an environ dictionary to the __call__ method of a “traverser” (e.g. an object that implements pyramid.interfaces.ITraverser such as an instance of pyramid.traversal.ResourceTreeTraverser) as its request argument now causes a deprecation warning to be emitted. Consumer code should pass a request object instead. The fact that passing an environ dict is permitted has been documentation-deprecated since repoze.bfg 1.1, and this capability will be removed entirely in a future version.

  • The following (undocumented, dictionary-like) methods of the pyramid.request.Request object have been deprecated: __contains__, __delitem__, __getitem__, __iter__, __setitem__, get, has_key, items, iteritems, itervalues, keys, pop, popitem, setdefault, update, and values. Usage of any of these methods will cause a deprecation warning to be emitted. These methods were added for internal compatibility in repoze.bfg 1.1 (code that currently expects a request object expected an environ object in BFG 1.0 and before). In a future version, these methods will be removed entirely.

  • A custom request factory is now required to return a request object that has a response attribute (or “reified”/lazy property) if the request is meant to be used in a view that uses a renderer. This response attribute should be an instance of the class pyramid.response.Response.

  • The JSON and string renderer factories now assign to request.response.content_type rather than request.response_content_type.

  • Each built-in renderer factory now determines whether it should change the content type of the response by comparing the response’s content type against the response’s default content type; if the content type is the default content type (usually text/html), the renderer changes the content type (to application/json or text/plain for JSON and string renderers respectively).

  • The pyramid.wsgi.wsgiapp2() now uses a slightly different method of figuring out how to “fix” SCRIPT_NAME and PATH_INFO for the downstream application. As a result, those values may differ slightly from the perspective of the downstream application (for example, SCRIPT_NAME will now never possess a trailing slash).

  • Previously, pyramid.request.Request inherited from webob.request.Request and implemented __getattr__, __setattr__ and __delattr__ itself in order to override “adhoc attr” WebOb behavior where attributes of the request are stored in the environ. Now, pyramid.request.Request inherits from (the more recent) webob.request.BaseRequest instead of webob.request.Request, which provides the same behavior. pyramid.request.Request no longer implements its own __getattr__, __setattr__ or __delattr__ as a result.

  • Deprecated pyramid.view.is_response() function in favor of (newly-added) pyramid.request.Request.is_response() method. Determining if an object is truly a valid response object now requires access to the registry, which is only easily available as a request attribute. The pyramid.view.is_response() function will still work until it is removed, but now may return an incorrect answer under some (very uncommon) circumstances.

  • pyramid.response.Response is now a subclass of webob.response.Response (in order to directly implement the pyramid.interfaces.IResponse interface, to speed up response generation).

  • The “exception response” objects importable from pyramid.httpexceptions (e.g. HTTPNotFound) are no longer just import aliases for classes that actually live in webob.exc. Instead, we’ve defined our own exception classes within the module that mirror and emulate the webob.exc exception response objects almost entirely. See Pyramid Uses its Own HTTP Exception Class Hierarchy Rather Than webob.exc in the Design Defense chapter for more information.

  • When visiting a URL that represented a static view which resolved to a subdirectory, the index.html of that subdirectory would not be served properly. Instead, a redirect to /subdir would be issued. This has been fixed, and now visiting a subdirectory that contains an index.html within a static view returns the index.html properly. See also https://github.com/Pylons/pyramid/issues/67.

  • Deprecated the pyramid.config.Configurator.set_renderer_globals_factory() method and the renderer_globals Configurator constructor parameter. Users should convert code using this feature to use a BeforeRender event. See the section Using The Before Render Event in the Hooks chapter.

  • In Pyramid 1.0, the pyramid.events.subscriber directive behaved contrary to the documentation when passed more than one interface object to its constructor. For example, when the following listener was registered:

    @subscriber(IFoo, IBar)
def expects_ifoo_events_and_ibar_events(event):
    print event

    The Events chapter docs claimed that the listener would be registered and listening for both IFoo and IBar events. Instead, it registered an “object event” subscriber which would only be called if an IObjectEvent was emitted where the object interface was IFoo and the event interface was IBar.

    The behavior now matches the documentation. If you were relying on the buggy behavior of the 1.0 subscriber directive in order to register an object event subscriber, you must now pass a sequence to indicate you’d like to register a subscriber for an object event. e.g.:

    @subscriber([IFoo, IBar])
def expects_object_event(object, event):
    print object, event

  • In 1.0, if a pyramid.events.BeforeRender event subscriber added a value via the __setitem__ or update methods of the event object with a key that already existed in the renderer globals dictionary, a KeyError was raised. With the deprecation of the “add_renderer_globals” feature of the configurator, there was no way to override an existing value in the renderer globals dictionary that already existed. Now, the event object will overwrite an older value that is already in the globals dictionary when its __setitem__ or update is called (as well as the new setdefault method), just like a plain old dictionary. As a result, for maximum interoperability with other third-party subscribers, if you write an event subscriber meant to be used as a BeforeRender subscriber, your subscriber code will now need to (using .get or __contains__ of the event object) ensure no value already exists in the renderer globals dictionary before setting an overriding value.

  • The pyramid.config.Configurator.add_route() method allowed two routes with the same route to be added without an intermediate call to pyramid.config.Configurator.commit(). If you now receive a ConfigurationError at startup time that appears to be add_route related, you’ll need to either a) ensure that all of your route names are unique or b) call config.commit() before adding a second route with the name of a previously added name or c) use a Configurator that works in autocommit mode.

Dependency Changes

  • Pyramid now depends on WebOb >= 1.0.2 as tests depend on the bugfix in that release: “Fix handling of WSGI environs with missing SCRIPT_NAME”. (Note that in reality, everyone should probably be using 1.0.4 or better though, as WebOb 1.0.2 and 1.0.3 were effectively brownbag releases.)

Documentation Enhancements

  • Added a section entitled Writing a Script to the “Command-Line Pyramid” chapter.
  • The ZODB + Traversal Wiki Tutorial was updated slightly.
  • The SQLAlchemy + URL Dispatch Wiki Tutorial was updated slightly.
  • Made pyramid.interfaces.IAuthenticationPolicy and pyramid.interfaces.IAuthorizationPolicy public interfaces, and they are now referred to within the pyramid.authentication and pyramid.authorization API docs.
  • Render the function definitions for each exposed interface in pyramid.interfaces.
  • Add missing docs reference to pyramid.config.Configurator.set_view_mapper() and refer to it within the documentation section entitled Using a View Mapper.
  • Added section to the “Environment Variables and .ini File Settings” chapter in the narrative documentation section entitled Adding A Custom Setting.
  • Added documentation for a multidict as pyramid.interfaces.IMultiDict.
  • Added a section to the “URL Dispatch” narrative chapter regarding the new “static” route feature entitled Static Routes.
  • Added API docs for pyramid.httpexceptions.exception_response().
  • Added HTTP Exceptions section to Views narrative chapter including a description of pyramid.httpexceptions.exception_response().
  • Added API docs for pyramid.authentication.SessionAuthenticationPolicy.

What’s New In Pyramid 1.0

This article explains the new features in Pyramid version 1.0 as compared to its predecessor, repoze.bfg 1.3. It also documents backwards incompatibilities between the two versions and deprecations added to Pyramid 1.0, as well as software dependency changes and notable documentation additions.

Major Feature Additions

The major feature additions in Pyramid 1.0 are:

  • New name and branding association with the Pylons Project.
  • BFG conversion script
  • Scaffold improvements
  • Terminology changes
  • Better platform compatibility and support
  • Direct built-in support for the Mako templating language.
  • Built-in support for sessions.
  • Updated URL dispatch features
  • Better imperative extensibility
  • ZCML externalized
  • Better support for global template variables during rendering
  • View mappers
  • Testing system improvements
  • Authentication support improvements
  • Documentation improvements
New Name and Branding

The name of repoze.bfg has been changed to Pyramid. The project is also now a subproject of a new entity, “The Pylons Project”. The Pylons Project is the project name for a collection of web-framework-related technologies. Pyramid was the first package in the Pylons Project. Other packages to the collection have been added over time, such as support packages useful for Pylons 1 users as well as ex-Zope users. Pyramid is the successor to both repoze.bfg and Pylons version 1.

The Pyramid codebase is derived almost entirely from repoze.bfg with some changes made for the sake of Pylons 1 compatibility.

Pyramid is technically backwards incompatible with repoze.bfg, as it has a new package name, so older imports from the repoze.bfg module will fail if you do nothing to your existing repoze.bfg application. However, you won’t have to do much to use your existing BFG applications on Pyramid. There’s automation which will change most of your import statements and ZCML declarations. See http://docs.pylonsproject.org/projects/pyramid/current/tutorials/bfg/index.html for upgrade instructions.

Pylons 1 users will need to do more work to use Pyramid, as Pyramid shares no “DNA” with Pylons. It is hoped that over time documentation and upgrade code will be developed to help Pylons 1 users transition to Pyramid more easily.

repoze.bfg version 1.3 will be its last major release. Minor updates will be made for critical bug fixes. Pylons version 1 will continue to see maintenance releases, as well.

The Repoze project will continue to exist. Repoze will be able to regain its original focus: bringing Zope technologies to WSGI. The popularity of repoze.bfg as its own web framework hindered this goal.

We hope that people are attracted at first by the spirit of cooperation demonstrated by the Pylons Project and the merging of development communities. It takes humility to sacrifice a little sovereignty and work together. The opposite, forking or splintering of projects, is much more common in the open source world. We feel there is a limited amount of oxygen in the space of “top-tier” Python web frameworks and we don’t do the Python community a service by over-crowding. By merging the repoze.bfg and the philosophically-similar Pylons communities, both gain an expanded audience and a stronger chance of future success.

BFG Conversion Script

The bfg2pyramid conversion script performs a mostly automated conversion of an existing repoze.bfg application to Pyramid. The process is described in Converting a repoze.bfg Application to Pyramid.

Scaffold Improvements
  • The scaffolds now have much nicer CSS and graphics.
  • The development.ini, generated by all scaffolds, is now configured to use the WebError interactive exception debugger by default.
  • All scaffolds have been normalized: each now uses the name main to represent the function that returns a WSGI application, and each now has roughly the same shape of development.ini style.
  • All preexisting scaffolds now use “imperative” configuration (starter, routesalchemy, alchemy, zodb) instead of ZCML configuration.
  • The pyramid_zodb, pyramid_routesalchemy and pyramid_alchemy scaffolds now use a default “commit veto” hook when configuring the repoze.tm2 transaction manager in development.ini. This prevents a transaction from being committed when the response status code is within the 400 or 500 ranges. See also http://docs.repoze.org/tm2/#using-a-commit-veto.
Terminology Changes

  • The Pyramid concept previously known as “model” is now known as “resource”. As a result, the following API renames have been made. Backwards compatibility shims for the old names have been left in place in all cases:

    pyramid.url.model_url ->
                  pyramid.url.resource_url

pyramid.traversal.find_model ->
                  pyramid.url.find_resource

pyramid.traversal.model_path ->
                  pyramid.traversal.resource_path

pyramid.traversal.model_path_tuple ->
                  pyramid.traversal.resource_path_tuple

pyramid.traversal.ModelGraphTraverser ->
                  pyramid.traversal.ResourceTreeTraverser

pyramid.config.Configurator.testing_models ->
                  pyramid.config.Configurator.testing_resources

pyramid.testing.registerModels ->
                  pyramid.testing.registerResources

pyramid.testing.DummyModel ->
                  pyramid.testing.DummyResource

  • All documentation which previously referred to “model” now refers to “resource”.

  • The starter scaffold now has a resources.py module instead of a models.py module.

  • Positional argument names of various APIs have been changed from model to resource.

  • The Pyramid concept previously known as “resource” is now known as “asset”. As a result, the following API changes were made. Backwards compatibility shims have been left in place as necessary:

    pyramid.config.Configurator.absolute_resource_spec ->
                  pyramid.config.Configurator.absolute_asset_spec

pyramid.config.Configurator.override_resource ->
                  pyramid.config.Configurator.override_asset

  • The (non-API) module previously known as pyramid.resource is now known as pyramid.asset.

  • All docs that previously referred to “resource specification” now refer to “asset specification”.

  • The setting previously known as BFG_RELOAD_RESOURCES (envvar) or reload_resources (config file) is now known, respectively, as PYRAMID_RELOAD_ASSETS and reload_assets.

Better Platform Compatibility and Support

We’ve made Pyramid’s test suite pass on both Jython and PyPy. However, Chameleon doesn’t work on either, so you’ll need to use Mako or Jinja2 templates on these platforms.

Sessions

Pyramid now has built-in sessioning support, documented in Sessions. The sessioning implementation is pluggable. It also provides flash messaging and cross-site-scripting prevention features.

Using request.session now returns a (dictionary-like) session object if a session factory has been configured.

A new argument to the Configurator constructor exists: session_factory and a new method on the configurator exists: pyramid.config.Configurator.set_session_factory().

Mako

In addition to Chameleon templating, Pyramid now also provides built-in support for Mako templating. See Templating With Mako Templates for more information.

URL Dispatch
  • URL Dispatch now allows for replacement markers to be located anywhere in the pattern, instead of immediately following a /.
  • URL Dispatch now uses the form {marker} to denote a replace marker in the route pattern instead of :marker. The old colon-style marker syntax is still accepted for backwards compatibility. The new format allows a regular expression for that marker location to be used instead of the default [^/]+, for example {marker:\d+} is now valid to require the marker to be digits.
  • Addded a new API pyramid.url.current_route_url(), which computes a URL based on the “current” route (if any) and its matchdict values.
  • Added a paster proute command which displays a summary of the routing table. See the narrative documentation section entitled Displaying All Application Routes.
  • Added debug_routematch configuration setting (settable in your .ini file) that logs matched routes including the matchdict and predicates.
  • Add a pyramid.url.route_path() API, allowing folks to generate relative URLs. Calling route_path is the same as calling pyramid.url.route_url() with the argument _app_url equal to the empty string.
  • Add a pyramid.request.Request.route_path() API. This is a convenience method of the request which calls pyramid.url.route_url().
  • Added class vars matchdict and matched_route to pyramid.request.Request. Each is set to None when a route isn’t matched during a request.
ZCML Externalized
  • The load_zcml method of a Configurator has been removed from the Pyramid core. Loading ZCML is now a feature of the pyramid_zcml package, which can be downloaded from PyPI. Documentation for the package should be available via http://pylonsproject.org/projects/pyramid_zcml/dev/, which describes how to add a configuration statement to your main block to reobtain this method. You will also need to add an install_requires dependency upon the pyramid_zcml distribution to your setup.py file.
  • The “Declarative Configuration” narrative chapter has been removed (it was moved to the pyramid_zcml package).
  • Most references to ZCML in narrative chapters have been removed or redirected to pyramid_zcml locations.
  • The starter_zcml paster scaffold has been moved to the pyramid_zcml package.
Imperative Two-Phase Configuration

To support application extensibility, the Pyramid Configurator, by default, now detects configuration conflicts and allows you to include configuration imperatively from other packages or modules. It also, by default, performs configuration in two separate phases. This allows you to ignore relative configuration statement ordering in some circumstances. See Advanced Configuration for more information.

The pyramid.config.Configurator.add_directive() allows framework extenders to add methods to the configurator, which allows extenders to avoid subclassing a Configurator just to add methods. See Adding Methods to the Configurator via add_directive for more info.

Surrounding application configuration with config.begin() and config.end() is no longer necessary. All scaffolds have been changed to no longer call these functions.

Better Support for Global Template Variables During Rendering

A new event type named pyramid.interfaces.IBeforeRender is now sent as an event before a renderer is invoked. Applications may now subscribe to the IBeforeRender event type in order to introspect the and modify the set of renderer globals before they are passed to a renderer. The event object iself has a dictionary-like interface that can be used for this purpose. For example:

from repoze.events import subscriber
from pyramid.interfaces import IRendererGlobalsEvent

@subscriber(IRendererGlobalsEvent)
def add_global(event):
    event['mykey'] = 'foo'
View Mappers

A “view mapper” subsystem has been extracted, which allows framework extenders to control how view callables are constructed and called. This feature is not useful for “civilians”, only for extension writers. See Using a View Mapper for more information.

Testing Support Improvements

The pyramid.testing.setUp() and pyramid.testing.tearDown() APIs have been undeprecated. They are now the canonical setup and teardown APIs for test configuration, replacing “direct” creation of a Configurator. This is a change designed to provide a facade that will protect against any future Configurator deprecations.

Authentication Support Improvements
  • The pyramid.interfaces.IAuthenticationPolicy interface now specifies an unauthenticated_userid method. This method supports an important optimization required by people who are using persistent storages which do not support object caching and whom want to create a “user object” as a request attribute.
  • A new API has been added to the pyramid.security module named unauthenticated_userid. This API function calls the unauthenticated_userid method of the effective security policy.
  • The class pyramid.authentication.AuthTktCookieHelper is now an API. This class can be used by third-party authentication policy developers to help in the mechanics of authentication cookie-setting.
  • The pyramid.authentication.AuthTktAuthenticationPolicy now accepts a tokens parameter via pyramid.security.remember(). The value must be a sequence of strings. Tokens are placed into the auth_tkt “tokens” field and returned in the auth_tkt cookie.
  • Added a wild_domain argument to pyramid.authentication.AuthTktAuthenticationPolicy, which defaults to True. If it is set to False, the feature of the policy which sets a cookie with a wilcard domain will be turned off.
Documentation Improvements
  • Casey Duncan, a good friend, and an excellent technical writer has given us the gift of professionally editing the entire Pyramid documentation set. Any faults in the documentation are the development team’s, and all improvements are his.
  • The “Resource Location and View Lookup” chapter has been replaced with a variant of Rob Miller’s “Much Ado About Traversal” (originally published at http://blog.nonsequitarian.org/2010/much-ado-about-traversal/).
  • Many users have contributed documentation fixes and improvements including Ben Bangert, Blaise Laflamme, Rob Miller, Mike Orr, Carlos de la Guardia, Paul Everitt, Tres Seaver, John Shipman, Marius Gedminas, Chris Rossi, Joachim Krebs, Xavier Spriet, Reed O’Brien, William Chambers, Charlie Choiniere, and Jamaludin Ahmad.

Minor Feature Additions

  • The settings dictionary passed to the Configurator is now available as config.registry.settings in configuration code and request.registry.settings in view code).
  • pyramid.config.Configurator.add_view() now accepts a decorator keyword argument, a callable which will decorate the view callable before it is added to the registry.
  • Allow static renderer provided during view registration to be overridden at request time via a request attribute named override_renderer, which should be the name of a previously registered renderer. Useful to provide “omnipresent” RPC using existing rendered views.
  • If a resource implements a __resource_url__ method, it will be called as the result of invoking the pyramid.url.resource_url() function to generate a URL, overriding the default logic. See Generating The URL Of A Resource for more information.
  • The name registry is now available in a pshell environment by default. It is the application registry object.
  • Added support for json on Google App Engine by catching NotImplementedError and importing simplejson from django.utils.
  • Added the pyramid.httpexceptions module, which is a facade for the webob.exc module.
  • New class: pyramid.response.Response. This is a pure facade for webob.Response (old code need not change to use this facade, it’s existence is mostly for vanity and documentation-generation purposes).
  • The request now has a new attribute: tmpl_context for benefit of Pylons users.
  • New API methods for pyramid.request.Request: model_url, route_url, and static_url. These are simple passthroughs for their respective functions in pyramid.url.

Backwards Incompatibilities

  • When a pyramid.exceptions.Forbidden error is raised, its status code now 403 Forbidden. It was previously 401 Unauthorized, for backwards compatibility purposes with repoze.bfg. This change will cause problems for users of Pyramid with repoze.who, which intercepts 401 Unauthorized by default, but allows 403 Forbidden to pass through. Those deployments will need to configure repoze.who to also react to 403 Forbidden. To do so, use a repoze.who challenge_decider that looks like this:

    import zope.interface
from repoze.who.interfaces import IChallengeDecider

def challenge_decider(environ, status, headers):
    return status.startswith('403') or status.startswith('401')
zope.interface.directlyProvides(challenge_decider, IChallengeDecider)

  • The paster bfgshell command is now known as paster pshell.

  • There is no longer an IDebugLogger object registered as a named utility with the name repoze.bfg.debug.

  • These deprecated APIs have been removed: pyramid.testing.registerViewPermission, pyramid.testing.registerRoutesMapper, pyramid.request.get_request, pyramid.security.Unauthorized, pyramid.view.view_execution_permitted, pyramid.view.NotFound

  • The Venusian “category” for all built-in Venusian decorators (e.g. subscriber and view_config/bfg_view) is now pyramid instead of bfg.

  • The pyramid.renderers.rendered_response function removed; use pyramid.renderers.render_to_response() instead.

  • Renderer factories now accept a renderer info object rather than an absolute resource specification or an absolute path. The object has the following attributes: name (the renderer= value), package (the ‘current package’ when the renderer configuration statement was found), type: the renderer type, registry: the current registry, and settings: the deployment settings dictionary. Third-party repoze.bfg renderer implementations that must be ported to Pyramid will need to account for this. This change was made primarily to support more flexible Mako template rendering.

  • The presence of the key repoze.bfg.message in the WSGI environment when an exception occurs is now deprecated. Instead, code which relies on this environ value should use the exception attribute of the request (e.g. request.exception[0]) to retrieve the message.

  • The values bfg_localizer and bfg_locale_name kept on the request during internationalization for caching purposes were never APIs. These however have changed to localizer and locale_name, respectively.

  • The default cookie_name value of the pyramid.authentication.AuthTktAuthenticationPolicy now defaults to auth_tkt (it used to default to repoze.bfg.auth_tkt).

  • The pyramid.testing.zcml_configure() API has been removed. It had been advertised as removed since repoze.bfg 1.2a1, but hadn’t actually been.

  • All environment variables which used to be prefixed with BFG_ are now prefixed with PYRAMID_ (e.g. BFG_DEBUG_NOTFOUND is now PYRAMID_DEBUG_NOTFOUND)

  • Since the pyramid.interfaces.IAuthenticationPolicy interface now specifies that a policy implementation must implement an unauthenticated_userid method, all third-party custom authentication policies now must implement this method. It, however, will only be called when the global function named pyramid.security.unauthenticated_userid() is invoked, so if you’re not invoking that, you will not notice any issues.

  • The configure_zcml setting within the deployment settings (within **settings passed to a Pyramid main function) has ceased to have any meaning.

  • The make_app function has been removed from the pyramid.router module. It continues life within the pyramid_zcml package. This leaves the pyramid.router module without any API functions.

Deprecations and Behavior Differences

  • pyramid.configuration.Configurator is now deprecated. Use pyramid.config.Configurator, passing its constructor autocommit=True instead. The pyramid.configuration.Configurator alias will live for a long time, as every application uses it, but its import now issues a deprecation warning. The pyramid.config.Configurator class has the same API as the pyramid.configuration.Configurator class, which it means to replace, except by default it is a non-autocommitting configurator. The now-deprecated pyramid.configuration.Configurator will autocommit every time a configuration method is called. The pyramid.configuration module remains, but it is deprecated. Use pyramid.config instead.
  • The pyramid.settings.get_settings() API is now deprecated. Use pyramid.threadlocals.get_current_registry().settings instead or use the settings attribute of the registry available from the request (request.registry.settings).
  • The decorator previously known as pyramid.view.bfg_view is now known most formally as pyramid.view.view_config in docs and scaffolds.
  • Obtaining the settings object via registry.{get|query}Utility(ISettings) is now deprecated. Instead, obtain the settings object via the registry.settings attribute. A backwards compatibility shim was added to the registry object to register the settings object as an ISettings utility when setattr(registry, 'settings', foo) is called, but it will be removed in a later release.
  • Obtaining the settings object via pyramid.settings.get_settings() is now deprecated. Obtain it instead as the settings attribute of the registry now (obtain the registry via pyramid.threadlocal.get_registry() or as request.registry).

Dependency Changes

  • Depend on Venusian >= 0.5 (for scanning conflict exception decoration).

Documentation Enhancements

  • Added a pyramid.httpexceptions API documentation chapter.
  • Added a pyramid.session API documentation chapter.
  • Added an API chapter for the pyramid.response module.
  • Added a Sessions narrative documentation chapter.
  • All documentation which previously referred to webob.Response now uses pyramid.response.Response instead.
  • The documentation has been overhauled to use imperative configuration, moving declarative configuration (ZCML) explanations to an external package, pyramid_zcml.
  • Removed zodbsessions tutorial chapter. It’s still useful, but we now have a SessionFactory abstraction which competes with it, and maintaining documentation on both ways to do it is a distraction.
  • Added an example of WebTest functional testing to the testing narrative chapter at Creating Functional Tests.
  • Extended the Resources chapter with examples of calls to resource-related APIs.
  • Add “Pyramid Provides More Than One Way to Do It” to Design Defense documentation.
  • The (weak) “Converting a CMF Application to Pyramid” tutorial has been removed from the tutorials section. It was moved to the pyramid_tutorials Github repository at http://docs.pylonsproject.org/projects/pyramid_tutorials/dev/.
  • Moved “Using ZODB With ZEO” and “Using repoze.catalog Within Pyramid” tutorials out of core documentation and into the Pyramid Tutorials site (http://docs.pylonsproject.org/projects/pyramid_tutorials/dev/).
  • Removed API documentation for deprecated pyramid.testing APIs named registerDummySecurityPolicy, registerResources, registerModels, registerEventListener, registerTemplateRenderer, registerDummyRenderer, registerView, registerUtility, registerAdapter, registerSubscriber, registerRoute, and registerSettings.

Narrative documentation

Narrative documentation in chapter form explaining how to use Pyramid.

Pyramid Introduction

Pyramid is a general, open source, Python web application development framework. Its primary goal is to make it easier for a Python developer to create web applications.

Frameworks vs. Libraries

A framework differs from a library in one very important way: library code is always called by code that you write, while a framework always calls code that you write. Using a set of libraries to create an application is usually easier than using a framework initially, because you can choose to cede control to library code you have not authored very selectively. But when you use a framework, you are required to cede a greater portion of control to code you have not authored: code that resides in the framework itself. You needn’t use a framework at all to create a web application using Python. A rich set of libraries already exists for the platform. In practice, however, using a framework to create an application is often more practical than rolling your own via a set of libraries if the framework provides a set of facilities that fits your application requirements.

Pyramid attempts to follow these design and engineering principles:

Simplicity
Pyramid takes a “pay only for what you eat” approach. You can get results even if you have only a partial understanding of Pyramid. It doesn’t force you to use any particular technology to produce an application, and we try to keep the core set of concepts that you need to understand to a minimum.
Minimalism
Pyramid tries to solve only the fundamental problems of creating a web application: the mapping of URLs to code, templating, security and serving static assets. We consider these to be the core activities that are common to nearly all web applications.
Documentation
Pyramid’s minimalism means that it is easier for us to maintain complete and up-to-date documentation. It is our goal that no aspect of Pyramid is undocumented.
Speed
Pyramid is designed to provide noticeably fast execution for common tasks such as templating and simple response generation. Although “hardware is cheap”, the limits of this approach become painfully evident when one finds him or herself responsible for managing a great many machines.
Reliability
Pyramid is developed conservatively and tested exhaustively. Where Pyramid source code is concerned, our motto is: “If it ain’t tested, it’s broke”.
Openness
As with Python, the Pyramid software is distributed under a permissive open source license.

What Makes Pyramid Unique

Understandably, people don’t usually want to hear about squishy engineering principles, they want to hear about concrete stuff that solves their problems. With that in mind, what would make someone want to use Pyramid instead of one of the many other web frameworks available today? What makes Pyramid unique?

This is a hard question to answer, because there are lots of excellent choices, and it’s actually quite hard to make a wrong choice, particularly in the Python web framework market. But one reasonable answer is this: you can write very small applications in Pyramid without needing to know a lot. “What?”, you say, “that can’t possibly be a unique feature, lots of other web frameworks let you do that!” Well, you’re right. But unlike many other systems, you can also write very large applications in Pyramid if you learn a little more about it. Pyramid will allow you to become productive quickly, and will grow with you; it won’t hold you back when your application is small and it won’t get in your way when your application becomes large. “Well that’s fine,” you say, “lots of other frameworks let me write large apps too.” Absolutely. But other Python web frameworks don’t seamlessly let you do both. They seem to fall into two non-overlapping categories: frameworks for “small apps” and frameworks for “big apps”. The “small app” frameworks typically sacrifice “big app” features, and vice versa.

We don’t think it’s a universally reasonable suggestion to write “small apps” in a “small framework” and “big apps” in a “big framework”. You can’t really know to what size every application will eventually grow. We don’t really want to have to rewrite a previously small application in another framework when it gets “too big”. We believe the current binary distinction between frameworks for small and large applications is just false; a well-designed framework should be able to be good at both. Pyramid strives to be that kind of framework.

To this end, Pyramid provides a set of features, that, combined, are unique amongst Python web frameworks. Lots of other frameworks contain some combination of these features; Pyramid of course actually stole many of them from those other frameworks. But Pyramid is the only one that has all of them in one place, documented appropriately, and useful a la carte without necessarily paying for the entire banquet. These are detailed below.

Single-file applications

You can write a Pyramid application that lives entirely in one Python file, not unlike existing Python microframeworks. This is beneficial for one-off prototyping, bug reproduction, and very small applications. These applications are easy to understand because all the information about the application lives in a single place, and you can deploy them without needing to understand much about Python distributions and packaging. Pyramid isn’t really marketed as a microframework, but it allows you to do almost everything that frameworks that are marketed as micro offer in very similar ways.

from paste.httpserver import serve
from pyramid.config import Configurator
from pyramid.response import Response


def hello_world(request):
    return Response('Hello %(name)s!' % request.matchdict)

if __name__ == '__main__':
    config = Configurator()
    config.add_route('hello', '/hello/{name}')
    config.add_view(hello_world, route_name='hello')
    app = config.make_wsgi_app()
    serve(app, host='0.0.0.0')

See also Creating Your First Pyramid Application.

Decorator-based configuration

If you like the idea of framework configuration statements living next to the code it configures, so you don’t have to constantly switch between files to refer to framework configuration when adding new code, you can use Pyramid decorators to localize the configuration. For example:

from pyramid.view import view_config
from pyramid.response import Response

@view_config(route_name='fred')
def fred_view(request):
    return Response('fred')

However, unlike some other systems, using decorators for Pyramid configuration does not make your application difficult to extend, test or reuse. The view_config decorator, for example, does not actually change the input or output of the function it decorates, so testing it is a “WYSIWYG” operation; you don’t need to understand the framework to test your own code, you just behave as if the decorator is not there. You can also instruct Pyramid to ignore some decorators, or use completely imperative configuration instead of decorators to add views. Pyramid decorators are inert instead of eager: you detect and activate them with a scan.

Example: Adding View Configuration Using the @view_config Decorator.

URL generation

Pyramid is capable of generating URLs for resources, routes, and static assets. Its URL generation APIs are easy to use and flexible. If you use Pyramid’s various APIs for generating URLs, you can change your configuration around arbitrarily without fear of breaking a link on one of your web pages.

Example: Generating Route URLs.

Static file serving

Pyramid is perfectly willing to serve static files itself. It won’t make you use some external web server to do that. You can even serve more than one set of static files in a single Pyramid web application (e.g. /static and /static2). You can also, optionally, place your files on an external web server and ask Pyramid to help you generate URLs to those files, so you can use Pyramid’s internal fileserving while doing development, and a faster static file server in production without changing any code.

Example: Serving Static Assets.

Debug Toolbar

Pyramid’s debug toolbar comes activated when you use a Pyramid scaffold to render a project. This toolbar overlays your application in the browser, and allows you access to framework data such as the routes configured, the last renderings performed, the current set of packages installed, SQLAlchemy queries run, logging data, and various other facts. When an exception occurs, you can use its interactive debugger to poke around right in your browser to try to determine the cause of the exception. It’s handy.

Example: The Debug Toolbar.

Debugging settings

Pyramid has debugging settings that allow you to print Pyramid runtime information to the console when things aren’t behaving as you’re expecting. For example, you can turn on “debug_notfound”, which prints an informative message to the console every time a URL does not match any view. You can turn on “debug_authorization”, which lets you know why a view execution was allowed or denied by printing a message to the console. These features are useful for those WTF moments.

There are also a number of paster commands that allow you to introspect the configuration of your system: paster proutes shows all configured routes for an application in the order they’ll be evaluated for matching; paster pviews shows all configured views for any given URL. These are also WTF-crushers in some circumstances.

Examples: Debugging View Authorization Failures and Command-Line Pyramid.

Add-ons

Pyramid has an extensive set of add-ons held to the same quality standards as the Pyramid core itself. Add-ons are packages which provide functionality that the Pyramid core doesn’t. Add-on packages already exist which let you easily send email, let you use the Jinja2 templating system, let you use XML-RPC or JSON-RPC, let you integrate with jQuery Mobile, etc.

Examples: http://docs.pylonsproject.org/docs/pyramid.html#pyramid-add-on-documentation

Class-based and function-based views

Pyramid has a structured, unified concept of a view callable. View callables can be functions, methods of classes, or even instances. When you add a new view callable, you can choose to make it a function or a method of a class; in either case, Pyramid treats it largely the same way. You can change your mind later, and move code between methods of classes and functions. A collection of similar view callables can be attached to a single class as methods, if that floats your boat, and they can share initialization code as necessary. All kinds of views are easy to understand and use and operate similarly. There is no phony distinction between them; they can be used for the same purposes.

Here’s a view callable defined as a function:

1
2
3
4
5
6
from pyramid.response import Response
from pyramid.view import view_config

@view_config(route_name='aview')
def aview(request):
    return Response('one')

Here’s a few views defined as methods of a class instead:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
from pyramid.response import Response
from pyramid.view import view_config

class AView(object):
    def __init__(self, request):
        self.request = request

    @view_config(route_name='view_one')
    def view_one(request):
        return Response('one')

    @view_config(route_name='view_two')
    def view_two(request):
        return Response('two')

See also @view_config Placement.

Asset specifications

Asset specifications are strings that contain both a Python package name and a file or directory name, e.g. MyPackage:static/index.html. Use of these specifications is omnipresent in Pyramid. An asset specification can refer to a template, a translation directory, or any other package-bound static resource. This makes a system built on Pyramid extensible, because you don’t have to rely on globals (“the static directory”) or lookup schemes (“the ordered set of template directories”) to address your files. You can move files around as necessary, and include other packages that may not share your system’s templates or static files without encountering conflicts.

Because asset specifications are used heavily in Pyramid, we’ve also provided a way to allow users to override assets. Say you love a system that someone else has created with Pyramid but you just need to change “that one template” to make it all better. No need to fork the application. Just override the asset specification for that template with your own inside a wrapper, and you’re good to go.

Examples: Understanding Asset Specifications and Overriding Assets.

Extensible templating

Pyramid has a structured API that allows for pluggability of “renderers”. Templating systems such as Mako, Genshi, Chameleon, and Jinja2 can be treated as renderers. Renderer bindings for all of these templating systems already exist for use in Pyramid. But if you’d rather use another, it’s not a big deal. Just copy the code from an existing renderer package, and plug in your favorite templating system. You’ll then be able to use that templating system from within Pyramid just as you’d use one of the “built-in” templating systems.

Pyramid does not make you use a single templating system exclusively. You can use multiple templating systems, even in the same project.

Example: Using Templates Directly.

Rendered views can return dictionaries

If you use a renderer, you don’t have to return a special kind of “webby” Response object from a view. Instead, you can return a dictionary instead, and Pyramid will take care of converting that dictionary to a Response using a template on your behalf. This makes the view easier to test, because you don’t have to parse HTML in your tests; just make an assertion instead that the view returns “the right stuff” in the dictionary it returns. You can write “real” unit tests instead of functionally testing all of your views.

For example, instead of:

1
2
3
4
5
 from pyramid.renderers import render_to_response

 def myview(request):
     return render_to_response('myapp:templates/mytemplate.pt', {'a':1},
                               request=request)

You can do this:

1
2
3
4
5
 from pyramid.view import view_config

 @view_config(renderer='myapp:templates/mytemplate.pt')
 def myview(request):
     return {'a':1}

When this view callable is called by Pyramid, the {'a':1} dictionary will be rendered to a response on your behalf. The string passed as renderer= above is an asset specification. It is in the form packagename:directoryname/filename.ext. In this case, it names the mytemplate.pt file in the templates directory within the myapp Python package. Asset specifications are omnipresent in Pyramid: see Asset specifications for more information.

Example: Renderers.

Event system

Pyramid emits events during its request processing lifecycle. You can subscribe any number of listeners to these events. For example, to be notified of a new request, you can subscribe to the NewRequest event. To be notified that a template is about to be rendered, you can subscribe to the BeforeRender event, and so forth. Using an event publishing system as a framework notification feature instead of hardcoded hook points tends to make systems based on that framework less brittle.

You can also use Pyramid’s event system to send your own events. For example, if you’d like to create a system that is itself a framework, and may want to notify subscribers that a document has just been indexed, you can create your own event type (DocumentIndexed perhaps) and send the event via Pyramid. Users of this framework can then subscribe to your event like they’d subscribe to the events that are normally sent by Pyramid itself.

Example: Using Events and Event Types.

Built-in internationalization

Pyramid ships with internationalization-related features in its core: localization, pluralization, and creating message catalogs from source files and templates. Pyramid allows for a plurality of message catalog via the use of translation domains: you can create a system that has its own translations without conflict with other translations in other domains.

Example: Internationalization and Localization.

HTTP caching

Pyramid provides an easy way to associate views with HTTP caching policies. You can just tell Pyramid to configure your view with an http_cache statement, and it will take care of the rest:

@view_config(http_cache=3600) # 60 minutes
def myview(request): ....

Pyramid will add appropriate Cache-Control and Expires headers to responses generated when this view is invoked.

See the add_view() method’s http_cache documentation for more information.

Sessions

Pyramid has built-in HTTP sessioning. This allows you to associate data with otherwise anonymous users between requests. Lots of systems do this. But Pyramid also allows you to plug in your own sessioning system by creating some code that adheres to a documented interface. Currently there is a binding package for the third-party Beaker sessioning system that does exactly this. But if you have a specialized need (perhaps you want to store your session data in MongoDB), you can. You can even switch between implementations without changing your application code.

Example: Sessions.

Speed

The Pyramid core is, as far as we can tell, at least marginally faster than any other existing Python web framework. It has been engineered from the ground up for speed. It only does as much work as absolutely necessary when you ask it to get a job done. Extraneous function calls and suboptimal algorithms in its core codepaths are avoided. It is feasible to get, for example, between 3500 and 4000 requests per second from a simple Pyramid view on commodity dual-core laptop hardware and an appropriate WSGI server (mod_wsgi or gunicorn). In any case, performance statistics are largely useless without requirements and goals, but if you need speed, Pyramid will almost certainly never be your application’s bottleneck; at least no more than Python will be a bottleneck.

Example: http://blog.curiasolutions.com/the-great-web-framework-shootout/

Exception views

Exceptions happen. Rather than deal with exceptions that might present themselves to a user in production in an ad-hoc way, Pyramid allows you to register an exception view. Exception views are like regular Pyramid views, but they’re only invoked when an exception “bubbles up” to Pyramid itself. For example, you might register an exception view for the Exception exception, which will catch all exceptions, and present a pretty “well, this is embarrassing” page. Or you might choose to register an exception view for only specific kinds of application-specific exceptions, such as an exception that happens when a file is not found, or an exception that happens when an action cannot be performed because the user doesn’t have permission to do something. In the former case, you can show a pretty “Not Found” page; in the latter case you might show a login form.

Example: Custom Exception Views.

No singletons

Pyramid is written in such a way that it requires your application to have exactly zero “singleton” data structures. Or, put another way, Pyramid doesn’t require you to construct any “mutable globals”. Or put even a different way, an import of a Pyramid application needn’t have any “import-time side effects”. This is esoteric-sounding, but if you’ve ever tried to cope with parameterizing a Django “settings.py” file for multiple installations of the same application, or if you’ve ever needed to monkey-patch some framework fixture so that it behaves properly for your use case, or if you’ve ever wanted to deploy your system using an asynchronous server, you’ll end up appreciating this feature. It just won’t be a problem. You can even run multiple copies of a similar but not identically configured Pyramid application within the same Python process. This is good for shared hosting environments, where RAM is at a premium.

View predicates and many views per route

Unlike many other systems, Pyramid allows you to associate more than one view per route. For example, you can create a route with the pattern /items and when the route is matched, you can shuffle off the request to one view if the request method is GET, another view if the request method is POST, etc. A system known as “view predicates” allows for this. Request method matching is the very most basic thing you can do with a view predicate. You can also associate views with other request parameters such as the elements in the query string, the Accept header, whether the request is an XHR request or not, and lots of other things. This feature allows you to keep your individual views “clean”; they won’t need much conditional logic, so they’ll be easier to test.

Example: View Configuration Parameters.

Transaction management

Pyramid’s scaffold system renders projects that include a transaction management system, stolen from Zope. When you use this transaction management system, you cease being responsible for committing your data anymore. Instead, Pyramid takes care of committing: it commits at the end of a request or aborts if there’s an exception. Why is that a good thing? Having a centralized place for transaction management is a great thing. If, instead of managing your transactions in a centralized place, you sprinkle session.commit calls in your application logic itself, you can wind up in a bad place. Wherever you manually commit data to your database, it’s likely that some of your other code is going to run after your commit. If that code goes on to do other important things after that commit, and an error happens in the later code, you can easily wind up with inconsistent data if you’re not extremely careful. Some data will have been written to the database that probably should not have. Having a centralized commit point saves you from needing to think about this; it’s great for lazy people who also care about data integrity. Either the request completes successfully, and all changes are committed, or it does not, and all changes are aborted.

Also, Pyramid’s transaction management system allows you to synchronize commits between multiple databases, and allows you to do things like conditionally send email if a transaction commits, but otherwise keep quiet.

Example: SQLAlchemy + URL Dispatch Wiki Tutorial (note the lack of commit statements anywhere in application code).

Configuration conflict detection

When a system is small, it’s reasonably easy to keep it all in your head. But when systems grow large, you may have hundreds or thousands of configuration statements which add a view, add a route, and so forth. Pyramid’s configuration system keeps track of your configuration statements, and if you accidentally add two that are identical, or Pyramid can’t make sense out of what it would mean to have both statements active at the same time, it will complain loudly at startup time. It’s not dumb though: it will automatically resolve conflicting configuration statements on its own if you use the configuration include() system: “more local” statements are preferred over “less local” ones. This allows you to intelligently factor large systems into smaller ones.

Example: Conflict Detection.

Configuration extensibility

Unlike other systems, Pyramid provides a structured “include” mechanism (see include()) that allows you to compose applications from multiple Python packages. All the configuration statements that can be performed in your “main” Pyramid application can also be performed by included packages including the addition of views, routes, subscribers, and even authentication and authorization policies. You can even extend or override an existing application by including another application’s configuration in your own, overriding or adding new views and routes to it. This has the potential to allow you to compose a big application out of many other smaller ones. For example, if you want to reuse an existing application that already has a bunch of routes, you can just use the include statement with a route_prefix; the new application will live within your application at a URL prefix. It’s not a big deal, and requires little up-front engineering effort.

For example:

1
2
3
4
5
6
7
from pyramid.config import Configurator

if __name__ == '__main__':
   config = Configurator()
   config.include('pyramid_jinja2')
   config.include('pyramid_exclog')
   config.include('some.other.guys.package', route_prefix='/someotherguy')

See also Including Configuration from External Sources and Rules for Building An Extensible Application

Flexible authentication and authorization

Pyramid includes a flexible, pluggable authentication and authorization system. No matter where your user data is stored, or what scheme you’d like to use to permit your users to access your data, you can use a predefined Pyramid plugpoint to plug in your custom authentication and authorization code. If you want to change these schemes later, you can just change it in one place rather than everywhere in your code. It also ships with prebuilt well-tested authentication and authorization schemes out of the box. But what if you don’t want to use Pyramid’s built-in system? You don’t have to. You can just write your own bespoke security code as you would in any other system.

Example: Enabling an Authorization Policy.

Traversal

Traversal is a concept stolen from Zope. It allows you to create a tree of resources, each of which can be addressed by one or more URLs. Each of those resources can have one or more views associated with it. If your data isn’t naturally treelike (or you’re unwilling to create a treelike representation of your data), you aren’t going to find traversal very useful. However, traversal is absolutely fantastic for sites that need to be arbitrarily extensible: it’s a lot easier to add a node to a tree than it is to shoehorn a route into an ordered list of other routes, or to create another entire instance of an application to service a department and glue code to allow disparate apps to share data. It’s a great fit for sites that naturally lend themselves to changing departmental hierarchies, such as content management systems and document management systems. Traversal also lends itself well to systems that require very granular security (“Bob can edit this document” as opposed to “Bob can edit documents”).

Example: Much Ado About Traversal.

Tweens

Pyramid has a sort of internal WSGI-middleware-ish pipeline that can be hooked by arbitrary add-ons named “tweens”. The debug toolbar is a “tween”, and the pyramid_tm transaction manager is also. Tweens are more useful than WSGI middleware in some circumstances because they run in the context of Pyramid itself, meaning you have access to templates and other renderers, a “real” request object, and other niceties.

Example: Registering “Tweens”.

View response adapters

A lot is made of the aesthetics of what kinds of objects you’re allowed to return from view callables in various frameworks. In a previous section in this document we showed you that, if you use a renderer, you can usually return a dictionary from a view callable instead of a full-on Response object. But some frameworks allow you to return strings or tuples from view callables. When frameworks allow for this, code looks slightly prettier, because fewer imports need to be done, and there is less code. For example, compare this:

1
2
def aview(request):
    return "Hello world!"

To this:

1
2
3
4
from pyramid.response import Response

def aview(request):
    return Response("Hello world!")

The former is “prettier”, right?

Out of the box, if you define the former view callable (the one that simply returns a string) in Pyramid, when it is executed, Pyramid will raise an exception. This is because “explicit is better than implicit”, in most cases, and by default, Pyramid wants you to return a Response object from a view callable. This is because there’s usually a heck of a lot more to a response object than just its body. But if you’re the kind of person who values such aesthetics, we have an easy way to allow for this sort of thing:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
from pyramid.config import Configurator
from pyramid.response import Response

def string_response_adapter(s):
    response = Response(s)
    response.content_type = 'text/html'
    return response

if __name__ == '__main__':
    config = Configurator()
    config.add_response_adapter(string_response_adapter, basestring)

Do that once in your Pyramid application at startup. Now you can return strings from any of your view callables, e.g.:

1
2
3
4
5
def helloview(request):
    return "Hello world!"

def goodbyeview(request):
    return "Goodbye world!"

Oh noes! What if you want to indicate a custom content type? And a custom status code? No fear:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
from pyramid.config import Configurator

def tuple_response_adapter(val):
    status_int, content_type, body = val
    response = Response(body)
    response.content_type = content_type
    response.status_int = status_int
    return response

def string_response_adapter(body):
    response = Response(body)
    response.content_type = 'text/html'
    response.status_int = 200
    return response

if __name__ == '__main__':
    config = Configurator()
    config.add_response_adapter(string_response_adapter, basestring)
    config.add_response_adapter(tuple_response_adapter, tuple)

Once this is done, both of these view callables will work:

1
2
3
4
5
def aview(request):
    return "Hello world!"

def anotherview(request):
    return (403, 'text/plain', "Forbidden")

Pyramid defaults to explicit behavior, because it’s the most generally useful, but provides hooks that allow you to adapt the framework to localized aesthetic desires.

See also Changing How Pyramid Treats View Responses.

“Global” response object

“Constructing these response objects in my view callables is such a chore! And I’m way too lazy to register a response adapter, as per the prior section,” you say. Fine. Be that way:

1
2
3
4
5
def aview(request):
    response = request.response
    response.body = 'Hello world!'
    response.content_type = 'text/plain'
    return response

See also Varying Attributes of Rendered Responses.

Automating repetitive configuration

Does Pyramid’s configurator allow you to do something, but you’re a little adventurous and just want it a little less verbose? Or you’d like to offer up some handy configuration feature to other Pyramid users without requiring that we change Pyramid? You can extend Pyramid’s Configurator with your own directives. For example, let’s say you find yourself calling pyramid.config.Configurator.add_view() repetitively. Usually you can take the boring away by using existing shortcuts, but let’s say that this is a case such a way that no existing shortcut works to take the boring away:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
from pyramid.config import Configurator

config = Configurator()
config.add_route('xhr_route', '/xhr/{id}')
config.add_view('my.package.GET_view', route_name='xhr_route',
                xhr=True,  permission='view', request_method='GET')
config.add_view('my.package.POST_view', route_name='xhr_route',
                xhr=True, permission='view', request_method='POST')
config.add_view('my.package.HEAD_view', route_name='xhr_route',
                xhr=True, permission='view', request_method='HEAD')

Pretty tedious right? You can add a directive to the Pyramid configurator to automate some of the tedium away:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
from pyramid.config import Configurator

def add_protected_xhr_views(config, module):
    module = config.maybe_dotted(module)
    for method in ('GET', 'POST', 'HEAD'):
        view = getattr(module, 'xhr_%s_view' % method, None)
        if view is not None:
            config.add_view(view, route_name='xhr_route', xhr=True,
                           permission='view', request_method=method)

config = Configurator()
config.add_directive('add_protected_xhr_views', add_protected_xhr_views)

Once that’s done, you can call the directive you’ve just added as a method of the Configurator object:

1
2
config.add_route('xhr_route', '/xhr/{id}')
config.add_protected_xhr_views('my.package')

Your previously repetitive configuration lines have now morphed into one line.

You can share your configuration code with others this way too by packaging it up and calling add_directive() from within a function called when another user uses the include() method against your code.

See also Adding Methods to the Configurator via add_directive.

Testing

Every release of Pyramid has 100% statement coverage via unit and integration tests, as measured by the coverage tool available on PyPI. It also has greater than 95% decision/condition coverage as measured by the instrumental tool available on PyPI. It is automatically tested by the Jenkins tool on Python 2.5, Python 2.6, Python 2.7, Jython and PyPy after each commit to its GitHub repository. Official Pyramid add-ons are held to a similar testing standard. We still find bugs in Pyramid and its official add-ons, but we’ve noticed we find a lot more of them while working on other projects that don’t have a good testing regime.

Example: http://jenkins.pylonsproject.org/

Support

It’s our goal that no Pyramid question go unanswered. Whether you ask a question on IRC, on the Pylons-discuss maillist, or on StackOverflow, you’re likely to get a reasonably prompt response. We don’t tolerate “support trolls” or other people who seem to get their rocks off by berating fellow users in our various offical support channels. We try to keep it well-lit and new-user-friendly.

Example: Visit irc://freenode.net#pyramid (the #pyramid channel on irc.freenode.net in an IRC client) or the pylons-discuss maillist at http://groups.google.com/group/pylons-discuss/ .

Documentation

It’s a constant struggle, but we try to maintain a balance between completeness and new-user-friendliness in the official narrative Pyramid documentation (concrete suggestions for improvement are always appreciated, by the way). We also maintain a “cookbook” of recipes, which are usually demonstrations of common integration scenarios, too specific to add to the official narrative docs. In any case, the Pyramid documentation is comprehensive.

Example: The rest of this documentation and the cookbook at http://docs.pylonsproject.org/projects/pyramid_cookbook/dev/ .

What Is The Pylons Project?

Pyramid is a member of the collection of software published under the Pylons Project. Pylons software is written by a loose-knit community of contributors. The Pylons Project website includes details about how Pyramid relates to the Pylons Project.

Pyramid and Other Web Frameworks

The first release of Pyramid’s predecessor (named repoze.bfg) was made in July of 2008. At the end of 2010, we changed the name of repoze.bfg to Pyramid. It was merged into the Pylons project as Pyramid in November of that year.

Pyramid was inspired by Zope, Pylons (version 1.0) and Django. As a result, Pyramid borrows several concepts and features from each, combining them into a unique web framework.

Many features of Pyramid trace their origins back to Zope. Like Zope applications, Pyramid applications can be easily extended: if you obey certain constraints, the application you produce can be reused, modified, re-integrated, or extended by third-party developers without forking the original application. The concepts of traversal and declarative security in Pyramid were pioneered first in Zope.

The Pyramid concept of URL dispatch is inspired by the Routes system used by Pylons version 1.0. Like Pylons version 1.0, Pyramid is mostly policy-free. It makes no assertions about which database you should use, and its built-in templating facilities are included only for convenience. In essence, it only supplies a mechanism to map URLs to view code, along with a set of conventions for calling those views. You are free to use third-party components that fit your needs in your applications.

The concept of view is used by Pyramid mostly as it would be by Django. Pyramid has a documentation culture more like Django’s than like Zope’s.

Like Pylons version 1.0, but unlike Zope, a Pyramid application developer may use completely imperative code to perform common framework configuration tasks such as adding a view or a route. In Zope, ZCML is typically required for similar purposes. In Grok, a Zope-based web framework, decorator objects and class-level declarations are used for this purpose. Out of the box, Pyramid supports imperative and decorator-based configuration; ZCML may be used via an add-on package named pyramid_zcml.

Also unlike Zope and unlike other “full-stack” frameworks such as Django, Pyramid makes no assumptions about which persistence mechanisms you should use to build an application. Zope applications are typically reliant on ZODB; Pyramid allows you to build ZODB applications, but it has no reliance on the ZODB software. Likewise, Django tends to assume that you want to store your application’s data in a relational database. Pyramid makes no such assumption; it allows you to use a relational database but doesn’t encourage or discourage the decision.

Other Python web frameworks advertise themselves as members of a class of web frameworks named model-view-controller frameworks. Insofar as this term has been claimed to represent a class of web frameworks, Pyramid also generally fits into this class.

You Say Pyramid is MVC, But Where’s The Controller?

The Pyramid authors believe that the MVC pattern just doesn’t really fit the web very well. In a Pyramid application, there is a resource tree, which represents the site structure, and views, which tend to present the data stored in the resource tree and a user-defined “domain model”. However, no facility provided by the framework actually necessarily maps to the concept of a “controller” or “model”. So if you had to give it some acronym, I guess you’d say Pyramid is actually an “RV” framework rather than an “MVC” framework. “MVC”, however, is close enough as a general classification moniker for purposes of comparison with other web frameworks.

Installing Pyramid

Before You Install

You will need Python version 2.5 or better to run Pyramid.

Python Versions

As of this writing, Pyramid has been tested under Python 2.5.5, Python 2.6.6, and Python 2.7.2. Pyramid does not run under any version of Python before 2.5, and does not yet run under Python 3.X.

Pyramid is known to run on all popular UNIX-like systems such as Linux, MacOS X, and FreeBSD as well as on Windows platforms. It is also known to run on Google’s App Engine, PyPy (1.5 and 1.6), and Jython (2.5.2).

Pyramid installation does not require the compilation of any C code, so you need only a Python interpreter that meets the requirements mentioned.

If You Don’t Yet Have A Python Interpreter (UNIX)

If your system doesn’t have a Python interpreter, and you’re on UNIX, you can either install Python using your operating system’s package manager or you can install Python from source fairly easily on any UNIX system that has development tools.

Package Manager Method

You can use your system’s “package manager” to install Python. Every system’s package manager is slightly different, but the “flavor” of them is usually the same.

For example, on an Ubuntu Linux system, to use the system package manager to install a Python 2.6 interpreter, use the following command:

$ sudo apt-get install python2.6-dev

Once these steps are performed, the Python interpreter will usually be invokable via python2.6 from a shell prompt.

Source Compile Method

It’s useful to use a Python interpreter that isn’t the “system” Python interpreter to develop your software. The authors of Pyramid tend not to use the system Python for development purposes; always a self-compiled one. Compiling Python is usually easy, and often the “system” Python is compiled with options that aren’t optimal for web development.

To compile software on your UNIX system, typically you need development tools. Often these can be installed via the package manager. For example, this works to do so on an Ubuntu Linux system:

$ sudo apt-get install build-essential

On Mac OS X, installing XCode has much the same effect.

Once you’ve got development tools installed on your system, you can install a Python 2.6 interpreter from source, on the same system, using the following commands:

[chrism@vitaminf ~]$ cd ~
[chrism@vitaminf ~]$ mkdir tmp
[chrism@vitaminf ~]$ mkdir opt
[chrism@vitaminf ~]$ cd tmp
[chrism@vitaminf tmp]$ wget \
       http://www.python.org/ftp/python/2.6.4/Python-2.6.4.tgz
[chrism@vitaminf tmp]$ tar xvzf Python-2.6.4.tgz
[chrism@vitaminf tmp]$ cd Python-2.6.4
[chrism@vitaminf Python-2.6.4]$ ./configure \
        --prefix=$HOME/opt/Python-2.6.4
[chrism@vitaminf Python-2.6.4]$ make; make install

Once these steps are performed, the Python interpreter will be invokable via $HOME/opt/Python-2.6.4/bin/python from a shell prompt.

If You Don’t Yet Have A Python Interpreter (Windows)

If your Windows system doesn’t have a Python interpreter, you’ll need to install it by downloading a Python 2.6-series interpreter executable from python.org’s download section (the files labeled “Windows Installer”). Once you’ve downloaded it, double click on the executable and accept the defaults during the installation process. You may also need to download and install the Python for Windows extensions.

Warning

After you install Python on Windows, you may need to add the C:\Python26 directory to your environment’s Path in order to make it possible to invoke Python from a command prompt by typing python. To do so, right click My Computer, select Properties –> Advanced Tab –> Environment Variables and add that directory to the end of the Path environment variable.

Installing Pyramid on a UNIX System

It is best practice to install Pyramid into a “virtual” Python environment in order to obtain isolation from any “system” packages you’ve got installed in your Python version. This can be done by using the virtualenv package. Using a virtualenv will also prevent Pyramid from globally installing versions of packages that are not compatible with your system Python.

To set up a virtualenv in which to install Pyramid, first ensure that setuptools is installed. Invoke import setuptools within the Python interpreter you’d like to run Pyramid under:

[chrism@vitaminf pyramid]$ python
Python 2.6.5 (r265:79063, Apr 29 2010, 00:31:32)
[GCC 4.4.3] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> import setuptools

If running import setuptools does not raise an ImportError, it means that setuptools is already installed into your Python interpreter. If import setuptools fails, you will need to install setuptools manually. Note that above we’re using a Python 2.6-series interpreter on Mac OS X; your output may differ if you’re using a later Python version or a different platform.

If you are using a “system” Python (one installed by your OS distributor or a 3rd-party packager such as Fink or MacPorts), you can usually install the setuptools package by using your system’s package manager. If you cannot do this, or if you’re using a self-installed version of Python, you will need to install setuptools “by hand”. Installing setuptools “by hand” is always a reasonable thing to do, even if your package manager already has a pre-chewed version of setuptools for installation.

To install setuptools by hand, first download ez_setup.py then invoke it using the Python interpreter into which you want to install setuptools.

$ python ez_setup.py

Once this command is invoked, setuptools should be installed on your system. If the command fails due to permission errors, you may need to be the administrative user on your system to successfully invoke the script. To remediate this, you may need to do:

$ sudo python ez_setup.py
Installing the virtualenv Package

Once you’ve got setuptools installed, you should install the virtualenv package. To install the virtualenv package into your setuptools-enabled Python interpreter, use the easy_install command.

$ easy_install virtualenv

This command should succeed, and tell you that the virtualenv package is now installed. If it fails due to permission errors, you may need to install it as your system’s administrative user. For example:

$ sudo easy_install virtualenv
Creating the Virtual Python Environment

Once the virtualenv package is installed in your Python, you can then create a virtual environment. To do so, invoke the following:

$ virtualenv --no-site-packages env
New python executable in env/bin/python
Installing setuptools.............done.

Warning

Using --no-site-packages when generating your virtualenv is very important. This flag provides the necessary isolation for running the set of packages required by Pyramid. If you do not specify --no-site-packages, it’s possible that Pyramid will not install properly into the virtualenv, or, even if it does, may not run properly, depending on the packages you’ve already got installed into your Python’s “main” site-packages dir.

Warning

do not use sudo to run the virtualenv script. It’s perfectly acceptable (and desirable) to create a virtualenv as a normal user.

You should perform any following commands that mention a “bin” directory from within the env virtualenv dir.

Installing Pyramid Into the Virtual Python Environment

After you’ve got your env virtualenv installed, you may install Pyramid itself using the following commands from within the virtualenv (env) directory you created in the last step.

$ cd env

.. parsed-literal::

$ bin/easy_install "pyramid==\ |release|\ "

The easy_install command will take longer than the previous ones to complete, as it downloads and installs a number of dependencies.

Installing Pyramid on a Windows System

  1. Install, or find Python 2.6 for your system.

  2. Install the Python for Windows extensions. Make sure to pick the right download for Python 2.6 and install it using the same Python installation from the previous step.

  3. Install latest setuptools distribution into the Python you obtained/installed/found in the step above: download ez_setup.py and run it using the python interpreter of your Python 2.6 installation using a command prompt:

    c:\> c:\Python26\python ez_setup.py

  4. Use that Python’s bin/easy_install to install virtualenv:

    c:\> c:\Python26\Scripts\easy_install virtualenv

  5. Use that Python’s virtualenv to make a workspace:

    c:\> c:\Python26\Scripts\virtualenv --no-site-packages env

  6. Switch to the env directory:

    c:\> cd env

  7. (Optional) Consider using Scripts\activate.bat to make your shell environment wired to use the virtualenv.

  8. Use easy_install pointed at the “current” index to get Pyramid and its direct dependencies installed:

    c:\env> Scripts\easy_install "pyramid==1.2.7"

Installing Pyramid on Google App Engine

Running Pyramid on Google’s App Engine documents the steps required to install a Pyramid application on Google App Engine.

Installing Pyramid on Jython

Pyramid is known to work under Jython version 2.5.1. Install Jython, and then follow the installation steps for Pyramid on your platform described in one of the sections entitled Installing Pyramid on a UNIX System or Installing Pyramid on a Windows System above, replacing the python command with jython as necessary. The steps are exactly the same except you should use the jython command name instead of the python command name.

One caveat exists to using Pyramid under Jython: the Chameleon templating engine does not work on Jython. However, the Mako templating system, which is also included with Pyramid, does work under Jython; use it instead.

What Gets Installed

When you easy_install Pyramid, various Zope libraries, various Chameleon libraries, WebOb, Paste, PasteScript, and PasteDeploy libraries are installed.

Additionally, as chronicled in Creating a Pyramid Project, scaffolds will be registered, which make it easy to start a new Pyramid project.

Creating Your First Pyramid Application

In this chapter, we will walk through the creation of a tiny Pyramid application. After we’re finished creating the application, we’ll explain in more detail how it works.

Hello World

Here’s one of the very simplest Pyramid applications:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
from paste.httpserver import serve
from pyramid.config import Configurator
from pyramid.response import Response


def hello_world(request):
    return Response('Hello %(name)s!' % request.matchdict)

if __name__ == '__main__':
    config = Configurator()
    config.add_route('hello', '/hello/{name}')
    config.add_view(hello_world, route_name='hello')
    app = config.make_wsgi_app()
    serve(app, host='0.0.0.0')

When this code is inserted into a Python script named helloworld.py and executed by a Python interpreter which has the Pyramid software installed, an HTTP server is started on TCP port 8080:

$ python helloworld.py
serving on 0.0.0.0:8080 view at http://127.0.0.1:8080

When port 8080 is visited by a browser on the URL /hello/world, the server will simply serve up the text “Hello world!”

Press Ctrl-C to stop the application.

Now that we have a rudimentary understanding of what the application does, let’s examine it piece-by-piece.

Imports

The above helloworld.py script uses the following set of import statements:

1
2
3
from paste.httpserver import serve
from pyramid.config import Configurator
from pyramid.response import Response

The script imports the Configurator class from the pyramid.config module. An instance of the Configurator class is later used to configure your Pyramid application.

Like many other Python web frameworks, Pyramid uses the WSGI protocol to connect an application and a web server together. The paste.httpserver server is used in this example as a WSGI server for convenience, as the paste package is a dependency of Pyramid itself.

The script also imports the pyramid.response.Response class for later use. An instance of this class will be used to create a web response.

View Callable Declarations

The above script, beneath its set of imports, defines a function named hello_world.

1
2
def hello_world(request):
    return Response('Hello %(name)s!' % request.matchdict)

This function doesn’t do anything very difficult. The functions accepts a single argument (request). The hello_world function returns an instance of the pyramid.response.Response. The single argument to the class’ constructor is value computed from arguments matched from the url route. This value becomes the body of the response.

This function is known as a view callable. A view callable accepts a single argument, request. It is expected to return a response object. A view callable doesn’t need to be a function; it can be represented via another type of object, like a class or an instance, but for our purposes here, a function serves us well.

A view callable is always called with a request object. A request object is a representation of an HTTP request sent to Pyramid via the active WSGI server.

A view callable is required to return a response object because a response object has all the information necessary to formulate an actual HTTP response; this object is then converted to text by the WSGI server which called Pyramid and it is sent back to the requesting browser. To return a response, each view callable creates an instance of the Response class. In the hello_world function, a string is passed as the body to the response.

Application Configuration

In the above script, the following code represents the configuration of this simple application. The application is configured using the previously defined imports and function definitions, placed within the confines of an if statement:

1
2
3
4
5
6
if __name__ == '__main__':
    config = Configurator()
    config.add_route('hello', '/hello/{name}')
    config.add_view(hello_world, route_name='hello')
    app = config.make_wsgi_app()
    serve(app, host='0.0.0.0')

Let’s break this down piece-by-piece.

Configurator Construction
1
2
if __name__ == '__main__':
    config = Configurator()

The if __name__ == '__main__': line in the code sample above represents a Python idiom: the code inside this if clause is not invoked unless the script containing this code is run directly from the operating system command line. For example, if the file named helloworld.py contains the entire script body, the code within the if statement will only be invoked when python helloworld.py is executed from the command line.

Using the if clause is necessary – or at least best practice – because code in a Python .py file may be eventually imported via the Python import statement by another .py file. .py files that are imported by other .py files are referred to as modules. By using the if __name__ == 'main': idiom, the script above is indicating that it does not want the code within the if statement to execute if this module is imported from another; the code within the if block should only be run during a direct script execution.

The config = Configurator() line above creates an instance of the Configurator class. The resulting config object represents an API which the script uses to configure this particular Pyramid application. Methods called on the Configurator will cause registrations to be made in an application registry associated with the application.

Adding Configuration
1
2
    config.add_route('hello', '/hello/{name}')
    config.add_view(hello_world, route_name='hello')

First line above calls the pyramid.config.Configurator.add_route() method, which registers a route to match any url path that begins with /hello/ followed by a string.

The second line, config.add_view(hello_world, route_name='hello'), registers the hello_world function as a view callable and makes sure that it will be called when the hello route is matched.

WSGI Application Creation
1
    app = config.make_wsgi_app()

After configuring views and ending configuration, the script creates a WSGI application via the pyramid.config.Configurator.make_wsgi_app() method. A call to make_wsgi_app implies that all configuration is finished (meaning all method calls to the configurator which set up views, and various other configuration settings have been performed). The make_wsgi_app method returns a WSGI application object that can be used by any WSGI server to present an application to a requestor. WSGI is a protocol that allows servers to talk to Python applications. We don’t discuss WSGI in any depth within this book, however, you can learn more about it by visiting wsgi.org.

The Pyramid application object, in particular, is an instance of a class representing a Pyramid router. It has a reference to the application registry which resulted from method calls to the configurator used to configure it. The router consults the registry to obey the policy choices made by a single application. These policy choices were informed by method calls to the Configurator made earlier; in our case, the only policy choices made were implied by calls to its add_view and add_route methods.

WSGI Application Serving
1
    serve(app, host='0.0.0.0')

Finally, we actually serve the application to requestors by starting up a WSGI server. We happen to use the paste.httpserver.serve() WSGI server runner, passing it the app object (a router) as the application we wish to serve. We also pass in an argument host=='0.0.0.0', meaning “listen on all TCP interfaces.” By default, the Paste HTTP server listens only on the 127.0.0.1 interface, which is problematic if you’re running the server on a remote system and you wish to access it with a web browser from a local system. We don’t specify a TCP port number to listen on; this means we want to use the default TCP port, which is 8080.

When this line is invoked, it causes the server to start listening on TCP port 8080. The server will serve requests forever, or at least until we stop it by killing the process which runs it (usually by pressing Ctrl-C in the terminal we used to start it).

Conclusion

Our hello world application is one of the simplest possible Pyramid applications, configured “imperatively”. We can see that it’s configured imperatively because the full power of Python is available to us as we perform configuration tasks.

References

For more information about the API of a Configurator object, see Configurator .

For more information about view configuration, see View Configuration.

Application Configuration

Most people already understand “configuration” as settings that influence the operation of an application. For instance, it’s easy to think of the values in a .ini file parsed at application startup time as “configuration”. However, if you’re reasonably open-minded, it’s easy to think of code as configuration too. Since Pyramid, like most other web application platforms, is a framework, it calls into code that you write (as opposed to a library, which is code that exists purely for you to call). The act of plugging application code that you’ve written into Pyramid is also referred to within this documentation as “configuration”; you are configuring Pyramid to call the code that makes up your application.

There are two ways to configure a Pyramid application: imperative configuration and declarative configuration. Both are described below.

Imperative Configuration

“Imperative configuration” just means configuration done by Python statements, one after the next. Here’s one of the simplest Pyramid applications, configured imperatively:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
from paste.httpserver import serve
from pyramid.config import Configurator
from pyramid.response import Response

def hello_world(request):
    return Response('Hello world!')

if __name__ == '__main__':
    config = Configurator()
    config.add_view(hello_world)
    app = config.make_wsgi_app()
    serve(app, host='0.0.0.0')

We won’t talk much about what this application does yet. Just note that the “configuration’ statements take place underneath the if __name__ == '__main__': stanza in the form of method calls on a Configurator object (e.g. config.add_view(...)). These statements take place one after the other, and are executed in order, so the full power of Python, including conditionals, can be employed in this mode of configuration.

Declarative Configuration

It’s sometimes painful to have all configuration done by imperative code, because often the code for a single application may live in many files. If the configuration is centralized in one place, you’ll need to have at least two files open at once to see the “big picture”: the file that represents the configuration, and the file that contains the implementation objects referenced by the configuration. To avoid this, Pyramid allows you to insert configuration decoration statements very close to code that is referred to by the declaration itself. For example:

1
2
3
4
5
6
from pyramid.response import Response
from pyramid.view import view_config

@view_config(name='hello', request_method='GET')
def hello(request):
    return Response('Hello')

The mere existence of configuration decoration doesn’t cause any configuration registration to be performed. Before it has any effect on the configuration of a Pyramid application, a configuration decoration within application code must be found through a process known as a scan.

For example, the pyramid.view.view_config decorator in the code example above adds an attribute to the hello function, making it available for a scan to find it later.

A scan of a module or a package and its subpackages for decorations happens when the pyramid.config.Configurator.scan() method is invoked: scanning implies searching for configuration declarations in a package and its subpackages. For example:

Starting A Scan

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
from paste.httpserver import serve
from pyramid.response import Response
from pyramid.view import view_config

@view_config()
def hello(request):
    return Response('Hello')

if __name__ == '__main__':
    from pyramid.config import Configurator
    config = Configurator()
    config.scan()
    app = config.make_wsgi_app()
    serve(app, host='0.0.0.0')

The scanning machinery imports each module and subpackage in a package or module recursively, looking for special attributes attached to objects defined within a module. These special attributes are typically attached to code via the use of a decorator. For example, the view_config decorator can be attached to a function or instance method.

Once scanning is invoked, and configuration decoration is found by the scanner, a set of calls are made to a Configurator on your behalf: these calls replace the need to add imperative configuration statements that don’t live near the code being configured.

The combination of configuration decoration and the invocation of a scan is collectively known as declarative configuration.

In the example above, the scanner translates the arguments to view_config into a call to the pyramid.config.Configurator.add_view() method, effectively:

1
config.add_view(hello)

Summary

There are two ways to configure a Pyramid application: declaratively and imperatively. You can choose the mode you’re most comfortable with; both are completely equivalent. Examples in this documentation will use both modes interchangeably.

Creating a Pyramid Project

As we saw in Creating Your First Pyramid Application, it’s possible to create a Pyramid application completely manually. However, it’s usually more convenient to use a scaffold to generate a basic Pyramid project.

A project is a directory that contains at least one Python package. You’ll use a scaffold to create a project, and you’ll create your application logic within a package that lives inside the project. Even if your application is extremely simple, it is useful to place code that drives the application within a package, because: 1) a package is more easily extended with new code and 2) an application that lives inside a package can also be distributed more easily than one which does not live within a package.

Pyramid comes with a variety of scaffolds that you can use to generate a project. Each scaffold makes different configuration assumptions about what type of application you’re trying to construct.

These scaffolds are rendered using the PasteDeploy paster create command.

Scaffolds Included with Pyramid

The convenience scaffolds included with Pyramid differ from each other on a number of axes:

  • the persistence mechanism they offer (no persistence mechanism, ZODB, or SQLAlchemy).
  • the mechanism they use to map URLs to code (traversal or URL dispatch).
  • whether or not the pyramid_beaker library is relied upon as the sessioning implementation (as opposed to no sessioning or default sessioning).

The included scaffolds are these:

pyramid_starter
URL mapping via traversal and no persistence mechanism.
pyramid_zodb
URL mapping via traversal and persistence via ZODB.
pyramid_routesalchemy
URL mapping via URL dispatch and persistence via SQLAlchemy
pyramid_alchemy
URL mapping via traversal and persistence via SQLAlchemy

Note

At this time, each of these scaffolds uses the Chameleon templating system, which is incompatible with Jython. To use scaffolds to build applications which will run on Jython, you can try the pyramid_jinja2_starter scaffold which ships as part of the pyramid_jinja2 package. You can also just use any above scaffold and replace the Chameleon template it includes with a Mako analogue.

Rather than use any of the above scaffolds, Pylons 1 users may feel more comfortable installing the Akhet development environment, which provides a scaffold named akhet. This scaffold configures a Pyramid application in a “Pylons-esque” way, including the use of a view handler to map URLs to code (a handler is much like a Pylons “controller”).

Creating the Project

In Installing Pyramid, you created a virtual Python environment via the virtualenv command. To start a Pyramid project, use the paster facility installed within the virtualenv. In Installing Pyramid we called the virtualenv directory env; the following command assumes that our current working directory is that directory. We’ll choose the pyramid_starter scaffold for this purpose.

On UNIX:

$ bin/paster create -t pyramid_starter

Or on Windows:

$ Scripts\paster.exe create -t pyramid_starter

The above command uses the paster create command to create a project with the pyramid_starter scaffold. To use a different scaffold, such as pyramid_routesalchemy, you’d just change the last argument. For example, on UNIX:

$ bin/paster create -t pyramid_routesalchemy

Or on Windows:

$ Scripts\paster.exe create -t pyramid_routesalchemy

paster create will ask you a single question: the name of the project. You should use a string without spaces and with only letters in it. Here’s sample output from a run of paster create on UNIX for a project we name MyProject:

$ bin/paster create -t pyramid_starter
Selected and implied templates:
  pyramid#pyramid_starter  pyramid starter project

Enter project name: MyProject
Variables:
  egg:      MyProject
  package:  myproject
  project:  MyProject
Creating template pyramid
Creating directory ./MyProject
# ... more output ...
Running /Users/chrism/projects/pyramid/bin/python setup.py egg_info

Note

You can skip the interrogative question about a project name during paster create by adding the project name to the command line, e.g. paster create -t pyramid_starter MyProject.

Note

You may encounter an error when using paster create if a dependent Python package is not installed. This will result in a traceback ending in pkg_resources.DistributionNotFound: <package name>. Simply run bin/easy_install (or Script\easy_install.exe on Windows), with the missing package name from the error message to work around this issue.

As a result of invoking the paster create command, a project is created in a directory named MyProject. That directory is a project directory. The setup.py file in that directory can be used to distribute your application, or install your application for deployment or development.

A PasteDeploy .ini file named development.ini will be created in the project directory. You will use this .ini file to configure a server, to run your application, and to debug your application. It sports configuration that enables an interactive debugger and settings optimized for development.

Another PasteDeploy .ini file named production.ini will also be created in the project directory. It sports configuration that disables any interactive debugger (to prevent inappropriate access and disclosure), and turns off a number of debugging settings. You can use this file to put your application into production.

The MyProject project directory contains an additional subdirectory named myproject (note the case difference) representing a Python package which holds very simple Pyramid sample code. This is where you’ll edit your application’s Python code and templates.

Installing your Newly Created Project for Development

To install a newly created project for development, you should cd to the newly created project directory and use the Python interpreter from the virtualenv you created during Installing Pyramid to invoke the command python setup.py develop

The file named setup.py will be in the root of the paster-generated project directory. The python you’re invoking should be the one that lives in the bin (or Scripts on Windows) directory of your virtual Python environment. Your terminal’s current working directory must be the newly created project directory.

On UNIX:

$ cd MyProject
$ ../bin/python setup.py develop

Or on Windows:

$ cd MyProject
$ ..\Scripts\python.exe setup.py develop

Elided output from a run of this command on UNIX is shown below:

$ cd MyProject
$ ../bin/python setup.py develop
...
Finished processing dependencies for MyProject==0.0

This will install a distribution representing your project into the interpreter’s library set so it can be found by import statements and by PasteDeploy commands such as paster serve, paster pshell, paster proutes and paster pviews.

Running The Tests For Your Application

To run unit tests for your application, you should invoke them using the Python interpreter from the virtualenv you created during Installing Pyramid (the python command that lives in the bin directory of your virtualenv).

On UNIX:

$ ../bin/python setup.py test -q

Or on Windows:

$ ..\Scripts\python.exe setup.py test -q

Here’s sample output from a test run on UNIX:

$ ../bin/python setup.py test -q
running test
running egg_info
writing requirements to MyProject.egg-info/requires.txt
writing MyProject.egg-info/PKG-INFO
writing top-level names to MyProject.egg-info/top_level.txt
writing dependency_links to MyProject.egg-info/dependency_links.txt
writing entry points to MyProject.egg-info/entry_points.txt
reading manifest file 'MyProject.egg-info/SOURCES.txt'
writing manifest file 'MyProject.egg-info/SOURCES.txt'
running build_ext
..
----------------------------------------------------------------------
Ran 1 test in 0.108s

OK

Note

The -q option is passed to the setup.py test command to limit the output to a stream of dots. If you don’t pass -q, you’ll see more verbose test result output (which normally isn’t very useful).

The tests themselves are found in the tests.py module in your paster create -generated project. Within a project generated by the pyramid_starter scaffold, a single sample test exists.

Running The Project Application

Once a project is installed for development, you can run the application it represents using the paster serve command against the generated configuration file. In our case, this file is named development.ini.

On UNIX:

$ ../bin/paster serve development.ini

On Windows:

$ ..\Scripts\paster.exe serve development.ini

Here’s sample output from a run of paster serve on UNIX:

$ ../bin/paster serve development.ini
Starting server in PID 16601.
serving on 0.0.0.0:6543 view at http://127.0.0.1:6543

By default, Pyramid applications generated from a scaffold will listen on TCP port 6543. You can shut down a server started this way by pressing Ctrl-C.

During development, it’s often useful to run paster serve using its --reload option. When --reload is passed to paster serve, changes to any Python module your project uses will cause the server to restart. This typically makes development easier, as changes to Python code made within a Pyramid application is not put into effect until the server restarts.

For example, on UNIX:

$ ../bin/paster serve development.ini --reload
Starting subprocess with file monitor
Starting server in PID 16601.
serving on 0.0.0.0:6543 view at http://127.0.0.1:6543

For more detailed information about the startup process, see Startup. For more information about environment variables and configuration file settings that influence startup and runtime behavior, see Environment Variables and .ini File Settings.

Viewing the Application

Once your application is running via paster serve, you may visit http://localhost:6543/ in your browser. You will see something in your browser like what is displayed in the following image:

_images/project.png

This is the page shown by default when you visit an unmodified paster create -generated pyramid_starter application in a browser.

The Debug Toolbar

If you click on the image shown at the right hand top of the page (“^DT”), you’ll be presented with a debug toolbar that provides various niceties while you’re developing. This image will float above every HTML page served by Pyramid while you develop an application, and allows you show the toolbar as necessary. Click on Hide to hide the toolbar and show the image again.

_images/project-debug.png

For more information about what the debug toolbar allows you to do, see the documentation for pyramid_debugtoolbar.

The debug toolbar will not be shown (and all debugging will be turned off) when you use the production.ini file instead of the development.ini ini file to run the application.

You can also turn the debug toolbar off by editing development.ini and commenting out the line pyramid.includes = pyramid_debugtoolbar. For example, instead of:

1
2
3
[app:main]
...
pyramid.includes = pyramid_debugtoolbar

Put a hash mark in front of the pyramid.includes line:

1
2
3
[app:main]
...
#pyramid.includes = pyramid_debugtoolbar

Then restart the application to see that the toolbar has been turned off.

The Project Structure

The pyramid_starter scaffold generated a project (named MyProject), which contains a Python package. The package is also named myproject, but it’s lowercased; the scaffold generates a project which contains a package that shares its name except for case.

All Pyramid paster -generated projects share a similar structure. The MyProject project we’ve generated has the following directory structure:

MyProject/
|-- CHANGES.txt
|-- development.ini
|-- MANIFEST.in
|-- myproject
|   |-- __init__.py
|   |-- resources.py
|   |-- static
|   |   |-- favicon.ico
|   |   |-- logo.png
|   |   `-- pylons.css
|   |-- templates
|   |   `-- mytemplate.pt
|   |-- tests.py
|   `-- views.py
|-- production.ini
|-- README.txt
|-- setup.cfg
`-- setup.py

The MyProject Project

The MyProject project directory is the distribution and deployment wrapper for your application. It contains both the myproject package representing your application as well as files used to describe, run, and test your application.

  1. CHANGES.txt describes the changes you’ve made to the application. It is conventionally written in ReStructuredText format.
  2. README.txt describes the application in general. It is conventionally written in ReStructuredText format.
  3. development.ini is a PasteDeploy configuration file that can be used to execute your application during development.
  4. production.ini is a PasteDeploy configuration file that can be used to execute your application in a production configuration.
  5. setup.cfg is a setuptools configuration file used by setup.py.
  6. MANIFEST.in is a distutils “manifest” file, naming which files should be included in a source distribution of the package when python setup.py sdist is run.
  7. setup.py is the file you’ll use to test and distribute your application. It is a standard setuptools setup.py file.
development.ini

The development.ini file is a PasteDeploy configuration file. Its purpose is to specify an application to run when you invoke paster serve, as well as the deployment settings provided to that application.

The generated development.ini file looks like so:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
[app:main]
use = egg:MyProject

pyramid.reload_templates = true
pyramid.debug_authorization = false
pyramid.debug_notfound = false
pyramid.debug_routematch = false
pyramid.debug_templates = true
pyramid.default_locale_name = en
pyramid.includes = pyramid_debugtoolbar

[server:main]
use = egg:Paste#http
host = 0.0.0.0
port = 6543

# Begin logging configuration

[loggers]
keys = root, myproject

[handlers]
keys = console

[formatters]
keys = generic

[logger_root]
level = INFO
handlers = console

[logger_myproject]
level = DEBUG
handlers =
qualname = myproject

[handler_console]
class = StreamHandler
args = (sys.stderr,)
level = NOTSET
formatter = generic

[formatter_generic]
format = %(asctime)s %(levelname)-5.5s [%(name)s] %(message)s

# End logging configuration

This file contains several sections including [app:main], [server:main] and several other sections related to logging configuration.

The [app:main] section represents configuration for your Pyramid application. The use setting is the only setting required to be present in the [app:main] section. Its default value, egg:MyProject, indicates that our MyProject project contains the application that should be served. Other settings added to this section are passed as keyword arguments to the function named main in our package’s __init__.py module. You can provide startup-time configuration parameters to your application by adding more settings to this section.

Note

See Entry Points and PasteDeploy .ini Files for more information about the meaning of the use = egg:MyProject value in this section.

The pyramid.reload_templates setting in the [app:main] section is a Pyramid -specific setting which is passed into the framework. If it exists, and its value is true, Chameleon and Mako template changes will not require an application restart to be detected. See Automatically Reloading Templates for more information.

The pyramid.debug_templates setting in the [app:main] section is a Pyramid -specific setting which is passed into the framework. If it exists, and its value is true, Chameleon template exceptions will contain more detailed and helpful information about the error than when this value is false. See Nicer Exceptions in Chameleon Templates for more information.

Warning

The pyramid.reload_templates and pyramid.debug_templates options should be turned off for production applications, as template rendering is slowed when either is turned on.

The pyramid.includes setting in the [app:main] section tells Pyramid to “include” configuration from another package. In this case, the line pyramid.includes = pyramid_debugtoolbar tells Pyramid to include configuration from the pyramid_debugtoolbar package. This turns on a debugging panel in development mode which will be shown on the right hand side of the screen. Including the debug toolbar will also make it possible to interactively debug exceptions when an error occurs.

Various other settings may exist in this section having to do with debugging or influencing runtime behavior of a Pyramid application. See Environment Variables and .ini File Settings for more information about these settings.

The name main in [app:main] signifies that this is the default application run by paster serve when it is invoked against this configuration file. The name main is a convention used by PasteDeploy signifying that it is the default application.

The [server:main] section of the configuration file configures a WSGI server which listens on TCP port 6543. It is configured to listen on all interfaces (0.0.0.0). This means that any remote system which has TCP access to your system can see your Pyramid application.

The sections that live between the markers # Begin logging configuration and # End logging configuration represent Python’s standard library logging module configuration for your application. The sections between these two markers are passed to the logging module’s config file configuration engine when the paster serve or paster pshell commands are executed. The default configuration sends application logging output to the standard error output of your terminal. For more information about logging configuration, see Logging.

See the PasteDeploy documentation for more information about other types of things you can put into this .ini file, such as other applications, middleware and alternate WSGI server implementations.

production.ini

The production.ini file is a PasteDeploy configuration file with a purpose much like that of development.ini. However, it disables the debug toolbar, and filters all log messages except those above the WARN level. It also turns off template development options such that templates are not automatically reloaded when changed, and turns off all debugging options. This file is appropriate to use instead of development.ini when you put your application into production.

It’s important to use production.ini (and not development.ini) to benchmark your application and put it into production. development.ini configures your system with a debug toolbar that helps development, but the inclusion of this toolbar slows down page rendering times by over an order of magnitude. The debug toolbar is also a potential security risk if you have it configured incorrectly.

MANIFEST.in

The MANIFEST.in file is a distutils configuration file which specifies the non-Python files that should be included when a distribution of your Pyramid project is created when you run python setup.py sdist. Due to the information contained in the default MANIFEST.in, an sdist of your Pyramid project will include .txt files, .ini files, .rst files, graphics files, and template files, as well as .py files. See http://docs.python.org/distutils/sourcedist.html#the-manifest-in-template for more information about the syntax and usage of MANIFEST.in.

Without the presence of a MANIFEST.in file or without checking your source code into a version control repository, setup.py sdist places only Python source files (files ending with a .py extension) into tarballs generated by python setup.py sdist. This means, for example, if your project was not checked into a setuptools-compatible source control system, and your project directory didn’t contain a MANIFEST.in file that told the sdist machinery to include *.pt files, the myproject/templates/mytemplate.pt file would not be included in the generated tarball.

Projects generated by Pyramid scaffolds include a default MANIFEST.in file. The MANIFEST.in file contains declarations which tell it to include files like *.pt, *.css and *.js in the generated tarball. If you include files with extensions other than the files named in the project’s MANIFEST.in and you don’t make use of a setuptools-compatible version control system, you’ll need to edit the MANIFEST.in file and include the statements necessary to include your new files. See http://docs.python.org/distutils/sourcedist.html#principle for more information about how to do this.

You can also delete MANIFEST.in from your project and rely on a setuptools feature which simply causes all files checked into a version control system to be put into the generated tarball. To allow this to happen, check all the files that you’d like to be distributed along with your application’s Python files into Subversion. After you do this, when you rerun setup.py sdist, all files checked into the version control system will be included in the tarball. If you don’t use Subversion, and instead use a different version control system, you may need to install a setuptools add-on such as setuptools-git or setuptools-hg for this behavior to work properly.

setup.py

The setup.py file is a setuptools setup file. It is meant to be run directly from the command line to perform a variety of functions, such as testing your application, packaging, and distributing your application.

Note

setup.py is the defacto standard which Python developers use to distribute their reusable code. You can read more about setup.py files and their usage in the Setuptools documentation and The Hitchhiker’s Guide to Packaging.

Our generated setup.py looks like this:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
import os

from setuptools import setup, find_packages

here = os.path.abspath(os.path.dirname(__file__))
README = open(os.path.join(here, 'README.txt')).read()
CHANGES = open(os.path.join(here, 'CHANGES.txt')).read()

requires = ['pyramid', 'pyramid_debugtoolbar']

setup(name='MyProject',
      version='0.0',
      description='MyProject',
      long_description=README + '\n\n' +  CHANGES,
      classifiers=[
        "Programming Language :: Python",
        "Framework :: Pylons",
        "Topic :: Internet :: WWW/HTTP",
        "Topic :: Internet :: WWW/HTTP :: WSGI :: Application",
        ],
      author='',
      author_email='',
      url='',
      keywords='web pyramid pylons',
      packages=find_packages(),
      include_package_data=True,
      zip_safe=False,
      install_requires=requires,
      tests_require=requires,
      test_suite="myproject",
      entry_points = """\
      [paste.app_factory]
      main = myproject:main
      """,
      paster_plugins=['pyramid'],
      )

The setup.py file calls the setuptools setup function, which does various things depending on the arguments passed to setup.py on the command line.

Within the arguments to this function call, information about your application is kept. While it’s beyond the scope of this documentation to explain everything about setuptools setup files, we’ll provide a whirlwind tour of what exists in this file in this section.

Your application’s name can be any string; it is specified in the name field. The version number is specified in the version value. A short description is provided in the description field. The long_description is conventionally the content of the README and CHANGES file appended together. The classifiers field is a list of Trove classifiers describing your application. author and author_email are text fields which probably don’t need any description. url is a field that should point at your application project’s URL (if any). packages=find_packages() causes all packages within the project to be found when packaging the application. include_package_data will include non-Python files when the application is packaged if those files are checked into version control. zip_safe indicates that this package is not safe to use as a zipped egg; instead it will always unpack as a directory, which is more convenient. install_requires and tests_require indicate that this package depends on the pyramid package. test_suite points at the package for our application, which means all tests found in the package will be run when setup.py test is invoked. We examined entry_points in our discussion of the development.ini file; this file defines the main entry point that represents our project’s application.

Usually you only need to think about the contents of the setup.py file when distributing your application to other people, when adding Python package dependencies, or when versioning your application for your own use. For fun, you can try this command now:

$ python setup.py sdist

This will create a tarball of your application in a dist subdirectory named MyProject-0.1.tar.gz. You can send this tarball to other people who want to install and use your application.

setup.cfg

The setup.cfg file is a setuptools configuration file. It contains various settings related to testing and internationalization:

Our generated setup.cfg looks like this:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
[nosetests]
match = ^test
nocapture = 1
cover-package = myproject
with-coverage = 1
cover-erase = 1

[compile_catalog]
directory = myproject/locale
domain = MyProject
statistics = true

[extract_messages]
add_comments = TRANSLATORS:
output_file = myproject/locale/MyProject.pot
width = 80

[init_catalog]
domain = MyProject
input_file = myproject/locale/MyProject.pot
output_dir = myproject/locale

[update_catalog]
domain = MyProject
input_file = myproject/locale/MyProject.pot
output_dir = myproject/locale
previous = true

The values in the default setup file allow various commonly-used internationalization commands and testing commands to work more smoothly.

The myproject Package

The myproject package lives inside the MyProject project. It contains:

  1. An __init__.py file signifies that this is a Python package. It also contains code that helps users run the application, including a main function which is used as a Paste entry point.
  2. A resources.py module, which contains resource code.
  3. A templates directory, which contains Chameleon (or other types of) templates.
  4. A tests.py module, which contains unit test code for the application.
  5. A views.py module, which contains view code for the application.

These are purely conventions established by the scaffold: Pyramid doesn’t insist that you name things in any particular way. However, it’s generally a good idea to follow Pyramid standards for naming, so that other Pyramid developers can get up to speed quickly on your code when you need help.

__init__.py

We need a small Python module that configures our application and which advertises an entry point for use by our PasteDeploy .ini file. This is the file named __init__.py. The presence of an __init__.py also informs Python that the directory which contains it is a package.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
from pyramid.config import Configurator
from myproject.resources import Root

def main(global_config, **settings):
    """ This function returns a Pyramid WSGI application.
    """
    config = Configurator(root_factory=Root, settings=settings)
    config.add_view('myproject.views.my_view',
                    context='myproject.resources.Root',
                    renderer='myproject:templates/mytemplate.pt')
    config.add_static_view('static', 'myproject:static')
    return config.make_wsgi_app()

  1. Line 1 imports the Configurator class from pyramid.config that we use later.

  2. Line 2 imports the Root class from myproject.resources that we use later.

  3. Lines 4-12 define a function named main that returns a Pyramid WSGI application. This function is meant to be called by the PasteDeploy framework as a result of running paster serve.

    Within this function, application configuration is performed.

    Lines 8-10 register a “default view” (a view that has no name attribute). It is registered so that it will be found when the context of the request is an instance of the myproject.resources.Root class. The first argument to add_view points at a Python function that does all the work for this view, also known as a view callable, via a dotted Python name. The view declaration also names a renderer, which in this case is a template that will be used to render the result of the view callable. This particular view declaration points at myproject:templates/mytemplate.pt, which is a asset specification that specifies the mytemplate.pt file within the templates directory of the myproject package. The template file it actually points to is a Chameleon ZPT template file.

    Line 11 registers a static view, which will serve up the files from the mypackage:static asset specification (the static directory of the mypackage package).

    Line 12 returns a WSGI application to the caller of the function (Paste).

views.py

Much of the heavy lifting in a Pyramid application is done by view callables. A view callable is the main tool of a Pyramid web application developer; it is a bit of code which accepts a request and which returns a response.

1
2
def my_view(request):
    return {'project':'MyProject'}

This bit of code was registered as the view callable within __init__.py (via add_view). add_view said that the default URL for instances that are of the class myproject.resources.Root should run this myproject.views.my_view() function.

This view callable function is handed a single piece of information: the request. The request is an instance of the WebOb Request class representing the browser’s request to our server.

This view returns a dictionary. When this view is invoked, a renderer converts the dictionary returned by the view into HTML, and returns the result as the response. This view is configured to invoke a renderer which uses a Chameleon ZPT template (mypackage:templates/my_template.pt, as specified in the __init__.py file call to add_view).

See Writing View Callables Which Use a Renderer for more information about how views, renderers, and templates relate and cooperate.

Note

Because our development.ini has a pyramid.reload_templates = true directive indicating that templates should be reloaded when they change, you won’t need to restart the application server to see changes you make to templates. During development, this is handy. If this directive had been false (or if the directive did not exist), you would need to restart the application server for each template change. For production applications, you should set your project’s pyramid.reload_templates to false to increase the speed at which templates may be rendered.

resources.py

The resources.py module provides the resource data and behavior for our application. Resources are objects which exist to provide site structure in applications which use traversal to map URLs to code. We write a class named Root that provides the behavior for the root resource.

1
2
3
class Root(object):
    def __init__(self, request):
        self.request = request
  1. Lines 1-3 define the Root class. The Root class is a “root resource factory” function that will be called by the Pyramid Router for each request when it wants to find the root of the resource tree.

In a “real” application, the Root object would likely not be such a simple object. Instead, it might be an object that could access some persistent data store, such as a database. Pyramid doesn’t make any assumption about which sort of data storage you’ll want to use, so the sample application uses an instance of myproject.resources.Root to represent the root.

static

This directory contains static assets which support the mytemplate.pt template. It includes CSS and images.

templates/mytemplate.pt

The single Chameleon template that exists in the project. Its contents are too long to show here, but it displays a default page when rendered. It is referenced by the call to add_view as the renderer attribute in the __init__ file. See Writing View Callables Which Use a Renderer for more information about renderers.

Templates are accessed and used by view configurations and sometimes by view functions themselves. See Using Templates Directly and Templates Used as Renderers via Configuration.

tests.py

The tests.py module includes unit tests for your application.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
import unittest

from pyramid import testing

class ViewTests(unittest.TestCase):
    def setUp(self):
        self.config = testing.setUp()

    def tearDown(self):
        testing.tearDown()

    def test_my_view(self):
        from myproject.views import my_view
        request = testing.DummyRequest()
        info = my_view(request)
        self.assertEqual(info['project'], 'MyProject')

This sample tests.py file has a single unit test defined within it. This test is executed when you run python setup.py test. You may add more tests here as you build your application. You are not required to write tests to use Pyramid, this file is simply provided as convenience and example.

See Unit, Integration, and Functional Testing for more information about writing Pyramid unit tests.

Modifying Package Structure

It is best practice for your application’s code layout to not stray too much from accepted Pyramid scaffold defaults. If you refrain from changing things very much, other Pyramid coders will be able to more quickly understand your application. However, the code layout choices made for you by a scaffold are in no way magical or required. Despite the choices made for you by any scaffold, you can decide to lay your code out any way you see fit.

For example, the configuration method named add_view() requires you to pass a dotted Python name or a direct object reference as the class or function to be used as a view. By default, the pyramid_starter scaffold would have you add view functions to the views.py module in your package. However, you might be more comfortable creating a views directory, and adding a single file for each view.

If your project package name was myproject and you wanted to arrange all your views in a Python subpackage within the myproject package named views instead of within a single views.py file, you might:

  • Create a views directory inside your mypackage package directory (the same directory which holds views.py).
  • Move the existing views.py file to a file inside the new views directory named, say, blog.py.
  • Create a file within the new views directory named __init__.py (it can be empty, this just tells Python that the views directory is a package.

Then change the __init__.py of your myproject project (not the __init__.py you just created in the views directory, the one in its parent directory). For example, from something like:

1
2
config.add_view('myproject.views.my_view',
                renderer='myproject:templates/mytemplate.pt')

To this:

1
2
config.add_view('myproject.views.blog.my_view',
                renderer='myproject:templates/mytemplate.pt')

You can then continue to add files to the views directory, and refer to view classes or functions within those files via the dotted name passed as the first argument to add_view. For example, if you added a file named anothermodule.py to the views subdirectory, and added a view callable named my_view to it:

1
2
config.add_view('myproject.views.anothermodule.my_view',
                renderer='myproject:templates/anothertemplate.pt')

This pattern can be used to rearrage code referred to by any Pyramid API argument which accepts a dotted Python name or direct object reference.

Using the Interactive Shell

It is possible to use a Python interpreter prompt loaded with a similar configuration as would be loaded if you were running your Pyramid application via paster serve. This can be a useful debugging tool. See The Interactive Shell for more details.

Using an Alternate WSGI Server

The code generated by a Pyramid scaffold assumes that you will be using the paster serve command to start your application while you do development. However, paster serve is by no means the only way to start up and serve a Pyramid application. As we saw in Creating Your First Pyramid Application, paster serve needn’t be invoked at all to run a Pyramid application. The use of paster serve to run a Pyramid application is purely conventional based on the output of its scaffold.

Any WSGI server is capable of running a Pyramid application. Some WSGI servers don’t require the PasteDeploy framework’s paster serve command to do server process management at all. Each WSGI server has its own documentation about how it creates a process to run an application, and there are many of them, so we cannot provide the details for each here. But the concepts are largely the same, whatever server you happen to use.

One popular production alternative to a paster-invoked server is mod_wsgi. You can also use mod_wsgi to serve your Pyramid application using the Apache web server rather than any “pure-Python” server that is started as a result of paster serve. See Running a Pyramid Application under mod_wsgi for details. However, it is usually easier to develop an application using a paster serve -invoked webserver, as exception and debugging output will be sent to the console.

Startup

When you cause a Pyramid application to start up in a console window, you’ll see something much like this show up on the console:

$ paster serve myproject/MyProject.ini
Starting server in PID 16601.
serving on 0.0.0.0:6543 view at http://127.0.0.1:6543

This chapter explains what happens between the time you press the “Return” key on your keyboard after typing paster serve myproject/MyProject.ini and the time the line serving on 0.0.0.0:6543 ... is output to your console.

The Startup Process

The easiest and best-documented way to start and serve a Pyramid application is to use the paster serve command against a PasteDeploy .ini file. This uses the .ini file to infer settings and starts a server listening on a port. For the purposes of this discussion, we’ll assume that you are using this command to run your Pyramid application.

Here’s a high-level time-ordered overview of what happens when you press return after running paster serve development.ini.

  1. The PasteDeploy paster command is invoked under your shell with the arguments serve and development.ini. As a result, the PasteDeploy framework recognizes that it is meant to begin to run and serve an application using the information contained within the development.ini file.

  2. The PasteDeploy framework finds a section named either [app:main], [pipeline:main], or [composite:main] in the .ini file. This section represents the configuration of a WSGI application that will be served. If you’re using a simple application (e.g. [app:main]), the application entry point or dotted Python name will be named on the use= line within the section’s configuration. If, instead of a simple application, you’re using a WSGI pipeline (e.g. a [pipeline:main] section), the application named on the “last” element will refer to your Pyramid application. If instead of a simple application or a pipeline, you’re using a Paste “composite” (e.g. [composite:main]), refer to the documentation for that particular composite to understand how to make it refer to your Pyramid application. In most cases, a Pyramid application built from a scaffold will have a single [app:main] section in it, and this will be the application served.

  3. The PasteDeploy framework finds all logging related configuration in the .ini file and uses it to configure the Python standard library logging system for this application.

  4. The application’s constructor (named by the entry point reference or dotted Python name on the use= line of the section representing your Pyramid application) is passed the key/value parameters mentioned within the section in which it’s defined. The constructor is meant to return a router instance, which is a WSGI application.

    For Pyramid applications, the constructor will be a function named main in the __init__.py file within the package in which your application lives. If this function succeeds, it will return a Pyramid router instance. Here’s the contents of an example __init__.py module:

     1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
    from pyramid.config import Configurator
from myproject.resources import Root

def main(global_config, **settings):
    """ This function returns a Pyramid WSGI application.
    """
    config = Configurator(root_factory=Root, settings=settings)
    config.add_view('myproject.views.my_view',
                    context='myproject.resources.Root',
                    renderer='myproject:templates/mytemplate.pt')
    config.add_static_view('static', 'myproject:static')
    return config.make_wsgi_app()

    Note that the constructor function accepts a global_config argument, which is a dictionary of key/value pairs mentioned in the [DEFAULT] section of an .ini file. It also accepts a **settings argument, which collects another set of arbitrary key/value pairs. The arbitrary key/value pairs received by this function in **settings will be composed of all the key/value pairs that are present in the [app:main] section (except for the use= setting) when this function is called by the PasteDeploy framework when you run paster serve.

    Our generated development.ini file looks like so:

     1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
    [app:main]
use = egg:MyProject

pyramid.reload_templates = true
pyramid.debug_authorization = false
pyramid.debug_notfound = false
pyramid.debug_routematch = false
pyramid.debug_templates = true
pyramid.default_locale_name = en
pyramid.includes = pyramid_debugtoolbar

[server:main]
use = egg:Paste#http
host = 0.0.0.0
port = 6543

# Begin logging configuration

[loggers]
keys = root, myproject

[handlers]
keys = console

[formatters]
keys = generic

[logger_root]
level = INFO
handlers = console

[logger_myproject]
level = DEBUG
handlers =
qualname = myproject

[handler_console]
class = StreamHandler
args = (sys.stderr,)
level = NOTSET
formatter = generic

[formatter_generic]
format = %(asctime)s %(levelname)-5.5s [%(name)s] %(message)s

# End logging configuration

    In this case, the myproject.__init__:main function referred to by the entry point URI egg:MyProject (see development.ini for more information about entry point URIs, and how they relate to callables), will receive the key/value pairs {'pyramid.reload_templates':'true', 'pyramid.debug_authorization':'false', 'pyramid.debug_notfound':'false', 'pyramid.debug_routematch':'false', 'pyramid.debug_templates':'true', 'pyramid.default_locale_name':'en'}.

  5. The main function first constructs a Configurator instance, passing a root resource factory (constructor) to it as its root_factory argument, and settings dictionary captured via the **settings kwarg as its settings argument.

    The root resource factory is invoked on every request to retrieve the application’s root resource. It is not called during startup, only when a request is handled.

    The settings dictionary contains all the options in the [app:main] section of our .ini file except the use option (which is internal to Paste) such as pyramid.reload_templates, pyramid.debug_authorization, etc.

  6. The main function then calls various methods on the instance of the class Configurator created in the previous step. The intent of calling these methods is to populate an application registry, which represents the Pyramid configuration related to the application.

  7. The make_wsgi_app() method is called. The result is a router instance. The router is associated with the application registry implied by the configurator previously populated by other methods run against the Configurator. The router is a WSGI application.

  8. A ApplicationCreated event is emitted (see Using Events for more information about events).

  9. Assuming there were no errors, the main function in myproject returns the router instance created by pyramid.config.Configurator.make_wsgi_app() back to PasteDeploy. As far as PasteDeploy is concerned, it is “just another WSGI application”.

  10. PasteDeploy starts the WSGI server defined within the [server:main] section. In our case, this is the Paste#http server (use = egg:Paste#http), and it will listen on all interfaces (host = 0.0.0.0), on port number 6543 (port = 6543). The server code itself is what prints serving on 0.0.0.0:6543 view at http://127.0.0.1:6543. The server serves the application, and the application is running, waiting to receive requests.

Deployment Settings

Note that an augmented version of the values passed as **settings to the Configurator constructor will be available in Pyramid view callable code as request.registry.settings. You can create objects you wish to access later from view code, and put them into the dictionary you pass to the configurator as settings. They will then be present in the request.registry.settings dictionary at application runtime.

URL Dispatch

URL dispatch provides a simple way to map URLs to view code using a simple pattern matching language. An ordered set of patterns is checked one-by-one. If one of the patterns matches the path information associated with a request, a particular view callable is invoked. A view callable is a specific bit of code, defined in your application, that receives the request and returns a response object.

High-Level Operational Overview

If route configuration is present in an application, the Pyramid Router checks every incoming request against an ordered set of URL matching patterns present in a route map.

If any route pattern matches the information in the request, Pyramid will invoke view lookup to find a matching view.

If no route pattern in the route map matches the information in the request provided in your application, Pyramid will fail over to using traversal to perform resource location and view lookup.

Route Configuration

Route configuration is the act of adding a new route to an application. A route has a name, which acts as an identifier to be used for URL generation. The name also allows developers to associate a view configuration with the route. A route also has a pattern, meant to match against the PATH_INFO portion of a URL (the portion following the scheme and port, e.g. /foo/bar in the URL http://localhost:8080/foo/bar). It also optionally has a factory and a set of route predicate attributes.

Configuring a Route to Match a View

The pyramid.config.Configurator.add_route() method adds a single route configuration to the application registry. Here’s an example:

# "config" below is presumed to be an instance of the
# pyramid.config.Configurator class; "myview" is assumed
# to be a "view callable" function
from views import myview
config.add_route('myroute', '/prefix/{one}/{two}')
config.add_view(myview, route_name='myroute')

When a view callable added to the configuration by way of add_view() bcomes associated with a route via its route_name predicate, that view callable will always be found and invoked when the associated route pattern matches during a request.

More commonly, you will not use any add_view statements in your project’s “setup” code, instead only using add_route statements using a scan for to associate view callables with routes. For example, if this is a portion of your project’s __init__.py:

# in your project's __init__.py (mypackage.__init__)

config.add_route('myroute', '/prefix/{one}/{two}')
config.scan('mypackage')

Note that we don’t call add_view() in this setup code. However, the above scan execution config.scan('mypackage') will pick up all configuration decoration, including any objects decorated with the pyramid.view.view_config decorator in the mypackage Python pakage. For example, if you have a views.py in your package, a scan will pick up any of its configuration decorators, so we can add one there that that references myroute as a route_name parameter:

# in your project's views.py module (mypackage.views)

from pyramid.view import view_config
from pyramid.response import Response

@view_config(route_name='myroute')
def myview(request):
    return Response('OK')

THe above combination of add_route and scan is completely equivalent to using the previous combination of add_route and add_view.

Route Pattern Syntax

The syntax of the pattern matching language used by Pyramid URL dispatch in the pattern argument is straightforward; it is close to that of the Routes system used by Pylons.

The pattern used in route configuration may start with a slash character. If the pattern does not start with a slash character, an implicit slash will be prepended to it at matching time. For example, the following patterns are equivalent:

{foo}/bar/baz

and:

/{foo}/bar/baz

A pattern segment (an individual item between / characters in the pattern) may either be a literal string (e.g. foo) or it may be a replacement marker (e.g. {foo}) or a certain combination of both. A replacement marker does not need to be preceded by a / character.

A replacement marker is in the format {name}, where this means “accept any characters up to the next slash character and use this as the name matchdict value.”

A replacement marker in a pattern must begin with an uppercase or lowercase ASCII letter or an underscore, and can be composed only of uppercase or lowercase ASCII letters, underscores, and numbers. For example: a, a_b, _b, and b9 are all valid replacement marker names, but 0a is not.

Note

A replacement marker could not start with an underscore until Pyramid 1.2. Previous versions required that the replacement marker start with an uppercase or lowercase letter.

A matchdict is the dictionary representing the dynamic parts extracted from a URL based on the routing pattern. It is available as request.matchdict. For example, the following pattern defines one literal segment (foo) and two replacement markers (baz, and bar):

foo/{baz}/{bar}

The above pattern will match these URLs, generating the following matchdicts:

foo/1/2        -> {'baz':u'1', 'bar':u'2'}
foo/abc/def    -> {'baz':u'abc', 'bar':u'def'}

It will not match the following patterns however:

foo/1/2/        -> No match (trailing slash)
bar/abc/def     -> First segment literal mismatch

The match for a segment replacement marker in a segment will be done only up to the first non-alphanumeric character in the segment in the pattern. So, for instance, if this route pattern was used:

foo/{name}.html

The literal path /foo/biz.html will match the above route pattern, and the match result will be {'name':u'biz'}. However, the literal path /foo/biz will not match, because it does not contain a literal .html at the end of the segment represented by {name}.html (it only contains biz, not biz.html).

To capture both segments, two replacement markers can be used:

foo/{name}.{ext}

The literal path /foo/biz.html will match the above route pattern, and the match result will be {'name': 'biz', 'ext': 'html'}. This occurs because there is a literal part of . (period) between the two replacement markers {name} and {ext}.

Replacement markers can optionally specify a regular expression which will be used to decide whether a path segment should match the marker. To specify that a replacement marker should match only a specific set of characters as defined by a regular expression, you must use a slightly extended form of replacement marker syntax. Within braces, the replacement marker name must be followed by a colon, then directly thereafter, the regular expression. The default regular expression associated with a replacement marker [^/]+ matches one or more characters which are not a slash. For example, under the hood, the replacement marker {foo} can more verbosely be spelled as {foo:[^/]+}. You can change this to be an arbitrary regular expression to match an arbitrary sequence of characters, such as {foo:\d+} to match only digits.

It is possible to use two replacement markers without any literal characters between them, for instance /{foo}{bar}. However, this would be a nonsensical pattern without specifying a custom regular expression to restrict what each marker captures.

Segments must contain at least one character in order to match a segment replacement marker. For example, for the URL /abc/:

  • /abc/{foo} will not match.
  • /{foo}/ will match.

Note that values representing matched path segments will be url-unquoted and decoded from UTF-8 into Unicode within the matchdict. So for instance, the following pattern:

foo/{bar}

When matching the following URL:

http://example.com/foo/La%20Pe%C3%B1a

The matchdict will look like so (the value is URL-decoded / UTF-8 decoded):

{'bar':u'La Pe\xf1a'}

Literal strings in the path segment should represent the decoded value of the PATH_INFO provided to Pyramid. You don’t want to use a URL-encoded value or a bytestring representing the literal’s UTF-8 in the pattern. For example, rather than this:

/Foo%20Bar/{baz}

You’ll want to use something like this:

/Foo Bar/{baz}

For patterns that contain “high-order” characters in its literals, you’ll want to use a Unicode value as the pattern as opposed to any URL-encoded or UTF-8-encoded value. For example, you might be tempted to use a bytestring pattern like this:

/La Pe\xc3\xb1a/{x}

But this will either cause an error at startup time or it won’t match properly. You’ll want to use a Unicode value as the pattern instead rather than raw bytestring escapes. You can use a high-order Unicode value as the pattern by using Python source file encoding plus the “real” character in the Unicode pattern in the source, like so:

/La Peña/{x}

Or you can ignore source file encoding and use equivalent Unicode escape characters in the pattern.

/La Pe\xf1a/{x}

Dynamic segment names cannot contain high-order characters, so this applies only to literals in the pattern.

If the pattern has a * in it, the name which follows it is considered a “remainder match”. A remainder match must come at the end of the pattern. Unlike segment replacement markers, it does not need to be preceded by a slash. For example:

foo/{baz}/{bar}*fizzle

The above pattern will match these URLs, generating the following matchdicts:

foo/1/2/           ->
         {'baz':u'1', 'bar':u'2', 'fizzle':()}

foo/abc/def/a/b/c  ->
         {'baz':u'abc', 'bar':u'def', 'fizzle':(u'a', u'b', u'c')}

Note that when a *stararg remainder match is matched, the value put into the matchdict is turned into a tuple of path segments representing the remainder of the path. These path segments are url-unquoted and decoded from UTF-8 into Unicode. For example, for the following pattern:

foo/*fizzle

When matching the following path:

/foo/La%20Pe%C3%B1a/a/b/c

Will generate the following matchdict:

{'fizzle':(u'La Pe\xf1a', u'a', u'b', u'c')}

By default, the *stararg will parse the remainder sections into a tuple split by segment. Changing the regular expression used to match a marker can also capture the remainder of the URL, for example:

foo/{baz}/{bar}{fizzle:.*}

The above pattern will match these URLs, generating the following matchdicts:

foo/1/2/           -> {'baz':u'1', 'bar':u'2', 'fizzle':()}
foo/abc/def/a/b/c  -> {'baz':u'abc', 'bar':u'def', 'fizzle': u'a/b/c')}

This occurs because the default regular expression for a marker is [^/]+ which will match everything up to the first /, while {fizzle:.*} will result in a regular expression match of .* capturing the remainder into a single value.

Route Declaration Ordering

Route configuration declarations are evaluated in a specific order when a request enters the system. As a result, the order of route configuration declarations is very important. The order that routes declarations are evaluated is the order in which they are added to the application at startup time. (This is unlike a different way of mapping URLs to code that Pyramid provides, named traversal, which does not depend on pattern ordering).

For routes added via the add_route method, the order that routes are evaluated is the order in which they are added to the configuration imperatively.

For example, route configuration statements with the following patterns might be added in the following order:

members/{def}
members/abc

In such a configuration, the members/abc pattern would never be matched. This is because the match ordering will always match members/{def} first; the route configuration with members/abc will never be evaluated.

Route Configuration Arguments

Route configuration add_route statements may specify a large number of arguments. They are documented as part of the API documentation at pyramid.config.Configurator.add_route().

Many of these arguments are route predicate arguments. A route predicate argument specifies that some aspect of the request must be true for the associated route to be considered a match during the route matching process. Examples of route predicate arguments are pattern, xhr, and request_method.

Other arguments are name and factory. These arguments represent neither predicates nor view configuration information.

Warning

Some arguments are view-configuration related arguments, such as view_renderer. These only have an effect when the route configuration names a view and these arguments have been deprecated as of Pyramid 1.1.

Route Matching

The main purpose of route configuration is to match (or not match) the PATH_INFO present in the WSGI environment provided during a request against a URL path pattern. PATH_INFO represents the path portion of the URL that was requested.

The way that Pyramid does this is very simple. When a request enters the system, for each route configuration declaration present in the system, Pyramid checks the request’s PATH_INFO against the pattern declared. This checking happens in the order that the routes were declared via pyramid.config.Configurator.add_route().

When a route configuration is declared, it may contain route predicate arguments. All route predicates associated with a route declaration must be True for the route configuration to be used for a given request during a check. If any predicate in the set of route predicate arguments provided to a route configuration returns False during a check, that route is skipped and route matching continues through the ordered set of routes.

If any route matches, the route matching process stops and the view lookup subsystem takes over to find the most reasonable view callable for the matched route. Most often, there’s only one view that will match (a view configured with a route_name argument matching the matched route). To gain a better understanding of how routes and views are associated in a real application, you can use the paster pviews command, as documented in Displaying Matching Views for a Given URL.

If no route matches after all route patterns are exhausted, Pyramid falls back to traversal to do resource location and view lookup.

The Matchdict

When the URL pattern associated with a particular route configuration is matched by a request, a dictionary named matchdict is added as an attribute of the request object. Thus, request.matchdict will contain the values that match replacement patterns in the pattern element. The keys in a matchdict will be strings. The values will be Unicode objects.

Note

If no route URL pattern matches, the matchdict object attached to the request will be None.

The Matched Route

When the URL pattern associated with a particular route configuration is matched by a request, an object named matched_route is added as an attribute of the request object. Thus, request.matched_route will be an object implementing the IRoute interface which matched the request. The most useful attribute of the route object is name, which is the name of the route that matched.

Note

If no route URL pattern matches, the matched_route object attached to the request will be None.

Routing Examples

Let’s check out some examples of how route configuration statements might be commonly declared, and what will happen if they are matched by the information present in a request.

Example 1

The simplest route declaration which configures a route match to directly result in a particular view callable being invoked:

1
2
 config.add_route('idea', 'site/{id}')
 config.add_view('mypackage.views.site_view', route_name='idea')

When a route configuration with a view attribute is added to the system, and an incoming request matches the pattern of the route configuration, the view callable named as the view attribute of the route configuration will be invoked.

In the case of the above example, when the URL of a request matches /site/{id}, the view callable at the Python dotted path name mypackage.views.site_view will be called with the request. In other words, we’ve associated a view callable directly with a route pattern.

When the /site/{id} route pattern matches during a request, the site_view view callable is invoked with that request as its sole argument. When this route matches, a matchdict will be generated and attached to the request as request.matchdict. If the specific URL matched is /site/1, the matchdict will be a dictionary with a single key, id; the value will be the string '1', ex.: {'id':'1'}.

The mypackage.views module referred to above might look like so:

1
2
3
4
from pyramid.response import Response

def site_view(request):
    return Response(request.matchdict['id'])

The view has access to the matchdict directly via the request, and can access variables within it that match keys present as a result of the route pattern.

See Views, and View Configuration for more information about views.

Example 2

Below is an example of a more complicated set of route statements you might add to your application:

1
2
3
4
5
6
7
config.add_route('idea', 'ideas/{idea}')
config.add_route('user', 'users/{user}')
config.add_route('tag', 'tags/{tags}')

config.add_view('mypackage.views.idea_view', route_name='idea')
config.add_view('mypackage.views.user_view', route_name='user')
config.add_view('mypackage.views.tag_view', route_name='tag')

The above configuration will allow Pyramid to service URLs in these forms:

/ideas/{idea}
/users/{user}
/tags/{tag}
  • When a URL matches the pattern /ideas/{idea}, the view callable available at the dotted Python pathname mypackage.views.idea_view will be called. For the specific URL /ideas/1, the matchdict generated and attached to the request will consist of {'idea':'1'}.
  • When a URL matches the pattern /users/{user}, the view callable available at the dotted Python pathname mypackage.views.user_view will be called. For the specific URL /users/1, the matchdict generated and attached to the request will consist of {'user':'1'}.
  • When a URL matches the pattern /tags/{tag}, the view callable available at the dotted Python pathname mypackage.views.tag_view will be called. For the specific URL /tags/1, the matchdict generated and attached to the request will consist of {'tag':'1'}.

In this example we’ve again associated each of our routes with a view callable directly. In all cases, the request, which will have a matchdict attribute detailing the information found in the URL by the process will be passed to the view callable.

Example 3

The context resource object passed in to a view found as the result of URL dispatch will, by default, be an instance of the object returned by the root factory configured at startup time (the root_factory argument to the Configurator used to configure the application).

You can override this behavior by passing in a factory argument to the add_route() method for a particular route. The factory should be a callable that accepts a request and returns an instance of a class that will be the context resource used by the view.

An example of using a route with a factory:

1
2
config.add_route('idea', 'ideas/{idea}', factory='myproject.resources.Idea')
config.add_view('myproject.views.idea_view', route_name='idea')

The above route will manufacture an Idea resource as a context, assuming that mypackage.resources.Idea resolves to a class that accepts a request in its __init__. For example:

1
2
3
class Idea(object):
    def __init__(self, request):
        pass

In a more complicated application, this root factory might be a class representing a SQLAlchemy model.

See Route Factories for more details about how to use route factories.

Matching the Root URL

It’s not entirely obvious how to use a route pattern to match the root URL (“/”). To do so, give the empty string as a pattern in a call to add_route():

1
config.add_route('root', '')

Or provide the literal string / as the pattern:

1
config.add_route('root', '/')

Generating Route URLs

Use the pyramid.request.Request.route_url() method to generate URLs based on route patterns. For example, if you’ve configured a route with the name “foo” and the pattern “{a}/{b}/{c}”, you might do this.

1
url = request.route_url('foo', a='1', b='2', c='3')

This would return something like the string http://example.com/1/2/3 (at least if the current protocol and hostname implied http://example.com).

To generate only the path portion of a URL from a route, use the pyramid.request.Request.route_path() API instead of route_url().

url = request.route_path('foo', a='1', b='2', c='3')

This will return the string /1/2/3 rather than a full URL.

Replacement values passed to route_url or route_path must be Unicode or bytestrings encoded in UTF-8. One exception to this rule exists: if you’re trying to replace a “remainder” match value (a *stararg replacement value), the value may be a tuple containing Unicode strings or UTF-8 strings.

Note that URLs and paths generated by route_path and route_url are always URL-quoted string types (they contain no non-ASCII characters). Therefore, if you’ve added a route like so:

config.add_route('la', u'/La Peña/{city}')

And you later generate a URL using route_path or route_url like so:

url = request.route_path('la', city=u'Québec')

You will wind up with the path encoded to UTF-8 and URL quoted like so:

/La%20Pe%C3%B1a/Qu%C3%A9bec

If you have a *stararg remainder dynamic part of your route pattern:

config.add_route('abc', 'a/b/c/*foo')

And you later generate a URL using route_path or route_url using a string as the replacement value:

url = request.route_path('abc', foo=u'Québec/biz')

The value you pass will be URL-quoted except for embedded slashes in the result:

/a/b/c/Qu%C3%A9bec/biz

You can get a similar result by passing a tuple composed of path elements:

url = request.route_path('abc', foo=(u'Québec', u'biz'))

Each value in the tuple will be url-quoted and joined by slashes in this case:

/a/b/c/Qu%C3%A9bec/biz

Static Routes

Routes may be added with a static keyword argument. For example:

1
2
config = Configurator()
config.add_route('page', '/page/{action}', static=True)

Routes added with a True static keyword argument will never be considered for matching at request time. Static routes are useful for URL generation purposes only. As a result, it is usually nonsensical to provide other non-name and non-pattern arguments to add_route() when static is passed as True, as none of the other arguments will ever be employed. A single exception to this rule is use of the pregenerator argument, which is not ignored when static is True.

Note

the static argument to add_route() is new as of Pyramid 1.1.

Redirecting to Slash-Appended Routes

For behavior like Django’s APPEND_SLASH=True, use the append_slash_notfound_view() view as the Not Found view in your application. Defining this view as the Not Found view is a way to automatically redirect requests where the URL lacks a trailing slash, but requires one to match the proper route. When configured, along with at least one other route in your application, this view will be invoked if the value of PATH_INFO does not already end in a slash, and if the value of PATH_INFO plus a slash matches any route’s pattern. In this case it does an HTTP redirect to the slash-appended PATH_INFO.

Let’s use an example, because this behavior is a bit magical. If the append_slash_notfound_view is configured in your application and your route configuration looks like so:

1
2
3
4
5
config.add_route('noslash', 'no_slash')
config.add_route('hasslash', 'has_slash/')

config.add_view('myproject.views.no_slash', route_name='noslash')
config.add_view('myproject.views.has_slash', route_name='hasslash')

If a request enters the application with the PATH_INFO value of /has_slash/, the second route will match. If a request enters the application with the PATH_INFO value of /has_slash, a route will be found by the slash-appending not found view. An HTTP redirect to /has_slash/ will be returned to the user’s browser.

If a request enters the application with the PATH_INFO value of /no_slash, the first route will match. However, if a request enters the application with the PATH_INFO value of /no_slash/, no route will match, and the slash-appending not found view will not find a matching route with an appended slash.

Warning

You should not rely on this mechanism to redirect POST requests. The redirect of the slash-appending not found view will turn a POST request into a GET, losing any POST data in the original request.

To configure the slash-appending not found view in your application, change the application’s startup configuration, adding the following stanza:

1
2
config.add_view('pyramid.view.append_slash_notfound_view',
                context='pyramid.httpexceptions.HTTPNotFound')

See pyramid.view and Changing the Not Found View for more information about the slash-appending not found view and for a more general description of how to configure a not found view.

Custom Not Found View With Slash Appended Routes

There can only be one Not Found view in any Pyramid application. Even if you use append_slash_notfound_view() as the Not Found view, Pyramid still must generate a 404 Not Found response when it cannot redirect to a slash-appended URL; this not found response will be visible to site users.

If you don’t care what this 404 response looks like, and only you need redirections to slash-appended route URLs, you may use the append_slash_notfound_view() object as the Not Found view as described above. However, if you wish to use a custom notfound view callable when a URL cannot be redirected to a slash-appended URL, you may wish to use an instance of the AppendSlashNotFoundViewFactory class as the Not Found view, supplying a view callable to be used as the custom notfound view as the first argument to its constructor. For instance:

1
2
3
4
5
6
7
8
from pyramid.httpexceptions import HTTPNotFound
from pyramid.view import AppendSlashNotFoundViewFactory

def notfound_view(context, request):
    return HTTPNotFound('It aint there, stop trying!')

custom_append_slash = AppendSlashNotFoundViewFactory(notfound_view)
config.add_view(custom_append_slash, context=HTTPNotFound)

The notfound_view supplied must adhere to the two-argument view callable calling convention of (context, request) (context will be the exception object).

Debugging Route Matching

It’s useful to be able to take a peek under the hood when requests that enter your application arent matching your routes as you expect them to. To debug route matching, use the PYRAMID_DEBUG_ROUTEMATCH environment variable or the pyramid.debug_routematch configuration file setting (set either to true). Details of the route matching decision for a particular request to the Pyramid application will be printed to the stderr of the console which you started the application from. For example:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
 [chrism@thinko pylonsbasic]$ PYRAMID_DEBUG_ROUTEMATCH=true \
                              bin/paster serve development.ini
 Starting server in PID 13586.
 serving on 0.0.0.0:6543 view at http://127.0.0.1:6543
 2010-12-16 14:45:19,956 no route matched for url \
                                     http://localhost:6543/wontmatch
 2010-12-16 14:45:20,010 no route matched for url \
                             http://localhost:6543/favicon.ico
 2010-12-16 14:41:52,084 route matched for url \
                             http://localhost:6543/static/logo.png; \
                             route_name: 'static/', ....

See Environment Variables and .ini File Settings for more information about how, and where to set these values.

You can also use the paster proutes command to see a display of all the routes configured in your application; for more information, see Displaying All Application Routes.

Using a Route Prefix to Compose Applications

Note

This feature is new as of Pyramid 1.2.

The pyramid.config.Configurator.include() method allows configuration statements to be included from separate files. See Rules for Building An Extensible Application for information about this method. Using pyramid.config.Configurator.include() allows you to build your application from small and potentially reusable components.

The pyramid.config.Configurator.include() method accepts an argument named route_prefix which can be useful to authors of URL-dispatch-based applications. If route_prefix is supplied to the include method, it must be a string. This string represents a route prefix that will be prepended to all route patterns added by the included configuration. Any calls to pyramid.config.Configurator.add_route() within the included callable will have their pattern prefixed with the value of route_prefix. This can be used to help mount a set of routes at a different location than the included callable’s author intended while still maintaining the same route names. For example:

1
2
3
4
5
6
7
8
from pyramid.config import Configurator

def users_include(config):
    config.add_route('show_users', '/show')

def main(global_config, **settings):
    config = Configurator()
    config.include(users_include, route_prefix='/users')

In the above configuration, the show_users route will have an effective route pattern of /users/show, instead of /show because the route_prefix argument will be prepended to the pattern. The route will then only match if the URL path is /users/show, and when the pyramid.request.Request.route_url() function is called with the route name show_users, it will generate a URL with that same path.

Route prefixes are recursive, so if a callable executed via an include itself turns around and includes another callable, the second-level route prefix will be prepended with the first:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
from pyramid.config import Configurator

def timing_include(config):
    config.add_route('show_times', /times')

def users_include(config):
    config.add_route('show_users', '/show')
    config.include(timing_include, route_prefix='/timing')

def main(global_config, **settings):
    config = Configurator()
    config.include(users_include, route_prefix='/users')

In the above configuration, the show_users route will still have an effective route pattern of /users/show. The show_times route however, will have an effective pattern of /users/timing/show_times.

Route prefixes have no impact on the requirement that the set of route names in any given Pyramid configuration must be entirely unique. If you compose your URL dispatch application out of many small subapplications using pyramid.config.Configurator.include(), it’s wise to use a dotted name for your route names, so they’ll be unlikely to conflict with other packages that may be added in the future. For example:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
from pyramid.config import Configurator

def timing_include(config):
    config.add_route('timing.show_times', /times')

def users_include(config):
    config.add_route('users.show_users', '/show')
    config.include(timing_include, route_prefix='/timing')

def main(global_config, **settings):
    config = Configurator()
    config.include(users_include, route_prefix='/users')

Custom Route Predicates

Each of the predicate callables fed to the custom_predicates argument of add_route() must be a callable accepting two arguments. The first argument passed to a custom predicate is a dictionary conventionally named info. The second argument is the current request object.

The info dictionary has a number of contained values: match is a dictionary: it represents the arguments matched in the URL by the route. route is an object representing the route which was matched (see pyramid.interfaces.IRoute for the API of such a route object).

info['match'] is useful when predicates need access to the route match. For example:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
def any_of(segment_name, *allowed):
    def predicate(info, request):
        if info['match'][segment_name] in allowed:
            return True
    return predicate

num_one_two_or_three = any_of('num', 'one', 'two', 'three')

config.add_route('route_to_num', '/{num}',
                 custom_predicates=(num_one_two_or_three,))

The above any_of function generates a predicate which ensures that the match value named segment_name is in the set of allowable values represented by allowed. We use this any_of function to generate a predicate function named num_one_two_or_three, which ensures that the num segment is one of the values one, two, or three , and use the result as a custom predicate by feeding it inside a tuple to the custom_predicates argument to add_route().

A custom route predicate may also modify the match dictionary. For instance, a predicate might do some type conversion of values:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
 def integers(*segment_names):
     def predicate(info, request):
         match = info['match']
         for segment_name in segment_names:
             try:
                 match[segment_name] = int(match[segment_name])
             except (TypeError, ValueError):
                 pass
         return True
     return predicate

 ymd_to_int = integers('year', 'month', 'day')

 config.add_route('ymd', '/{year}/{month}/{day}',
                  custom_predicates=(ymd_to_int,))

Note that a conversion predicate is still a predicate so it must return True or False; a predicate that does only conversion, such as the one we demonstrate above should unconditionally return True.

To avoid the try/except uncertainty, the route pattern can contain regular expressions specifying requirements for that marker. For instance:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
 def integers(*segment_names):
     def predicate(info, request):
         match = info['match']
         for segment_name in segment_names:
             match[segment_name] = int(match[segment_name])
         return True
     return predicate

 ymd_to_int = integers('year', 'month', 'day')

 config.add_route('ymd', '/{year:\d+}/{month:\d+}/{day:\d+}',
                  custom_predicates=(ymd_to_int,))

Now the try/except is no longer needed because the route will not match at all unless these markers match \d+ which requires them to be valid digits for an int type conversion.

The match dictionary passed within info to each predicate attached to a route will be the same dictionary. Therefore, when registering a custom predicate which modifies the match dict, the code registering the predicate should usually arrange for the predicate to be the last custom predicate in the custom predicate list. Otherwise, custom predicates which fire subsequent to the predicate which performs the match modification will receive the modified match dictionary.

Warning

It is a poor idea to rely on ordering of custom predicates to build a conversion pipeline, where one predicate depends on the side effect of another. For instance, it’s a poor idea to register two custom predicates, one which handles conversion of a value to an int, the next which handles conversion of that integer to some custom object. Just do all that in a single custom predicate.

The route object in the info dict is an object that has two useful attributes: name and pattern. The name attribute is the route name. The pattern attribute is the route pattern. An example of using the route in a set of route predicates:

1
2
3
4
5
6
7
8
 def twenty_ten(info, request):
     if info['route'].name in ('ymd', 'ym', 'y'):
         return info['match']['year'] == '2010'

 config.add_route('y', '/{year}', custom_predicates=(twenty_ten,))
 config.add_route('ym', '/{year}/{month}', custom_predicates=(twenty_ten,))
 config.add_route('ymd', '/{year}/{month}/{day}',
                  custom_predicates=(twenty_ten,))

The above predicate, when added to a number of route configurations ensures that the year match argument is ‘2010’ if and only if the route name is ‘ymd’, ‘ym’, or ‘y’.

You can also caption the predicates by setting the __text__ attribute. This will help you with the paster pviews command (see Displaying All Application Routes) and the pyramid_debugtoolbar.

If a predicate is a class just add __text__ property in a standard manner.

1
2
3
4
5
6
class DummyCustomPredicate1(object):
    def __init__(self):
        self.__text__ = 'my custom class predicate'

class DummyCustomPredicate2(object):
    __text__ = 'my custom class predicate'

If a predicate is a method you’ll need to assign it after method declaration (see PEP 232)

1
2
3
def custom_predicate():
    pass
custom_predicate.__text__ = 'my custom method predicate'

If a predicate is a classmethod using @classmethod will not work, but you can still easily do it by wrapping it in classmethod call.

1
2
3
4
def classmethod_predicate():
    pass
classmethod_predicate.__text__ = 'my classmethod predicate'
classmethod_predicate = classmethod(classmethod_predicate)

Same will work with staticmethod, just use staticmethod instead of classmethod.

See also pyramid.interfaces.IRoute for more API documentation about route objects.

Route Factories

Although it is not a particular common need in basic applications, a “route” configuration declaration can mention a “factory”. When that route matches a request, and a factory is attached to a route, the root factory passed at startup time to the Configurator is ignored; instead the factory associated with the route is used to generate a root object. This object will usually be used as the context resource of the view callable ultimately found via view lookup.

1
2
3
config.add_route('abc', '/abc',
                 factory='myproject.resources.root_factory')
config.add_view('myproject.views.theview', route_name='abc')

The factory can either be a Python object or a dotted Python name (a string) which points to such a Python object, as it is above.

In this way, each route can use a different factory, making it possible to supply a different context resource object to the view related to each particular route.

A factory must be a callable which accepts a request and returns an arbitrary Python object. For example, the below class can be used as a factory:

1
2
3
class Mine(object):
    def __init__(self, request):
        pass

A route factory is actually conceptually identical to the root factory described at The Resource Tree.

Supplying a different resource factory for each route is useful when you’re trying to use a Pyramid authorization policy to provide declarative, “context sensitive” security checks; each resource can maintain a separate ACL, as documented in Using Pyramid Security With URL Dispatch. It is also useful when you wish to combine URL dispatch with traversal as documented within Combining Traversal and URL Dispatch.

Using Pyramid Security With URL Dispatch

Pyramid provides its own security framework which consults an authorization policy before allowing any application code to be called. This framework operates in terms of an access control list, which is stored as an __acl__ attribute of a resource object. A common thing to want to do is to attach an __acl__ to the resource object dynamically for declarative security purposes. You can use the factory argument that points at a factory which attaches a custom __acl__ to an object at its creation time.

Such a factory might look like so:

1
2
3
4
5
6
class Article(object):
    def __init__(self, request):
       matchdict = request.matchdict
       article = matchdict.get('article', None)
       if article == '1':
           self.__acl__ = [ (Allow, 'editor', 'view') ]

If the route archives/{article} is matched, and the article number is 1, Pyramid will generate an Article context resource with an ACL on it that allows the editor principal the view permission. Obviously you can do more generic things than inspect the routes match dict to see if the article argument matches a particular string; our sample Article factory class is not very ambitious.

Note

See Security for more information about Pyramid security and ACLs.

Route View Callable Registration and Lookup Details

When a request enters the system which matches the pattern of the route, the usual result is simple: the view callable associated with the route is invoked with the request that caused the invocation.

For most usage, you needn’t understand more than this; how it works is an implementation detail. In the interest of completeness, however, we’ll explain how it does work in the this section. You can skip it if you’re uninterested.

When a view is associated with a route configuration, Pyramid ensures that a view configuration is registered that will always be found when the route pattern is matched during a request. To do so:

  • A special route-specific interface is created at startup time for each route configuration declaration.
  • When an add_view statement mentions a route name attribute, a view configuration is registered at startup time. This view configuration uses a route-specific interface as a request type.
  • At runtime, when a request causes any route to match, the request object is decorated with the route-specific interface.
  • The fact that the request is decorated with a route-specific interface causes the view lookup machinery to always use the view callable registered using that interface by the route configuration to service requests that match the route pattern.

As we can see from the above description, technically, URL dispatch doesn’t actually map a URL pattern directly to a view callable. Instead, URL dispatch is a resource location mechanism. A Pyramid resource location subsystem (i.e., URL dispatch or traversal) finds a resource object that is the context of a request. Once the context is determined, a separate subsystem named view lookup is then responsible for finding and invoking a view callable based on information available in the context and the request. When URL dispatch is used, the resource location and view lookup subsystems provided by Pyramid are still being utilized, but in a way which does not require a developer to understand either of them in detail.

If no route is matched using URL dispatch, Pyramid falls back to traversal to handle the request.

References

A tutorial showing how URL dispatch can be used to create a Pyramid application exists in SQLAlchemy + URL Dispatch Wiki Tutorial.

Views

One of the primary jobs of Pyramid is to find and invoke a view callable when a request reaches your application. View callables are bits of code which do something interesting in response to a request made to your application. They are the “meat” of any interesting web application.

Note

A Pyramid view callable is often referred to in conversational shorthand as a view. In this documentation, however, we need to use less ambiguous terminology because there are significant differences between view configuration, the code that implements a view callable, and the process of view lookup.

This chapter describes how view callables should be defined. We’ll have to wait until a following chapter (entitled View Configuration) to find out how we actually tell Pyramid to wire up view callables to particular URL patterns and other request circumstances.

View Callables

View callables are, at the risk of sounding obvious, callable Python objects. Specifically, view callables can be functions, classes, or instances that implement an __call__ method (making the instance callable).

View callables must, at a minimum, accept a single argument named request. This argument represents a Pyramid Request object. A request object represents a WSGI environment provided to Pyramid by the upstream WSGI server. As you might expect, the request object contains everything your application needs to know about the specific HTTP request being made.

A view callable’s ultimate responsibility is to create a Pyramid Response object. This can be done by creating a Response object in the view callable code and returning it directly or by raising special kinds of exceptions from within the body of a view callable.

Defining a View Callable as a Function

One of the easiest way to define a view callable is to create a function that accepts a single argument named request, and which returns a Response object. For example, this is a “hello world” view callable implemented as a function:

1
2
3
4
from pyramid.response import Response

def hello_world(request):
    return Response('Hello world!')

Defining a View Callable as a Class

A view callable may also be represented by a Python class instead of a function. When a view callable is a class, the calling semantics are slightly different than when it is a function or another non-class callable. When a view callable is a class, the class’ __init__ method is called with a request parameter. As a result, an instance of the class is created. Subsequently, that instance’s __call__ method is invoked with no parameters. Views defined as classes must have the following traits:

  • an __init__ method that accepts a request argument.
  • a __call__ (or other) method that accepts no parameters and which returns a response.

For example:

1
2
3
4
5
6
7
8
from pyramid.response import Response

class MyView(object):
    def __init__(self, request):
        self.request = request

    def __call__(self):
        return Response('hello')

The request object passed to __init__ is the same type of request object described in Defining a View Callable as a Function.

If you’d like to use a different attribute than __call__ to represent the method expected to return a response, you can use an attr value as part of the configuration for the view. See View Configuration Parameters. The same view callable class can be used in different view configuration statements with different attr values, each pointing at a different method of the class if you’d like the class to represent a collection of related view callables.

View Callable Responses

A view callable may return an object that implements the Pyramid Response interface. The easiest way to return something that implements the Response interface is to return a pyramid.response.Response object instance directly. For example:

1
2
3
4
from pyramid.response import Response

def view(request):
    return Response('OK')

Pyramid provides a range of different “exception” classes which inherit from pyramid.response.Response. For example, an instance of the class pyramid.httpexceptions.HTTPFound is also a valid response object because it inherits from Response. For examples, see HTTP Exceptions and Using a View Callable to Do an HTTP Redirect.

Note

You can also return objects from view callables that aren’t instances of pyramid.response.Response in various circumstances. This can be helpful when writing tests and when attempting to share code between view callables. See Renderers for the common way to allow for this. A much less common way to allow for view callables to return non-Response objects is documented in Changing How Pyramid Treats View Responses.

Using Special Exceptions In View Callables

Usually when a Python exception is raised within a view callable, Pyramid allows the exception to propagate all the way out to the WSGI server which invoked the application. It is usually caught and logged there.

However, for convenience, a special set of exceptions exists. When one of these exceptions is raised within a view callable, it will always cause Pyramid to generate a response. These are known as HTTP exception objects.

HTTP Exceptions

All classes documented in the pyramid.httpexceptions module documented as inheriting from the pryamid.httpexceptions.HTTPException are http exception objects. An instances of an HTTP exception object may either be returned or raised from within view code. In either case (return or raise) the instance will be used as as the view’s response.

For example, the pyramid.httpexceptions.HTTPUnauthorized exception can be raised. This will cause a response to be generated with a 401 Unauthorized status:

1
2
3
4
from pyramid.httpexceptions import HTTPUnauthorized

def aview(request):
    raise HTTPUnauthorized()

An HTTP exception, instead of being raised, can alternately be returned (HTTP exceptions are also valid response objects):

1
2
3
4
from pyramid.httpexceptions import HTTPUnauthorized

def aview(request):
    return HTTPUnauthorized()

A shortcut for creating an HTTP exception is the pyramid.httpexceptions.exception_response() function. This function accepts an HTTP status code and returns the corresponding HTTP exception. For example, instead of importing and constructing a HTTPUnauthorized response object, you can use the exception_response() function to construct and return the same object.

1
2
3
4
from pyramid.httpexceptions import exception_response

def aview(request):
    raise exception_response(401)

This is the case because 401 is the HTTP status code for “HTTP Unauthorized”. Therefore, raise exception_response(401) is functionally equivalent to raise HTTPUnauthorized(). Documentation which maps each HTTP response code to its purpose and its associated HTTP exception object is provided within pyramid.httpexceptions.

Note

The exception_response() function is new as of Pyramid 1.1.

How Pyramid Uses HTTP Exceptions

HTTP exceptions are meant to be used directly by application application developers. However, Pyramid itself will raise two HTTP exceptions at various points during normal operations: pyramid.httpexceptions.HTTPNotFound and pyramid.httpexceptions.HTTPForbidden. Pyramid will raise the HTTPNotFound exception are raised when it cannot find a view to service a request. Pyramid will raise the Forbidden exception or when authorization was forbidden by a security policy.

If HTTPNotFound is raised by Pyramid itself or within view code, the result of the Not Found View will be returned to the user agent which performed the request.

If HTTPForbidden is raised by Pyramid itself within view code, the result of the Forbidden View will be returned to the user agent which performed the request.

Custom Exception Views

The machinery which allows HTTP exceptions to be raised and caught by specialized views as described in Using Special Exceptions In View Callables can also be used by application developers to convert arbitrary exceptions to responses.

To register a view that should be called whenever a particular exception is raised from with Pyramid view code, use the exception class or one of its superclasses as the context of a view configuration which points at a view callable you’d like to generate a response.

For example, given the following exception class in a module named helloworld.exceptions:

1
2
3
class ValidationFailure(Exception):
    def __init__(self, msg):
        self.msg = msg

You can wire a view callable to be called whenever any of your other code raises a helloworld.exceptions.ValidationFailure exception:

1
2
3
4
5
6
7
8
from pyramid.view import view_config
from helloworld.exceptions import ValidationFailure

@view_config(context=ValidationFailure)
def failed_validation(exc, request):
    response =  Response('Failed validation: %s' % exc.msg)
    response.status_int = 500
    return response

Assuming that a scan was run to pick up this view registration, this view callable will be invoked whenever a helloworld.exceptions.ValidationFailure is raised by your application’s view code. The same exception raised by a custom root factory, a custom traverser, or a custom view or route predicate is also caught and hooked.

Other normal view predicates can also be used in combination with an exception view registration:

1
2
3
4
5
6
7
8
from pyramid.view import view_config
from helloworld.exceptions import ValidationFailure

@view_config(context=ValidationFailure, route_name='home')
def failed_validation(exc, request):
    response =  Response('Failed validation: %s' % exc.msg)
    response.status_int = 500
    return response

The above exception view names the route_name of home, meaning that it will only be called when the route matched has a name of home. You can therefore have more than one exception view for any given exception in the system: the “most specific” one will be called when the set of request circumstances match the view registration.

The only view predicate that cannot be used successfully when creating an exception view configuration is name. The name used to look up an exception view is always the empty string. Views registered as exception views which have a name will be ignored.

Note

Normal (i.e., non-exception) views registered against a context resource type which inherits from Exception will work normally. When an exception view configuration is processed, two views are registered. One as a “normal” view, the other as an “exception” view. This means that you can use an exception as context for a normal view.

Exception views can be configured with any view registration mechanism: @view_config decorator or imperative add_view styles.

Using a View Callable to Do an HTTP Redirect

You can issue an HTTP redirect by using the pyramid.httpexceptions.HTTPFound class. Raising or returning an instance of this class will cause the client to receive a “302 Found” response.

To do so, you can return a pyramid.httpexceptions.HTTPFound instance.

1
2
3
4
from pyramid.httpexceptions import HTTPFound

def myview(request):
    return HTTPFound(location='http://example.com')

Alternately, you can raise an HTTPFound exception instead of returning one.

1
2
3
4
from pyramid.httpexceptions import HTTPFound

def myview(request):
    raise HTTPFound(location='http://example.com')

When the instance is raised, it is caught by the default exception response handler and turned into a response.

Handling Form Submissions in View Callables (Unicode and Character Set Issues)

Most web applications need to accept form submissions from web browsers and various other clients. In Pyramid, form submission handling logic is always part of a view. For a general overview of how to handle form submission data using the WebOb API, see Request and Response Objects and “Query and POST variables” within the WebOb documentation. Pyramid defers to WebOb for its request and response implementations, and handling form submission data is a property of the request implementation. Understanding WebOb’s request API is the key to understanding how to process form submission data.

There are some defaults that you need to be aware of when trying to handle form submission data in a Pyramid view. Having high-order (i.e., non-ASCII) characters in data contained within form submissions is exceedingly common, and the UTF-8 encoding is the most common encoding used on the web for character data. Since Unicode values are much saner than working with and storing bytestrings, Pyramid configures the WebOb request machinery to attempt to decode form submission values into Unicode from UTF-8 implicitly. This implicit decoding happens when view code obtains form field values via the request.params, request.GET, or request.POST APIs (see pyramid.request for details about these APIs).

Note

Many people find the difference between Unicode and UTF-8 confusing. Unicode is a standard for representing text that supports most of the world’s writing systems. However, there are many ways that Unicode data can be encoded into bytes for transit and storage. UTF-8 is a specific encoding for Unicode, that is backwards-compatible with ASCII. This makes UTF-8 very convenient for encoding data where a large subset of that data is ASCII characters, which is largely true on the web. UTF-8 is also the standard character encoding for URLs.

As an example, let’s assume that the following form page is served up to a browser client, and its action points at some Pyramid view code:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
<html xmlns="http://www.w3.org/1999/xhtml">
  <head>
    <meta http-equiv="Content-Type" content="text/html; charset=UTF-8"/>
  </head>
  <form method="POST" action="myview">
    <div>
      <input type="text" name="firstname"/>
    </div>
    <div>
      <input type="text" name="lastname"/>
    </div>
    <input type="submit" value="Submit"/>
  </form>
</html>

The myview view code in the Pyramid application must expect that the values returned by request.params will be of type unicode, as opposed to type str. The following will work to accept a form post from the above form:

1
2
3
def myview(request):
    firstname = request.params['firstname']
    lastname = request.params['lastname']

But the following myview view code may not work, as it tries to decode already-decoded (unicode) values obtained from request.params:

1
2
3
4
5
def myview(request):
    # the .decode('utf-8') will break below if there are any high-order
    # characters in the firstname or lastname
    firstname = request.params['firstname'].decode('utf-8')
    lastname = request.params['lastname'].decode('utf-8')

For implicit decoding to work reliably, you should ensure that every form you render that posts to a Pyramid view explicitly defines a charset encoding of UTF-8. This can be done via a response that has a ;charset=UTF-8 in its Content-Type header; or, as in the form above, with a meta http-equiv tag that implies that the charset is UTF-8 within the HTML head of the page containing the form. This must be done explicitly because all known browser clients assume that they should encode form data in the same character set implied by Content-Type value of the response containing the form when subsequently submitting that form. There is no other generally accepted way to tell browser clients which charset to use to encode form data. If you do not specify an encoding explicitly, the browser client will choose to encode form data in its default character set before submitting it, which may not be UTF-8 as the server expects. If a request containing form data encoded in a non-UTF8 charset is handled by your view code, eventually the request code accessed within your view will throw an error when it can’t decode some high-order character encoded in another character set within form data, e.g., when request.params['somename'] is accessed.

If you are using the Response class to generate a response, or if you use the render_template_* templating APIs, the UTF-8 charset is set automatically as the default via the Content-Type header. If you return a Content-Type header without an explicit charset, a request will add a ;charset=utf-8 trailer to the Content-Type header value for you, for response content types that are textual (e.g. text/html, application/xml, etc) as it is rendered. If you are using your own response object, you will need to ensure you do this yourself.

Note

Only the values of request params obtained via request.params, request.GET or request.POST are decoded to Unicode objects implicitly in the Pyramid default configuration. The keys are still (byte) strings.

Alternate View Callable Argument/Calling Conventions

Usually, view callables are defined to accept only a single argument: request. However, view callables may alternately be defined as classes, functions, or any callable that accept two positional arguments: a context resource as the first argument and a request as the second argument.

The context and request arguments passed to a view function defined in this style can be defined as follows:

context

The resource object found via tree traversal or URL dispatch.
request
A Pyramid Request object representing the current WSGI request.

The following types work as view callables in this style:

  1. Functions that accept two arguments: context, and request, e.g.:

    1
2
3
4
    from pyramid.response import Response

def view(context, request):
        return Response('OK')

  2. Classes that have an __init__ method that accepts context, request and a __call__ method which accepts no arguments, e.g.:

    1
2
3
4
5
6
7
8
9
    from pyramid.response import Response

class view(object):
        def __init__(self, context, request):
                self.context = context
                self.request = request

        def __call__(self):
                return Response('OK')

  3. Arbitrary callables that have a __call__ method that accepts context, request, e.g.:

    1
2
3
4
5
6
    from pyramid.response import Response

class View(object):
        def __call__(self, context, request):
                return Response('OK')
view = View() # this is the view callable

This style of calling convention is most useful for traversal based applications, where the context object is frequently used within the view callable code itself.

No matter which view calling convention is used, the view code always has access to the context via request.context.

Pylons-1.0-Style “Controller” Dispatch

A package named pyramid_handlers (available from PyPI) provides an analogue of Pylons -style “controllers”, which are a special kind of view class which provides more automation when your application uses URL dispatch solely.

Renderers

A view callable needn’t always return a Response object. If a view happens to return something which does not implement the Pyramid Response interface, Pyramid will attempt to use a renderer to construct a response. For example:

1
2
3
4
5
from pyramid.view import view_config

@view_config(renderer='json')
def hello_world(request):
    return {'content':'Hello!'}

The above example returns a dictionary from the view callable. A dictionary does not implement the Pyramid response interface, so you might believe that this example would fail. However, since a renderer is associated with the view callable through its view configuration (in this case, using a renderer argument passed to view_config()), if the view does not return a Response object, the renderer will attempt to convert the result of the view to a response on the developer’s behalf.

Of course, if no renderer is associated with a view’s configuration, returning anything except an object which implements the Response interface will result in an error. And, if a renderer is used, whatever is returned by the view must be compatible with the particular kind of renderer used, or an error may occur during view invocation.

One exception exists: it is always OK to return a Response object, even when a renderer is configured. If a view callable returns a response object from a view that is configured with a renderer, the renderer is bypassed entirely.

Various types of renderers exist, including serialization renderers and renderers which use templating systems. See also Writing View Callables Which Use a Renderer.

Writing View Callables Which Use a Renderer

As we’ve seen, view callables needn’t always return a Response object. Instead, they may return an arbitrary Python object, with the expectation that a renderer will convert that object into a response instance on your behalf. Some renderers use a templating system; other renderers use object serialization techniques.

View configuration can vary the renderer associated with a view callable via the renderer attribute. For example, this call to add_view() associates the json renderer with a view callable:

1
config.add_view('myproject.views.my_view', renderer='json')

When this configuration is added to an application, the myproject.views.my_view view callable will now use a json renderer, which renders view return values to a JSON response serialization.

Other built-in renderers include renderers which use the Chameleon templating language to render a dictionary to a response. Additional renderers can be added by developers to the system as necessary (see Adding and Changing Renderers).

Views which use a renderer and return a non-Response value can vary non-body response attributes (such as headers and the HTTP status code) by attaching a property to the request.response attribute See Varying Attributes of Rendered Responses.

If the view callable associated with a view configuration returns a Response object directly, any renderer associated with the view configuration is ignored, and the response is passed back to Pyramid unchanged. For example, if your view callable returns an instance of the pyramid.response.Response class as a response, no renderer will be employed.

1
2
3
4
5
6
from pyramid.response import Response
from pyramid.view import view_config

@view_config(renderer='json')
def view(request):
    return Response('OK') # json renderer avoided

Likewise for an HTTP exception response:

1
2
3
4
5
6
from pyramid.httpexceptions import HTTPNotFound
from pyramid.view import view_config

@view_config(renderer='json')
def view(request):
    return HTTPFound(location='http://example.com') # json renderer avoided

You can of course also return the request.response attribute instead to avoid rendering:

1
2
3
4
5
6
from pyramid.view import view_config

@view_config(renderer='json')
def view(request):
    request.response.body = 'OK'
    return request.response # json renderer avoided

Built-In Renderers

Several built-in renderers exist in Pyramid. These renderers can be used in the renderer attribute of view configurations.

string: String Renderer

The string renderer is a renderer which renders a view callable result to a string. If a view callable returns a non-Response object, and the string renderer is associated in that view’s configuration, the result will be to run the object through the Python str function to generate a string. Note that if a Unicode object is returned by the view callable, it is not str() -ified.

Here’s an example of a view that returns a dictionary. If the string renderer is specified in the configuration for this view, the view will render the returned dictionary to the str() representation of the dictionary:

1
2
3
4
5
6
from pyramid.response import Response
from pyramid.view import view_config

@view_config(renderer='string')
def hello_world(request):
    return {'content':'Hello!'}

The body of the response returned by such a view will be a string representing the str() serialization of the return value:

1
{'content': 'Hello!'}

Views which use the string renderer can vary non-body response attributes by using the API of the request.response attribute. See Varying Attributes of Rendered Responses.

json: JSON Renderer

The json renderer renders view callable results to JSON. It passes the return value through the json.dumps standard library function, and wraps the result in a response object. It also sets the response content-type to application/json.

Here’s an example of a view that returns a dictionary. Since the json renderer is specified in the configuration for this view, the view will render the returned dictionary to a JSON serialization:

1
2
3
4
5
6
from pyramid.response import Response
from pyramid.view import view_config

@view_config(renderer='json')
def hello_world(request):
    return {'content':'Hello!'}

The body of the response returned by such a view will be a string representing the JSON serialization of the return value:

1
'{"content": "Hello!"}'

The return value needn’t be a dictionary, but the return value must contain values serializable by json.dumps().

You can configure a view to use the JSON renderer by naming json as the renderer argument of a view configuration, e.g. by using add_view():

1
2
3
4
config.add_view('myproject.views.hello_world',
                 name='hello',
                 context='myproject.resources.Hello',
                 renderer='json')

Views which use the JSON renderer can vary non-body response attributes by using the api of the request.response attribute. See Varying Attributes of Rendered Responses.

JSONP Renderer

Note

This feature is new in Pyramid 1.1.

pyramid.renderers.JSONP is a JSONP renderer factory helper which implements a hybrid json/jsonp renderer. JSONP is useful for making cross-domain AJAX requests.

Unlike other renderers, a JSONP renderer needs to be configured at startup time “by hand”. Configure a JSONP renderer using the pyramid.config.Configurator.add_renderer() method:

from pyramid.config import Configurator

config = Configurator()
config.add_renderer('jsonp', JSONP(param_name='callback'))

Once this renderer is registered via add_renderer() as above, you can use jsonp as the renderer= parameter to @view_config or pyramid.config.Configurator.add_view`():

from pyramid.view import view_config

@view_config(renderer='jsonp')
def myview(request):
    return {'greeting':'Hello world'}

When a view is called that uses a JSONP renderer:

  • If there is a parameter in the request’s HTTP query string (aka request.GET) that matches the param_name of the registered JSONP renderer (by default, callback), the renderer will return a JSONP response.
  • If there is no callback parameter in the request’s query string, the renderer will return a ‘plain’ JSON response.

Javscript library AJAX functionality will help you make JSONP requests. For example, JQuery has a getJSON function, and has equivalent (but more complicated) functionality in its ajax function.

For example (Javascript):

var api_url = 'http://api.geonames.org/timezoneJSON' +
              '?lat=38.301733840000004' +
              '&lng=-77.45869621' +
              '&username=fred' +
              '&callback=?';
jqhxr = $.getJSON(api_url);

The string callback=? above in the the url param to the JQuery getAjax function indicates to jQuery that the query should be made as a JSONP request; the callback parameter will be automatically filled in for you and used.

*.pt or *.txt: Chameleon Template Renderers

Two built-in renderers exist for Chameleon templates.

If the renderer attribute of a view configuration is an absolute path, a relative path or asset specification which has a final path element with a filename extension of .pt, the Chameleon ZPT renderer is used. See Chameleon ZPT Templates for more information about ZPT templates.

If the renderer attribute of a view configuration is an absolute path or a asset specification which has a final path element with a filename extension of .txt, the Chameleon text renderer is used. See Templating with Chameleon Text Templates for more information about Chameleon text templates.

The behavior of these renderers is the same, except for the engine used to render the template.

When a renderer attribute that names a template path or asset specification (e.g. myproject:templates/foo.pt or myproject:templates/foo.txt) is used, the view must return a Response object or a Python dictionary. If the view callable with an associated template returns a Python dictionary, the named template will be passed the dictionary as its keyword arguments, and the template renderer implementation will return the resulting rendered template in a response to the user. If the view callable returns anything but a Response object or a dictionary, an error will be raised.

Before passing keywords to the template, the keyword arguments derived from the dictionary returned by the view are augmented. The callable object – whatever object was used to define the view – will be automatically inserted into the set of keyword arguments passed to the template as the view keyword. If the view callable was a class, the view keyword will be an instance of that class. Also inserted into the keywords passed to the template are renderer_name (the string used in the renderer attribute of the directive), renderer_info (an object containing renderer-related information), context (the context resource of the view used to render the template), and request (the request passed to the view used to render the template).

Here’s an example view configuration which uses a Chameleon ZPT renderer:

1
2
3
4
5
6
 # config is an instance of pyramid.config.Configurator

 config.add_view('myproject.views.hello_world',
                 name='hello',
                 context='myproject.resources.Hello',
                 renderer='myproject:templates/foo.pt')

Here’s an example view configuration which uses a Chameleon text renderer:

1
2
3
4
 config.add_view('myproject.views.hello_world',
                 name='hello',
                 context='myproject.resources.Hello',
                 renderer='myproject:templates/foo.txt')

Views which use a Chameleon renderer can vary response attributes by using the API of the request.response attribute. See Varying Attributes of Rendered Responses.

*.mak or *.mako: Mako Template Renderer

The Mako template renderer renders views using a Mako template. When used, the view must return a Response object or a Python dictionary. The dictionary items will then be used in the global template space. If the view callable returns anything but a Response object or a dictionary, an error will be raised.

When using a renderer argument to a view configuration to specify a Mako template, the value of the renderer may be a path relative to the mako.directories setting (e.g. some/template.mak) or, alternately, it may be a asset specification (e.g. apackage:templates/sometemplate.mak). Mako templates may internally inherit other Mako templates using a relative filename or a asset specification as desired.

Here’s an example view configuration which uses a relative path:

1
2
3
4
5
6
 # config is an instance of pyramid.config.Configurator

 config.add_view('myproject.views.hello_world',
                 name='hello',
                 context='myproject.resources.Hello',
                 renderer='foo.mak')

It’s important to note that in Mako’s case, the ‘relative’ path name foo.mak above is not relative to the package, but is relative to the directory (or directories) configured for Mako via the mako.directories configuration file setting.

The renderer can also be provided in asset specification format. Here’s an example view configuration which uses one:

1
2
3
4
 config.add_view('myproject.views.hello_world',
                 name='hello',
                 context='myproject.resources.Hello',
                 renderer='mypackage:templates/foo.mak')

The above configuration will use the file named foo.mak in the templates directory of the mypackage package.

The Mako template renderer can take additional arguments beyond the standard pyramid.reload_templates setting, see the Environment Variables and .ini File Settings for additional Mako Template Render Settings.

Varying Attributes of Rendered Responses

Before a response constructed by a renderer is returned to Pyramid, several attributes of the request are examined which have the potential to influence response behavior.

View callables that don’t directly return a response should use the API of the pyramid.response.Response attribute available as request.response during their execution, to influence associated response behavior.

For example, if you need to change the response status from within a view callable that uses a renderer, assign the status attribute to the response attribute of the request before returning a result:

1
2
3
4
5
6
from pyramid.view import view_config

@view_config(name='gone', renderer='templates/gone.pt')
def myview(request):
    request.response.status = '404 Not Found'
    return {'URL':request.URL}

Note that mutations of request.response in views which return a Response object directly will have no effect unless the response object returned is request.response. For example, the following example calls request.response.set_cookie, but this call will have no effect, because a different Response object is returned.

1
2
3
4
5
from pyramid.response import Response

def view(request):
    request.response.set_cookie('abc', '123') # this has no effect
    return Response('OK') # because we're returning a different response

If you mutate request.response and you’d like the mutations to have an effect, you must return request.response:

1
2
3
def view(request):
    request.response.set_cookie('abc', '123')
    return request.response

For more information on attributes of the request, see the API documentation in pyramid.request. For more information on the API of request.response, see pyramid.request.Request.response.

Deprecated Mechanism to Vary Attributes of Rendered Responses

Warning

This section describes behavior deprecated in Pyramid 1.1.

In previous releases of Pyramid (1.0 and before), the request.response attribute did not exist. Instead, Pyramid required users to set special response_ -prefixed attributes of the request to influence response behavior. As of Pyramid 1.1, those request attributes are deprecated and their use will cause a deprecation warning to be issued when used. Until their existence is removed completely, we document them below, for benefit of people with older code bases.

response_content_type
Defines the content-type of the resulting response, e.g. text/xml.
response_headerlist
A sequence of tuples describing header values that should be set in the response, e.g. [('Set-Cookie', 'abc=123'), ('X-My-Header', 'foo')].
response_status
A WSGI-style status code (e.g. 200 OK) describing the status of the response.
response_charset
The character set (e.g. UTF-8) of the response.
response_cache_for
A value in seconds which will influence Cache-Control and Expires headers in the returned response. The same can also be achieved by returning various values in the response_headerlist, this is purely a convenience.

Adding and Changing Renderers

New templating systems and serializers can be associated with Pyramid renderer names. To this end, configuration declarations can be made which change an existing renderer factory, and which add a new renderer factory.

Renderers can be registered imperatively using the pyramid.config.Configurator.add_renderer() API.

For example, to add a renderer which renders views which have a renderer attribute that is a path that ends in .jinja2:

1
config.add_renderer('.jinja2', 'mypackage.MyJinja2Renderer')

The first argument is the renderer name. The second argument is a reference to an implementation of a renderer factory or a dotted Python name referring to such an object.

Adding a New Renderer

You may add a new renderer by creating and registering a renderer factory.

A renderer factory implementation is typically a class with the following interface:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
class RendererFactory:
    def __init__(self, info):
        """ Constructor: info will be an object having the
        following attributes: name (the renderer name), package
        (the package that was 'current' at the time the
        renderer was registered), type (the renderer type
        name), registry (the current application registry) and
        settings (the deployment settings dictionary). """

    def __call__(self, value, system):
        """ Call the renderer implementation with the value
        and the system value passed in as arguments and return
        the result (a string or unicode object).  The value is
        the return value of a view.  The system value is a
        dictionary containing available system values
        (e.g. view, context, and request). """

The formal interface definition of the info object passed to a renderer factory constructor is available as pyramid.interfaces.IRendererInfo.

There are essentially two different kinds of renderer factories:

  • A renderer factory which expects to accept an asset specification, or an absolute path, as the name attribute of the info object fed to its constructor. These renderer factories are registered with a name value that begins with a dot (.). These types of renderer factories usually relate to a file on the filesystem, such as a template.
  • A renderer factory which expects to accept a token that does not represent a filesystem path or an asset specification in the name attribute of the info object fed to its constructor. These renderer factories are registered with a name value that does not begin with a dot. These renderer factories are typically object serializers.

Asset Specifications

An asset specification is a colon-delimited identifier for an asset. The colon separates a Python package name from a package subpath. For example, the asset specification my.package:static/baz.css identifies the file named baz.css in the static subdirectory of the my.package Python package.

Here’s an example of the registration of a simple renderer factory via add_renderer():

1
2
3
# config is an instance of pyramid.config.Configurator

config.add_renderer(name='amf', factory='my.package.MyAMFRenderer')

Adding the above code to your application startup configuration will allow you to use the my.package.MyAMFRenderer renderer factory implementation in view configurations. Your application can use this renderer by specifying amf in the renderer attribute of a view configuration:

1
2
3
4
5
from pyramid.view import view_config

@view_config(renderer='amf')
def myview(request):
    return {'Hello':'world'}

At startup time, when a view configuration is encountered, which has a name attribute that does not contain a dot, the full name value is used to construct a renderer from the associated renderer factory. In this case, the view configuration will create an instance of an MyAMFRenderer for each view configuration which includes amf as its renderer value. The name passed to the MyAMFRenderer constructor will always be amf.

Here’s an example of the registration of a more complicated renderer factory, which expects to be passed a filesystem path:

1
2
config.add_renderer(name='.jinja2',
                    factory='my.package.MyJinja2Renderer')

Adding the above code to your application startup will allow you to use the my.package.MyJinja2Renderer renderer factory implementation in view configurations by referring to any renderer which ends in .jinja in the renderer attribute of a view configuration:

1
2
3
4
5
from pyramid.view import view_config

@view_config(renderer='templates/mytemplate.jinja2')
def myview(request):
    return {'Hello':'world'}

When a view configuration is encountered at startup time, which has a name attribute that does contain a dot, the value of the name attribute is split on its final dot. The second element of the split is typically the filename extension. This extension is used to look up a renderer factory for the configured view. Then the value of renderer is passed to the factory to create a renderer for the view. In this case, the view configuration will create an instance of a MyJinja2Renderer for each view configuration which includes anything ending with .jinja2 in its renderer value. The name passed to the MyJinja2Renderer constructor will be the full value that was set as renderer= in the view configuration.

Changing an Existing Renderer

You can associate more than one filename extension with the same existing renderer implementation as necessary if you need to use a different file extension for the same kinds of templates. For example, to associate the .zpt extension with the Chameleon ZPT renderer factory, use the pyramid.config.Configurator.add_renderer() method:

1
config.add_renderer('.zpt', 'pyramid.chameleon_zpt.renderer_factory')

After you do this, Pyramid will treat templates ending in both the .pt and .zpt filename extensions as Chameleon ZPT templates.

To change the default mapping in which files with a .pt extension are rendered via a Chameleon ZPT page template renderer, use a variation on the following in your application’s startup code:

1
config.add_renderer('.pt', 'mypackage.pt_renderer')

After you do this, the renderer factory in mypackage.pt_renderer will be used to render templates which end in .pt, replacing the default Chameleon ZPT renderer.

To associate a default renderer with all view configurations (even ones which do not possess a renderer attribute), pass None as the name attribute to the renderer tag:

1
config.add_renderer(None, 'mypackage.json_renderer_factory')

Overriding A Renderer At Runtime

Warning

This is an advanced feature, not typically used by “civilians”.

In some circumstances, it is necessary to instruct the system to ignore the static renderer declaration provided by the developer in view configuration, replacing the renderer with another after a request starts. For example, an “omnipresent” XML-RPC implementation that detects that the request is from an XML-RPC client might override a view configuration statement made by the user instructing the view to use a template renderer with one that uses an XML-RPC renderer. This renderer would produce an XML-RPC representation of the data returned by an arbitrary view callable.

To use this feature, create a NewRequest subscriber which sniffs at the request data and which conditionally sets an override_renderer attribute on the request itself, which is the name of a registered renderer. For example:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
from pyramid.event import subscriber
from pyramid.event import NewRequest

@subscriber(NewRequest)
def set_xmlrpc_params(event):
    request = event.request
    if (request.content_type == 'text/xml'
            and request.method == 'POST'
            and not 'soapaction' in request.headers
            and not 'x-pyramid-avoid-xmlrpc' in request.headers):
        params, method = parse_xmlrpc_request(request)
        request.xmlrpc_params, request.xmlrpc_method = params, method
        request.is_xmlrpc = True
        request.override_renderer = 'xmlrpc'
        return True

The result of such a subscriber will be to replace any existing static renderer configured by the developer with a (notional, nonexistent) XML-RPC renderer if the request appears to come from an XML-RPC client.

Templates

A template is a file on disk which can be used to render dynamic data provided by a view. Pyramid offers a number of ways to perform templating tasks out of the box, and provides add-on templating support through a set of bindings packages.

Out of the box, Pyramid provides templating via the Chameleon and Mako templating libraries. Chameleon provides support for two different types of templates: ZPT templates, and text templates.

Before discussing how built-in templates are used in detail, we’ll discuss two ways to render templates within Pyramid in general: directly, and via renderer configuration.

Using Templates Directly

The most straightforward way to use a template within Pyramid is to cause it to be rendered directly within a view callable. You may use whatever API is supplied by a given templating engine to do so.

Pyramid provides various APIs that allow you to render templates directly from within a view callable. For example, if there is a Chameleon ZPT template named foo.pt in a directory named templates in your application, you can render the template from within the body of a view callable like so:

1
2
3
4
5
6
from pyramid.renderers import render_to_response

def sample_view(request):
    return render_to_response('templates/foo.pt',
                              {'foo':1, 'bar':2},
                              request=request)

Warning

Earlier iterations of this documentation (pre-version-1.3) encouraged the application developer to use ZPT-specific APIs such as pyramid.chameleon_zpt.render_template_to_response() and pyramid.chameleon_zpt.render_template() to render templates directly. This style of rendering still works, but at least for purposes of this documentation, those functions are deprecated. Application developers are encouraged instead to use the functions available in the pyramid.renderers module to perform rendering tasks. This set of functions works to render templates for all renderer extensions registered with Pyramid.

The sample_view view callable function above returns a response object which contains the body of the templates/foo.pt template. In this case, the templates directory should live in the same directory as the module containing the sample_view function. The template author will have the names foo and bar available as top-level names for replacement or comparison purposes.

In the example above, the path templates/foo.pt is relative to the directory containing the file which defines the view configuration. In this case, this is the directory containing the file that defines the sample_view function. Although a renderer path is usually just a simple relative pathname, a path named as a renderer can be absolute, starting with a slash on UNIX or a drive letter prefix on Windows.

Warning

Only Chameleon templates support defining a renderer for a template relative to the location of the module where the view callable is defined. Mako templates, and other templating system bindings work differently. In particular, Mako templates use a “lookup path” as defined by the mako.directories configuration file instead of treating relative paths as relative to the current view module. See Templating With Mako Templates.

The path can alternately be a asset specification in the form some.dotted.package_name:relative/path. This makes it possible to address template assets which live in another package. For example:

1
2
3
4
5
6
from pyramid.renderers import render_to_response

def sample_view(request):
    return render_to_response('mypackage:templates/foo.pt',
                              {'foo':1, 'bar':2},
                              request=request)

An asset specification points at a file within a Python package. In this case, it points at a file named foo.pt within the templates directory of the mypackage package. Using a asset specification instead of a relative template name is usually a good idea, because calls to render_to_response using asset specifications will continue to work properly if you move the code containing them around.

Note

Mako templating system bindings also respect absolute asset specifications as an argument to any of the render* commands. If a template name defines a : (colon) character and is not an absolute path, it is treated as an absolute asset specification.

In the examples above we pass in a keyword argument named request representing the current Pyramid request. Passing a request keyword argument will cause the render_to_response function to supply the renderer with more correct system values (see System Values Used During Rendering), because most of the information required to compose proper system values is present in the request. If your template relies on the name request or context, or if you’ve configured special renderer globals, make sure to pass request as a keyword argument in every call to to a pyramid.renderers.render_* function.

Every view must return a response object, except for views which use a renderer named via view configuration (which we’ll see shortly). The pyramid.renderers.render_to_response() function is a shortcut function that actually returns a response object. This allows the example view above to simply return the result of its call to render_to_response() directly.

Obviously not all APIs you might call to get response data will return a response object. For example, you might render one or more templates to a string that you want to use as response data. The pyramid.renderers.render() API renders a template to a string. We can manufacture a response object directly, and use that string as the body of the response:

1
2
3
4
5
6
7
8
9
from pyramid.renderers import render
from pyramid.response import Response

def sample_view(request):
    result = render('mypackage:templates/foo.pt',
                    {'foo':1, 'bar':2},
                    request=request)
    response = Response(result)
    return response

Because view callable functions are typically the only code in Pyramid that need to know anything about templates, and because view functions are very simple Python, you can use whatever templating system you’re most comfortable with within Pyramid. Install the templating system, import its API functions into your views module, use those APIs to generate a string, then return that string as the body of a Pyramid Response object.

For example, here’s an example of using “raw” Mako from within a Pyramid view:

1
2
3
4
5
6
7
8
from mako.template import Template
from pyramid.response import Response

def make_view(request):
    template = Template(filename='/templates/template.mak')
    result = template.render(name=request.params['name'])
    response = Response(result)
    return response

You probably wouldn’t use this particular snippet in a project, because it’s easier to use the Mako renderer bindings which already exist in Pyramid. But if your favorite templating system is not supported as a renderer extension for Pyramid, you can create your own simple combination as shown above.

Note

If you use third-party templating languages without cooperating Pyramid bindings directly within view callables, the auto-template-reload strategy explained in Automatically Reloading Templates will not be available, nor will the template asset overriding capability explained in Overriding Assets be available, nor will it be possible to use any template using that language as a renderer. However, it’s reasonably easy to write custom templating system binding packages for use under Pyramid so that templates written in the language can be used as renderers. See Adding and Changing Renderers for instructions on how to create your own template renderer and Available Add-On Template System Bindings for example packages.

If you need more control over the status code and content-type, or other response attributes from views that use direct templating, you may set attributes on the response that influence these values.

Here’s an example of changing the content-type and status of the response object returned by render_to_response():

1
2
3
4
5
6
7
8
9
from pyramid.renderers import render_to_response

def sample_view(request):
    response = render_to_response('templates/foo.pt',
                                  {'foo':1, 'bar':2},
                                  request=request)
    response.content_type = 'text/plain'
    response.status_int = 204
    return response

Here’s an example of manufacturing a response object using the result of render() (a string):

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
from pyramid.renderers import render
from pyramid.response import Response

def sample_view(request):
    result = render('mypackage:templates/foo.pt',
                    {'foo':1, 'bar':2},
                    request=request)
    response = Response(result)
    response.content_type = 'text/plain'
    return response

System Values Used During Rendering

When a template is rendered using render_to_response() or render(), the renderer representing the template will be provided with a number of system values. These values are provided in a dictionary to the renderer and include:

context
The current Pyramid context if request was provided as a keyword argument, or None.
request
The request provided as a keyword argument.
renderer_name
The renderer name used to perform the rendering, e.g. mypackage:templates/foo.pt.
renderer_info
An object implementing the pyramid.interfaces.IRendererInfo interface. Basically, an object with the following attributes: name, package and type.

You can define more values which will be passed to every template executed as a result of rendering by defining renderer globals.

What any particular renderer does with these system values is up to the renderer itself, but most template renderers, including Chameleon and Mako renderers, make these names available as top-level template variables.

Templates Used as Renderers via Configuration

An alternative to using render_to_response() to render templates manually in your view callable code, is to specify the template as a renderer in your view configuration. This can be done with any of the templating languages supported by Pyramid.

To use a renderer via view configuration, specify a template asset specification as the renderer argument, or attribute to the view configuration of a view callable. Then return a dictionary from that view callable. The dictionary items returned by the view callable will be made available to the renderer template as top-level names.

The association of a template as a renderer for a view configuration makes it possible to replace code within a view callable that handles the rendering of a template.

Here’s an example of using a view_config decorator to specify a view configuration that names a template renderer:

1
2
3
4
5
from pyramid.view import view_config

@view_config(renderer='templates/foo.pt')
def my_view(request):
    return {'foo':1, 'bar':2}

Note

You do not need to supply the request value as a key in the dictionary result returned from a renderer-configured view callable. Pyramid automatically supplies this value for you so that the “most correct” system values are provided to the renderer.

Warning

The renderer argument to the @view_config configuration decorator shown above is the template path. In the example above, the path templates/foo.pt is relative. Relative to what, you ask? Because we’re using a Chameleon renderer, it means “relative to the directory in which the file which defines the view configuration lives”. In this case, this is the directory containing the file that defines the my_view function. View-configuration-relative asset specifications work only in Chameleon, not in Mako templates.

Similar renderer configuration can be done imperatively. See Writing View Callables Which Use a Renderer. See also Built-In Renderers.

Although a renderer path is usually just a simple relative pathname, a path named as a renderer can be absolute, starting with a slash on UNIX or a drive letter prefix on Windows. The path can alternately be an asset specification in the form some.dotted.package_name:relative/path, making it possible to address template assets which live in another package.

Not just any template from any arbitrary templating system may be used as a renderer. Bindings must exist specifically for Pyramid to use a templating language template as a renderer. Currently, Pyramid has built-in support for two Chameleon templating languages: ZPT and text, and the Mako templating system. See Built-In Renderers for a discussion of their details. Pyramid also supports the use of Jinja2 templates as renderers. See Available Add-On Template System Bindings.

Why Use A Renderer via View Configuration

Using a renderer in view configuration is usually a better way to render templates than using any rendering API directly from within a view callable because it makes the view callable more unit-testable. Views which use templating or rendering APIs directly must return a Response object. Making testing assertions about response objects is typically an indirect process, because it means that your test code often needs to somehow parse information out of the response body (often HTML). View callables configured with renderers externally via view configuration typically return a dictionary, as above. Making assertions about results returned in a dictionary is almost always more direct and straightforward than needing to parse HTML.

By default, views rendered via a template renderer return a Response object which has a status code of 200 OK, and a content-type of text/html. To vary attributes of the response of a view that uses a renderer, such as the content-type, headers, or status attributes, you must use the API of the pyramid.response.Response object exposed as request.response within the view before returning the dictionary. See Varying Attributes of Rendered Responses for more information.

The same set of system values are provided to templates rendered via a renderer view configuration as those provided to templates rendered imperatively. See System Values Used During Rendering.

Chameleon ZPT Templates

Like Zope, Pyramid uses ZPT (Zope Page Templates) as its default templating language. However, Pyramid uses a different implementation of the ZPT specification than Zope does: the Chameleon templating engine. The Chameleon engine complies largely with the Zope Page Template template specification. However, it is significantly faster.

The language definition documentation for Chameleon ZPT-style templates is available from the Chameleon website.

Warning

Chameleon only works on CPython platforms and Google App Engine. On Jython and other non-CPython platforms, you should use Mako (see Templating With Mako Templates) or pyramid_jinja2 instead. See Available Add-On Template System Bindings.

Given a Chameleon ZPT template named foo.pt in a directory in your application named templates, you can render the template as a renderer like so:

1
2
3
4
5
from pyramid.view import view_config

@view_config(renderer='templates/foo.pt')
def my_view(request):
    return {'foo':1, 'bar':2}

See also Built-In Renderers for more general information about renderers, including Chameleon ZPT renderers.

A Sample ZPT Template

Here’s what a simple Chameleon ZPT template used under Pyramid might look like:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
 <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN"
     "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
 <html xmlns="http://www.w3.org/1999/xhtml"
       xmlns:tal="http://xml.zope.org/namespaces/tal">
 <head>
     <meta http-equiv="content-type" content="text/html; charset=utf-8" />
     <title>${project} Application</title>
 </head>
   <body>
      <h1 class="title">Welcome to <code>${project}</code>, an
       application generated by the <a
       href="http://docs.pylonsproject.org/projects/pyramid/current/"
      >pyramid</a> web
       application framework.</h1>
   </body>
 </html>

Note the use of Genshi -style ${replacements} above. This is one of the ways that Chameleon ZPT differs from standard ZPT. The above template expects to find a project key in the set of keywords passed in to it via render() or render_to_response(). Typical ZPT attribute-based syntax (e.g. tal:content and tal:replace) also works in these templates.

Using ZPT Macros in Pyramid

When a renderer is used to render a template, Pyramid makes at least two top-level names available to the template by default: context and request. One of the common needs in ZPT-based templates is to use one template’s “macros” from within a different template. In Zope, this is typically handled by retrieving the template from the context. But the context in Pyramid is a resource object, and templates cannot usually be retrieved from resources. To use macros in Pyramid, you need to make the macro template itself available to the rendered template by passing the macro template, or even the macro itself, into the rendered template. To do this you can use the pyramid.renderers.get_renderer() API to retrieve the macro template, and pass it into the template being rendered via the dictionary returned by the view. For example, using a view configuration via a view_config decorator that uses a renderer:

1
2
3
4
5
6
7
from pyramid.renderers import get_renderer
from pyramid.view import view_config

@view_config(renderer='templates/mytemplate.pt')
def my_view(request):
    main = get_renderer('templates/master.pt').implementation()
    return {'main':main}

Where templates/master.pt might look like so:

1
2
3
4
5
6
7
8
9
 <html xmlns="http://www.w3.org/1999/xhtml"
       xmlns:tal="http://xml.zope.org/namespaces/tal"
       xmlns:metal="http://xml.zope.org/namespaces/metal">
   <span metal:define-macro="hello">
     <h1>
       Hello <span metal:define-slot="name">Fred</span>!
     </h1>
   </span>
 </html>

And templates/mytemplate.pt might look like so:

1
2
3
4
5
6
7
 <html xmlns="http://www.w3.org/1999/xhtml"
       xmlns:tal="http://xml.zope.org/namespaces/tal"
       xmlns:metal="http://xml.zope.org/namespaces/metal">
   <span metal:use-macro="main.macros['hello']">
     <span metal:fill-slot="name">Chris</span>
   </span>
 </html>

Templating with Chameleon Text Templates

Pyramid also allows for the use of templates which are composed entirely of non-XML text via Chameleon. To do so, you can create templates that are entirely composed of text except for ${name} -style substitution points.

Here’s an example usage of a Chameleon text template. Create a file on disk named mytemplate.txt in your project’s templates directory with the following contents:

Hello, ${name}!

Then in your project’s views.py module, you can create a view which renders this template:

1
2
3
4
5
from pyramid.view import view_config

@view_config(renderer='templates/mytemplate.txt')
def my_view(request):
    return {'name':'world'}

When the template is rendered, it will show:

Hello, world!

If you’d rather use templates directly within a view callable (without the indirection of using a renderer), see pyramid.chameleon_text for the API description.

See also Built-In Renderers for more general information about renderers, including Chameleon text renderers.

Side Effects of Rendering a Chameleon Template

When a Chameleon template is rendered from a file, the templating engine writes a file in the same directory as the template file itself as a kind of cache, in order to do less work the next time the template needs to be read from disk. If you see “strange” .py files showing up in your templates directory (or otherwise directly “next” to your templates), it is due to this feature.

If you’re using a version control system such as Subversion, you should configure it to ignore these files. Here’s the contents of the author’s svn propedit svn:ignore . in each of my templates directories.

*.pt.py
*.txt.py

Note that I always name my Chameleon ZPT template files with a .pt extension and my Chameleon text template files with a .txt extension so that these svn:ignore patterns work.

Nicer Exceptions in Chameleon Templates

The exceptions raised by Chameleon templates when a rendering fails are sometimes less than helpful. Pyramid allows you to configure your application development environment so that exceptions generated by Chameleon during template compilation and execution will contain nicer debugging information.

Warning

Template-debugging behavior is not recommended for production sites as it slows renderings; it’s usually only desirable during development.

In order to turn on template exception debugging, you can use an environment variable setting or a configuration file setting.

To use an environment variable, start your application under a shell using the PYRAMID_DEBUG_TEMPLATES operating system environment variable set to 1, For example:

$ PYRAMID_DEBUG_TEMPLATES=1 bin/paster serve myproject.ini

To use a setting in the application .ini file for the same purpose, set the pyramid.debug_templates key to true within the application’s configuration section, e.g.:

1
2
3
[app:main]
use = egg:MyProject
pyramid.debug_templates = true

With template debugging off, a NameError exception resulting from rendering a template with an undefined variable (e.g. ${wrong}) might end like this:

File "...", in __getitem__
  raise NameError(key)
NameError: wrong

Note that the exception has no information about which template was being rendered when the error occured. But with template debugging on, an exception resulting from the same problem might end like so:

RuntimeError: Caught exception rendering template.
 - Expression: ``wrong``
 - Filename:   /home/fred/env/proj/proj/templates/mytemplate.pt
 - Arguments:  renderer_name: proj:templates/mytemplate.pt
               template: <PageTemplateFile - at 0x1d2ecf0>
               xincludes: <XIncludes - at 0x1d3a130>
               request: <Request - at 0x1d2ecd0>
               project: proj
               macros: <Macros - at 0x1d3aed0>
               context: <MyResource None at 0x1d39130>
               view: <function my_view at 0x1d23570>

NameError: wrong

The latter tells you which template the error occurred in, as well as displaying the arguments passed to the template itself.

Note

Turning on pyramid.debug_templates has the same effect as using the Chameleon environment variable CHAMELEON_DEBUG. See Chameleon Environment Variables for more information.

Chameleon Template Internationalization

See Chameleon Template Support for Translation Strings for information about supporting internationalized units of text within Chameleon templates.

Templating With Mako Templates

Mako is a templating system written by Mike Bayer. Pyramid has built-in bindings for the Mako templating system. The language definition documentation for Mako templates is available from the Mako website.

To use a Mako template, given a Mako template file named foo.mak in the templates subdirectory in your application package named mypackage, you can configure the template as a renderer like so:

1
2
3
4
5
from pyramid.view import view_config

@view_config(renderer='foo.mak')
def my_view(request):
    return {'project':'my project'}

For the above view callable to work, the following setting needs to be present in the application stanza of your configuration’s ini file:

mako.directories = mypackage:templates

This lets the Mako templating system know that it should look for templates in the templates subdirectory of the mypackage Python package. See Mako Template Render Settings for more information about the mako.directories setting and other Mako-related settings that can be placed into the application’s ini file.

A Sample Mako Template

Here’s what a simple Mako template used under Pyramid might look like:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
 <html>
 <head>
     <title>${project} Application</title>
 </head>
   <body>
      <h1 class="title">Welcome to <code>${project}</code>, an
       application generated by the <a
       href="http://docs.pylonsproject.org/projects/pyramid/current/"
      >pyramid</a> web application framework.</h1>
   </body>
 </html>

This template doesn’t use any advanced features of Mako, only the ${} replacement syntax for names that are passed in as renderer globals. See the the Mako documentation to use more advanced features.

Automatically Reloading Templates

It’s often convenient to see changes you make to a template file appear immediately without needing to restart the application process. Pyramid allows you to configure your application development environment so that a change to a template will be automatically detected, and the template will be reloaded on the next rendering.

Warning

Auto-template-reload behavior is not recommended for production sites as it slows rendering slightly; it’s usually only desirable during development.

In order to turn on automatic reloading of templates, you can use an environment variable, or a configuration file setting.

To use an environment variable, start your application under a shell using the PYRAMID_RELOAD_TEMPLATES operating system environment variable set to 1, For example:

$ PYRAMID_RELOAD_TEMPLATES=1 bin/paster serve myproject.ini

To use a setting in the application .ini file for the same purpose, set the pyramid.reload_templates key to true within the application’s configuration section, e.g.:

1
2
3
[app:main]
use = egg:MyProject
pyramid.reload_templates = true

Available Add-On Template System Bindings

Jinja2 template bindings are available for Pyramid in the pyramid_jinja2 package. You can get the latest release of this package from the Python package index (pypi).

View Configuration

View lookup is the Pyramid subsystem responsible for finding and invoking a view callable. View configuration controls how view lookup operates in your application. During any given request, view configuration information is compared against request data by the view lookup subsystem in order to find the “best” view callable for that request.

In earlier chapters, you have been exposed to a few simple view configuration declarations without much explanation. In this chapter we will explore the subject in detail.

Mapping a Resource or URL Pattern to a View Callable

A developer makes a view callable available for use within a Pyramid application via view configuration. A view configuration associates a view callable with a set of statements that determine the set of circumstances which must be true for the view callable to be invoked.

A view configuration statement is made about information present in the context resource and the request.

View configuration is performed in one of two ways:

View Configuration Parameters

All forms of view configuration accept the same general types of arguments.

Many arguments supplied during view configuration are view predicate arguments. View predicate arguments used during view configuration are used to narrow the set of circumstances in which view lookup will find a particular view callable.

View predicate attributes are an important part of view configuration that enables the view lookup subsystem to find and invoke the appropriate view. The greater number of predicate attributes possessed by a view’s configuration, the more specific the circumstances need to be before the registered view callable will be invoked. The fewer number of predicates which are supplied to a particular view configuration, the more likely it is that the associated view callable will be invoked. A view with five predicates will always be found and evaluated before a view with two, for example. All predicates must match for the associated view to be called.

This does not mean however, that Pyramid “stops looking” when it finds a view registration with predicates that don’t match. If one set of view predicates does not match, the “next most specific” view (if any) is consulted for predicates, and so on, until a view is found, or no view can be matched up with the request. The first view with a set of predicates all of which match the request environment will be invoked.

If no view can be found with predicates which allow it to be matched up with the request, Pyramid will return an error to the user’s browser, representing a “not found” (404) page. See Changing the Not Found View for more information about changing the default notfound view.

Other view configuration arguments are non-predicate arguments. These tend to modify the response of the view callable or prevent the view callable from being invoked due to an authorization policy. The presence of non-predicate arguments in a view configuration does not narrow the circumstances in which the view callable will be invoked.

Non-Predicate Arguments
permission

The name of a permission that the user must possess in order to invoke the view callable. See Configuring View Security for more information about view security and permissions.

If permission is not supplied, no permission is registered for this view (it’s accessible by any caller).

attr

The view machinery defaults to using the __call__ method of the view callable (or the function itself, if the view callable is a function) to obtain a response. The attr value allows you to vary the method attribute used to obtain the response. For example, if your view was a class, and the class has a method named index and you wanted to use this method instead of the class’ __call__ method to return the response, you’d say attr="index" in the view configuration for the view. This is most useful when the view definition is a class.

If attr is not supplied, None is used (implying the function itself if the view is a function, or the __call__ callable attribute if the view is a class).

renderer

Denotes the renderer implementation which will be used to construct a response from the associated view callable’s return value. (see also Renderers).

This is either a single string term (e.g. json) or a string implying a path or asset specification (e.g. templates/views.pt) naming a renderer implementation. If the renderer value does not contain a dot (.), the specified string will be used to look up a renderer implementation, and that renderer implementation will be used to construct a response from the view return value. If the renderer value contains a dot (.), the specified term will be treated as a path, and the filename extension of the last element in the path will be used to look up the renderer implementation, which will be passed the full path.

When the renderer is a path, although a path is usually just a simple relative pathname (e.g. templates/foo.pt, implying that a template named “foo.pt” is in the “templates” directory relative to the directory of the current package), a path can be absolute, starting with a slash on UNIX or a drive letter prefix on Windows. The path can alternately be a asset specification in the form some.dotted.package_name:relative/path, making it possible to address template assets which live in a separate package.

The renderer attribute is optional. If it is not defined, the “null” renderer is assumed (no rendering is performed and the value is passed back to the upstream Pyramid machinery unchanged). Note that if the view callable itself returns a response (see View Callable Responses), the specified renderer implementation is never called.

http_cache

When you supply an http_cache value to a view configuration, the Expires and Cache-Control headers of a response generated by the associated view callable are modified. The value for http_cache may be one of the following:

  • A nonzero integer. If it’s a nonzero integer, it’s treated as a number of seconds. This number of seconds will be used to compute the Expires header and the Cache-Control: max-age parameter of responses to requests which call this view. For example: http_cache=3600 instructs the requesting browser to ‘cache this response for an hour, please’.
  • A datetime.timedelta instance. If it’s a datetime.timedelta instance, it will be converted into a number of seconds, and that number of seconds will be used to compute the Expires header and the Cache-Control: max-age parameter of responses to requests which call this view. For example: http_cache=datetime.timedelta(days=1) instructs the requesting browser to ‘cache this response for a day, please’.
  • Zero (0). If the value is zero, the Cache-Control and Expires headers present in all responses from this view will be composed such that client browser cache (and any intermediate caches) are instructed to never cache the response.
  • A two-tuple. If it’s a two tuple (e.g. http_cache=(1, {'public':True})), the first value in the tuple may be a nonzero integer or a datetime.timedelta instance; in either case this value will be used as the number of seconds to cache the response. The second value in the tuple must be a dictionary. The values present in the dictionary will be used as input to the Cache-Control response header. For example: http_cache=(3600, {'public':True}) means ‘cache for an hour, and add public to the Cache-Control header of the response’. All keys and values supported by the webob.cachecontrol.CacheControl interface may be added to the dictionary. Supplying {'public':True} is equivalent to calling response.cache_control.public = True.

Providing a non-tuple value as http_cache is equivalent to calling response.cache_expires(value) within your view’s body.

Providing a two-tuple value as http_cache is equivalent to calling response.cache_expires(value[0], **value[1]) within your view’s body.

If you wish to avoid influencing, the Expires header, and instead wish to only influence Cache-Control headers, pass a tuple as http_cache with the first element of None, e.g.: (None, {'public':True}).

wrapper

The view name of a different view configuration which will receive the response body of this view as the request.wrapped_body attribute of its own request, and the response returned by this view as the request.wrapped_response attribute of its own request. Using a wrapper makes it possible to “chain” views together to form a composite response. The response of the outermost wrapper view will be returned to the user. The wrapper view will be found as any view is found: see View Configuration. The “best” wrapper view will be found based on the lookup ordering: “under the hood” this wrapper view is looked up via pyramid.view.render_view_to_response(context, request, 'wrapper_viewname'). The context and request of a wrapper view is the same context and request of the inner view.

If wrapper is not supplied, no wrapper view is used.

decorator
A dotted Python name to a function (or the function itself) which will be used to decorate the registered view callable. The decorator function will be called with the view callable as a single argument. The view callable it is passed will accept (context, request). The decorator must return a replacement view callable which also accepts (context, request).
mapper
A Python object or dotted Python name which refers to a view mapper, or None. By default it is None, which indicates that the view should use the default view mapper. This plug-point is useful for Pyramid extension developers, but it’s not very useful for ‘civilians’ who are just developing stock Pyramid applications. Pay no attention to the man behind the curtain.
Predicate Arguments

These arguments modify view lookup behavior. In general, the more predicate arguments that are supplied, the more specific, and narrower the usage of the configured view.

name

The view name required to match this view callable. A name argument is typically only used when your application uses traversal. Read Traversal to understand the concept of a view name.

If name is not supplied, the empty string is used (implying the default view).

context

An object representing a Python class that the context resource must be an instance of or the interface that the context resource must provide in order for this view to be found and called. This predicate is true when the context resource is an instance of the represented class or if the context resource provides the represented interface; it is otherwise false.

If context is not supplied, the value None, which matches any resource, is used.

route_name

If route_name is supplied, the view callable will be invoked only when the named route has matched.

This value must match the name of a route configuration declaration (see URL Dispatch) that must match before this view will be called. Note that the route configuration referred to by route_name will usually have a *traverse token in the value of its pattern, representing a part of the path that will be used by traversal against the result of the route’s root factory.

If route_name is not supplied, the view callable will only have a chance of being invoked if no other route was matched. This is when the request/context pair found via resource location does not indicate it matched any configured route.

request_type

This value should be an interface that the request must provide in order for this view to be found and called.

If request_type is not supplied, the value None is used, implying any request type.

This is an advanced feature, not often used by “civilians”.

request_method

This value can be one of the strings GET, POST, PUT, DELETE, or HEAD representing an HTTP REQUEST_METHOD. A view declaration with this argument ensures that the view will only be called when the request’s method attribute (aka the REQUEST_METHOD of the WSGI environment) string matches the supplied value.

If request_method is not supplied, the view will be invoked regardless of the REQUEST_METHOD of the WSGI environment.

request_param

This value can be any string. A view declaration with this argument ensures that the view will only be called when the request has a key in the request.params dictionary (an HTTP GET or POST variable) that has a name which matches the supplied value.

If the value supplied has a = sign in it, e.g. request_param="foo=123", then the key (foo) must both exist in the request.params dictionary, and the value must match the right hand side of the expression (123) for the view to “match” the current request.

If request_param is not supplied, the view will be invoked without consideration of keys and values in the request.params dictionary.

match_param

Note

This feature is new as of Pyramid 1.2.

This param may be either a single string of the format “key=value” or a dict of key/value pairs.

This argument ensures that the view will only be called when the request has key/value pairs in its matchdict that equal those supplied in the predicate. e.g. match_param="action=edit" would require the ``action parameter in the matchdict match the right hande side of the expression (edit) for the view to “match” the current request.

If the match_param is a dict, every key/value pair must match for the predicate to pass.

If match_param is not supplied, the view will be invoked without consideration of the keys and values in request.matchdict.

containment

This value should be a reference to a Python class or interface that a parent object in the context resource’s lineage must provide in order for this view to be found and called. The resources in your resource tree must be “location-aware” to use this feature.

If containment is not supplied, the interfaces and classes in the lineage are not considered when deciding whether or not to invoke the view callable.

See Location-Aware Resources for more information about location-awareness.

xhr

This value should be either True or False. If this value is specified and is True, the WSGI environment must possess an HTTP_X_REQUESTED_WITH (aka X-Requested-With) header that has the value XMLHttpRequest for the associated view callable to be found and called. This is useful for detecting AJAX requests issued from jQuery, Prototype and other Javascript libraries.

If xhr is not specified, the HTTP_X_REQUESTED_WITH HTTP header is not taken into consideration when deciding whether or not to invoke the associated view callable.

accept

The value of this argument represents a match query for one or more mimetypes in the Accept HTTP request header. If this value is specified, it must be in one of the following forms: a mimetype match token in the form text/plain, a wildcard mimetype match token in the form text/* or a match-all wildcard mimetype match token in the form */*. If any of the forms matches the Accept header of the request, this predicate will be true.

If accept is not specified, the HTTP_ACCEPT HTTP header is not taken into consideration when deciding whether or not to invoke the associated view callable.

header

This value represents an HTTP header name or a header name/value pair.

If header is specified, it must be a header name or a headername:headervalue pair.

If header is specified without a value (a bare header name only, e.g. If-Modified-Since), the view will only be invoked if the HTTP header exists with any value in the request.

If header is specified, and possesses a name/value pair (e.g. User-Agent:Mozilla/.*), the view will only be invoked if the HTTP header exists and the HTTP header matches the value requested. When the headervalue contains a : (colon), it will be considered a name/value pair (e.g. User-Agent:Mozilla/.* or Host:localhost). The value portion should be a regular expression.

Whether or not the value represents a header name or a header name/value pair, the case of the header name is not significant.

If header is not specified, the composition, presence or absence of HTTP headers is not taken into consideration when deciding whether or not to invoke the associated view callable.

path_info

This value represents a regular expression pattern that will be tested against the PATH_INFO WSGI environment variable to decide whether or not to call the associated view callable. If the regex matches, this predicate will be True.

If path_info is not specified, the WSGI PATH_INFO is not taken into consideration when deciding whether or not to invoke the associated view callable.

custom_predicates

If custom_predicates is specified, it must be a sequence of references to custom predicate callables. Use custom predicates when no set of predefined predicates do what you need. Custom predicates can be combined with predefined predicates as necessary. Each custom predicate callable should accept two arguments: context and request and should return either True or False after doing arbitrary evaluation of the context resource and/or the request. If all callables return True, the associated view callable will be considered viable for a given request.

If custom_predicates is not specified, no custom predicates are used.

Adding View Configuration Using the @view_config Decorator

Warning

Using this feature tends to slows down application startup slightly, as more work is performed at application startup to scan for view configuration declarations. For maximum startup performance, use the view configuration method described in Adding View Configuration Using add_view() instead.

The view_config decorator can be used to associate view configuration information with a function, method, or class that acts as a Pyramid view callable.

Here’s an example of the view_config decorator that lives within a Pyramid application module views.py:

1
2
3
4
5
6
7
from resources import MyResource
from pyramid.view import view_config
from pyramid.response import Response

@view_config(route_name='ok', request_method='POST', permission='read')
def my_view(request):
    return Response('OK')

Using this decorator as above replaces the need to add this imperative configuration stanza:

1
2
config.add_view('mypackage.views.my_view', route_name='ok',
                request_method='POST', permission='read')

All arguments to view_config may be omitted. For example:

1
2
3
4
5
6
7
from pyramid.response import Response
from pyramid.view import view_config

@view_config()
def my_view(request):
    """ My view """
    return Response()

Such a registration as the one directly above implies that the view name will be my_view, registered with a context argument that matches any resource type, using no permission, registered against requests with any request method, request type, request param, route name, or containment.

The mere existence of a @view_config decorator doesn’t suffice to perform view configuration. All that the decorator does is “annotate” the function with your configuration declarations, it doesn’t process them. To make Pyramid process your pyramid.view.view_config declarations, you must use the scan method of a pyramid.config.Configurator:

1
2
3
# config is assumed to be an instance of the
# pyramid.config.Configurator class
config.scan()

Please see Declarative Configuration for detailed information about what happens when code is scanned for configuration declarations resulting from use of decorators like view_config.

See pyramid.config for additional API arguments to the scan() method. For example, the method allows you to supply a package argument to better control exactly which code will be scanned.

All arguments to the view_config decorator mean precisely the same thing as they would if they were passed as arguments to the pyramid.config.Configurator.add_view() method save for the view argument. Usage of the view_config decorator is a form of declarative configuration, while pyramid.config.Configurator.add_view() is a form of imperative configuration. However, they both do the same thing.

@view_config Placement

A view_config decorator can be placed in various points in your application.

If your view callable is a function, it may be used as a function decorator:

1
2
3
4
5
6
from pyramid.view import view_config
from pyramid.response import Response

@view_config(route_name='edit')
def edit(request):
    return Response('edited!')

If your view callable is a class, the decorator can also be used as a class decorator in Python 2.6 and better (Python 2.5 and below do not support class decorators). All the arguments to the decorator are the same when applied against a class as when they are applied against a function. For example:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
from pyramid.response import Response
from pyramid.view import view_config

@view_config(route_name='hello')
class MyView(object):
    def __init__(self, request):
        self.request = request

    def __call__(self):
        return Response('hello')

You can use the view_config decorator as a simple callable to manually decorate classes in Python 2.5 and below without the decorator syntactic sugar, if you wish:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
from pyramid.response import Response
from pyramid.view import view_config

class MyView(object):
    def __init__(self, request):
        self.request = request

    def __call__(self):
        return Response('hello')

my_view = view_config(route_name='hello')(MyView)

More than one view_config decorator can be stacked on top of any number of others. Each decorator creates a separate view registration. For example:

1
2
3
4
5
6
7
from pyramid.view import view_config
from pyramid.response import Response

@view_config(route_name='edit')
@view_config(route_name='change')
def edit(request):
    return Response('edited!')

This registers the same view under two different names.

The decorator can also be used against a method of a class:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
from pyramid.response import Response
from pyramid.view import view_config

class MyView(object):
    def __init__(self, request):
        self.request = request

    @view_config(route_name='hello')
    def amethod(self):
        return Response('hello')

When the decorator is used against a method of a class, a view is registered for the class, so the class constructor must accept an argument list in one of two forms: either it must accept a single argument request or it must accept two arguments, context, request.

The method which is decorated must return a response.

Using the decorator against a particular method of a class is equivalent to using the attr parameter in a decorator attached to the class itself. For example, the above registration implied by the decorator being used against the amethod method could be spelled equivalently as the below:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
from pyramid.response import Response
from pyramid.view import view_config

@view_config(attr='amethod', route_name='hello')
class MyView(object):
    def __init__(self, request):
        self.request = request

    def amethod(self):
        return Response('hello')
Adding View Configuration Using add_view()

The pyramid.config.Configurator.add_view() method within pyramid.config is used to configure a view “imperatively” (without a view_config decorator). The arguments to this method are very similar to the arguments that you provide to the view_config decorator. For example:

1
2
3
4
5
6
7
8
from pyramid.response import Response

def hello_world(request):
    return Response('hello!')

# config is assumed to be an instance of the
# pyramid.config.Configurator class
config.add_view(hello_world, route_name='hello')

The first argument, view, is required. It must either be a Python object which is the view itself or a dotted Python name to such an object. In the above example, view is the hello_world function. All other arguments are optional. See pyramid.config.Configurator.add_view() for more information.

When you use only add_view() to add view configurations, you don’t need to issue a scan in order for the view configuration to take effect.

Configuring View Security

If an authorization policy is active, any permission attached to a view configuration found during view lookup will be verified. This will ensure that the currently authenticated user possesses that permission against the context resource before the view function is actually called. Here’s an example of specifying a permission in a view configuration using add_view():

1
2
3
4
5
# config is an instance of pyramid.config.Configurator

config.add_route('add', '/add.html', factory='mypackage.Blog')
config.add_view('myproject.views.add_entry', route_name='add',
                permission='add')

When an authorization policy is enabled, this view will be protected with the add permission. The view will not be called if the user does not possess the add permission relative to the current context. Instead the forbidden view result will be returned to the client as per Protecting Views with Permissions.

NotFound Errors

It’s useful to be able to debug NotFound error responses when they occur unexpectedly due to an application registry misconfiguration. To debug these errors, use the PYRAMID_DEBUG_NOTFOUND environment variable or the pyramid.debug_notfound configuration file setting. Details of why a view was not found will be printed to stderr, and the browser representation of the error will include the same information. See Environment Variables and .ini File Settings for more information about how, and where to set these values.

Influencing HTTP Caching

Note

This feature is new in Pyramid 1.1.

When a non-None http_cache argument is passed to a view configuration, Pyramid will set Expires and Cache-Control response headers in the resulting response, causing browsers to cache the response data for some time. See http_cache in Non-Predicate Arguments for the its allowable values and what they mean.

Sometimes it’s undesirable to have these headers set as the result of returning a response from a view, even though you’d like to decorate the view with a view configuration decorator that has http_cache. Perhaps there’s an alternate branch in your view code that returns a response that should never be cacheable, while the “normal” branch returns something that should always be cacheable. If this is the case, set the prevent_auto attribute of the response.cache_control object to a non-False value. For example, the below view callable is configured with a @view_config decorator that indicates any response from the view should be cached for 3600 seconds. However, the view itself prevents caching from taking place unless there’s a should_cache GET or POST variable:

from pyramid.view import view_config

@view_config(http_cache=3600)
def view(request):
    response = Response()
    if not 'should_cache' in request.params:
        response.cache_control.prevent_auto = True
    return response

Note that the http_cache machinery will overwrite or add to caching headers you set within the view itself unless you use preserve_auto.

You can also turn of the effect of http_cache entirely for the duration of a Pyramid application lifetime. To do so, set the PYRAMID_PREVENT_HTTP_CACHE environment variable or the pyramid.prevent_http_cache configuration value setting to a true value. For more information, see Preventing HTTP Caching.

Note that setting pyramid.prevent_http_cache will have no effect on caching headers that your application code itself sets. It will only prevent caching headers that would have been set by the Pyramid HTTP caching machinery invoked as the result of the http_cache argument to view configuration.

Debugging View Configuration

See Displaying Matching Views for a Given URL for information about how to display each of the view callables that might match for a given URL. This can be an effective way to figure out why a particular view callable is being called instead of the one you’d like to be called.

Static Assets

An asset is any file contained within a Python package which is not a Python source code file. For example, each of the following is an asset:

  • a GIF image file contained within a Python package or contained within any subdirectory of a Python package.
  • a CSS file contained within a Python package or contained within any subdirectory of a Python package.
  • a JavaScript source file contained within a Python package or contained within any subdirectory of a Python package.
  • A directory within a package that does not have an __init__.py in it (if it possessed an __init__.py it would be a package).
  • a Chameleon or Mako template file contained within a Python package.

The use of assets is quite common in most web development projects. For example, when you create a Pyramid application using one of the available scaffolds, as described in Creating the Project, the directory representing the application contains a Python package. Within that Python package, there are directories full of files which are static assets. For example, there’s a static directory which contains .css, .js, and .gif files. These asset files are delivered when a user visits an application URL.

Understanding Asset Specifications

Let’s imagine you’ve created a Pyramid application that uses a Chameleon ZPT template via the pyramid.renderers.render_to_response() API. For example, the application might address the asset using the asset specification myapp:templates/some_template.pt using that API within a views.py file inside a myapp package:

1
2
from pyramid.renderers import render_to_response
render_to_response('myapp:templates/some_template.pt', {}, request)

“Under the hood”, when this API is called, Pyramid attempts to make sense out of the string myapp:templates/some_template.pt provided by the developer. This string is an asset specification. It is composed of two parts:

  • The package name (myapp)
  • The asset name (templates/some_template.pt), relative to the package directory.

The two parts are separated by the colon character.

Pyramid uses the Python pkg_resources API to resolve the package name and asset name to an absolute (operating-system-specific) file name. It eventually passes this resolved absolute filesystem path to the Chameleon templating engine, which then uses it to load, parse, and execute the template file.

There is a second form of asset specification: a relative asset specification. Instead of using an “absolute” asset specification which includes the package name, in certain circumstances you can omit the package name from the specification. For example, you might be able to use templates/mytemplate.pt instead of myapp:templates/some_template.pt. Such asset specifications are usually relative to a “current package.” The “current package” is usually the package which contains the code that uses the asset specification. Pyramid APIs which accept relative asset specifications typically describe what the asset is relative to in their individual documentation.

Serving Static Assets

Pyramid makes it possible to serve up static asset files from a directory on a filesystem to an application user’s browser. Use the pyramid.config.Configurator.add_static_view() to instruct Pyramid to serve static assets such as JavaScript and CSS files. This mechanism makes a directory of static files available at a name relative to the application root URL, e.g. /static or as an external URL.

Note

add_static_view() cannot serve a single file, nor can it serve a directory of static files directly relative to the root URL of a Pyramid application. For these features, see Advanced: Serving Static Assets Using a View Callable.

Here’s an example of a use of add_static_view() that will serve files up from the /var/www/static directory of the computer which runs the Pyramid application as URLs beneath the /static URL prefix.

1
2
# config is an instance of pyramid.config.Configurator
config.add_static_view(name='static', path='/var/www/static')

The name prepresents a URL prefix. In order for files that live in the path directory to be served, a URL that requests one of them must begin with that prefix. In the example above, name is static, and path is /var/www/static. In English, this means that you wish to serve the files that live in /var/www/static as sub-URLs of the /static URL prefix. Therefore, the file /var/www/static/foo.css will be returned when the user visits your application’s URL /static/foo.css.

A static directory named at path may contain subdirectories recursively, and any subdirectories may hold files; these will be resolved by the static view as you would expect. The Content-Type header returned by the static view for each particular type of file is dependent upon its file extension.

By default, all files made available via add_static_view() are accessible by completely anonymous users. Simple authorization can be required, however. To protect a set of static files using a permission, in addition to passing the required name and path arguments, also pass the permission keyword argument to add_static_view(). The value of the permission argument represents the permission that the user must have relative to the current context when the static view is invoked. A user will be required to possess this permission to view any of the files represented by path of the static view. If your static assets must be protected by a more complex authorization scheme, see Advanced: Serving Static Assets Using a View Callable.

Here’s another example that uses an asset specification instead of an absolute path as the path argument. To convince add_static_view() to serve files up under the /static URL from the a/b/c/static directory of the Python package named some_package, we can use a fully qualified asset specification as the path:

1
2
# config is an instance of pyramid.config.Configurator
config.add_static_view(name='static', path='some_package:a/b/c/static')

The path provided to add_static_view() may be a fully qualified asset specification or an absolute path.

Instead of representing a URL prefix, the name argument of a call to add_static_view() can alternately be a URL. Each of examples we’ve seen so far have shown usage of the name argument as a URL prefix. However, when name is a URL, static assets can be served from an external webserver. In this mode, the name is used as the URL prefix when generating a URL using pyramid.request.Request.static_url().

For example, add_static_view() may be fed a name argument which is http://example.com/images:

1
2
3
# config is an instance of pyramid.config.Configurator
config.add_static_view(name='http://example.com/images',
                       path='mypackage:images')

Because add_static_view() is provided with a name argument that is the URL http://example.com/images, subsequent calls to static_url() with paths that start with the path argument passed to add_static_view() will generate a URL something like http://example.com/images/logo.png. The external webserver listening on example.com must be itself configured to respond properly to such a request. The static_url() API is discussed in more detail later in this chapter.

Generating Static Asset URLs

When a add_static_view() method is used to register a static asset directory, a special helper API named pyramid.request.Request.static_url() can be used to generate the appropriate URL for an asset that lives in one of the directories named by the static registration path attribute.

For example, let’s assume you create a set of static declarations like so:

1
2
config.add_static_view(name='static1', path='mypackage:assets/1')
config.add_static_view(name='static2', path='mypackage:assets/2')

These declarations create URL-accessible directories which have URLs that begin with /static1 and /static2, respectively. The assets in the assets/1 directory of the mypackage package are consulted when a user visits a URL which begins with /static1, and the assets in the assets/2 directory of the mypackage package are consulted when a user visits a URL which begins with /static2.

You needn’t generate the URLs to static assets “by hand” in such a configuration. Instead, use the static_url() API to generate them for you. For example:

1
2
3
4
5
6
7
8
from pyramid.chameleon_zpt import render_template_to_response

def my_view(request):
    css_url = request.static_url('mypackage:assets/1/foo.css')
    js_url = request.static_url('mypackage:assets/2/foo.js')
    return render_template_to_response('templates/my_template.pt',
                                       css_url = css_url,
                                       js_url = js_url)

If the request “application URL” of the running system is http://example.com, the css_url generated above would be: http://example.com/static1/foo.css. The js_url generated above would be http://example.com/static2/foo.js.

One benefit of using the static_url() function rather than constructing static URLs “by hand” is that if you need to change the name of a static URL declaration, the generated URLs will continue to resolve properly after the rename.

URLs may also be generated by static_url() to static assets that live outside the Pyramid application. This will happen when the add_static_view() API associated with the path fed to static_url() is a URL instead of a view name. For example, the name argument may be http://example.com while the the path given may be mypackage:images:

1
2
config.add_static_view(name='http://example.com/images',
                       path='mypackage:images')

Under such a configuration, the URL generated by static_url for assets which begin with mypackage:images will be prefixed with http://example.com/images:

1
2
request.static_url('mypackage:images/logo.png')
# -> http://example.com/images/logo.png

Using static_url() in conjunction with a add_static_view() makes it possible to put static media on a separate webserver during production (if the name argument to add_static_view() is a URL), while keeping static media package-internal and served by the development webserver during development (if the name argument to add_static_view() is a URL prefix). To create such a circumstance, we suggest using the pyramid.registry.Registry.settings API in conjunction with a setting in the application .ini file named media_location. Then set the value of media_location to either a prefix or a URL depending on whether the application is being run in development or in production (use a different .ini file for production than you do for development). This is just a suggestion for a pattern; any setting name other than media_location could be used.

Advanced: Serving Static Assets Using a View Callable

For more flexibility, static assets can be served by a view callable which you register manually. For example, if you’re using URL dispatch, you may want static assets to only be available as a fallback if no previous route matches. Alternately, you might like to serve a particular static asset manually, because its download requires authentication.

Note that you cannot use the static_url() API to generate URLs against assets made accessible by registering a custom static view.

Root-Relative Custom Static View (URL Dispatch Only)

The pyramid.static.static_view helper class generates a Pyramid view callable. This view callable can serve static assets from a directory. An instance of this class is actually used by the add_static_view() configuration method, so its behavior is almost exactly the same once it’s configured.

Warning

The following example will not work for applications that use traversal, it will only work if you use URL dispatch exclusively. The root-relative route we’ll be registering will always be matched before traversal takes place, subverting any views registered via add_view (at least those without a route_name). A static_view static view cannot be made root-relative when you use traversal unless it’s registered as a Not Found view.

To serve files within a directory located on your filesystem at /path/to/static/dir as the result of a “catchall” route hanging from the root that exists at the end of your routing table, create an instance of the static_view class inside a static.py file in your application root as below.

1
2
from pyramid.static import static
static_view = static_view('/path/to/static/dir', use_subpath=True)

Note

For better cross-system flexibility, use an asset specification as the argument to static_view instead of a physical absolute filesystem path, e.g. mypackage:static instead of /path/to/mypackage/static.

Subsequently, you may wire the files that are served by this view up to be accessible as /<filename> using a configuration method in your application’s startup code.

1
2
3
4
5
# .. every other add_route declaration should come
# before this one, as it will, by default, catch all requests

config.add_route('catchall_static', '/*subpath')
config.add_view('myapp.static.static_view', route_name='catchall_static')

The special name *subpath above is used by the static_view view callable to signify the path of the file relative to the directory you’re serving.

Registering A View Callable to Serve a “Static” Asset

You can register a simple view callable to serve a single static asset. To do so, do things “by hand”. First define the view callable.

1
2
3
4
5
6
7
import os
from pyramid.response import Response

def favicon_view(request):
    here = os.path.dirname(__file__)
    icon = open(os.path.join(here, 'static', 'favicon.ico'))
    return Response(content_type='image/x-icon', app_iter=icon)

The above bit of code within favicon_view computes “here”, which is a path relative to the Python file in which the function is defined. It then uses the Python open function to obtain a file handle to a file within “here” named static, and returns a response using the open the file handle as the response’s app_iter. It makes sure to set the right content_type too.

You might register such a view via configuration as a view callable that should be called as the result of a traversal:

1
config.add_view('myapp.views.favicon_view', name='favicon.ico')

Or you might register it to be the view callable for a particular route:

1
2
config.add_route('favicon', '/favicon.ico')
config.add_view('myapp.views.favicon_view', route_name='favicon')

Because this is a simple view callable, it can be protected with a permission or can be configured to respond under different circumstances using view predicate arguments.

Overriding Assets

It can often be useful to override specific assets from “outside” a given Pyramid application. For example, you may wish to reuse an existing Pyramid application more or less unchanged. However, some specific template file owned by the application might have inappropriate HTML, or some static asset (such as a logo file or some CSS file) might not be appropriate. You could just fork the application entirely, but it’s often more convenient to just override the assets that are inappropriate and reuse the application “as is”. This is particularly true when you reuse some “core” application over and over again for some set of customers (such as a CMS application, or some bug tracking application), and you want to make arbitrary visual modifications to a particular application deployment without forking the underlying code.

To this end, Pyramid contains a feature that makes it possible to “override” one asset with one or more other assets. In support of this feature, a Configurator API exists named pyramid.config.Configurator.override_asset(). This API allows you to override the following kinds of assets defined in any Python package:

  • Individual Chameleon templates.
  • A directory containing multiple Chameleon templates.
  • Individual static files served up by an instance of the pyramid.static.static_view helper class.
  • A directory of static files served up by an instance of the pyramid.static.static_view helper class.
  • Any other asset (or set of assets) addressed by code that uses the setuptools pkg_resources API.
The override_asset API

An individual call to override_asset() can override a single asset. For example:

1
2
3
config.override_asset(
         to_override='some.package:templates/mytemplate.pt',
         override_with='another.package:othertemplates/anothertemplate.pt')

The string value passed to both to_override and override_with sent to the override_asset API is called an asset specification. The colon separator in a specification separates the package name from the asset name. The colon and the following asset name are optional. If they are not specified, the override attempts to resolve every lookup into a package from the directory of another package. For example:

1
2
config.override_asset(to_override='some.package',
                      override_with='another.package')

Individual subdirectories within a package can also be overridden:

1
2
config.override_asset(to_override='some.package:templates/',
                      override_with='another.package:othertemplates/')

If you wish to override a directory with another directory, you must make sure to attach the slash to the end of both the to_override specification and the override_with specification. If you fail to attach a slash to the end of a specification that points to a directory, you will get unexpected results.

You cannot override a directory specification with a file specification, and vice versa: a startup error will occur if you try. You cannot override an asset with itself: a startup error will occur if you try.

Only individual package assets may be overridden. Overrides will not traverse through subpackages within an overridden package. This means that if you want to override assets for both some.package:templates, and some.package.views:templates, you will need to register two overrides.

The package name in a specification may start with a dot, meaning that the package is relative to the package in which the configuration construction file resides (or the package argument to the Configurator class construction). For example:

1
2
config.override_asset(to_override='.subpackage:templates/',
                      override_with='another.package:templates/')

Multiple calls to override_asset which name a shared to_override but a different override_with specification can be “stacked” to form a search path. The first asset that exists in the search path will be used; if no asset exists in the override path, the original asset is used.

Asset overrides can actually override assets other than templates and static files. Any software which uses the pkg_resources.get_resource_filename(), pkg_resources.get_resource_stream() or pkg_resources.get_resource_string() APIs will obtain an overridden file when an override is used.

Request and Response Objects

Note

This chapter is adapted from a portion of the WebOb documentation, originally written by Ian Bicking.

Pyramid uses the WebOb package as a basis for its request and response object implementations. The request object that is passed to a Pyramid view is an instance of the pyramid.request.Request class, which is a subclass of webob.Request. The response returned from a Pyramid view renderer is an instance of the pyramid.response.Response class, which is a subclass of the webob.Response class. Users can also return an instance of pyramid.response.Response directly from a view as necessary.

WebOb is a project separate from Pyramid with a separate set of authors and a fully separate set of documentation. Pyramid adds some functionality to the standard WebOb request, which is documented in the pyramid.request API documentation.

WebOb provides objects for HTTP requests and responses. Specifically it does this by wrapping the WSGI request environment and response status, header list, and app_iter (body) values.

WebOb request and response objects provide many conveniences for parsing WSGI requests and forming WSGI responses. WebOb is a nice way to represent “raw” WSGI requests and responses; however, we won’t cover that use case in this document, as users of Pyramid don’t typically need to use the WSGI-related features of WebOb directly. The reference documentation shows many examples of creating requests and using response objects in this manner, however.

Request

The request object is a wrapper around the WSGI environ dictionary. This dictionary contains keys for each header, keys that describe the request (including the path and query string), a file-like object for the request body, and a variety of custom keys. You can always access the environ with req.environ.

Some of the most important/interesting attributes of a request object:

req.method:
The request method, e.g., 'GET', 'POST'
req.GET:
A multidict with all the variables in the query string.
req.POST:
A multidict with all the variables in the request body. This only has variables if the request was a POST and it is a form submission.
req.params:
A multidict with a combination of everything in req.GET and req.POST.
req.body:
The contents of the body of the request. This contains the entire request body as a string. This is useful when the request is a POST that is not a form submission, or a request like a PUT. You can also get req.body_file for a file-like object.
req.json_body
The JSON-decoded contents of the body of the request. See Dealing With A JSON-Encoded Request Body.
req.cookies:
A simple dictionary of all the cookies.
req.headers:
A dictionary of all the headers. This dictionary is case-insensitive.
req.urlvars and req.urlargs:
req.urlvars are the keyword parameters associated with the request URL. req.urlargs are the positional parameters. These are set by products like Routes and Selector.

Also, for standard HTTP request headers there are usually attributes, for instance: req.accept_language, req.content_length, req.user_agent, as an example. These properties expose the parsed form of each header, for whatever parsing makes sense. For instance, req.if_modified_since returns a datetime object (or None if the header is was not provided).

Note

Full API documentation for the Pyramid request object is available in pyramid.request.

Special Attributes Added to the Request by Pyramid

In addition to the standard WebOb attributes, Pyramid adds special attributes to every request: context, registry, root, subpath, traversed, view_name, virtual_root, virtual_root_path, session, and tmpl_context, matchdict, and matched_route. These attributes are documented further within the pyramid.request.Request API documentation.

URLs

In addition to these attributes, there are several ways to get the URL of the request. I’ll show various values for an example URL http://localhost/app/blog?id=10, where the application is mounted at http://localhost/app.

req.url:
The full request URL, with query string, e.g., http://localhost/app/blog?id=10
req.host:
The host information in the URL, e.g., localhost
req.host_url:
The URL with the host, e.g., http://localhost
req.application_url:
The URL of the application (just the SCRIPT_NAME portion of the path, not PATH_INFO). E.g., http://localhost/app
req.path_url:
The URL of the application including the PATH_INFO. e.g., http://localhost/app/blog
req.path:
The URL including PATH_INFO without the host or scheme. e.g., /app/blog
req.path_qs:
The URL including PATH_INFO and the query string. e.g, /app/blog?id=10
req.query_string:
The query string in the URL, e.g., id=10
req.relative_url(url, to_application=False):
Gives a URL, relative to the current URL. If to_application is True, then resolves it relative to req.application_url.
Methods

There are methods of request objects documented in pyramid.request.Request but you’ll find that you won’t use very many of them. Here are a couple that might be useful:

Request.blank(base_url):
Creates a new request with blank information, based at the given URL. This can be useful for subrequests and artificial requests. You can also use req.copy() to copy an existing request, or for subrequests req.copy_get() which copies the request but always turns it into a GET (which is safer to share for subrequests).
req.get_response(wsgi_application):
This method calls the given WSGI application with this request, and returns a pyramid.response.Response object. You can also use this for subrequests, or testing.
Unicode

Many of the properties in the request object will return unicode values if the request encoding/charset is provided. The client can indicate the charset with something like Content-Type: application/x-www-form-urlencoded; charset=utf8, but browsers seldom set this. You can set the charset with req.charset = 'utf8', or during instantiation with Request(environ, charset='utf8'). If you subclass Request you can also set charset as a class-level attribute.

If it is set, then req.POST, req.GET, req.params, and req.cookies will contain unicode strings. Each has a corresponding req.str_* (e.g., req.str_POST) that is always a str, and never unicode.

Multidict

Several attributes of a WebOb request are “multidict”; structures (such as request.GET, request.POST, and request.params). A multidict is a dictionary where a key can have multiple values. The quintessential example is a query string like ?pref=red&pref=blue; the pref variable has two values: red and blue.

In a multidict, when you do request.GET['pref'] you’ll get back only 'blue' (the last value of pref). Sometimes returning a string, and sometimes returning a list, is the cause of frequent exceptions. If you want all the values back, use request.GET.getall('pref'). If you want to be sure there is one and only one value, use request.GET.getone('pref'), which will raise an exception if there is zero or more than one value for pref.

When you use operations like request.GET.items() you’ll get back something like [('pref', 'red'), ('pref', 'blue')]. All the key/value pairs will show up. Similarly request.GET.keys() returns ['pref', 'pref']. Multidict is a view on a list of tuples; all the keys are ordered, and all the values are ordered.

API documentation for a multidict exists as pyramid.interfaces.IMultiDict.

Dealing With A JSON-Encoded Request Body

Note

this feature is new as of Pyramid 1.1.

pyramid.request.Request.json_body is a property that returns a JSON -decoded representation of the request body. If the request does not have a body, or the body is not a properly JSON-encoded value, an exception will be raised when this attribute is accessed.

This attribute is useful when you invoke a Pyramid view callable via e.g. jQuery’s $.ajax function, which has the potential to send a request with a JSON-encoded body.

Using request.json_body is equivalent to:

from json import loads
loads(request.body, encoding=request.charset)

Here’s how to construct an AJAX request in Javascript using jQuery that allows you to use the request.json_body attribute when the request is sent to a Pyramid application:

jQuery.ajax({type:'POST',
             url: 'http://localhost:6543/', // the pyramid server
             data: JSON.stringify({'a':1}),
             contentType: 'application/json; charset=utf-8'});

When such a request reaches a view in your application, the request.json_body attribute will be available in the view callable body.

@view_config(renderer='string')
def aview(request):
    print request.json_body
    return 'OK'

For the above view, printed to the console will be:

{u'a': 1}

For bonus points, here’s a bit of client-side code that will produce a request that has a body suitable for reading via request.json_body using Python’s urllib2 instead of a Javascript AJAX request:

import urllib2
import json

json_payload = json.dumps({'a':1})
headers = {'Content-Type':'application/json; charset=utf-8'}
req = urllib2.Request('http://localhost:6543/', json_payload, headers)
resp = urllib2.urlopen(req)
Cleaning Up After a Request

Sometimes it’s required that some cleanup be performed at the end of a request when a database connection is involved.

For example, let’s say you have a mypackage Pyramid application package that uses SQLAlchemy, and you’d like the current SQLAlchemy database session to be removed after each request. Put the following in the mypackage.__init__ module:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
from mypackage.models import DBSession

from pyramid.events import subscriber
from pyramid.events import NewRequest

def cleanup_callback(request):
    DBSession.remove()

@subscriber(NewRequest)
def add_cleanup_callback(event):
    event.request.add_finished_callback(cleanup_callback)

Registering the cleanup_callback finished callback at the start of a request (by causing the add_cleanup_callback to receive a pyramid.events.NewRequest event at the start of each request) will cause the DBSession to be removed whenever request processing has ended. Note that in the example above, for the pyramid.events.subscriber decorator to “work”, the pyramid.config.Configurator.scan() method must be called against your mypackage package during application initialization.

Note

This is only an example. In particular, it is not necessary to cause DBSession.remove to be called in an application generated from any Pyramid scaffold, because these all use the pyramid_tm package. The cleanup done by DBSession.remove is unnecessary when pyramid_tm middleware is configured into the application.

More Details

More detail about the request object API is available in:

  • The pyramid.request.Request API documentation.
  • The WebOb documentation. All methods and attributes of a webob.Request documented within the WebOb documentation will work with request objects created by Pyramid.

Response

The Pyramid response object can be imported as pyramid.response.Response. This class is a subclass of the webob.Response class. The subclass does not add or change any functionality, so the WebOb Response documentation will be completely relevant for this class as well.

A response object has three fundamental parts:

response.status:
The response code plus reason message, like '200 OK'. To set the code without a message, use status_int, i.e.: response.status_int = 200.
response.headerlist:
A list of all the headers, like [('Content-Type', 'text/html')]. There’s a case-insensitive multidict in response.headers that also allows you to access these same headers.
response.app_iter:
An iterable (such as a list or generator) that will produce the content of the response. This is also accessible as response.body (a string), response.unicode_body (a unicode object, informed by response.charset), and response.body_file (a file-like object; writing to it appends to app_iter).

Everything else in the object typically derives from this underlying state. Here are some highlights:

response.content_type
The content type not including the charset parameter. Typical use: response.content_type = 'text/html'.
response.charset:
The charset parameter of the content-type, it also informs encoding in response.unicode_body. response.content_type_params is a dictionary of all the parameters.
response.set_cookie(key, value, max_age=None, path='/', ...):
Set a cookie. The keyword arguments control the various cookie parameters. The max_age argument is the length for the cookie to live in seconds (you may also use a timedelta object). The Expires key will also be set based on the value of max_age.
response.delete_cookie(key, path='/', domain=None):
Delete a cookie from the client. This sets max_age to 0 and the cookie value to ''.
response.cache_expires(seconds=0):
This makes this response cacheable for the given number of seconds, or if seconds is 0 then the response is uncacheable (this also sets the Expires header).
response(environ, start_response):
The response object is a WSGI application. As an application, it acts according to how you create it. It can do conditional responses if you pass conditional_response=True when instantiating (or set that attribute later). It can also do HEAD and Range requests.
Headers

Like the request, most HTTP response headers are available as properties. These are parsed, so you can do things like response.last_modified = os.path.getmtime(filename).

The details are available in the extracted Response documentation.

Instantiating the Response

Of course most of the time you just want to make a response. Generally any attribute of the response can be passed in as a keyword argument to the class; e.g.:

1
2
from pyramid.response import Response
response = Response(body='hello world!', content_type='text/plain')

The status defaults to '200 OK'. The content_type does not default to anything, though if you subclass pyramid.response.Response and set default_content_type you can override this behavior.

Exception Responses

To facilitate error responses like 404 Not Found, the module pyramid.httpexceptions contains classes for each kind of error response. These include boring, but appropriate error bodies. The exceptions exposed by this module, when used under Pyramid, should be imported from the pyramid.httpexceptions module. This import location contains subclasses and replacements that mirror those in the webob.exc module.

Each class is named pyramid.httpexceptions.HTTP*, where * is the reason for the error. For instance, pyramid.httpexceptions.HTTPNotFound subclasses pyramid.Response, so you can manipulate the instances in the same way. A typical example is:

1
2
3
4
5
6
from pyramid.httpexceptions import HTTPNotFound
from pyramid.httpexceptions import HTTPMovedPermanently

response = HTTPNotFound('There is no such resource')
# or:
response = HTTPMovedPermanently(location=new_url)
More Details

More details about the response object API are available in the pyramid.response documentation. More details about exception responses are in the pyramid.httpexceptions API documentation. The WebOb documentation is also useful.

Sessions

A session is a namespace which is valid for some period of continual activity that can be used to represent a user’s interaction with a web application.

This chapter describes how to configure sessions, what session implementations Pyramid provides out of the box, how to store and retrieve data from sessions, and two session-specific features: flash messages, and cross-site request forgery attack prevention.

Using The Default Session Factory

In order to use sessions, you must set up a session factory during your Pyramid configuration.

A very basic, insecure sample session factory implementation is provided in the Pyramid core. It uses a cookie to store session information. This implementation has the following limitation:

  • The session information in the cookies used by this implementation is not encrypted, so it can be viewed by anyone with access to the cookie storage of the user’s browser or anyone with access to the network along which the cookie travels.
  • The maximum number of bytes that are storable in a serialized representation of the session is fewer than 4000. This is suitable only for very small data sets.

It is digitally signed, however, and thus its data cannot easily be tampered with.

You can configure this session factory in your Pyramid application by using the session_factory argument to the Configurator class:

1
2
3
4
5
from pyramid.session import UnencryptedCookieSessionFactoryConfig
my_session_factory = UnencryptedCookieSessionFactoryConfig('itsaseekreet')

from pyramid.config import Configurator
config = Configurator(session_factory = my_session_factory)

Warning

Note the very long, very explicit name for UnencryptedCookieSessionFactoryConfig. It’s trying to tell you that this implementation is, by default, unencrypted. You should not use it when you keep sensitive information in the session object, as the information can be easily read by both users of your application and third parties who have access to your users’ network traffic. Use a different session factory implementation (preferably one which keeps session data on the server) for anything but the most basic of applications where “session security doesn’t matter”.

Using a Session Object

Once a session factory has been configured for your application, you can access session objects provided by the session factory via the session attribute of any request object. For example:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
from pyramid.response import Response

def myview(request):
    session = request.session
    if 'abc' in session:
        session['fred'] = 'yes'
    session['abc'] = '123'
    if 'fred' in session:
        return Response('Fred was in the session')
    else:
        return Response('Fred was not in the session')

You can use a session much like a Python dictionary. It supports all dictionary methods, along with some extra attributes, and methods.

Extra attributes:

created
An integer timestamp indicating the time that this session was created.
new
A boolean. If new is True, this session is new. Otherwise, it has been constituted from data that was already serialized.

Extra methods:

changed()
Call this when you mutate a mutable value in the session namespace. See the gotchas below for details on when, and why you should call this.
invalidate()
Call this when you want to invalidate the session (dump all data, and – perhaps – set a clearing cookie).

The formal definition of the methods and attributes supported by the session object are in the pyramid.interfaces.ISession documentation.

Some gotchas:

  • Keys and values of session data must be pickleable. This means, typically, that they are instances of basic types of objects, such as strings, lists, dictionaries, tuples, integers, etc. If you place an object in a session data key or value that is not pickleable, an error will be raised when the session is serialized.
  • If you place a mutable value (for example, a list or a dictionary) in a session object, and you subsequently mutate that value, you must call the changed() method of the session object. In this case, the session has no way to know that is was modified. However, when you modify a session object directly, such as setting a value (i.e., __setitem__), or removing a key (e.g., del or pop), the session will automatically know that it needs to re-serialize its data, thus calling changed() is unnecessary. There is no harm in calling changed() in either case, so when in doubt, call it after you’ve changed sessioning data.

Using Alternate Session Factories

At the time of this writing, exactly one alternate session factory implementation exists, named pyramid_beaker. This is a session factory that uses the Beaker library as a backend. Beaker has support for file-based sessions, database based sessions, and encrypted cookie-based sessions. See http://github.com/Pylons/pyramid_beaker for more information about pyramid_beaker.

Creating Your Own Session Factory

If none of the default or otherwise available sessioning implementations for Pyramid suit you, you may create your own session object by implementing a session factory. Your session factory should return a session. The interfaces for both types are available in pyramid.interfaces.ISessionFactory and pyramid.interfaces.ISession. You might use the cookie implementation in the pyramid.session module as inspiration.

Flash Messages

“Flash messages” are simply a queue of message strings stored in the session. To use flash messaging, you must enable a session factory as described in Using The Default Session Factory or Using Alternate Session Factories.

Flash messaging has two main uses: to display a status message only once to the user after performing an internal redirect, and to allow generic code to log messages for single-time display without having direct access to an HTML template. The user interface consists of a number of methods of the session object.

Using the session.flash Method

To add a message to a flash message queue, use a session object’s flash() method:

request.session.flash('mymessage')

The flash() method appends a message to a flash queue, creating the queue if necessary.

flash() accepts three arguments:

flash(message, queue='', allow_duplicate=True)

The message argument is required. It represents a message you wish to later display to a user. It is usually a string but the message you provide is not modified in any way.

The queue argument allows you to choose a queue to which to append the message you provide. This can be used to push different kinds of messages into flash storage for later display in different places on a page. You can pass any name for your queue, but it must be a string. Each queue is independent, and can be popped by pop_flash() or examined via peek_flash() separately. queue defaults to the empty string. The empty string represents the default flash message queue.

request.session.flash(msg, 'myappsqueue')

The allow_duplicate argument defaults to True. If this is False, and you attempt to add a message value which is already present in the queue, it will not be added.

Using the session.pop_flash Method

Once one or more messages have been added to a flash queue by the session.flash() API, the session.pop_flash() API can be used to pop an entire queue and return it for use.

To pop a particular queue of messages from the flash object, use the session object’s pop_flash() method. This returns a list of the messages that were added to the flash queue, and empties the queue.

pop_flash(queue='')
1
2
3
>>> request.session.flash('info message')
>>> request.session.pop_flash()
['info message']

Calling session.pop_flash() again like above without a corresponding call to session.flash() will return an empty list, because the queue has already been popped.

1
2
3
4
5
>>> request.session.flash('info message')
>>> request.session.pop_flash()
['info message']
>>> request.session.pop_flash()
[]
Using the session.peek_flash Method

Once one or more messages has been added to a flash queue by the session.flash() API, the session.peek_flash() API can be used to “peek” at that queue. Unlike session.pop_flash(), the queue is not popped from flash storage.

peek_flash(queue='')
1
2
3
4
5
6
7
8
9
>>> request.session.flash('info message')
>>> request.session.peek_flash()
['info message']
>>> request.session.peek_flash()
['info message']
>>> request.session.pop_flash()
['info message']
>>> request.session.peek_flash()
[]

Preventing Cross-Site Request Forgery Attacks

Cross-site request forgery attacks are a phenomenon whereby a user with an identity on your website might click on a URL or button on another website which secretly redirects the user to your application to perform some command that requires elevated privileges.

You can avoid most of these attacks by making sure that the correct CSRF token has been set in an Pyramid session object before performing any actions in code which requires elevated privileges that is invoked via a form post. To use CSRF token support, you must enable a session factory as described in Using The Default Session Factory or Using Alternate Session Factories.

Using the session.get_csrf_token Method

To get the current CSRF token from the session, use the session.get_csrf_token() method.

token = request.session.get_csrf_token()

The session.get_csrf_token() method accepts no arguments. It returns a CSRF token string. If session.get_csrf_token() or session.new_csrf_token() was invoked previously for this session, the existing token will be returned. If no CSRF token previously existed for this session, a new token will be will be set into the session and returned. The newly created token will be opaque and randomized.

You can use the returned token as the value of a hidden field in a form that posts to a method that requires elevated privileges. The handler for the form post should use session.get_csrf_token() again to obtain the current CSRF token related to the user from the session, and compare it to the value of the hidden form field. For example, if your form rendering included the CSRF token obtained via session.get_csrf_token() as a hidden input field named csrf_token:

1
2
3
token = request.session.get_csrf_token()
if token != request.POST['csrf_token']:
    raise ValueError('CSRF token did not match')
Using the session.new_csrf_token Method

To explicitly add a new CSRF token to the session, use the session.new_csrf_token() method. This differs only from session.get_csrf_token() inasmuch as it clears any existing CSRF token, creates a new CSRF token, sets the token into the session, and returns the token.

token = request.session.new_csrf_token()

Using Events

An event is an object broadcast by the Pyramid framework at interesting points during the lifetime of an application. You don’t need to use events in order to create most Pyramid applications, but they can be useful when you want to perform slightly advanced operations. For example, subscribing to an event can allow you to run some code as the result of every new request.

Events in Pyramid are always broadcast by the framework. However, they only become useful when you register a subscriber. A subscriber is a function that accepts a single argument named event:

1
2
def mysubscriber(event):
    print event

The above is a subscriber that simply prints the event to the console when it’s called.

The mere existence of a subscriber function, however, is not sufficient to arrange for it to be called. To arrange for the subscriber to be called, you’ll need to use the pyramid.config.Configurator.add_subscriber() method or you’ll need to use the pyramid.events.subscriber() decorator to decorate a function found via a scan.

Configuring an Event Listener Imperatively

You can imperatively configure a subscriber function to be called for some event type via the add_subscriber() method (see also Configurator):

1
2
3
4
5
6
7
8
from pyramid.events import NewRequest

from subscribers import mysubscriber

# "config" below is assumed to be an instance of a
# pyramid.config.Configurator object

config.add_subscriber(mysubscriber, NewRequest)

The first argument to add_subscriber() is the subscriber function (or a dotted Python name which refers to a subscriber callable); the second argument is the event type.

Configuring an Event Listener Using a Decorator

You can configure a subscriber function to be called for some event type via the pyramid.events.subscriber() function.

1
2
3
4
5
6
from pyramid.events import NewRequest
from pyramid.events import subscriber

@subscriber(NewRequest)
def mysubscriber(event):
        event.request.foo = 1

When the subscriber() decorator is used a scan must be performed against the package containing the decorated function for the decorator to have any effect.

Either of the above registration examples implies that every time the Pyramid framework emits an event object that supplies an pyramid.events.NewRequest interface, the mysubscriber function will be called with an event object.

As you can see, a subscription is made in terms of a class (such as pyramid.events.NewResponse). The event object sent to a subscriber will always be an object that possesses an interface. For pyramid.events.NewResponse, that interface is pyramid.interfaces.INewResponse. The interface documentation provides information about available attributes and methods of the event objects.

The return value of a subscriber function is ignored. Subscribers to the same event type are not guaranteed to be called in any particular order relative to each other.

All the concrete Pyramid event types are documented in the pyramid.events API documentation.

An Example

If you create event listener functions in a subscribers.py file in your application like so:

1
2
3
4
5
def handle_new_request(event):
    print 'request', event.request

def handle_new_response(event):
    print 'response', event.response

You may configure these functions to be called at the appropriate times by adding the following code to your application’s configuration startup:

1
2
3
4
5
6
# config is an instance of pyramid.config.Configurator

config.add_subscriber('myproject.subscribers.handle_new_request',
                      'pyramid.events.NewRequest')
config.add_subscriber('myproject.subscribers.handle_new_response',
                      'pyramid.events.NewResponse')

Either mechanism causes the functions in subscribers.py to be registered as event subscribers. Under this configuration, when the application is run, each time a new request or response is detected, a message will be printed to the console.

Each of our subscriber functions accepts an event object and prints an attribute of the event object. This begs the question: how can we know which attributes a particular event has?

We know that pyramid.events.NewRequest event objects have a request attribute, which is a request object, because the interface defined at pyramid.interfaces.INewRequest says it must. Likewise, we know that pyramid.interfaces.NewResponse events have a response attribute, which is a response object constructed by your application, because the interface defined at pyramid.interfaces.INewResponse says it must (pyramid.events.NewResponse objects also have a request).

Environment Variables and .ini File Settings

Pyramid behavior can be configured through a combination of operating system environment variables and .ini configuration file application section settings. The meaning of the environment variables and the configuration file settings overlap.

Note

Where a configuration file setting exists with the same meaning as an environment variable, and both are present at application startup time, the environment variable setting takes precedence.

The term “configuration file setting name” refers to a key in the .ini configuration for your application. The configuration file setting names documented in this chapter are reserved for Pyramid use. You should not use them to indicate application-specific configuration settings.

Reloading Templates

When this value is true, templates are automatically reloaded whenever they are modified without restarting the application, so you can see changes to templates take effect immediately during development. This flag is meaningful to Chameleon and Mako templates, as well as most third-party template rendering extensions.

Environment Variable Name Config File Setting Name
PYRAMID_RELOAD_TEMPLATES
pyramid.reload_templates
or reload_templates

Reloading Assets

Don’t cache any asset file data when this value is true. See also Overriding Assets.

Environment Variable Name Config File Setting Name
PYRAMID_RELOAD_ASSETS pyramid.reload_assets or reload_assets

Note

For backwards compatibility purposes, aliases can be used for configurating asset reloading: PYRAMID_RELOAD_RESOURCES (envvar) and pyramid.reload_resources (config file).

Debugging Authorization

Print view authorization failure and success information to stderr when this value is true. See also Debugging View Authorization Failures.

Environment Variable Name Config File Setting Name
PYRAMID_DEBUG_AUTHORIZATION pyramid.debug_authorization or debug_authorization

Debugging Not Found Errors

Print view-related NotFound debug messages to stderr when this value is true. See also NotFound Errors.

Environment Variable Name Config File Setting Name
PYRAMID_DEBUG_NOTFOUND pyramid.debug_notfound or debug_notfound

Debugging Route Matching

Print debugging messages related to url dispatch route matching when this value is true. See also Debugging Route Matching.

Environment Variable Name Config File Setting Name
PYRAMID_DEBUG_ROUTEMATCH pyramid.debug_routematch or debug_routematch

Preventing HTTP Caching

Prevent the http_cache view configuration argument from having any effect globally in this process when this value is true. No http caching-related response headers will be set by the Pyramid http_cache view configuration feature when this is true. See also Influencing HTTP Caching.

Environment Variable Name Config File Setting Name
PYRAMID_PREVENT_HTTP_CACHE pyramid.prevent_http_cache or prevent_http_cache

Debugging All

Turns on all debug* settings.

Environment Variable Name Config File Setting Name
PYRAMID_DEBUG_ALL pyramid.debug_all or debug_all

Reloading All

Turns on all reload* settings.

Environment Variable Name Config File Setting Name
PYRAMID_RELOAD_ALL pyramid.reload_all or reload_all

Default Locale Name

The value supplied here is used as the default locale name when a locale negotiator is not registered. See also Localization-Related Deployment Settings.

Environment Variable Name Config File Setting Name
PYRAMID_DEFAULT_LOCALE_NAME pyramid.default_locale_name or default_locale_name

Including Packages

pyramid.includes instructs your application to include other packages. Using the setting is equivalent to using the pyramid.config.Configurator.include() method.

Config File Setting Name
pyramid.includes

The value supplied as pyramid.includes should be a sequence. The sequence can take several different forms.

  1. It can be a string.

    If it is a string, the package names can be separated by spaces:

      package1 package2 package3

The package names can also be separated by carriage returns::

   package1
   package2
   package3

  2. It can be a Python list, where the values are strings:

    ['package1', 'package2', 'package3']

Each value in the sequence should be a dotted Python name.

pyramid.includes vs. pyramid.config.Configurator.include()

Two methods exist for including packages: pyramid.includes and pyramid.config.Configurator.include(). This section explains their equivalence.

Using PasteDeploy

Using the following pyramid.includes setting in the PasteDeploy .ini file in your application:

[app:main]
pyramid.includes = pyramid_debugtoolbar
                   pyramid_tm

Is equivalent to using the following statements in your configuration code:

1
2
3
4
5
6
7
8
from pyramid.config import Configurator

def main(global_config, **settings):
    config = Configurator(settings=settings)
    # ...
    config.include('pyramid_debugtoolbar')
    config.include('pyramid_tm')
    # ...

It is fine to use both or either form.

Plain Python

Using the following pyramid.includes setting in your plain-Python Pyramid application:

1
2
3
4
5
from pyramid.config import Configurator

if __name__ == '__main__':
    settings = {'pyramid.includes':'pyramid_debugtoolbar pyramid_tm'}
    config = Configurator(settings=settings)

Is equivalent to using the following statements in your configuration code:

1
2
3
4
5
6
7
from pyramid.config import Configurator

if __name__ == '__main__':
    settings = {}
    config = Configurator(settings=settings)
    config.include('pyramid_debugtoolbar')
    config.include('pyramid_tm')

It is fine to use both or either form.

Explicit Tween Configuration

This value allows you to perform explicit tween ordering in your configuration. Tweens are bits of code used by add-on authors to extend Pyramid. They form a chain, and require ordering.

Ideally, you won’t need to use the pyramid.tweens setting at all. Tweens are generally ordered and included “implicitly” when an add-on package which registers a tween is “included”. Packages are included when you name a pyramid.includes setting in your configuration or when you call pyramid.config.Configuration.include().

Authors of included add-ons provide “implicit” tween configuration ordering hints to Pyramid when their packages are included. However, the implicit tween ordering is only best-effort. Pyramid will attempt to provide an implicit order of tweens as best it can using hints provided by add-on authors, but because it’s only best-effort, if very precise tween ordering is required, the only surefire way to get it is to use an explicit tween order. You may be required to inspect your tween ordering (see Displaying “Tweens”) and add a pyramid.tweens configuration value at the behest of an add-on author.

Config File Setting Name
pyramid.tweens

The value supplied as pyramid.tweens should be a sequence. The sequence can take several different forms.

  1. It can be a string.

    If it is a string, the tween names can be separated by spaces:

      pkg.tween_factory1 pkg.tween_factory2 pkg.tween_factory3

The tween names can also be separated by carriage returns::

   pkg.tween_factory1
   pkg.tween_factory2
   pkg.tween_factory3

  2. It can be a Python list, where the values are strings:

    ['pkg.tween_factory1', 'pkg.tween_factory2', 'pkg.tween_factory3']

Each value in the sequence should be a dotted Python name.

Paste Configuration vs. Plain-Python Configuration

Using the following pyramid.tweens setting in the PasteDeploy .ini file in your application:

[app:main]
pyramid.tweens = pyramid_debugtoolbar.toolbar.tween_factory
                 pyramid.tweens.excview_tween_factory
                 pyramid_tm.tm_tween_factory

Is equivalent to using the following statements in your configuration code:

1
2
3
4
5
6
7
8
9
from pyramid.config import Configurator

def main(global_config, **settings):
    settings['pyramid.tweens'] = [
            'pyramid_debugtoolbar.toolbar.tween_factory',
            'pyramid.tweebs.excview_tween_factory',
            'pyramid_tm.tm_tween_factory',
             ]
    config = Configurator(settings=settings)

It is fine to use both or either form.

Mako Template Render Settings

Mako derives additional settings to configure its template renderer that should be set when using it. Many of these settings are optional and only need to be set if they should be different from the default. The Mako Template Renderer uses a subclass of Mako’s template lookup and accepts several arguments to configure it.

Mako Directories

The value(s) supplied here are passed in as the template directories. They should be in asset specification format, for example: my.package:templates.

Config File Setting Name
mako.directories
Mako Module Directory

The value supplied here tells Mako where to store compiled Mako templates. If omitted, compiled templates will be stored in memory. This value should be an absolute path, for example: %(here)s/data/templates would use a directory called data/templates in the same parent directory as the INI file.

Config File Setting Name
mako.module_directory
Mako Input Encoding

The encoding that Mako templates are assumed to have. By default this is set to utf-8. If you wish to use a different template encoding, this value should be changed accordingly.

Config File Setting Name
mako.input_encoding
Mako Error Handler

A callable (or a dotted Python name which names a callable) which is called whenever Mako compile or runtime exceptions occur. The callable is passed the current context as well as the exception. If the callable returns True, the exception is considered to be handled, else it is re-raised after the function completes. Is used to provide custom error-rendering functions.

Config File Setting Name
mako.error_handler
Mako Default Filters

List of string filter names that will be applied to all Mako expressions.

Config File Setting Name
mako.default_filters
Mako Import

String list of Python statements, typically individual “import” lines, which will be placed into the module level preamble of all generated Python modules.

Config File Setting Name
mako.imports
Mako Strict Undefined

true or false, representing the “strict undefined” behavior of Mako (see Mako Context Variables). By default, this is false.

Config File Setting Name
mako.strict_undefined
Mako Preprocessor

A callable (or a dotted Python name which names a callable) which is called to preprocess the source before the template is called. The callable will be passed the full template source before it is parsed. The return result of the callable will be used as the template source code.

Note

This feature is new in Pyramid 1.1.

Config File Setting Name
mako.preprocessor

Examples

Let’s presume your configuration file is named MyProject.ini, and there is a section representing your application named [app:main] within the file that represents your Pyramid application. The configuration file settings documented in the above “Config File Setting Name” column would go in the [app:main] section. Here’s an example of such a section:

1
2
3
4
[app:main]
use = egg:MyProject
pyramid.reload_templates = true
pyramid.debug_authorization = true

You can also use environment variables to accomplish the same purpose for settings documented as such. For example, you might start your Pyramid application using the following command line:

$ PYRAMID_DEBUG_AUTHORIZATION=1 PYRAMID_RELOAD_TEMPLATES=1 \
       bin/paster serve MyProject.ini

If you started your application this way, your Pyramid application would behave in the same manner as if you had placed the respective settings in the [app:main] section of your application’s .ini file.

If you want to turn all debug settings (every setting that starts with pyramid.debug_). on in one fell swoop, you can use PYRAMID_DEBUG_ALL=1 as an environment variable setting or you may use pyramid.debug_all=true in the config file. Note that this does not affect settings that do not start with pyramid.debug_* such as pyramid.reload_templates.

If you want to turn all pyramid.reload settings (every setting that starts with pyramid.reload_) on in one fell swoop, you can use PYRAMID_RELOAD_ALL=1 as an environment variable setting or you may use pyramid.reload_all=true in the config file. Note that this does not affect settings that do not start with pyramid.reload_* such as pyramid.debug_notfound.

Note

Specifying configuration settings via environment variables is generally most useful during development, where you may wish to augment or override the more permanent settings in the configuration file. This is useful because many of the reload and debug settings may have performance or security (i.e., disclosure) implications that make them undesirable in a production environment.

Understanding the Distinction Between reload_templates and reload_assets

The difference between pyramid.reload_assets and pyramid.reload_templates is a bit subtle. Templates are themselves also treated by Pyramid as asset files (along with other static files), so the distinction can be confusing. It’s helpful to read Overriding Assets for some context about assets in general.

When pyramid.reload_templates is true, Pyramid takes advantage of the underlying templating systems’ ability to check for file modifications to an individual template file. When pyramid.reload_templates is true but pyramid.reload_assets is not true, the template filename returned by the pkg_resources package (used under the hood by asset resolution) is cached by Pyramid on the first request. Subsequent requests for the same template file will return a cached template filename. The underlying templating system checks for modifications to this particular file for every request. Setting pyramid.reload_templates to True doesn’t affect performance dramatically (although it should still not be used in production because it has some effect).

However, when pyramid.reload_assets is true, Pyramid will not cache the template filename, meaning you can see the effect of changing the content of an overridden asset directory for templates without restarting the server after every change. Subsequent requests for the same template file may return different filenames based on the current state of overridden asset directories. Setting pyramid.reload_assets to True affects performance dramatically, slowing things down by an order of magnitude for each template rendering. However, it’s convenient to enable when moving files around in overridden asset directories. pyramid.reload_assets makes the system very slow when templates are in use. Never set pyramid.reload_assets to True on a production system.

Adding A Custom Setting

From time to time, you may need to add a custom setting to your application. Here’s how:

  • If you’re using an .ini file, change the .ini file, adding the setting to the [app:foo] section representing your Pyramid application. For example:

    [app:main]
# .. other settings
debug_frobnosticator = True

  • In the main() function that represents the place that your Pyramid WSGI application is created, anticipate that you’ll be getting this key/value pair as a setting and do any type conversion necessary.

    If you’ve done any type conversion of your custom value, reset the converted values into the settings dictionary before you pass the dictionary as settings to the Configurator. For example:

    def main(global_config, **settings):
    # ...
    from pyramid.settings import asbool
    debug_frobnosticator = asbool(settings.get(
               'debug_frobnosticator', 'false'))
    settings['debug_frobnosticator'] = debug_frobnosticator
    config = Configurator(settings=settings)

    Note

    It’s especially important that you mutate the settings dictionary with the converted version of the variable before passing it to the Configurator: the configurator makes a copy of settings, it doesn’t use the one you pass directly.

  • When creating an includeme function that will be later added to your application’s configuration you may access the settings dictionary through the instance of the Configurator that is passed into the function as its only argument. For Example:

def includeme(config):
    settings = config.registry.settings
    debug_frobnosticator = settings['debug_frobnosticator']

  • In the runtime code that you need to access the new settings value, find the value in the registry.settings dictionary and use it. In view code (or any other code that has access to the request), the easiest way to do this is via request.registry.settings. For example:

    settings = request.registry.settings
debug_frobnosticator = settings['debug_frobnosticator']

    If you wish to use the value in code that does not have access to the request and you wish to use the value, you’ll need to use the pyramid.threadlocal.get_current_registry() API to obtain the current registry, then ask for its settings attribute. For example:

    registry = pyramid.threadlocal.get_current_registry()
settings = registry.settings
debug_frobnosticator = settings['debug_frobnosticator']

Logging

Pyramid allows you to make use of the Python standard library logging module. This chapter describes how to configure logging and how to send log messages to loggers that you’ve configured.

Warning

This chapter assumes you’ve used a scaffold to create a project which contains development.ini and production.ini files which help configure logging. All of the scaffolds which ship along with Pyramid do this. If you’re not using a scaffold, or if you’ve used a third-party scaffold which does not create these files, the configuration information in this chapter will not be applicable.

Logging Configuration

A Pyramid project created from a scaffold is configured to allow you to send messages to Python standard library logging package loggers from within your application. In particular, the PasteDeploy development.ini and production.ini files created when you use a scaffold include a basic configuration for the Python logging package.

PasteDeploy .ini files use the Python standard library ConfigParser format; this the same format used as the Python logging module’s Configuration file format. The application-related and logging-related sections in the configuration file can coexist peacefully, and the logging-related sections in the file are used from when you run paster serve.

The paster serve command calls the logging.fileConfig function using the specified ini file if it contains a [loggers] section (all of the scaffold-generated .ini files do). logging.fileConfig reads the logging configuration from the ini file upon which paster serve was invoked.

Default logging configuration is provided in both the default development.ini and the production.ini file. The logging configuration in the development.ini file is as follows:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
# Begin logging configuration

[loggers]
keys = root, {{package_logger}}

[handlers]
keys = console

[formatters]
keys = generic

[logger_root]
level = INFO
handlers = console

[logger_{{package_logger}}]
level = DEBUG
handlers =
qualname = {{package}}

[handler_console]
class = StreamHandler
args = (sys.stderr,)
level = NOTSET
formatter = generic

[formatter_generic]
format = %(asctime)s %(levelname)-5.5s [%(name)s][%(threadName)s] %(message)s

# End logging configuration

The production.ini file uses the WARN level in its logger configuration, but it is otherwise identical.

The name {{package_logger}} above will be replaced with the name of your project’s package, which is derived from the name you provide to your project. For instance, if you do:

1
paster create -t pyramid_starter MyApp

The logging configuration will literally be:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
# Begin logging configuration

[loggers]
keys = root, myapp

[handlers]
keys = console

[formatters]
keys = generic

[logger_root]
level = INFO
handlers = console

[logger_myapp]
level = DEBUG
handlers =
qualname = myapp

[handler_console]
class = StreamHandler
args = (sys.stderr,)
level = NOTSET
formatter = generic

[formatter_generic]
format = %(asctime)s %(levelname)-5.5s [%(name)s][%(threadName)s] %(message)s

# End logging configuration

In this logging configuration:

  • a logger named root is created that logs messages at a level above or equal to the INFO level to stderr, with the following format:

    2007-08-17 15:04:08,704 INFO [packagename]
                             Loading resource, id: 86

  • a logger named myapp is configured that logs messages sent at a level above or equal to DEBUG to stderr in the same format as the root logger.

The root logger will be used by all applications in the Pyramid process that ask for a logger (via logging.getLogger) that has a name which begins with anything except your project’s package name (e.g. myapp). The logger with the same name as your package name is reserved for your own usage in your Pyramid application. Its existence means that you can log to a known logging location from any Pyramid application generated via a scaffold.

Pyramid and many other libraries (such as Beaker, SQLAlchemy, Paste) log a number of messages to the root logger for debugging purposes. Switching the root logger level to DEBUG reveals them:

[logger_root]
#level = INFO
level = DEBUG
handlers = console

Some scaffolds configure additional loggers for additional subsystems they use (such as SQLALchemy). Take a look at the production.ini and development.ini files rendered when you create a project from a scaffold.

Sending Logging Messages

Python’s special __name__ variable refers to the current module’s fully qualified name. From any module in a package named myapp, the __name__ builtin variable will always be something like myapp, or myapp.subpackage or myapp.package.subpackage if your project is named myapp. Sending a message to this logger will send it to the myapp logger.

To log messages to the package-specific logger configured in your .ini file, simply create a logger object using the __name__ builtin and call methods on it.

1
2
3
4
5
6
7
8
9
 import logging
 log = logging.getLogger(__name__)

 def myview(request):
     content_type = 'text/plain'
     content = 'Hello World!'
     log.debug('Returning: %s (content-type: %s)', content, content_type)
     request.response.content_type = content_type
     return request.response

This will result in the following printed to the console, on stderr:

16:20:20,440 DEBUG [myapp.views] Returning: Hello World!
                   (content-type: text/plain)

Filtering log messages

Often there’s too much log output to sift through, such as when switching the root logger’s level to DEBUG.

An example: you’re diagnosing database connection issues in your application and only want to see SQLAlchemy’s DEBUG messages in relation to database connection pooling. You can leave the root logger’s level at the less verbose INFO level and set that particular SQLAlchemy logger to DEBUG on its own, apart from the root logger:

[logger_sqlalchemy.pool]
level = DEBUG
handlers =
qualname = sqlalchemy.pool

then add it to the list of loggers:

[loggers]
keys = root, myapp, sqlalchemy.pool

No handlers need to be configured for this logger as by default non root loggers will propagate their log records up to their parent logger’s handlers. The root logger is the top level parent of all loggers.

This technique is used in the default development.ini. The root logger’s level is set to INFO, whereas the application’s log level is set to DEBUG:

# Begin logging configuration

[loggers]
keys = root, myapp

[logger_myapp]
level = DEBUG
handlers =
qualname = helloworld

All of the child loggers of the myapp logger will inherit the DEBUG level unless they’re explicitly set differently. Meaning the myapp.views, myapp.models (and all your app’s modules’) loggers by default have an effective level of DEBUG too.

For more advanced filtering, the logging module provides a Filter object; however it cannot be used directly from the configuration file.

Advanced Configuration

To capture log output to a separate file, use a FileHandler (or a RotatingFileHandler):

[handler_filelog]
class = FileHandler
args = ('%(here)s/myapp.log','a')
level = INFO
formatter = generic

Before it’s recognized, it needs to be added to the list of handlers:

[handlers]
keys = console, myapp, filelog

and finally utilized by a logger.

[logger_root]
level = INFO
handlers = console, filelog

These final 3 lines of configuration directs all of the root logger’s output to the myapp.log as well as the console.

Logging Exceptions

To log (or email) exceptions generated by your Pyramid application, use the pyramid_exclog package. Details about its configuration are in its documentation.

Request Logging with Paste’s TransLogger

Paste provides the TransLogger middleware for logging requests using the Apache Combined Log Format. TransLogger combined with a FileHandler can be used to create an access.log file similar to Apache’s.

Like any standard middleware with a Paste entry point, TransLogger can be configured to wrap your application using .ini file syntax. First, rename your Pyramid .ini file’s [app:main] section to [app:mypyramidapp], then add a [filter:translogger] section, then use a [pipeline:main] section file to form a WSGI pipeline with both the translogger and your application in it. For instance, change from this:

[app:main]
use = egg:MyProject

To this:

[app:mypyramidapp]
use = egg:MyProject

[filter:translogger]
paste.filter_app_factory = egg:Paste#translogger
setup_console_handler = False

[pipeline:main]
pipeline = translogger
           mypyramidapp

Using PasteDeploy this way to form and serve a pipeline is equivalent to wrapping your app in a TransLogger instance via the bottom the main function of your project’s __init__ file:

...
app = config.make_wsgi_app()
from paste.translogger import TransLogger
app = TransLogger(app, setup_console_handler=False)
return app

TransLogger will automatically setup a logging handler to the console when called with no arguments, so it ‘just works’ in environments that don’t configure logging. Since we’ve configured our own logging handlers, we need to disable that option via setup_console_handler = False.

With the filter in place, TransLogger’s logger (named the ‘wsgi’ logger) will propagate its log messages to the parent logger (the root logger), sending its output to the console when we request a page:

00:50:53,694 INFO [myapp.views] Returning: Hello World!
                  (content-type: text/plain)
00:50:53,695 INFO [wsgi] 192.168.1.111 - - [11/Aug/2011:20:09:33 -0700] "GET /hello
HTTP/1.1" 404 - "-"
"Mozilla/5.0 (Macintosh; U; Intel Mac OS X; en-US; rv:1.8.1.6) Gecko/20070725
Firefox/2.0.0.6"

To direct TransLogger to an access.log FileHandler, we need to add that FileHandler to the wsgi logger’s list of handlers:

# Begin logging configuration

[loggers]
keys = root, myapp, wsgi

[logger_wsgi]
level = INFO
handlers = handler_accesslog
qualname = wsgi
propagate = 0

[handler_accesslog]
class = FileHandler
args = ('%(here)s/access.log','a')
level = INFO
formatter = generic

As mentioned above, non-root loggers by default propagate their log records to the root logger’s handlers (currently the console handler). Setting propagate to 0 (false) here disables this; so the wsgi logger directs its records only to the accesslog handler.

Finally, there’s no need to use the generic formatter with TransLogger as TransLogger itself provides all the information we need. We’ll use a formatter that passes-through the log messages as is:

[formatters]
keys = generic, accesslog

[formatter_accesslog]
format = %(message)s

Then wire this new accesslog formatter into the FileHandler:

[handler_accesslog]
class = FileHandler
args = ('%(here)s/access.log','a')
level = INFO
formatter = accesslog

Paste

Packages generated via a scaffold make use of a system created by Ian Bicking named Paste. Paste provides the following features:

  • A way to declare WSGI application configuration in an .ini file (PasteDeploy).
  • A WSGI server runner (paster serve) which can accept PasteDeploy .ini file values as input.
  • A mechanism for rendering scaffolds into projects (paster create).

Paste is not a particularly integral part of Pyramid. It’s more or less used directly only in projects created from scaffolds. It’s possible to create a Pyramid application which does not use Paste at all. We show a Pyramid application that doesn’t use Paste in Creating Your First Pyramid Application. However, all Pyramid scaffolds use the system, to provide new developers with a standardized way of starting, stopping, and setting deployment values. This chapter is not a replacement for documentation about Paste or PasteDeploy; it only contextualizes the use of Paste within Pyramid. For detailed documentation, see http://pythonpaste.org.

PasteDeploy

PasteDeploy is the system that Pyramid uses to allow deployment settings to be spelled using an .ini configuration file format. It also allows the paster serve command to work. Its configuration format provides a convenient place to define application deployment settings and WSGI server settings, and its server runner allows you to stop and start a Pyramid application easily.

Entry Points and PasteDeploy .ini Files

In the Creating a Pyramid Project chapter, we breezed over the meaning of a configuration line in the deployment.ini file. This was the use = egg:MyProject line in the [app:main] section. We breezed over it because it’s pretty confusing and “too much information” for an introduction to the system. We’ll try to give it a bit of attention here. Let’s see the config file again:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
[app:main]
use = egg:MyProject

pyramid.reload_templates = true
pyramid.debug_authorization = false
pyramid.debug_notfound = false
pyramid.debug_routematch = false
pyramid.debug_templates = true
pyramid.default_locale_name = en
pyramid.includes = pyramid_debugtoolbar

[server:main]
use = egg:Paste#http
host = 0.0.0.0
port = 6543

# Begin logging configuration

[loggers]
keys = root, myproject

[handlers]
keys = console

[formatters]
keys = generic

[logger_root]
level = INFO
handlers = console

[logger_myproject]
level = DEBUG
handlers =
qualname = myproject

[handler_console]
class = StreamHandler
args = (sys.stderr,)
level = NOTSET
formatter = generic

[formatter_generic]
format = %(asctime)s %(levelname)-5.5s [%(name)s] %(message)s

# End logging configuration

The line in [app:main] above that says use = egg:MyProject is actually shorthand for a longer spelling: use = egg:MyProject#main. The #main part is omitted for brevity, as #main is a default defined by PasteDeploy. egg:MyProject#main is a string which has meaning to PasteDeploy. It points at a setuptools entry point named main defined in the MyProject project.

Take a look at the generated setup.py file for this project.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
import os

from setuptools import setup, find_packages

here = os.path.abspath(os.path.dirname(__file__))
README = open(os.path.join(here, 'README.txt')).read()
CHANGES = open(os.path.join(here, 'CHANGES.txt')).read()

requires = ['pyramid', 'pyramid_debugtoolbar']

setup(name='MyProject',
      version='0.0',
      description='MyProject',
      long_description=README + '\n\n' +  CHANGES,
      classifiers=[
        "Programming Language :: Python",
        "Framework :: Pylons",
        "Topic :: Internet :: WWW/HTTP",
        "Topic :: Internet :: WWW/HTTP :: WSGI :: Application",
        ],
      author='',
      author_email='',
      url='',
      keywords='web pyramid pylons',
      packages=find_packages(),
      include_package_data=True,
      zip_safe=False,
      install_requires=requires,
      tests_require=requires,
      test_suite="myproject",
      entry_points = """\
      [paste.app_factory]
      main = myproject:main
      """,
      paster_plugins=['pyramid'],
      )

Note that the entry_point line in setup.py points at a string which looks a lot like an .ini file. This string representation of an .ini file has a section named [paste.app_factory]. Within this section, there is a key named main (the entry point name) which has a value myproject:main. The key main is what our egg:MyProject#main value of the use section in our config file is pointing at, although it is actually shortened to egg:MyProject there. The value represents a dotted Python name path, which refers to a callable in our myproject package’s __init__.py module.

The egg: prefix in egg:MyProject indicates that this is an entry point URI specifier, where the “scheme” is “egg”. An “egg” is created when you run setup.py install or setup.py develop within your project.

In English, this entry point can thus be referred to as a “Paste application factory in the MyProject project which has the entry point named main where the entry point refers to a main function in the mypackage module”. Indeed, if you open up the __init__.py module generated within any scaffold-generated package, you’ll see a main function. This is the function called by PasteDeploy when the paster serve command is invoked against our application. It accepts a global configuration object and returns an instance of our application.

[DEFAULTS] Section of a PasteDeploy .ini File

You can add a [DEFAULT] section to your PasteDeploy .ini file. Such a section should consists of global parameters that are shared by all the applications, servers and middleware defined within the configuration file. The values in a [DEFAULT] section will be passed to your application’s main function as global_config (see the reference to the main function in __init__.py).

Command-Line Pyramid

Your Pyramid application can be controlled and inspected using a variety of command-line utilities. These utilities are documented in this chapter.

Displaying Matching Views for a Given URL

For a big application with several views, it can be hard to keep the view configuration details in your head, even if you defined all the views yourself. You can use the paster pviews command in a terminal window to print a summary of matching routes and views for a given URL in your application. The paster pviews command accepts two arguments. The first argument to pviews is the path to your application’s .ini file and section name inside the .ini file which points to your application. This should be of the format config_file#section_name. The second argument is the URL to test for matching views. The section_name may be omitted; if it is, it’s considered to be main.

Here is an example for a simple view configuration using traversal:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
$ ../bin/paster pviews development.ini#tutorial /FrontPage

URL = /FrontPage

    context: <tutorial.models.Page object at 0xa12536c>
    view name:

    View:
    -----
    tutorial.views.view_page
    required permission = view

The output always has the requested URL at the top and below that all the views that matched with their view configuration details. In this example only one view matches, so there is just a single View section. For each matching view, the full code path to the associated view callable is shown, along with any permissions and predicates that are part of that view configuration.

A more complex configuration might generate something like this:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
$ ../bin/paster pviews development.ini#shootout /about

URL = /about

    context: <shootout.models.RootFactory object at 0xa56668c>
    view name: about

    Route:
    ------
    route name: about
    route pattern: /about
    route path: /about
    subpath:
    route predicates (request method = GET)

        View:
        -----
        shootout.views.about_view
        required permission = view
        view predicates (request_param testing, header X/header)

    Route:
    ------
    route name: about_post
    route pattern: /about
    route path: /about
    subpath:
    route predicates (request method = POST)

        View:
        -----
        shootout.views.about_view_post
        required permission = view
        view predicates (request_param test)

        View:
        -----
        shootout.views.about_view_post2
        required permission = view
        view predicates (request_param test2)

In this case, we are dealing with a URL dispatch application. This specific URL has two matching routes. The matching route information is displayed first, followed by any views that are associated with that route. As you can see from the second matching route output, a route can be associated with more than one view.

For a URL that doesn’t match any views, paster pviews will simply print out a Not found message.

The Interactive Shell

Once you’ve installed your program for development using setup.py develop, you can use an interactive Python shell to execute expressions in a Python environment exactly like the one that will be used when your application runs “for real”. To do so, use the paster pshell command.

The argument to pshell follows the format config_file#section_name where config_file is the path to your application’s .ini file and section_name is the app section name inside the .ini file which points to your application. For example, if your application .ini file might have a [app:main] section that looks like so:

1
2
3
4
5
6
7
[app:main]
use = egg:MyProject
pyramid.reload_templates = true
pyramid.debug_authorization = false
pyramid.debug_notfound = false
pyramid.debug_templates = true
pyramid.default_locale_name = en

If so, you can use the following command to invoke a debug shell using the name MyProject as a section name:

chrism@thinko env26]$ bin/paster pshell starter/development.ini#MyProject
Python 2.6.5 (r265:79063, Apr 29 2010, 00:31:32)
[GCC 4.4.3] on linux2
Type "help" for more information.

Environment:
  app          The WSGI application.
  registry     Active Pyramid registry.
  request      Active request object.
  root         Root of the default resource tree.
  root_factory Default root factory used to create `root`.

>>> root
<myproject.resources.MyResource object at 0x445270>
>>> registry
<Registry myproject>
>>> registry.settings['pyramid.debug_notfound']
False
>>> from myproject.views import my_view
>>> from pyramid.request import Request
>>> r = Request.blank('/')
>>> my_view(r)
{'project': 'myproject'}

The WSGI application that is loaded will be available in the shell as the app global. Also, if the application that is loaded is the Pyramid app with no surrounding middleware, the root object returned by the default root factory, registry, and request will be available.

You can also simply rely on the main default section name by omitting any hash after the filename:

chrism@thinko env26]$ bin/paster pshell starter/development.ini

Press Ctrl-D to exit the interactive shell (or Ctrl-Z on Windows).

Extending the Shell

It is convenient when using the interactive shell often to have some variables significant to your application already loaded as globals when you start the pshell. To facilitate this, pshell will look for a special [pshell] section in your INI file and expose the subsequent key/value pairs to the shell. Each key is a variable name that will be global within the pshell session; each value is a dotted Python name. If specified, the special key setup should be a dotted Python name pointing to a callable that accepts the dictionary of globals that will be loaded into the shell. This allows for some custom initializing code to be executed each time the pshell is run. The setup callable can also be specified from the commandline using the --setup option which will override the key in the INI file.

For example, you want to expose your model to the shell, along with the database session so that you can mutate the model on an actual database. Here, we’ll assume your model is stored in the myapp.models package.

1
2
3
4
5
[pshell]
setup = myapp.lib.pshell.setup
m = myapp.models
session = myapp.models.DBSession
t = transaction

By defining the setup callable, we will create the module myapp.lib.pshell containing a callable named setup that will receive the global environment before it is exposed to the shell. Here we mutate the environment’s request as well as add a new value containing a WebTest version of the application to which we can easily submit requests.

1
2
3
4
5
6
7
# myapp/lib/pshell.py
from webtest import TestApp

def setup(env):
    env['request'].host = 'www.example.com'
    env['request'].scheme = 'https'
    env['testapp'] = TestApp(env['app'])

When this INI file is loaded, the extra variables m, session and t will be available for use immediately. Since a setup callable was also specified, it is executed and a new variable testapp is exposed, and the request is configured to generate urls from the host http://www.example.com. For example:

chrism@thinko env26]$ bin/paster pshell starter/development.ini
Python 2.6.5 (r265:79063, Apr 29 2010, 00:31:32)
[GCC 4.4.3] on linux2
Type "help" for more information.

Environment:
  app          The WSGI application.
  registry     Active Pyramid registry.
  request      Active request object.
  root         Root of the default resource tree.
  root_factory Default root factory used to create `root`.
  testapp      <webtest.TestApp object at ...>

Custom Variables:
  m            myapp.models
  session      myapp.models.DBSession
  t            transaction

>>> testapp.get('/')
<200 OK text/html body='<!DOCTYPE...l>\n'/3337>
>>> request.route_url('home')
'https://www.example.com/'
IPython or bpython

If you have IPython or bpython or both installed in the interpreter you use to invoke the pshell command, pshell will autodiscover them and use the first respectively found in this order : IPython, bpython, standard Python interpreter. However you could specifically invoke one of your choice with the -p choice or --python-shell choice option.

[chrism@vitaminf shellenv]$ ../bin/pshell -p ipython | bpython | python \
                             development.ini#MyProject

Displaying All Application Routes

You can use the paster proutes command in a terminal window to print a summary of routes related to your application. Much like the paster pshell command (see The Interactive Shell), the paster proutes command accepts one argument with the format config_file#section_name. The config_file is the path to your application’s .ini file, and section_name is the app section name inside the .ini file which points to your application. By default, the section_name is main and can be omitted.

For example:

1
2
3
4
5
6
7
8
[chrism@thinko MyProject]$ ../bin/paster proutes development.ini#MyProject
Name            Pattern                        View
----            -------                        ----
home            /                              <function my_view>
home2           /                              <function my_view>
another         /another                       None
static/         static/*subpath                <static_view object>
catchall        /*subpath                      <function static_view>

paster proutes generates a table. The table has three columns: a Name column, a Pattern column, and a View column. The items listed in the Name column are route names, the items listed in the Pattern column are route patterns, and the items listed in the View column are representations of the view callable that will be invoked when a request matches the associated route pattern. The view column may show None if no associated view callable could be found. If no routes are configured within your application, nothing will be printed to the console when paster proutes is executed.

Displaying “Tweens”

A tween is a bit of code that sits between the main Pyramid application request handler and the WSGI application which calls it. A user can get a representation of both the implicit tween ordering (the ordering specified by calls to pyramid.config.Configurator.add_tween()) and the explicit tween ordering (specified by the pyramid.tweens configuration setting) orderings using the paster ptweens command. Tween factories will show up represented by their standard Python dotted name in the paster ptweens output.

For example, here’s the paster pwteens command run against a system configured without any explicit tweens:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
[chrism@thinko pyramid]$ paster ptweens development.ini
"pyramid.tweens" config value NOT set (implicitly ordered tweens used)

Implicit Tween Chain

Position    Name                                                Alias
--------    ----                                                -----
-           -                                                   INGRESS
0           pyramid_debugtoolbar.toolbar.toolbar_tween_factory  pdbt
1           pyramid.tweens.excview_tween_factory                excview
-           -                                                   MAIN

Here’s the paster pwteens command run against a system configured with explicit tweens defined in its development.ini file:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
[chrism@thinko pyramid]$ paster ptweens development.ini
"pyramid.tweens" config value set (explicitly ordered tweens used)

Explicit Tween Chain (used)

Position    Name
--------    ----
-           INGRESS
0           starter.tween_factory2
1           starter.tween_factory1
2           pyramid.tweens.excview_tween_factory
-           MAIN

Implicit Tween Chain (not used)

Position    Name                                                Alias
--------    ----                                                -----
-           -                                                   INGRESS
0           pyramid_debugtoolbar.toolbar.toolbar_tween_factory  pdbt
1           pyramid.tweens.excview_tween_factory                excview
-           -                                                   MAIN

Here’s the application configuration section of the development.ini used by the above paster ptweens command which reprorts that the explicit tween chain is used:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
[app:main]
use = egg:starter
reload_templates = true
debug_authorization = false
debug_notfound = false
debug_routematch = false
debug_templates = true
default_locale_name = en
pyramid.include = pyramid_debugtoolbar
pyramid.tweens = starter.tween_factory2
                 starter.tween_factory1
                 pyramid.tweens.excview_tween_factory

See Registering “Tweens” for more information about tweens.

Writing a Script

All web applications are, at their hearts, systems which accept a request and return a response. When a request is accepted by a Pyramid application, the system receives state from the request which is later relied on by your application code. For example, one view callable may assume it’s working against a request that has a request.matchdict of a particular composition, while another assumes a different composition of the matchdict.

In the meantime, it’s convenient to be able to write a Python script that can work “in a Pyramid environment”, for instance to update database tables used by your Pyramid application. But a “real” Pyramid environment doesn’t have a completely static state independent of a request; your application (and Pyramid itself) is almost always reliant on being able to obtain information from a request. When you run a Python script that simply imports code from your application and tries to run it, there just is no request data, because there isn’t any real web request. Therefore some parts of your application and some Pyramid APIs will not work.

For this reason, Pyramid makes it possible to run a script in an environment much like the environment produced when a particular request reaches your Pyramid application. This is achieved by using the pyramid.paster.bootstrap() command in the body of your script.

Note

This feature is new as of Pyramid 1.1.

In the simplest case, pyramid.paster.bootstrap() can be used with a single argument, which accepts the PasteDeploy .ini file representing Pyramid your application configuration as a single argument:

from pyramid.paster import bootstrap
env = bootstrap('/path/to/my/development.ini')
print env['request'].route_url('home')

pyramid.paster.bootstrap() returns a dictionary containing framework-related information. This dictionary will always contain a request object as its request key.

The following keys are available in the env dictionary returned by pyramid.paster.bootstrap():

request

A pyramid.request.Request object implying the current request state for your script.

app

The WSGI application object generated by bootstrapping.

root

The resource root of your Pyramid application. This is an object generated by the root factory configured in your application.

registry

The application registry of your Pyramid application.

closer

A parameterless callable that can be used to pop an internal Pyramid threadlocal stack (used by pyramid.threadlocal.get_current_registry() and pyramid.threadlocal.get_current_request()) when your scripting job is finished.

Let’s assume that the /path/to/my/development.ini file used in the example above looks like so:

[pipeline:main]
pipeline = translogger
           another

[filter:translogger]
filter_app_factory = egg:Paste#translogger
setup_console_handler = False
logger_name = wsgi

[app:another]
use = egg:MyProject

The configuration loaded by the above bootstrap example will use the configuration implied by the [pipeline:main] section of your configuration file by default. Specifying /path/to/my/development.ini is logically equivalent to specifying /path/to/my/development.ini#main. In this case, we’ll be using a configuration that includes an app object which is wrapped in the Paste “translogger” middleware (which logs requests to the console).

You can also specify a particular section of the PasteDeploy .ini file to load instead of main:

from pyramid.paster import bootstrap
env = bootstrap('/path/to/my/development.ini#another')
print env['request'].route_url('home')

The above example specifies the another app, pipeline, or composite section of your PasteDeploy configuration file. The app object present in the env dictionary returned by pyramid.paster.bootstrap() will be a Pyramid router.

Changing the Request

By default, Pyramid will generate a request object in the env dictionary for the URL http://localhost:80/. This means that any URLs generated by Pyramid during the execution of your script will be anchored here. This is generally not what you want.

So how do we make Pyramid generate the correct URLs?

Assuming that you have a route configured in your application like so:

config.add_route('verify', '/verify/{code}')

You need to inform the Pyramid environment that the WSGI application is handling requests from a certain base. For example, we want to mount our application at example.com/prefix and the generated URLs should use HTTPS. This can be done by mutating the request object:

from pyramid.paster import bootstrap
env = bootstrap('/path/to/my/development.ini#another')
env['request'].host = 'example.com'
env['request'].scheme = 'https'
env['request'].script_name = '/prefix'
print env['request'].application_url
# will print 'https://example.com/prefix/another/url'

Now you can readily use Pyramid’s APIs for generating URLs:

env['request'].route_url('verify', code='1337')
# will return 'https://example.com/prefix/verify/1337'
Cleanup

When your scripting logic finishes, it’s good manners (but not required) to call the closer callback:

from pyramid.paster import bootstrap
env = bootstrap('/path/to/my/development.ini')

# .. do stuff ...

env['closer']()
Setting Up Logging

By default, pyramid.paster.bootstrap() does not configure logging parameters present in the configuration file. If you’d like to configure logging based on [logger] and related sections in the configuration file, use the following command:

import logging.config
logging.config.fileConfig('/path/to/my/development.ini')

Internationalization and Localization

Internationalization (i18n) is the act of creating software with a user interface that can potentially be displayed in more than one language or cultural context. Localization (l10n) is the process of displaying the user interface of an internationalized application in a particular language or cultural context.

Pyramid offers internationalization and localization subsystems that can be used to translate the text of buttons, error messages and other software- and template-defined values into the native language of a user of your application.

Creating a Translation String

While you write your software, you can insert specialized markup into your Python code that makes it possible for the system to translate text values into the languages used by your application’s users. This markup creates a translation string. A translation string is an object that behaves mostly like a normal Unicode object, except that it also carries around extra information related to its job as part of the Pyramid translation machinery.

Using The TranslationString Class

The most primitive way to create a translation string is to use the pyramid.i18n.TranslationString callable:

1
2
from pyramid.i18n import TranslationString
ts = TranslationString('Add')

This creates a Unicode-like object that is a TranslationString.

Note

For people more familiar with Zope i18n, a TranslationString is a lot like a zope.i18nmessageid.Message object. It is not a subclass, however. For people more familiar with Pylons or Django i18n, using a TranslationString is a lot like using “lazy” versions of related gettext APIs.

The first argument to TranslationString is the msgid; it is required. It represents the key into the translation mappings provided by a particular localization. The msgid argument must be a Unicode object or an ASCII string. The msgid may optionally contain replacement markers. For instance:

1
2
from pyramid.i18n import TranslationString
ts = TranslationString('Add ${number}')

Within the string above, ${number} is a replacement marker. It will be replaced by whatever is in the mapping for a translation string. The mapping may be supplied at the same time as the replacement marker itself:

1
2
from pyramid.i18n import TranslationString
ts = TranslationString('Add ${number}', mapping={'number':1})

Any number of replacement markers can be present in the msgid value, any number of times. Only markers which can be replaced by the values in the mapping will be replaced at translation time. The others will not be interpolated and will be output literally.

A translation string should also usually carry a domain. The domain represents a translation category to disambiguate it from other translations of the same msgid, in case they conflict.

1
2
3
from pyramid.i18n import TranslationString
ts = TranslationString('Add ${number}', mapping={'number':1},
                       domain='form')

The above translation string named a domain of form. A translator function will often use the domain to locate the right translator file on the filesystem which contains translations for a given domain. In this case, if it were trying to translate our msgid to German, it might try to find a translation from a gettext file within a translation directory like this one:

locale/de/LC_MESSAGES/form.mo

In other words, it would want to take translations from the form.mo translation file in the German language.

Finally, the TranslationString constructor accepts a default argument. If a default argument is supplied, it replaces usages of the msgid as the default value for the translation string. When default is None, the msgid value passed to a TranslationString is used as an implicit message identifier. Message identifiers are matched with translations in translation files, so it is often useful to create translation strings with “opaque” message identifiers unrelated to their default text:

1
2
3
from pyramid.i18n import TranslationString
ts = TranslationString('add-number', default='Add ${number}',
                        domain='form', mapping={'number':1})

When default text is used, Default text objects may contain replacement values.

Using the TranslationStringFactory Class

Another way to generate a translation string is to use the TranslationStringFactory object. This object is a translation string factory. Basically a translation string factory presets the domain value of any translation string generated by using it. For example:

1
2
3
from pyramid.i18n import TranslationStringFactory
_ = TranslationStringFactory('pyramid')
ts = _('Add ${number}', msgid='add-number', mapping={'number':1})

Note

We assigned the translation string factory to the name _. This is a convention which will be supported by translation file generation tools.

After assigning _ to the result of a TranslationStringFactory(), the subsequent result of calling _ will be a TranslationString instance. Even though a domain value was not passed to _ (as would have been necessary if the TranslationString constructor were used instead of a translation string factory), the domain attribute of the resulting translation string will be pyramid. As a result, the previous code example is completely equivalent (except for spelling) to:

1
2
3
from pyramid.i18n import TranslationString as _
ts = _('Add ${number}', msgid='add-number', mapping={'number':1},
       domain='pyramid')

You can set up your own translation string factory much like the one provided above by using the TranslationStringFactory class. For example, if you’d like to create a translation string factory which presets the domain value of generated translation strings to form, you’d do something like this:

1
2
3
from pyramid.i18n import TranslationStringFactory
_ = TranslationStringFactory('form')
ts = _('Add ${number}', msgid='add-number', mapping={'number':1})

Creating a unique domain for your application via a translation string factory is best practice. Using your own unique translation domain allows another person to reuse your application without needing to merge your translation files with his own. Instead, he can just include your package’s translation directory via the pyramid.config.Configurator.add_translation_dirs() method.

Note

For people familiar with Zope internationalization, a TranslationStringFactory is a lot like a zope.i18nmessageid.MessageFactory object. It is not a subclass, however.

Working With gettext Translation Files

The basis of Pyramid translation services is GNU gettext. Once your application source code files and templates are marked up with translation markers, you can work on translations by creating various kinds of gettext files.

Note

The steps a developer must take to work with gettext message catalog files within a Pyramid application are very similar to the steps a Pylons developer must take to do the same. See the Pylons internationalization documentation for more information.

GNU gettext uses three types of files in the translation framework, .pot files, .po files and .mo files.

.pot (Portable Object Template) files

A .pot file is created by a program which searches through your project’s source code and which picks out every message identifier passed to one of the _() functions (eg. translation string constructions). The list of all message identifiers is placed into a .pot file, which serves as a template for creating .po files.

.po (Portable Object) files

The list of messages in a .pot file are translated by a human to a particular language; the result is saved as a .po file.

.mo (Machine Object) files

A .po file is turned into a machine-readable binary file, which is the .mo file. Compiling the translations to machine code makes the localized program run faster.

The tools for working with gettext translation files related to a Pyramid application is Babel and Lingua. Lingua is a Balel extension that provides support for scraping i18n references out of Python and Chameleon files.

Installing Babel and Lingua

In order for the commands related to working with gettext translation files to work properly, you will need to have Babel and Lingua installed into the same environment in which Pyramid is installed.

Installation on UNIX

If the virtualenv into which you’ve installed your Pyramid application lives in /my/virtualenv, you can install Babel and Lingua like so:

$ cd /my/virtualenv
$ bin/easy_install Babel lingua
Installation on Windows

If the virtualenv into which you’ve installed your Pyramid application lives in C:\my\virtualenv, you can install Babel and Lingua like so:

C> cd \my\virtualenv
C> Scripts\easy_install Babel lingua
Changing the setup.py

You need to add a few boilerplate lines to your application’s setup.py file in order to properly generate gettext files from your application.

Note

See Creating a Pyramid Project to learn about about the composition of an application’s setup.py file.

In particular, add the Babel and lingua distributions to the install_requires list and insert a set of references to Babel message extractors within the call to setuptools.setup() inside your application’s setup.py file:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
 setup(name="mypackage",
       # ...
       install_requires = [
             # ...
             'Babel',
             'lingua',
             ],
       message_extractors = { '.': [
             ('**.py',   'lingua_python', None ),
             ('**.pt',   'lingua_xml', None ),
             ]},
       )

The message_extractors stanza placed into the setup.py file causes the Babel message catalog extraction machinery to also consider *.pt files when doing message id extraction.

Extracting Messages from Code and Templates

Once Babel and Lingua are installed and your application’s setup.py file has the correct message extractor references, you may extract a message catalog template from the code and Chameleon templates which reside in your Pyramid application. You run a setup.py command to extract the messages:

$ cd /place/where/myapplication/setup.py/lives
$ mkdir -p myapplication/locale
$ python setup.py extract_messages

The message catalog .pot template will end up in:

myapplication/locale/myapplication.pot.

Translation Domains

The name myapplication above in the filename myapplication.pot denotes the translation domain of the translations that must be performed to localize your application. By default, the translation domain is the project name of your Pyramid application.

To change the translation domain of the extracted messages in your project, edit the setup.cfg file of your application, The default setup.cfg file of a Paster-generated Pyramid application has stanzas in it that look something like the following:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
 [compile_catalog]
 directory = myproject/locale
 domain = MyProject
 statistics = true

 [extract_messages]
 add_comments = TRANSLATORS:
 output_file = myproject/locale/MyProject.pot
 width = 80

 [init_catalog]
 domain = MyProject
 input_file = myproject/locale/MyProject.pot
 output_dir = myproject/locale

 [update_catalog]
 domain = MyProject
 input_file = myproject/locale/MyProject.pot
 output_dir = myproject/locale
 previous = true

In the above example, the project name is MyProject. To indicate that you’d like the domain of your translations to be mydomain instead, change the setup.cfg file stanzas to look like so:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
 [compile_catalog]
 directory = myproject/locale
 domain = mydomain
 statistics = true

 [extract_messages]
 add_comments = TRANSLATORS:
 output_file = myproject/locale/mydomain.pot
 width = 80

 [init_catalog]
 domain = mydomain
 input_file = myproject/locale/mydomain.pot
 output_dir = myproject/locale

 [update_catalog]
 domain = mydomain
 input_file = myproject/locale/mydomain.pot
 output_dir = myproject/locale
 previous = true
Initializing a Message Catalog File

Once you’ve extracted messages into a .pot file (see Extracting Messages from Code and Templates), to begin localizing the messages present in the .pot file, you need to generate at least one .po file. A .po file represents translations of a particular set of messages to a particular locale. Initialize a .po file for a specific locale from a pre-generated .pot template by using the setup.py init_catalog command:

$ cd /place/where/myapplication/setup.py/lives
$ python setup.py init_catalog -l es

By default, the message catalog .po file will end up in:

myapplication/locale/es/LC_MESSAGES/myapplication.po.

Once the file is there, it can be worked on by a human translator. One tool which may help with this is Poedit.

Note that Pyramid itself ignores the existence of all .po files. For a running application to have translations available, a .mo file must exist. See Compiling a Message Catalog File.

Updating a Catalog File

If more translation strings are added to your application, or translation strings change, you will need to update existing .po files based on changes to the .pot file, so that the new and changed messages can also be translated or re-translated.

First, regenerate the .pot file as per Extracting Messages from Code and Templates. Then use the setup.py update_catalog command.

$ cd /place/where/myapplication/setup.py/lives
$ python setup.py update_catalog
Compiling a Message Catalog File

Finally, to prepare an application for performing actual runtime translations, compile .po files to .mo files:

$ cd /place/where/myapplication/setup.py/lives
$ python setup.py compile_catalog

This will create a .mo file for each .po file in your application. As long as the translation directory in which the .mo file ends up in is configured into your application, these translations will be available to Pyramid.

Using a Localizer

A localizer is an object that allows you to perform translation or pluralization “by hand” in an application. You may use the pyramid.i18n.get_localizer() function to obtain a localizer. This function will return either the localizer object implied by the active locale negotiator or a default localizer object if no explicit locale negotiator is registered.

1
2
3
4
from pyramid.i18n import get_localizer

def aview(request):
    locale = get_localizer(request)

Note

If you need to create a localizer for a locale use the pyramid.i18n.make_localizer() function.

Performing a Translation

A localizer has a translate method which accepts either a translation string or a Unicode string and which returns a Unicode object representing the translation. So, generating a translation in a view component of an application might look like so:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
from pyramid.i18n import get_localizer
from pyramid.i18n import TranslationString

ts = TranslationString('Add ${number}', mapping={'number':1},
                       domain='pyramid')

def aview(request):
    localizer = get_localizer(request)
    translated = localizer.translate(ts) # translation string
    # ... use translated ...

The get_localizer() function will return a pyramid.i18n.Localizer object bound to the locale name represented by the request. The translation returned from its pyramid.i18n.Localizer.translate() method will depend on the domain attribute of the provided translation string as well as the locale of the localizer.

Note

If you’re using Chameleon templates, you don’t need to pre-translate translation strings this way. See Chameleon Template Support for Translation Strings.

Performing a Pluralization

A localizer has a pluralize method with the following signature:

1
2
def pluralize(singular, plural, n, domain=None, mapping=None):
    ...

The singular and plural arguments should each be a Unicode value representing a message identifier. n should be an integer. domain should be a translation domain, and mapping should be a dictionary that is used for replacement value interpolation of the translated string. If n is plural for the current locale, pluralize will return a Unicode translation for the message id plural, otherwise it will return a Unicode translation for the message id singular.

The arguments provided as singular and/or plural may also be translation string objects, but the domain and mapping information attached to those objects is ignored.

1
2
3
4
5
6
from pyramid.i18n import get_localizer

def aview(request):
    localizer = get_localizer(request)
    translated = localizer.pluralize('Item', 'Items', 1, 'mydomain')
    # ... use translated ...

Obtaining the Locale Name for a Request

You can obtain the locale name related to a request by using the pyramid.i18n.get_locale_name() function.

1
2
3
4
from pyramid.i18n import get_locale_name

def aview(request):
    locale_name = get_locale_name(request)

This returns the locale name negotiated by the currently active locale negotiator or the default locale name if the locale negotiator returns None. You can change the default locale name by changing the pyramid.default_locale_name setting; see Default Locale Name.

Once get_locale_name() is first run, the locale name is stored on the request object. Subsequent calls to get_locale_name() will return the stored locale name without invoking the locale negotiator. To avoid this caching, you can use the pyramid.i18n.negotiate_locale_name() function:

1
2
3
4
from pyramid.i18n import negotiate_locale_name

def aview(request):
    locale_name = negotiate_locale_name(request)

You can also obtain the locale name related to a request using the locale_name attribute of a localizer.

1
2
3
4
5
from pyramid.i18n import get_localizer

def aview(request):
    localizer = get_localizer(request)
    locale_name = localizer.locale_name

Obtaining the locale name as an attribute of a localizer is equivalent to obtaining a locale name by calling the get_locale_name() function.

Performing Date Formatting and Currency Formatting

Pyramid does not itself perform date and currency formatting for different locales. However, Babel can help you do this via the babel.core.Locale class. The Babel documentation for this class provides minimal information about how to perform date and currency related locale operations. See Installing Babel and Lingua for information about how to install Babel.

The babel.core.Locale class requires a locale name as an argument to its constructor. You can use Pyramid APIs to obtain the locale name for a request to pass to the babel.core.Locale constructor; see Obtaining the Locale Name for a Request. For example:

1
2
3
4
5
6
from babel.core import Locale
from pyramid.i18n import get_locale_name

def aview(request):
    locale_name = get_locale_name(request)
    locale = Locale(locale_name)

Chameleon Template Support for Translation Strings

When a translation string is used as the subject of textual rendering by a Chameleon template renderer, it will automatically be translated to the requesting user’s language if a suitable translation exists. This is true of both the ZPT and text variants of the Chameleon template renderers.

For example, in a Chameleon ZPT template, the translation string represented by “some_translation_string” in each example below will go through translation before being rendered:

1
<span tal:content="some_translation_string"/>

1
<span tal:replace="some_translation_string"/>

1
<span>${some_translation_string}</span>

1
<a tal:attributes="href some_translation_string">Click here</a>

The features represented by attributes of the i18n namespace of Chameleon will also consult the Pyramid translations. See http://chameleon.repoze.org/docs/latest/i18n.html#the-i18n-namespace.

Note

Unlike when Chameleon is used outside of Pyramid, when it is used within Pyramid, it does not support use of the zope.i18n translation framework. Applications which use Pyramid should use the features documented in this chapter rather than zope.i18n.

Third party Pyramid template renderers might not provide this support out of the box and may need special code to do an equivalent. For those, you can always use the more manual translation facility described in Performing a Translation.

Mako Pyramid I18N Support

There exists a recipe within the Pyramid Cookbook named “Mako Internationalization” which explains how to add idiomatic I18N support to Mako templates.

“Detecting” Available Languages

Other systems provide an API that returns the set of “available languages” as indicated by the union of all languages in all translation directories on disk at the time of the call to the API.

It is by design that Pyramid doesn’t supply such an API. Instead, the application itself is responsible for knowing the “available languages”. The rationale is this: any particular application deployment must always know which languages it should be translatable to anyway, regardless of which translation files are on disk.

Here’s why: it’s not a given that because translations exist in a particular language within the registered set of translation directories that this particular deployment wants to allow translation to that language. For example, some translations may exist but they may be incomplete or incorrect. Or there may be translations to a language but not for all translation domains.

Any nontrivial application deployment will always need to be able to selectively choose to allow only some languages even if that set of languages is smaller than all those detected within registered translation directories. The easiest way to allow for this is to make the application entirely responsible for knowing which languages are allowed to be translated to instead of relying on the framework to divine this information from translation directory file info.

You can set up a system to allow a deployer to select available languages based on convention by using the pyramid.settings mechanism:

Allow a deployer to modify your application’s PasteDeploy .ini file:

1
2
3
4
[app:main]
use = egg:MyProject
# ...
available_languages = fr de en ru

Then as a part of the code of a custom locale negotiator:

1
2
3
from pyramid.threadlocal import get_current_registry
settings = get_current_registry().settings
languages = settings['available_languages'].split()

This is only a suggestion. You can create your own “available languages” configuration scheme as necessary.

Activating Translation

By default, a Pyramid application performs no translation. To turn translation on, you must:

Adding a Translation Directory

gettext is the underlying machinery behind the Pyramid translation machinery. A translation directory is a directory organized to be useful to gettext. A translation directory usually includes a listing of language directories, each of which itself includes an LC_MESSAGES directory. Each LC_MESSAGES directory should contain one or more .mo files. Each .mo file represents a message catalog, which is used to provide translations to your application.

Adding a translation directory registers all of its constituent message catalog files within your Pyramid application to be available to use for translation services. This includes all of the .mo files found within all LC_MESSAGES directories within each locale directory in the translation directory.

You can add a translation directory imperatively by using the pyramid.config.Configurator.add_translation_dirs() during application startup. For example:

1
2
3
from pyramid.config import Configurator
config.add_translation_dirs('my.application:locale/',
                            'another.application:locale/')

A message catalog in a translation directory added via add_translation_dirs() will be merged into translations from a message catalog added earlier if both translation directories contain translations for the same locale and translation domain.

Setting the Locale

When the default locale negotiator (see The Default Locale Negotiator) is in use, you can inform Pyramid of the current locale name by doing any of these things before any translations need to be performed:

  • Set the _LOCALE_ attribute of the request to a valid locale name (usually directly within view code). E.g. request._LOCALE_ = 'de'.
  • Ensure that a valid locale name value is in the request.params dictionary under the key named _LOCALE_. This is usually the result of passing a _LOCALE_ value in the query string or in the body of a form post associated with a request. For example, visiting http://my.application?_LOCALE_=de.
  • Ensure that a valid locale name value is in the request.cookies dictionary under the key named _LOCALE_. This is usually the result of setting a _LOCALE_ cookie in a prior response, e.g. response.set_cookie('_LOCALE_', 'de').

Note

If this locale negotiation scheme is inappropriate for a particular application, you can configure a custom locale negotiator function into that application as required. See Using a Custom Locale Negotiator.

Locale Negotiators

A locale negotiator informs the operation of a localizer by telling it what locale name is related to a particular request. A locale negotiator is a bit of code which accepts a request and which returns a locale name. It is consulted when pyramid.i18n.Localizer.translate() or pyramid.i18n.Localizer.pluralize() is invoked. It is also consulted when get_locale_name() or negotiate_locale_name() is invoked.

The Default Locale Negotiator

Most applications can make use of the default locale negotiator, which requires no additional coding or configuration.

The default locale negotiator implementation named default_locale_negotiator uses the following set of steps to dermine the locale name.

  • First, the negotiator looks for the _LOCALE_ attribute of the request object (possibly set directly by view code or by a listener for an event).
  • Then it looks for the request.params['_LOCALE_'] value.
  • Then it looks for the request.cookies['_LOCALE_'] value.
  • If no locale can be found via the request, it falls back to using the default locale name (see Localization-Related Deployment Settings).
  • Finally, if the default locale name is not explicitly set, it uses the locale name en.
Using a Custom Locale Negotiator

Locale negotiation is sometimes policy-laden and complex. If the (simple) default locale negotiation scheme described in Activating Translation is inappropriate for your application, you may create and a special locale negotiator. Subsequently you may override the default locale negotiator by adding your newly created locale negotiator to your application’s configuration.

A locale negotiator is simply a callable which accepts a request and returns a single locale name or None if no locale can be determined.

Here’s an implementation of a simple locale negotiator:

1
2
3
 def my_locale_negotiator(request):
     locale_name = request.params.get('my_locale')
     return locale_name

If a locale negotiator returns None, it signifies to Pyramid that the default application locale name should be used.

You may add your newly created locale negotiator to your application’s configuration by passing an object which can act as the negotiator (or a dotted Python name referring to the object) as the locale_negotiator argument of the Configurator instance during application startup. For example:

1
2
from pyramid.config import Configurator
config = Configurator(locale_negotiator=my_locale_negotiator)

Alternately, use the pyramid.config.Configurator.set_locale_negotiator() method.

For example:

1
2
3
from pyramid.config import Configurator
config = Configurator()
config.set_locale_negotiator(my_locale_negotiator)

Virtual Hosting

“Virtual hosting” is, loosely, the act of serving a Pyramid application or a portion of a Pyramid application under a URL space that it does not “naturally” inhabit.

Pyramid provides facilities for serving an application under a URL “prefix”, as well as serving a portion of a traversal based application under a root URL.

Hosting an Application Under a URL Prefix

Pyramid supports a common form of virtual hosting whereby you can host a Pyramid application as a “subset” of some other site (e.g. under http://example.com/mypyramidapplication/ as opposed to under http://example.com/).

If you use a “pure Python” environment, this functionality is provided by Paste’s urlmap “composite” WSGI application. Alternately, you can use mod_wsgi to serve your application, which handles this virtual hosting translation for you “under the hood”.

If you use the urlmap composite application “in front” of a Pyramid application or if you use mod_wsgi to serve up a Pyramid application, nothing special needs to be done within the application for URLs to be generated that contain a prefix. paste.urlmap and mod_wsgi manipulate the WSGI environment in such a way that the PATH_INFO and SCRIPT_NAME variables are correct for some given prefix.

Here’s an example of a PasteDeploy configuration snippet that includes a urlmap composite.

1
2
3
4
5
6
[app:mypyramidapp]
use = egg:mypyramidapp

[composite:main]
use = egg:Paste#urlmap
/pyramidapp = mypyramidapp

This “roots” the Pyramid application at the prefix /pyramidapp and serves up the composite as the “main” application in the file.

Note

If you’re using an Apache server to proxy to a Paste urlmap composite, you may have to use the ProxyPreserveHost directive to pass the original HTTP_HOST header along to the application, so URLs get generated properly. As of this writing the urlmap composite does not seem to respect the HTTP_X_FORWARDED_HOST parameter, which will contain the original host header even if HTTP_HOST is incorrect.

If you use mod_wsgi, you do not need to use a composite application in your .ini file. The WSGIScriptAlias configuration setting in a mod_wsgi configuration does the work for you:

1
WSGIScriptAlias /pyramidapp /Users/chrism/projects/modwsgi/env/pyramid.wsgi

In the above configuration, we root a Pyramid application at /pyramidapp within the Apache configuration.

Virtual Root Support

Pyramid also supports “virtual roots”, which can be used in traversal -based (but not URL dispatch -based) applications.

Virtual root support is useful when you’d like to host some resource in a Pyramid resource tree as an application under a URL pathname that does not include the resource path itself. For example, you might want to serve the object at the traversal path /cms as an application reachable via http://example.com/ (as opposed to http://example.com/cms).

To specify a virtual root, cause an environment variable to be inserted into the WSGI environ named HTTP_X_VHM_ROOT with a value that is the absolute pathname to the resource object in the resource tree that should behave as the “root” resource. As a result, the traversal machinery will respect this value during traversal (prepending it to the PATH_INFO before traversal starts), and the pyramid.request.Request.resource_url() API will generate the “correct” virtually-rooted URLs.

An example of an Apache mod_proxy configuration that will host the /cms subobject as http://www.example.com/ using this facility is below:

1
2
3
4
5
6
7
8
9
 NameVirtualHost *:80

 <VirtualHost *:80>
   ServerName www.example.com
   RewriteEngine On
   RewriteRule ^/(.*) http://127.0.0.1:6543/$1 [L,P]
   ProxyPreserveHost on
   RequestHeader add X-Vhm-Root /cms
 </VirtualHost>

Note

Use of the RequestHeader directive requires that the Apache mod_headers module be available in the Apache environment you’re using.

For a Pyramid application running under mod_wsgi, the same can be achieved using SetEnv:

1
2
3
 <Location />
   SetEnv HTTP_X_VHM_ROOT /cms
 </Location>

Setting a virtual root has no effect when using an application based on URL dispatch.

Further Documentation and Examples

The API documentation in pyramid.traversal documents a pyramid.traversal.virtual_root() API. When called, it returns the virtual root object (or the physical root object if no virtual root has been specified).

Running a Pyramid Application under mod_wsgi has detailed information about using mod_wsgi to serve Pyramid applications.

Unit, Integration, and Functional Testing

Unit testing is, not surprisingly, the act of testing a “unit” in your application. In this context, a “unit” is often a function or a method of a class instance. The unit is also referred to as a “unit under test”.

The goal of a single unit test is to test only some permutation of the “unit under test”. If you write a unit test that aims to verify the result of a particular codepath through a Python function, you need only be concerned about testing the code that lives in the function body itself. If the function accepts a parameter that represents a complex application “domain object” (such as a resource, a database connection, or an SMTP server), the argument provided to this function during a unit test need not be and likely should not be a “real” implementation object. For example, although a particular function implementation may accept an argument that represents an SMTP server object, and the function may call a method of this object when the system is operating normally that would result in an email being sent, a unit test of this codepath of the function does not need to test that an email is actually sent. It just needs to make sure that the function calls the method of the object provided as an argument that would send an email if the argument happened to be the “real” implementation of an SMTP server object.

An integration test, on the other hand, is a different form of testing in which the interaction between two or more “units” is explicitly tested. Integration tests verify that the components of your application work together. You might make sure that an email was actually sent in an integration test.

A functional test is a form of integration test in which the application is run “literally”. You would have to make sure that an email was actually sent in a functional test, because it tests your code end to end.

It is often considered best practice to write each type of tests for any given codebase. Unit testing often provides the opportunity to obtain better “coverage”: it’s usually possible to supply a unit under test with arguments and/or an environment which causes all of its potential codepaths to be executed. This is usually not as easy to do with a set of integration or functional tests, but integration and functional testing provides a measure of assurance that your “units” work together, as they will be expected to when your application is run in production.

The suggested mechanism for unit and integration testing of a Pyramid application is the Python unittest module. Although this module is named unittest, it is actually capable of driving both unit and integration tests. A good unittest tutorial is available within Dive Into Python by Mark Pilgrim.

Pyramid provides a number of facilities that make unit, integration, and functional tests easier to write. The facilities become particularly useful when your code calls into Pyramid -related framework functions.

Test Set Up and Tear Down

Pyramid uses a “global” (actually thread local) data structure to hold on to two items: the current request and the current application registry. These data structures are available via the pyramid.threadlocal.get_current_request() and pyramid.threadlocal.get_current_registry() functions, respectively. See Thread Locals for information about these functions and the data structures they return.

If your code uses these get_current_* functions or calls Pyramid code which uses get_current_* functions, you will need to call pyramid.testing.setUp() in your test setup and you will need to call pyramid.testing.tearDown() in your test teardown. setUp() pushes a registry onto the thread local stack, which makes the get_current_* functions work. It returns a Configurator object which can be used to perform extra configuration required by the code under test. tearDown() pops the thread local stack.

Normally when a Configurator is used directly with the main block of a Pyramid application, it defers performing any “real work” until its .commit method is called (often implicitly by the pyramid.config.Configurator.make_wsgi_app() method). The Configurator returned by setUp() is an autocommitting Configurator, however, which performs all actions implied by methods called on it immediately. This is more convenient for unit-testing purposes than needing to call pyramid.config.Configurator.commit() in each test after adding extra configuration statements.

The use of the setUp() and tearDown() functions allows you to supply each unit test method in a test case with an environment that has an isolated registry and an isolated request for the duration of a single test. Here’s an example of using this feature:

1
2
3
4
5
6
7
8
9
import unittest
from pyramid import testing

class MyTest(unittest.TestCase):
    def setUp(self):
        self.config = testing.setUp()

    def tearDown(self):
        testing.tearDown()

The above will make sure that get_current_registry() called within a test case method of MyTest will return the application registry associated with the config Configurator instance. Each test case method attached to MyTest will use an isolated registry.

The setUp() and tearDown() functions accepts various arguments that influence the environment of the test. See the pyramid.testing chapter for information about the extra arguments supported by these functions.

If you also want to make get_current_request() return something other than None during the course of a single test, you can pass a request object into the pyramid.testing.setUp() within the setUp method of your test:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
import unittest
from pyramid import testing

class MyTest(unittest.TestCase):
    def setUp(self):
        request = testing.DummyRequest()
        self.config = testing.setUp(request=request)

    def tearDown(self):
        testing.tearDown()

If you pass a request object into pyramid.testing.setUp() within your test case’s setUp, any test method attached to the MyTest test case that directly or indirectly calls get_current_request() will receive the request object. Otherwise, during testing, get_current_request() will return None. We use a “dummy” request implementation supplied by pyramid.testing.DummyRequest because it’s easier to construct than a “real” Pyramid request object.

What?

Thread local data structures are always a bit confusing, especially when they’re used by frameworks. Sorry. So here’s a rule of thumb: if you don’t know whether you’re calling code that uses the get_current_registry() or get_current_request() functions, or you don’t care about any of this, but you still want to write test code, just always call pyramid.testing.setUp() in your test’s setUp method and pyramid.testing.tearDown() in your tests’ tearDown method. This won’t really hurt anything if the application you’re testing does not call any get_current* function.

Using the Configurator and pyramid.testing APIs in Unit Tests

The Configurator API and the pyramid.testing module provide a number of functions which can be used during unit testing. These functions make configuration declaration calls to the current application registry, but typically register a “stub” or “dummy” feature in place of the “real” feature that the code would call if it was being run normally.

For example, let’s imagine you want to unit test a Pyramid view function.

1
2
3
4
5
6
7
from pyramid.security import has_permission
from pyramid.httpexceptions import HTTPForbidden

def view_fn(request):
    if not has_permission('edit', request.context, request):
        raise HTTPForbidden
    return {'greeting':'hello'}

Without doing anything special during a unit test, the call to has_permission() in this view function will always return a True value. When a Pyramid application starts normally, it will populate a application registry using configuration declaration calls made against a Configurator. But if this application registry is not created and populated (e.g. by initializing the configurator with an authorization policy), like when you invoke application code via a unit test, Pyramid API functions will tend to either fail or return default results. So how do you test the branch of the code in this view function that raises HTTPForbidden?

The testing API provided by Pyramid allows you to simulate various application registry registrations for use under a unit testing framework without needing to invoke the actual application configuration implied by its main function. For example, if you wanted to test the above view_fn (assuming it lived in the package named my.package), you could write a unittest.TestCase that used the testing API.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
import unittest
from pyramid import testing

class MyTest(unittest.TestCase):
    def setUp(self):
        self.config = testing.setUp()

    def tearDown(self):
        testing.tearDown()

    def test_view_fn_forbidden(self):
        from pyramid.httpexceptions import HTTPForbidden
        from my.package import view_fn
        self.config.testing_securitypolicy(userid='hank',
                                           permissive=False)
        request = testing.DummyRequest()
        request.context = testing.DummyResource()
        self.assertRaises(HTTPForbidden, view_fn, request)

    def test_view_fn_allowed(self):
        from my.package import view_fn
        self.config.testing_securitypolicy(userid='hank',
                                           permissive=True)
        request = testing.DummyRequest()
        request.context = testing.DummyResource()
        response = view_fn(request)
        self.assertEqual(response, {'greeting':'hello'})

In the above example, we create a MyTest test case that inherits from unittest.TestCase. If it’s in our Pyramid application, it will be found when setup.py test is run. It has two test methods.

The first test method, test_view_fn_forbidden tests the view_fn when the authentication policy forbids the current user the edit permission. Its third line registers a “dummy” “non-permissive” authorization policy using the testing_securitypolicy() method, which is a special helper method for unit testing.

We then create a pyramid.testing.DummyRequest object which simulates a WebOb request object API. A pyramid.testing.DummyRequest is a request object that requires less setup than a “real” Pyramid request. We call the function being tested with the manufactured request. When the function is called, pyramid.security.has_permission() will call the “dummy” authentication policy we’ve registered through testing_securitypolicy(), which denies access. We check that the view function raises a HTTPForbidden error.

The second test method, named test_view_fn_allowed tests the alternate case, where the authentication policy allows access. Notice that we pass different values to testing_securitypolicy() to obtain this result. We assert at the end of this that the view function returns a value.

Note that the test calls the pyramid.testing.setUp() function in its setUp method and the pyramid.testing.tearDown() function in its tearDown method. We assign the result of pyramid.testing.setUp() as config on the unittest class. This is a Configurator object and all methods of the configurator can be called as necessary within tests. If you use any of the Configurator APIs during testing, be sure to use this pattern in your test case’s setUp and tearDown; these methods make sure you’re using a “fresh” application registry per test run.

See the pyramid.testing chapter for the entire Pyramid -specific testing API. This chapter describes APIs for registering a security policy, registering resources at paths, registering event listeners, registering views and view permissions, and classes representing “dummy” implementations of a request and a resource.

See also the various methods of the Configurator documented in pyramid.config that begin with the testing_ prefix.

Creating Integration Tests

In Pyramid, a unit test typically relies on “mock” or “dummy” implementations to give the code under test only enough context to run.

“Integration testing” implies another sort of testing. In the context of a Pyramid, integration test, the test logic tests the functionality of some code and its integration with the rest of the Pyramid framework.

In Pyramid applications that are plugins to Pyramid, you can create an integration test by including it’s includeme function via pyramid.config.Configurator.include() in the test’s setup code. This causes the entire Pyramid environment to be set up and torn down as if your application was running “for real”. This is a heavy-hammer way of making sure that your tests have enough context to run properly, and it tests your code’s integration with the rest of Pyramid.

Let’s demonstrate this by showing an integration test for a view. The below test assumes that your application’s package name is myapp, and that there is a views module in the app with a function with the name my_view in it that returns the response ‘Welcome to this application’ after accessing some values that require a fully set up environment.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
import unittest

from pyramid import testing

class ViewIntegrationTests(unittest.TestCase):
    def setUp(self):
        """ This sets up the application registry with the
        registrations your application declares in its ``includeme``
        function.
        """
        import myapp
        self.config = testing.setUp()
        self.config.include('myapp')

    def tearDown(self):
        """ Clear out the application registry """
        testing.tearDown()

    def test_my_view(self):
        from myapp.views import my_view
        request = testing.DummyRequest()
        result = my_view(request)
        self.assertEqual(result.status, '200 OK')
        body = result.app_iter[0]
        self.failUnless('Welcome to' in body)
        self.assertEqual(len(result.headerlist), 2)
        self.assertEqual(result.headerlist[0],
                         ('Content-Type', 'text/html; charset=UTF-8'))
        self.assertEqual(result.headerlist[1], ('Content-Length',
                                                str(len(body))))

Unless you cannot avoid it, you should prefer writing unit tests that use the Configurator API to set up the right “mock” registrations rather than creating an integration test. Unit tests will run faster (because they do less for each test) and the result of a unit test is usually easier to make assertions about.

Creating Functional Tests

Functional tests test your literal application.

The below test assumes that your application’s package name is myapp, and that there is view that returns an HTML body when the root URL is invoked. It further assumes that you’ve added a tests_require dependency on the WebTest package within your setup.py file. WebTest is a functional testing package written by Ian Bicking.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
import unittest

class FunctionalTests(unittest.TestCase):
    def setUp(self):
        from myapp import main
        app = main({})
        from webtest import TestApp
        self.testapp = TestApp(app)

    def test_root(self):
        res = self.testapp.get('/', status=200)
        self.failUnless('Pyramid' in res.body)

When this test is run, each test creates a “real” WSGI application using the main function in your myapp.__init__ module and uses WebTest to wrap that WSGI application. It assigns the result to self.testapp. In the test named test_root, we use the testapp’s get method to invoke the root URL. We then assert that the returned HTML has the string Pyramid in it.

See the WebTest documentation for further information about the methods available to a webtest.TestApp instance.

Resources

A resource is an object that represents a “place” in a tree related to your application. Every Pyramid application has at least one resource object: the root resource. Even if you don’t define a root resource manually, a default one is created for you. The root resource is the root of a resource tree. A resource tree is a set of nested dictionary-like objects which you can use to represent your website’s structure.

In an application which uses traversal to map URLs to code, the resource tree structure is used heavily to map each URL to a view callable. When traversal is used, Pyramid will walk through the resource tree by traversing through its nested dictionary structure in order to find a context resource. Once a context resource is found, the context resource and data in the request will be used to find a view callable.

In an application which uses URL dispatch, the resource tree is only used indirectly, and is often “invisible” to the developer. In URL dispatch applications, the resource “tree” is often composed of only the root resource by itself. This root resource sometimes has security declarations attached to it, but is not required to have any. In general, the resource tree is much less important in applications that use URL dispatch than applications that use traversal.

In “Zope-like” Pyramid applications, resource objects also often store data persistently, and offer methods related to mutating that persistent data. In these kinds of applications, resources not only represent the site structure of your website, but they become the domain model of the application.

Also:

  • The context and containment predicate arguments to add_view() (or a view_config() decorator) reference a resource class or resource interface.
  • A root factory returns a resource.
  • A resource is exposed to view code as the context of a view.
  • Various helpful Pyramid API methods expect a resource as an argument (e.g. resource_url() and others).

Defining a Resource Tree

When traversal is used (as opposed to a purely url dispatch based application), Pyramid expects to be able to traverse a tree composed of resources (the resource tree). Traversal begins at a root resource, and descends into the tree recursively, trying each resource’s __getitem__ method to resolve a path segment to another resource object. Pyramid imposes the following policy on resource instances in the tree:

  • A container resource (a resource which contains other resources) must supply a __getitem__ method which is willing to resolve a unicode name to a sub-resource. If a sub-resource by a particular name does not exist in a container resource, __getitem__ method of the container resource must raise a KeyError. If a sub-resource by that name does exist, the container’s __getitem__ should return the sub-resource.
  • Leaf resources, which do not contain other resources, must not implement a __getitem__, or if they do, their __getitem__ method must always raise a KeyError.

See Traversal for more information about how traversal works against resource instances.

Here’s a sample resource tree, represented by a variable named root:

1
2
3
4
 class Resource(dict):
     pass

 root = Resource({'a':Resource({'b':Resource({'c':Resource()})})})

The resource tree we’ve created above is represented by a dictionary-like root object which has a single child named 'a'. 'a' has a single child named 'b', and 'b' has a single child named 'c', which has no children. It is therefore possible to access the 'c' leaf resource like so:

1
root['a']['b']['c']

If you returned the above root object from a root factory, the path /a/b/c would find the 'c' object in the resource tree as the result of traversal.

In this example, each of the resources in the tree is of the same class. This is not a requirement. Resource elements in the tree can be of any type. We used a single class to represent all resources in the tree for the sake of simplicity, but in a “real” app, the resources in the tree can be arbitrary.

Although the example tree above can service a traversal, the resource instances in the above example are not aware of location, so their utility in a “real” application is limited. To make best use of built-in Pyramid API facilities, your resources should be “location-aware”. The next section details how to make resources location-aware.

Location-Aware Resources

In order for certain Pyramid location, security, URL-generation, and traversal APIs to work properly against the resources in a resource tree, all resources in the tree must be location -aware. This means they must have two attributes: __parent__ and __name__.

The __parent__ attribute of a location-aware resource should be a reference to the resource’s parent resource instance in the tree. The __name__ attribute should be the name with which a resource’s parent refers to the resource via __getitem__.

The __parent__ of the root resource should be None and its __name__ should be the empty string. For instance:

1
2
3
class MyRootResource(object):
    __name__ = ''
    __parent__ = None

A resource returned from the root resource’s __getitem__ method should have a __parent__ attribute that is a reference to the root resource, and its __name__ attribute should match the name by which it is reachable via the root resource’s __getitem__. A container resource within the root resource should have a __getitem__ that returns resources with a __parent__ attribute that points at the container, and these subobjects should have a __name__ attribute that matches the name by which they are retrieved from the container via __getitem__. This pattern continues recursively “up” the tree from the root.

The __parent__ attributes of each resource form a linked list that points “downwards” toward the root. This is analogous to the .. entry in filesystem directories. If you follow the __parent__ values from any resource in the resource tree, you will eventually come to the root resource, just like if you keep executing the cd .. filesystem command, eventually you will reach the filesystem root directory.

Warning

If your root resource has a __name__ argument that is not None or the empty string, URLs returned by the resource_url() function and paths generated by the resource_path() and resource_path_tuple() APIs will be generated improperly. The value of __name__ will be prepended to every path and URL generated (as opposed to a single leading slash or empty tuple element).

Using pyramid_traversalwrapper

If you’d rather not manage the __name__ and __parent__ attributes of your resources “by hand”, an add-on package named pyramid_traversalwrapper can help.

In order to use this helper feature, you must first install the pyramid_traversalwrapper package (available via PyPI), then register its ModelGraphTraverser as the traversal policy, rather than the default Pyramid traverser. The package contains instructions for doing so.

Once Pyramid is configured with this feature, you will no longer need to manage the __parent__ and __name__ attributes on resource objects “by hand”. Instead, as necessary, during traversal Pyramid will wrap each resource (even the root resource) in a LocationProxy which will dynamically assign a __name__ and a __parent__ to the traversed resource (based on the last traversed resource and the name supplied to __getitem__). The root resource will have a __name__ attribute of None and a __parent__ attribute of None.

Applications which use tree-walking Pyramid APIs require location-aware resources. These APIs include (but are not limited to) resource_url(), find_resource(), find_root(), find_interface(), resource_path(), resource_path_tuple(), or traverse(), virtual_root(), and (usually) has_permission() and principals_allowed_by_permission().

In general, since so much Pyramid infrastructure depends on location-aware resources, it’s a good idea to make each resource in your tree location-aware.

Generating The URL Of A Resource

If your resources are location aware, you can use the pyramid.request.Request.resource_url() API to generate a URL for the resource. This URL will use the resource’s position in the parent tree to create a resource path, and it will prefix the path with the current application URL to form a fully-qualified URL with the scheme, host, port, and path. You can also pass extra arguments to resource_url() to influence the generated URL.

The simplest call to resource_url() looks like this:

1
url = request.resource_url(resource)

The request in the above example is an instance of a Pyramid request object.

If the resource referred to as resource in the above example was the root resource, and the host that was used to contact the server was example.com, the URL generated would be http://example.com/. However, if the resource was a child of the root resource named a, the generated URL would be http://example.com/a/.

A slash is appended to all resource URLs when resource_url() is used to generate them in this simple manner, because resources are “places” in the hierarchy, and URLs are meant to be clicked on to be visited. Relative URLs that you include on HTML pages rendered as the result of the default view of a resource are more apt to be relative to these resources than relative to their parent.

You can also pass extra elements to resource_url():

1
url = request.resource_url(resource, 'foo', 'bar')

If the resource referred to as resource in the above example was the root resource, and the host that was used to contact the server was example.com, the URL generated would be http://example.com/foo/bar. Any number of extra elements can be passed to resource_url() as extra positional arguments. When extra elements are passed, they are appended to the resource’s URL. A slash is not appended to the final segment when elements are passed.

You can also pass a query string:

1
url = request.resource_url(resource, query={'a':'1'})

If the resource referred to as resource in the above example was the root resource, and the host that was used to contact the server was example.com, the URL generated would be http://example.com/?a=1.

When a virtual root is active, the URL generated by resource_url() for a resource may be “shorter” than its physical tree path. See Virtual Root Support for more information about virtually rooting a resource.

For more information about generating resource URLs, see the documentation for pyramid.request.Request.resource_url().

Overriding Resource URL Generation

If a resource object implements a __resource_url__ method, this method will be called when resource_url() is called to generate a URL for the resource, overriding the default URL returned for the resource by resource_url().

The __resource_url__ hook is passed two arguments: request and info. request is the request object passed to resource_url(). info is a dictionary with two keys:

physical_path
The “physical path” computed for the resource, as defined by pyramid.traversal.resource_path(resource).
virtual_path
The “virtual path” computed for the resource, as defined by Virtual Root Support. This will be identical to the physical path if virtual rooting is not enabled.

The __resource_url__ method of a resource should return a string representing a URL. If it cannot override the default, it should return None. If it returns None, the default URL will be returned.

Here’s an example __resource_url__ method.

1
2
3
class Resource(object):
    def __resource_url__(self, request, info):
        return request.application_url + info['virtual_path']

The above example actually just generates and returns the default URL, which would have been what was returned anyway, but your code can perform arbitrary logic as necessary. For example, your code may wish to override the hostname or port number of the generated URL.

Note that the URL generated by __resource_url__ should be fully qualified, should end in a slash, and should not contain any query string or anchor elements (only path elements) to work best with resource_url().

Generating the Path To a Resource

pyramid.traversal.resource_path() returns a string object representing the absolute physical path of the resource object based on its position in the resource tree. Each segment of the path is separated with a slash character.

1
2
from pyramid.traversal import resource_path
url = resource_path(resource)

If resource in the example above was accessible in the tree as root['a']['b'], the above example would generate the string /a/b.

Any positional arguments passed in to resource_path() will be appended as path segments to the end of the resource path.

1
2
from pyramid.traversal import resource_path
url = resource_path(resource, 'foo', 'bar')

If resource in the example above was accessible in the tree as root['a']['b'], the above example would generate the string /a/b/foo/bar.

The resource passed in must be location-aware.

The presence or absence of a virtual root has no impact on the behavior of resource_path().

Finding a Resource by Path

If you have a string path to a resource, you can grab the resource from that place in the application’s resource tree using pyramid.traversal.find_resource().

You can resolve an absolute path by passing a string prefixed with a / as the path argument:

1
2
from pyramid.traversal import find_resource
url = find_resource(anyresource, '/path')

Or you can resolve a path relative to the resource you pass in by passing a string that isn’t prefixed by /:

1
2
from pyramid.traversal import find_resource
url = find_resource(anyresource, 'path')

Often the paths you pass to find_resource() are generated by the resource_path() API. These APIs are “mirrors” of each other.

If the path cannot be resolved when calling find_resource() (if the respective resource in the tree does not exist), a KeyError will be raised.

See the pyramid.traversal.find_resource() documentation for more information about resolving a path to a resource.

Obtaining the Lineage of a Resource

pyramid.location.lineage() returns a generator representing the lineage of the location aware resource object.

The lineage() function returns the resource it is passed, then each parent of the resource, in order. For example, if the resource tree is composed like so:

1
2
3
4
5
class Thing(object): pass

thing1 = Thing()
thing2 = Thing()
thing2.__parent__ = thing1

Calling lineage(thing2) will return a generator. When we turn it into a list, we will get:

1
2
list(lineage(thing2))
[ <Thing object at thing2>, <Thing object at thing1> ]

The generator returned by lineage() first returns the resource it was passed unconditionally. Then, if the resource supplied a __parent__ attribute, it returns the resource represented by resource.__parent__. If that resource has a __parent__ attribute, return that resource’s parent, and so on, until the resource being inspected either has no __parent__ attribute or has a __parent__ attribute of None.

See the documentation for pyramid.location.lineage() for more information.

Determining if a Resource is In The Lineage of Another Resource

Use the pyramid.location.inside() function to determine if one resource is in the lineage of another resource.

For example, if the resource tree is:

1
2
3
4
5
class Thing(object): pass

a = Thing()
b = Thing()
b.__parent__ = a

Calling inside(b, a) will return True, because b has a lineage that includes a. However, calling inside(a, b) will return False because a does not have a lineage that includes b.

The argument list for inside() is (resource1, resource2). resource1 is ‘inside’ resource2 if resource2 is a lineage ancestor of resource1. It is a lineage ancestor if its parent (or one of its parent’s parents, etc.) is an ancestor.

See pyramid.location.inside() for more information.

Finding the Root Resource

Use the pyramid.traversal.find_root() API to find the root resource. The root resource is the root resource of the resource tree. The API accepts a single argument: resource. This is a resource that is location aware. It can be any resource in the tree for which you want to find the root.

For example, if the resource tree is:

1
2
3
4
5
class Thing(object): pass

a = Thing()
b = Thing()
b.__parent__ = a

Calling find_root(b) will return a.

The root resource is also available as request.root within view callable code.

The presence or absence of a virtual root has no impact on the behavior of find_root(). The root object returned is always the physical root object.

Resources Which Implement Interfaces

Resources can optionally be made to implement an interface. An interface is used to tag a resource object with a “type” that can later be referred to within view configuration and by pyramid.traversal.find_interface().

Specifying an interface instead of a class as the context or containment predicate arguments within view configuration statements makes it possible to use a single view callable for more than one class of resource object. If your application is simple enough that you see no reason to want to do this, you can skip reading this section of the chapter.

For example, here’s some code which describes a blog entry which also declares that the blog entry implements an interface.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
import datetime
from zope.interface import implements
from zope.interface import Interface

class IBlogEntry(Interface):
    pass

class BlogEntry(object):
    implements(IBlogEntry)
    def __init__(self, title, body, author):
        self.title = title
        self.body = body
        self.author = author
        self.created = datetime.datetime.now()

This resource consists of two things: the class which defines the resource constructor as the class BlogEntry, and an interface attached to the class via an implements statement at class scope using the IBlogEntry interface as its sole argument.

The interface object used must be an instance of a class that inherits from zope.interface.Interface.

A resource class may implement zero or more interfaces. You specify that a resource implements an interface by using the zope.interface.implements() function at class scope. The above BlogEntry resource implements the IBlogEntry interface.

You can also specify that a particular resource instance provides an interface, as opposed to its class. When you declare that a class implements an interface, all instances of that class will also provide that interface. However, you can also just say that a single object provides the interface. To do so, use the zope.interface.directlyProvides() function:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
import datetime
from zope.interface import directlyProvides
from zope.interface import Interface

class IBlogEntry(Interface):
    pass

class BlogEntry(object):
    def __init__(self, title, body, author):
        self.title = title
        self.body = body
        self.author = author
        self.created = datetime.datetime.now()

entry = BlogEntry('title', 'body', 'author')
directlyProvides(entry, IBlogEntry)

zope.interface.directlyProvides() will replace any existing interface that was previously provided by an instance. If a resource object already has instance-level interface declarations that you don’t want to replace, use the zope.interface.alsoProvides() function:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
import datetime
from zope.interface import alsoProvides
from zope.interface import directlyProvides
from zope.interface import Interface

class IBlogEntry1(Interface):
    pass

class IBlogEntry2(Interface):
    pass

class BlogEntry(object):
    def __init__(self, title, body, author):
        self.title = title
        self.body = body
        self.author = author
        self.created = datetime.datetime.now()

entry = BlogEntry('title', 'body', 'author')
directlyProvides(entry, IBlogEntry1)
alsoProvides(entry, IBlogEntry2)

zope.interface.alsoProvides() will augment the set of interfaces directly provided by an instance instead of overwriting them like zope.interface.directlyProvides() does.

For more information about how resource interfaces can be used by view configuration, see Using Resource Interfaces In View Configuration.

Finding a Resource With a Class or Interface in Lineage

Use the find_interface() API to locate a parent that is of a particular Python class, or which implements some interface.

For example, if your resource tree is composed as follows:

1
2
3
4
5
6
class Thing1(object): pass
class Thing2(object): pass

a = Thing1()
b = Thing2()
b.__parent__ = a

Calling find_interface(a, Thing1) will return the a resource because a is of class Thing1 (the resource passed as the first argument is considered first, and is returned if the class or interface spec matches).

Calling find_interface(b, Thing1) will return the a resource because a is of class Thing1 and a is the first resource in b‘s lineage of this class.

Calling find_interface(b, Thing2) will return the b resource.

The second argument to find_interface may also be a interface instead of a class. If it is an interface, each resource in the lineage is checked to see if the resource implements the specificed interface (instead of seeing if the resource is of a class). See also Resources Which Implement Interfaces.

Pyramid API Functions That Act Against Resources

A resource object is used as the context provided to a view. See Traversal and URL Dispatch for more information about how a resource object becomes the context.

The APIs provided by pyramid.traversal are used against resource objects. These functions can be used to find the “path” of a resource, the root resource in a resource tree, or to generate a URL for a resource.

The APIs provided by pyramid.location are used against resources. These can be used to walk down a resource tree, or conveniently locate one resource “inside” another.

Some APIs in pyramid.security accept a resource object as a parameter. For example, the has_permission() API accepts a resource object as one of its arguments; the ACL is obtained from this resource or one of its ancestors. Other APIs in the pyramid.security module also accept context as an argument, and a context is always a resource.

Much Ado About Traversal

Note

This chapter was adapted, with permission, from a blog post by Rob Miller, originally published at http://blog.nonsequitarian.org/2010/much-ado-about-traversal/ .

Traversal is an alternative to URL dispatch which allows Pyramid applications to map URLs to code.

Note

Ex-Zope users whom are already familiar with traversal and view lookup conceptually may want to skip directly to the Traversal chapter, which discusses technical details. This chapter is mostly aimed at people who have previous Pylons experience or experience in another framework which does not provide traversal, and need an introduction to the “why” of traversal.

Some folks who have been using Pylons and its Routes-based URL matching for a long time are being exposed for the first time, via Pyramid, to new ideas such as “traversal” and “view lookup” as a way to route incoming HTTP requests to callable code. Some of the same folks believe that traversal is hard to understand. Others question its usefulness; URL matching has worked for them so far, why should they even consider dealing with another approach, one which doesn’t fit their brain and which doesn’t provide any immediately obvious value?

You can be assured that if you don’t want to understand traversal, you don’t have to. You can happily build Pyramid applications with only URL dispatch. However, there are some straightforward, real-world use cases that are much more easily served by a traversal-based approach than by a pattern-matching mechanism. Even if you haven’t yet hit one of these use cases yourself, understanding these new ideas is worth the effort for any web developer so you know when you might want to use them. Traversal is actually a straightforward metaphor easily comprehended by anyone who’s ever used a run-of-the-mill file system with folders and files.

URL Dispatch

Let’s step back and consider the problem we’re trying to solve. An HTTP request for a particular path has been routed to our web application. The requested path will possibly invoke a specific view callable function defined somewhere in our app. We’re trying to determine which callable function, if any, should be invoked for a given requested URL.

Many systems, including Pyramid, offer a simple solution. They offer the concept of “URL matching”. URL matching approaches this problem by parsing the URL path and comparing the results to a set of registered “patterns”, defined by a set of regular expressions, or some other URL path templating syntax. Each pattern is mapped to a callable function somewhere; if the request path matches a specific pattern, the associated function is called. If the request path matches more than one pattern, some conflict resolution scheme is used, usually a simple order precedence so that the first match will take priority over any subsequent matches. If a request path doesn’t match any of the defined patterns, a “404 Not Found” response is returned.

In Pyramid, we offer an implementation of URL matching which we call URL dispatch. Using Pyramid syntax, we might have a match pattern such as /{userid}/photos/{photoid}, mapped to a photo_view() function defined somewhere in our code. Then a request for a path such as /joeschmoe/photos/photo1 would be a match, and the photo_view() function would be invoked to handle the request. Similarly, /{userid}/blog/{year}/{month}/{postid} might map to a blog_post_view() function, so /joeschmoe/blog/2010/12/urlmatching would trigger the function, which presumably would know how to find and render the urlmatching blog post.

Historical Refresher

Now that we’ve refreshed our understanding of URL dispatch, we’ll dig in to the idea of traversal. Before we do, though, let’s take a trip down memory lane. If you’ve been doing web work for a while, you may remember a time when we didn’t have fancy web frameworks like Pylons and Pyramid. Instead, we had general purpose HTTP servers that primarily served files off of a file system. The “root” of a given site mapped to a particular folder somewhere on the file system. Each segment of the request URL path represented a subdirectory. The final path segment would be either a directory or a file, and once the server found the right file it would package it up in an HTTP response and send it back to the client. So serving up a request for /joeschmoe/photos/photo1 literally meant that there was a joeschmoe folder somewhere, which contained a photos folder, which in turn contained a photo1 file. If at any point along the way we find that there is not a folder or file matching the requested path, we return a 404 response.

As the web grew more dynamic, however, a little bit of extra complexity was added. Technologies such as CGI and HTTP server modules were developed. Files were still looked up on the file system, but if the file ended with (for example) .cgi or .php, or if it lived in a special folder, instead of simply sending the file to the client the server would read the file, execute it using an interpreter of some sort, and then send the output from this process to the client as the final result. The server configuration specified which files would trigger some dynamic code, with the default case being to just serve the static file.

Traversal (aka Resource Location)

Believe it or not, if you understand how serving files from a file system works, you understand traversal. And if you understand that a server might do something different based on what type of file a given request specifies, then you understand view lookup.

The major difference between file system lookup and traversal is that a file system lookup steps through nested directories and files in a file system tree, while traversal steps through nested dictionary-type objects in a resource tree. Let’s take a detailed look at one of our example paths, so we can see what I mean:

The path /joeschmoe/photos/photo1, has four segments: /, joeschmoe, photos and photo1. With file system lookup we might have a root folder (/) containing a nested folder (joeschmoe), which contains another nested folder (photos), which finally contains a JPG file (photo1). With traversal, we instead have a dictionary-like root object. Asking for the joeschmoe key gives us another dictionary-like object. Asking this in turn for the photos key gives us yet another mapping object, which finally (hopefully) contains the resource that we’re looking for within its values, referenced by the photo1 key.

In pure Python terms, then, the traversal or “resource location” portion of satisfying the /joeschmoe/photos/photo1 request will look something like this pseudocode:

get_root()['joeschmoe']['photos']['photo1']

get_root() is some function that returns a root traversal resource. If all of the specified keys exist, then the returned object will be the resource that is being requested, analogous to the JPG file that was retrieved in the file system example. If a KeyError is generated anywhere along the way, Pyramid will return 404. (This isn’t precisely true, as you’ll see when we learn about view lookup below, but the basic idea holds.)

What Is a “Resource”?

“Files on a file system I understand”, you might say. “But what are these nested dictionary things? Where do these objects, these ‘resources’, live? What are they?”

Since Pyramid is not a highly opinionated framework, it makes no restriction on how a resource is implemented; a developer can implement them as he wishes. One common pattern used is to persist all of the resources, including the root, in a database as a graph. The root object is a dictionary-like object. Dictionary-like objects in Python supply a __getitem__ method which is called when key lookup is done. Under the hood, when adict is a dictionary-like object, Python translates adict['a'] to adict.__getitem__('a'). Try doing this in a Python interpreter prompt if you don’t believe us:

1
2
3
4
5
6
7
8
9
Python 2.4.6 (#2, Apr 29 2010, 00:31:48)
[GCC 4.4.3] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> adict = {}
>>> adict['a'] = 1
>>> adict['a']
1
>>> adict.__getitem__('a')
1

The dictionary-like root object stores the ids of all of its subresources as keys, and provides a __getitem__ implementation that fetches them. So get_root() fetches the unique root object, while get_root()['joeschmoe'] returns a different object, also stored in the database, which in turn has its own subresources and __getitem__ implementation, etc. These resources might be persisted in a relational database, one of the many “NoSQL” solutions that are becoming popular these days, or anywhere else, it doesn’t matter. As long as the returned objects provide the dictionary-like API (i.e. as long as they have an appropriately implemented __getitem__ method) then traversal will work.

In fact, you don’t need a “database” at all. You could use plain dictionaries, with your site’s URL structure hard-coded directly in the Python source. Or you could trivially implement a set of objects with __getitem__ methods that search for files in specific directories, and thus precisely recreate the traditional mechanism of having the URL path mapped directly to a folder structure on the file system. Traversal is in fact a superset of file system lookup.

Note

See the chapter entitled Resources for a more technical overview of resources.

View Lookup

At this point we’re nearly there. We’ve covered traversal, which is the process by which a specific resource is retrieved according to a specific URL path. But what is “view lookup”?

The need for view lookup is simple: there is more than one possible action that you might want to take after finding a resource. With our photo example, for instance, you might want to view the photo in a page, but you might also want to provide a way for the user to edit the photo and any associated metadata. We’ll call the former the view view, and the latter will be the edit view. (Original, I know.) Pyramid has a centralized view application registry where named views can be associated with specific resource types. So in our example, we’ll assume that we’ve registered view and edit views for photo objects, and that we’ve specified the view view as the default, so that /joeschmoe/photos/photo1/view and /joeschmoe/photos/photo1 are equivalent. The edit view would sensibly be provided by a request for /joeschmoe/photos/photo1/edit.

Hopefully it’s clear that the first portion of the edit view’s URL path is going to resolve to the same resource as the non-edit version, specifically the resource returned by get_root()['joeschmoe']['photos']['photo1']. But traveral ends there; the photo1 resource doesn’t have an edit key. In fact, it might not even be a dictionary-like object, in which case photo1['edit'] would be meaningless. When the Pyramid resource location has been resolved to a leaf resource, but the entire request path has not yet been expended, the very next path segment is treated as a view name. The registry is then checked to see if a view of the given name has been specified for a resource of the given type. If so, the view callable is invoked, with the resource passed in as the related context object (also available as request.context). If a view callable could not be found, Pyramid will return a “404 Not Found” response.

You might conceptualize a request for /joeschmoe/photos/photo1/edit as ultimately converted into the following piece of Pythonic pseudocode:

context = get_root()['joeschmoe']['photos']['photo1']
view_callable = get_view(context, 'edit')
request.context = context
view_callable(request)

The get_root and get_view functions don’t really exist. Internally, Pyramid does something more complicated. But the example above is a reasonable approximation of the view lookup algorithm in pseudocode.

Use Cases

Why should we care about traversal? URL matching is easier to explain, and it’s good enough, right?

In some cases, yes, but certainly not in all cases. So far we’ve had very structured URLs, where our paths have had a specific, small number of pieces, like this:

/{userid}/{typename}/{objectid}[/{view_name}]

In all of the examples thus far, we’ve hard coded the typename value, assuming that we’d know at development time what names were going to be used (“photos”, “blog”, etc.). But what if we don’t know what these names will be? Or, worse yet, what if we don’t know anything about the structure of the URLs inside a user’s folder? We could be writing a CMS where we want the end user to be able to arbitrarily add content and other folders inside his folder. He might decide to nest folders dozens of layers deep. How will you construct matching patterns that could account for every possible combination of paths that might develop?

It might be possible, but it certainly won’t be easy. The matching patterns are going to become complex quickly as you try to handle all of the edge cases.

With traversal, however, it’s straightforward. Twenty layers of nesting would be no problem. Pyramid will happily call __getitem__ as many times as it needs to, until it runs out of path segments or until a resource raises a KeyError. Each resource only needs to know how to fetch its immediate children, the traversal algorithm takes care of the rest. Also, since the structure of the resource tree can live in the database and not in the code, it’s simple to let users modify the tree at runtime to set up their own personalized “directory” structures.

Another use case in which traversal shines is when there is a need to support a context-dependent security policy. One example might be a document management infrastructure for a large corporation, where members of different departments have varying access levels to the various other departments’ files. Reasonably, even specific files might need to be made available to specific individuals. Traversal does well here if your resources actually represent the data objects related to your documents, because the idea of a resource authorization is baked right into the code resolution and calling process. Resource objects can store ACLs, which can be inherited and/or overridden by the subresources.

If each resource can thus generate a context-based ACL, then whenever view code is attempting to perform a sensitive action, it can check against that ACL to see whether the current user should be allowed to perform the action. In this way you achieve so called “instance based” or “row level” security which is considerably harder to model using a traditional tabular approach. Pyramid actively supports such a scheme, and in fact if you register your views with guard permissions and use an authorization policy, Pyramid can check against a resource’s ACL when deciding whether or not the view itself is available to the current user.

In summary, there are entire classes of problems that are more easily served by traversal and view lookup than by URL dispatch. If your problems don’t require it, great: stick with URL dispatch. But if you’re using Pyramid and you ever find that you do need to support one of these use cases, you’ll be glad you have traversal in your toolkit.

Note

It is even possible to mix and match traversal with URL dispatch in the same Pyramid application. See the Combining Traversal and URL Dispatch chapter for details.

Traversal

A traversal uses the URL (Universal Resource Locator) to find a resource located in a resource tree, which is a set of nested dictionary-like objects. Traversal is done by using each segment of the path portion of the URL to navigate through the resource tree. You might think of this as looking up files and directories in a file system. Traversal walks down the path until it finds a published resource, analogous to a file system “directory” or “file”. The resource found as the result of a traversal becomes the context of the request. Then, the view lookup subsystem is used to find some view code willing to “publish” this resource by generating a response.

Using Traversal to map a URL to code is optional. It is often less easy to understand than URL dispatch, so if you’re a rank beginner, it probably makes sense to use URL dispatch to map URLs to code instead of traversal. In that case, you can skip this chapter.

Traversal Details

Traversal is dependent on information in a request object. Every request object contains URL path information in the PATH_INFO portion of the WSGI environment. The PATH_INFO string is the portion of a request’s URL following the hostname and port number, but before any query string elements or fragment element. For example the PATH_INFO portion of the URL http://example.com:8080/a/b/c?foo=1 is /a/b/c.

Traversal treats the PATH_INFO segment of a URL as a sequence of path segments. For example, the PATH_INFO string /a/b/c is converted to the sequence ['a', 'b', 'c'].

This path sequence is then used to descend through the resource tree, looking up a resource for each path segment. Each lookup uses the __getitem__ method of a resource in the tree.

For example, if the path info sequence is ['a', 'b', 'c']:

  • Traversal starts by acquiring the root resource of the application by calling the root factory. The root factory can be configured to return whatever object is appropriate as the traversal root of your application.
  • Next, the first element ('a') is popped from the path segment sequence and is used as a key to lookup the corresponding resource in the root. This invokes the root resource’s __getitem__ method using that value ('a') as an argument.
  • If the root resource “contains” a resource with key 'a', its __getitem__ method will return it. The context temporarily becomes the “A” resource.
  • The next segment ('b') is popped from the path sequence, and the “A” resource’s __getitem__ is called with that value ('b') as an argument; we’ll presume it succeeds.
  • The “A” resource’s __getitem__ returns another resource, which we’ll call “B”. The context temporarily becomes the “B” resource.

Traversal continues until the path segment sequence is exhausted or a path element cannot be resolved to a resource. In either case, the context resource is the last object that the traversal successfully resolved. If any resource found during traversal lacks a __getitem__ method, or if its __getitem__ method raises a KeyError, traversal ends immediately, and that resource becomes the context.

The results of a traversal also include a view name. If traversal ends before the path segment sequence is exhausted, the view name is the next remaining path segment element. If the traversal expends all of the path segments, then the view name is the empty string ('').

The combination of the context resource and the view name found via traversal is used later in the same request by the view lookup subsystem to find a view callable. How Pyramid performs view lookup is explained within the View Configuration chapter.

The Resource Tree

The resource tree is a set of nested dictionary-like resource objects that begins with a root resource. In order to use traversal to resolve URLs to code, your application must supply a resource tree to Pyramid.

In order to supply a root resource for an application the Pyramid Router is configured with a callback known as a root factory. The root factory is supplied by the application, at startup time, as the root_factory argument to the Configurator.

The root factory is a Python callable that accepts a request object, and returns the root object of the resource tree. A function, or class is typically used as an application’s root factory. Here’s an example of a simple root factory class:

1
2
3
class Root(dict):
    def __init__(self, request):
        pass

Here’s an example of using this root factory within startup configuration, by passing it to an instance of a Configurator named config:

1
config = Configurator(root_factory=Root)

The root_factory argument to the Configurator constructor registers this root factory to be called to generate a root resource whenever a request enters the application. The root factory registered this way is also known as the global root factory. A root factory can alternately be passed to the Configurator as a dotted Python name which can refer to a root factory defined in a different module.

If no root factory is passed to the Pyramid Configurator constructor, or if the root_factory value specified is None, a default root factory is used. The default root factory always returns a resource that has no child resources; it is effectively empty.

Usually a root factory for a traversal-based application will be more complicated than the above Root class; in particular it may be associated with a database connection or another persistence mechanism.

Emulating the Default Root Factory

For purposes of understanding the default root factory better, we’ll note that you can emulate the default root factory by using this code as an explicit root factory in your application setup:

1
2
3
4
5
class Root(object):
    def __init__(self, request):
        pass

config = Configurator(root_factory=Root)

The default root factory is just a really stupid object that has no behavior or state. Using traversal against an application that uses the resource tree supplied by the default root resource is not very interesting, because the default root resource has no children. Its availability is more useful when you’re developing an application using URL dispatch.

Note

If the items contained within the resource tree are “persistent” (they have state that lasts longer than the execution of a single process), they become analogous to the concept of domain model objects used by many other frameworks.

The resource tree consists of container resources and leaf resources. There is only one difference between a container resource and a leaf resource: container resources possess a __getitem__ method (making it “dictionary-like”) while leaf resources do not. The __getitem__ method was chosen as the signifying difference between the two types of resources because the presence of this method is how Python itself typically determines whether an object is “containerish” or not (dictionary objects are “containerish”).

Each container resource is presumed to be willing to return a child resource or raise a KeyError based on a name passed to its __getitem__.

Leaf-level instances must not have a __getitem__. If instances that you’d like to be leaves already happen to have a __getitem__ through some historical inequity, you should subclass these resource types and cause their __getitem__ methods to simply raise a KeyError. Or just disuse them and think up another strategy.

Usually, the traversal root is a container resource, and as such it contains other resources. However, it doesn’t need to be a container. Your resource tree can be as shallow or as deep as you require.

In general, the resource tree is traversed beginning at its root resource using a sequence of path elements described by the PATH_INFO of the current request; if there are path segments, the root resource’s __getitem__ is called with the next path segment, and it is expected to return another resource. The resulting resource’s __getitem__ is called with the very next path segment, and it is expected to return another resource. This happens ad infinitum until all path segments are exhausted.

The Traversal Algorithm

This section will attempt to explain the Pyramid traversal algorithm. We’ll provide a description of the algorithm, a diagram of how the algorithm works, and some example traversal scenarios that might help you understand how the algorithm operates against a specific resource tree.

We’ll also talk a bit about view lookup. The View Configuration chapter discusses view lookup in detail, and it is the canonical source for information about views. Technically, view lookup is a Pyramid subsystem that is separated from traversal entirely. However, we’ll describe the fundamental behavior of view lookup in the examples in the next few sections to give you an idea of how traversal and view lookup cooperate, because they are almost always used together.

A Description of The Traversal Algorithm

When a user requests a page from your traversal-powered application, the system uses this algorithm to find a context resource and a view name.

  1. The request for the page is presented to the Pyramid router in terms of a standard WSGI request, which is represented by a WSGI environment and a WSGI start_response callable.

  2. The router creates a request object based on the WSGI environment.

  3. The root factory is called with the request. It returns a root resource.

  4. The router uses the WSGI environment’s PATH_INFO information to determine the path segments to traverse. The leading slash is stripped off PATH_INFO, and the remaining path segments are split on the slash character to form a traversal sequence.

    The traversal algorithm by default attempts to first URL-unquote and then Unicode-decode each path segment derived from PATH_INFO from its natural byte string (str type) representation. URL unquoting is performed using the Python standard library urllib.unquote function. Conversion from a URL-decoded string into Unicode is attempted using the UTF-8 encoding. If any URL-unquoted path segment in PATH_INFO is not decodeable using the UTF-8 decoding, a TypeError is raised. A segment will be fully URL-unquoted and UTF8-decoded before it is passed in to the __getitem__ of any resource during traversal.

    Thus, a request with a PATH_INFO variable of /a/b/c maps to the traversal sequence [u'a', u'b', u'c'].

  5. Traversal begins at the root resource returned by the root factory. For the traversal sequence [u'a', u'b', u'c'], the root resource’s __getitem__ is called with the name 'a'. Traversal continues through the sequence. In our example, if the root resource’s __getitem__ called with the name a returns a resource (aka resource “A”), that resource’s __getitem__ is called with the name 'b'. If resource “A” returns a resource “B” when asked for 'b', resource B’s __getitem__ is then asked for the name 'c', and may return resource “C”.

  6. Traversal ends when a) the entire path is exhausted or b) when any resouce raises a KeyError from its __getitem__ or c) when any non-final path element traversal does not have a __getitem__ method (resulting in a AttributeError) or d) when any path element is prefixed with the set of characters @@ (indicating that the characters following the @@ token should be treated as a view name).

  7. When traversal ends for any of the reasons in the previous step, the last resource found during traversal is deemed to be the context. If the path has been exhausted when traversal ends, the view name is deemed to be the empty string (''). However, if the path was not exhausted before traversal terminated, the first remaining path segment is treated as the view name.

  8. Any subsequent path elements after the view name is found are deemed the subpath. The subpath is always a sequence of path segments that come from PATH_INFO that are “left over” after traversal has completed.

Once the context resource, the view name, and associated attributes such as the subpath are located, the job of traversal is finished. It passes back the information it obtained to its caller, the Pyramid Router, which subsequently invokes view lookup with the context and view name information.

The traversal algorithm exposes two special cases:

  • You will often end up with a view name that is the empty string as the result of a particular traversal. This indicates that the view lookup machinery should look up the default view. The default view is a view that is registered with no name or a view which is registered with a name that equals the empty string.
  • If any path segment element begins with the special characters @@ (think of them as goggles), the value of that segment minus the goggle characters is considered the view name immediately and traversal stops there. This allows you to address views that may have the same names as resource names in the tree unambiguously.

Finally, traversal is responsible for locating a virtual root. A virtual root is used during “virtual hosting”; see the Virtual Hosting chapter for information. We won’t speak more about it in this chapter.

_images/resourcetreetraverser.png
Traversal Algorithm Examples

No one can be expected to understand the traversal algorithm by analogy and description alone, so let’s examine some traversal scenarios that use concrete URLs and resource tree compositions.

Let’s pretend the user asks for http://example.com/foo/bar/baz/biz/buz.txt. The request’s PATH_INFO in that case is /foo/bar/baz/biz/buz.txt. Let’s further pretend that when this request comes in that we’re traversing the following resource tree:

/--
   |
   |-- foo
        |
        ----bar

Here’s what happens:

  • traversal traverses the root, and attempts to find “foo”, which it finds.
  • traversal traverses “foo”, and attempts to find “bar”, which it finds.
  • traversal traverses “bar”, and attempts to find “baz”, which it does not find (the “bar” resource raises a KeyError when asked for “baz”).

The fact that it does not find “baz” at this point does not signify an error condition. It signifies that:

  • the context is the “bar” resource (the context is the last resource found during traversal).
  • the view name is baz
  • the subpath is ('biz', 'buz.txt')

At this point, traversal has ended, and view lookup begins.

Because it’s the “context” resource, the view lookup machinery examines “bar” to find out what “type” it is. Let’s say it finds that the context is a Bar type (because “bar” happens to be an instance of the class Bar). Using the view name (baz) and the type, view lookup asks the application registry this question:

Let’s say that view lookup finds no matching view type. In this circumstance, the Pyramid router returns the result of the not found view and the request ends.

However, for this tree:

/--
   |
   |-- foo
        |
        ----bar
             |
             ----baz
                    |
                    biz

The user asks for http://example.com/foo/bar/baz/biz/buz.txt

  • traversal traverses “foo”, and attempts to find “bar”, which it finds.
  • traversal traverses “bar”, and attempts to find “baz”, which it finds.
  • traversal traverses “baz”, and attempts to find “biz”, which it finds.
  • traversal traverses “biz”, and attempts to find “buz.txt” which it does not find.

The fact that it does not find a resource related to “buz.txt” at this point does not signify an error condition. It signifies that:

  • the context is the “biz” resource (the context is the last resource found during traversal).
  • the view name is “buz.txt”
  • the subpath is an empty sequence ( () ).

At this point, traversal has ended, and view lookup begins.

Because it’s the “context” resource, the view lookup machinery examines the “biz” resource to find out what “type” it is. Let’s say it finds that the resource is a Biz type (because “biz” is an instance of the Python class Biz). Using the view name (buz.txt) and the type, view lookup asks the application registry this question:

Let’s say that question is answered by the application registry; in such a situation, the application registry returns a view callable. The view callable is then called with the current WebOb request as the sole argument: request; it is expected to return a response.

The Example View Callables Accept Only a Request; How Do I Access the Context Resource?

Most of the examples in this book assume that a view callable is typically passed only a request object. Sometimes your view callables need access to the context resource, especially when you use traversal. You might use a supported alternate view callable argument list in your view callables such as the (context, request) calling convention described in Alternate View Callable Argument/Calling Conventions. But you don’t need to if you don’t want to. In view callables that accept only a request, the context resource found by traversal is available as the context attribute of the request object, e.g. request.context. The view name is available as the view_name attribute of the request object, e.g. request.view_name. Other Pyramid -specific request attributes are also available as described in Special Attributes Added to the Request by Pyramid.

Using Resource Interfaces In View Configuration

Instead of registering your views with a context that names a Python resource class, you can optionally register a view callable with a context which is an interface. An interface can be attached arbitrarily to any resource object. View lookup treats context interfaces specially, and therefore the identity of a resource can be divorced from that of the class which implements it. As a result, associating a view with an interface can provide more flexibility for sharing a single view between two or more different implementations of a resource type. For example, if two resource objects of different Python class types share the same interface, you can use the same view configuration to specify both of them as a context.

In order to make use of interfaces in your application during view dispatch, you must create an interface and mark up your resource classes or instances with interface declarations that refer to this interface.

To attach an interface to a resource class, you define the interface and use the zope.interface.implements() function to associate the interface with the class.

1
2
3
4
5
6
7
8
from zope.interface import Interface
from zope.interface import implements

class IHello(Interface):
    """ A marker interface """

class Hello(object):
    implements(IHello)

To attach an interface to a resource instance, you define the interface and use the zope.interface.alsoProvides() function to associate the interface with the instance. This function mutates the instance in such a way that the interface is attached to it.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
from zope.interface import Interface
from zope.interface import alsoProvides

class IHello(Interface):
    """ A marker interface """

class Hello(object):
    pass

def make_hello():
    hello = Hello()
    alsoProvides(hello, IHello)
    return hello

Regardless of how you associate an interface, with a resource instance, or a resource class, the resulting code to associate that interface with a view callable is the same. Assuming the above code that defines an IHello interface lives in the root of your application, and its module is named “resources.py”, the interface declaration below will associate the mypackage.views.hello_world view with resources that implement, or provide, this interface.

1
2
3
4
# config is an instance of pyramid.config.Configurator

config.add_view('mypackage.views.hello_world', name='hello.html',
                context='mypackage.resources.IHello')

Any time a resource that is determined to be the context provides this interface, and a view named hello.html is looked up against it as per the URL, the mypackage.views.hello_world view callable will be invoked.

Note, in cases where a view is registered against a resource class, and a view is also registered against an interface that the resource class implements, an ambiguity arises. Views registered for the resource class take precedence over any views registered for any interface the resource class implements. Thus, if one view configuration names a context of both the class type of a resource, and another view configuration names a context of interface implemented by the resource’s class, and both view configurations are otherwise identical, the view registered for the context’s class will “win”.

For more information about defining resources with interfaces for use within view configuration, see Resources Which Implement Interfaces.

References

A tutorial showing how traversal can be used within a Pyramid application exists in ZODB + Traversal Wiki Tutorial.

See the View Configuration chapter for detailed information about view lookup.

The pyramid.traversal module contains API functions that deal with traversal, such as traversal invocation from within application code.

The pyramid.request.Request.resource_url() method generates a URL when given a resource retrieved from a resource tree.

Security

Pyramid provides an optional declarative authorization system that can prevent a view from being invoked based on an authorization policy. Before a view is invoked, the authorization system can use the credentials in the request along with the context resource to determine if access will be allowed. Here’s how it works at a high level:

  • A request is generated when a user visits the application.
  • Based on the request, a context resource is located through resource location. A context is located differently depending on whether the application uses traversal or URL dispatch, but a context is ultimately found in either case. See the URL Dispatch chapter for more information.
  • A view callable is located by view lookup using the context as well as other attributes of the request.
  • If an authentication policy is in effect, it is passed the request; it returns some number of principal identifiers.
  • If an authorization policy is in effect and the view configuration associated with the view callable that was found has a permission associated with it, the authorization policy is passed the context, some number of principal identifiers returned by the authentication policy, and the permission associated with the view; it will allow or deny access.
  • If the authorization policy allows access, the view callable is invoked.
  • If the authorization policy denies access, the view callable is not invoked; instead the forbidden view is invoked.

Security in Pyramid, unlike many systems, cleanly and explicitly separates authentication and authorization. Authentication is merely the mechanism by which credentials provided in the request are resolved to one or more principal identifiers. These identifiers represent the users and groups in effect during the request. Authorization then determines access based on the principal identifiers, the view callable being invoked, and the context resource.

Authorization is enabled by modifying your application to include an authentication policy and authorization policy. Pyramid comes with a variety of implementations of these policies. To provide maximal flexibility, Pyramid also allows you to create custom authentication policies and authorization policies.

Enabling an Authorization Policy

By default, Pyramid enables no authorization policy. All views are accessible by completely anonymous users. In order to begin protecting views from execution based on security settings, you need to enable an authorization policy.

Enabling an Authorization Policy Imperatively

Passing an authorization_policy argument to the constructor of the Configurator class enables an authorization policy.

You must also enable an authentication policy in order to enable the authorization policy. This is because authorization, in general, depends upon authentication. Use the authentication_policy argument to the Configurator class during application setup to specify an authentication policy.

For example:

1
2
3
4
5
6
7
from pyramid.config import Configurator
from pyramid.authentication import AuthTktAuthenticationPolicy
from pyramid.authorization import ACLAuthorizationPolicy
authentication_policy = AuthTktAuthenticationPolicy('seekrit')
authorization_policy = ACLAuthorizationPolicy()
config = Configurator(authentication_policy=authentication_policy,
                      authorization_policy=authorization_policy)

Note

the authentication_policy and authorization_policy arguments may also be passed to the Configurator as dotted Python name values, each representing the dotted name path to a suitable implementation global defined at Python module scope.

The above configuration enables a policy which compares the value of an “auth ticket” cookie passed in the request’s environment which contains a reference to a single principal against the principals present in any ACL found in the resource tree when attempting to call some view.

While it is possible to mix and match different authentication and authorization policies, it is an error to pass an authentication policy without the authorization policy or vice versa to a Configurator constructor.

See also the pyramid.authorization and pyramid.authentication modules for alternate implementations of authorization and authentication policies.

Protecting Views with Permissions

To protect a view callable from invocation based on a user’s security settings when a particular type of resource becomes the context, you must pass a permission to view configuration. Permissions are usually just strings, and they have no required composition: you can name permissions whatever you like.

For example, the following view declaration protects the view named add_entry.html when the context resource is of type Blog with the add permission using the pyramid.config.Configurator.add_view() API:

1
2
3
4
5
6
# config is an instance of pyramid.config.Configurator

config.add_view('mypackage.views.blog_entry_add_view',
                name='add_entry.html',
                context='mypackage.resources.Blog',
                permission='add')

The equivalent view registration including the add permission name may be performed via the @view_config decorator:

1
2
3
4
5
6
7
from pyramid.view import view_config
from resources import Blog

@view_config(context=Blog, name='add_entry.html', permission='add')
def blog_entry_add_view(request):
    """ Add blog entry code goes here """
    pass

As a result of any of these various view configuration statements, if an authorization policy is in place when the view callable is found during normal application operations, the requesting user will need to possess the add permission against the context resource in order to be able to invoke the blog_entry_add_view view. If he does not, the Forbidden view will be invoked.

Setting a Default Permission

If a permission is not supplied to a view configuration, the registered view will always be executable by entirely anonymous users: any authorization policy in effect is ignored.

In support of making it easier to configure applications which are “secure by default”, Pyramid allows you to configure a default permission. If supplied, the default permission is used as the permission string to all view registrations which don’t otherwise name a permission argument.

These APIs are in support of configuring a default permission for an application:

  • The default_permission constructor argument to the Configurator constructor.
  • The pyramid.config.Configurator.set_default_permission() method.

When a default permission is registered:

  • If a view configuration names an explicit permission, the default permission is ignored for that view registration, and the view-configuration-named permission is used.
  • If a view configuration names the permission pyramid.security.NO_PERMISSION_REQUIRED, the default permission is ignored, and the view is registered without a permission (making it available to all callers regardless of their credentials).

Warning

When you register a default permission, all views (even exception view views) are protected by a permission. For all views which are truly meant to be anonymously accessible, you will need to associate the view’s configuration with the pyramid.security.NO_PERMISSION_REQUIRED permission.

Assigning ACLs to your Resource Objects

When the default Pyramid authorization policy determines whether a user possesses a particular permission with respect to a resource, it examines the ACL associated with the resource. An ACL is associated with a resource by adding an __acl__ attribute to the resource object. This attribute can be defined on the resource instance if you need instance-level security, or it can be defined on the resource class if you just need type-level security.

For example, an ACL might be attached to the resource for a blog via its class:

1
2
3
4
5
6
7
8
9
from pyramid.security import Everyone
from pyramid.security import Allow

class Blog(object):
    __acl__ = [
        (Allow, Everyone, 'view'),
        (Allow, 'group:editors', 'add'),
        (Allow, 'group:editors', 'edit'),
        ]

Or, if your resources are persistent, an ACL might be specified via the __acl__ attribute of an instance of a resource:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
from pyramid.security import Everyone
from pyramid.security import Allow

class Blog(object):
    pass

blog = Blog()

blog.__acl__ = [
        (Allow, Everyone, 'view'),
        (Allow, 'group:editors', 'add'),
        (Allow, 'group:editors', 'edit'),
        ]

Whether an ACL is attached to a resource’s class or an instance of the resource itself, the effect is the same. It is useful to decorate individual resource instances with an ACL (as opposed to just decorating their class) in applications such as “CMS” systems where fine-grained access is required on an object-by-object basis.

Elements of an ACL

Here’s an example ACL:

1
2
3
4
5
6
7
8
from pyramid.security import Everyone
from pyramid.security import Allow

__acl__ = [
        (Allow, Everyone, 'view'),
        (Allow, 'group:editors', 'add'),
        (Allow, 'group:editors', 'edit'),
        ]

The example ACL indicates that the pyramid.security.Everyone principal – a special system-defined principal indicating, literally, everyone – is allowed to view the blog, the group:editors principal is allowed to add to and edit the blog.

Each element of an ACL is an ACE or access control entry. For example, in the above code block, there are three ACEs: (Allow, Everyone, 'view'), (Allow, 'group:editors', 'add'), and (Allow, 'group:editors', 'edit').

The first element of any ACE is either pyramid.security.Allow, or pyramid.security.Deny, representing the action to take when the ACE matches. The second element is a principal. The third argument is a permission or sequence of permission names.

A principal is usually a user id, however it also may be a group id if your authentication system provides group information and the effective authentication policy policy is written to respect group information. For example, the pyramid.authentication.RepozeWho1AuthenicationPolicy respects group information if you configure it with a callback.

Each ACE in an ACL is processed by an authorization policy in the order dictated by the ACL. So if you have an ACL like this:

1
2
3
4
5
6
7
8
from pyramid.security import Everyone
from pyramid.security import Allow
from pyramid.security import Deny

__acl__ = [
    (Allow, Everyone, 'view'),
    (Deny, Everyone, 'view'),
    ]

The default authorization policy will allow everyone the view permission, even though later in the ACL you have an ACE that denies everyone the view permission. On the other hand, if you have an ACL like this:

1
2
3
4
5
6
7
8
from pyramid.security import Everyone
from pyramid.security import Allow
from pyramid.security import Deny

__acl__ = [
    (Deny, Everyone, 'view'),
    (Allow, Everyone, 'view'),
    ]

The authorization policy will deny everyone the view permission, even though later in the ACL is an ACE that allows everyone.

The third argument in an ACE can also be a sequence of permission names instead of a single permission name. So instead of creating multiple ACEs representing a number of different permission grants to a single group:editors group, we can collapse this into a single ACE, as below.

1
2
3
4
5
6
7
from pyramid.security import Everyone
from pyramid.security import Allow

__acl__ = [
    (Allow, Everyone, 'view'),
    (Allow, 'group:editors', ('add', 'edit')),
    ]

Special Principal Names

Special principal names exist in the pyramid.security module. They can be imported for use in your own code to populate ACLs, e.g. pyramid.security.Everyone.

pyramid.security.Everyone

Literally, everyone, no matter what. This object is actually a string “under the hood” (system.Everyone). Every user “is” the principal named Everyone during every request, even if a security policy is not in use.

pyramid.security.Authenticated

Any user with credentials as determined by the current security policy. You might think of it as any user that is “logged in”. This object is actually a string “under the hood” (system.Authenticated).

Special Permissions

Special permission names exist in the pyramid.security module. These can be imported for use in ACLs.

pyramid.security.ALL_PERMISSIONS

An object representing, literally, all permissions. Useful in an ACL like so: (Allow, 'fred', ALL_PERMISSIONS). The ALL_PERMISSIONS object is actually a stand-in object that has a __contains__ method that always returns True, which, for all known authorization policies, has the effect of indicating that a given principal “has” any permission asked for by the system.

Special ACEs

A convenience ACE is defined representing a deny to everyone of all permissions in pyramid.security.DENY_ALL. This ACE is often used as the last ACE of an ACL to explicitly cause inheriting authorization policies to “stop looking up the traversal tree” (effectively breaking any inheritance). For example, an ACL which allows only fred the view permission for a particular resource despite what inherited ACLs may say when the default authorization policy is in effect might look like so:

1
2
3
4
from pyramid.security import Allow
from pyramid.security import DENY_ALL

__acl__ = [ (Allow, 'fred', 'view'), DENY_ALL ]

“Under the hood”, the pyramid.security.DENY_ALL ACE equals the following:

1
2
from pyramid.security import ALL_PERMISSIONS
__acl__ = [ (Deny, Everyone, ALL_PERMISSIONS) ]

ACL Inheritance and Location-Awareness

While the default authorization policy is in place, if a resource object does not have an ACL when it is the context, its parent is consulted for an ACL. If that object does not have an ACL, its parent is consulted for an ACL, ad infinitum, until we’ve reached the root and there are no more parents left.

In order to allow the security machinery to perform ACL inheritance, resource objects must provide location-awareness. Providing location-awareness means two things: the root object in the resource tree must have a __name__ attribute and a __parent__ attribute.

1
2
3
class Blog(object):
    __name__ = ''
    __parent__ = None

An object with a __parent__ attribute and a __name__ attribute is said to be location-aware. Location-aware objects define an __parent__ attribute which points at their parent object. The root object’s __parent__ is None.

See pyramid.location for documentations of functions which use location-awareness. See also Location-Aware Resources.

Changing the Forbidden View

When Pyramid denies a view invocation due to an authorization denial, the special forbidden view is invoked. “Out of the box”, this forbidden view is very plain. See Changing the Forbidden View within Using Hooks for instructions on how to create a custom forbidden view and arrange for it to be called when view authorization is denied.

Debugging View Authorization Failures

If your application in your judgment is allowing or denying view access inappropriately, start your application under a shell using the PYRAMID_DEBUG_AUTHORIZATION environment variable set to 1. For example:

$ PYRAMID_DEBUG_AUTHORIZATION=1 bin/paster serve myproject.ini

When any authorization takes place during a top-level view rendering, a message will be logged to the console (to stderr) about what ACE in which ACL permitted or denied the authorization based on authentication information.

This behavior can also be turned on in the application .ini file by setting the pyramid.debug_authorization key to true within the application’s configuration section, e.g.:

1
2
3
[app:main]
use = egg:MyProject
pyramid.debug_authorization = true

With this debug flag turned on, the response sent to the browser will also contain security debugging information in its body.

Debugging Imperative Authorization Failures

The pyramid.security.has_permission() API is used to check security within view functions imperatively. It returns instances of objects that are effectively booleans. But these objects are not raw True or False objects, and have information attached to them about why the permission was allowed or denied. The object will be one of pyramid.security.ACLAllowed, pyramid.security.ACLDenied, pyramid.security.Allowed, or pyramid.security.Denied, as documented in pyramid.security. At the very minimum these objects will have a msg attribute, which is a string indicating why the permission was denied or allowed. Introspecting this information in the debugger or via print statements when a call to has_permission() fails is often useful.

Creating Your Own Authentication Policy

Pyramid ships with a number of useful out-of-the-box security policies (see pyramid.authentication). However, creating your own authentication policy is often necessary when you want to control the “horizontal and vertical” of how your users authenticate. Doing so is a matter of creating an instance of something that implements the following interface:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
class IAuthenticationPolicy(object):
    """ An object representing a Pyramid authentication policy. """

    def authenticated_userid(self, request):
        """ Return the authenticated userid or ``None`` if no
        authenticated userid can be found. This method of the policy
        should ensure that a record exists in whatever persistent store is
        used related to the user (the user should not have been deleted);
        if a record associated with the current id does not exist in a
        persistent store, it should return ``None``."""

    def unauthenticated_userid(self, request):
        """ Return the *unauthenticated* userid.  This method performs the
        same duty as ``authenticated_userid`` but is permitted to return the
        userid based only on data present in the request; it needn't (and
        shouldn't) check any persistent store to ensure that the user record
        related to the request userid exists."""

    def effective_principals(self, request):
        """ Return a sequence representing the effective principals
        including the userid and any groups belonged to by the current
        user, including 'system' groups such as
        ``pyramid.security.Everyone`` and
        ``pyramid.security.Authenticated``. """

    def remember(self, request, principal, **kw):
        """ Return a set of headers suitable for 'remembering' the
        principal named ``principal`` when set in a response.  An
        individual authentication policy and its consumers can decide
        on the composition and meaning of **kw. """

    def forget(self, request):
        """ Return a set of headers suitable for 'forgetting' the
        current user on subsequent requests. """

After you do so, you can pass an instance of such a class into the Configurator class at configuration time as authentication_policy to use it.

Creating Your Own Authorization Policy

An authorization policy is a policy that allows or denies access after a user has been authenticated. By default, Pyramid will use the pyramid.authorization.ACLAuthorizationPolicy if an authentication policy is activated and an authorization policy isn’t otherwise specified.

In some cases, it’s useful to be able to use a different authorization policy than the default ACLAuthorizationPolicy. For example, it might be desirable to construct an alternate authorization policy which allows the application to use an authorization mechanism that does not involve ACL objects.

Pyramid ships with only a single default authorization policy, so you’ll need to create your own if you’d like to use a different one. Creating and using your own authorization policy is a matter of creating an instance of an object that implements the following interface:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
class IAuthorizationPolicy(object):
    """ An object representing a Pyramid authorization policy. """
    def permits(self, context, principals, permission):
        """ Return ``True`` if any of the ``principals`` is allowed the
        ``permission`` in the current ``context``, else return ``False``
        """

    def principals_allowed_by_permission(self, context, permission):
        """ Return a set of principal identifiers allowed by the
        ``permission`` in ``context``.  This behavior is optional; if you
        choose to not implement it you should define this method as
        something which raises a ``NotImplementedError``.  This method
        will only be called when the
        ``pyramid.security.principals_allowed_by_permission`` API is
        used."""

After you do so, you can pass an instance of such a class into the Configurator class at configuration time as authorization_policy to use it.

Combining Traversal and URL Dispatch

When you write most Pyramid applications, you’ll be using one or the other of two available resource location subsystems: traversal or URL dispatch. However, to solve a limited set of problems, it’s useful to use both traversal and URL dispatch together within the same application. Pyramid makes this possible via hybrid applications.

Warning

Reasoning about the behavior of a “hybrid” URL dispatch + traversal application can be challenging. To successfully reason about using URL dispatch and traversal together, you need to understand URL pattern matching, root factories, and the traversal algorithm, and the potential interactions between them. Therefore, we don’t recommend creating an application that relies on hybrid behavior unless you must.

A Review of Non-Hybrid Applications

When used according to the tutorials in its documentation Pyramid is a “dual-mode” framework: the tutorials explain how to create an application in terms of using either url dispatch or traversal. This chapter details how you might combine these two dispatch mechanisms, but we’ll review how they work in isolation before trying to combine them.

URL Dispatch Only

An application that uses url dispatch exclusively to map URLs to code will often have statements like this within application startup configuration:

1
2
3
4
5
6
7
# config is an instance of pyramid.config.Configurator

config.add_route('foobar', '{foo}/{bar}')
config.add_route('bazbuz', '{baz}/{buz}')

config.add_view('myproject.views.foobar', route_name='foobar')
config.add_view('myproject.views.bazbuz', route_name='bazbuz')

Each route corresponds to one or more view callables. Each view callable is associated with a route by passing a route_name parameter that matches its name during a call to add_view(). When a route is matched during a request, view lookup is used to match the request to its associated view callable. The presence of calls to add_route() signify that an application is using URL dispatch.

Traversal Only

An application that uses only traversal will have view configuration declarations that look like this:

1
2
3
4
 # config is an instance of pyramid.config.Configurator

 config.add_view('mypackage.views.foobar', name='foobar')
 config.add_view('mypackage.views.bazbuz', name='bazbuz')

When the above configuration is applied to an application, the mypackage.views.foobar view callable above will be called when the URL /foobar is visited. Likewise, the view mypackage.views.bazbuz will be called when the URL /bazbuz is visited.

Typically, an application that uses traversal exclusively won’t perform any calls to pyramid.config.Configurator.add_route() in its startup code.

Hybrid Applications

Either traversal or url dispatch alone can be used to create a Pyramid application. However, it is also possible to combine the concepts of traversal and url dispatch when building an application: the result is a hybrid application. In a hybrid application, traversal is performed after a particular route has matched.

A hybrid application is a lot more like a “pure” traversal-based application than it is like a “pure” URL-dispatch based application. But unlike in a “pure” traversal-based application, in a hybrid application, traversal is performed during a request after a route has already matched. This means that the URL pattern that represents the pattern argument of a route must match the PATH_INFO of a request, and after the route pattern has matched, most of the “normal” rules of traversal with respect to resource location and view lookup apply.

There are only four real differences between a purely traversal-based application and a hybrid application:

  • In a purely traversal based application, no routes are defined; in a hybrid application, at least one route will be defined.
  • In a purely traversal based application, the root object used is global, implied by the root factory provided at startup time; in a hybrid application, the root object at which traversal begins may be varied on a per-route basis.
  • In a purely traversal-based application, the PATH_INFO of the underlying WSGI environment is used wholesale as a traversal path; in a hybrid application, the traversal path is not the entire PATH_INFO string, but a portion of the URL determined by a matching pattern in the matched route configuration’s pattern.
  • In a purely traversal based application, view configurations which do not mention a route_name argument are considered during view lookup; in a hybrid application, when a route is matched, only view configurations which mention that route’s name as a route_name are considered during view lookup.

More generally, a hybrid application is a traversal-based application except:

  • the traversal root is chosen based on the route configuration of the route that matched instead of from the root_factory supplied during application startup configuration.
  • the traversal path is chosen based on the route configuration of the route that matched rather than from the PATH_INFO of a request.
  • the set of views that may be chosen during view lookup when a route matches are limited to those which specifically name a route_name in their configuration that is the same as the matched route’s name.

To create a hybrid mode application, use a route configuration that implies a particular root factory and which also includes a pattern argument that contains a special dynamic part: either *traverse or *subpath.

The Root Object for a Route Match

A hybrid application implies that traversal is performed during a request after a route has matched. Traversal, by definition, must always begin at a root object. Therefore it’s important to know which root object will be traversed after a route has matched.

Figuring out which root object results from a particular route match is straightforward. When a route is matched:

  • If the route’s configuration has a factory argument which points to a root factory callable, that callable will be called to generate a root object.
  • If the route’s configuration does not have a factory argument, the global root factory will be called to generate a root object. The global root factory is the callable implied by the root_factory argument passed to the Configurator at application startup time.
  • If a root_factory argument is not provided to the Configurator at startup time, a default root factory is used. The default root factory is used to generate a root object.

Note

Root factories related to a route were explained previously within Route Factories. Both the global root factory and default root factory were explained previously within The Resource Tree.

Using *traverse In a Route Pattern

A hybrid application most often implies the inclusion of a route configuration that contains the special token *traverse at the end of a route’s pattern:

1
config.add_route('home', '{foo}/{bar}/*traverse')

A *traverse token at the end of the pattern in a route’s configuration implies a “remainder” capture value. When it is used, it will match the remainder of the path segments of the URL. This remainder becomes the path used to perform traversal.

Note

The *remainder route pattern syntax is explained in more detail within Route Pattern Syntax.

A hybrid mode application relies more heavily on traversal to do resource location and view lookup than most examples indicate within URL Dispatch.

Because the pattern of the above route ends with *traverse, when this route configuration is matched during a request, Pyramid will attempt to use traversal against the root object implied by the root factory that is implied by the route’s configuration. Since no root_factory argument is explicitly specified for this route, this will either be the global root factory for the application, or the default root factory. Once traversal has found a context resource, view lookup will be invoked in almost exactly the same way it would have been invoked in a “pure” traversal-based application.

Let’s assume there is no global root factory configured in this application. The default root factory cannot be traversed: it has no useful __getitem__ method. So we’ll need to associate this route configuration with a custom root factory in order to create a useful hybrid application. To that end, let’s imagine that we’ve created a root factory that looks like so in a module named routes.py:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
class Resource(object):
    def __init__(self, subobjects):
       self.subobjects = subobjects

    def __getitem__(self, name):
       return self.subobjects[name]

root = Resource(
        {'a': Resource({'b': Resource({'c': Resource({})})})}
       )

def root_factory(request):
    return root

Above, we’ve defined a (bogus) resource tree that can be traversed, and a root_factory function that can be used as part of a particular route configuration statement:

1
2
config.add_route('home', '{foo}/{bar}/*traverse',
                 factory='mypackage.routes.root_factory')

The factory above points at the function we’ve defined. It will return an instance of the Resource class as a root object whenever this route is matched. Instances of the Resource class can be used for tree traversal because they have a __getitem__ method that does something nominally useful. Since traversal uses __getitem__ to walk the resources of a resource tree, using traversal against the root resource implied by our route statement is a reasonable thing to do.

Note

We could have also used our root_factory function as the root_factory argument of the Configurator constructor, instead of associating it with a particular route inside the route’s configuration. Every hybrid route configuration that is matched but which does not name a factory attribute will use the use global root_factory function to generate a root object.

When the route configuration named home above is matched during a request, the matchdict generated will be based on its pattern: {foo}/{bar}/*traverse. The “capture value” implied by the *traverse element in the pattern will be used to traverse the resource tree in order to find a context resource, starting from the root object returned from the root factory. In the above example, the root object found will be the instance named root in routes.py.

If the URL that matched a route with the pattern {foo}/{bar}/*traverse, is http://example.com/one/two/a/b/c, the traversal path used against the root object will be a/b/c. As a result, Pyramid will attempt to traverse through the edges 'a', 'b', and 'c', beginning at the root object.

In our above example, this particular set of traversal steps will mean that the context resource of the view would be the Resource object we’ve named 'c' in our bogus resource tree and the view name resulting from traversal will be the empty string; if you need a refresher about why this outcome is presumed, see The Traversal Algorithm.

At this point, a suitable view callable will be found and invoked using view lookup as described in View Configuration, but with a caveat: in order for view lookup to work, we need to define a view configuration that will match when view lookup is invoked after a route matches:

1
2
3
config.add_route('home', '{foo}/{bar}/*traverse',
                 factory='mypackage.routes.root_factory')
config.add_view('mypackage.views.myview', route_name='home')

Note that the above call to add_view() includes a route_name argument. View configurations that include a route_name argument are meant to associate a particular view declaration with a route, using the route’s name, in order to indicate that the view should only be invoked when the route matches.

Calls to add_view() may pass a route_name attribute, which refers to the value of an existing route’s name argument. In the above example, the route name is home, referring to the name of the route defined above it.

The above mypackage.views.myview view callable will be invoked when:

  • the route named “home” is matched
  • the view name resulting from traversal is the empty string.
  • the context resource is any object.

It is also possible to declare alternate views that may be invoked when a hybrid route is matched:

1
2
3
4
5
config.add_route('home', '{foo}/{bar}/*traverse',
                 factory='mypackage.routes.root_factory')
config.add_view('mypackage.views.myview', route_name='home')
config.add_view('mypackage.views.another_view', route_name='home',
                name='another')

The add_view call for mypackage.views.another_view above names a different view and, more importantly, a different view name. The above mypackage.views.another_view view will be invoked when:

  • the route named “home” is matched
  • the view name resulting from traversal is another.
  • the context resource is any object.

For instance, if the URL http://example.com/one/two/a/another is provided to an application that uses the previously mentioned resource tree, the mypackage.views.another view callable will be called instead of the mypackage.views.myview view callable because the view name will be another instead of the empty string.

More complicated matching can be composed. All arguments to route configuration statements and view configuration statements are supported in hybrid applications (such as predicate arguments).

Using the traverse Argument In a Route Definition

Rather than using the *traverse remainder marker in a pattern, you can use the traverse argument to the add_route() method.

When you use the *traverse remainder marker, the traversal path is limited to being the remainder segments of a request URL when a route matches. However, when you use the traverse argument or attribute, you have more control over how to compose a traversal path.

Here’s a use of the traverse pattern in a call to add_route():

1
2
config.add_route('abc', '/articles/{article}/edit',
                 traverse='/{article}')

The syntax of the traverse argument is the same as it is for pattern.

If, as above, the pattern provided is /articles/{article}/edit, and the traverse argument provided is /{article}, when a request comes in that causes the route to match in such a way that the article match value is 1 (when the request URI is /articles/1/edit), the traversal path will be generated as /1. This means that the root object’s __getitem__ will be called with the name 1 during the traversal phase. If the 1 object exists, it will become the context of the request. The Traversal chapter has more information about traversal.

If the traversal path contains segment marker names which are not present in the pattern argument, a runtime error will occur. The traverse pattern should not contain segment markers that do not exist in the path.

Note that the traverse argument is ignored when attached to a route that has a *traverse remainder marker in its pattern.

Traversal will begin at the root object implied by this route (either the global root, or the object returned by the factory associated with this route).

Making Global Views Match

By default, only view configurations that mention a route_name will be found during view lookup when a route that has a *traverse in its pattern matches. You can allow views without a route_name attribute to match a route by adding the use_global_views flag to the route definition. For example, the myproject.views.bazbuz view below will be found if the route named abc below is matched and the PATH_INFO is /abc/bazbuz, even though the view configuration statement does not have the route_name="abc" attribute.

1
2
config.add_route('abc', '/abc/*traverse', use_global_views=True)
config.add_view('myproject.views.bazbuz', name='bazbuz')
Using *subpath in a Route Pattern

There are certain extremely rare cases when you’d like to influence the traversal subpath when a route matches without actually performing traversal. For instance, the pyramid.wsgi.wsgiapp2() decorator and the pyramid.static.static_view helper attempt to compute PATH_INFO from the request’s subpath when its use_subpath argument is True, so it’s useful to be able to influence this value.

When *subpath exists in a pattern, no path is actually traversed, but the traversal algorithm will return a subpath list implied by the capture value of *subpath. You’ll see this pattern most commonly in route declarations that look like this:

1
2
3
4
5
6
from pryamid.static import static_view

www = static_view('mypackage:static', use_subpath=True)

config.add_route('static', '/static/*subpath')
config.add_view(www, route_name='static')

mypackage.views.www is an instance of pyramid.static.static_view. This effectively tells the static helper to traverse everything in the subpath as a filename.

Corner Cases

A number of corner case “gotchas” exist when using a hybrid application. We’ll detail them here.

Registering a Default View for a Route That Has a view Attribute

Warning

As of Pyramid 1.1 this section is slated to be removed in a later documentation release because the the ability to add views directly to the route configuration by passing a view argument to add_route has been deprecated.

It is an error to provide both a view argument to a route configuration and a view configuration which names a route_name that has no name value or the empty name value. For example, this pair of declarations will generate a conflict error at startup time.

1
2
3
config.add_route('home', '{foo}/{bar}/*traverse',
                 view='myproject.views.home')
config.add_view('myproject.views.another', route_name='home')

This is because the view argument to the add_route() above is an implicit default view when that route matches. add_route calls don’t need to supply a view attribute. For example, this add_route call:

1
2
config.add_route('home', '{foo}/{bar}/*traverse',
                 view='myproject.views.home')

Can also be spelled like so:

1
2
config.add_route('home', '{foo}/{bar}/*traverse')
config.add_view('myproject.views.home', route_name='home')

The two spellings are logically equivalent. In fact, the former is just a syntactical shortcut for the latter.

Binding Extra Views Against a Route Configuration that Doesn’t Have a *traverse Element In Its Pattern

Here’s another corner case that just makes no sense:

1
2
3
config.add_route('abc', '/abc', view='myproject.views.abc')
config.add_view('myproject.views.bazbuz', name='bazbuz',
                route_name='abc')

The above view declaration is useless, because it will never be matched when the route it references has matched. Only the view associated with the route itself (myproject.views.abc) will ever be invoked when the route matches, because the default view is always invoked when a route matches and when no post-match traversal is performed.

To make the above view declaration useful, the special *traverse token must end the route’s pattern. For example:

1
2
3
config.add_route('abc', '/abc/*traverse', view='myproject.views.abc')
config.add_view('myproject.views.bazbuz', name='bazbuz',
                route_name='abc')

With the above configuration, the myproject.views.bazbuz view will be invoked when the request URI is /abc/bazbuz, assuming there is no object contained by the root object with the key bazbuz. A different request URI, such as /abc/foo/bar, would invoke the default myproject.views.abc view.

Using Hooks

“Hooks” can be used to influence the behavior of the Pyramid framework in various ways.

Changing the Not Found View

When Pyramid can’t map a URL to view code, it invokes a not found view, which is a view callable. A default notfound view exists. The default not found view can be overridden through application configuration.

The not found view callable is a view callable like any other. The view configuration which causes it to be a “not found” view consists only of naming the pyramid.httpexceptions.HTTPNotFound class as the context of the view configuration.

If your application uses imperative configuration, you can replace the Not Found view by using the pyramid.config.Configurator.add_view() method to register an “exception view”:

1
2
3
from pyramid.httpexceptions import HTTPNotFound
from helloworld.views import notfound_view
config.add_view(notfound_view, context=HTTPNotFound)

Replace helloworld.views.notfound_view with a reference to the view callable you want to use to represent the Not Found view.

Like any other view, the notfound view must accept at least a request parameter, or both context and request. The request is the current request representing the denied action. The context (if used in the call signature) will be the instance of the HTTPNotFound exception that caused the view to be called.

Here’s some sample code that implements a minimal NotFound view callable:

1
2
3
4
from pyramid.httpexceptions import HTTPNotFound

def notfound_view(request):
    return HTTPNotFound()

Note

When a NotFound view callable is invoked, it is passed a request. The exception attribute of the request will be an instance of the HTTPNotFound exception that caused the not found view to be called. The value of request.exception.message will be a value explaining why the not found error was raised. This message will be different when the pyramid.debug_notfound environment setting is true than it is when it is false.

Warning

When a NotFound view callable accepts an argument list as described in Alternate View Callable Argument/Calling Conventions, the context passed as the first argument to the view callable will be the HTTPNotFound exception instance. If available, the resource context will still be available as request.context.

Changing the Forbidden View

When Pyramid can’t authorize execution of a view based on the authorization policy in use, it invokes a forbidden view. The default forbidden response has a 403 status code and is very plain, but the view which generates it can be overridden as necessary.

The forbidden view callable is a view callable like any other. The view configuration which causes it to be a “forbidden” view consists only of naming the pyramid.httpexceptions.HTTPForbidden class as the context of the view configuration.

You can replace the forbidden view by using the pyramid.config.Configurator.add_view() method to register an “exception view”:

1
2
3
from helloworld.views import forbidden_view
from pyramid.httpexceptions import HTTPForbidden
config.add_view(forbidden_view, context=HTTPForbidden)

Replace helloworld.views.forbidden_view with a reference to the Python view callable you want to use to represent the Forbidden view.

Like any other view, the forbidden view must accept at least a request parameter, or both context and request. The context (available as request.context if you’re using the request-only view argument pattern) is the context found by the router when the view invocation was denied. The request is the current request representing the denied action.

Here’s some sample code that implements a minimal forbidden view:

1
2
3
4
5
from pyramid.views import view_config
from pyramid.response import Response

def forbidden_view(request):
    return Response('forbidden')

Note

When a forbidden view callable is invoked, it is passed a request. The exception attribute of the request will be an instance of the HTTPForbidden exception that caused the forbidden view to be called. The value of request.exception.message will be a value explaining why the forbidden was raised and request.exception.result will be extended information about the forbidden exception. These messages will be different when the pyramid.debug_authorization environment setting is true than it is when it is false.

Changing the Request Factory

Whenever Pyramid handles a WSGI request, it creates a request object based on the WSGI environment it has been passed. By default, an instance of the pyramid.request.Request class is created to represent the request object.

The class (aka “factory”) that Pyramid uses to create a request object instance can be changed by passing a request_factory argument to the constructor of the configurator. This argument can be either a callable or a dotted Python name representing a callable.

1
2
3
4
5
6
from pyramid.request import Request

class MyRequest(Request):
    pass

config = Configurator(request_factory=MyRequest)

If you’re doing imperative configuration, and you’d rather do it after you’ve already constructed a configurator it can also be registered via the pyramid.config.Configurator.set_request_factory() method:

1
2
3
4
5
6
7
8
from pyramid.config import Configurator
from pyramid.request import Request

class MyRequest(Request):
    pass

config = Configurator()
config.set_request_factory(MyRequest)

Using The Before Render Event

Subscribers to the pyramid.events.BeforeRender event may introspect and modify the set of renderer globals before they are passed to a renderer. This event object iself has a dictionary-like interface that can be used for this purpose. For example:

1
2
3
4
5
6
 from pyramid.events import subscriber
 from pyramid.events import BeforeRender

 @subscriber(BeforeRender)
 def add_global(event):
     event['mykey'] = 'foo'

An object of this type is sent as an event just before a renderer is invoked (but after the application-level renderer globals factory added via set_renderer_globals_factory, if any, has injected its own keys into the renderer globals dictionary).

If a subscriber attempts to add a key that already exist in the renderer globals dictionary, a KeyError is raised. This limitation is enforced because event subscribers do not possess any relative ordering. The set of keys added to the renderer globals dictionary by all pyramid.events.BeforeRender subscribers and renderer globals factories must be unique.

See the API documentation for the BeforeRender event interface at pyramid.interfaces.IBeforeRender.

Another (deprecated) mechanism which allows event subscribers more control when adding renderer global values exists in Adding Renderer Globals (Deprecated).

Adding Renderer Globals (Deprecated)

Warning

this feature is deprecated as of Pyramid 1.1. A non-deprecated mechanism which allows event subscribers to add renderer global values is documented in Using The Before Render Event.

Whenever Pyramid handles a request to perform a rendering (after a view with a renderer= configuration attribute is invoked, or when any of the methods beginning with render within the pyramid.renderers module are called), renderer globals can be injected into the system values sent to the renderer. By default, no renderer globals are injected, and the “bare” system values (such as request, context, and renderer_name) are the only values present in the system dictionary passed to every renderer.

A callback that Pyramid will call every time a renderer is invoked can be added by passing a renderer_globals_factory argument to the constructor of the configurator. This callback can either be a callable object or a dotted Python name representing such a callable.

1
2
3
4
5
def renderer_globals_factory(system):
    return {'a': 1}

config = Configurator(
         renderer_globals_factory=renderer_globals_factory)

Such a callback must accept a single positional argument (notionally named system) which will contain the original system values. It must return a dictionary of values that will be merged into the system dictionary. See System Values Used During Rendering for description of the values present in the system dictionary.

If you’re doing imperative configuration, and you’d rather do it after you’ve already constructed a configurator it can also be registered via the pyramid.config.Configurator.set_renderer_globals_factory() method:

1
2
3
4
5
6
7
from pyramid.config import Configurator

def renderer_globals_factory(system):
    return {'a': 1}

config = Configurator()
config.set_renderer_globals_factory(renderer_globals_factory)

Using Response Callbacks

Unlike many other web frameworks, Pyramid does not eagerly create a global response object. Adding a response callback allows an application to register an action to be performed against whatever response object is returned by a view, usually in order to mutate the response.

The pyramid.request.Request.add_response_callback() method is used to register a response callback.

A response callback is a callable which accepts two positional parameters: request and response. For example:

1
2
3
4
5
def cache_callback(request, response):
    """Set the cache_control max_age for the response"""
    if request.exception is not None:
        response.cache_control.max_age = 360
request.add_response_callback(cache_callback)

No response callback is called if an unhandled exception happens in application code, or if the response object returned by a view callable is invalid. Response callbacks are, however, invoked when a exception view is rendered successfully: in such a case, the request.exception attribute of the request when it enters a response callback will be an exception object instead of its default value of None.

Response callbacks are called in the order they’re added (first-to-most-recently-added). All response callbacks are called after the NewResponse event is sent. Errors raised by response callbacks are not handled specially. They will be propagated to the caller of the Pyramid router application.

A response callback has a lifetime of a single request. If you want a response callback to happen as the result of every request, you must re-register the callback into every new request (perhaps within a subscriber of a NewRequest event).

Using Finished Callbacks

A finished callback is a function that will be called unconditionally by the Pyramid router at the very end of request processing. A finished callback can be used to perform an action at the end of a request unconditionally.

The pyramid.request.Request.add_finished_callback() method is used to register a finished callback.

A finished callback is a callable which accepts a single positional parameter: request. For example:

1
2
3
4
5
6
7
8
9
import transaction

def commit_callback(request):
    '''commit or abort the transaction associated with request'''
    if request.exception is not None:
        transaction.abort()
    else:
        transaction.commit()
request.add_finished_callback(commit_callback)

Finished callbacks are called in the order they’re added (first-to-most-recently-added). Finished callbacks (unlike a response callback) are always called, even if an exception happens in application code that prevents a response from being generated.

The set of finished callbacks associated with a request are called very late in the processing of that request; they are essentially the very last thing called by the router before a request “ends”. They are called after response processing has already occurred in a top-level finally: block within the router request processing code. As a result, mutations performed to the request provided to a finished callback will have no meaningful effect, because response processing will have already occurred, and the request’s scope will expire almost immediately after all finished callbacks have been processed.

It is often necessary to tell whether an exception occurred within view callable code from within a finished callback: in such a case, the request.exception attribute of the request when it enters a response callback will be an exception object instead of its default value of None.

Errors raised by finished callbacks are not handled specially. They will be propagated to the caller of the Pyramid router application.

A finished callback has a lifetime of a single request. If you want a finished callback to happen as the result of every request, you must re-register the callback into every new request (perhaps within a subscriber of a NewRequest event).

Changing the Traverser

The default traversal algorithm that Pyramid uses is explained in The Traversal Algorithm. Though it is rarely necessary, this default algorithm can be swapped out selectively for a different traversal pattern via configuration.

1
2
3
4
5
from pyramid.interfaces import ITraverser
from zope.interface import Interface
from myapp.traversal import Traverser

config.registry.registerAdapter(Traverser, (Interface,), ITraverser)

In the example above, myapp.traversal.Traverser is assumed to be a class that implements the following interface:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
class Traverser(object):
    def __init__(self, root):
        """ Accept the root object returned from the root factory """

    def __call__(self, request):
        """ Return a dictionary with (at least) the keys ``root``,
        ``context``, ``view_name``, ``subpath``, ``traversed``,
        ``virtual_root``, and ``virtual_root_path``.  These values are
        typically the result of a resource tree traversal.  ``root``
        is the physical root object, ``context`` will be a resource
        object, ``view_name`` will be the view name used (a Unicode
        name), ``subpath`` will be a sequence of Unicode names that
        followed the view name but were not traversed, ``traversed``
        will be a sequence of Unicode names that were traversed
        (including the virtual root path, if any) ``virtual_root``
        will be a resource object representing the virtual root (or the
        physical root if traversal was not performed), and
        ``virtual_root_path`` will be a sequence representing the
        virtual root path (a sequence of Unicode names) or None if
        traversal was not performed.

        Extra keys for special purpose functionality can be added as
        necessary.

        All values returned in the dictionary will be made available
        as attributes of the ``request`` object.
        """

More than one traversal algorithm can be active at the same time. For instance, if your root factory returns more than one type of object conditionally, you could claim that an alternate traverser adapter is for only one particular class or interface. When the root factory returned an object that implemented that class or interface, a custom traverser would be used. Otherwise, the default traverser would be used. For example:

1
2
3
4
5
6
from pyramid.interfaces import ITraverser
from zope.interface import Interface
from myapp.traversal import Traverser
from myapp.resources import MyRoot

config.registry.registerAdapter(Traverser, (MyRoot,), ITraverser)

If the above stanza was added to a Pyramid __init__.py file’s main function, Pyramid would use the myapp.traversal.Traverser only when the application root factory returned an instance of the myapp.resources.MyRoot object. Otherwise it would use the default Pyramid traverser to do traversal.

Changing How pyramid.request.Request.resource_url() Generates a URL

When you add a traverser as described in Changing the Traverser, it’s often convenient to continue to use the pyramid.request.Request.resource_url() API. However, since the way traversal is done will have been modified, the URLs it generates by default may be incorrect.

If you’ve added a traverser, you can change how resource_url() generates a URL for a specific type of resource by adding a registerAdapter call for pyramid.interfaces.IContextURL to your application:

1
2
3
4
5
6
7
from pyramid.interfaces import ITraverser
from zope.interface import Interface
from myapp.traversal import URLGenerator
from myapp.resources import MyRoot

config.registry.registerAdapter(URLGenerator, (MyRoot, Interface),
                                IContextURL)

In the above example, the myapp.traversal.URLGenerator class will be used to provide services to resource_url() any time the context passed to resource_url is of class myapp.resources.MyRoot. The second argument in the (MyRoot, Interface) tuple represents the type of interface that must be possessed by the request (in this case, any interface, represented by zope.interface.Interface).

The API that must be implemented by a class that provides IContextURL is as follows:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
from zope.interface import Interface

class IContextURL(Interface):
    """ An adapter which deals with URLs related to a context.
    """
    def __init__(self, context, request):
        """ Accept the context and request """

    def virtual_root(self):
        """ Return the virtual root object related to a request and the
        current context"""

    def __call__(self):
        """ Return a URL that points to the context """

The default context URL generator is available for perusal as the class pyramid.traversal.TraversalContextURL in the traversal module of the Pylons GitHub Pyramid repository.

Changing How Pyramid Treats View Responses

It is possible to control how Pyramid treats the result of calling a view callable on a per-type basis by using a hook involving pyramid.config.Configurator.add_response_adapter() or the response_adapter decorator.

Note

This feature is new as of Pyramid 1.1.

Pyramid, in various places, adapts the result of calling a view callable to the IResponse interface to ensure that the object returned by the view callable is a “true” response object. The vast majority of time, the result of this adaptation is the result object itself, as view callables written by “civilians” who read the narrative documentation contained in this manual will always return something that implements the IResponse interface. Most typically, this will be an instance of the pyramid.response.Response class or a subclass. If a civilian returns a non-Response object from a view callable that isn’t configured to use a renderer, he will typically expect the router to raise an error. However, you can hook Pyramid in such a way that users can return arbitrary values from a view callable by providing an adapter which converts the arbitrary return value into something that implements IResponse.

For example, if you’d like to allow view callables to return bare string objects (without requiring a a renderer to convert a string to a response object), you can register an adapter which converts the string to a Response:

1
2
3
4
5
6
7
8
9
from pyramid.response import Response

def string_response_adapter(s):
    response = Response(s)
    return response

# config is an instance of pyramid.config.Configurator

config.add_response_adapter(string_response_adapter, str)

Likewise, if you want to be able to return a simplified kind of response object from view callables, you can use the IResponse hook to register an adapter to the more complex IResponse interface:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
from pyramid.response import Response

class SimpleResponse(object):
    def __init__(self, body):
        self.body = body

def simple_response_adapter(simple_response):
    response = Response(simple_response.body)
    return response

# config is an instance of pyramid.config.Configurator

config.add_response_adapter(simple_response_adapter, SimpleResponse)

If you want to implement your own Response object instead of using the pyramid.response.Response object in any capacity at all, you’ll have to make sure the object implements every attribute and method outlined in pyramid.interfaces.IResponse and you’ll have to ensure that it’s marked up with zope.interface.implements(IResponse):

1
2
3
4
5
6
7
from pyramid.interfaces import IResponse
from zope.interface import implements

class MyResponse(object):
    implements(IResponse)
    # ... an implementation of every method and attribute
    # documented in IResponse should follow ...

When an alternate response object implementation is returned by a view callable, if that object asserts that it implements IResponse (via zope.interface.implements(IResponse)) , an adapter needn’t be registered for the object; Pyramid will use it directly.

An IResponse adapter for webob.Response (as opposed to pyramid.response.Response) is registered by Pyramid by default at startup time, as by their nature, instances of this class (and instances of subclasses of the class) will natively provide IResponse. The adapter registered for webob.Response simply returns the response object.

Instead of using pyramid.config.Configurator.add_response_adapter(), you can use the pyramid.response.response_adapter decorator:

1
2
3
4
5
6
7
from pyramid.response import Response
from pyramid.response import response_adapter

@response_adapter(str)
def string_response_adapter(s):
    response = Response(s)
    return response

The above example, when scanned, has the same effect as:

config.add_response_adapter(string_response_adapter, str)

The response_adapter decorator will have no effect until activated by a scan.

Using a View Mapper

The default calling conventions for view callables are documented in the Views chapter. You can change the way users define view callables by employing a view mapper.

A view mapper is an object that accepts a set of keyword arguments and which returns a callable. The returned callable is called with the view callable object. The returned callable should itself return another callable which can be called with the “internal calling protocol” (context, request).

You can use a view mapper in a number of ways:

  • by setting a __view_mapper__ attribute (which is the view mapper object) on the view callable itself
  • by passing the mapper object to pyramid.config.Configurator.add_view() (or its declarative/decorator equivalents) as the mapper argument.
  • by registering a default view mapper.

Here’s an example of a view mapper that emulates (somewhat) a Pylons “controller”. The mapper is initialized with some keyword arguments. Its __call__ method accepts the view object (which will be a class). It uses the attr keyword argument it is passed to determine which attribute should be used as an action method. The wrapper method it returns accepts (context, request) and returns the result of calling the action method with keyword arguments implied by the matchdict after popping the action out of it. This somewhat emulates the Pylons style of calling action methods with routing parameters pulled out of the route matching dict as keyword arguments.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
# framework

class PylonsControllerViewMapper(object):
    def __init__(self, **kw):
        self.kw = kw

    def __call__(self, view):
        attr = self.kw['attr']
        def wrapper(context, request):
            matchdict = request.matchdict.copy()
            matchdict.pop('action', None)
            inst = view()
            meth = getattr(inst, attr)
            return meth(**matchdict)
        return wrapper

class BaseController(object):
    __view_mapper__ = PylonsControllerViewMapper

A user might make use of these framework components like so:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
# user application

from pyramid.response import Response
from pyramid.config import Configurator
import pyramid_handlers
from paste.httpserver import serve

class MyController(BaseController):
    def index(self, id):
        return Response(id)

if __name__ == '__main__':
    config = Configurator()
    config.include(pyramid_handlers)
    config.add_handler('one', '/{id}', MyController, action='index')
    config.add_handler('two', '/{action}/{id}', MyController)
    serve(config.make_wsgi_app())

The pyramid.config.Configurator.set_view_mapper() method can be used to set a default view mapper (overriding the superdefault view mapper used by Pyramid itself).

A single view registration can use a view mapper by passing the mapper as the mapper argument to add_view().

Registering Configuration Decorators

Decorators such as view_config don’t change the behavior of the functions or classes they’re decorating. Instead, when a scan is performed, a modified version of the function or class is registered with Pyramid.

You may wish to have your own decorators that offer such behaviour. This is possible by using the Venusian package in the same way that it is used by Pyramid.

By way of example, let’s suppose you want to write a decorator that registers the function it wraps with a Zope Component Architecture “utility” within the application registry provided by Pyramid. The application registry and the utility inside the registry is likely only to be available once your application’s configuration is at least partially completed. A normal decorator would fail as it would be executed before the configuration had even begun.

However, using Venusian, the decorator could be written as follows:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
import venusian
from mypackage.interfaces import IMyUtility

class registerFunction(object):

    def __init__(self, path):
        self.path = path

    def register(self, scanner, name, wrapped):
        registry = scanner.config.registry
        registry.getUtility(IMyUtility).register(
            self.path, wrapped)

    def __call__(self, wrapped):
        venusian.attach(wrapped, self.register)
        return wrapped

This decorator could then be used to register functions throughout your code:

1
2
3
@registerFunction('/some/path')
def my_function():
   do_stuff()

However, the utility would only be looked up when a scan was performed, enabling you to set up the utility in advance:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
from paste.httpserver import serve
from pyramid.config import Configurator
from mypackage.interfaces import IMyUtility

class UtilityImplementation:

    implements(IMyUtility)

    def __init__(self):
       self.registrations = {}

    def register(self, path, callable_):
       self.registrations[path] = callable_

if __name__ == '__main__':
    config = Configurator()
    config.registry.registerUtility(UtilityImplementation())
    config.scan()
    app = config.make_wsgi_app()
    serve(app, host='0.0.0.0')

For full details, please read the Venusian documentation.

Registering “Tweens”

Note

Tweens are a feature which were added in Pyramid 1.2. They are not available in any previous version.

A tween (a contraction of the word “between”) is a bit of code that sits between the Pyramid router’s main request handling function and the upstream WSGI component that uses Pyramid as its “app”. This is a feature that may be used by Pyramid framework extensions, to provide, for example, Pyramid-specific view timing support bookkeeping code that examines exceptions before they are returned to the upstream WSGI application. Tweens behave a bit like WSGI middleware but they have the benefit of running in a context in which they have access to the Pyramid application registry as well as the Pyramid rendering machinery.

Creating a Tween Factory

To make use of tweens, you must construct a “tween factory”. A tween factory must be a globally importable callable which accepts two arguments: handler and registry. handler will be the either the main Pyramid request handling function or another tween. registry will be the Pyramid application registry represented by this Configurator. A tween factory must return a tween when it is called.

A tween is a callable which accepts a request object and returns a response object.

Here’s an example of a tween factory:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
 # in a module named myapp.tweens

 import time
 from pyramid.settings import asbool
 import logging

 log = logging.getLogger(__name__)

 def timing_tween_factory(handler, registry):
     if asbool(registry.settings.get('do_timing')):
         # if timing support is enabled, return a wrapper
         def timing_tween(request):
             start = time.time()
             try:
                 response = handler(request)
             finally:
                 end = time.time()
                 log.debug('The request took %s seconds' %
                           (end - start))
             return response
         return timing_tween
     # if timing support is not enabled, return the original
     # handler
     return handler

If you remember, a tween is an object which accepts a request object and which returns a response argument. The request argument to a tween will be the request created by Pyramid’s router when it receives a WSGI request. The response object will be generated by the downstream Pyramid application and it should be returned by the tween.

In the above example, the tween factory defines a timing_tween tween and returns it if asbool(registry.settings.get('do_timing')) is true. It otherwise simply returns the handler it was given. The registry.settings attribute is a handle to the deployment settings provided by the user (usually in an .ini file). In this case, if the user has defined a do_timing setting, and that setting is True, the user has said she wants to do timing, so the tween factory returns the timing tween; it otherwise just returns the handler it has been provided, preventing any timing.

The example timing tween simply records the start time, calls the downstream handler, logs the number of seconds consumed by the downstream handler, and returns the response.

Registering an Implicit Tween Factory

Once you’ve created a tween factory, you can register it into the implicit tween chain using the pyramid.config.Configurator.add_tween() method using its dotted Python name.

Here’s an example of registering the a tween factory as an “implicit” tween in a Pyramid application:

1
2
3
 from pyramid.config import Configurator
 config = Configurator()
 config.add_tween('myapp.tweens.timing_tween_factory')

Note that you must use a dotted Python name as the first argument to pyramid.config.Configurator.add_tween(); this must point at a tween factory. You cannot pass the tween factory object itself to the method: it must be dotted Python name that points to a globally importable object. In the above example, we assume that a timing_tween_factory tween factory was defined in a module named myapp.tweens, so the tween factory is importable as myapp.tweens.timing_tween_factory.

When you use pyramid.config.Configurator.add_tween(), you’re instructing the system to use your tween factory at startup time unless the user has provided an explicit tween list in his configuration. This is what’s meant by an “implicit” tween. A user can always elect to supply an explicit tween list, reordering or disincluding implicitly added tweens. See Explicit Tween Ordering for more information about explicit tween ordering.

If more than one call to pyramid.config.Configurator.add_tween() is made within a single application configuration, the tweens will be chained together at application startup time. The first tween factory added via add_tween will be called with the Pyramid exception view tween factory as its handler argument, then the tween factory added directly after that one will be called with the result of the first tween factory as its handler argument, and so on, ad infinitum until all tween factories have been called. The Pyramid router will use the outermost tween produced by this chain (the tween generated by the very last tween factory added) as its request handler function. For example:

1
2
3
4
5
 from pyramid.config import Configurator

 config = Configurator()
 config.add_tween('myapp.tween_factory1')
 config.add_tween('myapp.tween_factory2')

The above example will generate an implicit tween chain that looks like this:

INGRESS (implicit)
myapp.tween_factory2
myapp.tween_factory1
pyramid.tweens.excview_tween_factory (implicit)
MAIN (implicit)
Suggesting Implicit Tween Ordering

By default, as described above, the ordering of the chain is controlled entirely by the relative ordering of calls to pyramid.config.Configurator.add_tween(). However, the caller of add_tween can provide an optional hint that can influence the implicit tween chain ordering by supplying under or over (or both) arguments to add_tween(). These hints are only used used when an explicit tween ordering is not used. See Explicit Tween Ordering for a description of how to set an explicit tween ordering.

Allowable values for under or over (or both) are:

  • None (the default).
  • A dotted Python name to a tween factory: a string representing the predicted dotted name of a tween factory added in a call to add_tween in the same configuration session.
  • One of the constants pyramid.tweens.MAIN, pyramid.tweens.INGRESS, or pyramid.tweens.EXCVIEW.
  • An iterable of any combination of the above. This allows the user to specify fallbacks if the desired tween is not included, as well as compatibility with multiple other tweens.

Effectively, under means “closer to the main Pyramid application than”, over means “closer to the request ingress than”.

For example, the following call to add_tween() will attempt to place the tween factory represented by myapp.tween_factory directly ‘above’ (in paster ptweens order) the main Pyramid request handler.

1
2
3
import pyramid.tweens

config.add_tween('myapp.tween_factory', over=pyramid.tweens.MAIN)

The above example will generate an implicit tween chain that looks like this:

INGRESS (implicit)
pyramid.tweens.excview_tween_factory (implicit)
myapp.tween_factory
MAIN (implicit)

Likewise, calling the following call to add_tween() will attempt to place this tween factory ‘above’ the main handler but ‘below’ a separately added tween factory:

1
2
3
4
5
6
7
import pyramid.tweens

config.add_tween('myapp.tween_factory1',
                 over=pyramid.tweens.MAIN)
config.add_tween('myapp.tween_factory2',
                 over=pyramid.tweens.MAIN,
                 under='myapp.tween_factory1')

The above example will generate an implicit tween chain that looks like this:

INGRESS (implicit)
pyramid.tweens.excview_tween_factory (implicit)
myapp.tween_factory1
myapp.tween_factory2
MAIN (implicit)

Specifying neither over nor under is equivalent to specifying under=INGRESS.

If all options for under (or over) cannot be found in the current configuration, it is an error. If some options are specified purely for compatibilty with other tweens, just add a fallback of MAIN or INGRESS. For example, under=('someothertween', 'someothertween2', INGRESS). This constraint will require the tween to be located under both the ‘someothertween’ tween, the ‘someothertween2’ tween, and INGRESS. If any of these is not in the current configuration, this constraint will only organize itself based on the tweens that are present.

Explicit Tween Ordering

Implicit tween ordering is obviously only best-effort. Pyramid will attempt to provide an implicit order of tweens as best it can using hints provided by calls to add_tween(), but because it’s only best-effort, if very precise tween ordering is required, the only surefire way to get it is to use an explicit tween order. The deploying user can override the implicit tween inclusion and ordering implied by calls to add_tween() entirely by using the pyramid.tweens settings value. When used, this settings value must be a list of Python dotted names which will override the ordering (and inclusion) of tween factories in the implicit tween chain. For example:

1
2
3
4
5
6
7
8
9
[app:main]
use = egg:MyApp
pyramid.reload_templates = true
pyramid.debug_authorization = false
pyramid.debug_notfound = false
pyramid.debug_routematch = false
pyramid.debug_templates = true
pyramid.tweens = myapp.my_cool_tween_factory
                 pyramid.tweens.excview_tween_factory

In the above configuration, calls made during configuration to pyramid.config.Configurator.add_tween() are ignored, and the user is telling the system to use the tween factories he has listed in the pyramid.tweens configuration setting (each is a dotted Python name which points to a tween factory) instead of any tween factories added via pyramid.config.Configurator.add_tween(). The first tween factory in the pyramid.tweens list will be used as the producer of the effective Pyramid request handling function; it will wrap the tween factory declared directly “below” it, ad infinitum. The “main” Pyramid request handler is implicit, and always “at the bottom”.

Note

Pyramid’s own exception view handling logic is implemented as a tween factory function: pyramid.tweens.excview_tween_factory(). If Pyramid exception view handling is desired, and tween factories are specified via the pyramid.tweens configuration setting, the pyramid.tweens.excview_tween_factory() function must be added to the pyramid.tweens configuration setting list explicitly. If it is not present, Pyramid will not perform exception view handling.

Tween Conflicts and Ordering Cycles

Pyramid will prevent the same tween factory from being added to the tween chain more than once using configuration conflict detection. If you wish to add the same tween factory more than once in a configuration, you should either: a) use a tween factory that is a separate globally importable instance object from the factory that it conflicts with b) use a function or class as a tween factory with the same logic as the other tween factory it conflicts with but with a different __name__ attribute or c) call pyramid.config.Configurator.commit() between calls to pyramid.config.Configurator.add_tween().

If a cycle is detected in implicit tween ordering when over and under are used in any call to “add_tween”, an exception will be raised at startup time.

Displaying Tween Ordering

The paster ptweens command-line utility can be used to report the current implict and explicit tween chains used by an application. See Displaying “Tweens”.

Advanced Configuration

To support application extensibility, the Pyramid Configurator, by default, detects configuration conflicts and allows you to include configuration imperatively from other packages or modules. It also, by default, performs configuration in two separate phases. This allows you to ignore relative configuration statement ordering in some circumstances.

Conflict Detection

Here’s a familiar example of one of the simplest Pyramid applications, configured imperatively:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
from paste.httpserver import serve
from pyramid.config import Configurator
from pyramid.response import Response

def hello_world(request):
    return Response('Hello world!')

if __name__ == '__main__':
    config = Configurator()
    config.add_view(hello_world)
    app = config.make_wsgi_app()
    serve(app, host='0.0.0.0')

When you start this application, all will be OK. However, what happens if we try to add another view to the configuration with the same set of predicate arguments as one we’ve already added?

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
from paste.httpserver import serve
from pyramid.config import Configurator
from pyramid.response import Response

def hello_world(request):
    return Response('Hello world!')

def goodbye_world(request):
    return Response('Goodbye world!')

if __name__ == '__main__':
    config = Configurator()

    config.add_view(hello_world, name='hello')

    # conflicting view configuration
    config.add_view(goodbye_world, name='hello')

    app = config.make_wsgi_app()
    serve(app, host='0.0.0.0')

The application now has two conflicting view configuration statements. When we try to start it again, it won’t start. Instead, we’ll receive a traceback that ends something like this:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
Traceback (most recent call last):
  File "app.py", line 12, in <module>
    app = config.make_wsgi_app()
  File "pyramid/config.py", line 839, in make_wsgi_app
    self.commit()
  File "pyramid/pyramid/config.py", line 473, in commit
    self._ctx.execute_actions()
  ... more code ...
pyramid.exceptions.ConfigurationConflictError:
        Conflicting configuration actions
  For: ('view', None, '', None, <InterfaceClass pyramid.interfaces.IView>,
        None, None, None, None, None, False, None, None, None)
 ('app.py', 14, '<module>', 'config.add_view(hello_world)')
 ('app.py', 17, '<module>', 'config.add_view(hello_world)')

This traceback is trying to tell us:

  • We’ve got conflicting information for a set of view configuration statements (The For: line).
  • There are two statements which conflict, shown beneath the For: line: config.add_view(hello_world. 'hello') on line 14 of app.py, and config.add_view(goodbye_world, 'hello') on line 17 of app.py.

These two configuration statements are in conflict because we’ve tried to tell the system that the set of predicate values for both view configurations are exactly the same. Both the hello_world and goodbye_world views are configured to respond under the same set of circumstances. This circumstance: the view name (represented by the name= predicate) is hello.

This presents an ambiguity that Pyramid cannot resolve. Rather than allowing the circumstance to go unreported, by default Pyramid raises a ConfigurationConflictError error and prevents the application from running.

Conflict detection happens for any kind of configuration: imperative configuration or configuration that results from the execution of a scan.

Manually Resolving Conflicts

There are a number of ways to manually resolve conflicts: the “right” way, by strategically using pyramid.config.Configurator.commit(), or by using an “autocommitting” configurator.

The Right Thing

The most correct way to resolve conflicts is to “do the needful”: change your configuration code to not have conflicting configuration statements. The details of how this is done depends entirely on the configuration statements made by your application. Use the detail provided in the ConfigurationConflictError to track down the offending conflicts and modify your configuration code accordingly.

If you’re getting a conflict while trying to extend an existing application, and that application has a function which performs configuration like this one:

1
2
def add_routes(config):
    config.add_route(...)

Don’t call this function directly with config as an argument. Instead, use pyramid.config.Configuration.include():

1
config.include(add_routes)

Using include() instead of calling the function directly provides a modicum of automated conflict resolution, with the configuration statements you define in the calling code overriding those of the included function. See also Automatic Conflict Resolution and Including Configuration from External Sources.

Using config.commit()

You can manually commit a configuration by using the commit() method between configuration calls. For example, we prevent conflicts from occurring in the application we examined previously as the result of adding a commit. Here’s the application that generates conflicts:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
from paste.httpserver import serve
from pyramid.config import Configurator
from pyramid.response import Response

def hello_world(request):
    return Response('Hello world!')

def goodbye_world(request):
    return Response('Goodbye world!')

if __name__ == '__main__':
    config = Configurator()

    config.add_view(hello_world, name='hello')

    # conflicting view configuration
    config.add_view(goodbye_world, name='hello')

    app = config.make_wsgi_app()
    serve(app, host='0.0.0.0')

We can prevent the two add_view calls from conflicting by issuing a call to commit() between them:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
from paste.httpserver import serve
from pyramid.config import Configurator
from pyramid.response import Response

def hello_world(request):
    return Response('Hello world!')

def goodbye_world(request):
    return Response('Goodbye world!')

if __name__ == '__main__':
    config = Configurator()

    config.add_view(hello_world, name='hello')

    config.commit() # commit any pending configuration actions

    # no-longer-conflicting view configuration
    config.add_view(goodbye_world, name='hello')

    app = config.make_wsgi_app()
    serve(app, host='0.0.0.0')

In the above example we’ve issued a call to commit() between the two add_view calls. commit() will cause any pending configuration statements.

Calling commit() is safe at any time. It executes all pending configuration actions and leaves the configuration action list “clean”.

Note that commit() has no effect when you’re using an autocommitting configurator (see Using An Autocommitting Configurator).

Using An Autocommitting Configurator

You can also use a heavy hammer to circumvent conflict detection by using a configurator constructor parameter: autocommit=True. For example:

1
2
3
4
from pyramid.config import Configurator

if __name__ == '__main__':
    config = Configurator(autocommit=True)

When the autocommit parameter passed to the Configurator is True, conflict detection (and Two-Phase Configuration) is disabled. Configuration statements will be executed immediately, and succeeding statements will override preceding ones.

commit() has no effect when autocommit is True.

If you use a Configurator in code that performs unit testing, it’s usually a good idea to use an autocommitting Configurator, because you are usually unconcerned about conflict detection or two-phase configuration in test code.

Automatic Conflict Resolution

If your code uses the include() method to include external configuration, some conflicts are automatically resolved. Configuration statements that are made as the result of an “include” will be overridden by configuration statements that happen within the caller of the “include” method.

Automatic conflict resolution supports this goal: if a user wants to reuse a Pyramid application, and they want to customize the configuration of this application without hacking its code “from outside”, they can “include” a configuration function from the package and override only some of its configuration statements within the code that does the include. No conflicts will be generated by configuration statements within the code which does the including, even if configuration statements in the included code would conflict if it was moved “up” to the calling code.

Methods Which Provide Conflict Detection

These are the methods of the configurator which provide conflict detection:

add_view(), add_route(), add_renderer(), set_request_factory(), set_renderer_globals_factory(), set_locale_negotiator() and set_default_permission().

add_static_view() also indirectly provides conflict detection, because it’s implemented in terms of the conflict-aware add_route and add_view methods.

Including Configuration from External Sources

Some application programmers will factor their configuration code in such a way that it is easy to reuse and override configuration statements. For example, such a developer might factor out a function used to add routes to his application:

1
2
def add_routes(config):
    config.add_route(...)

Rather than calling this function directly with config as an argument. Instead, use pyramid.config.Configuration.include():

1
config.include(add_routes)

Using include rather than calling the function directly will allow Automatic Conflict Resolution to work.

include() can also accept a module as an argument:

1
2
3
import myapp

config.include(myapp)

For this to work properly, the myapp module must contain a callable with the special name includeme, which should perform configuration (like the add_routes callable we showed above as an example).

include() can also accept a dotted Python name to a function or a module.

Two-Phase Configuration

When a non-autocommitting Configurator is used to do configuration (the default), configuration execution happens in two phases. In the first phase, “eager” configuration actions (actions that must happen before all others, such as registering a renderer) are executed, and discriminators are computed for each of the actions that depend on the result of the eager actions. In the second phase, the discriminators of all actions are compared to do conflict detection.

Due to this, for configuration methods that have no internal ordering constraints, execution order of configuration method calls is not important. For example, the relative ordering of add_view() and add_renderer() is unimportant when a non-autocommitting configurator is used. This code snippet:

1
2
config.add_view('some.view', renderer='path_to_custom/renderer.rn')
config.add_renderer('.rn', SomeCustomRendererFactory)

Has the same result as:

1
2
config.add_renderer('.rn', SomeCustomRendererFactory)
config.add_view('some.view', renderer='path_to_custom/renderer.rn')

Even though the view statement depends on the registration of a custom renderer, due to two-phase configuration, the order in which the configuration statements are issued is not important. add_view will be able to find the .rn renderer even if add_renderer is called after add_view.

The same is untrue when you use an autocommitting configurator (see Using An Autocommitting Configurator). When an autocommitting configurator is used, two-phase configuration is disabled, and configuration statements must be ordered in dependency order.

Some configuration methods, such as add_route() have internal ordering constraints: the routes they imply require relative ordering. Such ordering constraints are not absolved by two-phase configuration. Routes are still added in configuration execution order.

Adding Methods to the Configurator via add_directive

Framework extension writers can add arbitrary methods to a Configurator by using the pyramid.config.Configurator.add_directive() method of the configurator. This makes it possible to extend a Pyramid configurator in arbitrary ways, and allows it to perform application-specific tasks more succinctly.

The add_directive() method accepts two positional arguments: a method name and a callable object. The callable object is usually a function that takes the configurator instance as its first argument and accepts other arbitrary positional and keyword arguments. For example:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
from pyramid.events import NewRequest
from pyramid.config import Configurator

def add_newrequest_subscriber(config, subscriber):
    config.add_subscriber(subscriber, NewRequest).

if __name__ == '__main__':
   config = Configurator()
   config.add_directive('add_newrequest_subscriber',
                        add_newrequest_subscriber)

Once add_directive() is called, a user can then call the method by its given name as if it were a built-in method of the Configurator:

1
2
3
4
def mysubscriber(event):
   print event.request

config.add_newrequest_subscriber(mysubscriber)

A call to add_directive() is often “hidden” within an includeme function within a “frameworky” package meant to be included as per Including Configuration from External Sources via include(). For example, if you put this code in a package named pyramid_subscriberhelpers:

1
2
3
def includeme(config)
   config.add_directive('add_newrequest_subscriber',
                        add_newrequest_subscriber)

The user of the add-on package pyramid_subscriberhelpers would then be able to install it and subsequently do:

1
2
3
4
5
6
7
def mysubscriber(event):
   print event.request

from pyramid.config import Configurator
config = Configurator()
config.include('pyramid_subscriberhelpers')
config.add_newrequest_subscriber(mysubscriber)

Extending An Existing Pyramid Application

If a Pyramid developer has obeyed certain constraints while building an application, a third party should be able to change the application’s behavior without needing to modify its source code. The behavior of a Pyramid application that obeys certain constraints can be overridden or extended without modification.

We’ll define some jargon here for the benefit of identifying the parties involved in such an effort.

Developer
The original application developer.
Integrator
Another developer who wishes to reuse the application written by the original application developer in an unanticipated context. He may also wish to modify the original application without changing the original application’s source code.

The Difference Between “Extensible” and “Pluggable” Applications

Other web frameworks, such as Django, advertise that they allow developers to create “pluggable applications”. They claim that if you create an application in a certain way, it will be integratable in a sensible, structured way into another arbitrarily-written application or project created by a third-party developer.

Pyramid, as a platform, does not claim to provide such a feature. The platform provides no guarantee that you can create an application and package it up such that an arbitrary integrator can use it as a subcomponent in a larger Pyramid application or project. Pyramid does not mandate the constraints necessary for such a pattern to work satisfactorily. Because Pyramid is not very “opinionated”, developers are able to use wildly different patterns and technologies to build an application. A given Pyramid application may happen to be reusable by a particular third party integrator, because the integrator and the original developer may share similar base technology choices (such as the use of a particular relational database or ORM). But the same application may not be reusable by a different developer, because he has made different technology choices which are incompatible with the original developer’s.

As a result, the concept of a “pluggable application” is left to layers built above Pyramid, such as a “CMS” layer or “application server” layer. Such layers are apt to provide the necessary “opinions” (such as mandating a storage layer, a templating system, and a structured, well-documented pattern of registering that certain URLs map to certain bits of code) which makes the concept of a “pluggable application” possible. “Pluggable applications”, thus, should not plug in to Pyramid itself but should instead plug into a system written atop Pyramid.

Although it does not provide for “pluggable applications”, Pyramid does provide a rich set of mechanisms which allows for the extension of a single existing application. Such features can be used by frameworks built using Pyramid as a base. All Pyramid applications may not be pluggable, but all Pyramid applications are extensible.

Rules for Building An Extensible Application

There is only one rule you need to obey if you want to build a maximally extensible Pyramid application: as a developer, you should factor any overrideable imperative configuration you’ve created into functions which can be used via pyramid.config.Configurator.include() rather than inlined as calls to methods of a Configurator within the main function in your application’s __init__.py. For example, rather than:

1
2
3
4
5
6
from pyramid.config import Configurator

if __name__ == '__main__':
    config = Configurator()
    config.add_view('myapp.views.view1', name='view1')
    config.add_view('myapp.views.view2', name='view2')

You should do move the calls to add_view outside of the (non-reusable) if __name__ == '__main__' block, and into a reusable function:

1
2
3
4
5
6
7
8
9
from pyramid.config import Configurator

if __name__ == '__main__':
    config = Configurator()
    config.include(add_views)

def add_views(config):
    config.add_view('myapp.views.view1', name='view1')
    config.add_view('myapp.views.view2', name='view2')

Doing this allows an integrator to maximally reuse the configuration statements that relate to your application by allowing him to selectively include or disinclude the configuration functions you’ve created from an “override package”.

Alternately, you can use ZCML for the purpose of making configuration extensible and overrideable. ZCML declarations that belong to an application can be overridden and extended by integrators as necessary in a similar fashion. If you use only ZCML to configure your application, it will automatically be maximally extensible without any manual effort. See pyramid_zcml for information about using ZCML.

Fundamental Plugpoints

The fundamental “plug points” of an application developed using Pyramid are routes, views, and assets. Routes are declarations made using the pyramid.config.Configurator.add_route() method. Views are declarations made using the pyramid.config.Configurator.add_view() method. Assets are files that are accessed by Pyramid using the pkg_resources API such as static files and templates via a asset specification. Other directives and configurator methods also deal in routes, views, and assets. For example, the add_handler directive of the pyramid_handlers package adds a single route, and some number of views.

Extending an Existing Application

The steps for extending an existing application depend largely on whether the application does or does not use configuration decorators and/or imperative code.

If The Application Has Configuration Decorations

You’ve inherited a Pyramid application which you’d like to extend or override that uses pyramid.view.view_config decorators or other configuration decoration decorators.

If you just want to extend the application, you can run a scan against the application’s package, then add additional configuration that registers more views or routes.

1
2
3
if __name__ == '__main__':
    config.scan('someotherpackage')
    config.add_view('mypackage.views.myview', name='myview')

If you want to override configuration in the application, you may need to run pyramid.config.Configurator.commit() after performing the scan of the original package, then add additional configuration that registers more views or routes which performs overrides.

1
2
3
4
if __name__ == '__main__':
    config.scan('someotherpackage')
    config.commit()
    config.add_view('mypackage.views.myview', name='myview')

Once this is done, you should be able to extend or override the application like any other (see Extending the Application).

You can alternately just prevent a scan from happening (by omitting any call to the pyramid.config.Configurator.scan() method). This will cause the decorators attached to objects in the target application to do nothing. At this point, you will need to convert all the configuration done in decorators into equivalent imperative configuration or ZCML and add that configuration or ZCML to a separate Python package as described in Extending the Application.

Extending the Application

To extend or override the behavior of an existing application, you will need to create a new package which includes the configuration of the old package, and you’ll perhaps need to create implementations of the types of things you’d like to override (such as views), which are referred to within the original package.

The general pattern for extending an existing application looks something like this:

  • Create a new Python package. The easiest way to do this is to create a new Pyramid application using the scaffold mechanism. See Creating the Project for more information.
  • In the new package, create Python files containing views and other overridden elements, such as templates and static assets as necessary.
  • Install the new package into the same Python environment as the original application (e.g. python setup.py develop or python setup.py install).
  • Change the main function in the new package’s __init__.py to include the original Pyramid application’s configuration functions via pyramid.config.Configurator.include() statements or a scan.
  • Wire the new views and assets created in the new package up using imperative registrations within the main function of the __init__.py file of the new application. These wiring should happen after including the configuration functions of the old application. These registrations will extend or override any registrations performed by the original application. See Overriding Views, Overriding Routes and Overriding Assets.
Overriding Views

The view configuration declarations you make which override application behavior will usually have the same view predicate attributes as the original you wish to override. These <view> declarations will point at “new” view code, in the override package you’ve created. The new view code itself will usually be cut-n-paste copies of view callables from the original application with slight tweaks.

For example, if the original application has the following configure_views configuration method:

1
2
 def configure_views(config):
     config.add_view('theoriginalapp.views.theview', name='theview')

You can override the first view configuration statement made by configure_views within the override package, after loading the original configuration function:

1
2
3
4
5
6
7
from pyramid.config import Configurator
from originalapp import configure_views

if __name == '__main__':
    config = Configurator()
    config.include(configure_views)
    config.add_view('theoverrideapp.views.theview', name='theview')

In this case, the theoriginalapp.views.theview view will never be executed. Instead, a new view, theoverrideapp.views.theview will be executed instead, when request circumstances dictate.

A similar pattern can be used to extend the application with add_view declarations. Just register a new view against some other set of predicates to make sure the URLs it implies are available on some other page rendering.

Overriding Routes

Route setup is currently typically performed in a sequence of ordered calls to add_route(). Because these calls are ordered relative to each other, and because this ordering is typically important, you should retain their relative ordering when performing an override. Typically, this means copying all the add_route statements into the override package’s file and changing them as necessary. Then disinclude any add_route statements from the original application.

Overriding Assets

Assets are files on the filesystem that are accessible within a Python package. An entire chapter is devoted to assets: Static Assets. Within this chapter is a section named Overriding Assets. This section of that chapter describes in detail how to override package assets with other assets by using the pyramid.config.Configurator.override_asset() method. Add such override_asset calls to your override package’s __init__.py to perform overrides.

Request Processing

Once a Pyramid application is up and running, it is ready to accept requests and return responses.

What happens from the time a WSGI request enters a Pyramid application through to the point that Pyramid hands off a response back to WSGI for upstream processing?

  1. A user initiates a request from his browser to the hostname and port number of the WSGI server used by the Pyramid application.
  2. The WSGI server used by the Pyramid application passes the WSGI environment to the __call__ method of the Pyramid router object.
  3. A request object is created based on the WSGI environment.
  4. The application registry and the request object created in the last step are pushed on to the thread local stack that Pyramid uses to allow the functions named get_current_request() and get_current_registry() to work.
  5. A NewRequest event is sent to any subscribers.
  6. If any route has been defined within application configuration, the Pyramid router calls a URL dispatch “route mapper.” The job of the mapper is to examine the request to determine whether any user-defined route matches the current WSGI environment. The router passes the request as an argument to the mapper.
  7. If any route matches, the request is mutated; a matchdict and matched_route attributes are added to the request object; the former contains a dictionary representing the matched dynamic elements of the request’s PATH_INFO value, the latter contains the IRoute object representing the route which matched. The root object associated with the route found is also generated: if the route configuration which matched has an associated a factory argument, this factory is used to generate the root object, otherwise a default root factory is used.
  8. If a route match was not found, and a root_factory argument was passed to the Configurator constructor, that callable is used to generate the root object. If the root_factory argument passed to the Configurator constructor was None, a default root factory is used to generate a root object.
  9. The Pyramid router calls a “traverser” function with the root object and the request. The traverser function attempts to traverse the root object (using any existing __getitem__ on the root object and subobjects) to find a context. If the root object has no __getitem__ method, the root itself is assumed to be the context. The exact traversal algorithm is described in Traversal. The traverser function returns a dictionary, which contains a context and a view name as well as other ancillary information.
  10. The request is decorated with various names returned from the traverser (such as context, view_name, and so forth), so they can be accessed via e.g. request.context within view code.
  11. A ContextFound event is sent to any subscribers.
  12. Pyramid looks up a view callable using the context, the request, and the view name. If a view callable doesn’t exist for this combination of objects (based on the type of the context, the type of the request, and the value of the view name, and any predicate attributes applied to the view configuration), Pyramid raises a HTTPNotFound exception, which is meant to be caught by a surrounding exception view.
  13. If a view callable was found, Pyramid attempts to call the view function.
  14. If an authorization policy is in use, and the view was protected by a permission, Pyramid passes the context, the request, and the view_name to a function which determines whether the view being asked for can be executed by the requesting user, based on credential information in the request and security information attached to the context. If it returns True, Pyramid calls the view callable to obtain a response. If it returns False, it raises a HTTPForbidden exception, which is meant to be called by a surrounding exception view.
  15. If any exception was raised within a root factory, by traversal, by a view callable or by Pyramid itself (such as when it raises HTTPNotFound or HTTPForbidden), the router catches the exception, and attaches it to the request as the exception attribute. It then attempts to find a exception view for the exception that was caught. If it finds an exception view callable, that callable is called, and is presumed to generate a response. If an exception view that matches the exception cannot be found, the exception is reraised.
  16. The following steps occur only when a response could be successfully generated by a normal view callable or an exception view callable. Pyramid will attempt to execute any response callback functions attached via add_response_callback(). A NewResponse event is then sent to any subscribers. The response object’s __call__ method is then used to generate a WSGI response. The response is sent back to the upstream WSGI server.
  17. Pyramid will attempt to execute any finished callback functions attached via add_finished_callback().
  18. The thread local stack is popped.
_images/router.png

This is a very high-level overview that leaves out various details. For more detail about subsystems invoked by the Pyramid router such as traversal, URL dispatch, views, and event processing, see URL Dispatch, Views, and Using Events.

Thread Locals

A thread local variable is a variable that appears to be a “global” variable to an application which uses it. However, unlike a true global variable, one thread or process serving the application may receive a different value than another thread or process when that variable is “thread local”.

When a request is processed, Pyramid makes two thread local variables available to the application: a “registry” and a “request”.

Why and How Pyramid Uses Thread Local Variables

How are thread locals beneficial to Pyramid and application developers who use Pyramid? Well, usually they’re decidedly not. Using a global or a thread local variable in any application usually makes it a lot harder to understand for a casual reader. Use of a thread local or a global is usually just a way to avoid passing some value around between functions, which is itself usually a very bad idea, at least if code readability counts as an important concern.

For historical reasons, however, thread local variables are indeed consulted by various Pyramid API functions. For example, the implementation of the pyramid.security function named authenticated_userid() retrieves the thread local application registry as a matter of course to find an authentication policy. It uses the pyramid.threadlocal.get_current_registry() function to retrieve the application registry, from which it looks up the authentication policy; it then uses the authentication policy to retrieve the authenticated user id. This is how Pyramid allows arbitrary authentication policies to be “plugged in”.

When they need to do so, Pyramid internals use two API functions to retrieve the request and application registry: get_current_request() and get_current_registry(). The former returns the “current” request; the latter returns the “current” registry. Both get_current_* functions retrieve an object from a thread-local data structure. These API functions are documented in pyramid.threadlocal.

These values are thread locals rather than true globals because one Python process may be handling multiple simultaneous requests or even multiple Pyramid applications. If they were true globals, Pyramid could not handle multiple simultaneous requests or allow more than one Pyramid application instance to exist in a single Python process.

Because one Pyramid application is permitted to call another Pyramid application from its own view code (perhaps as a WSGI app with help from the pyramid.wsgi.wsgiapp2() decorator), these variables are managed in a stack during normal system operations. The stack instance itself is a threading.local.

During normal operations, the thread locals stack is managed by a Router object. At the beginning of a request, the Router pushes the application’s registry and the request on to the stack. At the end of a request, the stack is popped. The topmost request and registry on the stack are considered “current”. Therefore, when the system is operating normally, the very definition of “current” is defined entirely by the behavior of a pyramid Router.

However, during unit testing, no Router code is ever invoked, and the definition of “current” is defined by the boundary between calls to the pyramid.config.Configurator.begin() and pyramid.config.Configurator.end() methods (or between calls to the pyramid.testing.setUp() and pyramid.testing.tearDown() functions). These functions push and pop the threadlocal stack when the system is under test. See Test Set Up and Tear Down for the definitions of these functions.

Scripts which use Pyramid machinery but never actually start a WSGI server or receive requests via HTTP such as scripts which use the pyramid.scripting API will never cause any Router code to be executed. However, the pyramid.scripting APIs also push some values on to the thread locals stack as a matter of course. Such scripts should expect the get_current_request() function to always return None, and should expect the get_current_registry() function to return exactly the same application registry for every request.

Why You Shouldn’t Abuse Thread Locals

You probably should almost never use the get_current_request() or get_current_registry() functions, except perhaps in tests. In particular, it’s almost always a mistake to use get_current_request or get_current_registry in application code because its usage makes it possible to write code that can be neither easily tested nor scripted. Inappropriate usage is defined as follows:

  • get_current_request should never be called within the body of a view callable, or within code called by a view callable. View callables already have access to the request (it’s passed in to each as request).
  • get_current_request should never be called in resource code. If a resource needs access to the request, it should be passed the request by a view callable.
  • get_current_request function should never be called because it’s “easier” or “more elegant” to think about calling it than to pass a request through a series of function calls when creating some API design. Your application should instead almost certainly pass data derived from the request around rather than relying on being able to call this function to obtain the request in places that actually have no business knowing about it. Parameters are meant to be passed around as function arguments, this is why they exist. Don’t try to “save typing” or create “nicer APIs” by using this function in the place where a request is required; this will only lead to sadness later.
  • Neither get_current_request nor get_current_registry should ever be called within application-specific forks of third-party library code. The library you’ve forked almost certainly has nothing to do with Pyramid, and making it dependent on Pyramid (rather than making your pyramid application depend upon it) means you’re forming a dependency in the wrong direction.

Use of the get_current_request() function in application code is still useful in very limited circumstances. As a rule of thumb, usage of get_current_request is useful within code which is meant to eventually be removed. For instance, you may find yourself wanting to deprecate some API that expects to be passed a request object in favor of one that does not expect to be passed a request object. But you need to keep implementations of the old API working for some period of time while you deprecate the older API. So you write a “facade” implementation of the new API which calls into the code which implements the older API. Since the new API does not require the request, your facade implementation doesn’t have local access to the request when it needs to pass it into the older API implementation. After some period of time, the older implementation code is disused and the hack that uses get_current_request is removed. This would be an appropriate place to use the get_current_request.

Use of the get_current_registry() function should be limited to testing scenarios. The registry made current by use of the pyramid.config.Configurator.begin() method during a test (or via pyramid.testing.setUp()) when you do not pass one in is available to you via this API.

Using the Zope Component Architecture in Pyramid

Under the hood, Pyramid uses a Zope Component Architecture component registry as its application registry. The Zope Component Architecture is referred to colloquially as the “ZCA.”

The zope.component API used to access data in a traditional Zope application can be opaque. For example, here is a typical “unnamed utility” lookup using the zope.component.getUtility() global API as it might appear in a traditional Zope application:

1
2
3
from pyramid.interfaces import ISettings
from zope.component import getUtility
settings = getUtility(ISettings)

After this code runs, settings will be a Python dictionary. But it’s unlikely that any “civilian” will be able to figure this out just by reading the code casually. When the zope.component.getUtility API is used by a developer, the conceptual load on a casual reader of code is high.

While the ZCA is an excellent tool with which to build a framework such as Pyramid, it is not always the best tool with which to build an application due to the opacity of the zope.component APIs. Accordingly, Pyramid tends to hide the presence of the ZCA from application developers. You needn’t understand the ZCA to create a Pyramid application; its use is effectively only a framework implementation detail.

However, developers who are already used to writing Zope applications often still wish to use the ZCA while building a Pyramid application; pyramid makes this possible.

Using the ZCA Global API in a Pyramid Application

Zope uses a single ZCA registry – the “global” ZCA registry – for all Zope applications that run in the same Python process, effectively making it impossible to run more than one Zope application in a single process.

However, for ease of deployment, it’s often useful to be able to run more than a single application per process. For example, use of a Paste “composite” allows you to run separate individual WSGI applications in the same process, each answering requests for some URL prefix. This makes it possible to run, for example, a TurboGears application at /turbogears and a Pyramid application at /pyramid, both served up using the same WSGI server within a single Python process.

Most production Zope applications are relatively large, making it impractical due to memory constraints to run more than one Zope application per Python process. However, a Pyramid application may be very small and consume very little memory, so it’s a reasonable goal to be able to run more than one Pyramid application per process.

In order to make it possible to run more than one Pyramid application in a single process, Pyramid defaults to using a separate ZCA registry per application.

While this services a reasonable goal, it causes some issues when trying to use patterns which you might use to build a typical Zope application to build a Pyramid application. Without special help, ZCA “global” APIs such as zope.component.getUtility and zope.component.getSiteManager will use the ZCA “global” registry. Therefore, these APIs will appear to fail when used in a Pyramid application, because they’ll be consulting the ZCA global registry rather than the component registry associated with your Pyramid application.

There are three ways to fix this: by disusing the ZCA global API entirely, by using pyramid.config.Configurator.hook_zca() or by passing the ZCA global registry to the Configurator constructor at startup time. We’ll describe all three methods in this section.

Disusing the Global ZCA API

ZCA “global” API functions such as zope.component.getSiteManager, zope.component.getUtility, zope.component.getAdapter, and zope.component.getMultiAdapter aren’t strictly necessary. Every component registry has a method API that offers the same functionality; it can be used instead. For example, presuming the registry value below is a Zope Component Architecture component registry, the following bit of code is equivalent to zope.component.getUtility(IFoo):

1
registry.getUtility(IFoo)

The full method API is documented in the zope.component package, but it largely mirrors the “global” API almost exactly.

If you are willing to disuse the “global” ZCA APIs and use the method interface of a registry instead, you need only know how to obtain the Pyramid component registry.

There are two ways of doing so:

  • use the pyramid.threadlocal.get_current_registry() function within Pyramid view or resource code. This will always return the “current” Pyramid application registry.
  • use the attribute of the request object named registry in your Pyramid view code, eg. request.registry. This is the ZCA component registry related to the running Pyramid application.

See Thread Locals for more information about pyramid.threadlocal.get_current_registry().

Enabling the ZCA Global API by Using hook_zca

Consider the following bit of idiomatic Pyramid startup code:

1
2
3
4
5
6
7
from zope.component import getGlobalSiteManager
from pyramid.config import Configurator

def app(global_settings, **settings):
    config = Configurator(settings=settings)
    config.include('some.other.package')
    return config.make_wsgi_app()

When the app function above is run, a Configurator is constructed. When the configurator is created, it creates a new application registry (a ZCA component registry). A new registry is constructed whenever the registry argument is omitted when a Configurator constructor is called, or when a registry argument with a value of None is passed to a Configurator constructor.

During a request, the application registry created by the Configurator is “made current”. This means calls to get_current_registry() in the thread handling the request will return the component registry associated with the application.

As a result, application developers can use get_current_registry to get the registry and thus get access to utilities and such, as per Disusing the Global ZCA API. But they still cannot use the global ZCA API. Without special treatment, the ZCA global APIs will always return the global ZCA registry (the one in zope.component.globalregistry.base).

To “fix” this and make the ZCA global APIs use the “current” Pyramid registry, you need to call hook_zca() within your setup code. For example:

1
2
3
4
5
6
7
8
from zope.component import getGlobalSiteManager
from pyramid.config import Configurator

def app(global_settings, **settings):
    config = Configurator(settings=settings)
    config.hook_zca()
    config.include('some.other.application')
    return config.make_wsgi_app()

We’ve added a line to our original startup code, line number 6, which calls config.hook_zca(). The effect of this line under the hood is that an analogue of the following code is executed:

1
2
3
from zope.component import getSiteManager
from pyramid.threadlocal import get_current_registry
getSiteManager.sethook(get_current_registry)

This causes the ZCA global API to start using the Pyramid application registry in threads which are running a Pyramid request.

Calling hook_zca is usually sufficient to “fix” the problem of being able to use the global ZCA API within a Pyramid application. However, it also means that a Zope application that is running in the same process may start using the Pyramid global registry instead of the Zope global registry, effectively inverting the original problem. In such a case, follow the steps in the next section, Enabling the ZCA Global API by Using The ZCA Global Registry.

Enabling the ZCA Global API by Using The ZCA Global Registry

You can tell your Pyramid application to use the ZCA global registry at startup time instead of constructing a new one:

1
2
3
4
5
6
7
8
9
from zope.component import getGlobalSiteManager
from pyramid.config import Configurator

def app(global_settings, **settings):
    globalreg = getGlobalSiteManager()
    config = Configurator(registry=globalreg)
    config.setup_registry(settings=settings)
    config.include('some.other.application')
    return config.make_wsgi_app()

Lines 5, 6, and 7 above are the interesting ones. Line 5 retrieves the global ZCA component registry. Line 6 creates a Configurator, passing the global ZCA registry into its constructor as the registry argument. Line 7 “sets up” the global registry with Pyramid-specific registrations; this is code that is normally executed when a registry is constructed rather than created, but we must call it “by hand” when we pass an explicit registry.

At this point, Pyramid will use the ZCA global registry rather than creating a new application-specific registry; since by default the ZCA global API will use this registry, things will work as you might expect a Zope app to when you use the global ZCA API.

Tutorials

Detailed tutorials explaining how to use Pyramid to build various types of applications and how to deploy Pyramid applications to various platforms.

SQLAlchemy + URL Dispatch Wiki Tutorial

This tutorial introduces a SQLAlchemy and url dispatch -based Pyramid application to a developer familiar with Python, and will be most familiar to developers who have used the Pylons 1.X web framework. When the tutorial is finished, the developer will have created a basic Wiki application with authentication.

For cut and paste purposes, the source code for all stages of this tutorial can be browsed at http://github.com/Pylons/pyramid/tree/1.2-branch/docs/tutorials/wiki2/src/.

Background

This tutorial presents a Pyramid application that uses technologies which will be familiar to someone with Pylons experience. It uses SQLAlchemy as a persistence mechanism and url dispatch to map URLs to code. It can also be followed by people without any prior Python web framework experience.

To code along with this tutorial, the developer will need a UNIX machine with development tools (Mac OS X with XCode, any Linux or BSD variant, etc) or a Windows system of any kind.

Have fun!

Installation

This tutorial assumes that Python and virtualenv are already installed and working in your system. If you need help setting this up, you should refer to the chapters on Installing Pyramid.

Preparation

Please take the following steps to prepare for the tutorial. The steps are slightly different depending on whether you’re using UNIX or Windows.

Preparation, UNIX

  1. Install SQLite3 and its development packages if you don’t already have them installed. Usually this is via your system’s package manager. For example, on a Debian Linux system, do sudo apt-get install libsqlite3-dev.

  2. Use your Python’s virtualenv to make a workspace:

    $ path/to/my/Python-2.6/bin/virtualenv --no-site-packages pyramidtut

  3. Switch to the pyramidtut directory:

    $ cd pyramidtut

  4. Use easy_install to get Pyramid and its direct dependencies installed:

    $ bin/easy_install pyramid

  5. Use easy_install to install various packages from PyPI.

    $ bin/easy_install docutils nose coverage zope.sqlalchemy \
          SQLAlchemy pyramid_tm
Preparation, Windows

  1. Use your Python’s virtualenv to make a workspace:

    c:\> c:\Python26\Scripts\virtualenv --no-site-packages pyramidtut

  2. Switch to the pyramidtut directory:

    c:\> cd pyramidtut

  3. Use easy_install to get Pyramid and its direct dependencies installed:

    c:\pyramidtut> Scripts\easy_install pyramid

  4. Use easy_install to install various packages from PyPI.

    c:\pyramidtut> Scripts\easy_install docutils \
         nose coverage zope.sqlalchemy SQLAlchemy pyramid_tm
Making a Project

Your next step is to create a project. Pyramid supplies a variety of scaffolds to generate sample projects. We will use the pyramid_routesalchemy scaffold, which generates an application that uses SQLAlchemy and URL dispatch.

The below instructions assume your current working directory is the “virtualenv” named “pyramidtut”.

On UNIX:

$ bin/paster create -t pyramid_routesalchemy tutorial

On Windows:

c:\pyramidtut> Scripts\paster create -t pyramid_routesalchemy tutorial

Note

If you are using Windows, the pyramid_routesalchemy scaffold may not deal gracefully with installation into a location that contains spaces in the path. If you experience startup problems, try putting both the virtualenv and the project into directories that do not contain spaces in their paths.

Installing the Project in “Development Mode”

In order to do development on the project easily, you must “register” the project as a development egg in your workspace using the setup.py develop command. In order to do so, cd to the “tutorial” directory you created in Making a Project, and run the “setup.py develop” command using virtualenv Python interpreter.

On UNIX:

$ cd tutorial
$ ../bin/python setup.py develop

On Windows:

c:\pyramidtut> cd tutorial
c:\pyramidtut\tutorial> ..\Scripts\python setup.py develop
Running the Tests

After you’ve installed the project in development mode, you may run the tests for the project.

On UNIX:

$ ../bin/python setup.py test -q

On Windows:

c:\pyramidtut\tutorial> ..\Scripts\python setup.py test -q
Starting the Application

Start the application.

On UNIX:

$ ../bin/paster serve development.ini --reload

On Windows:

c:\pyramidtut\tutorial> ..\Scripts\paster serve development.ini --reload
Exposing Test Coverage Information

You can run the nosetests command to see test coverage information. This runs the tests in the same way that setup.py test does but provides additional “coverage” information, exposing which lines of your project are “covered” (or not covered) by the tests.

To get this functionality working, we’ll need to install a couple of other packages into our virtualenv: nose and coverage:

On UNIX:

$ ../bin/easy_install nose coverage

On Windows:

c:\pyramidtut\tutorial> ..\Scripts\easy_install nose coverage

Once nose and coverage are installed, we can actually run the coverage tests.

On UNIX:

$ ../bin/nosetests --cover-package=tutorial --cover-erase --with-coverage

On Windows:

c:\pyramidtut\tutorial> ..\Scripts\nosetests --cover-package=tutorial ^
      --cover-erase --with-coverage

Looks like our package’s models module doesn’t quite have 100% test coverage.

Visit the Application in a Browser

In a browser, visit http://localhost:6543/. You will see the generated application’s default page.

One thing you’ll notice is the “debug toolbar” icon on right hand side of the page. You can read more about the purpose of the icon at The Debug Toolbar. It allows you to get information about your application while you develop.

Decisions the pyramid_routesalchemy Scaffold Has Made For You

Creating a project using the pyramid_routesalchemy scaffold makes the following assumptions:

  • you are willing to use SQLAlchemy as a database access tool
  • you are willing to use url dispatch to map URLs to code.

Note

Pyramid supports any persistent storage mechanism (e.g. object database or filesystem files, etc). It also supports an additional mechanism to map URLs to code (traversal). However, for the purposes of this tutorial, we’ll only be using url dispatch and SQLAlchemy.

Basic Layout

The starter files generated by the pyramid_routesalchemy scaffold are basic, but they provide a good orientation for the high-level patterns common to most url dispatch -based Pyramid projects.

The source code for this tutorial stage can be browsed at http://github.com/Pylons/pyramid/tree/1.2-branch/docs/tutorials/wiki2/src/basiclayout/.

App Startup with __init__.py

A directory on disk can be turned into a Python package by containing an __init__.py file. Even if empty, this marks a directory as a Python package. We use __init__.py both as a package marker and to contain configuration code.

The generated development.ini file is read by paster which looks for the application module in the use variable of the app:main section. The entry point is defined in the Setuptools configuration of this module, specifically in the setup.py file. For this tutorial, the entry point is defined as tutorial:main and points to a function named main.

First we need some imports to support later code:

1
2
3
4
5
from pyramid.config import Configurator
from sqlalchemy import engine_from_config

from tutorial.models import initialize_sql

Next we define the main function and create a SQLAlchemy database engine from the sqlalchemy. prefixed settings in the development.ini file’s [app:main] section. This will be a URI (something like sqlite://):

1
2
3
4
def main(global_config, **settings):
    """ This function returns a Pyramid WSGI application.
    """
    engine = engine_from_config(settings, 'sqlalchemy.')

We then initialize our SQL database using SQLAlchemy, passing it the engine:

    initialize_sql(engine)

The next step is to construct a Configurator:

    config = Configurator(settings=settings)

settings is passed to the Configurator as a keyword argument with the dictionary values passed by PasteDeploy as the **settings argument. This will be a dictionary of settings parsed from the .ini file, which contains deployment-related values such as pyramid.reload_templates, db_string, etc.

We now can call pyramid.config.Configurator.add_static_view() with the arguments static (the name), and tutorial:static (the path):

    config.add_static_view('static', 'tutorial:static', cache_max_age=3600)

This registers a static resource view which will match any URL that starts with /static/. This will serve up static resources for us from within the static directory of our tutorial package, in this case, via http://localhost:6543/static/ and below. With this declaration, we’re saying that any URL that starts with /static should go to the static view; any remainder of its path (e.g. the /foo in /static/foo) will be used to compose a path to a static file resource, such as a CSS file.

Using the configurator we can also register a route configuration via the pyramid.config.Configurator.add_route() method that will be used when the URL is /:

    config.add_route('home', '/')

Since this route has a pattern equalling / it is the route that will be matched when the URL / is visted, e.g. http://localhost:6543/.

Mapping the home route to code is done by registering a view. You will use pyramid.config.Configurator.add_view() in URL dispatch to register views for the routes, mapping your patterns to code:

    config.add_view('tutorial.views.my_view', route_name='home',
                    renderer='templates/mytemplate.pt')

The first positional add_view argument tutorial.views.my_view is the dotted name to a function we write (generated by the pyramid_routesalchemy scaffold) that is given a request object and which returns a response or a dictionary. This view also names a renderer, which is a template which lives in the templates subdirectory of the package. When the tutorial.views.my_view view returns a dictionary, a renderer will use this template to create a response.

Finally, we use the pyramid.config.Configurator.make_wsgi_app() method to return a WSGI application:

    return config.make_wsgi_app()

Our final __init__.py file will look like this:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
from pyramid.config import Configurator
from sqlalchemy import engine_from_config

from tutorial.models import initialize_sql

def main(global_config, **settings):
    """ This function returns a Pyramid WSGI application.
    """
    engine = engine_from_config(settings, 'sqlalchemy.')
    initialize_sql(engine)
    config = Configurator(settings=settings)
    config.add_static_view('static', 'tutorial:static', cache_max_age=3600)
    config.add_route('home', '/')
    config.add_view('tutorial.views.my_view', route_name='home',
                    renderer='templates/mytemplate.pt')
    return config.make_wsgi_app()
Content Models with models.py

In a SQLAlchemy-based application, a model object is an object composed by querying the SQL database which backs an application. SQLAlchemy is an “object relational mapper” (an ORM). The models.py file is where the pyramid_routesalchemy scaffold put the classes that implement our models.

Let’s take a look. First, we need some imports to support later code.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
import transaction

from sqlalchemy import Column
from sqlalchemy import Integer
from sqlalchemy import Unicode

from sqlalchemy.exc import IntegrityError
from sqlalchemy.ext.declarative import declarative_base

from sqlalchemy.orm import scoped_session
from sqlalchemy.orm import sessionmaker

from zope.sqlalchemy import ZopeTransactionExtension

Next we set up a SQLAlchemy “DBSession” object:

1
2
DBSession = scoped_session(sessionmaker(
                             extension=ZopeTransactionExtension()))

We also need to create a declarative Base object to use as a base class for our model:

Base = declarative_base()

To give a simple example of a model class, we define one named MyModel:

1
2
3
4
5
6
7
8
9
class MyModel(Base):
    __tablename__ = 'models'
    id = Column(Integer, primary_key=True)
    name = Column(Unicode(255), unique=True)
    value = Column(Integer)

    def __init__(self, name, value):
        self.name = name
        self.value = value

Our sample model has an __init__ that takes a two arguments (name, and value). It stores these values as self.name and self.value within the __init__ function itself. The MyModel class also has a __tablename__ attribute. This informs SQLAlchemy which table to use to store the data representing instances of this class.

Next we define a function named populate which adds a single model instance into our SQL storage and commits a transaction:

1
2
3
4
5
6
def populate():
    session = DBSession()
    model = MyModel(name=u'root',value=55)
    session.add(model)
    session.flush()
    transaction.commit()

The function doesn’t do a lot in this case, but it’s there to illustrate how an application requiring many objects to be set up could work.

Lastly we have a function named initialize_sql which receives a SQL database engine and binds it to our SQLAlchemy DBSession object. It also calls the populate function, to do initial database population. This is the initialization function that is called from __init__.py above.

1
2
3
4
5
6
7
8
def initialize_sql(engine):
    DBSession.configure(bind=engine)
    Base.metadata.bind = engine
    Base.metadata.create_all(engine)
    try:
        populate()
    except IntegrityError:
        transaction.abort()

Here is the complete source for models.py:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
import transaction

from sqlalchemy import Column
from sqlalchemy import Integer
from sqlalchemy import Unicode

from sqlalchemy.exc import IntegrityError
from sqlalchemy.ext.declarative import declarative_base

from sqlalchemy.orm import scoped_session
from sqlalchemy.orm import sessionmaker

from zope.sqlalchemy import ZopeTransactionExtension

DBSession = scoped_session(sessionmaker(
                             extension=ZopeTransactionExtension()))
Base = declarative_base()

class MyModel(Base):
    __tablename__ = 'models'
    id = Column(Integer, primary_key=True)
    name = Column(Unicode(255), unique=True)
    value = Column(Integer)

    def __init__(self, name, value):
        self.name = name
        self.value = value

def populate():
    session = DBSession()
    model = MyModel(name=u'root',value=55)
    session.add(model)
    session.flush()
    transaction.commit()
    
def initialize_sql(engine):
    DBSession.configure(bind=engine)
    Base.metadata.bind = engine
    Base.metadata.create_all(engine)
    try:
        populate()
    except IntegrityError:
        transaction.abort()

Defining the Domain Model

The first change we’ll make to our stock paster-generated application will be to define a domain model constructor representing a wiki page. We’ll do this inside our models.py file.

The source code for this tutorial stage can be browsed at http://github.com/Pylons/pyramid/tree/1.2-branch/docs/tutorials/wiki2/src/models/.

Making Edits to models.py

Note

There is nothing automagically special about the filename models.py. A project may have many models throughout its codebase in arbitrarily-named files. Files implementing models often have model in their filenames (or they may live in a Python subpackage of your application package named models) , but this is only by convention.

The first thing we want to do is remove the stock MyModel class from the generated models.py file. The MyModel class is only a sample and we’re not going to use it.

Next, we’ll remove the sqlalchemy.Unicode import and replace it with sqlalchemy.Text.

1
from sqlalchemy import Text

Then, we’ll add a Page class. Because this is a SQLAlchemy application, this class should inherit from an instance of sqlalchemy.ext.declarative.declarative_base. Declarative SQLAlchemy models are easier to use than directly-mapped ones.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
class Page(Base):
    """ The SQLAlchemy declarative model class for a Page object. """
    __tablename__ = 'pages'
    id = Column(Integer, primary_key=True)
    name = Column(Text, unique=True)
    data = Column(Text)

    def __init__(self, name, data):
        self.name = name
        self.data = data

As you can see, our Page class has a class level attribute __tablename__ which equals the string 'pages'. This means that SQLAlchemy will store our wiki data in a SQL table named pages. Our Page class will also have class-level attributes named id, name and data (all instances of sqlalchemy.Column). These will map to columns in the pages table. The id attribute will be the primary key in the table. The name attribute will be a text attribute, each value of which needs to be unique within the column. The data attribute is a text attribute that will hold the body of each page.

We’ll also remove our populate function. We’ll inline the populate step into initialize_sql, changing our initialize_sql function to add a FrontPage object to our database at startup time.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
def initialize_sql(engine):
    DBSession.configure(bind=engine)
    Base.metadata.bind = engine
    Base.metadata.create_all(engine)
    try:
        transaction.begin()
        session = DBSession()
        page = Page('FrontPage', 'This is the front page')
        session.add(page)
        transaction.commit()
    except IntegrityError:
        # already created
        transaction.abort()

Here, we’re using a slightly different binding syntax. It is otherwise largely the same as the initialize_sql in the paster-generated models.py.

Our DBSession assignment stays the same as the original generated models.py.

Looking at the Result of all Our Edits to models.py

The result of all of our edits to models.py will end up looking something like this:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
import transaction

from sqlalchemy import Column
from sqlalchemy import Integer
from sqlalchemy import Text

from sqlalchemy.exc import IntegrityError
from sqlalchemy.ext.declarative import declarative_base

from sqlalchemy.orm import scoped_session
from sqlalchemy.orm import sessionmaker

from zope.sqlalchemy import ZopeTransactionExtension

DBSession = scoped_session(sessionmaker(
                             extension=ZopeTransactionExtension()))
Base = declarative_base()

class Page(Base):
    """ The SQLAlchemy declarative model class for a Page object. """
    __tablename__ = 'pages'
    id = Column(Integer, primary_key=True)
    name = Column(Text, unique=True)
    data = Column(Text)

    def __init__(self, name, data):
        self.name = name
        self.data = data

def initialize_sql(engine):
    DBSession.configure(bind=engine)
    Base.metadata.bind = engine
    Base.metadata.create_all(engine)
    try:
        transaction.begin()
        session = DBSession()
        page = Page('FrontPage', 'This is the front page')
        session.add(page)
        transaction.commit()
    except IntegrityError:
        # already created
        transaction.abort()
Viewing the Application in a Browser

We can’t. At this point, our system is in a “non-runnable” state; we’ll need to change view-related files in the next chapter to be able to start the application successfully. If you try to start the application, you’ll wind up with a Python traceback on your console that ends with this exception:

ImportError: cannot import name MyModel

This will also happen if you attempt to run the tests.

Defining Views

A view callable in a url dispatch -based Pyramid application is typically a simple Python function that accepts a single parameter named request. A view callable is assumed to return a response object.

Note

A Pyramid view can also be defined as callable which accepts two arguments: a context and a request. You’ll see this two-argument pattern used in other Pyramid tutorials and applications. Either calling convention will work in any Pyramid application; the calling conventions can be used interchangeably as necessary. In url dispatch based applications, however, the context object is rarely used in the view body itself, so within this tutorial we define views as callables that accept only a request to avoid the visual “noise”. If you do need the context within a view function that only takes the request as a single argument, you can obtain it via request.context.

The request passed to every view that is called as the result of a route match has an attribute named matchdict that contains the elements placed into the URL by the pattern of a route statement. For instance, if a call to pyramid.config.Configurator.add_route() in __init__.py had the pattern {one}/{two}, and the URL at http://example.com/foo/bar was invoked, matching this pattern, the matchdict dictionary attached to the request passed to the view would have a 'one' key with the value 'foo' and a 'two' key with the value 'bar'.

The source code for this tutorial stage can be browsed at http://github.com/Pylons/pyramid/tree/1.2-branch/docs/tutorials/wiki2/src/views/.

Declaring Dependencies in Our setup.py File

The view code in our application will depend on a package which is not a dependency of the original “tutorial” application. The original “tutorial” application was generated by the paster create command; it doesn’t know about our custom application requirements. We need to add a dependency on the docutils package to our tutorial package’s setup.py file by assigning this dependency to the install_requires parameter in the setup function.

Our resulting setup.py should look like so:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
import os
import sys

from setuptools import setup, find_packages

here = os.path.abspath(os.path.dirname(__file__))
README = open(os.path.join(here, 'README.txt')).read()
CHANGES = open(os.path.join(here, 'CHANGES.txt')).read()

requires = [
    'pyramid',
    'SQLAlchemy',
    'transaction',
    'pyramid_tm',
    'pyramid_debugtoolbar',
    'zope.sqlalchemy',
    'docutils',
    ]

if sys.version_info[:3] < (2,5,0):
    requires.append('pysqlite')

setup(name='tutorial',
      version='0.0',
      description='tutorial',
      long_description=README + '\n\n' +  CHANGES,
      classifiers=[
        "Programming Language :: Python",
        "Framework :: Pylons",
        "Topic :: Internet :: WWW/HTTP",
        "Topic :: Internet :: WWW/HTTP :: WSGI :: Application",
        ],
      author='',
      author_email='',
      url='',
      keywords='web wsgi bfg pylons pyramid',
      packages=find_packages(),
      include_package_data=True,
      zip_safe=False,
      test_suite='tutorial',
      install_requires = requires,
      entry_points = """\
      [paste.app_factory]
      main = tutorial:main
      """,
      paster_plugins=['pyramid'],
      )

Note

After these new dependencies are added, you will need to rerun python setup.py develop inside the root of the tutorial package to obtain and register the newly added dependency package.

Adding View Functions

We’ll get rid of our my_view view function in our views.py file. It’s only an example and isn’t relevant to our application.

Then we’re going to add four view callable functions to our views.py module. One view callable (named view_wiki) will display the wiki itself (it will answer on the root URL), another named view_page will display an individual page, another named add_page will allow a page to be added, and a final view callable named edit_page will allow a page to be edited. We’ll describe each one briefly and show the resulting views.py file afterward.

Note

There is nothing special about the filename views.py. A project may have many view callables throughout its codebase in arbitrarily-named files. Files implementing view callables often have view in their filenames (or may live in a Python subpackage of your application package named views), but this is only by convention.

The view_wiki view function

The view_wiki function is the default view that will be called when a request is made to the root URL of our wiki. It always redirects to a URL which represents the path to our “FrontPage”.

1
2
3
def view_wiki(request):
    return HTTPFound(location = request.route_url('view_page',
                                                  pagename='FrontPage'))

The view_wiki function returns an instance of the pyramid.httpexceptions.HTTPFound class (instances of which implement the pyramid.interfaces.IResponse interface like pyramid.response.Response does), It will use the pyramid.request.Request.route_url() API to construct a URL to the FrontPage page (e.g. http://localhost:6543/FrontPage), and will use it as the “location” of the HTTPFound response, forming an HTTP redirect.

The view_page view function

The view_page function will be used to show a single page of our wiki. It renders the ReStructuredText body of a page (stored as the data attribute of a Page object) as HTML. Then it substitutes an HTML anchor for each WikiWord reference in the rendered HTML using a compiled regular expression.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
def view_page(request):
    pagename = request.matchdict['pagename']
    session = DBSession()
    page = session.query(Page).filter_by(name=pagename).first()
    if page is None:
        return HTTPNotFound('No such page')

    def check(match):
        word = match.group(1)
        exists = session.query(Page).filter_by(name=word).all()
        if exists:
            view_url = request.route_url('view_page', pagename=word)
            return '<a href="%s">%s</a>' % (view_url, word)
        else:
            add_url = request.route_url('add_page', pagename=word)
            return '<a href="%s">%s</a>' % (add_url, word)

    content = publish_parts(page.data, writer_name='html')['html_body']
    content = wikiwords.sub(check, content)
    edit_url = request.route_url('edit_page', pagename=pagename)
    return dict(page=page, content=content, edit_url=edit_url)

The curried function named check is used as the first argument to wikiwords.sub, indicating that it should be called to provide a value for each WikiWord match found in the content. If the wiki already contains a page with the matched WikiWord name, the check function generates a view link to be used as the substitution value and returns it. If the wiki does not already contain a page with with the matched WikiWord name, the function generates an “add” link as the substitution value and returns it.

As a result, the content variable is now a fully formed bit of HTML containing various view and add links for WikiWords based on the content of our current page object.

We then generate an edit URL (because it’s easier to do here than in the template), and we return a dictionary with a number of arguments. The fact that this view returns a dictionary (as opposed to a response object) is a cue to Pyramid that it should try to use a renderer associated with the view configuration to render a template. In our case, the template which will be rendered will be the templates/view.pt template, as per the configuration put into effect in __init__.py.

The add_page view function

The add_page function will be invoked when a user clicks on a WikiWord which isn’t yet represented as a page in the system. The check function within the view_page view generates URLs to this view. It also acts as a handler for the form that is generated when we want to add a page object. The matchdict attribute of the request passed to the add_page view will have the values we need to construct URLs and find model objects.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
def add_page(request):
    name = request.matchdict['pagename']
    if 'form.submitted' in request.params:
        session = DBSession()
        body = request.params['body']
        page = Page(name, body)
        session.add(page)
        return HTTPFound(location = request.route_url('view_page',
                                                      pagename=name))
    save_url = request.route_url('add_page', pagename=name)
    page = Page('', '')
    return dict(page=page, save_url=save_url)

The matchdict will have a 'pagename' key that matches the name of the page we’d like to add. If our add view is invoked via, e.g. http://localhost:6543/add_page/SomeName, the value for 'pagename' in the matchdict will be 'SomeName'.

If the view execution is not a result of a form submission (if the expression 'form.submitted' in request.params is False), the view callable renders a template. To do so, it generates a “save url” which the template uses as the form post URL during rendering. We’re lazy here, so we’re trying to use the same template (templates/edit.pt) for the add view as well as the page edit view, so we create a dummy Page object in order to satisfy the edit form’s desire to have some page object exposed as page, and Pyramid will render the template associated with this view to a response.

If the view execution is a result of a form submission (if the expression 'form.submitted' in request.params is True), we scrape the page body from the form data, create a Page object with this page body and the name taken from matchdict['pagename'], and save it into the database using session.add. We then redirect back to the view_page view for the newly created page.

The edit_page view function

The edit_page function will be invoked when a user clicks the “Edit this Page” button on the view form. It renders an edit form but it also acts as the handler for the form it renders. The matchdict attribute of the request passed to the edit_page view will have a 'pagename' key matching the name of the page the user wants to edit.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
def edit_page(request):
    name = request.matchdict['pagename']
    session = DBSession()
    page = session.query(Page).filter_by(name=name).one()
    if 'form.submitted' in request.params:
        page.data = request.params['body']
        session.add(page)
        return HTTPFound(location = request.route_url('view_page',
                                                      pagename=name))
    return dict(
        page=page,
        save_url = request.route_url('edit_page', pagename=name),
        )

If the view execution is not a result of a form submission (if the expression 'form.submitted' in request.params is False), the view simply renders the edit form, passing the page object and a save_url which will be used as the action of the generated form.

If the view execution is a result of a form submission (if the expression 'form.submitted' in request.params is True), the view grabs the body element of the request parameters and sets it as the data attribute of the page object. It then redirects to the view_page view of the wiki page.

Viewing the Result of all Our Edits to views.py

The result of all of our edits to views.py will leave it looking like this:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
import re

from docutils.core import publish_parts

from pyramid.httpexceptions import HTTPFound, HTTPNotFound

from tutorial.models import DBSession
from tutorial.models import Page

# regular expression used to find WikiWords
wikiwords = re.compile(r"\b([A-Z]\w+[A-Z]+\w+)")

def view_wiki(request):
    return HTTPFound(location = request.route_url('view_page',
                                                  pagename='FrontPage'))

def view_page(request):
    pagename = request.matchdict['pagename']
    session = DBSession()
    page = session.query(Page).filter_by(name=pagename).first()
    if page is None:
        return HTTPNotFound('No such page')

    def check(match):
        word = match.group(1)
        exists = session.query(Page).filter_by(name=word).all()
        if exists:
            view_url = request.route_url('view_page', pagename=word)
            return '<a href="%s">%s</a>' % (view_url, word)
        else:
            add_url = request.route_url('add_page', pagename=word)
            return '<a href="%s">%s</a>' % (add_url, word)

    content = publish_parts(page.data, writer_name='html')['html_body']
    content = wikiwords.sub(check, content)
    edit_url = request.route_url('edit_page', pagename=pagename)
    return dict(page=page, content=content, edit_url=edit_url)

def add_page(request):
    name = request.matchdict['pagename']
    if 'form.submitted' in request.params:
        session = DBSession()
        body = request.params['body']
        page = Page(name, body)
        session.add(page)
        return HTTPFound(location = request.route_url('view_page',
                                                      pagename=name))
    save_url = request.route_url('add_page', pagename=name)
    page = Page('', '')
    return dict(page=page, save_url=save_url)

def edit_page(request):
    name = request.matchdict['pagename']
    session = DBSession()
    page = session.query(Page).filter_by(name=name).one()
    if 'form.submitted' in request.params:
        page.data = request.params['body']
        session.add(page)
        return HTTPFound(location = request.route_url('view_page',
                                                      pagename=name))
    return dict(
        page=page,
        save_url = request.route_url('edit_page', pagename=name),
        )
Adding Templates

The views we’ve added all reference a template. Each template is a Chameleon ZPT template. These templates will live in the templates directory of our tutorial package.

The view.pt Template

The view.pt template is used for viewing a single wiki page. It is used by the view_page view function. It should have a div that is “structure replaced” with the content value provided by the view. It should also have a link on the rendered page that points at the “edit” URL (the URL which invokes the edit_page view for the page being viewed).

Once we’re done with the view.pt template, it will look a lot like the below:

<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN"
  "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en"
      xmlns:tal="http://xml.zope.org/namespaces/tal">
<head>
  <title>${page.name} - Pyramid tutorial wiki (based on
    TurboGears 20-Minute Wiki)</title>
  <meta http-equiv="Content-Type" content="text/html;charset=UTF-8"/>
  <meta name="keywords" content="python web application" />
  <meta name="description" content="pyramid web application" />
  <link rel="shortcut icon"
        href="${request.static_url('tutorial:static/favicon.ico')}" />
  <link rel="stylesheet"
        href="${request.static_url('tutorial:static/pylons.css')}"
        type="text/css" media="screen" charset="utf-8" />
  <!--[if lte IE 6]>
  <link rel="stylesheet"
        href="${request.static_url('tutorial:static/ie6.css')}"
        type="text/css" media="screen" charset="utf-8" />
  <![endif]-->
</head>
<body>
  <div id="wrap">
    <div id="top-small">
      <div class="top-small align-center">
        <div>
          <img width="220" height="50" alt="pyramid"
        src="${request.static_url('tutorial:static/pyramid-small.png')}" />
        </div>
      </div>
    </div>
    <div id="middle">
      <div class="middle align-right">
        <div id="left" class="app-welcome align-left">
          Viewing <b><span tal:replace="page.name">Page Name
              Goes Here</span></b><br/>
          You can return to the
          <a href="${request.application_url}">FrontPage</a>.<br/>
        </div>
        <div id="right" class="app-welcome align-right"></div>
      </div>
    </div>
    <div id="bottom">
      <div class="bottom">
        <div tal:replace="structure content">
          Page text goes here.
        </div>
        <p>
          <a tal:attributes="href edit_url" href="">
            Edit this page
          </a>
        </p>
      </div>
    </div>
  </div>
  <div id="footer">
    <div class="footer"
         >&copy; Copyright 2008-2011, Agendaless Consulting.</div>
  </div>
</body>
</html>

Note

The names available for our use in a template are always those that are present in the dictionary returned by the view callable. But our templates make use of a request object that none of our tutorial views return in their dictionary. This value appears as if “by magic”. However, request is one of several names that are available “by default” in a template when a template renderer is used. See *.pt or *.txt: Chameleon Template Renderers for more information about other names that are available by default in a template when a Chameleon template is used as a renderer.

The edit.pt Template

The edit.pt template is used for adding and editing a wiki page. It is used by the add_page and edit_page view functions. It should display a page containing a form that POSTs back to the “save_url” argument supplied by the view. The form should have a “body” textarea field (the page data), and a submit button that has the name “form.submitted”. The textarea in the form should be filled with any existing page data when it is rendered.

Once we’re done with the edit.pt template, it will look a lot like the below:

<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN"
  "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en"
      xmlns:tal="http://xml.zope.org/namespaces/tal">
<head>
  <title>${page.name} - Pyramid tutorial wiki (based on
    TurboGears 20-Minute Wiki)</title>
  <meta http-equiv="Content-Type" content="text/html;charset=UTF-8"/>
  <meta name="keywords" content="python web application" />
  <meta name="description" content="pyramid web application" />
  <link rel="shortcut icon"
        href="${request.static_url('tutorial:static/favicon.ico')}" />
  <link rel="stylesheet"
        href="${request.static_url('tutorial:static/pylons.css')}"
        type="text/css" media="screen" charset="utf-8" />
  <!--[if lte IE 6]>
  <link rel="stylesheet"
        href="${request.static_url('tutorial:static/ie6.css')}"
        type="text/css" media="screen" charset="utf-8" />
  <![endif]-->
</head>
<body>
  <div id="wrap">
    <div id="top-small">
      <div class="top-small align-center">
        <div>
          <img width="220" height="50" alt="pyramid"
        src="${request.static_url('tutorial:static/pyramid-small.png')}" />
        </div>
      </div>
    </div>
    <div id="middle">
      <div class="middle align-right">
        <div id="left" class="app-welcome align-left">
          Editing <b><span tal:replace="page.name">Page Name Goes
            Here</span></b><br/>
          You can return to the
          <a href="${request.application_url}">FrontPage</a>.<br/>
        </div>
        <div id="right" class="app-welcome align-right"></div>
      </div>
    </div>
    <div id="bottom">
      <div class="bottom">
        <form action="${save_url}" method="post">
          <textarea name="body" tal:content="page.data" rows="10"
                    cols="60"/><br/>
          <input type="submit" name="form.submitted" value="Save"/>
        </form>
      </div>
    </div>
  </div>
  <div id="footer">
    <div class="footer"
         >&copy; Copyright 2008-2011, Agendaless Consulting.</div>
  </div>
</body>
</html>
Static Assets

Our templates name a single static asset named pylons.css. We don’t need to create this file within our package’s static directory because it was provided at the time we created the project. This file is a little too long to replicate within the body of this guide, however it is available online.

This CSS file will be accessed via e.g. http://localhost:6543/static/pylons.css by virtue of the call to add_static_view directive we’ve made in the __init__.py file. Any number and type of static assets can be placed in this directory (or subdirectories) and are just referred to by URL or by using the convenience method static_url e.g. request.static_url('{{package}}:static/foo.css') within templates.

Mapping Views to URLs in __init__.py

The __init__.py file contains pyramid.config.Configurator.add_view() calls which serve to map routes via url dispatch to views. First, we’ll get rid of the existing route created by the template using the name 'home'. It’s only an example and isn’t relevant to our application.

We then need to add four calls to add_route. Note that the ordering of these declarations is very important. route declarations are matched in the order they’re found in the __init__.py file.

  1. Add a declaration which maps the pattern / (signifying the root URL) to the route named view_wiki.
  2. Add a declaration which maps the pattern /{pagename} to the route named view_page. This is the regular view for a page.
  3. Add a declaration which maps the pattern /add_page/{pagename} to the route named add_page. This is the add view for a new page.
  4. Add a declaration which maps the pattern /{pagename}/edit_page to the route named edit_page. This is the edit view for a page.

After we’ve defined the routes for our application, we can register views to handle the processing and rendering that needs to happen when each route is requested.

  1. Add a declaration which maps the view_wiki route to the view named view_wiki in our views.py file. This is the default view for the wiki.
  2. Add a declaration which maps the view_page route to the view named view_page in our views.py file.
  3. Add a declaration which maps the add_page route to the view named add_page in our views.py file.
  4. Add a declaration which maps the edit_page route to the view named edit_page in our views.py file.

As a result of our edits, the __init__.py file should look something like so:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
from pyramid.config import Configurator
from sqlalchemy import engine_from_config

from tutorial.models import initialize_sql

def main(global_config, **settings):
    """ This function returns a WSGI application.
    """
    engine = engine_from_config(settings, 'sqlalchemy.')
    initialize_sql(engine)
    config = Configurator(settings=settings)
    config.add_static_view('static', 'tutorial:static', cache_max_age=3600)
    config.add_route('view_wiki', '/')
    config.add_route('view_page', '/{pagename}')
    config.add_route('add_page', '/add_page/{pagename}')
    config.add_route('edit_page', '/{pagename}/edit_page')
    config.add_view('tutorial.views.view_wiki', route_name='view_wiki')
    config.add_view('tutorial.views.view_page', route_name='view_page',
                    renderer='tutorial:templates/view.pt')
    config.add_view('tutorial.views.add_page', route_name='add_page',
                    renderer='tutorial:templates/edit.pt')
    config.add_view('tutorial.views.edit_page', route_name='edit_page',
                    renderer='tutorial:templates/edit.pt')
    return config.make_wsgi_app()
Viewing the Application in a Browser

We can finally examine our application in a browser. The views we’ll try are as follows:

  • Visiting http://localhost:6543 in a browser invokes the view_wiki view. This always redirects to the view_page view of the FrontPage page object.
  • Visiting http://localhost:6543/FrontPage in a browser invokes the view_page view of the front page page object.
  • Visiting http://localhost:6543/FrontPage/edit_page in a browser invokes the edit view for the front page object.
  • Visiting http://localhost:6543/add_page/SomePageName in a browser invokes the add view for a page.

Try generating an error within the body of a view by adding code to the top of it that generates an exception (e.g. raise Exception('Forced Exception')). Then visit the error-raising view in a browser. You should see an interactive exception handler in the browser which allows you to examine values in a post-mortem mode.

Adding Authorization

Our application currently allows anyone with access to the server to view, edit, and add pages to our wiki. For purposes of demonstration we’ll change our application to allow only people whom possess a specific username (editor) to add and edit wiki pages but we’ll continue allowing anyone with access to the server to view pages. Pyramid provides facilities for authorization and authentication. We’ll make use of both features to provide security to our application.

We will add an authentication policy and an authorization policy to our application registry, add a security.py module, create a root factory with an ACL, and add permission declarations to the edit_page and add_page views.

Then we will add login and logout views, and modify the existing views to make them return a logged_in flag to the renderer.

Finally, we will add a login.pt template and change the existing view.pt and edit.pt to show a “Logout” link when not logged in.

The source code for this tutorial stage can be browsed at http://github.com/Pylons/pyramid/tree/1.2-branch/docs/tutorials/wiki2/src/authorization/.

Changing __init__.py For Authorization

We’re going to be making several changes to our __init__.py file which will help us configure an authorization policy.

Adding A Root Factory

We’re going to start to use a custom root factory within our __init__.py file. The objects generated by the root factory will be used as the context of each request to our application. We do this to allow Pyramid declarative security to work properly. The context object generated by the root factory during a request will be decorated with security declarations. When we begin to use a custom root factory to generate our contexts, we can begin to make use of the declarative security features of Pyramid.

We’ll modify our __init__.py, passing in a root factory to our Configurator constructor. We’ll point it at a new class we create inside our models.py file. Add the following statements to your models.py file:

1
2
3
4
5
6
7
8
from pyramid.security import Allow
from pyramid.security import Everyone

class RootFactory(object):
    __acl__ = [ (Allow, Everyone, 'view'),
                (Allow, 'group:editors', 'edit') ]
    def __init__(self, request):
        pass

The RootFactory class we’ve just added will be used by Pyramid to construct a context object. The context is attached to the request object passed to our view callables as the context attribute.

The context object generated by our root factory will possess an __acl__ attribute that allows pyramid.security.Everyone (a special principal) to view all pages, while allowing only a principal named group:editors to edit and add pages. The __acl__ attribute attached to a context is interpreted specially by Pyramid as an access control list during view callable execution. See Assigning ACLs to your Resource Objects for more information about what an ACL represents.

We’ll pass the RootFactory we created in the step above in as the root_factory argument to a Configurator.

Configuring an Authorization Policy

For any Pyramid application to perform authorization, we need to add a security.py module (we’ll do that shortly) and we’ll need to change our __init__.py file to add an authentication policy and an authorization policy which uses the security.py file for a callback.

We’ll change our __init__.py file to enable an AuthTktAuthenticationPolicy and an ACLAuthorizationPolicy to enable declarative security checking. We need to import the new policies:

1
2
3
from pyramid.authentication import AuthTktAuthenticationPolicy
from pyramid.authorization import ACLAuthorizationPolicy
from tutorial.security import groupfinder

Then, we’ll add those policies to the configuration:

1
2
3
4
5
6
7
    authn_policy = AuthTktAuthenticationPolicy(
        'sosecret', callback=groupfinder)
    authz_policy = ACLAuthorizationPolicy()
    config = Configurator(settings=settings,
                          root_factory='tutorial.models.RootFactory',
                          authentication_policy=authn_policy,
                          authorization_policy=authz_policy)

Note that that the pyramid.authentication.AuthTktAuthenticationPolicy constructor accepts two arguments: secret and callback. secret is a string representing an encryption key used by the “authentication ticket” machinery represented by this policy: it is required. The callback is a groupfinder function in the current directory’s security.py file. We haven’t added that module yet, but we’re about to.

We’ll also change __init__.py, adding a call to pyramid.config.Configurator.add_view() that points at our login view callable. This is also known as a forbidden view:

1
2
3
4
    config.add_route('login', '/login')
    config.add_view('tutorial.login.login',
                    context='pyramid.httpexceptions.HTTPForbidden',
                    renderer='tutorial:templates/login.pt')

A forbidden view configures our newly created login view to show up when Pyramid detects that a view invocation can not be authorized.

A logout view callable will allow users to log out later:

1
2
    config.add_route('logout', '/logout')
    config.add_view('tutorial.login.logout', route_name='logout')

We’ll also add permission arguments with the value edit to the edit_page and add_page views. This indicates that the view callables which these views reference cannot be invoked without the authenticated user possessing the edit permission with respect to the current context.

1
2
3
4
    config.add_view('tutorial.views.add_page', route_name='add_page',
                    renderer='tutorial:templates/edit.pt', permission='edit')
    config.add_view('tutorial.views.edit_page', route_name='edit_page',
                    renderer='tutorial:templates/edit.pt', permission='edit')

Adding these permission arguments causes Pyramid to make the assertion that only users who possess the effective edit permission at the time of the request may invoke those two views. We’ve granted the group:editors principal the edit permission at the root model via its ACL, so only the a user whom is a member of the group named group:editors will able to invoke the views associated with the add_page or edit_page routes.

Viewing Your Changes

When we’re done configuring a root factory, adding an authorization policy, and adding views, your application’s __init__.py will look like this:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
from pyramid.config import Configurator
from pyramid.authentication import AuthTktAuthenticationPolicy
from pyramid.authorization import ACLAuthorizationPolicy

from sqlalchemy import engine_from_config

from tutorial.models import initialize_sql
from tutorial.security import groupfinder

def main(global_config, **settings):
    """ This function returns a WSGI application.
    """
    engine = engine_from_config(settings, 'sqlalchemy.')
    initialize_sql(engine)
    authn_policy = AuthTktAuthenticationPolicy(
        'sosecret', callback=groupfinder)
    authz_policy = ACLAuthorizationPolicy()
    config = Configurator(settings=settings,
                          root_factory='tutorial.models.RootFactory',
                          authentication_policy=authn_policy,
                          authorization_policy=authz_policy)
    config.add_static_view('static', 'tutorial:static', cache_max_age=3600)

    config.add_route('view_wiki', '/')
    config.add_route('login', '/login')
    config.add_route('logout', '/logout')
    config.add_route('view_page', '/{pagename}')
    config.add_route('add_page', '/add_page/{pagename}')
    config.add_route('edit_page', '/{pagename}/edit_page')

    config.add_view('tutorial.views.view_wiki', route_name='view_wiki')
    config.add_view('tutorial.login.login', route_name='login', 
                    renderer='tutorial:templates/login.pt')
    config.add_view('tutorial.login.logout', route_name='logout')
    config.add_view('tutorial.views.view_page', route_name='view_page',
                    renderer='tutorial:templates/view.pt')
    config.add_view('tutorial.views.add_page', route_name='add_page',
                    renderer='tutorial:templates/edit.pt', permission='edit')
    config.add_view('tutorial.views.edit_page', route_name='edit_page',
                    renderer='tutorial:templates/edit.pt', permission='edit')
    config.add_view('tutorial.login.login',
                    context='pyramid.httpexceptions.HTTPForbidden',
                    renderer='tutorial:templates/login.pt')
    return config.make_wsgi_app()
Adding security.py

Add a security.py module within your package (in the same directory as __init__.py, views.py, etc.) with the following content:

1
2
3
4
5
6
7
8
USERS = {'editor':'editor',
          'viewer':'viewer'}
GROUPS = {'editor':['group:editors']}

def groupfinder(userid, request):
    if userid in USERS:
        return GROUPS.get(userid, [])

The groupfinder function defined here is an authentication policy “callback”; it is a callable that accepts a userid and a request. If the userid exists in the system, the callback will return a sequence of group identifiers (or an empty sequence if the user isn’t a member of any groups). If the userid does not exist in the system, the callback will return None. In a production system, user and group data will most often come from a database, but here we use “dummy” data to represent user and groups sources. Note that the editor user is a member of the group:editors group in our dummy group data (the GROUPS data structure).

We’ve given the editor user membership to the group:editors by mapping him to this group in the GROUPS data structure (GROUPS = {'editor':['group:editors']}). Since the groupfinder function consults the GROUPS data structure, this will mean that, as a result of the ACL attached to the root returned by the root factory, and the permission associated with the add_page and edit_page views, the editor user should be able to add and edit pages.

Adding Login and Logout Views

We’ll add a login view callable which renders a login form and processes the post from the login form, checking credentials.

We’ll also add a logout view callable to our application and provide a link to it. This view will clear the credentials of the logged in user and redirect back to the front page.

We’ll add a different file (for presentation convenience) to add login and the logout view callables. Add a file named login.py to your application (in the same directory as views.py) with the following content:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
from pyramid.httpexceptions import HTTPFound
from pyramid.security import remember
from pyramid.security import forget

from tutorial.security import USERS

def login(request):
    login_url = request.route_url('login')
    referrer = request.url
    if referrer == login_url:
        referrer = '/' # never use the login form itself as came_from
    came_from = request.params.get('came_from', referrer)
    message = ''
    login = ''
    password = ''
    if 'form.submitted' in request.params:
        login = request.params['login']
        password = request.params['password']
        if USERS.get(login) == password:
            headers = remember(request, login)
            return HTTPFound(location = came_from,
                             headers = headers)
        message = 'Failed login'

    return dict(
        message = message,
        url = request.application_url + '/login',
        came_from = came_from,
        login = login,
        password = password,
        )
    
def logout(request):
    headers = forget(request)
    return HTTPFound(location = request.route_url('view_wiki'),
                     headers = headers)
Changing Existing Views

Then we need to change each of our view_page, edit_page and add_page views in views.py to pass a “logged in” parameter to its template. We’ll add something like this to each view body:

1
2
from pyramid.security import authenticated_userid
logged_in = authenticated_userid(request)

We’ll then change the return value of these views to pass the resulting `logged_in` value to the template, e.g.:

1
2
3
4
return dict(page = page,
            content = content,
            logged_in = logged_in,
            edit_url = edit_url)
Adding the login.pt Template

Add a login.pt template to your templates directory. It’s referred to within the login view we just added to login.py.

<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN"
  "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en"
      xmlns:tal="http://xml.zope.org/namespaces/tal">
<head>
  <title>Login - Pyramid tutorial wiki (based on TurboGears
    20-Minute Wiki)</title>
  <meta http-equiv="Content-Type" content="text/html;charset=UTF-8"/>
  <meta name="keywords" content="python web application" />
  <meta name="description" content="pyramid web application" />
  <link rel="shortcut icon"
        href="${request.static_url('tutorial:static/favicon.ico')}" />
  <link rel="stylesheet"
        href="${request.static_url('tutorial:static/pylons.css')}"
        type="text/css" media="screen" charset="utf-8" />
  <!--[if lte IE 6]>
  <link rel="stylesheet"
        href="${request.static_url('tutorial:static/ie6.css')}"
        type="text/css" media="screen" charset="utf-8" />
  <![endif]-->
</head>
<body>
  <div id="wrap">
    <div id="top-small">
      <div class="top-small align-center">
        <div>
          <img width="220" height="50" alt="pyramid"
        src="${request.static_url('tutorial:static/pyramid-small.png')}" />
        </div>
      </div>
    </div>
    <div id="middle">
      <div class="middle align-right">
        <div id="left" class="app-welcome align-left">
          <b>Login</b><br/>
          <span tal:replace="message"/>
        </div>
        <div id="right" class="app-welcome align-right"></div>
      </div>
    </div>
    <div id="bottom">
      <div class="bottom">
        <form action="${url}" method="post">
          <input type="hidden" name="came_from" value="${came_from}"/>
          <input type="text" name="login" value="${login}"/><br/>
          <input type="password" name="password"
                 value="${password}"/><br/>
          <input type="submit" name="form.submitted" value="Log In"/>
        </form>
      </div>
    </div>
  </div>
  <div id="footer">
    <div class="footer"
         >&copy; Copyright 2008-2011, Agendaless Consulting.</div>
  </div>
</body>
</html>
Change view.pt and edit.pt

We’ll also need to change our edit.pt and view.pt templates to display a “Logout” link if someone is logged in. This link will invoke the logout view.

To do so we’ll add this to both templates within the <div id="right" class="app-welcome align-right"> div:

<span tal:condition="logged_in">
   <a href="${request.application_url}/logout">Logout</a>
</span>
Seeing Our Changes To views.py and our Templates

Our views.py module will look something like this when we’re done:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
import re

from docutils.core import publish_parts

from pyramid.httpexceptions import HTTPFound, HTTPNotFound
from pyramid.security import authenticated_userid

from tutorial.models import DBSession
from tutorial.models import Page

# regular expression used to find WikiWords
wikiwords = re.compile(r"\b([A-Z]\w+[A-Z]+\w+)")

def view_wiki(request):
    return HTTPFound(location = request.route_url('view_page',
                                                  pagename='FrontPage'))

def view_page(request):
    pagename = request.matchdict['pagename']
    session = DBSession()
    page = session.query(Page).filter_by(name=pagename).first()
    if page is None:
        return HTTPNotFound('No such page')

    def check(match):
        word = match.group(1)
        exists = session.query(Page).filter_by(name=word).all()
        if exists:
            view_url = request.route_url('view_page', pagename=word)
            return '<a href="%s">%s</a>' % (view_url, word)
        else:
            add_url = request.route_url('add_page', pagename=word)
            return '<a href="%s">%s</a>' % (add_url, word)

    content = publish_parts(page.data, writer_name='html')['html_body']
    content = wikiwords.sub(check, content)
    edit_url = request.route_url('edit_page', pagename=pagename)
    logged_in = authenticated_userid(request)
    return dict(page=page, content=content, edit_url=edit_url,
                logged_in=logged_in)

def add_page(request):
    name = request.matchdict['pagename']
    if 'form.submitted' in request.params:
        session = DBSession()
        body = request.params['body']
        page = Page(name, body)
        session.add(page)
        return HTTPFound(location = request.route_url('view_page',
                                                      pagename=name))
    save_url = request.route_url('add_page', pagename=name)
    page = Page('', '')
    logged_in = authenticated_userid(request)
    return dict(page=page, save_url=save_url, logged_in=logged_in)

def edit_page(request):
    name = request.matchdict['pagename']
    session = DBSession()
    page = session.query(Page).filter_by(name=name).one()
    if 'form.submitted' in request.params:
        page.data = request.params['body']
        session.add(page)
        return HTTPFound(location = request.route_url('view_page',
                                                      pagename=name))

    logged_in = authenticated_userid(request)
    return dict(
        page=page,
        save_url = request.route_url('edit_page', pagename=name),
        logged_in = logged_in,
        )

Our edit.pt template will look something like this when we’re done:

<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN"
  "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en"
      xmlns:tal="http://xml.zope.org/namespaces/tal">
<head>
  <title>${page.name} - Pyramid tutorial wiki (based on
    TurboGears 20-Minute Wiki)</title>
  <meta http-equiv="Content-Type" content="text/html;charset=UTF-8"/>
  <meta name="keywords" content="python web application" />
  <meta name="description" content="pyramid web application" />
  <link rel="shortcut icon"
        href="${request.static_url('tutorial:static/favicon.ico')}" />
  <link rel="stylesheet"
        href="${request.static_url('tutorial:static/pylons.css')}"
        type="text/css" media="screen" charset="utf-8" />
  <!--[if lte IE 6]>
  <link rel="stylesheet"
        href="${request.static_url('tutorial:static/ie6.css')}"
        type="text/css" media="screen" charset="utf-8" />
  <![endif]-->
</head>
<body>
  <div id="wrap">
    <div id="top-small">
      <div class="top-small align-center">
        <div>
          <img width="220" height="50" alt="pyramid"
        src="${request.static_url('tutorial:static/pyramid-small.png')}" />
        </div>
      </div>
    </div>
    <div id="middle">
      <div class="middle align-right">
        <div id="left" class="app-welcome align-left">
          Editing <b><span tal:replace="page.name">Page Name
            Goes Here</span></b><br/>
          You can return to the
          <a href="${request.application_url}">FrontPage</a>.<br/>
        </div>
        <div id="right" class="app-welcome align-right">
          <span tal:condition="logged_in">
              <a href="${request.application_url}/logout">Logout</a>
          </span>
        </div>
      </div>
    </div>
    <div id="bottom">
      <div class="bottom">
        <form action="${save_url}" method="post">
          <textarea name="body" tal:content="page.data" rows="10"
                    cols="60"/><br/>
          <input type="submit" name="form.submitted" value="Save"/>
        </form>
      </div>
    </div>
  </div>
  <div id="footer">
    <div class="footer"
         >&copy; Copyright 2008-2011, Agendaless Consulting.</div>
  </div>
</body>
</html>

Our view.pt template will look something like this when we’re done:

<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN"
  "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en"
      xmlns:tal="http://xml.zope.org/namespaces/tal">
<head>
  <title>${page.name} - Pyramid tutorial wiki (based on
    TurboGears 20-Minute Wiki)</title>
  <meta http-equiv="Content-Type" content="text/html;charset=UTF-8"/>
  <meta name="keywords" content="python web application" />
  <meta name="description" content="pyramid web application" />
  <link rel="shortcut icon"
        href="${request.static_url('tutorial:static/favicon.ico')}" />
  <link rel="stylesheet"
        href="${request.static_url('tutorial:static/pylons.css')}"
        type="text/css" media="screen" charset="utf-8" />
  <!--[if lte IE 6]>
  <link rel="stylesheet"
        href="${request.static_url('tutorial:static/ie6.css')}"
        type="text/css" media="screen" charset="utf-8" />
  <![endif]-->
</head>
<body>
  <div id="wrap">
    <div id="top-small">
      <div class="top-small align-center">
        <div>
          <img width="220" height="50" alt="pyramid"
        src="${request.static_url('tutorial:static/pyramid-small.png')}" />
        </div>
      </div>
    </div>
    <div id="middle">
      <div class="middle align-right">
        <div id="left" class="app-welcome align-left">
          Viewing <b><span tal:replace="page.name">Page Name
            Goes Here</span></b><br/>
          You can return to the
          <a href="${request.application_url}">FrontPage</a>.<br/>
        </div>
        <div id="right" class="app-welcome align-right">
          <span tal:condition="logged_in">
            <a href="${request.application_url}/logout">Logout</a>
          </span>
        </div>
      </div>
    </div>
    <div id="bottom">
      <div class="bottom">
        <div tal:replace="structure content">
          Page text goes here.
        </div>
        <p>
          <a tal:attributes="href edit_url" href="">
            Edit this page
          </a>
        </p>
      </div>
    </div>
  </div>
  <div id="footer">
    <div class="footer"
         >&copy; Copyright 2008-2011, Agendaless Consulting.</div>
  </div>
</body>
</html>
Viewing the Application in a Browser

We can finally examine our application in a browser. The views we’ll try are as follows:

  • Visiting http://localhost:6543/ in a browser invokes the view_wiki view. This always redirects to the view_page view of the FrontPage page object. It is executable by any user.
  • Visiting http://localhost:6543/FrontPage in a browser invokes the view_page view of the FrontPage page object.
  • Visiting http://localhost:6543/FrontPage/edit_page in a browser invokes the edit view for the FrontPage object. It is executable by only the editor user. If a different user (or the anonymous user) invokes it, a login form will be displayed. Supplying the credentials with the username editor, password editor will display the edit page form.
  • Visiting http://localhost:6543/add_page/SomePageName in a browser invokes the add view for a page. It is executable by only the editor user. If a different user (or the anonymous user) invokes it, a login form will be displayed. Supplying the credentials with the username editor, password editor will display the edit page form.
  • After logging in (as a result of hitting an edit or add page and submitting the login form with the editor credentials), we’ll see a Logout link in the upper right hand corner. When we click it, we’re logged out, and redirected back to the front page.

Adding Tests

We will now add tests for the models and the views and a few functional tests in the tests.py. Tests ensure that an application works, and that it continues to work after some changes are made in the future.

Testing the Models

We write a test class for the model class Page and another test class for the initialize_sql function.

To do so, we’ll retain the tutorial.tests.ViewTests class provided as a result of the pyramid_routesalchemy project generator. We’ll add two test classes: one for the Page model named PageModelTests, one for the initialize_sql function named InitializeSqlTests.

Testing the Views

We’ll modify our tests.py file, adding tests for each view function we added above. As a result, we’ll delete the ViewTests test in the file, and add four other test classes: ViewWikiTests, ViewPageTests, AddPageTests, and EditPageTests. These test the view_wiki, view_page, add_page, and edit_page views respectively.

Functional tests

We test the whole application, covering security aspects that are not tested in the unit tests, like logging in, logging out, checking that the viewer user cannot add or edit pages, but the editor user can, and so on.

Viewing the results of all our edits to tests.py

Once we’re done with the tests.py module, it will look a lot like the below:

  1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
import unittest

from pyramid import testing


def _initTestingDB():
    from tutorial.models import DBSession
    from tutorial.models import Base
    from sqlalchemy import create_engine
    engine = create_engine('sqlite:///:memory:')
    DBSession.configure(bind=engine)
    Base.metadata.bind = engine
    Base.metadata.create_all(engine)
    return DBSession

def _registerRoutes(config):
    config.add_route('view_page', '{pagename}')
    config.add_route('edit_page', '{pagename}/edit_page')
    config.add_route('add_page', 'add_page/{pagename}')


class PageModelTests(unittest.TestCase):

    def setUp(self):
        self.session = _initTestingDB()

    def tearDown(self):
        self.session.remove()

    def _getTargetClass(self):
        from tutorial.models import Page
        return Page

    def _makeOne(self, name='SomeName', data='some data'):
        return self._getTargetClass()(name, data)

    def test_constructor(self):
        instance = self._makeOne()
        self.assertEqual(instance.name, 'SomeName')
        self.assertEqual(instance.data, 'some data')

class InitializeSqlTests(unittest.TestCase):

    def setUp(self):
        from tutorial.models import DBSession
        DBSession.remove()

    def tearDown(self):
        from tutorial.models import DBSession
        DBSession.remove()

    def _callFUT(self, engine):
        from tutorial.models import initialize_sql
        return initialize_sql(engine)

    def test_it(self):
        from sqlalchemy import create_engine
        engine = create_engine('sqlite:///:memory:')
        self._callFUT(engine)
        from tutorial.models import DBSession, Page
        self.assertEqual(DBSession.query(Page).one().data,
            'This is the front page')

class ViewWikiTests(unittest.TestCase):
    def setUp(self):
        self.config = testing.setUp()

    def tearDown(self):
        testing.tearDown()

    def _callFUT(self, request):
        from tutorial.views import view_wiki
        return view_wiki(request)

    def test_it(self):
        _registerRoutes(self.config)
        request = testing.DummyRequest()
        response = self._callFUT(request)
        self.assertEqual(response.location, 'http://example.com/FrontPage')

class ViewPageTests(unittest.TestCase):
    def setUp(self):
        self.session = _initTestingDB()
        self.config = testing.setUp()

    def tearDown(self):
        self.session.remove()
        testing.tearDown()

    def _callFUT(self, request):
        from tutorial.views import view_page
        return view_page(request)

    def test_it(self):
        from tutorial.models import Page
        request = testing.DummyRequest()
        request.matchdict['pagename'] = 'IDoExist'
        page = Page('IDoExist', 'Hello CruelWorld IDoExist')
        self.session.add(page)
        _registerRoutes(self.config)
        info = self._callFUT(request)
        self.assertEqual(info['page'], page)
        self.assertEqual(
            info['content'],
            '<div class="document">\n'
            '<p>Hello <a href="http://example.com/add_page/CruelWorld">'
            'CruelWorld</a> '
            '<a href="http://example.com/IDoExist">'
            'IDoExist</a>'
            '</p>\n</div>\n')
        self.assertEqual(info['edit_url'],
            'http://example.com/IDoExist/edit_page')

class AddPageTests(unittest.TestCase):
    def setUp(self):
        self.session = _initTestingDB()
        self.config = testing.setUp()

    def tearDown(self):
        self.session.remove()
        testing.tearDown()

    def _callFUT(self, request):
        from tutorial.views import add_page
        return add_page(request)

    def test_it_notsubmitted(self):
        _registerRoutes(self.config)
        request = testing.DummyRequest()
        request.matchdict = {'pagename':'AnotherPage'}
        info = self._callFUT(request)
        self.assertEqual(info['page'].data,'')
        self.assertEqual(info['save_url'],
                         'http://example.com/add_page/AnotherPage')

    def test_it_submitted(self):
        from tutorial.models import Page
        _registerRoutes(self.config)
        request = testing.DummyRequest({'form.submitted':True,
                                        'body':'Hello yo!'})
        request.matchdict = {'pagename':'AnotherPage'}
        self._callFUT(request)
        page = self.session.query(Page).filter_by(name='AnotherPage').one()
        self.assertEqual(page.data, 'Hello yo!')

class EditPageTests(unittest.TestCase):
    def setUp(self):
        self.session = _initTestingDB()
        self.config = testing.setUp()

    def tearDown(self):
        self.session.remove()
        testing.tearDown()

    def _callFUT(self, request):
        from tutorial.views import edit_page
        return edit_page(request)

    def test_it_notsubmitted(self):
        from tutorial.models import Page
        _registerRoutes(self.config)
        request = testing.DummyRequest()
        request.matchdict = {'pagename':'abc'}
        page = Page('abc', 'hello')
        self.session.add(page)
        info = self._callFUT(request)
        self.assertEqual(info['page'], page)
        self.assertEqual(info['save_url'],
            'http://example.com/abc/edit_page')

    def test_it_submitted(self):
        from tutorial.models import Page
        _registerRoutes(self.config)
        request = testing.DummyRequest({'form.submitted':True,
            'body':'Hello yo!'})
        request.matchdict = {'pagename':'abc'}
        page = Page('abc', 'hello')
        self.session.add(page)
        response = self._callFUT(request)
        self.assertEqual(response.location, 'http://example.com/abc')
        self.assertEqual(page.data, 'Hello yo!')

class FunctionalTests(unittest.TestCase):

    viewer_login = '/login?login=viewer&password=viewer' \
                   '&came_from=FrontPage&form.submitted=Login'
    viewer_wrong_login = '/login?login=viewer&password=incorrect' \
                   '&came_from=FrontPage&form.submitted=Login'
    editor_login = '/login?login=editor&password=editor' \
                   '&came_from=FrontPage&form.submitted=Login'

    def setUp(self):
        from tutorial import main
        settings = { 'sqlalchemy.url': 'sqlite:///:memory:'}
        app = main({}, **settings)
        from webtest import TestApp
        self.testapp = TestApp(app)

    def tearDown(self):
        del self.testapp
        from tutorial.models import DBSession
        DBSession.remove()

    def test_root(self):
        res = self.testapp.get('/', status=302)
        self.assertEqual(res.location, 'http://localhost/FrontPage')

    def test_FrontPage(self):
        res = self.testapp.get('/FrontPage', status=200)
        self.assertTrue('FrontPage' in res.body)

    def test_unexisting_page(self):
        self.testapp.get('/SomePage', status=404)

    def test_successful_log_in(self):
        res = self.testapp.get(self.viewer_login, status=302)
        self.assertEqual(res.location, 'http://localhost/FrontPage')

    def test_failed_log_in(self):
        res = self.testapp.get(self.viewer_wrong_login, status=200)
        self.assertTrue('login' in res.body)

    def test_logout_link_present_when_logged_in(self):
        self.testapp.get(self.viewer_login, status=302)
        res = self.testapp.get('/FrontPage', status=200)
        self.assertTrue('Logout' in res.body)

    def test_logout_link_not_present_after_logged_out(self):
        self.testapp.get(self.viewer_login, status=302)
        self.testapp.get('/FrontPage', status=200)
        res = self.testapp.get('/logout', status=302)
        self.assertTrue('Logout' not in res.body)

    def test_anonymous_user_cannot_edit(self):
        res = self.testapp.get('/FrontPage/edit_page', status=200)
        self.assertTrue('Login' in res.body)

    def test_anonymous_user_cannot_add(self):
        res = self.testapp.get('/add_page/NewPage', status=200)
        self.assertTrue('Login' in res.body)

    def test_viewer_user_cannot_edit(self):
        self.testapp.get(self.viewer_login, status=302)
        res = self.testapp.get('/FrontPage/edit_page', status=200)
        self.assertTrue('Login' in res.body)

    def test_viewer_user_cannot_add(self):
        self.testapp.get(self.viewer_login, status=302)
        res = self.testapp.get('/add_page/NewPage', status=200)
        self.assertTrue('Login' in res.body)

    def test_editors_member_user_can_edit(self):
        self.testapp.get(self.editor_login, status=302)
        res = self.testapp.get('/FrontPage/edit_page', status=200)
        self.assertTrue('Editing' in res.body)

    def test_editors_member_user_can_add(self):
        self.testapp.get(self.editor_login, status=302)
        res = self.testapp.get('/add_page/NewPage', status=200)
        self.assertTrue('Editing' in res.body)

    def test_editors_member_user_can_view(self):
        self.testapp.get(self.editor_login, status=302)
        res = self.testapp.get('/FrontPage', status=200)
        self.assertTrue('FrontPage' in res.body)
Running the Tests

We can run these tests by using setup.py test in the same way we did in Running the Tests. However, first we must edit our setup.py to include a dependency on WebTest, which we’ve used in our tests.py. Change the requires list in setup.py to include WebTest.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
requires = [
    'pyramid',
    'SQLAlchemy',
    'transaction',
    'pyramid_tm',
    'pyramid_debugtoolbar',
    'zope.sqlalchemy',
    'docutils',
    'WebTest', # add this
    ]

After we’ve added a dependency on WebTest in setup.py, we need to rerun setup.py develop to get WebTest installed into our virtualenv. Assuming our shell’s current working directory is the “tutorial” distribution directory:

On UNIX:

$ ../bin/python setup.py develop

On Windows:

c:\pyramidtut\tutorial> ..\Scripts\python setup.py develop

Once that command has completed successfully, we can run the tests themselves:

On UNIX:

$ ../bin/python setup.py test -q

On Windows:

c:\pyramidtut\tutorial> ..\Scripts\python setup.py test -q

The expected result looks something like:

......................
----------------------------------------------------------------------
Ran 22 tests in 2.700s

OK

Distributing Your Application

Once your application works properly, you can create a “tarball” from it by using the setup.py sdist command. The following commands assume your current working directory is the tutorial package we’ve created and that the parent directory of the tutorial package is a virtualenv representing a Pyramid environment.

On UNIX:

$ ../bin/python setup.py sdist

On Windows:

c:\pyramidtut> ..\Scripts\python setup.py sdist

The output of such a command will be something like:

running sdist
# ... more output ...
creating dist
tar -cf dist/tutorial-0.1.tar tutorial-0.1
gzip -f9 dist/tutorial-0.1.tar
removing 'tutorial-0.1' (and everything under it)

Note that this command creates a tarball in the “dist” subdirectory named tutorial-0.1.tar.gz. You can send this file to your friends to show them your cool new application. They should be able to install it by pointing the easy_install command directly at it. Or you can upload it to PyPI and share it with the rest of the world, where it can be downloaded via easy_install remotely like any other package people download from PyPI.

ZODB + Traversal Wiki Tutorial

This tutorial introduces a traversal -based Pyramid application to a developer familiar with Python. It will be most familiar to developers with previous Zope experience. When we’re done with the tutorial, the developer will have created a basic Wiki application with authentication.

For cut and paste purposes, the source code for all stages of this tutorial can be browsed at http://github.com/Pylons/pyramid/tree/1.2-branch/docs/tutorials/wiki/src/.

Background

This version of the Pyramid wiki tutorial presents a Pyramid application that uses technologies which will be familiar to someone with Zope experience. It uses ZODB as a persistence mechanism and traversal to map URLs to code. It can also be followed by people without any prior Python web framework experience.

To code along with this tutorial, the developer will need a UNIX machine with development tools (Mac OS X with XCode, any Linux or BSD variant, etc) or a Windows system of any kind.

Have fun!

Installation

For the most part, the installation process for this tutorial duplicates the steps described in Installing Pyramid and Creating a Pyramid Project, however it also explains how to install additional libraries for tutorial purposes.

Preparation

Please take the following steps to prepare for the tutorial. The steps to prepare for the tutorial are slightly different depending on whether you’re using UNIX or Windows.

Preparation, UNIX

  1. If you don’t already have a Python 2.6 interpreter installed on your system, obtain, install, or find Python 2.6 for your system.

  2. Make sure the Python development headers are installed on your system. If you’ve installed Python from source, these will already be installed. If you’re using a system Python, you may have to install a python-dev package (e.g. apt-get python-dev). The headers are not required for Pyramid itself, just for dependencies of the tutorial.

  3. Install the latest setuptools into the Python you obtained/installed/found in the step above: download ez_setup.py and run it using the python interpreter of your Python 2.6 installation:

    $ /path/to/my/Python-2.6/bin/python ez_setup.py

  4. Use that Python’s bin/easy_install to install virtualenv:

    $ /path/to/my/Python-2.6/bin/easy_install virtualenv

  5. Use that Python’s virtualenv to make a workspace:

    $ path/to/my/Python-2.6/bin/virtualenv --no-site-packages \
          pyramidtut

  6. Switch to the pyramidtut directory:

    $ cd pyramidtut

  7. (Optional) Consider using source bin/activate to make your shell environment wired to use the virtualenv.

  8. Use easy_install to get Pyramid and its direct dependencies installed:

    $ bin/easy_install pyramid

  9. Use easy_install to install docutils, pyramid_tm, pyramid_zodbconn, pyramid_debugtoolbar, nose and coverage:

    $ bin/easy_install docutils pyramid_tm pyramid_zodbconn \
          pyramid_debugtoolbar nose coverage
Preparation, Windows

  1. Install, or find Python 2.6 for your system.

  2. Install the latest setuptools into the Python you obtained/installed/found in the step above: download ez_setup.py and run it using the python interpreter of your Python 2.6 installation using a command prompt:

    c:\> c:\Python26\python ez_setup.py

  3. Use that Python’s bin/easy_install to install virtualenv:

    c:\> c:\Python26\Scripts\easy_install virtualenv

  4. Use that Python’s virtualenv to make a workspace:

    c:\> c:\Python26\Scripts\virtualenv --no-site-packages pyramidtut

  5. Switch to the pyramidtut directory:

    c:\> cd pyramidtut

  6. (Optional) Consider using bin\activate.bat to make your shell environment wired to use the virtualenv.

  7. Use easy_install to get Pyramid and its direct dependencies installed:

    c:\pyramidtut> Scripts\easy_install pyramid

  8. Use easy_install to install docutils, pyramid_tm, pyramid_zodbconn, pyramid_debugtoolbar, nose and coverage:

    c:\pyramidtut> Scripts\easy_install docutils pyramid_tm \
      pyramid_zodbconn pyramid_debugtoolbar nose coverage
Making a Project

Your next step is to create a project. Pyramid supplies a variety of scaffolds to generate sample projects. For this tutorial, we will use the ZODB -oriented scaffold named pyramid_zodb.

The below instructions assume your current working directory is the “virtualenv” named “pyramidtut”.

On UNIX:

$ bin/paster create -t pyramid_zodb tutorial

On Windows:

c:\pyramidtut> Scripts\paster create -t pyramid_zodb tutorial

Note

If you are using Windows, the pyramid_zodb Paster scaffold doesn’t currently deal gracefully with installation into a location that contains spaces in the path. If you experience startup problems, try putting both the virtualenv and the project into directories that do not contain spaces in their paths.

Installing the Project in “Development Mode”

In order to do development on the project easily, you must “register” the project as a development egg in your workspace using the setup.py develop command. In order to do so, cd to the “tutorial” directory you created in Making a Project, and run the “setup.py develop” command using virtualenv Python interpreter.

On UNIX:

$ cd tutorial
$ ../bin/python setup.py develop

On Windows:

C:\pyramidtut> cd tutorial
C:\pyramidtut\tutorial> ..\Scripts\python setup.py develop
Running the Tests

After you’ve installed the project in development mode, you may run the tests for the project.

On UNIX:

$ ../bin/python setup.py test -q

On Windows:

c:\pyramidtut\tutorial> ..\Scripts\python setup.py test -q
Starting the Application

Start the application.

On UNIX:

$ ../bin/paster serve development.ini --reload

On Windows:

c:\pyramidtut\tutorial> ..\Scripts\paster serve development.ini --reload
Exposing Test Coverage Information

You can run the nosetests command to see test coverage information. This runs the tests in the same way that setup.py test does but provides additional “coverage” information, exposing which lines of your project are “covered” (or not covered) by the tests.

On UNIX:

$ ../bin/nosetests --cover-package=tutorial --cover-erase --with-coverage

On Windows:

c:\pyramidtut\tutorial> ..\Scripts\nosetests --cover-package=tutorial ^
     --cover-erase --with-coverage

Looks like the code in the pyramid_zodb scaffold for ZODB projects is missing some test coverage, particularly in the file named models.py.

Visit the Application in a Browser

In a browser, visit http://localhost:6543/. You will see the generated application’s default page.

One thing you’ll notice is the “debug toolbar” icon on right hand side of the page. You can read more about the purpose of the icon at The Debug Toolbar. It allows you to get information about your application while you develop.

Decisions the pyramid_zodb Scaffold Has Made For You

Creating a project using the pyramid_zodb scaffold makes the following assumptions:

  • you are willing to use ZODB as persistent storage
  • you are willing to use traversal to map URLs to code.

Note

Pyramid supports any persistent storage mechanism (e.g. a SQL database or filesystem files, etc). Pyramid also supports an additional mechanism to map URLs to code (URL dispatch). However, for the purposes of this tutorial, we’ll only be using traversal and ZODB.

Basic Layout

The starter files generated by the pyramid_zodb scaffold are basic, but they provide a good orientation for the high-level patterns common to most traversal -based Pyramid (and ZODB based) projects.

The source code for this tutorial stage can be browsed via http://github.com/Pylons/pyramid/tree/1.2-branch/docs/tutorials/wiki/src/basiclayout/.

App Startup with __init__.py

A directory on disk can be turned into a Python package by containing an __init__.py file. Even if empty, this marks a directory as a Python package. Our application uses __init__.py as both a package marker, as well as to contain application configuration code.

When you run the application using the paster command using the development.ini generated config file, the application configuration points at a Setuptools entry point described as egg:tutorial. In our application, because the application’s setup.py file says so, this entry point happens to be the main function within the file named __init__.py:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
from pyramid.config import Configurator
from pyramid_zodbconn import get_connection
from tutorial.models import appmaker

def root_factory(request):
    conn = get_connection(request)
    return appmaker(conn.root())

def main(global_config, **settings):
    """ This function returns a Pyramid WSGI application.
    """
    config = Configurator(root_factory=root_factory, settings=settings)
    config.add_static_view('static', 'tutorial:static', cache_max_age=3600)
    config.scan('tutorial')
    return config.make_wsgi_app()
  1. Lines 1-3. Perform some dependency imports.
  2. Lines 5-7 Define a root factory for our Pyramid application.
  3. Line 12. We construct a Configurator with a root factory and the settings keywords parsed by PasteDeploy. The root factory is named get_root.
  4. Line 13. Register a ‘static view’ which answers requests which start with with URL path /static using the pyramid.config.Configurator.add_static_view method(). This statement registers a view that will serve up static assets, such as CSS and image files, for us, in this case, at http://localhost:6543/static/ and below. The first argument is the “name” static, which indicates that the URL path prefix of the view will be /static. the The second argument of this tag is the “path”, which is an asset specification, so it finds the resources it should serve within the static directory inside the tutorial package.
  5. Line 14. Perform a scan. A scan will find configuration decoration, such as view configuration decorators (e.g. @view_config) in the source code of the tutorial package and will take actions based on these decorators. The argument to scan() is the package name to scan, which is tutorial.
  6. Line 15. Use the pyramid.config.Configurator.make_wsgi_app() method to return a WSGI application.
Resources and Models with models.py

Pyramid uses the word resource to describe objects arranged hierarchically in a resource tree. This tree is consulted by traversal to map URLs to code. In this application, the resource tree represents the site structure, but it also represents the domain model of the application, because each resource is a node stored persistently in a ZODB database. The models.py file is where the pyramid_zodb scaffold put the classes that implement our resource objects, each of which happens also to be a domain model object.

Here is the source for models.py:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
from persistent.mapping import PersistentMapping

class MyModel(PersistentMapping):
    __parent__ = __name__ = None

def appmaker(zodb_root):
    if not 'app_root' in zodb_root:
        app_root = MyModel()
        zodb_root['app_root'] = app_root
        import transaction
        transaction.commit()
    return zodb_root['app_root']

  1. Lines 3-4. The MyModel resource class is implemented here. Instances of this class will be capable of being persisted in ZODB because the class inherits from the persistent.mapping.PersistentMapping class. The __parent__ and __name__ are important parts of the traversal protocol. By default, have these as None indicating that this is the root object.

  2. Lines 6-12. appmaker is used to return the application root object. It is called on every request to the Pyramid application. It also performs bootstrapping by creating an application root (inside the ZODB root object) if one does not already exist. It is used by the “root_factory” we’ve defined in our __init__.py.

    We do so by first seeing if the database has the persistent application root. If not, we make an instance, store it, and commit the transaction. We then return the application root object.

Views With views.py

Our scaffold generated a default views.py on our behalf. It contains a single view, which is used to render the page shown when you visit the URL http://localhost:6543/.

Here is the source for views.py:

1
2
3
4
5
6
7
from pyramid.view import view_config
from tutorial.models import MyModel

@view_config(context=MyModel,
             renderer='tutorial:templates/mytemplate.pt')
def my_view(request):
    return {'project':'tutorial'}

Let’s try to understand the components in this module:

  1. Lines 1-2. Perform some dependency imports.

  2. Line 4. Use the pyramid.view.view_config() configuration decoration to perform a view configuration registration. This view configuration registration will be activated when the application is started. It will be activated by virtue of it being found as the result of a scan (when Line 17 of __init__.py is run).

    The @view_config decorator accepts a number of keyword arguments. We use two keyword arguments here: context and renderer.

    The context argument signifies that the decorated view callable should only be run when traversal finds the tutorial.models.MyModel resource to be the context of a request. In English, this means that when the URL / is visited, because MyModel is the root model, this view callable will be invoked.

    The renderer argument names an asset specification of tutorial:templates/mytemplate.pt. This asset specification points at a Chameleon template which lives in the mytemplate.pt file within the templates directory of the tutorial package. And indeed if you look in the templates directory of this package, you’ll see a mytemplate.pt template file, which renders the default home page of the generated project.

    Since this call to @view_config doesn’t pass a name argument, the my_view function which it decorates represents the “default” view callable used when the context is of the type MyModel.

  3. Lines 5-6. We define a view callable named my_view, which we decorated in the step above. This view callable is a function we write generated by the pyramid_zodb scaffold that is given a request and which returns a dictionary. The mytemplate.pt renderer named by the asset specification in the step above will convert this dictionary to a response on our behalf.

    The function returns the dictionary {'project':'tutorial'}. This dictionary is used by the template named by the mytemplate.pt asset specification to fill in certain values on the page.

Configuration in development.ini

The development.ini (in the tutorial project directory, as opposed to the tutorial package directory) looks like this:

[app:main]
use = egg:tutorial
pyramid.reload_templates = true
pyramid.debug_authorization = false
pyramid.debug_notfound = false
pyramid.debug_routematch = false
pyramid.debug_templates = true
pyramid.default_locale_name = en
pyramid.includes = pyramid_debugtoolbar
                   pyramid_zodbconn
                   pyramid_tm
tm.attempts = 3
zodbconn.uri = file://%(here)s/Data.fs?connection_cache_size=20000

[server:main]
use = egg:Paste#http
host = 0.0.0.0
port = 6543

# Begin logging configuration

[loggers]
keys = root

[handlers]
keys = console

[formatters]
keys = generic

[logger_root]
level = INFO
handlers = console

[handler_console]
class = StreamHandler
args = (sys.stderr,)
level = NOTSET
formatter = generic

[formatter_generic]
format = %(asctime)s %(levelname)-5.5s [%(name)s] %(message)s

# End logging configuration

Note the existence of an [app:main] section which specifies our WSGI application. Our ZODB database settings are specified as the zodbconn.uri setting within this section. This value, and the other values within this section are passed as **settings to the main function we defined in __init__.py when the server is started via paster serve.

Defining the Domain Model

The first change we’ll make to our stock paster-generated application will be to define two resource constructors, one representing a wiki page, and another representing the wiki as a mapping of wiki page names to page objects. We’ll do this inside our models.py file.

Because we’re using ZODB to represent our resource tree, each of these resource constructors represents a domain model object, so we’ll call these constructors “model constructors”. Both our Page and Wiki constructors will be class objects. A single instance of the “Wiki” class will serve as a container for “Page” objects, which will be instances of the “Page” class.

The source code for this tutorial stage can be browsed via http://github.com/Pylons/pyramid/tree/1.2-branch/docs/tutorials/wiki/src/models/.

Deleting the Database

In the next step, we’re going to remove the MyModel Python model class from our models.py file. Since this class is referred to within our persistent storage (represented on disk as a file named Data.fs), we’ll have strange things happen the next time we want to visit the application in a browser. Remove the Data.fs from the tutorial directory before proceeding any further. It’s always fine to do this as long as you don’t care about the content of the database; the database itself will be recreated as necessary.

Making Edits to models.py

Note

There is nothing automagically special about the filename models.py. A project may have many models throughout its codebase in arbitrarily-named files. Files implementing models often have model in their filenames, or they may live in a Python subpackage of your application package named models, but this is only by convention.

The first thing we want to do is remove the MyModel class from the generated models.py file. The MyModel class is only a sample and we’re not going to use it.

Then, we’ll add a Wiki class. We want it to inherit from the persistent.mapping.PersistentMapping class because it provides mapping behavior, and it makes sure that our Wiki page is stored as a “first-class” persistent object in our ZODB database.

Our Wiki class should have two attributes set to None at class scope: __parent__ and __name__. If a model has a __parent__ attribute of None in a traversal-based Pyramid application, it means that it’s the root model. The __name__ of the root model is also always None.

Then we’ll add a Page class. This class should inherit from the persistent.Persistent class. We’ll also give it an __init__ method that accepts a single parameter named data. This parameter will contain the ReStructuredText body representing the wiki page content. Note that Page objects don’t have an initial __name__ or __parent__ attribute. All objects in a traversal graph must have a __name__ and a __parent__ attribute. We don’t specify these here because both __name__ and __parent__ will be set by by a view function when a Page is added to our Wiki mapping.

As a last step, we want to change the appmaker function in our models.py file so that the root resource of our application is a Wiki instance. We’ll also slot a single page object (the front page) into the Wiki within the appmaker. This will provide traversal a resource tree to work against when it attempts to resolve URLs to resources.

Looking at the Result of Our Edits to models.py

The result of all of our edits to models.py will end up looking something like this:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
from persistent import Persistent
from persistent.mapping import PersistentMapping

class Wiki(PersistentMapping):
    __name__ = None
    __parent__ = None

class Page(Persistent):
    def __init__(self, data):
        self.data = data

def appmaker(zodb_root):
    if not 'app_root' in zodb_root:
        app_root = Wiki()
        frontpage = Page('This is the front page')
        app_root['FrontPage'] = frontpage
        frontpage.__name__ = 'FrontPage'
        frontpage.__parent__ = app_root
        zodb_root['app_root'] = app_root
        import transaction
        transaction.commit()
    return zodb_root['app_root']
Viewing the Application in a Browser

We can’t. At this point, our system is in a “non-runnable” state; we’ll need to change view-related files in the next chapter to be able to start the application successfully. If you try to start the application, you’ll wind up with a Python traceback on your console that ends with this exception:

ImportError: cannot import name MyModel

This will also happen if you attempt to run the tests.

Defining Views

A view callable in a traversal -based Pyramid application is typically a simple Python function that accepts two parameters: context and request. A view callable is assumed to return a response object.

Note

A Pyramid view can also be defined as callable which accepts only a request argument. You’ll see this one-argument pattern used in other Pyramid tutorials and applications. Either calling convention will work in any Pyramid application; the calling conventions can be used interchangeably as necessary. In traversal based applications, URLs are mapped to a context resource, and since our resource tree also represents our application’s “domain model”, we’re often interested in the context, because it represents the persistent storage of our application. For this reason, in this tutorial we define views as callables that accept context in the callable argument list. If you do need the context within a view function that only takes the request as a single argument, you can obtain it via request.context.

We’re going to define several view callable functions, then wire them into Pyramid using some view configuration.

The source code for this tutorial stage can be browsed via http://github.com/Pylons/pyramid/tree/1.2-branch/docs/tutorials/wiki/src/views/.

Declaring Dependencies in Our setup.py File

The view code in our application will depend on a package which is not a dependency of the original “tutorial” application. The original “tutorial” application was generated by the paster create command; it doesn’t know about our custom application requirements. We need to add a dependency on the docutils package to our tutorial package’s setup.py file by assigning this dependency to the install_requires parameter in the setup function.

Our resulting setup.py should look like so:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
import os

from setuptools import setup, find_packages

here = os.path.abspath(os.path.dirname(__file__))
README = open(os.path.join(here, 'README.txt')).read()
CHANGES = open(os.path.join(here, 'CHANGES.txt')).read()

requires = [
    'pyramid',
    'pyramid_zodbconn',
    'pyramid_tm',
    'pyramid_debugtoolbar',
    'ZODB3',
    'docutils',
    ]

setup(name='tutorial',
      version='0.0',
      description='tutorial',
      long_description=README + '\n\n' +  CHANGES,
      classifiers=[
        "Intended Audience :: Developers",
        "Framework :: Pylons",
        "Programming Language :: Python",
        "Topic :: Internet :: WWW/HTTP",
        "Topic :: Internet :: WWW/HTTP :: WSGI :: Application",
        ],
      author='',
      author_email='',
      url='',
      keywords='web pylons pyramid',
      packages=find_packages(),
      include_package_data=True,
      zip_safe=False,
      install_requires=requires,
      tests_require=requires,
      test_suite="tutorial",
      entry_points = """\
      [paste.app_factory]
      main = tutorial:main
      """,
      paster_plugins=['pyramid'],
      )

Note

After these new dependencies are added, you will need to rerun python setup.py develop inside the root of the tutorial package to obtain and register the newly added dependency package.

Adding View Functions

We’re going to add four view callable functions to our views.py module. One view named view_wiki will display the wiki itself (it will answer on the root URL), another named view_page will display an individual page, another named add_page will allow a page to be added, and a final view named edit_page will allow a page to be edited.

Note

There is nothing special about the filename views.py. A project may have many view callables throughout its codebase in arbitrarily-named files. Files implementing view callables often have view in their filenames (or may live in a Python subpackage of your application package named views), but this is only by convention.

The view_wiki view function

The view_wiki function will be configured to respond as the default view callable for a Wiki resource. We’ll provide it with a @view_config decorator which names the class tutorial.models.Wiki as its context. This means that when a Wiki resource is the context, and no view name exists in the request, this view will be used. The view configuration associated with view_wiki does not use a renderer because the view callable always returns a response object rather than a dictionary. No renderer is necessary when a view returns a response object.

The view_wiki view callable always redirects to the URL of a Page resource named “FrontPage”. To do so, it returns an instance of the pyramid.httpexceptions.HTTPFound class (instances of which implement the pyramid.interfaces.IResponse interface like pyramid.response.Response does). The pyramid.request.Request.resource_url() API. pyramid.request.Request.resource_url() constructs a URL to the FrontPage page resource (e.g. http://localhost:6543/FrontPage), and uses it as the “location” of the HTTPFound response, forming an HTTP redirect.

The view_page view function

The view_page function will be configured to respond as the default view of a Page resource. We’ll provide it with a @view_config decorator which names the class tutorial.models.Page as its context. This means that when a Page resource is the context, and no view name exists in the request, this view will be used. We inform Pyramid this view will use the templates/view.pt template file as a renderer.

The view_page function generates the ReStructuredText body of a page (stored as the data attribute of the context passed to the view; the context will be a Page resource) as HTML. Then it substitutes an HTML anchor for each WikiWord reference in the rendered HTML using a compiled regular expression.

The curried function named check is used as the first argument to wikiwords.sub, indicating that it should be called to provide a value for each WikiWord match found in the content. If the wiki (our page’s __parent__) already contains a page with the matched WikiWord name, the check function generates a view link to be used as the substitution value and returns it. If the wiki does not already contain a page with with the matched WikiWord name, the function generates an “add” link as the substitution value and returns it.

As a result, the content variable is now a fully formed bit of HTML containing various view and add links for WikiWords based on the content of our current page resource.

We then generate an edit URL (because it’s easier to do here than in the template), and we wrap up a number of arguments in a dictionary and return it.

The arguments we wrap into a dictionary include page, content, and edit_url. As a result, the template associated with this view callable (via renderer= in its configuration) will be able to use these names to perform various rendering tasks. The template associated with this view callable will be a template which lives in templates/view.pt.

Note the contrast between this view callable and the view_wiki view callable. In the view_wiki view callable, we unconditionally return a response object. In the view_page view callable, we return a dictionary. It is always fine to return a response object from a Pyramid view. Returning a dictionary is allowed only when there is a renderer associated with the view callable in the view configuration.

The add_page view function

The add_page function will be configured to respond when the context resource is a Wiki and the view name is add_page. We’ll provide it with a @view_config decorator which names the string add_page as its view name (via name=), the class tutorial.models.Wiki as its context, and the renderer named templates/edit.pt. This means that when a Wiki resource is the context, and a view name named add_page exists as the result of traversal, this view will be used. We inform Pyramid this view will use the templates/edit.pt template file as a renderer. We share the same template between add and edit views, thus edit.pt instead of add.pt.

The add_page function will be invoked when a user clicks on a WikiWord which isn’t yet represented as a page in the system. The check function within the view_page view generates URLs to this view. It also acts as a handler for the form that is generated when we want to add a page resource. The context of the add_page view is always a Wiki resource (not a Page resource).

The request subpath in Pyramid is the sequence of names that are found after the view name in the URL segments given in the PATH_INFO of the WSGI request as the result of traversal. If our add view is invoked via, e.g. http://localhost:6543/add_page/SomeName, the subpath will be a tuple: ('SomeName',).

The add view takes the zeroth element of the subpath (the wiki page name), and aliases it to the name attribute in order to know the name of the page we’re trying to add.

If the view rendering is not a result of a form submission (if the expression 'form.submitted' in request.params is False), the view renders a template. To do so, it generates a “save url” which the template use as the form post URL during rendering. We’re lazy here, so we’re trying to use the same template (templates/edit.pt) for the add view as well as the page edit view. To do so, we create a dummy Page resource object in order to satisfy the edit form’s desire to have some page object exposed as page, and we’ll render the template to a response.

If the view rendering is a result of a form submission (if the expression 'form.submitted' in request.params is True), we scrape the page body from the form data, create a Page object using the name in the subpath and the page body, and save it into “our context” (the Wiki) using the __setitem__ method of the context. We then redirect back to the view_page view (the default view for a page) for the newly created page.

The edit_page view function

The edit_page function will be configured to respond when the context is a Page resource and the view name is edit_page. We’ll provide it with a @view_config decorator which names the string edit_page as its view name (via name=), the class tutorial.models.Page as its context, and the renderer named templates/edit.pt. This means that when a Page resource is the context, and a view name exists as the result of traversal named edit_page, this view will be used. We inform Pyramid this view will use the templates/edit.pt template file as a renderer.

The edit_page function will be invoked when a user clicks the “Edit this Page” button on the view form. It renders an edit form but it also acts as the form post view callable for the form it renders. The context of the edit_page view will always be a Page resource (never a Wiki resource).

If the view execution is not a result of a form submission (if the expression 'form.submitted' in request.params is False), the view simply renders the edit form, passing the page resource, and a save_url which will be used as the action of the generated form.

If the view execution is a result of a form submission (if the expression 'form.submitted' in request.params is True), the view grabs the body element of the request parameter and sets it as the data attribute of the page context. It then redirects to the default view of the context (the page), which will always be the view_page view.

Viewing the Result of all Our Edits to views.py

The result of all of our edits to views.py will leave it looking like this:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
from docutils.core import publish_parts
import re

from pyramid.httpexceptions import HTTPFound
from pyramid.view import view_config

from tutorial.models import Page

# regular expression used to find WikiWords
wikiwords = re.compile(r"\b([A-Z]\w+[A-Z]+\w+)")

@view_config(context='tutorial.models.Wiki')
def view_wiki(context, request):
    return HTTPFound(location=request.resource_url(context, 'FrontPage'))

@view_config(context='tutorial.models.Page',
             renderer='tutorial:templates/view.pt')
def view_page(context, request):
    wiki = context.__parent__

    def check(match):
        word = match.group(1)
        if word in wiki:
            page = wiki[word]
            view_url = request.resource_url(page)
            return '<a href="%s">%s</a>' % (view_url, word)
        else:
            add_url = request.application_url + '/add_page/' + word 
            return '<a href="%s">%s</a>' % (add_url, word)

    content = publish_parts(context.data, writer_name='html')['html_body']
    content = wikiwords.sub(check, content)
    edit_url = request.resource_url(context, 'edit_page')
    return dict(page = context, content = content, edit_url = edit_url)

@view_config(name='add_page', context='tutorial.models.Wiki',
             renderer='tutorial:templates/edit.pt')
def add_page(context, request):
    name = request.subpath[0]
    if 'form.submitted' in request.params:
        body = request.params['body']
        page = Page(body)
        page.__name__ = name
        page.__parent__ = context
        context[name] = page
        return HTTPFound(location = request.resource_url(page))
    save_url = request.resource_url(context, 'add_page', name)
    page = Page('')
    page.__name__ = name
    page.__parent__ = context
    return dict(page = page, save_url = save_url)

@view_config(name='edit_page', context='tutorial.models.Page',
             renderer='tutorial:templates/edit.pt')
def edit_page(context, request):
    if 'form.submitted' in request.params:
        context.data = request.params['body']
        return HTTPFound(location = request.resource_url(context))

    return dict(page = context,
                save_url = request.resource_url(context, 'edit_page'))
Adding Templates

Most view callables we’ve added expected to be rendered via a template. The default templating systems in Pyramid are Chameleon and Mako. Chameleon is a variant of ZPT, which is an XML-based templating language. Mako is a non-XML-based templating language. Because we had to pick one, we chose Chameleon for this tutorial.

The templates we create will live in the templates directory of our tutorial package. Chameleon templates must have a .pt extension to be recognized as such.

The view.pt Template

The view.pt template is used for viewing a single Page. It is used by the view_page view function. It should have a div that is “structure replaced” with the content value provided by the view. It should also have a link on the rendered page that points at the “edit” URL (the URL which invokes the edit_page view for the page being viewed).

Once we’re done with the view.pt template, it will look a lot like the below:

<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN"
  "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en"
      xmlns:tal="http://xml.zope.org/namespaces/tal">
<head>
  <title>${page.__name__} - Pyramid tutorial wiki (based on
      TurboGears 20-Minute Wiki)</title>
  <meta http-equiv="Content-Type" content="text/html;charset=UTF-8"/>
  <meta name="keywords" content="python web application" />
  <meta name="description" content="pyramid web application" />
  <link rel="shortcut icon"
        href="${request.static_url('tutorial:static/favicon.ico')}" />
  <link rel="stylesheet"
        href="${request.static_url('tutorial:static/pylons.css')}"
        type="text/css" media="screen" charset="utf-8" />
  <!--[if lte IE 6]>
  <link rel="stylesheet"
        href="${request.static_url('tutorial:static/ie6.css')}"
        type="text/css" media="screen" charset="utf-8" />
  <![endif]-->
</head>
<body>
  <div id="wrap">
    <div id="top-small">
      <div class="top-small align-center">
        <div>
          <img width="220" height="50" alt="pyramid"
        src="${request.static_url('tutorial:static/pyramid-small.png')}" />
        </div>
      </div>
    </div>
    <div id="middle">
      <div class="middle align-right">
        <div id="left" class="app-welcome align-left">
          Viewing <b><span tal:replace="page.__name__">Page Name Goes
          Here</span></b><br/>
          You can return to the
          <a href="${request.application_url}">FrontPage</a>.<br/>
        </div>
        <div id="right" class="app-welcome align-right"></div>
      </div>
    </div>
    <div id="bottom">
      <div class="bottom">
        <div tal:replace="structure content">
          Page text goes here.
        </div>
        <p>
          <a tal:attributes="href edit_url" href="">
            Edit this page
          </a>
        </p>
      </div>
    </div>
  </div>
  <div id="footer">
    <div class="footer"
         >&copy; Copyright 2008-2011, Agendaless Consulting.</div>
  </div>
</body>
</html>

Note

The names available for our use in a template are always those that are present in the dictionary returned by the view callable. But our templates make use of a request object that none of our tutorial views return in their dictionary. This value appears as if “by magic”. However, request is one of several names that are available “by default” in a template when a template renderer is used. See *.pt or *.txt: Chameleon Template Renderers for more information about other names that are available by default in a template when a template is used as a renderer.

The edit.pt Template

The edit.pt template is used for adding and editing a Page. It is used by the add_page and edit_page view functions. It should display a page containing a form that POSTs back to the “save_url” argument supplied by the view. The form should have a “body” textarea field (the page data), and a submit button that has the name “form.submitted”. The textarea in the form should be filled with any existing page data when it is rendered.

Once we’re done with the edit.pt template, it will look a lot like the below:

<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN"
  "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en"
      xmlns:tal="http://xml.zope.org/namespaces/tal">
<head>
  <title>${page.__name__} - Pyramid tutorial wiki (based on
    TurboGears 20-Minute Wiki)</title>
  <meta http-equiv="Content-Type" content="text/html;charset=UTF-8"/>
  <meta name="keywords" content="python web application" />
  <meta name="description" content="pyramid web application" />
  <link rel="shortcut icon"
        href="${request.static_url('tutorial:static/favicon.ico')}" />
  <link rel="stylesheet"
        href="${request.static_url('tutorial:static/pylons.css')}"
        type="text/css" media="screen" charset="utf-8" />
  <!--[if lte IE 6]>
  <link rel="stylesheet"
        href="${request.static_url('tutorial:static/ie6.css')}"
        type="text/css" media="screen" charset="utf-8" />
  <![endif]-->
</head>
<body>
  <div id="wrap">
    <div id="top-small">
      <div class="top-small align-center">
        <div>
          <img width="220" height="50" alt="pyramid"
        src="${request.static_url('tutorial:static/pyramid-small.png')}" />
        </div>
      </div>
    </div>
    <div id="middle">
      <div class="middle align-right">
        <div id="left" class="app-welcome align-left">
          Editing <b><span tal:replace="page.__name__">Page Name Goes
            Here</span></b><br/>
          You can return to the
          <a href="${request.application_url}">FrontPage</a>.<br/>
        </div>
        <div id="right" class="app-welcome align-right"></div>
      </div>
    </div>
    <div id="bottom">
      <div class="bottom">
        <form action="${save_url}" method="post">
          <textarea name="body" tal:content="page.data" rows="10"
                    cols="60"/><br/>
          <input type="submit" name="form.submitted" value="Save"/>
        </form>
      </div>
    </div>
  </div>
  <div id="footer">
    <div class="footer"
         >&copy; Copyright 2008-2011, Agendaless Consulting.</div>
  </div>
</body>
</html>
Static Assets

Our templates name a single static asset named pylons.css. We don’t need to create this file within our package’s static directory because it was provided at the time we created the project. This file is a little too long to replicate within the body of this guide, however it is available online.

This CSS file will be accessed via e.g. http://localhost:6543/static/pylons.css by virtue of the call to add_static_view directive we’ve made in the __init__.py file. Any number and type of static assets can be placed in this directory (or subdirectories) and are just referred to by URL or by using the convenience method static_url e.g. request.static_url('{{package}}:static/foo.css') within templates.

Viewing the Application in a Browser

We can finally examine our application in a browser. The views we’ll try are as follows:

  • Visiting http://localhost:6543/ in a browser invokes the view_wiki view. This always redirects to the view_page view of the FrontPage Page resource.
  • Visiting http://localhost:6543/FrontPage/ in a browser invokes the view_page view of the front page resource. This is because it’s the default view (a view without a name) for Page resources.
  • Visiting http://localhost:6543/FrontPage/edit_page in a browser invokes the edit view for the FrontPage Page resource.
  • Visiting http://localhost:6543/add_page/SomePageName in a browser invokes the add view for a Page.
  • To generate an error, visit http://localhost:6543/add_page which will generate an IndexError for the expression request.subpath[0]. You’ll see an interactive traceback facility provided by pyramid_debugtoolbar.

Adding Authorization

Our application currently allows anyone with access to the server to view, edit, and add pages to our wiki. For purposes of demonstration we’ll change our application to allow people whom are members of a group named group:editors to add and edit wiki pages but we’ll continue allowing anyone with access to the server to view pages. Pyramid provides facilities for authorization and authentication. We’ll make use of both features to provide security to our application.

We will add an authentication policy and an authorization policy to our application registry, add a security.py module and give our root resource an ACL.

Then we will add login and logout views, and modify the existing views to make them return a logged_in flag to the renderer and add permission declarations to their view_config decorators.

Finally, we will add a login.pt template and change the existing view.pt and edit.pt to show a “Logout” link when not logged in.

The source code for this tutorial stage can be browsed via http://github.com/Pylons/pyramid/tree/1.2-branch/docs/tutorials/wiki/src/authorization/.

Adding Authentication and Authorization Policies

We’ll change our package’s __init__.py file to enable an AuthTktAuthenticationPolicy and an ACLAuthorizationPolicy to enable declarative security checking. We need to import the new policies:

1
2
3
from pyramid.authentication import AuthTktAuthenticationPolicy
from pyramid.authorization import ACLAuthorizationPolicy
from tutorial.security import groupfinder

Then, we’ll add those policies to the configuration:

1
2
3
4
5
6
    authn_policy = AuthTktAuthenticationPolicy(secret='sosecret',
                                               callback=groupfinder)
    authz_policy = ACLAuthorizationPolicy()
    config = Configurator(root_factory=root_factory, settings=settings,
                          authentication_policy=authn_policy,
                          authorization_policy=authz_policy)

Note that the creation of an AuthTktAuthenticationPolicy requires two arguments: secret and callback. secret is a string representing an encryption key used by the “authentication ticket” machinery represented by this policy: it is required. The callback is a reference to a groupfinder function in the tutorial package’s security.py file. We haven’t added that module yet, but we’re about to.

When you’re done, your __init__.py will look like so:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
from pyramid.config import Configurator
from pyramid_zodbconn import get_connection

from pyramid.authentication import AuthTktAuthenticationPolicy
from pyramid.authorization import ACLAuthorizationPolicy

from tutorial.models import appmaker
from tutorial.security import groupfinder

def root_factory(request):
    conn = get_connection(request)
    return appmaker(conn.root())

def main(global_config, **settings):
    """ This function returns a WSGI application.
    
    It is usually called by the PasteDeploy framework during 
    ``paster serve``.
    """
    authn_policy = AuthTktAuthenticationPolicy(secret='sosecret',
                                               callback=groupfinder)
    authz_policy = ACLAuthorizationPolicy()
    config = Configurator(root_factory=root_factory, settings=settings,
                          authentication_policy=authn_policy,
                          authorization_policy=authz_policy)
    config.add_static_view('static', 'tutorial:static', cache_max_age=3600)
    config.scan('tutorial')
    return config.make_wsgi_app()
Adding security.py

Add a security.py module within your package (in the same directory as __init__.py, views.py, etc.) with the following content:

1
2
3
4
5
6
7
8
USERS = {'editor':'editor',
          'viewer':'viewer'}
GROUPS = {'editor':['group:editors']}

def groupfinder(userid, request):
    if userid in USERS:
        return GROUPS.get(userid, [])

The groupfinder function defined here is an authentication policy “callback”; it is a callable that accepts a userid and a request. If the userid exists in the system, the callback will return a sequence of group identifiers (or an empty sequence if the user isn’t a member of any groups). If the userid does not exist in the system, the callback will return None. In a production system, user and group data will most often come from a database, but here we use “dummy” data to represent user and groups sources. Note that the editor user is a member of the group:editors group in our dummy group data (the GROUPS data structure).

Giving Our Root Resource an ACL

We need to give our root resource object an ACL. This ACL will be sufficient to provide enough information to the Pyramid security machinery to challenge a user who doesn’t have appropriate credentials when he attempts to invoke the add_page or edit_page views.

We need to perform some imports at module scope in our models.py file:

1
2
from pyramid.security import Allow
from pyramid.security import Everyone

Our root resource object is a Wiki instance. We’ll add the following line at class scope to our Wiki class:

1
2
__acl__ = [ (Allow, Everyone, 'view'),
            (Allow, 'group:editors', 'edit') ]

It’s only happenstance that we’re assigning this ACL at class scope. An ACL can be attached to an object instance too; this is how “row level security” can be achieved in Pyramid applications. We actually only need one ACL for the entire system, however, because our security requirements are simple, so this feature is not demonstrated.

Our resulting models.py file will now look like so:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
from persistent import Persistent
from persistent.mapping import PersistentMapping

from pyramid.security import Allow
from pyramid.security import Everyone

class Wiki(PersistentMapping):
    __name__ = None
    __parent__ = None
    __acl__ = [ (Allow, Everyone, 'view'),
                (Allow, 'group:editors', 'edit') ]

class Page(Persistent):
    def __init__(self, data):
        self.data = data

def appmaker(zodb_root):
    if not 'app_root' in zodb_root:
        app_root = Wiki()
        frontpage = Page('This is the front page')
        app_root['FrontPage'] = frontpage
        frontpage.__name__ = 'FrontPage'
        frontpage.__parent__ = app_root
        zodb_root['app_root'] = app_root
        import transaction
        transaction.commit()
    return zodb_root['app_root']
Adding Login and Logout Views

We’ll add a login view which renders a login form and processes the post from the login form, checking credentials.

We’ll also add a logout view to our application and provide a link to it. This view will clear the credentials of the logged in user and redirect back to the front page.

We’ll add a different file (for presentation convenience) to add login and logout views. Add a file named login.py to your application (in the same directory as views.py) with the following content:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
from pyramid.httpexceptions import HTTPFound

from pyramid.security import remember
from pyramid.security import forget
from pyramid.view import view_config

from tutorial.security import USERS

@view_config(context='tutorial.models.Wiki', name='login',
             renderer='templates/login.pt')
@view_config(context='pyramid.httpexceptions.HTTPForbidden',
             renderer='templates/login.pt')
def login(request):
    login_url = request.resource_url(request.context, 'login')
    referrer = request.url
    if referrer == login_url:
        referrer = '/' # never use the login form itself as came_from
    came_from = request.params.get('came_from', referrer)
    message = ''
    login = ''
    password = ''
    if 'form.submitted' in request.params:
        login = request.params['login']
        password = request.params['password']
        if USERS.get(login) == password:
            headers = remember(request, login)
            return HTTPFound(location = came_from,
                             headers = headers)
        message = 'Failed login'

    return dict(
        message = message,
        url = request.application_url + '/login',
        came_from = came_from,
        login = login,
        password = password,
        )
    
@view_config(context='tutorial.models.Wiki', name='logout')
def logout(request):
    headers = forget(request)
    return HTTPFound(location = request.resource_url(request.context),
                     headers = headers)

Note that the login view callable in the login.py file has two view configuration decorators. The order of these decorators is unimportant. Each just adds a different view configuration for the login view callable.

The first view configuration decorator configures the login view callable so it will be invoked when someone visits /login (when the context is a Wiki and the view name is login). The second decorator (with context of pyramid.httpexceptions.HTTPForbidden) specifies a forbidden view. This configures our login view to be presented to the user when Pyramid detects that a view invocation can not be authorized. Because we’ve configured a forbidden view, the login view callable will be invoked whenever one of our users tries to execute a view callable that they are not allowed to invoke as determined by the authorization policy in use. In our application, for example, this means that if a user has not logged in, and he tries to add or edit a Wiki page, he will be shown the login form. Before being allowed to continue on to the add or edit form, he will have to provide credentials that give him permission to add or edit via this login form.

Changing Existing Views

Then we need to change each of our view_page, edit_page and add_page views in views.py to pass a “logged in” parameter into its template. We’ll add something like this to each view body:

1
2
from pyramid.security import authenticated_userid
logged_in = authenticated_userid(request)

We’ll then change the return value of each view that has an associated renderer to pass the resulting logged_in value to the template. For example:

1
2
3
4
return dict(page = context,
            content = content,
            logged_in = logged_in,
            edit_url = edit_url)
Adding permission Declarations to our view_config Decorators

To protect each of our views with a particular permission, we need to pass a permission argument to each of our pyramid.view.view_config decorators. To do so, within views.py:

  • We add permission='view' to the decorator attached to the view_wiki and view_page view functions. This makes the assertion that only users who possess the view permission against the context resource at the time of the request may invoke these views. We’ve granted pyramid.security.Everyone the view permission at the root model via its ACL, so everyone will be able to invoke the view_wiki and view_page views.
  • We add permission='edit' to the decorator attached to the add_page and edit_page view functions. This makes the assertion that only users who possess the effective edit permission against the context resource at the time of the request may invoke these views. We’ve granted the group:editors principal the edit permission at the root model via its ACL, so only a user whom is a member of the group named group:editors will able to invoke the add_page or edit_page views. We’ve likewise given the editor user membership to this group via the security.py file by mapping him to the group:editors group in the GROUPS data structure (GROUPS = {'editor':['group:editors']}); the groupfinder function consults the GROUPS data structure. This means that the editor user can add and edit pages.
Adding the login.pt Template

Add a login.pt template to your templates directory. It’s referred to within the login view we just added to login.py.

<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN"
  "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en"
      xmlns:tal="http://xml.zope.org/namespaces/tal">
<head>
  <title>Login - Pyramid tutorial wiki (based on TurboGears
    20-Minute Wiki)</title>
  <meta http-equiv="Content-Type" content="text/html;charset=UTF-8"/>
  <meta name="keywords" content="python web application" />
  <meta name="description" content="pyramid web application" />
  <link rel="shortcut icon"
        href="${request.static_url('tutorial:static/favicon.ico')}" />
  <link rel="stylesheet"
        href="${request.static_url('tutorial:static/pylons.css')}"
        type="text/css" media="screen" charset="utf-8" />
  <!--[if lte IE 6]>
  <link rel="stylesheet"
        href="${request.static_url('tutorial:static/ie6.css')}"
        type="text/css" media="screen" charset="utf-8" />
  <![endif]-->
</head>
<body>
  <div id="wrap">
    <div id="top-small">
      <div class="top-small align-center">
        <div>
          <img width="220" height="50" alt="pyramid"
        src="${request.static_url('tutorial:static/pyramid-small.png')}" />
        </div>
      </div>
    </div>
    <div id="middle">
      <div class="middle align-right">
        <div id="left" class="app-welcome align-left">
          <b>Login</b><br/>
          <span tal:replace="message"/>
        </div>
        <div id="right" class="app-welcome align-right"></div>
      </div>
    </div>
    <div id="bottom">
      <div class="bottom">
        <form action="${url}" method="post">
          <input type="hidden" name="came_from" value="${came_from}"/>
          <input type="text" name="login" value="${login}"/><br/>
          <input type="password" name="password"
                 value="${password}"/><br/>
          <input type="submit" name="form.submitted" value="Log In"/>
        </form>
      </div>
    </div>
  </div>
  <div id="footer">
    <div class="footer"
         >&copy; Copyright 2008-2011, Agendaless Consulting.</div>
  </div>
</body>
</html>
Change view.pt and edit.pt

We’ll also need to change our edit.pt and view.pt templates to display a “Logout” link if someone is logged in. This link will invoke the logout view.

To do so we’ll add this to both templates within the <div id="right" class="app-welcome align-right"> div:

<span tal:condition="logged_in">
   <a href="${request.application_url}/logout">Logout</a>
</span>
Seeing Our Changes To views.py and our Templates

Our views.py module will look something like this when we’re done:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
from docutils.core import publish_parts
import re

from pyramid.httpexceptions import HTTPFound
from pyramid.view import view_config
from pyramid.security import authenticated_userid

from tutorial.models import Page

# regular expression used to find WikiWords
wikiwords = re.compile(r"\b([A-Z]\w+[A-Z]+\w+)")

@view_config(context='tutorial.models.Wiki', permission='view')
def view_wiki(context, request):
    return HTTPFound(location=request.resource_url(context, 'FrontPage'))

@view_config(context='tutorial.models.Page',
             renderer='templates/view.pt', permission='view')
def view_page(context, request):
    wiki = context.__parent__

    def check(match):
        word = match.group(1)
        if word in wiki:
            page = wiki[word]
            view_url = request.resource_url(page)
            return '<a href="%s">%s</a>' % (view_url, word)
        else:
            add_url = request.application_url + '/add_page/' + word 
            return '<a href="%s">%s</a>' % (add_url, word)

    content = publish_parts(context.data, writer_name='html')['html_body']
    content = wikiwords.sub(check, content)
    edit_url = request.resource_url(context, 'edit_page')

    logged_in = authenticated_userid(request)

    return dict(page = context, content = content, edit_url = edit_url,
                logged_in = logged_in)

@view_config(name='add_page', context='tutorial.models.Wiki',
             renderer='templates/edit.pt',
             permission='edit')
def add_page(context, request):
    name = request.subpath[0]
    if 'form.submitted' in request.params:
        body = request.params['body']
        page = Page(body)
        page.__name__ = name
        page.__parent__ = context
        context[name] = page
        return HTTPFound(location = request.resource_url(page))
    save_url = request.resource_url(context, 'add_page', name)
    page = Page('')
    page.__name__ = name
    page.__parent__ = context

    logged_in = authenticated_userid(request)

    return dict(page = page, save_url = save_url, logged_in = logged_in)

@view_config(name='edit_page', context='tutorial.models.Page',
             renderer='templates/edit.pt',
             permission='edit')
def edit_page(context, request):
    if 'form.submitted' in request.params:
        context.data = request.params['body']
        return HTTPFound(location = request.resource_url(context))

    logged_in = authenticated_userid(request)

    return dict(page = context,
                save_url = request.resource_url(context, 'edit_page'),
                logged_in = logged_in)

Our edit.pt template will look something like this when we’re done:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN"
  "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en"
      xmlns:tal="http://xml.zope.org/namespaces/tal">
<head>
  <title>${page.__name__} - Pyramid tutorial wiki (based on
      TurboGears 20-Minute Wiki)</title>
  <meta http-equiv="Content-Type" content="text/html;charset=UTF-8"/>
  <meta name="keywords" content="python web application" />
  <meta name="description" content="pyramid web application" />
  <link rel="shortcut icon"
        href="${request.static_url('tutorial:static/favicon.ico')}" />
  <link rel="stylesheet"
        href="${request.static_url('tutorial:static/pylons.css')}"
        type="text/css" media="screen" charset="utf-8" />
  <!--[if lte IE 6]>
  <link rel="stylesheet"
        href="${request.static_url('tutorial:static/ie6.css')}"
        type="text/css" media="screen" charset="utf-8" />
  <![endif]-->
</head>
<body>
  <div id="wrap">
    <div id="top-small">
      <div class="top-small align-center">
        <div>
          <img width="220" height="50" alt="pyramid"
        src="${request.static_url('tutorial:static/pyramid-small.png')}" />
        </div>
      </div>
    </div>
    <div id="middle">
      <div class="middle align-right">
        <div id="left" class="app-welcome align-left">
          Editing <b><span tal:replace="page.__name__">Page Name
            Goes Here</span></b><br/>
          You can return to the
          <a href="${request.application_url}">FrontPage</a>.<br/>
        </div>
        <div id="right" class="app-welcome align-right">
          <span tal:condition="logged_in">
              <a href="${request.application_url}/logout">Logout</a>
          </span>
        </div>
      </div>
    </div>
    <div id="bottom">
      <div class="bottom">
        <form action="${save_url}" method="post">
          <textarea name="body" tal:content="page.data" rows="10"
                    cols="60"/><br/>
          <input type="submit" name="form.submitted" value="Save"/>
        </form>
      </div>
    </div>
  </div>
  <div id="footer">
    <div class="footer"
         >&copy; Copyright 2008-2011, Agendaless Consulting.</div>
  </div>
</body>
</html>

Our view.pt template will look something like this when we’re done:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN"
  "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en"
      xmlns:tal="http://xml.zope.org/namespaces/tal">
<head>
  <title>${page.__name__} - Pyramid tutorial wiki (based on
    TurboGears 20-Minute Wiki)</title>
  <meta http-equiv="Content-Type" content="text/html;charset=UTF-8"/>
  <meta name="keywords" content="python web application" />
  <meta name="description" content="pyramid web application" />
  <link rel="shortcut icon"
        href="${request.static_url('tutorial:static/favicon.ico')}" />
  <link rel="stylesheet"
        href="${request.static_url('tutorial:static/pylons.css')}"
        type="text/css" media="screen" charset="utf-8" />
  <!--[if lte IE 6]>
  <link rel="stylesheet"
        href="${request.static_url('tutorial:static/ie6.css')}"
        type="text/css" media="screen" charset="utf-8" />
  <![endif]-->
</head>
<body>
  <div id="wrap">
    <div id="top-small">
      <div class="top-small align-center">
        <div>
          <img width="220" height="50" alt="pyramid"
        src="${request.static_url('tutorial:static/pyramid-small.png')}" />
        </div>
      </div>
    </div>
    <div id="middle">
      <div class="middle align-right">
        <div id="left" class="app-welcome align-left">
          Viewing <b><span tal:replace="page.__name__">Page Name
            Goes Here</span></b><br/>
          You can return to the
          <a href="${request.application_url}">FrontPage</a>.<br/>
        </div>
        <div id="right" class="app-welcome align-right">
          <span tal:condition="logged_in">
            <a href="${request.application_url}/logout">Logout</a>
          </span>
        </div>
      </div>
    </div>
    <div id="bottom">
      <div class="bottom">
        <div tal:replace="structure content">
          Page text goes here.
        </div>
        <p>
          <a tal:attributes="href edit_url" href="">
            Edit this page
          </a>
        </p>
      </div>
    </div>
  </div>
  <div id="footer">
    <div class="footer"
         >&copy; Copyright 2008-2011, Agendaless Consulting.</div>
  </div>
</body>
</html>
Viewing the Application in a Browser

We can finally examine our application in a browser. The views we’ll try are as follows:

  • Visiting http://localhost:6543/ in a browser invokes the view_wiki view. This always redirects to the view_page view of the FrontPage page resource. It is executable by any user.
  • Visiting http://localhost:6543/FrontPage/ in a browser invokes the view_page view of the FrontPage Page resource. This is because it’s the default view (a view without a name) for Page resources. It is executable by any user.
  • Visiting http://localhost:6543/FrontPage/edit_page in a browser invokes the edit view for the FrontPage Page resource. It is executable by only the editor user. If a different user (or the anonymous user) invokes it, a login form will be displayed. Supplying the credentials with the username editor, password editor will show the edit page form being displayed.
  • Visiting http://localhost:6543/add_page/SomePageName in a browser invokes the add view for a page. It is executable by only the editor user. If a different user (or the anonymous user) invokes it, a login form will be displayed. Supplying the credentials with the username editor, password editor will show the edit page form being displayed.
  • After logging in (as a result of hitting an edit or add page and submitting the login form with the editor credentials), we’ll see a Logout link in the upper right hand corner. When we click it, we’re logged out, and redirected back to the front page.

Adding Tests

We will now add tests for the models and the views and a few functional tests in the tests.py. Tests ensure that an application works, and that it continues to work after some changes are made in the future.

Testing the Models

We write tests for the model classes and the appmaker. Changing tests.py, we’ll write a separate test class for each model class, and we’ll write a test class for the appmaker.

To do so, we’ll retain the tutorial.tests.ViewTests class provided as a result of the pyramid_zodb project generator. We’ll add three test classes: one for the Page model named PageModelTests, one for the Wiki model named WikiModelTests, and one for the appmaker named AppmakerTests.

Testing the Views

We’ll modify our tests.py file, adding tests for each view function we added above. As a result, we’ll delete the ViewTests test in the file, and add four other test classes: ViewWikiTests, ViewPageTests, AddPageTests, and EditPageTests. These test the view_wiki, view_page, add_page, and edit_page views respectively.

Functional tests

We test the whole application, covering security aspects that are not tested in the unit tests, like logging in, logging out, checking that the viewer user cannot add or edit pages, but the editor user can, and so on.

Viewing the results of all our edits to tests.py

Once we’re done with the tests.py module, it will look a lot like the below:

  1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
import unittest

from pyramid import testing

class PageModelTests(unittest.TestCase):

    def _getTargetClass(self):
        from tutorial.models import Page
        return Page

    def _makeOne(self, data=u'some data'):
        return self._getTargetClass()(data=data)

    def test_constructor(self):
        instance = self._makeOne()
        self.assertEqual(instance.data, u'some data')

class WikiModelTests(unittest.TestCase):

    def _getTargetClass(self):
        from tutorial.models import Wiki
        return Wiki

    def _makeOne(self):
        return self._getTargetClass()()

    def test_it(self):
        wiki = self._makeOne()
        self.assertEqual(wiki.__parent__, None)
        self.assertEqual(wiki.__name__, None)

class AppmakerTests(unittest.TestCase):
    def _callFUT(self, zodb_root):
        from tutorial.models import appmaker
        return appmaker(zodb_root)

    def test_it(self):
        root = {}
        self._callFUT(root)
        self.assertEqual(root['app_root']['FrontPage'].data,
                         'This is the front page')

class ViewWikiTests(unittest.TestCase):
    def test_it(self):
        from tutorial.views import view_wiki
        context = testing.DummyResource()
        request = testing.DummyRequest()
        response = view_wiki(context, request)
        self.assertEqual(response.location, 'http://example.com/FrontPage')

class ViewPageTests(unittest.TestCase):
    def _callFUT(self, context, request):
        from tutorial.views import view_page
        return view_page(context, request)

    def test_it(self):
        wiki = testing.DummyResource()
        wiki['IDoExist'] = testing.DummyResource()
        context = testing.DummyResource(data='Hello CruelWorld IDoExist')
        context.__parent__ = wiki
        context.__name__ = 'thepage'
        request = testing.DummyRequest()
        info = self._callFUT(context, request)
        self.assertEqual(info['page'], context)
        self.assertEqual(
            info['content'],
            '<div class="document">\n'
            '<p>Hello <a href="http://example.com/add_page/CruelWorld">'
            'CruelWorld</a> '
            '<a href="http://example.com/IDoExist/">'
            'IDoExist</a>'
            '</p>\n</div>\n')
        self.assertEqual(info['edit_url'],
                         'http://example.com/thepage/edit_page')


class AddPageTests(unittest.TestCase):
    def _callFUT(self, context, request):
        from tutorial.views import add_page
        return add_page(context, request)

    def test_it_notsubmitted(self):
        context = testing.DummyResource()
        request = testing.DummyRequest()
        request.subpath = ['AnotherPage']
        info = self._callFUT(context, request)
        self.assertEqual(info['page'].data,'')
        self.assertEqual(
            info['save_url'],
            request.resource_url(context, 'add_page', 'AnotherPage'))

    def test_it_submitted(self):
        context = testing.DummyResource()
        request = testing.DummyRequest({'form.submitted':True,
                                        'body':'Hello yo!'})
        request.subpath = ['AnotherPage']
        self._callFUT(context, request)
        page = context['AnotherPage']
        self.assertEqual(page.data, 'Hello yo!')
        self.assertEqual(page.__name__, 'AnotherPage')
        self.assertEqual(page.__parent__, context)

class EditPageTests(unittest.TestCase):
    def _callFUT(self, context, request):
        from tutorial.views import edit_page
        return edit_page(context, request)

    def test_it_notsubmitted(self):
        context = testing.DummyResource()
        request = testing.DummyRequest()
        info = self._callFUT(context, request)
        self.assertEqual(info['page'], context)
        self.assertEqual(info['save_url'],
                         request.resource_url(context, 'edit_page'))

    def test_it_submitted(self):
        context = testing.DummyResource()
        request = testing.DummyRequest({'form.submitted':True,
                                        'body':'Hello yo!'})
        response = self._callFUT(context, request)
        self.assertEqual(response.location, 'http://example.com/')
        self.assertEqual(context.data, 'Hello yo!')

class FunctionalTests(unittest.TestCase):

    viewer_login = '/login?login=viewer&password=viewer' \
                   '&came_from=FrontPage&form.submitted=Login'
    viewer_wrong_login = '/login?login=viewer&password=incorrect' \
                   '&came_from=FrontPage&form.submitted=Login'
    editor_login = '/login?login=editor&password=editor' \
                   '&came_from=FrontPage&form.submitted=Login'

    def setUp(self):
        import tempfile
        import os.path
        from tutorial import main
        self.tmpdir = tempfile.mkdtemp()

        dbpath = os.path.join( self.tmpdir, 'test.db')
        uri = 'file://' + dbpath
        settings = { 'zodbconn.uri' : uri ,
                     'pyramid.includes': ['pyramid_zodbconn', 'pyramid_tm'] }

        app = main({}, **settings)
        self.db = app.registry.zodb_database
        from webtest import TestApp
        self.testapp = TestApp(app)

    def tearDown(self):
        import shutil
        self.db.close()
        shutil.rmtree( self.tmpdir )

    def test_root(self):
        res = self.testapp.get('/', status=302)
        self.assertEqual(res.location, 'http://localhost/FrontPage')

    def test_FrontPage(self):
        res = self.testapp.get('/FrontPage', status=200)
        self.assertTrue('FrontPage' in res.body)

    def test_unexisting_page(self):
        res = self.testapp.get('/SomePage', status=404)
        self.assertTrue('Not Found' in res.body)

    def test_successful_log_in(self):
        res = self.testapp.get( self.viewer_login, status=302)
        self.assertEqual(res.location, 'http://localhost/FrontPage')

    def test_failed_log_in(self):
        res = self.testapp.get( self.viewer_wrong_login, status=200)
        self.assertTrue('login' in res.body)

    def test_logout_link_present_when_logged_in(self):
        res = self.testapp.get( self.viewer_login, status=302)
        res = self.testapp.get('/FrontPage', status=200)
        self.assertTrue('Logout' in res.body)

    def test_logout_link_not_present_after_logged_out(self):
        res = self.testapp.get( self.viewer_login, status=302)
        res = self.testapp.get('/FrontPage', status=200)
        res = self.testapp.get('/logout', status=302)
        self.assertTrue('Logout' not in res.body)

    def test_anonymous_user_cannot_edit(self):
        res = self.testapp.get('/FrontPage/edit_page', status=200)
        self.assertTrue('Login' in res.body)

    def test_anonymous_user_cannot_add(self):
        res = self.testapp.get('/add_page/NewPage', status=200)
        self.assertTrue('Login' in res.body)

    def test_viewer_user_cannot_edit(self):
        res = self.testapp.get( self.viewer_login, status=302)
        res = self.testapp.get('/FrontPage/edit_page', status=200)
        self.assertTrue('Login' in res.body)

    def test_viewer_user_cannot_add(self):
        res = self.testapp.get( self.viewer_login, status=302)
        res = self.testapp.get('/add_page/NewPage', status=200)
        self.assertTrue('Login' in res.body)

    def test_editors_member_user_can_edit(self):
        res = self.testapp.get( self.editor_login, status=302)
        res = self.testapp.get('/FrontPage/edit_page', status=200)
        self.assertTrue('Editing' in res.body)

    def test_editors_member_user_can_add(self):
        res = self.testapp.get( self.editor_login, status=302)
        res = self.testapp.get('/add_page/NewPage', status=200)
        self.assertTrue('Editing' in res.body)

    def test_editors_member_user_can_view(self):
        res = self.testapp.get( self.editor_login, status=302)
        res = self.testapp.get('/FrontPage', status=200)
        self.assertTrue('FrontPage' in res.body)
Running the Tests

We can run these tests by using setup.py test in the same way we did in Running the Tests. However, first we must edit our setup.py to include a dependency on WebTest, which we’ve used in our tests.py. Change the requires list in setup.py to include WebTest.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
requires = [
    'pyramid',
    'pyramid_zodbconn',
    'pyramid_tm',
    'pyramid_debugtoolbar',
    'ZODB3',
    'docutils',
    'WebTest', # add this
    ]

After we’ve added a dependency on WebTest in setup.py, we need to rerun setup.py develop to get WebTest installed into our virtualenv. Assuming our shell’s current working directory is the “tutorial” distribution directory:

On UNIX:

$ ../bin/python setup.py develop

On Windows:

c:\pyramidtut\tutorial> ..\Scripts\python setup.py develop

Once that command has completed successfully, we can run the tests themselves:

On UNIX:

$ ../bin/python setup.py test -q

On Windows:

c:\pyramidtut\tutorial> ..\Scripts\python setup.py test -q

The expected result looks something like:

.........
----------------------------------------------------------------------
Ran 23 tests in 1.653s

OK

Distributing Your Application

Once your application works properly, you can create a “tarball” from it by using the setup.py sdist command. The following commands assume your current working directory is the tutorial package we’ve created and that the parent directory of the tutorial package is a virtualenv representing a Pyramid environment.

On UNIX:

$ ../bin/python setup.py sdist

On Windows:

c:\pyramidtut> ..\Scripts\python setup.py sdist

The output of such a command will be something like:

running sdist
# .. more output ..
creating dist
tar -cf dist/tutorial-0.1.tar tutorial-0.1
gzip -f9 dist/tutorial-0.1.tar
removing 'tutorial-0.1' (and everything under it)

Note that this command creates a tarball in the “dist” subdirectory named tutorial-0.1.tar.gz. You can send this file to your friends to show them your cool new application. They should be able to install it by pointing the easy_install command directly at it. Or you can upload it to PyPI and share it with the rest of the world, where it can be downloaded via easy_install remotely like any other package people download from PyPI.

Converting a repoze.bfg Application to Pyramid

Prior iterations of Pyramid were released as a package named repoze.bfg. repoze.bfg users are encouraged to upgrade their deployments to Pyramid, as, after the first final release of Pyramid, further feature development on repoze.bfg will cease.

Most existing repoze.bfg applications can be converted to a Pyramid application in a completely automated fashion. However, if your application depends on packages which are not “core” parts of repoze.bfg but which nonetheless have repoze.bfg in their names (e.g. repoze.bfg.skins, repoze.bfg.traversalwrapper, repoze.bfg.jinja2), you will need to find an analogue for each. For example, by the time you read this, there will be a pyramid_jinja2 package, which can be used instead of repoze.bfg.jinja2. If an analogue does not seem to exist for a repoze.bfg add-on package that your application uses, please email the Pylons-devel maillist; we’ll convert the package to a Pyramid analogue for you.

Here’s how to convert a repoze.bfg application to a Pyramid application:

  1. Ensure that your application works under repoze.bfg version 1.3 or better. See http://docs.repoze.org/bfg/1.3/narr/install.html for repoze.bfg 1.3 installation instructions. If your application has an automated test suite, run it while your application is using repoze.bfg 1.3+. Otherwise, test it manually. It is only safe to proceed to the next step once your application works under repoze.bfg 1.3+.

    If your application has a proper set of dependencies, and a standard automated test suite, you might test your repoze.bfg application against repoze.bfg 1.3 like so:

    $ bfgenv/bin/python setup.py test

    bfgenv above will be the virtualenv into which you’ve installed repoze.bfg 1.3.

  2. Install Pyramid into a separate virtualenv as per the instructions in Installing Pyramid. The Pyramid virtualenv should be separate from the one you’ve used to install repoze.bfg. A quick way to do this:

    $ cd ~
$ virtualenv --no-site-packages pyramidenv
$ cd pyramidenv
$ bin/easy_install pyramid

  3. Put a copy of your repoze.bfg application into a temporary location (perhaps by checking a fresh copy of the application out of a version control repository). For example:

    $ cd /tmp
$ svn co http://my.server/my/bfg/application/trunk bfgapp

  4. Use the bfg2pyramid script present in the bin directory of the Pyramid virtualenv to convert all repoze.bfg Python import statements into compatible Pyramid import statements. bfg2pyramid will also fix ZCML directive usages of common repoze.bfg directives. You invoke bfg2pyramid by passing it the path of the copy of your application. The path passed should contain a “setup.py” file, representing your repoze.bfg application’s setup script. bfg2pyramid will change the copy of the application in place.

    $ ~/pyramidenv/bfg2pyramid /tmp/bfgapp

    bfg2pyramid will convert the following repoze.bfg application aspects to Pyramid compatible analogues:

    • Python import statements naming repoze.bfg APIs will be converted to Pyramid compatible import statements. Every Python file beneath the top-level path will be visited and converted recursively, except Python files which live in directories which start with a . (dot).
    • Each ZCML file found (recursively) within the path will have the default xmlns attribute attached to the configure tag changed from http://namespaces.repoze.org/bfg to http://pylonshq.com/pyramid. Every ZCML file beneath the top-level path (files ending with .zcml) will be visited and converted recursively, except ZCML files which live in directories which start with a . (dot).
    • ZCML files which contain directives that have attributes which name a repoze.bfg API module or attribute of an API module (e.g. context="repoze.bfg.exceptions.NotFound") will be converted to Pyramid compatible ZCML attributes (e.g. context="pyramid.exceptions.NotFound). Every ZCML file beneath the top-level path (files ending with .zcml) will be visited and converted recursively, except ZCML files which live in directories which start with a . (dot).

  5. Edit the setup.py file of the application you’ve just converted (if you’ve been using the example paths, this will be /tmp/bfgapp/setup.py) to depend on the pyramid distribution instead the of repoze.bfg distribution in its install_requires list. If you used a scaffold to create the repoze.bfg application, you can do so by changing the requires line near the top of the setup.py file. The original may look like this:

    requires = ['repoze.bfg', ... other dependencies ...]

    Edit the setup.py so it has:

    requires = ['pyramid', ... other dependencies ...]

    All other install-requires and tests-requires dependencies save for the one on repoze.bfg can remain the same.

  6. Convert any install_requires dependencies your application has on other add-on packages which have repoze.bfg in their names to Pyramid compatible analogues (e.g. repoze.bfg.jinja2 should be replaced with pyramid_jinja2). You may need to adjust configuration options and/or imports in your repoze.bfg application after replacing these add-ons. Read the documentation of the Pyramid add-on package for information.

  7. Only if you use ZCML and add-ons which use ZCML: The default xmlns of the configure tag in ZCML has changed. The bfg2pyramid script effects the default namespace change (it changes the configure tag default xmlns from http://namespaces.repoze.org/bfg to http://pylonshq.com/pyramid).

    This means that uses of add-ons which define ZCML directives in the http://namespaces.repoze.org/bfg namespace will begin to “fail” (they’re actually not really failing, but your ZCML assumes that they will always be used within a configure tag which names the http://namespaces.repoze.org/bfg namespace as its default xmlns). Symptom: when you attempt to start the application, an error such as ConfigurationError: ('Unknown directive', u'http://namespaces.repoze.org/bfg', u'workflow') is printed to the console and the application fails to start. In such a case, either add an xmlns="http://namespaces.repoze.org/bfg" attribute to each tag which causes a failure, or define a namespace alias in the configure tag and prefix each failing tag. For example, change this “failing” tag instance:

    <configure xmlns="http://pylonshq.com/pyramid">
    <failingtag attr="foo"/>
</configure>

    To this, which will begin to succeed:

    <configure xmlns="http://pylonshq.com/pyramid"
           xmlns:bfg="http://namespaces.repoze.org/bfg">
    <bfg:failingtag attr="foo"/>
</configure>

    You will also need to add the pyramid_zcml package to your setup.py install_requires list. In Pyramid, ZCML configuration became an optional add-on supported by the pyramid_zcml package.

  8. Retest your application using Pyramid. This might be as easy as:

    $ cd /tmp/bfgapp
$ ~/pyramidenv/bin/python setup.py test

  9. Fix any test failures.

  10. Fix any code which generates deprecation warnings.

  11. Start using the converted version of your application. Celebrate.

Two terminological changes have been made to Pyramid which make its documentation and newer APIs different than those of repoze.bfg. The concept that BFG called model is called resource in Pyramid and the concept that BFG called resource is called asset in Pyramid. Various APIs have changed as a result (although all have backwards compatible shims). Additionally, the environment variables that influenced server behavior which used to be prefixed with BFG_ (such as BFG_DEBUG_NOTFOUND) must now be prefixed with PYRAMID_.

Running Pyramid on Google’s App Engine

It is possible to run a Pyramid application on Google’s App Engine. Content from this tutorial was contributed by YoungKing, based on the “appengine-monkey” tutorial for Pylons. This tutorial is written in terms of using the command line on a UNIX system; it should be possible to perform similar actions on a Windows system.

  1. Download Google’s App Engine SDK and install it on your system.

  2. Use Subversion to check out the source code for appengine-monkey.

    $ svn co http://appengine-monkey.googlecode.com/svn/trunk/ \
    appengine-monkey

  3. Use appengine_homedir.py script in appengine-monkey to create a virtualenv for your application.

    $ export GAE_PATH=/usr/local/google_appengine
$ python2.5 /path/to/appengine-monkey/appengine-homedir.py --gae \
  $GAE_PATH pyramidapp

    Note that $GAE_PATH should be the path where you have unpacked the App Engine SDK. (On Mac OS X at least, /usr/local/google_appengine is indeed where the installer puts it).

    This will set up an environment in pyramidapp/, with some tools installed in pyramidapp/bin. There will also be a directory pyramidapp/app/ which is the directory you will upload to appengine.

  4. Install Pyramid into the virtualenv

    $ cd pyramidapp/
$ bin/easy_install pyramid

    This will install Pyramid in the environment.

  5. Create your application

    We’ll use the standard way to create a Pyramid application, but we’ll have to move some files around when we are done. The below commands assume your current working directory is the pyramidapp virtualenv directory you created in the third step above:

    $ cd app
$ rm -rf pyramidapp
$ bin/paster create -t pyramid_starter pyramidapp
$ mv pyramidapp aside
$ mv aside/pyramidapp .
$ rm -rf aside

  6. Edit config.py

    Edit the APP_NAME and APP_ARGS settings within config.py. The APP_NAME must be pyramidapp:main, and the APP_ARGS must be ({},). Any other settings in config.py should remain the same.

    APP_NAME = 'pyramidapp:main'
APP_ARGS = ({},)

  7. Edit runner.py

    To prevent errors for import site, add this code stanza before import site in app/runner.py:

    import sys
sys.path = [path for path in sys.path if 'site-packages' not in path]
import site

    You will also need to comment out the line that starts with assert sys.path in the file.

    # comment the sys.path assertion out
# assert sys.path[:len(cur_sys_path)] == cur_sys_path, (
#   "addsitedir() caused entries to be prepended to sys.path")

    For GAE development environment 1.3.0 or better, you will also need the following somewhere near the top of the runner.py file to fix a compatibility issue with appengine-monkey:

    import os
os.mkdir = None

  8. Run the application. dev_appserver.py is typically installed by the SDK in the global path but you need to be sure to run it with Python 2.5 (or whatever version of Python your GAE SDK expects).

    1
2
    $ cd ../..
$ python2.5 /usr/local/bin/dev_appserver.py pyramidapp/app/

    Startup success looks something like this:

    [chrism@vitaminf pyramid_gae]$ python2.5 \
              /usr/local/bin/dev_appserver.py \
              pyramidapp/app/
INFO     2009-05-03 22:23:13,887 appengine_rpc.py:157] # ... more...
Running application pyramidapp on port 8080: http://localhost:8080

    You may need to run “Make Symlinks” from the Google App Engine Launcher GUI application if your system doesn’t already have the dev_appserver.py script sitting around somewhere.

  9. Hack on your pyramid application, using a normal run, debug, restart process. For tips on how to use the pdb module within Google App Engine, see this blog post. In particular, you can create a function like so and call it to drop your console into a pdb trace:

    1
2
3
4
5
    def set_trace():
    import pdb, sys
    debugger = pdb.Pdb(stdin=sys.__stdin__,
        stdout=sys.__stdout__)
    debugger.set_trace(sys._getframe().f_back)

  10. Sign up for a GAE account and create an application. You’ll need a mobile phone to accept an SMS in order to receive authorization.

  11. Edit the application’s ID in app.yaml to match the application name you created during GAE account setup.

    application: mycoolpyramidapp

  12. Upload the application

    $ python2.5 /usr/local/bin/appcfg.py update pyramidapp/app

    You almost certainly won’t hit the 3000-file GAE file number limit when invoking this command. If you do, however, it will look like so:

    HTTPError: HTTP Error 400: Bad Request
Rolling back the update.
Error 400: --- begin server output ---
Max number of files and blobs is 3000.
--- end server output ---

    If you do experience this error, you will be able to get around this by zipping libraries. You can use pip to create zipfiles from packages. See Zipping Files Via Pip for more information about this.

    A successful upload looks like so:

    [chrism@vitaminf pyramidapp]$ python2.5 /usr/local/bin/appcfg.py \
                              update ../pyramidapp/app/
Scanning files on local disk.
Scanned 500 files.
# ... more output ...
Will check again in 16 seconds.
Checking if new version is ready to serve.
Closing update: new version is ready to start serving.
Uploading index definitions.

  13. Visit http://<yourapp>.appspot.com in a browser.

Zipping Files Via Pip

If you hit the Google App Engine 3000-file limit, you may need to create zipfile archives out of some distributions installed in your application’s virtualenv.

First, see which packages are available for zipping:

$ bin/pip zip -l

This shows your zipped packages (by default, none) and your unzipped packages. You can zip a package like so:

$ bin/pip zip pytz-2009g-py2.5.egg

Note that it requires the whole egg file name. For a Pyramid app, the following packages are good candidates to be zipped.

  • Chameleon
  • zope.i18n

Once the zipping procedure is finished you can try uploading again.

Running a Pyramid Application under mod_wsgi

mod_wsgi is an Apache module developed by Graham Dumpleton. It allows WSGI programs to be served using the Apache web server.

This guide will outline broad steps that can be used to get a Pyramid application running under Apache via mod_wsgi. This particular tutorial was developed under Apple’s Mac OS X platform (Snow Leopard, on a 32-bit Mac), but the instructions should be largely the same for all systems, delta specific path information for commands and files.

Note

Unfortunately these instructions almost certainly won’t work for deploying a Pyramid application on a Windows system using mod_wsgi. If you have experience with Pyramid and mod_wsgi on Windows systems, please help us document this experience by submitting documentation to the Pylons-devel maillist.

  1. The tutorial assumes you have Apache already installed on your system. If you do not, install Apache 2.X for your platform in whatever manner makes sense.

  2. Once you have Apache installed, install mod_wsgi. Use the (excellent) installation instructions for your platform into your system’s Apache installation.

  3. Install virtualenv into the Python which mod_wsgi will run using the easy_install program.

    $ sudo /usr/bin/easy_install-2.6 virtualenv

    This command may need to be performed as the root user.

  4. Create a virtualenv which we’ll use to install our application.

    $ cd ~
$ mkdir modwsgi
$ cd modwsgi
$ /usr/local/bin/virtualenv --no-site-packages env

  5. Install Pyramid into the newly created virtualenv:

    $ cd ~/modwsgi/env
$ bin/easy_install pyramid

  6. Create and install your Pyramid application. For the purposes of this tutorial, we’ll just be using the pyramid_starter application as a baseline application. Substitute your existing Pyramid application as necessary if you already have one.

    $ cd ~/modwsgi/env
$ bin/paster create -t pyramid_starter myapp
$ cd myapp
$ ../bin/python setup.py install

  7. Within the virtualenv directory (~/modwsgi/env), create a script named pyramid.wsgi. Give it these contents:

    from pyramid.paster import get_app
application = get_app(
  '/Users/chrism/modwsgi/env/myapp/production.ini', 'main')

    The first argument to get_app is the project Paste configuration file name. It’s best to use the production.ini file provided by your scaffold, as it contains settings appropriate for production. The second is the name of the section within the .ini file that should be loaded by mod_wsgi. The assignment to the name application is important: mod_wsgi requires finding such an assignment when it opens the file.

  8. Make the pyramid.wsgi script executable.

    $ cd ~/modwsgi/env
$ chmod 755 pyramid.wsgi

  9. Edit your Apache configuration and add some stuff. I happened to create a file named /etc/apache2/other/modwsgi.conf on my own system while installing Apache, so this stuff went in there.

    # Use only 1 Python sub-interpreter.  Multiple sub-interpreters
# play badly with C extensions.
WSGIApplicationGroup %{GLOBAL}
WSGIPassAuthorization On
WSGIDaemonProcess pyramid user=chrism group=staff threads=4 \
   python-path=/Users/chrism/modwsgi/env/lib/python2.6/site-packages
WSGIScriptAlias /myapp /Users/chrism/modwsgi/env/pyramid.wsgi

<Directory /Users/chrism/modwsgi/env>
  WSGIProcessGroup pyramid
  Order allow,deny
  Allow from all
</Directory>

  10. Restart Apache

    $ sudo /usr/sbin/apachectl restart

  11. Visit http://localhost/myapp in a browser. You should see the sample application rendered in your browser.

mod_wsgi has many knobs and a great variety of deployment modes. This is just one representation of how you might use it to serve up a Pyramid application. See the mod_wsgi configuration documentation for more in-depth configuration information.

Reference Material

Reference material includes documentation for every Pyramid API.

API Documentation

Comprehensive reference material for every public API exposed by Pyramid is available within this chapter. The API documentation is organized alphabetically by module name.

pyramid.authorization

pyramid.authentication

Authentication Policies
Helper Classes

pyramid.chameleon_text

These APIs will will work against template files which contain simple ${Genshi} - style replacement markers.

The API of pyramid.chameleon_text is identical to that of pyramid.chameleon_zpt; only its import location is different. If you need to import an API functions from this module as well as the pyramid.chameleon_zpt module within the same view file, use the as feature of the Python import statement, e.g.:

1
2
from pyramid.chameleon_zpt import render_template as zpt_render
from pyramid.chameleon_text import render_template as text_render

pyramid.chameleon_zpt

These APIs will work against files which supply template text which matches the ZPT specification.

The API of pyramid.chameleon_zpt is identical to that of pyramid.chameleon_text; only its import location is different. If you need to import an API functions from this module as well as the pyramid.chameleon_text module within the same view file, use the as feature of the Python import statement, e.g.:

1
2
from pyramid.chameleon_zpt import render_template as zpt_render
from pyramid.chameleon_text import render_template as text_render

pyramid.config

pyramid.events

Functions
Event Types

See Using Events for more information about how to register code which subscribes to these events.

pyramid.exceptions

pyramid.httpexceptions

pyramid.i18n

See Internationalization and Localization for more information about using Pyramid internationalization and localization services within an application.

pyramid.interfaces

Other Interfaces

pyramid.location

pyramid.paster

pyramid.renderers

null_renderer

An object that can be used in advanced integration cases as input to the view configuration renderer= argument. When the null renderer is used as a view renderer argument, Pyramid avoids converting the view callable result into a Response object. This is useful if you want to reuse the view configuration and lookup machinery outside the context of its use by the Pyramid router (e.g. the package named pyramid_rpc does this).

pyramid.request

Note

For information about the API of a multidict structure (such as that used as request.GET, request.POST, and request.params), see pyramid.interfaces.IMultiDict.

pyramid.scripting

pyramid.security

Authentication API Functions
Authorization API Functions
Constants
Everyone

The special principal id named ‘Everyone’. This principal id is granted to all requests. Its actual value is the string ‘system.Everyone’.

Authenticated

The special principal id named ‘Authenticated’. This principal id is granted to all requests which contain any other non-Everyone principal id (according to the authentication policy). Its actual value is the string ‘system.Authenticated’.

ALL_PERMISSIONS

An object that can be used as the permission member of an ACE which matches all permissions unconditionally. For example, an ACE that uses ALL_PERMISSIONS might be composed like so: ('Deny', 'system.Everyone', ALL_PERMISSIONS).

DENY_ALL

A convenience shorthand ACE that defines ('Deny', 'system.Everyone', ALL_PERMISSIONS). This is often used as the last ACE in an ACL in systems that use an “inheriting” security policy, representing the concept “don’t inherit any other ACEs”.

NO_PERMISSION_REQUIRED

A special permission which indicates that the view should always be executable by entirely anonymous users, regardless of the default permission, bypassing any authorization policy that may be in effect. Its actual value is the string ‘__no_permission_required__’.

Return Values
Allow

The ACE “action” (the first element in an ACE e.g. (Allow, Everyone, 'read') that means allow access. A sequence of ACEs makes up an ACL. It is a string, and it’s actual value is “Allow”.

Deny

The ACE “action” (the first element in an ACE e.g. (Deny, 'george', 'read') that means deny access. A sequence of ACEs makes up an ACL. It is a string, and it’s actual value is “Deny”.

pyramid.session

pyramid.settings

pyramid.static

pyramid.testing

pyramid.threadlocal

pyramid.traversal

pyramid.tweens

pyramid.url

pyramid.view

pyramid.wsgi

Detailed Change History

Pyramid Change History

Next release

  • Internal: catch unhashable discriminators early (raise an error instead of allowing them to find their way into resolveConflicts).
  • The match_param view predicate now accepts a string or a tuple. This replaces the broken behavior of accepting a dict. See https://github.com/Pylons/pyramid/issues/425 for more information.
Backwards Incompatibilities
  • The match_param view predicate no longer accepts a dict. This will have no negative affect because the implementation was broken for dict-based arguments.

1.2.7 (2012-01-20)

Bug Fixes

1.2.6 (2012-01-05)

Bug Fixes

  • Literal portions of route patterns were not URL-quoted when route_url or route_path was used to generate a URL or path.

  • The result of route_path or route_url might have been unicode or str depending on the input. It is now guaranteed to always be str.

  • URL matching when the pattern contained non-ASCII characters in literal parts was indeterminate. Now the pattern supplied to add_route is assumed to be either: a unicode value, or a str value that contains only ASCII characters. If you now want to match the path info from a URL that contains high order characters, you can pass the Unicode representation of the decoded path portion in the pattern.

  • When using a traverse= route predicate, traversal would fail with a URLDecodeError if there were any high-order characters in the traversal pattern or in the matched dynamic segments.

  • Using a dynamic segment named traverse in a route pattern like this:

    config.add_route('trav_route', 'traversal/{traverse:.*}')

    Would cause a UnicodeDecodeError when the route was matched and the matched portion of the URL contained any high-order characters. See https://github.com/Pylons/pyramid/issues/385 .

  • When using a *traverse stararg in a route pattern, a URL that matched that possessed a @@ in its name (signifying a view name) would be inappropriately quoted by the traversal machinery during traversal, resulting in the view not being found properly. See https://github.com/Pylons/pyramid/issues/382 and https://github.com/Pylons/pyramid/issues/375 .

Backwards Incompatibilities

  • String values passed to route_url or route_path that are meant to replace “remainder” matches will now be URL-quoted except for embedded slashes. For example:

    config.add_route('remain', '/foo*remainder')
request.route_path('remain', remainder='abc / def')
# -> '/foo/abc%20/%20def'

    Previously string values passed as remainder replacements were tacked on untouched, without any URL-quoting. But this doesn’t really work logically if the value passed is Unicode (raw unicode cannot be placed in a URL or in a path) and it is inconsistent with the rest of the URL generation machinery if the value is a string (it won’t be quoted unless by the caller).

    Some folks will have been relying on the older behavior to tack on query string elements and anchor portions of the URL; sorry, you’ll need to change your code to use the _query and/or _anchor arguments to route_path or route_url to do this now.

  • If you pass a bytestring that contains non-ASCII characters to add_route as a pattern, it will now fail at startup time. Use Unicode instead.

1.2.5 (2011-12-14)

Features
  • Backport the egg:pyramid#wsgiref and egg:pyramid#cherrypy server runners from the 1.3 branch. This change is to primarily allow scaffolding authors to rely on the wsgiref entry point for projects that are meant to work on both Pyramid 1.2.X and 1.3.X.

1.2.4 (2011-12-06)

Features
  • bpython interpreter compatibility in pshell. See the “Command-Line Pyramid” narrative docs chapter for more information.
Bug Fixes
  • Prevent a scaffold rendering from being named site (conflicts with Python internal site.py).
  • Forward compatibility for pyramid_zcml >= 0.8 + zope.configuration >= 3.8.0.
Backward Incompatibilities
  • The pshell command (see “paster pshell”) no longer accepts a --disable-ipython command-line argument. Instead, it accepts a -p or --python-shell argument, which can be any of the values python, ipython or bpython.

1.2.3 (2011-11-20)

  • 1.2.2 was a brownbag, containing a stray intr.py.

1.2.2 (2011-11-20)

Features
  • Backport from master: a mako.directories setting is no longer required to use Mako templates. Rationale: Mako template renderers can be specified using an absolute asset spec. An entire application can be written with such asset specs, requiring no ordered lookup path.
Bug Fixes
  • Backport from master: The pryamid.view.view_config decorator did not accept a match_params predicate argument. See https://github.com/Pylons/pyramid/pull/308
  • Backport fixes from master regarding URL decoding. URL segments are no-longer “double-decoded” during traversal and when encountered in a route subpath (or other star-arg pattern). As a result, a new API named pyramid.traversal.traversal_path_info was added to the system. This function accepts an already-URL-decoded string and returns a tuple of Unicode objects. This API is used internally by Pyramid in all places that pyramid.traversal.traversal_path used to be used. The traversal_path function remains for backwards compatibility, however, and can still be used when a path is encoded. See https://github.com/Pylons/pyramid/issues/349 for more information.
  • Backport fix from master: request.static_url now generates URL-quoted URLs when fed a path argument which contains characters that are unsuitable for URLs. See https://github.com/Pylons/pyramid/issues/349 for more information.
  • Backport from master: fix request.json_body to deal with alternate request charsets.
  • Backport from master: the AuthTktCookieHelper could potentially generate Unicode headers inappropriately when the tokens argument to remember was used. See https://github.com/Pylons/pyramid/pull/314.
  • Backport from master: the AuthTktAuthenticationPolicy did not use a timing-attack-aware string comparator. See https://github.com/Pylons/pyramid/pull/320 for more info.
  • Backport from master: the DummySession in pyramid.testing now generates a new CSRF token if one doesn’t yet exist.
Testing
  • Make tox use WebOb 1.1 for Python 2.5-based systems (WebOb 1.2 is 2.6+).

1.2.1 (2011-09-28)

Features
Bug Fixes
Documentation
  • Fix ..note and ..warning directives to run properly under newer Sphinx.

1.2 (2011-09-12)

Features

1.2b3 (2011-09-11)

Bug Fixes

1.2b2 (2011-09-08)

Bug Fixes
  • The 1.2b1 tarball was a brownbag (particularly for Windows users) because it contained filenames with stray quotation marks in inappropriate places. We depend on setuptools-git to produce release tarballs, and when it was run to produce the 1.2b1 tarball, it didn’t yet cope well with files present in git repositories with high-order characters in their filenames.
Documentation
  • Minor tweaks to the “Introduction” narrative chapter example app and wording.

1.2b1 (2011-09-08)

Bug Fixes
  • Sometimes falling back from territory translations (de_DE) to language translations (de) would not work properly when using a localizer. See https://github.com/Pylons/pyramid/issues/263
  • The static file serving machinery could not serve files that started with a . (dot) character.
  • Static files with high-order (super-ASCII) characters in their names could not be served by a static view. The static file serving machinery inappropriately URL-quoted path segments in filenames when asking for files from the filesystem.
  • Within pyramid.traversal.traversal_path , canonicalize URL segments from UTF-8 to Unicode before checking whether a segment matches literally one of ., the empty string, or .. in case there’s some sneaky way someone might tunnel those strings via UTF-8 that don’t match the literals before decoded.
Documentation
  • Added a “What Makes Pyramid Unique” section to the Introduction narrative chapter.

1.2a6 (2011-09-06)

Bug Fixes
Internal
  • Internalize code previously depended upon as imports from the paste.auth module (futureproof).
  • Replaced use of paste.urlparser.StaticURLParser with a derivative of Chris Rossi’s “happy” static file serving code (futureproof).
  • Fixed test suite; on some systems tests would fail due to indeterminate test run ordering and a double-push-single-pop of a shared test variable.
Behavior Differences
  • An ETag header is no longer set when serving a static file. A Last-Modified header is set instead.
  • Static file serving no longer supports the wsgi.file_wrapper extension.
  • Instead of returning a 403 Forbidden error when a static file is served that cannot be accessed by the Pyramid process’ user due to file permissions, an IOError (or similar) will be raised.
Scaffolds
  • All scaffolds now send the cache_max_age parameter to the add_static_view method.

1.2a5 (2011-09-04)

Bug Fixes
Dependencies
  • The zope.configuration package is no longer a dependency.

1.2a4 (2011-09-02)

Features
  • Support an onerror keyword argument to pyramid.config.Configurator.scan(). This onerror keyword argument is passed to venusian.Scanner.scan() to influence error behavior when an exception is raised during scanning.
  • The request_method predicate argument to pyramid.config.Configurator.add_view and pyramid.config.Configurator.add_route is now permitted to be a tuple of HTTP method names. Previously it was restricted to being a string representing a single HTTP method name.
  • Undeprecated pyramid.traversal.find_model, pyramid.traversal.model_path, pyramid.traversal.model_path_tuple, and pyramid.url.model_url, which were all deprecated in Pyramid 1.0. There’s just not much cost to keeping them around forever as aliases to their renamed resource_* prefixed functions.
  • Undeprecated pyramid.view.bfg_view, which was deprecated in Pyramid 1.0. This is a low-cost alias to pyramid.view.view_config which we’ll just keep around forever.
Dependencies
  • Pyramid now requires Venusian 1.0a1 or better to support the onerror keyword argument to pyramid.config.Configurator.scan.

1.2a3 (2011-08-29)

Bug Fixes
  • Pyramid did not properly generate static URLs using pyramid.url.static_url when passed a caller-package relative path due to a refactoring done in 1.2a1.
  • The settings object emitted a deprecation warning any time __getattr__ was called upon it. However, there are legitimate situations in which __getattr__ is called on arbitrary objects (e.g. hasattr). Now, the settings object only emits the warning upon successful lookup.
Internal
  • Use config.with_package in view_config decorator rather than manufacturing a new renderer helper (cleanup).

1.2a2 (2011-08-27)

Bug Fixes
  • When a renderers= argument is not specified to the Configurator constructor, eagerly register and commit the default renderer set. This permits the overriding of the default renderers, which was broken in 1.2a1 without a commit directly after Configurator construction.
  • Mako rendering exceptions had the wrong value for an error message.
  • An include could not set a root factory successfully because the Configurator constructor unconditionally registered one that would be treated as if it were “the word of the user”.
Features
  • A session factory can now be passed in using the dotted name syntax.

1.2a1 (2011-08-24)

Features

  • The [pshell] section in an ini configuration file now treats a setup key as a dotted name that points to a callable that is passed the bootstrap environment. It can mutate the environment as necessary for great justice.

  • A new configuration setting named pyramid.includes is now available. It is described in the “Environment Variables and .ini Files Settings” narrative documentation chapter.

  • Added a route_prefix argument to the pyramid.config.Configurator.include method. This argument allows you to compose URL dispatch applications together. See the section entitled “Using a Route Prefix to Compose Applications” in the “URL Dispatch” narrative documentation chapter.

  • Added a pyramid.security.NO_PERMISSION_REQUIRED constant for use in permission= statements to view configuration. This constant has a value of the string __no_permission_required__. This string value was previously referred to in documentation; now the documentation uses the constant.

  • Added a decorator-based way to configure a response adapter: pyramid.response.response_adapter. This decorator has the same use as pyramid.config.Configurator.add_response_adapter but it’s declarative.

  • The pyramid.events.BeforeRender event now has an attribute named rendering_val. This can be used to introspect the value returned by a view in a BeforeRender subscriber.

  • New configurator directive: pyramid.config.Configurator.add_tween. This directive adds a “tween”. A “tween” is used to wrap the Pyramid router’s primary request handling function. This is a feature may be used by Pyramid framework extensions, to provide, for example, view timing support and as a convenient place to hang bookkeeping code.

    Tweens are further described in the narrative docs section in the Hooks chapter, named “Registering Tweens”.

  • New paster command paster ptweens, which prints the current “tween” configuration for an application. See the section entitled “Displaying Tweens” in the Command-Line Pyramid chapter of the narrative documentation for more info.

  • The Pyramid debug logger now uses the standard logging configuration (usually set up by Paste as part of startup). This means that output from e.g. debug_notfound, debug_authorization, etc. will go to the normal logging channels. The logger name of the debug logger will be the package name of the caller of the Configurator’s constructor.

  • A new attribute is available on request objects: exc_info. Its value will be None until an exception is caught by the Pyramid router, after which it will be the result of sys.exc_info().

  • pyramid.testing.DummyRequest now implements the add_finished_callback and add_response_callback methods.

  • New methods of the pyramid.config.Configurator class: set_authentication_policy and set_authorization_policy. These are meant to be consumed mostly by add-on authors.

  • New Configurator method: set_root_factory.

  • Pyramid no longer eagerly commits some default configuration statements at Configurator construction time, which permits values passed in as constructor arguments (e.g. authentication_policy and authorization_policy) to override the same settings obtained via an “include”.

  • Better Mako rendering exceptions via pyramid.mako_templating.MakoRenderingException

  • New request methods: current_route_url, current_route_path, and static_path.

  • New functions in pyramid.url: current_route_path and static_path.

  • The pyramid.request.Request.static_url API (and its brethren pyramid.request.Request.static_path, pyramid.url.static_url, and pyramid.url.static_path) now accept an asbolute filename as a “path” argument. This will generate a URL to an asset as long as the filename is in a directory which was previously registered as a static view. Previously, trying to generate a URL to an asset using an absolute file path would raise a ValueError.

  • The RemoteUserAuthenticationPolicy ``, ``AuthTktAuthenticationPolicy, and SessionAuthenticationPolicy constructors now accept an additional keyword argument named debug. By default, this keyword argument is False. When it is True, debug information will be sent to the Pyramid debug logger (usually on stderr) when the authenticated_userid or effective_principals method is called on any of these policies. The output produced can be useful when trying to diagnose authentication-related problems.

  • New view predicate: match_param. Example: a view added via config.add_view(aview, match_param='action=edit') will be called only when the request.matchdict has a value inside it named action with a value of edit.

Internal
  • The Pyramid “exception view” machinery is now implemented as a “tween” (pyramid.tweens.excview_tween_factory).
  • WSGIHTTPException (HTTPFound, HTTPNotFound, etc) now has a new API named “prepare” which renders the body and content type when it is provided with a WSGI environ. Required for debug toolbar.
  • Once __call__ or prepare is called on a WSGIHTTPException, the body will be set, and subsequent calls to __call__ will always return the same body. Delete the body attribute to rerender the exception body.
  • Previously the pyramid.events.BeforeRender event wrapped a dictionary (it addressed it as its _system attribute). Now it is a dictionary (it inherits from dict), and it’s the value that is passed to templates as a top-level dictionary.
  • The route_url, route_path, resource_url, static_url, and current_route_url functions in the pyramid.url package now delegate to a method on the request they’ve been passed, instead of the other way around. The pyramid.request.Request object now inherits from a mixin named pyramid.url.URLMethodsMixin to make this possible, and all url/path generation logic is embedded in this mixin.
  • Refactor pyramid.config into a package.
  • Removed the _set_security_policies method of the Configurator.
  • Moved the StaticURLInfo class from pyramid.static to pyramid.config.views.
  • Move the Settings class from pyramid.settings to pyramid.config.settings.
  • Move the OverrideProvider, PackageOverrides, DirectoryOverride, and FileOverride classes from pyramid.asset to pyramid.config.assets.
Deprecations
  • All Pyramid-related deployment settings (e.g. debug_all, debug_notfound) are now meant to be prefixed with the prefix pyramid.. For example: debug_all -> pyramid.debug_all. The old non-prefixed settings will continue to work indefinitely but supplying them may eventually print a deprecation warning. All scaffolds and tutorials have been changed to use prefixed settings.
  • The settings dictionary now raises a deprecation warning when you attempt to access its values via __getattr__ instead of via __getitem__.
Backwards Incompatibilities

  • If a string is passed as the debug_logger parameter to a Configurator, that string is considered to be the name of a global Python logger rather than a dotted name to an instance of a logger.

  • The pyramid.config.Configurator.include method now accepts only a single callable argument (a sequence of callables used to be permitted). If you are passing more than one callable to pyramid.config.Configurator.include, it will break. You now must now instead make a separate call to the method for each callable. This change was introduced to support the route_prefix feature of include.

  • It may be necessary to more strictly order configuration route and view statements when using an “autocommitting” Configurator. In the past, it was possible to add a view which named a route name before adding a route with that name when you used an autocommitting configurator. For example:

    config = Configurator(autocommit=True)
config.add_view('my.pkg.someview', route_name='foo')
config.add_route('foo', '/foo')

    The above will raise an exception when the view attempts to add itself. Now you must add the route before adding the view:

    config = Configurator(autocommit=True)
config.add_route('foo', '/foo')
config.add_view('my.pkg.someview', route_name='foo')

    This won’t effect “normal” users, only people who have legacy BFG codebases that used an autommitting configurator and possibly tests that use the configurator API (the configurator returned by pyramid.testing.setUp is an autocommitting configurator). The right way to get around this is to use a non-autocommitting configurator (the default), which does not have these directive ordering requirements.

  • The pyramid.config.Configurator.add_route directive no longer returns a route object. This change was required to make route vs. view configuration processing work properly.

Documentation
  • Narrative and API documentation which used the route_url, route_path, resource_url, static_url, and current_route_url functions in the pyramid.url package have now been changed to use eponymous methods of the request instead.
  • Added a section entitled “Using a Route Prefix to Compose Applications” to the “URL Dispatch” narrative documentation chapter.
  • Added a new module to the API docs: pyramid.tweens.
  • Added a “Registering Tweens” section to the “Hooks” narrative chapter.
  • Added a “Displaying Tweens” section to the “Command-Line Pyramid” narrative chapter.
  • Added documentation for the pyramid.tweens and pyramid.includes configuration settings to the “Environment Variables and .ini Files Settings” chapter.
  • Added a Logging chapter to the narrative docs (based on the Pylons logging docs, thanks Phil).
  • Added a Paste chapter to the narrative docs (moved content from the Project chapter).
  • Added the pyramid.interfaces.IDict interface representing the methods of a dictionary, for documentation purposes only (IMultiDict and IBeforeRender inherit from it).
  • All tutorials now use - The route_url, route_path, resource_url, static_url, and current_route_url methods of the request rather than the function variants imported from pyramid.url.
  • The ZODB wiki tutorial now uses the pyramid_zodbconn package rather than the repoze.zodbconn package to provide ZODB integration.
Dependency Changes
  • Pyramid now relies on PasteScript >= 1.7.4. This version contains a feature important for allowing flexible logging configuration.
Scaffolds
  • All scaffolds now use the pyramid_tm package rather than the repoze.tm2 middleware to manage transaction management.
  • The ZODB scaffold now uses the pyramid_zodbconn package rather than the repoze.zodbconn package to provide ZODB integration.
  • All scaffolds now use the pyramid_debugtoolbar package rather than the WebError package to provide interactive debugging features.
  • Projects created via a scaffold no longer depend on the WebError package at all; configuration in the production.ini file which used to require its error_catcher middleware has been removed. Configuring error catching / email sending is now the domain of the pyramid_exclog package (see https://docs.pylonsproject.org/projects/pyramid_exclog/dev/).
Bug Fixes

1.1 (2011-07-22)

Features
  • Added the pyramid.renderers.null_renderer object as an API. The null renderer is an object that can be used in advanced integration cases as input to the view configuration renderer= argument. When the null renderer is used as a view renderer argument, Pyramid avoids converting the view callable result into a Response object. This is useful if you want to reuse the view configuration and lookup machinery outside the context of its use by the Pyramid router. This feature was added for consumption by the pyramid_rpc package, which uses view configuration and lookup outside the context of a router in exactly this way. pyramid_rpc has been broken under 1.1 since 1.1b1; adding it allows us to make it work again.
  • Change all scaffolding templates that point to docs.pylonsproject.org to use /projects/pyramid/current rather than /projects/pyramid/dev.
Internals
  • Remove compat code that served only the purpose of providing backwards compatibility with Python 2.4.
  • Add a deprecation warning for non-API function pyramid.renderers.renderer_from_name which has seen use in the wild.
  • Add a clone method to pyramid.renderers.RendererHelper for use by the pyramid.view.view_config decorator.
Documentation
  • Fixed two typos in wiki2 (SQLA + URL Dispatch) tutorial.
  • Reordered chapters in narrative section for better new user friendliness.
  • Added more indexing markers to sections in documentation.

1.1b4 (2011-07-18)

Documentation
  • Added a section entitled “Writing a Script” to the “Command-Line Pyramid” chapter.
Backwards Incompatibilities
  • We added the pyramid.scripting.make_request API too hastily in 1.1b3. It has been removed. Sorry for any inconvenience. Use the pyramid.request.Request.blank API instead.
Features
  • The paster pshell, paster pviews, and paster proutes commands each now under the hood uses pyramid.paster.bootstrap, which makes it possible to supply an .ini file without naming the “right” section in the file that points at the actual Pyramid application. Instead, you can generally just run paster {pshell|proutes|pviews} development.ini and it will do mostly the right thing.
Bug Fixes
  • Omit custom environ variables when rendering a custom exception template in pyramid.httpexceptions.WSGIHTTPException._set_default_attrs; stringifying thse may trigger code that should not be executed; see https://github.com/Pylons/pyramid/issues/239

1.1b3 (2011-07-15)

Features

  • Fix corner case to ease semifunctional testing of views: create a new rendererinfo to clear out old registry on a rescan. See https://github.com/Pylons/pyramid/pull/234.

  • New API class: pyramid.static.static_view. This supersedes the deprecated pyramid.view.static class. pyramid.static.static_view by default serves up documents as the result of the request’s path_info, attribute rather than it’s subpath attribute (the inverse was true of pyramid.view.static, and still is). pyramid.static.static_view exposes a use_subpath flag for use when you want the static view to behave like the older deprecated version.

  • A new API function pyramid.paster.bootstrap has been added to make writing scripts that bootstrap a Pyramid environment easier, e.g.:

    from pyramid.paster import bootstrap
info = bootstrap('/path/to/my/development.ini')
request = info['request']
print request.route_url('myroute')

  • A new API function pyramid.scripting.prepare has been added. It is a lower-level analogue of pyramid.paster.boostrap that accepts a request and a registry instead of a config file argument, and is used for the same purpose:

    from pyramid.scripting import prepare
info = prepare(registry=myregistry)
request = info['request']
print request.route_url('myroute')

  • A new API function pyramid.scripting.make_request has been added. The resulting request will have a registry attribute. It is meant to be used in conjunction with pyramid.scripting.prepare and/or pyramid.paster.bootstrap (both of which accept a request as an argument):

    from pyramid.scripting import make_request
request = make_request('/')

  • New API attribute pyramid.config.global_registries is an iterable object that contains references to every Pyramid registry loaded into the current process via pyramid.config.Configurator.make_app. It also has a last attribute containing the last registry loaded. This is used by the scripting machinery, and is available for introspection.

Deprecations
  • The pyramid.view.static class has been deprecated in favor of the newer pyramid.static.static_view class. A deprecation warning is raised when it is used. You should replace it with a reference to pyramid.static.static_view with the use_subpath=True argument.
Bug Fixes
  • Without a mo-file loaded for the combination of domain/locale, pyramid.i18n.Localizer.pluralize run using that domain/locale combination raised an inscrutable “translations object has no attr ‘plural’” error. Now, instead it “works” (it uses a germanic pluralization by default). It’s nonsensical to try to pluralize something without translations for that locale/domain available, but this behavior matches the behavior of pyramid.i18n.Localizer.translate so it’s at least consistent; see https://github.com/Pylons/pyramid/issues/235.

1.1b2 (2011-07-13)

Features
  • New environment setting PYRAMID_PREVENT_HTTP_CACHE and new configuration file value prevent_http_cache. These are synomymous and allow you to prevent HTTP cache headers from being set by Pyramid’s http_cache machinery globally in a process. see the “Influencing HTTP Caching” section of the “View Configuration” narrative chapter and the detailed documentation for this setting in the “Environment Variables and Configuration Settings” narrative chapter.
Behavior Changes
  • Previously, If a BeforeRender event subscriber added a value via the __setitem__ or update methods of the event object with a key that already existed in the renderer globals dictionary, a KeyError was raised. With the deprecation of the “add_renderer_globals” feature of the configurator, there was no way to override an existing value in the renderer globals dictionary that already existed. Now, the event object will overwrite an older value that is already in the globals dictionary when its __setitem__ or update is called (as well as the new setdefault method), just like a plain old dictionary. As a result, for maximum interoperability with other third-party subscribers, if you write an event subscriber meant to be used as a BeforeRender subscriber, your subscriber code will now need to (using .get or __contains__ of the event object) ensure no value already exists in the renderer globals dictionary before setting an overriding value.
Bug Fixes
  • The Configurator.add_route method allowed two routes with the same route to be added without an intermediate config.commit(). If you now receive a ConfigurationError at startup time that appears to be add_route related, you’ll need to either a) ensure that all of your route names are unique or b) call config.commit() before adding a second route with the name of a previously added name or c) use a Configurator that works in autocommit mode.
  • The pyramid_routesalchemy and pyramid_alchemy scaffolds inappropriately used DBSession.rollback() instead of transaction.abort() in one place.
  • We now clear request.response before we invoke an exception view; an exception view will be working with a request.response that has not been touched by any code prior to the exception.
  • Views associated with routes with spaces in the route name may not have been looked up correctly when using Pyramid with zope.interface 3.6.4 and better. See https://github.com/Pylons/pyramid/issues/232.
Documentation
  • Wiki2 (SQLAlchemy + URL Dispatch) tutorial models.initialize_sql didn’t match the pyramid_routesalchemy scaffold function of the same name; it didn’t get synchronized when it was changed in the scaffold.
  • New documentation section in View Configuration narrative chapter: “Influencing HTTP Caching”.

1.1b1 (2011-07-10)

Features

  • It is now possible to invoke paster pshell even if the paste ini file section name pointed to in its argument is not actually a Pyramid WSGI application. The shell will work in a degraded mode, and will warn the user. See “The Interactive Shell” in the “Creating a Pyramid Project” narrative documentation section.

  • paster pshell now offers more built-in global variables by default (including app and settings). See “The Interactive Shell” in the “Creating a Pyramid Project” narrative documentation section.

  • It is now possible to add a [pshell] section to your application’s .ini configuration file, which influences the global names available to a pshell session. See “Extending the Shell” in the “Creating a Pyramid Project” narrative documentation chapter.

  • The config.scan method has grown a **kw argument. kw argument represents a set of keyword arguments to pass to the Venusian Scanner object created by Pyramid. (See the Venusian documentation for more information about Scanner).

  • New request property: json_body. This property will return the JSON-decoded variant of the request body. If the request body is not well-formed JSON, this property will raise an exception.

  • A new value http_cache can be used as a view configuration parameter.

    When you supply an http_cache value to a view configuration, the Expires and Cache-Control headers of a response generated by the associated view callable are modified. The value for http_cache may be one of the following:

    • A nonzero integer. If it’s a nonzero integer, it’s treated as a number of seconds. This number of seconds will be used to compute the Expires header and the Cache-Control: max-age parameter of responses to requests which call this view. For example: http_cache=3600 instructs the requesting browser to ‘cache this response for an hour, please’.
    • A datetime.timedelta instance. If it’s a datetime.timedelta instance, it will be converted into a number of seconds, and that number of seconds will be used to compute the Expires header and the Cache-Control: max-age parameter of responses to requests which call this view. For example: http_cache=datetime.timedelta(days=1) instructs the requesting browser to ‘cache this response for a day, please’.
    • Zero (0). If the value is zero, the Cache-Control and Expires headers present in all responses from this view will be composed such that client browser cache (and any intermediate caches) are instructed to never cache the response.
    • A two-tuple. If it’s a two tuple (e.g. http_cache=(1, {'public':True})), the first value in the tuple may be a nonzero integer or a datetime.timedelta instance; in either case this value will be used as the number of seconds to cache the response. The second value in the tuple must be a dictionary. The values present in the dictionary will be used as input to the Cache-Control response header. For example: http_cache=(3600, {'public':True}) means ‘cache for an hour, and add public to the Cache-Control header of the response’. All keys and values supported by the webob.cachecontrol.CacheControl interface may be added to the dictionary. Supplying {'public':True} is equivalent to calling response.cache_control.public = True.

    Providing a non-tuple value as http_cache is equivalent to calling response.cache_expires(value) within your view’s body.

    Providing a two-tuple value as http_cache is equivalent to calling response.cache_expires(value[0], **value[1]) within your view’s body.

    If you wish to avoid influencing, the Expires header, and instead wish to only influence Cache-Control headers, pass a tuple as http_cache with the first element of None, e.g.: (None, {'public':True}).

Bug Fixes
  • Framework wrappers of the original view (such as http_cached and so on) relied on being able to trust that the response they were receiving was an IResponse. It wasn’t always, because the response was resolved by the router instead of early in the view wrapping process. This has been fixed.
Documentation
  • Added a section in the “Webob” chapter named “Dealing With A JSON-Encoded Request Body” (usage of request.json_body).
Behavior Changes
  • The paster pshell, paster proutes, and paster pviews commands now take a single argument in the form /path/to/config.ini#sectionname rather than the previous 2-argument spelling /path/to/config.ini sectionname. #sectionname may be omitted, in which case #main is assumed.

1.1a4 (2011-07-01)

Bug Fixes

  • pyramid.testing.DummyRequest now raises deprecation warnings when attributes deprecated for pyramid.request.Request are accessed (like response_content_type). This is for the benefit of folks running unit tests which use DummyRequest instead of a “real” request, so they know things are deprecated without necessarily needing a functional test suite.

  • The pyramid.events.subscriber directive behaved contrary to the documentation when passed more than one interface object to its constructor. For example, when the following listener was registered:

    @subscriber(IFoo, IBar)
def expects_ifoo_events_and_ibar_events(event):
    print event

    The Events chapter docs claimed that the listener would be registered and listening for both IFoo and IBar events. Instead, it registered an “object event” subscriber which would only be called if an IObjectEvent was emitted where the object interface was IFoo and the event interface was IBar.

    The behavior now matches the documentation. If you were relying on the buggy behavior of the 1.0 subscriber directive in order to register an object event subscriber, you must now pass a sequence to indicate you’d like to register a subscriber for an object event. e.g.:

    @subscriber([IFoo, IBar])
def expects_object_event(object, event):
    print object, event
Features
  • Add JSONP renderer (see “JSONP renderer” in the Renderers chapter of the documentation).
Deprecations
  • Deprecated the set_renderer_globals_factory method of the Configurator and the renderer_globals Configurator constructor parameter.
Documentation
  • The Wiki and Wiki2 tutorial “Tests” chapters each had two bugs: neither did told the user to depend on WebTest, and 2 tests failed in each as the result of changes to Pyramid itself. These issues have been fixed.
  • Move 1.0.X CHANGES.txt entries to HISTORY.txt.

1.1a3 (2011-06-26)

Features
Bug fixes
  • Pyramid would raise an AttributeError in the Configurator when attempting to set a __text__ attribute on a custom predicate that was actually a classmethod. See https://github.com/Pylons/pyramid/pull/217 .
  • Accessing or setting deprecated response_* attrs on request (e.g. response_content_type) now issues a deprecation warning at access time rather than at rendering time.

1.1a2 (2011-06-22)

Bug Fixes
  • 1.1a1 broke Akhet by not providing a backwards compatibility import shim for pyramid.paster.PyramidTemplate. Now one has been added, although a deprecation warning is emitted when Akhet imports it.
  • If multiple specs were provided in a single call to config.add_translation_dirs, the directories were inserted into the beginning of the directory list in the wrong order: they were inserted in the reverse of the order they were provided in the *specs list (items later in the list were added before ones earlier in the list). This is now fixed.
Backwards Incompatibilities

  • The pyramid Router attempted to set a value into the key environ['repoze.bfg.message'] when it caught a view-related exception for backwards compatibility with applications written for repoze.bfg during error handling. It did this by using code that looked like so:

    # "why" is an exception object
try:
    msg = why[0]
except:
    msg = ''

environ['repoze.bfg.message'] = msg

    Use of the value environ['repoze.bfg.message'] was docs-deprecated in Pyramid 1.0. Our standing policy is to not remove features after a deprecation for two full major releases, so this code was originally slated to be removed in Pyramid 1.2. However, computing the repoze.bfg.message value was the source of at least one bug found in the wild (https://github.com/Pylons/pyramid/issues/199), and there isn’t a foolproof way to both preserve backwards compatibility and to fix the bug. Therefore, the code which sets the value has been removed in this release. Code in exception views which relies on this value’s presence in the environment should now use the exception attribute of the request (e.g. request.exception[0]) to retrieve the message instead of relying on request.environ['repoze.bfg.message'].

1.1a1 (2011-06-20)

Documentation
  • The term “template” used to refer to both “paster templates” and “rendered templates” (templates created by a rendering engine. i.e. Mako, Chameleon, Jinja, etc.). “Paster templates” will now be refered to as “scaffolds”, whereas the name for “rendered templates” will remain as “templates.”
  • The wiki (ZODB+Traversal) tutorial was updated slightly.
  • The wiki2 (SQLA+URL Dispatch) tutorial was updated slightly.
  • Make pyramid.interfaces.IAuthenticationPolicy and pyramid.interfaces.IAuthorizationPolicy public interfaces, and refer to them within the pyramid.authentication and pyramid.authorization API docs.
  • Render the function definitions for each exposed interface in pyramid.interfaces.
  • Add missing docs reference to pyramid.config.Configurator.set_view_mapper and refer to it within Hooks chapter section named “Using a View Mapper”.
  • Added section to the “Environment Variables and .ini File Settings” chapter in the narrative documentation section entitled “Adding a Custom Setting”.
  • Added documentation for a “multidict” (e.g. the API of request.POST) as interface API documentation.
  • Added a section to the “URL Dispatch” narrative chapter regarding the new “static” route feature.
  • Added “What’s New in Pyramid 1.1” to HTML rendering of documentation.
  • Added API docs for pyramid.authentication.SessionAuthenticationPolicy.
  • Added API docs for pyramid.httpexceptions.exception_response.
  • Added “HTTP Exceptions” section to Views narrative chapter including a description of pyramid.httpexceptions.exception_response.
Features

  • Add support for language fallbacks: when trying to translate for a specific territory (such as en_GB) fall back to translations for the language (ie en). This brings the translation behaviour in line with GNU gettext and fixes partially translated texts when using C extensions.

  • New authentication policy: pyramid.authentication.SessionAuthenticationPolicy, which uses a session to store credentials.

  • Accessing the response attribute of a pyramid.request.Request object (e.g. request.response within a view) now produces a new pyramid.response.Response object. This feature is meant to be used mainly when a view configured with a renderer needs to set response attributes: all renderers will use the Response object implied by request.response as the response object returned to the router.

    request.response can also be used by code in a view that does not use a renderer, however the response object that is produced by request.response must be returned when a renderer is not in play (it is not a “global” response).

  • Integers and longs passed as elements to pyramid.url.resource_url or pyramid.request.Request.resource_url e.g. resource_url(context, request, 1, 2) (1 and 2 are the elements) will now be converted implicitly to strings in the result. Previously passing integers or longs as elements would cause a TypeError.

  • pyramid_alchemy paster template now uses query.get rather than query.filter_by to take better advantage of identity map caching.

  • pyramid_alchemy paster template now has unit tests.

  • Added pyramid.i18n.make_localizer API (broken out from get_localizer guts).

  • An exception raised by a NewRequest event subscriber can now be caught by an exception view.

  • It is now possible to get information about why Pyramid raised a Forbidden exception from within an exception view. The ACLDenied object returned by the permits method of each stock authorization policy (pyramid.interfaces.IAuthorizationPolicy.permits) is now attached to the Forbidden exception as its result attribute. Therefore, if you’ve created a Forbidden exception view, you can see the ACE, ACL, permission, and principals involved in the request as eg. context.result.permission, context.result.acl, etc within the logic of the Forbidden exception view.

  • Don’t explicitly prevent the timeout from being lower than the reissue_time when setting up an AuthTktAuthenticationPolicy (previously such a configuration would raise a ValueError, now it’s allowed, although typically nonsensical). Allowing the nonsensical configuration made the code more understandable and required fewer tests.

  • A new paster command named paster pviews was added. This command prints a summary of potentially matching views for a given path. See the section entitled “Displaying Matching Views for a Given URL” in the “View Configuration” chapter of the narrative documentation for more information.

  • The add_route method of the Configurator now accepts a static argument. If this argument is True, the added route will never be considered for matching when a request is handled. Instead, it will only be useful for URL generation via route_url and route_path. See the section entitled “Static Routes” in the URL Dispatch narrative chapter for more information.

  • A default exception view for the context pyramid.interfaces.IExceptionResponse is now registered by default. This means that an instance of any exception response class imported from pyramid.httpexceptions (such as HTTPFound) can now be raised from within view code; when raised, this exception view will render the exception to a response.

  • A function named pyramid.httpexceptions.exception_response is a shortcut that can be used to create HTTP exception response objects using an HTTP integer status code.

  • The Configurator now accepts an additional keyword argument named exceptionresponse_view. By default, this argument is populated with a default exception view function that will be used when a response is raised as an exception. When None is passed for this value, an exception view for responses will not be registered. Passing None returns the behavior of raising an HTTP exception to that of Pyramid 1.0 (the exception will propagate to middleware and to the WSGI server).

  • The pyramid.request.Request class now has a ResponseClass interface which points at pyramid.response.Response.

  • The pyramid.response.Response class now has a RequestClass interface which points at pyramid.request.Request.

  • It is now possible to return an arbitrary object from a Pyramid view callable even if a renderer is not used, as long as a suitable adapter to pyramid.interfaces.IResponse is registered for the type of the returned object by using the new pyramid.config.Configurator.add_response_adapter API. See the section in the Hooks chapter of the documentation entitled “Changing How Pyramid Treats View Responses”.

  • The Pyramid router will now, by default, call the __call__ method of WebOb response objects when returning a WSGI response. This means that, among other things, the conditional_response feature of WebOb response objects will now behave properly.

  • New method named pyramid.request.Request.is_response. This method should be used instead of the pyramid.view.is_response function, which has been deprecated.

Bug Fixes
  • URL pattern markers used in URL dispatch are permitted to specify a custom regex. For example, the pattern /{foo:\d+} means to match /12345 (foo==12345 in the match dictionary) but not /abc. However, custom regexes in a pattern marker which used squiggly brackets did not work. For example, /{foo:\d{4}} would fail to match /1234 and /{foo:\d{1,2}} would fail to match /1 or /11. One level of inner squiggly brackets is now recognized so that the prior two patterns given as examples now work. See also https://github.com/Pylons/pyramid/issues/#issue/123.
  • Don’t send port numbers along with domain information in cookies set by AuthTktCookieHelper (see https://github.com/Pylons/pyramid/issues/131).
  • pyramid.url.route_path (and the shortcut pyramid.request.Request.route_url method) now include the WSGI SCRIPT_NAME at the front of the path if it is not empty (see https://github.com/Pylons/pyramid/issues/135).
  • pyramid.testing.DummyRequest now has a script_name attribute (the empty string).
  • Don’t quote :@&+$, symbols in *elements passed to pyramid.url.route_url or pyramid.url.resource_url (see https://github.com/Pylons/pyramid/issues#issue/141).
  • Include SCRIPT_NAME in redirects issued by pyramid.view.append_slash_notfound_view (see https://github.com/Pylons/pyramid/issues#issue/149).
  • Static views registered with config.add_static_view which also included a permission keyword argument would not work as expected, because add_static_view also registered a route factory internally. Because a route factory was registered internally, the context checked by the Pyramid permission machinery never had an ACL. add_static_view no longer registers a route with a factory, so the default root factory will be used.
  • config.add_static_view now passes extra keyword arguments it receives to config.add_route (calling add_static_view is mostly logically equivalent to adding a view of the type pyramid.static.static_view hooked up to a route with a subpath). This makes it possible to pass e.g., factory= to add_static_view to protect a particular static view with a custom ACL.
  • testing.DummyRequest used the wrong registry (the global registry) as self.registry if a dummy request was created before testing.setUp was executed (testing.setUp pushes a local registry onto the threadlocal stack). Fixed by implementing registry as a property for DummyRequest instead of eagerly assigning an attribute. See also https://github.com/Pylons/pyramid/issues/165
  • When visiting a URL that represented a static view which resolved to a subdirectory, the index.html of that subdirectory would not be served properly. Instead, a redirect to /subdir would be issued. This has been fixed, and now visiting a subdirectory that contains an index.html within a static view returns the index.html properly. See also https://github.com/Pylons/pyramid/issues/67.
  • Redirects issued by a static view did not take into account any existing SCRIPT_NAME (such as one set by a url mapping composite). Now they do.
  • The pyramid.wsgi.wsgiapp2 decorator did not take into account the SCRIPT_NAME in the origin request.
  • The pyramid.wsgi.wsgiapp2 decorator effectively only worked when it decorated a view found via traversal; it ignored the PATH_INFO that was part of a url-dispatch-matched view.
Deprecations

  • Deprecated all assignments to request.response_* attributes (for example request.response_content_type = 'foo' is now deprecated). Assignments and mutations of assignable request attributes that were considered by the framework for response influence are now deprecated: response_content_type, response_headerlist, response_status, response_charset, and response_cache_for. Instead of assigning these to the request object for later detection by the rendering machinery, users should use the appropriate API of the Response object created by accessing request.response (e.g. code which does request.response_content_type = 'abc' should be changed to request.response.content_type = 'abc').

  • Passing view-related parameters to pyramid.config.Configurator.add_route is now deprecated. Previously, a view was permitted to be connected to a route using a set of view* parameters passed to the add_route method of the Configurator. This was a shorthand which replaced the need to perform a subsequent call to add_view. For example, it was valid (and often recommended) to do:

    config.add_route('home', '/', view='mypackage.views.myview',
                  view_renderer='some/renderer.pt')

    Passing view* arguments to add_route is now deprecated in favor of connecting a view to a predefined route via Configurator.add_view using the route’s route_name parameter. As a result, the above example should now be spelled:

    config.add_route('home', '/')
config.add_view('mypackage.views.myview', route_name='home')
                renderer='some/renderer.pt')

    This deprecation was done to reduce confusion observed in IRC, as well as to (eventually) reduce documentation burden (see also https://github.com/Pylons/pyramid/issues/164). A deprecation warning is now issued when any view-related parameter is passed to Configurator.add_route.

  • Passing an environ dictionary to the __call__ method of a “traverser” (e.g. an object that implements pyramid.interfaces.ITraverser such as an instance of pyramid.traversal.ResourceTreeTraverser) as its request argument now causes a deprecation warning to be emitted. Consumer code should pass a request object instead. The fact that passing an environ dict is permitted has been documentation-deprecated since repoze.bfg 1.1, and this capability will be removed entirely in a future version.

  • The following (undocumented, dictionary-like) methods of the pyramid.request.Request object have been deprecated: __contains__, __delitem__, __getitem__, __iter__, __setitem__, get, has_key, items, iteritems, itervalues, keys, pop, popitem, setdefault, update, and values. Usage of any of these methods will cause a deprecation warning to be emitted. These methods were added for internal compatibility in repoze.bfg 1.1 (code that currently expects a request object expected an environ object in BFG 1.0 and before). In a future version, these methods will be removed entirely.

  • Deprecated pyramid.view.is_response function in favor of (newly-added) pyramid.request.Request.is_response method. Determining if an object is truly a valid response object now requires access to the registry, which is only easily available as a request attribute. The pyramid.view.is_response function will still work until it is removed, but now may return an incorrect answer under some (very uncommon) circumstances.

Behavior Changes
  • The default Mako renderer is now configured to escape all HTML in expression tags. This is intended to help prevent XSS attacks caused by rendering unsanitized input from users. To revert this behavior in user’s templates, they need to filter the expression through the ‘n’ filter. For example, ${ myhtml | n }. See https://github.com/Pylons/pyramid/issues/193.
  • A custom request factory is now required to return a request object that has a response attribute (or “reified”/lazy property) if they the request is meant to be used in a view that uses a renderer. This response attribute should be an instance of the class pyramid.response.Response.
  • The JSON and string renderer factories now assign to request.response.content_type rather than request.response_content_type.
  • Each built-in renderer factory now determines whether it should change the content type of the response by comparing the response’s content type against the response’s default content type; if the content type is the default content type (usually text/html), the renderer changes the content type (to application/json or text/plain for JSON and string renderers respectively).
  • The pyramid.wsgi.wsgiapp2 now uses a slightly different method of figuring out how to “fix” SCRIPT_NAME and PATH_INFO for the downstream application. As a result, those values may differ slightly from the perspective of the downstream application (for example, SCRIPT_NAME will now never possess a trailing slash).
  • Previously, pyramid.request.Request inherited from webob.request.Request and implemented __getattr__, __setattr__ and __delattr__ itself in order to overidde “adhoc attr” WebOb behavior where attributes of the request are stored in the environ. Now, pyramid.request.Request object inherits from (the more recent) webob.request.BaseRequest instead of webob.request.Request, which provides the same behavior. pyramid.request.Request no longer implements its own __getattr__, __setattr__ or __delattr__ as a result.
  • pyramid.response.Response is now a subclass of webob.response.Response (in order to directly implement the pyramid.interfaces.IResponse interface).
  • The “exception response” objects importable from pyramid.httpexceptions (e.g. HTTPNotFound) are no longer just import aliases for classes that actually live in webob.exc. Instead, we’ve defined our own exception classes within the module that mirror and emulate the webob.exc exception response objects almost entirely. See the “Design Defense” doc section named “Pyramid Uses its Own HTTP Exception Classes” for more information.
Backwards Incompatibilities
  • Pyramid no longer supports Python 2.4. Python 2.5 or better is required to run Pyramid 1.1+.
  • The Pyramid router now, by default, expects response objects returned from view callables to implement the pyramid.interfaces.IResponse interface. Unlike the Pyramid 1.0 version of this interface, objects which implement IResponse now must define a __call__ method that accepts environ and start_response, and which returns an app_iter iterable, among other things. Previously, it was possible to return any object which had the three WebOb app_iter, headerlist, and status attributes as a response, so this is a backwards incompatibility. It is possible to get backwards compatibility back by registering an adapter to IResponse from the type of object you’re now returning from view callables. See the section in the Hooks chapter of the documentation entitled “Changing How Pyramid Treats View Responses”.
  • The pyramid.interfaces.IResponse interface is now much more extensive. Previously it defined only app_iter, status and headerlist; now it is basically intended to directly mirror the webob.Response API, which has many methods and attributes.
  • The pyramid.httpexceptions classes named HTTPFound, HTTPMultipleChoices, HTTPMovedPermanently, HTTPSeeOther, HTTPUseProxy, and HTTPTemporaryRedirect now accept location as their first positional argument rather than detail. This means that you can do, e.g. return pyramid.httpexceptions.HTTPFound('http://foo') rather than return pyramid.httpexceptions.HTTPFound(location='http//foo') (the latter will of course continue to work).
Dependencies
  • Pyramid now depends on WebOb >= 1.0.2 as tests depend on the bugfix in that release: “Fix handling of WSGI environs with missing SCRIPT_NAME”. (Note that in reality, everyone should probably be using 1.0.4 or better though, as WebOb 1.0.2 and 1.0.3 were effectively brownbag releases.)

1.0 (2011-01-30)

Documentation
  • Fixed bug in ZODB Wiki tutorial (missing dependency on docutils in “models” step within setup.py).
  • Removed API documentation for pyramid.testing APIs named registerDummySecurityPolicy, registerResources, registerModels, registerEventListener, registerTemplateRenderer, registerDummyRenderer, registerView, registerUtility, registerAdapter, registerSubscriber, registerRoute, and registerSettings.
  • Moved “Using ZODB With ZEO” and “Using repoze.catalog Within Pyramid” tutorials out of core documentation and into the Pyramid Tutorials site (http://docs.pylonsproject.org/projects/pyramid_tutorials/dev/).
  • Changed “Cleaning up After a Request” section in the URL Dispatch chapter to use request.add_finished_callback instead of jamming an object with a __del__ into the WSGI environment.
  • Remove duplication of add_route API documentation from URL Dispatch narrative chapter.
  • Remove duplication of API and narrative documentation in pyramid.view.view_config API docs by pointing to pyramid.config.add_view documentation and narrative chapter documentation.
  • Removed some API documentation duplicated in narrative portions of documentation
  • Removed “Overall Flow of Authentication” from SQLAlchemy + URL Dispatch wiki tutorial due to print space concerns (moved to Pyramid Tutorials site).
Bug Fixes
  • Deprecated-since-BFG-1.2 APIs from pyramid.testing now properly emit deprecation warnings.
  • Added egg:repoze.retry#retry middleware to the WSGI pipeline in ZODB templates (retry ZODB conflict errors which occur in normal operations).
  • Removed duplicate implementations of is_response. Two competing implementations existed: one in pyramid.config and one in pyramid.view. Now the one defined in pyramid.view is used internally by pyramid.config and continues to be advertised as an API.

1.0b3 (2011-01-28)

Bug Fixes
  • Use &copy; instead of copyright symbol in paster templates / tutorial templates for the benefit of folks who cutnpaste and save to a non-UTF8 format.
  • pyramid.view.append_slash_notfound_view now preserves GET query parameters across redirects.
Documentation
  • Beef up documentation related to set_default_permission: explicitly mention that default permissions also protect exception views.
  • Paster templates and tutorials now use spaces instead of tabs in their HTML templates.

1.0b2 (2011-01-24)

Bug Fixes
  • The production.ini generated by all paster templates now have an effective logging level of WARN, which prevents e.g. SQLAlchemy statement logging and other inappropriate output.
  • The production.ini of the pyramid_routesalchemy and pyramid_alchemy paster templates did not have a sqlalchemy logger section, preventing paster serve production.ini from working.
  • The pyramid_routesalchemy and pyramid_alchemy paster templates used the {{package}} variable in a place where it should have used the {{project}} variable, causing applications created with uppercase letters e.g. paster create -t pyramid_routesalchemy Dibbus to fail to start when paster serve development.ini was used against the result. See https://github.com/Pylons/pyramid/issues/#issue/107
  • The render_view method of pyramid.renderers.RendererHelper passed an incorrect value into the renderer for renderer_info. It now passes an instance of RendererHelper instead of a dictionary, which is consistent with other usages. See https://github.com/Pylons/pyramid/issues#issue/106
  • A bug existed in the pyramid.authentication.AuthTktCookieHelper which would break any usage of an AuthTktAuthenticationPolicy when one was configured to reissue its tokens (reissue_time < timeout / max_age). Symptom: ValueError: ('Invalid token %r', ''). See https://github.com/Pylons/pyramid/issues#issue/108.

1.0b1 (2011-01-21)

Features
  • The AuthTktAuthenticationPolicy now accepts a tokens parameter via pyramid.security.remember. The value must be a sequence of strings. Tokens are placed into the auth_tkt “tokens” field and returned in the auth_tkt cookie.
  • Add wild_domain argument to AuthTktAuthenticationPolicy, which defaults to True. If it is set to False, the feature of the policy which sets a cookie with a wilcard domain will be turned off.
  • Add a MANIFEST.in file to each paster template. See https://github.com/Pylons/pyramid/issues#issue/95
Bug Fixes
  • testing.setUp now adds a settings attribute to the registry (both when it’s passed a registry without any settings and when it creates one).
  • The testing.setUp function now takes a settings argument, which should be a dictionary. Its values will subsequently be available on the returned config object as config.registry.settings.
Documentation
  • Added “What’s New in Pyramid 1.0” chapter to HTML rendering of documentation.
  • Merged caseman-master narrative editing branch, many wording fixes and extensions.
  • Fix deprecated example showing chameleon_zpt API call in testing narrative chapter.
  • Added “Adding Methods to the Configurator via add_directive” section to Advanced Configuration narrative chapter.
  • Add docs for add_finished_callback, add_response_callback, route_path, route_url, and static_url methods to pyramid.request.Request API docs.
  • Add (minimal) documentation about using I18N within Mako templates to “Internationalization and Localization” narrative chapter.
  • Move content of “Forms” chapter back to “Views” chapter; I can’t think of a better place to put it.
  • Slightly improved interface docs for IAuthorizationPolicy.
  • Minimally explain usage of custom regular expressions in URL dispatch replacement markers within URL Dispatch chapter.
Deprecations
  • Using the pyramid.view.bfg_view alias for pyramid.view.view_config (a backwards compatibility shim) now issues a deprecation warning.
Backwards Incompatibilities
  • Using testing.setUp now registers an ISettings utility as a side effect. Some test code which queries for this utility after testing.setUp via queryAdapter will expect a return value of None. This code will need to be changed.
  • When a pyramid.exceptions.Forbidden error is raised, its status code now 403 Forbidden. It was previously 401 Unauthorized, for backwards compatibility purposes with repoze.bfg. This change will cause problems for users of Pyramid with repoze.who, which intercepts 401 Unauthorized by default, but allows 403 Forbidden to pass through. Those deployments will need to configure repoze.who to also react to 403 Forbidden.
  • The default value for the cookie_on_exception parameter to pyramid.session.UnencyrptedCookieSessionFactory is now True. This means that when view code causes an exception to be raised, and the session has been mutated, a cookie will be sent back in the response. Previously its default value was False.
Paster Templates
  • The pyramid_zodb, pyramid_routesalchemy and pyramid_alchemy paster templates now use a default “commit veto” hook when configuring the repoze.tm2 transaction manager in development.ini. This prevents a transaction from being committed when the response status code is within the 400 or 500 ranges. See also http://docs.repoze.org/tm2/#using-a-commit-veto.

1.0a10 (2011-01-18)

Bug Fixes
  • URL dispatch now properly handles a .* or * appearing in a regex match when used inside brackets. Resolves issue #90.
Backwards Incompatibilities
  • The add_handler method of a Configurator has been removed from the Pyramid core. Handlers are now a feature of the pyramid_handlers package, which can be downloaded from PyPI. Documentation for the package should be available via http://pylonsproject.org/projects/pyramid_handlers/dev/, which describes how to add a configuration statement to your main block to reobtain this method. You will also need to add an install_requires dependency upon pyramid_handlers to your setup.py file.
  • The load_zcml method of a Configurator has been removed from the Pyramid core. Loading ZCML is now a feature of the pyramid_zcml package, which can be downloaded from PyPI. Documentation for the package should be available via http://pylonsproject.org/projects/pyramid_zcml/dev/, which describes how to add a configuration statement to your main block to reobtain this method. You will also need to add an install_requires dependency upon pyramid_zcml to your setup.py file.
  • The pyramid.includes subpackage has been removed. ZCML files which use include the package pyramid.includes (e.g. <include package="pyramid.includes"/>) now must include the pyramid_zcml package instead (e.g. <include package="pyramid_zcml"/>).
  • The pyramid.view.action decorator has been removed from the Pyramid core. Handlers are now a feature of the pyramid_handlers package. It should now be imported from pyramid_handlers e.g. from pyramid_handlers import action.
  • The handler ZCML directive has been removed. It is now a feature of the pyramid_handlers package.
  • The pylons_minimal, pylons_basic and pylons_sqla paster templates were removed. Use pyramid_sqla (available from PyPI) as a generic replacement for Pylons-esque development.
  • The make_app function has been removed from the pyramid.router module. It continues life within the pyramid_zcml package. This leaves the pyramid.router module without any API functions.
  • The configure_zcml setting within the deployment settings (within **settings passed to a Pyramid main function) has ceased to have any meaning.
Features
  • pyramid.testing.setUp and pyramid.testing.tearDown have been undeprecated. They are now the canonical setup and teardown APIs for test configuration, replacing “direct” creation of a Configurator. This is a change designed to provide a facade that will protect against any future Configurator deprecations.
  • Add charset attribute to pyramid.testing.DummyRequest (unconditionally UTF-8).
  • Add add_directive method to configurator, which allows framework extenders to add methods to the configurator (ala ZCML directives).
  • When Configurator.include is passed a module as an argument, it defaults to attempting to find and use a callable named includeme within that module. This makes it possible to use config.include('some.module') rather than config.include('some.module.somefunc') as long as the include function within some.module is named includeme.
  • The bfg2pyramid script now converts ZCML include tags that have repoze.bfg.includes as a package attribute to the value pyramid_zcml. For example, <include package="repoze.bfg.includes"> will be converted to <include package="pyramid_zcml">.
Paster Templates
  • All paster templates now use pyramid.testing.setUp and pyramid.testing.tearDown rather than creating a Configurator “by hand” within their tests.py module, as per decision in features above.
  • The starter_zcml paster template has been moved to the pyramid_zcml package.
Documentation
  • The wiki and wiki2 tutorials now use pyramid.testing.setUp and pyramid.testing.tearDown rather than creating a Configurator “by hand”, as per decision in features above.
  • The “Testing” narrative chapter now explains pyramid.testing.setUp and pyramid.testing.tearDown instead of Configurator creation and Configurator.begin() and Configurator.end().
  • Document the request.override_renderer attribute within the narrative “Renderers” chapter in a section named “Overriding A Renderer at Runtime”.
  • The “Declarative Configuration” narrative chapter has been removed (it was moved to the pyramid_zcml package).
  • Most references to ZCML in narrative chapters have been removed or redirected to pyramid_zcml locations.
Deprecations
  • Deprecation warnings related to import of the following API functions were added: pyramid.traversal.find_model, pyramid.traversal.model_path, pyramid.traversal.model_path_tuple, pyramid.url.model_url. The instructions emitted by the deprecation warnings instruct the developer to change these method spellings to their resource equivalents. This is a consequence of the mass concept rename of “model” to “resource” performed in 1.0a7.

1.0a9 (2011-01-08)

Bug Fixes
  • The proutes command tried too hard to resolve the view for printing, resulting in exceptions when an exceptional root factory was encountered. Instead of trying to resolve the view, if it cannot, it will now just print <unknown>.
  • The self argument was included in new methods of the ISession interface signature, causing pyramid_beaker tests to fail.
  • Readd pyramid.traversal.model_path_tuple as an alias for pyramid.traversal.resource_path_tuple for backwards compatibility.
Features
  • Add a new API pyramid.url.current_route_url, which computes a URL based on the “current” route (if any) and its matchdict values.
  • config.add_view now accepts a decorator keyword argument, a callable which will decorate the view callable before it is added to the registry.
  • If a handler class provides an __action_decorator__ attribute (usually a classmethod or staticmethod), use that as the decorator for each view registration for that handler.
  • The pyramid.interfaces.IAuthenticationPolicy interface now specifies an unauthenticated_userid method. This method supports an important optimization required by people who are using persistent storages which do not support object caching and whom want to create a “user object” as a request attribute.
  • A new API has been added to the pyramid.security module named unauthenticated_userid. This API function calls the unauthenticated_userid method of the effective security policy.
  • An unauthenticated_userid method has been added to the dummy authentication policy returned by pyramid.config.Configurator.testing_securitypolicy. It returns the same thing as that the dummy authentication policy’s authenticated_userid method.
  • The class pyramid.authentication.AuthTktCookieHelper is now an API. This class can be used by third-party authentication policy developers to help in the mechanics of authentication cookie-setting.
  • New constructor argument to Configurator: default_view_mapper. Useful to create systems that have alternate view calling conventions. A view mapper allows objects that are meant to be used as view callables to have an arbitrary argument list and an arbitrary result. The object passed as default_view_mapper should implement the pyramid.interfaces.IViewMapperFactory interface.
  • add a set_view_mapper API to Configurator. Has the same result as passing default_view_mapper to the Configurator constructor.
  • config.add_view now accepts a mapper keyword argument, which should either be None, a string representing a Python dotted name, or an object which is an IViewMapperFactory. This feature is not useful for “civilians”, only for extension writers.
  • Allow static renderer provided during view registration to be overridden at request time via a request attribute named override_renderer, which should be the name of a previously registered renderer. Useful to provide “omnipresent” RPC using existing rendered views.
  • Instances of pyramid.testing.DummyRequest now have a session object, which is mostly a dictionary, but also implements the other session API methods for flash and CSRF.
Backwards Incompatibilities
  • Since the pyramid.interfaces.IAuthenticationPolicy interface now specifies that a policy implementation must implement an unauthenticated_userid method, all third-party custom authentication policies now must implement this method. It, however, will only be called when the global function named pyramid.security.unauthenticated_userid is invoked, so if you’re not invoking that, you will not notice any issues.
  • pyramid.interfaces.ISession.get_csrf_token now mandates that an implementation should return a new token if one doesn’t already exist in the session (previously it would return None). The internal sessioning implementation has been changed.
Documentation
  • The (weak) “Converting a CMF Application to Pyramid” tutorial has been removed from the tutorials section. It was moved to the pyramid_tutorials Github repository.
  • The “Resource Location and View Lookup” chapter has been replaced with a variant of Rob Miller’s “Much Ado About Traversal” (originally published at http://blog.nonsequitarian.org/2010/much-ado-about-traversal/).
  • Many minor wording tweaks and refactorings (merged Casey Duncan’s docs fork, in which he is working on general editing).
  • Added (weak) description of new view mapper feature to Hooks narrative chapter.
  • Split views chapter into 2: View Callables and View Configuration.
  • Reorder Renderers and Templates chapters after View Callables but before View Configuration.
  • Merge Session Objects, Cross-Site Request Forgery, and Flash Messaging chapter into a single Sessions chapter.
  • The Wiki and Wiki2 tutorials now have much nicer CSS and graphics.
Internals
  • The “view derivation” code is now factored into a set of classes rather than a large number of standalone functions (a side effect of the view mapper refactoring).
  • The pyramid.renderer.RendererHelper class has grown a render_view method, which is used by the default view mapper (a side effect of the view mapper refactoring).
  • The object passed as renderer to the “view deriver” is now an instance of pyramid.renderers.RendererHelper rather than a dictionary (a side effect of view mapper refactoring).
  • The class used as the “page template” in pyramid.chameleon_text was removed, in preference to using a Chameleon-inbuilt version.
  • A view callable wrapper registered in the registry now contains an __original_view__ attribute which references the original view callable (or class).
  • The (non-API) method of all internal authentication policy implementations previously named _get_userid is now named unauthenticated_userid, promoted to an API method. If you were overriding this method, you’ll now need to override it as unauthenticated_userid instead.
  • Remove (non-API) function of config.py named _map_view.

1.0a8 (2010-12-27)

Bug Fixes
  • The name registry was not available in the paster pshell environment under IPython.
Features
  • If a resource implements a __resource_url__ method, it will be called as the result of invoking the pyramid.url.resource_url function to generate a URL, overriding the default logic. See the new “Generating The URL Of A Resource” section within the Resources narrative chapter.
  • Added flash messaging, as described in the “Flash Messaging” narrative documentation chapter.
  • Added CSRF token generation, as described in the narrative chapter entitled “Preventing Cross-Site Request Forgery Attacks”.
  • Prevent misunderstanding of how the view and view_permission arguments to add_route work by raising an exception during configuration if view-related arguments exist but no view argument is passed.
  • Add paster proute command which displays a summary of the routing table. See the narrative documentation section within the “URL Dispatch” chapter entitled “Displaying All Application Routes”.
Paster Templates
  • The pyramid_zodb Paster template no longer employs ZCML. Instead, it is based on scanning.
Documentation
  • Added “Generating The URL Of A Resource” section to the Resources narrative chapter (includes information about overriding URL generation using __resource_url__).
  • Added “Generating the Path To a Resource” section to the Resources narrative chapter.
  • Added “Finding a Resource by Path” section to the Resources narrative chapter.
  • Added “Obtaining the Lineage of a Resource” to the Resources narrative chapter.
  • Added “Determining if a Resource is In The Lineage of Another Resource” to Resources narrative chapter.
  • Added “Finding the Root Resource” to Resources narrative chapter.
  • Added “Finding a Resource With a Class or Interface in Lineage” to Resources narrative chapter.
  • Added a “Flash Messaging” narrative documentation chapter.
  • Added a narrative chapter entitled “Preventing Cross-Site Request Forgery Attacks”.
  • Changed the “ZODB + Traversal Wiki Tutorial” based on changes to pyramid_zodb Paster template.
  • Added “Advanced Configuration” narrative chapter which documents how to deal with configuration conflicts, two-phase configuration, include and commit.
  • Fix API documentation rendering for pyramid.view.static
  • Add “Pyramid Provides More Than One Way to Do It” to Design Defense documentation.
  • Changed “Static Assets” narrative chapter: clarify that name represents a prefix unless it’s a URL, added an example of a root-relative static view fallback for URL dispatch, added an example of creating a simple view that returns the body of a file.
  • Move ZCML usage in Hooks chapter to Declarative Configuration chapter.
  • Merge “Static Assets” chapter into the “Assets” chapter.
  • Added narrative documentation section within the “URL Dispatch” chapter entitled “Displaying All Application Routes” (for paster proutes command).

1.0a7 (2010-12-20)

Terminology Changes

  • The Pyramid concept previously known as “model” is now known as “resource”. As a result:

    • The following API changes have been made:

      pyramid.url.model_url ->
                  pyramid.url.resource_url

pyramid.traversal.find_model ->
                  pyramid.url.find_resource

pyramid.traversal.model_path ->
                  pyramid.traversal.resource_path

pyramid.traversal.model_path_tuple ->
                  pyramid.traversal.resource_path_tuple

pyramid.traversal.ModelGraphTraverser ->
                  pyramid.traversal.ResourceTreeTraverser

pyramid.config.Configurator.testing_models ->
                  pyramid.config.Configurator.testing_resources

pyramid.testing.registerModels ->
                  pyramid.testing.registerResources

pyramid.testing.DummyModel ->
                  pyramid.testing.DummyResource
    • All documentation which previously referred to “model” now refers to “resource”.
    • The starter and starter_zcml paster templates now have a resources.py module instead of a models.py module.
    • Positional argument names of various APIs have been changed from model to resource.

    Backwards compatibility shims have been left in place in all cases. They will continue to work “forever”.

  • The Pyramid concept previously known as “resource” is now known as “asset”. As a result:

    • The (non-API) module previously known as pyramid.resource is now known as pyramid.asset.

    • All docs that previously referred to “resource specification” now refer to “asset specification”.

    • The following API changes were made:

      pyramid.config.Configurator.absolute_resource_spec ->
                  pyramid.config.Configurator.absolute_asset_spec

pyramid.config.Configurator.override_resource ->
                  pyramid.config.Configurator.override_asset

    • The ZCML directive previously known as resource is now known as asset.

    • The setting previously known as BFG_RELOAD_RESOURCES (envvar) or reload_resources (config file) is now known, respectively, as PYRAMID_RELOAD_ASSETS and reload_assets.

    Backwards compatibility shims have been left in place in all cases. They will continue to work “forever”.

Bug Fixes
  • Make it possible to succesfully run all tests via nosetests command directly (rather than indirectly via python setup.py nosetests).
  • When a configuration conflict is encountered during scanning, the conflict exception now shows the decorator information that caused the conflict.
Features
  • Added debug_routematch configuration setting that logs matched routes (including the matchdict and predicates).
  • The name registry is now available in a pshell environment by default. It is the application registry object.
Environment
  • All environment variables which used to be prefixed with BFG_ are now prefixed with PYRAMID_ (e.g. BFG_DEBUG_NOTFOUND is now PYRAMID_DEBUG_NOTFOUND)
Documentation
  • Added “Debugging Route Matching” section to the urldispatch narrative documentation chapter.
  • Added reference to PYRAMID_DEBUG_ROUTEMATCH envvar and debug_routematch config file setting to the Environment narrative docs chapter.
  • Changed “Project” chapter slightly to expand on use of paster pshell.
  • Direct Jython users to Mako rather than Jinja2 in “Install” narrative chapter.
  • Many changes to support terminological renaming of “model” to “resource” and “resource” to “asset”.
  • Added an example of WebTest functional testing to the testing narrative chapter.
  • Rearranged chapter ordering by popular demand (URL dispatch first, then traversal). Put hybrid chapter after views chapter.
  • Split off “Renderers” as its own chapter from “Views” chapter in narrative documentation.
Paster Templates
  • Added debug_routematch = false to all paster templates.
Dependencies
  • Depend on Venusian >= 0.5 (for scanning conflict exception decoration).

1.0a6 (2010-12-15)

Bug Fixes
  • 1.0a5 introduced a bug when pyramid.config.Configurator.scan was used without a package argument (e.g. config.scan() as opposed to config.scan('packagename'). The symptoms were: lots of deprecation warnings printed to the console about imports of deprecated Pyramid functions and classes and non-detection of view callables decorated with view_config decorators. This has been fixed.
  • Tests now pass on Windows (no bugs found, but a few tests in the test suite assumed UNIX path segments in filenames).
Documentation
  • If you followed it to-the-letter, the ZODB+Traversal Wiki tutorial would instruct you to run a test which would fail because the view callable generated by the pyramid_zodb tutorial used a one-arg view callable, but the test in the sample code used a two-arg call.
  • Updated ZODB+Traversal tutorial setup.py of all steps to match what’s generated by pyramid_zodb.
  • Fix reference to repoze.bfg.traversalwrapper in “Models” chapter (point at pyramid_traversalwrapper instead).

1.0a5 (2010-12-14)

Features
  • Add a handler ZCML directive. This directive does the same thing as pyramid.configuration.add_handler.
  • A new module named pyramid.config was added. It subsumes the duties of the older pyramid.configuration module.
  • The new pyramid.config.Configurator` class has API methods that the older ``pyramid.configuration.Configurator class did not: with_context (a classmethod), include, action, and commit. These methods exist for imperative application extensibility purposes.
  • The pyramid.testing.setUp function now accepts an autocommit keyword argument, which defaults to True. If it is passed False, the Config object returned by setUp will be a non-autocommiting Config object.
  • Add logging configuration to all paster templates.
  • pyramid_alchemy, pyramid_routesalchemy, and pylons_sqla paster templates now use idiomatic SQLAlchemy configuration in their respective .ini files and Python code.
  • pyramid.testing.DummyRequest now has a class variable, query_string, which defaults to the empty string.
  • Add support for json on GAE by catching NotImplementedError and importing simplejson from django.utils.
  • The Mako renderer now accepts a resource specification for mako.module_directory.
  • New boolean Mako settings variable mako.strict_undefined. See Mako Context Variables for its meaning.
Dependencies
  • Depend on Mako 0.3.6+ (we now require the strict_undefined feature).
Bug Fixes
  • When creating a Configurator from within a paster pshell session, you were required to pass a package argument although package is not actually required. If you didn’t pass package, you would receive an error something like KeyError: '__name__' emanating from the pyramid.path.caller_module function. This has now been fixed.
  • The pyramid_routesalchemy paster template’s unit tests failed (AssertionError: 'SomeProject' != 'someproject'). This is fixed.
  • Make default renderer work (renderer factory registered with no name, which is active for every view unless the view names a specific renderer).
  • The Mako renderer did not properly turn the mako.imports, mako.default_filters, and mako.imports settings into lists.
  • The Mako renderer did not properly convert the mako.error_handler setting from a dotted name to a callable.
Documentation
  • Merged many wording, readability, and correctness changes to narrative documentation chapters from https://github.com/caseman/pyramid (up to and including “Models” narrative chapter).
  • “Sample Applications” section of docs changed to note existence of Cluegun, Shootout and Virginia sample applications, ported from their repoze.bfg origin packages.
  • SQLAlchemy+URLDispatch tutorial updated to integrate changes to pyramid_routesalchemy template.
  • Add pyramid.interfaces.ITemplateRenderer interface to Interfaces API chapter (has implementation() method, required to be used when getting at Chameleon macros).
  • Add a “Modifying Package Structure” section to the project narrative documentation chapter (explain turning a module into a package).
  • Documentation was added for the new handler ZCML directive in the ZCML section.
Deprecations

  • pyramid.configuration.Configurator is now deprecated. Use pyramid.config.Configurator, passing its constructor autocommit=True instead. The pyramid.configuration.Configurator alias will live for a long time, as every application uses it, but its import now issues a deprecation warning. The pyramid.config.Configurator class has the same API as pyramid.configuration.Configurator class, which it means to replace, except by default it is a non-autocommitting configurator. The now-deprecated pyramid.configuration.Configurator will autocommit every time a configuration method is called.

    The pyramid.configuration module remains, but it is deprecated. Use pyramid.config instead.

1.0a4 (2010-11-21)

Features
  • URL Dispatch now allows for replacement markers to be located anywhere in the pattern, instead of immediately following a /.
  • URL Dispatch now uses the form {marker} to denote a replace marker in the route pattern instead of :marker. The old colon-style marker syntax is still accepted for backwards compatibility. The new format allows a regular expression for that marker location to be used instead of the default [^/]+, for example {marker:\d+} is now valid to require the marker to be digits.
  • Add a pyramid.url.route_path API, allowing folks to generate relative URLs. Calling route_path is the same as calling pyramid.url.route_url with the argument _app_url equal to the empty string.
  • Add a pyramid.request.Request.route_path API. This is a convenience method of the request which calls pyramid.url.route_url.
  • Make test suite pass on Jython (requires PasteScript trunk, presumably to be 1.7.4).
  • Make test suite pass on PyPy (Chameleon doesn’t work).
  • Surrounding application configuration with config.begin() and config.end() is no longer necessary. All paster templates have been changed to no longer call these functions.
  • Fix configurator to not convert ImportError to ConfigurationError if the import that failed was unrelated to the import requested via a dotted name when resolving dotted names (such as view dotted names).
Documentation
  • SQLAlchemy+URLDispatch and ZODB+Traversal tutorials have been updated to not call config.begin() or config.end().
Bug Fixes
  • Add deprecation warnings to import of pyramid.chameleon_text and pyramid.chameleon_zpt of get_renderer, get_template, render_template, and render_template_to_response.
  • Add deprecation warning for import of pyramid.zcml.zcml_configure and pyramid.zcml.file_configure.
  • The pyramid_alchemy paster template had a typo, preventing an import from working.
  • Fix apparent failures when calling pyramid.traversal.find_model(root, path) or pyramid.traversal.traverse(path) when path is (erroneously) a Unicode object. The user is meant to pass these APIs a string object, never a Unicode object. In practice, however, users indeed pass Unicode. Because the string that is passed must be ASCII encodeable, now, if they pass a Unicode object, its data is eagerly converted to an ASCII string rather than being passed along to downstream code as a convenience to the user and to prevent puzzling second-order failures from cropping up (all failures will occur within pyramid.traversal.traverse rather than later down the line as the result of calling e.g. traversal_path).
Backwards Incompatibilities
  • The pyramid.testing.zcml_configure API has been removed. It had been advertised as removed since repoze.bfg 1.2a1, but hadn’t actually been.
Deprecations
  • The pyramid.settings.get_settings API is now deprecated. Use pyramid.threadlocals.get_current_registry().settings instead or use the settings attribute of the registry available from the request (request.registry.settings).
Documentation
  • Removed zodbsessions tutorial chapter. It’s still useful, but we now have a SessionFactory abstraction which competes with it, and maintaining documentation on both ways to do it is a distraction.
Internal
  • Replace Twill with WebTest in internal integration tests (avoid deprecation warnings generated by Twill).

1.0a3 (2010-11-16)

Features
  • Added Mako TemplateLookup settings for mako.error_handler, mako.default_filters, and mako.imports.
  • Normalized all paster templates: each now uses the name main to represent the function that returns a WSGI application, each now uses WebError, each now has roughly the same shape of development.ini style.
  • Added class vars matchdict and matched_route to pyramid.request.Request. Each is set to None.
  • New API method: pyramid.settings.asbool.
  • New API methods for pyramid.request.Request: model_url, route_url, and static_url. These are simple passthroughs for their respective functions in pyramid.url.
  • The settings object which used to be available only when request.settings.get_settings was called is now available as registry.settings (e.g. request.registry.settings in view code).
Bug Fixes
  • The pylons_* paster templates erroneously used the {squiggly} routing syntax as the pattern supplied to add_route. This style of routing is not supported. They were replaced with :colon style route patterns.
  • The pylons_* paster template used the same string (your_app_secret_string) for the session.secret setting in the generated development.ini. This was a security risk if left unchanged in a project that used one of the templates to produce production applications. It now uses a randomly generated string.
Documentation
  • ZODB+traversal wiki (wiki) tutorial updated due to changes to pyramid_zodb paster template.
  • SQLAlchemy+urldispach wiki (wiki2) tutorial updated due to changes to pyramid_routesalchemy paster template.
  • Documented the matchdict and matched_route attributes of the request object in the Request API documentation.
Deprecations
  • Obtaining the settings object via registry.{get|query}Utility(ISettings) is now deprecated. Instead, obtain the settings object via the registry.settings attribute. A backwards compatibility shim was added to the registry object to register the settings object as an ISettings utility when setattr(registry, 'settings', foo) is called, but it will be removed in a later release.
  • Obtaining the settings object via pyramid.settings.get_settings is now deprecated. Obtain it as the settings attribute of the registry now (obtain the registry via pyramid.threadlocal.get_registry or as request.registry).
Behavior Differences
  • Internal: ZCML directives no longer call get_current_registry() if there’s a registry attribute on the ZCML context (kill off use of threadlocals).
  • Internal: Chameleon template renderers now accept two arguments: path and lookup. Lookup will be an instance of a lookup class which supplies (late-bound) arguments for debug, reload, and translate. Any third-party renderers which use (the non-API) function pyramid.renderers.template_renderer_factory will need to adjust their implementations to obey the new callback argument list. This change was to kill off inappropriate use of threadlocals.

1.0a2 (2010-11-09)

Documentation
  • All references to events by interface (e.g. pyramid.interfaces.INewRequest) have been changed to reference their concrete classes (e.g. pyramid.events.NewRequest) in documentation about making subscriptions.
  • All references to Pyramid-the-application were changed from mod-pyramid to app-Pyramid. A custom role setting was added to docs/conf.py to allow for this. (internal)

1.0a1 (2010-11-05)

Features (delta from BFG 1.3)

  • Mako templating renderer supports resource specification format for template lookups and within Mako templates. Absolute filenames must be used in Pyramid to avoid this lookup process.

  • Add pyramid.httpexceptions module, which is a facade for the webob.exc module.

  • Direct built-in support for the Mako templating language.

  • A new configurator method exists: add_handler. This method adds a Pylons-style “view handler” (such a thing used to be called a “controller” in Pylons 1.0).

  • New argument to configurator: session_factory.

  • New method on configurator: set_session_factory

  • Using request.session now returns a (dictionary-like) session object if a session factory has been configured.

  • The request now has a new attribute: tmpl_context for benefit of Pylons users.

  • The decorator previously known as pyramid.view.bfg_view is now known most formally as pyramid.view.view_config in docs and paster templates. An import of pyramid.view.bfg_view, however, will continue to work “forever”.

  • New API methods in pyramid.session: signed_serialize and signed_deserialize.

  • New interface: pyramid.interfaces.IRendererInfo. An object of this type is passed to renderer factory constructors (see “Backwards Incompatibilities”).

  • New event type: pyramid.interfaces.IBeforeRender. An object of this type is sent as an event before a renderer is invoked (but after the application-level renderer globals factory added via pyramid.configurator.configuration.set_renderer_globals_factory, if any, has injected its own keys). Applications may now subscribe to the IBeforeRender event type in order to introspect the and modify the set of renderer globals before they are passed to a renderer. The event object iself has a dictionary-like interface that can be used for this purpose. For example:

    from repoze.events import subscriber
from pyramid.interfaces import IRendererGlobalsEvent

@subscriber(IRendererGlobalsEvent)
def add_global(event):
    event['mykey'] = 'foo'

    If a subscriber attempts to add a key that already exist in the renderer globals dictionary, a KeyError is raised. This limitation is due to the fact that subscribers cannot be ordered relative to each other. The set of keys added to the renderer globals dictionary by all subscribers and app-level globals factories must be unique.

  • New class: pyramid.response.Response. This is a pure facade for webob.Response (old code need not change to use this facade, it’s existence is mostly for vanity and documentation-generation purposes).

  • All preexisting paster templates (except zodb) now use “imperative” configuration (starter, routesalchemy, alchemy).

  • A new paster template named pyramid_starter_zcml exists, which uses declarative configuration.

Documentation (delta from BFG 1.3)
  • Added a pyramid.httpexceptions API documentation chapter.
  • Added a pyramid.session API documentation chapter.
  • Added a Session Objects narrative documentation chapter.
  • Added an API chapter for the pyramid.personality module.
  • Added an API chapter for the pyramid.response module.
  • All documentation which previously referred to webob.Response now uses pyramid.response.Response instead.
  • The documentation has been overhauled to use imperative configuration, moving declarative configuration (ZCML) explanations to a separate narrative chapter declarative.rst.
  • The ZODB Wiki tutorial was updated to take into account changes to the pyramid_zodb paster template.
  • The SQL Wiki tutorial was updated to take into account changes to the pyramid_routesalchemy paster template.
Backwards Incompatibilities (with BFG 1.3)

  • There is no longer an IDebugLogger registered as a named utility with the name repoze.bfg.debug.

  • The logger which used to have the name of repoze.bfg.debug now has the name pyramid.debug.

  • The deprecated API pyramid.testing.registerViewPermission has been removed.

  • The deprecated API named pyramid.testing.registerRoutesMapper has been removed.

  • The deprecated API named pyramid.request.get_request was removed.

  • The deprecated API named pyramid.security.Unauthorized was removed.

  • The deprecated API named pyramid.view.view_execution_permitted was removed.

  • The deprecated API named pyramid.view.NotFound was removed.

  • The bfgshell paster command is now named pshell.

  • The Venusian “category” for all built-in Venusian decorators (e.g. subscriber and view_config/bfg_view) is now pyramid instead of bfg.

  • pyramid.renderers.rendered_response function removed; use render_pyramid.renderers.render_to_response instead.

  • Renderer factories now accept a renderer info object rather than an absolute resource specification or an absolute path. The object has the following attributes: name (the renderer= value), package (the ‘current package’ when the renderer configuration statement was found), type: the renderer type, registry: the current registry, and settings: the deployment settings dictionary.

    Third-party repoze.bfg renderer implementations that must be ported to Pyramid will need to account for this.

    This change was made primarily to support more flexible Mako template rendering.

  • The presence of the key repoze.bfg.message in the WSGI environment when an exception occurs is now deprecated. Instead, code which relies on this environ value should use the exception attribute of the request (e.g. request.exception[0]) to retrieve the message.

  • The values bfg_localizer and bfg_locale_name kept on the request during internationalization for caching purposes were never APIs. These however have changed to localizer and locale_name, respectively.

  • The default cookie_name value of the authtktauthenticationpolicy ZCML now defaults to auth_tkt (it used to default to repoze.bfg.auth_tkt).

  • The default cookie_name value of the pyramid.authentication.AuthTktAuthenticationPolicy constructor now defaults to auth_tkt (it used to default to repoze.bfg.auth_tkt).

  • The request_type argument to the view ZCML directive, the pyramid.configuration.Configurator.add_view method, or the pyramid.view.view_config decorator (nee bfg_view) is no longer permitted to be one of the strings GET, HEAD, PUT, POST or DELETE, and now must always be an interface. Accepting the method-strings as request_type was a backwards compatibility strategy servicing repoze.bfg 1.0 applications. Use the request_method parameter instead to specify that a view a string request-method predicate.

repoze.bfg Change History (previous name for Pyramid)

1.3b1 (2010-10-25)

Features
  • The paster template named bfg_routesalchemy has been updated to use SQLAlchemy declarative syntax. Thanks to Ergo^.
Bug Fixes
  • When a renderer factory could not be found, a misleading error message was raised if the renderer name was not a string.
Documentation
  • The “”bfgwiki2” (SQLAlchemy + url dispatch) tutorial has been updated slightly. In particular, the source packages no longer attempt to use a private index, and the recommended Python version is now 2.6. It was also updated to take into account the changes to the bfg_routesalchemy template used to set up an environment.
  • The “bfgwiki” (ZODB + traversal) tutorial has been updated slightly. In particular, the source packages no longer attempt to use a private index, and the recommended Python version is now 2.6.

1.3a15 (2010-09-30)

Features
  • The repoze.bfg.traversal.traversal_path API now eagerly attempts to encode a Unicode path into ASCII before attempting to split it and decode its segments. This is for convenience, effectively to allow a (stored-as-Unicode-in-a-database, or retrieved-as-Unicode-from-a-request-parameter) Unicode path to be passed to find_model, which eventually internally uses the traversal_path function under the hood. In version 1.2 and prior, if the path was Unicode, that Unicode was split on slashes and each resulting segment value was Unicode. An inappropriate call to the decode() method of a resulting Unicode path segment could cause a UnicodeDecodeError to occur even if the Unicode representation of the path contained no ‘high order’ characters (it effectively did a “double decode”). By converting the Unicode path argument to ASCII before we attempt to decode and split, genuine errors will occur in a more obvious place while also allowing us to handle (for convenience) the case that it’s a Unicode representation formed entirely from ASCII-compatible characters.

1.3a14 (2010-09-14)

Bug Fixes
  • If an exception view was registered through the legacy set_notfound_view or set_forbidden_view APIs, the context sent to the view was incorrect (could be None inappropriately).
Features
  • Compatibility with WebOb 1.0.
Requirements
  • Now requires WebOb >= 1.0.
Backwards Incompatibilities
  • Due to changes introduced WebOb 1.0, the repoze.bfg.request.make_request_ascii event subscriber no longer works, so it has been removed. This subscriber was meant to be used in a deployment so that code written before BFG 0.7.0 could run unchanged. At this point, such code will need to be rewritten to expect Unicode from request.GET, request.POST and request.params or it will need to be changed to use request.str_POST, request.str_GET and/or request.str_params instead of the non-str versions of same, as the non-str versions of the same APIs always now perform decoding to Unicode.
Errata
  • A prior changelog entry asserted that the INewResponse event was not sent to listeners if the response was not “valid” (if a view or renderer returned a response object that did not have a status/headers/app_iter). This is not true in this release, nor was it true in 1.3a13.

1.3a13 (2010-09-14)

Bug Fixes
  • The traverse route predicate could not successfully generate a traversal path.
Features
  • In support of making it easier to configure applications which are “secure by default”, a default permission feature was added. If supplied, the default permission is used as the permission string to all view registrations which don’t otherwise name a permission. These APIs are in support of that:
    • A new constructor argument was added to the Configurator: default_permission.
    • A new method was added to the Configurator: set_default_permission.
    • A new ZCML directive was added: default_permission.
  • Add a new request API: request.add_finished_callback. Finished callbacks are called by the router unconditionally near the very end of request processing. See the “Using Finished Callbacks” section of the “Hooks” narrative chapter of the documentation for more information.
  • A request.matched_route attribute is now added to the request when a route has matched. Its value is the “route” object that matched (see the IRoute interface within repoze.bfg.interfaces API documentation for the API of a route object).
  • The exception attribute of the request is now set slightly earlier and in a slightly different set of scenarios, for benefit of “finished callbacks” and “response callbacks”. In previous versions, the exception attribute of the request was not set at all if an exception view was not found. In this version, the request.exception attribute is set immediately when an exception is caught by the router, even if an exception view could not be found.
  • The add_route method of a Configurator now accepts a pregenerator argument. The pregenerator for the resulting route is called by route_url in order to adjust the set of arguments passed to it by the user for special purposes, such as Pylons ‘subdomain’ support. It will influence the URL returned by route_url. See the repoze.bfg.interfaces.IRoutePregenerator interface for more information.
Backwards Incompatibilities
  • The router no longer sets the value wsgiorg.routing_args into the environ when a route matches. The value used to be something like ((), matchdict). This functionality was only ever obliquely referred to in change logs; it was never documented as an API.
  • The exception attribute of the request now defaults to None. In prior versions, the request.exception attribute did not exist if an exception was not raised by user code during request processing; it only began existence once an exception view was found.
Deprecations
  • The repoze.bfg.interfaces.IWSGIApplicationCreatedEvent event interface was renamed to repoze.bfg.interfaces.IApplicationCreated. Likewise, the repoze.bfg.events.WSGIApplicationCreatedEvent class was renamed to repoze.bfg.events.ApplicationCreated. The older aliases will continue to work indefinitely.
  • The repoze.bfg.interfaces.IAfterTraversal event interface was renamed to repoze.bfg.interfaces.IContextFound. Likewise, the repoze.bfg.events.AfterTraversal class was renamed to repoze.bfg.events.ContextFound. The older aliases will continue to work indefinitely.
  • References to the WSGI environment values bfg.routes.matchdict and bfg.routes.route were removed from documentation. These will stick around internally for several more releases, but it is request.matchdict and request.matched_route are now the “official” way to obtain the matchdict and the route object which resulted in the match.
Documentation
  • Added documentation for the default_permission ZCML directive.
  • Added documentation for the default_permission constructor value and the set_default_permission method in the Configurator API documentation.
  • Added a new section to the “security” chapter named “Setting a Default Permission”.
  • Document renderer_globals_factory and request_factory arguments to Configurator constructor.
  • Added two sections to the “Hooks” chapter of the documentation: “Using Response Callbacks” and “Using Finished Callbacks”.
  • Added documentation of the request.exception attribute to the repoze.bfg.request.Request API documentation.
  • Added glossary entries for “response callback” and “finished callback”.
  • The “Request Processing” narrative chapter has been updated to note finished and response callback steps.
  • New interface in interfaces API documentation: IRoutePregenerator.
  • Added a “The Matched Route” section to the URL Dispatch narrative docs chapter, detailing the matched_route attribute.

1.3a12 (2010-09-08)

Bug Fixes
  • Fix a bug in repoze.bfg.url.static_url URL generation: if two resource specifications were used to create two separate static views, but they shared a common prefix, it was possible that static_url would generate an incorrect URL.
  • Fix another bug in repoze.bfg.static_url URL generation: too many slashes in generated URL.
  • Prevent a race condition which could result in a RuntimeError when rendering a Chameleon template that has not already been rendered once. This would usually occur directly after a restart, when more than one person or thread is trying to execute the same view at the same time: https://bugs.launchpad.net/karl3/+bug/621364
Features
  • The argument to repoze.bfg.configuration.Configurator.add_route which was previously called path is now called pattern for better explicability. For backwards compatibility purposes, passing a keyword argument named path to add_route will still work indefinitely.
  • The path attribute to the ZCML route directive is now named pattern for better explicability. The older path attribute will continue to work indefinitely.
Documentation
  • All narrative, API, and tutorial docs which referred to a route pattern as a path have now been updated to refer to them as a pattern.
  • The repoze.bfg.interfaces API documentation page is now rendered via repoze.sphinx.autointerface.
  • The URL Dispatch narrative chapter now refers to the interfaces chapter to explain the API of an IRoute object.
Paster Templates
  • The routesalchemy template has been updated to use pattern in its route declarations rather than path.
Dependencies
  • tests_require now includes repoze.sphinx.autointerface as a dependency.
Internal
  • Add an API to the Configurator named get_routes_mapper. This returns an object implementing the IRoutesMapper interface.
  • The repoze.bfg.urldispatch.RoutesMapper object now has a get_route method which returns a single Route object or None.
  • A new interface repoze.bfg.interfaces.IRoute was added. The repoze.bfg.urldispatch.Route object implements this interface.
  • The canonical attribute for accessing the routing pattern from a route object is now pattern rather than path.
  • Use hash() rather than id() when computing the “phash” of a custom route/view predicate in order to allow the custom predicate some control over which predicates are “equal”.
  • Use response.headerlist.append instead of response.headers.add in repoze.bfg.request.add_global_response_headers in case the response is not a WebOb response.
  • The repoze.bfg.urldispatch.Route constructor (not an API) now accepts a different ordering of arguments. Previously it was (pattern, name, factory=None, predicates=()). It is now (name, pattern, factory=None, predicates=()). This is in support of consistency with configurator.add_route.
  • The repoze.bfg.urldispatch.RoutesMapper.connect method (not an API) now accepts a different ordering of arguments. Previously it was (pattern, name, factory=None, predicates=()). It is now (name, pattern, factory=None, predicates=()). This is in support of consistency with configurator.add_route.

1.3a11 (2010-09-05)

Bug Fixes
  • Process the response callbacks and the NewResponse event earlier, to enable mutations to the response to take effect.

1.3a10 (2010-09-05)

Features

  • A new repoze.bfg.request.Request.add_response_callback API has been added. This method is documented in the new repoze.bfg.request API chapter. It can be used to influence response values before a concrete response object has been created.

  • The repoze.bfg.interfaces.INewResponse interface now includes a request attribute; as a result, a handler for INewResponse now has access to the request which caused the response.

  • Each of the follow methods of the Configurator now allow the below-named arguments to be passed as “dotted name strings” (e.g. “foo.bar.baz”) rather than as actual implementation objects that must be imported:

    setup_registry

    root_factory, authentication_policy, authorization_policy, debug_logger, locale_negotiator, request_factory, renderer_globals_factory

    add_subscriber

    subscriber, iface

    derive_view

    view

    add_view

    view, for_, context, request_type, containment

    add_route()

    view, view_for, factory, for_, view_context

    scan

    package

    add_renderer

    factory

    set_forbidden_view

    view

    set_notfound_view

    view

    set_request_factory

    factory

    set_renderer_globals_factory()

    factory

    set_locale_negotiator

    negotiator

    testing_add_subscriber

    event_iface

Bug Fixes
  • The route pattern registered internally for a a local “static view” (either via the static ZCML directive or via the add_static_view method of the configurator) was incorrect. It was regsistered for e.g. static*traverse, while it should have been registered for static/*traverse. Symptom: two static views could not reliably be added to a system when they both shared the same path prefix (e.g. /static and /static2).
Backwards Incompatibilities
  • The INewResponse event is now not sent to listeners if the response returned by view code (or a renderer) is not a “real” response (e.g. if it does not have .status, .headerlist and .app_iter attribtues).
Documentation
  • Add an API chapter for the repoze.bfg.request module, which includes documentation for the repoze.bfg.request.Request class (the “request object”).
  • Modify the “Request and Response” narrative chapter to reference the new repoze.bfg.request API chapter. Some content was moved from this chapter into the API documentation itself.
  • Various changes to denote that Python dotted names are now allowed as input to Configurator methods.
Internal
  • The (internal) feature which made it possible to attach a global_response_headers attribute to the request (which was assumed to contain a sequence of header key/value pairs which would later be added to the response by the router), has been removed. The functionality of repoze.bfg.request.Request.add_response_callback takes its place.
  • The repoze.bfg.events.NewResponse class’s construct has changed: it now must be created with (request, response) rather than simply (response).

1.3a9 (2010-08-22)

Features
  • The Configurator now accepts a dotted name string to a package as a package constructor argument. The package argument was previously required to be a package object (not a dotted name string).
  • The repoze.bfg.configuration.Configurator.with_package method was added. This method returns a new Configurator using the same application registry as the configurator object it is called upon. The new configurator is created afresh with its package constructor argument set to the value passed to with_package. This feature will make it easier for future BFG versions to allow dotted names as arguments in places where currently only object references are allowed (the work to allow dotted names isntead of object references everywhere has not yet been done, however).
  • The new repoze.bfg.configuration.Configurator.maybe_dotted method resolves a Python dotted name string supplied as its dotted argument to a global Python object. If the value cannot be resolved, a repoze.bfg.configuration.ConfigurationError is raised. If the value supplied as dotted is not a string, the value is returned unconditionally without any resolution attempted.
  • The new repoze.bfg.configuration.Configurator.absolute_resource_spec method resolves a potentially relative “resource specification” string into an absolute version. If the value supplied as relative_spec is not a string, the value is returned unconditionally without any resolution attempted.
Backwards Incompatibilities
  • The functions in repoze.bfg.renderers named render and render_to_response introduced in 1.3a6 previously took a set of **values arguments for the values to be passed to the renderer. This was wrong, as renderers don’t need to accept only dictionaries (they can accept any type of object). Now, the value sent to the renderer must be supplied as a positional argument named value. The request argument is still a keyword argument, however.
  • The functions in repoze.bfg.renderers named render and render_to_response now accept an additonal keyword argument named package.
  • The get_renderer API in repoze.bfg.renderers now accepts a package argument.
Documentation
  • The ZCML include directive docs were incorrect: they specified filename rather than (the correct) file as an allowable attribute.
Internal
  • The repoze.bfg.resource.resolve_resource_spec function can now accept a package object as its pname argument instead of just a package name.
  • The _renderer_factory_from_name and _renderer_from_name methods of the Configurator were removed. These were never APIs.
  • The _render, _render_to_response and _make_response functions with repoze.bfg.render (added in 1.3a6) have been removed.
  • A new helper class repoze.bfg.renderers.RendererHelper was added.
  • The _map_view function of repoze.bfg.configuration now takes only a renderer_name argument instead of both a renderer and renderer``_name argument.  It also takes a ``package argument now.
  • Use imp.get_suffixes indirection in repoze.bfg.path.package_name instead of hardcoded .py .pyc and .pyo to use for comparison when attemtping to decide if a directory is a package.
  • Make tests runnable again under Jython (although they do not all pass currently).
  • The reify decorator now maintains the docstring of the function it wraps.

1.3a8 (2010-08-08)

Features

  • New public interface: repoze.bfg.exceptions.IExceptionResponse. This interface is provided by all internal exception classes (such as repoze.bfg.exceptions.NotFound and repoze.bfg.exceptions.Forbidden), instances of which are both exception objects and can behave as WSGI response objects. This interface is made public so that exception classes which are also valid WSGI response factories can be configured to implement them or exception instances which are also or response instances can be configured to provide them.

  • New API class: repoze.bfg.view.AppendSlashNotFoundViewFactory.

    There can only be one Not Found view in any repoze.bfg application. Even if you use repoze.bfg.view.append_slash_notfound_view as the Not Found view, repoze.bfg still must generate a 404 Not Found response when it cannot redirect to a slash-appended URL; this not found response will be visible to site users.

    If you don’t care what this 404 response looks like, and you only need redirections to slash-appended route URLs, you may use the repoze.bfg.view.append_slash_notfound_view object as the Not Found view. However, if you wish to use a custom notfound view callable when a URL cannot be redirected to a slash-appended URL, you may wish to use an instance of the repoze.bfg.view.AppendSlashNotFoundViewFactory class as the Not Found view, supplying the notfound view callable as the first argument to its constructor. For instance:

    from repoze.bfg.exceptions import NotFound
from repoze.bfg.view import AppendSlashNotFoundViewFactory

def notfound_view(context, request):
    return HTTPNotFound('It aint there, stop trying!')

custom_append_slash = AppendSlashNotFoundViewFactory(notfound_view)
config.add_view(custom_append_slash, context=NotFound)

    The notfound_view supplied must adhere to the two-argument view callable calling convention of (context, request) (context will be the exception object).

Documentation
  • Expanded the “Cleaning Up After a Request” section of the URL Dispatch narrative chapter.
  • Expanded the “Redirecting to Slash-Appended Routes” section of the URL Dispatch narrative chapter.
Internal
  • Previously, two default view functions were registered at Configurator setup (one for repoze.bfg.exceptions.NotFound named default_notfound_view and one for repoze.bfg.exceptions.Forbidden named default_forbidden_view) to render internal exception responses. Those default view functions have been removed, replaced with a generic default view function which is registered at Configurator setup for the repoze.bfg.interfaces.IExceptionResponse interface that simply returns the exception instance; the NotFound and Forbidden classes are now still exception factories but they are also response factories which generate instances that implement the new repoze.bfg.interfaces.IExceptionResponse interface.

1.3a7 (2010-08-01)

Features
  • The repoze.bfg.configuration.Configurator.add_route API now returns the route object that was added.
  • A repoze.bfg.events.subscriber decorator was added. This decorator decorates module-scope functions, which are then treated as event listeners after a scan() is performed. See the Events narrative documentation chapter and the repoze.bfg.events module documentation for more information.
Bug Fixes
  • When adding a view for a route which did not yet exist (“did not yet exist” meaning, temporally, a view was added with a route name for a route which had not yet been added via add_route), the value of the custom_predicate argument to add_view was lost. Symptom: wrong view matches when using URL dispatch and custom view predicates together.
  • Pattern matches for a :segment marker in a URL dispatch route pattern now always match at least one character. See “Backwards Incompatibilities” below in this changelog.
Backwards Incompatibilities
  • A bug existed in the regular expression to do URL matching. As an example, the URL matching machinery would cause the pattern /{foo} to match the root URL / resulting in a match dictionary of {'foo':u''} or the pattern /{fud}/edit might match the URL ``//edit resulting in a match dictionary of {'fud':u''}. It was always the intent that :segment markers in the pattern would need to match at least one character, and never match the empty string. This, however, means that in certain circumstances, a routing match which your application inadvertently depended upon may no longer happen.
Documentation
  • Added description of the repoze.bfg.events.subscriber decorator to the Events narrative chapter.
  • Added repoze.bfg.events.subscriber API documentation to repoze.bfg.events API docs.
  • Added a section named “Zope 3 Enforces ‘TTW’ Authorization Checks By Default; BFG Does Not” to the “Design Defense” chapter.

1.3a6 (2010-07-25)

Features

  • New argument to repoze.bfg.configuration.Configurator.add_route and the route ZCML directive: traverse. If you would like to cause the context to be something other than the root object when this route matches, you can spell a traversal pattern as the traverse argument. This traversal pattern will be used as the traversal path: traversal will begin at the root object implied by this route (either the global root, or the object returned by the factory associated with this route).

    The syntax of the traverse argument is the same as it is for path. For example, if the path provided is articles/:article/edit, and the traverse argument provided is /:article, when a request comes in that causes the route to match in such a way that the article match value is ‘1’ (when the request URI is /articles/1/edit), the traversal path will be generated as /1. This means that the root object’s __getitem__ will be called with the name 1 during the traversal phase. If the 1 object exists, it will become the context of the request. The Traversal narrative has more information about traversal.

    If the traversal path contains segment marker names which are not present in the path argument, a runtime error will occur. The traverse pattern should not contain segment markers that do not exist in the path.

    A similar combining of routing and traversal is available when a route is matched which contains a *traverse remainder marker in its path. The traverse argument allows you to associate route patterns with an arbitrary traversal path without using a a *traverse remainder marker; instead you can use other match information.

    Note that the traverse argument is ignored when attached to a route that has a *traverse remainder marker in its path.

  • A new method of the Configurator exists: set_request_factory. If used, this method will set the factory used by the repoze.bfg router to create all request objects.

  • The Configurator constructor takes an additional argument: request_factory. If used, this argument will set the factory used by the repoze.bfg router to create all request objects.

  • The Configurator constructor takes an additional argument: request_factory. If used, this argument will set the factory used by the repoze.bfg router to create all request objects.

  • A new method of the Configurator exists: set_renderer_globals_factory. If used, this method will set the factory used by the repoze.bfg router to create renderer globals.

  • A new method of the Configurator exists: get_settings. If used, this method will return the current settings object (performs the same job as the repoze.bfg.settings.get_settings API).

  • The Configurator constructor takes an additional argument: renderer_globals_factory. If used, this argument will set the factory used by the repoze.bfg router to create renderer globals.

  • Add repoze.bfg.renderers.render, repoze.bfg.renderers.render_to_response and repoze.bfg.renderers.get_renderer functions. These are imperative APIs which will use the same rendering machinery used by view configurations with a renderer= attribute/argument to produce a rendering or renderer. Because these APIs provide a central API for all rendering, they now form the preferred way to perform imperative template rendering. Using functions named render_* from modules such as repoze.bfg.chameleon_zpt and repoze.bfg.chameleon_text is now discouraged (although not deprecated). The code the backing older templating-system-specific APIs now calls into the newer repoze.bfg.renderer code.

  • The repoze.bfg.configuration.Configurator.testing_add_template has been renamed to testing_add_renderer. A backwards compatibility alias is present using the old name.

Documentation
  • The Hybrid narrative chapter now contains a description of the traverse route argument.
  • The Hooks narrative chapter now contains sections about changing the request factory and adding a renderer globals factory.
  • The API documentation includes a new module: repoze.bfg.renderers.
  • The Templates chapter was updated; all narrative that used templating-specific APIs within examples to perform rendering (such as the repoze.bfg.chameleon_zpt.render_template_to_response method) was changed to use repoze.bfg.renderers.render_* functions.
Bug Fixes
  • The header predicate (when used as either a view predicate or a route predicate) had a problem when specified with a name/regex pair. When the header did not exist in the headers dictionary, the regex match could be fed None, causing it to throw a TypeError: expected string or buffer exception. Now, the predicate returns False as intended.
Deprecations

  • The repoze.bfg.renderers.rendered_response function was never an official API, but may have been imported by extensions in the wild. It is officially deprecated in this release. Use repoze.bfg.renderers.render_to_response instead.

  • The following APIs are documentation deprecated (meaning they are officially deprecated in documentation but do not raise a deprecation error upon their usage, and may continue to work for an indefinite period of time):

    In the repoze.bfg.chameleon_zpt module: get_renderer, get_template, render_template, render_template_to_response. The suggested alternatives are documented within the docstrings of those methods (which are still present in the documentation).

    In the repoze.bfg.chameleon_text module: get_renderer, get_template, render_template, render_template_to_response. The suggested alternatives are documented within the docstrings of those methods (which are still present in the documentation).

    In general, to perform template-related functions, one should now use the various methods in the repoze.bfg.renderers module.

Backwards Incompatibilities

  • A new internal exception class (not an API) named repoze.bfg.exceptions.PredicateMismatch now exists. This exception is currently raised when no constituent view of a multiview can be called (due to no predicate match). Previously, in this situation, a repoze.bfg.exceptions.NotFound was raised. We provide backwards compatibility for code that expected a NotFound to be raised when no predicates match by causing repoze.bfg.exceptions.PredicateMismatch to inherit from NotFound. This will cause any exception view registered for NotFound to be called when a predicate mismatch occurs, as was the previous behavior.

    There is however, one perverse case that will expose a backwards incompatibility. If 1) you had a view that was registered as a member of a multiview 2) this view explicitly raised a NotFound exception in order to proceed to the next predicate check in the multiview, that code will now behave differently: rather than skipping to the next view match, a NotFound will be raised to the top-level exception handling machinery instead. For code to be depending upon the behavior of a view raising NotFound to proceed to the next predicate match, would be tragic, but not impossible, given that NotFound is a public interface. repoze.bfg.exceptions.PredicateMismatch is not a public API and cannot be depended upon by application code, so you should not change your view code to raise PredicateMismatch. Instead, move the logic which raised the NotFound exception in the view out into a custom view predicate.

  • If, when you run your application’s unit test suite under BFG 1.3, a KeyError naming a template or a ValueError indicating that a ‘renderer factory’ is not registered may is raised (e.g. ValueError: No factory for renderer named '.pt' when looking up karl.views:templates/snippets.pt), you may need to perform some extra setup in your test code.

    The best solution is to use the repoze.bfg.configuration.Configurator.testing_add_renderer (or, alternately the deprecated repoze.bfg.testing.registerTemplateRenderer or registerDummyRenderer) API within the code comprising each individual unit test suite to register a “dummy” renderer for each of the templates and renderers used by code under test. For example:

    config = Configurator()
config.testing_add_renderer('karl.views:templates/snippets.pt')

    This will register a basic dummy renderer for this particular missing template. The testing_add_renderer API actually returns the renderer, but if you don’t care about how the render is used, you don’t care about having a reference to it either.

    A more rough way to solve the issue exists. It causes the “real” template implementations to be used while the system is under test, which is suboptimal, because tests will run slower, and unit tests won’t actually be unit tests, but it is easier. Always ensure you call the setup_registry() method of the Configurator . Eg:

    reg = MyRegistry()
config = Configurator(registry=reg)
config.setup_registry()

    Calling setup_registry only has an effect if you’re passing in a registry argument to the Configurator constructor. setup_registry is called by the course of normal operations anyway if you do not pass in a registry.

    If your test suite isn’t using a Configurator yet, and is still using the older repoze.bfg.testing APIs name setUp or cleanUp, these will register the renderers on your behalf.

    A variant on the symptom for this theme exists: you may already be dutifully registering a dummy template or renderer for a template used by the code you’re testing using testing_register_renderer or registerTemplateRenderer, but (perhaps unbeknownst to you) the code under test expects to be able to use a “real” template renderer implementation to retrieve or render another template that you forgot was being rendered as a side effect of calling the code you’re testing. This happened to work because it found the real template while the system was under test previously, and now it cannot. The solution is the same.

    It may also help reduce confusion to use a resource specification to specify the template path in the test suite and code rather than a relative path in either. A resource specification is unambiguous, while a relative path needs to be relative to “here”, where “here” isn’t always well-defined (“here” in a test suite may or may not be the same as “here” in the code under test).

1.3a5 (2010-07-14)

Features
  • New internal exception: repoze.bfg.exceptions.URLDecodeError. This URL is a subclass of the built-in Python exception named UnicodeDecodeError.
  • When decoding a URL segment to Unicode fails, the exception raised is now repoze.bfg.exceptions.URLDecodeError instead of UnicodeDecodeError. This makes it possible to register an exception view invoked specifically when repoze.bfg cannot decode a URL.
Bug Fixes
  • Fix regression in repoze.bfg.configuration.Configurator.add_static_view. Before 1.3a4, view names that contained a slash were supported as route prefixes. 1.3a4 broke this by trying to treat them as full URLs.
Documentation
  • The repoze.bfg.exceptions.URLDecodeError exception was added to the exceptions chapter of the API documentation.
Backwards Incompatibilities
  • in previous releases, when a URL could not be decoded from UTF-8 during traversal, a TypeError was raised. Now the error which is raised is a repoze.bfg.exceptions.URLDecodeError.

1.3a4 (2010-07-03)

Features

  • Undocumented hook: make get_app and get_root of the repoze.bfg.paster.BFGShellCommand hookable in cases where endware may interfere with the default versions.

  • In earlier versions, a custom route predicate associated with a url dispatch route (each of the predicate functions fed to the custom_predicates argument of repoze.bfg.configuration.Configurator.add_route) has always required a 2-positional argument signature, e.g. (context, request). Before this release, the context argument was always None.

    As of this release, the first argument passed to a predicate is now a dictionary conventionally named info consisting of route, and match. match is a dictionary: it represents the arguments matched in the URL by the route. route is an object representing the route which was matched.

    This is useful when predicates need access to the route match. For example:

    def any_of(segment_name, *args):
    def predicate(info, request):
        if info['match'][segment_name] in args:
            return True
    return predicate

num_one_two_or_three = any_of('num, 'one', 'two', 'three')

add_route('num', '/:num', custom_predicates=(num_one_two_or_three,))

    The route object is an object that has two useful attributes: name and path. The name attribute is the route name. The path attribute is the route pattern. An example of using the route in a set of route predicates:

    def twenty_ten(info, request):
    if info['route'].name in ('ymd', 'ym', 'y'):
        return info['match']['year'] == '2010'

add_route('y', '/:year', custom_predicates=(twenty_ten,))
add_route('ym', '/:year/:month', custom_predicates=(twenty_ten,))
add_route('ymd', '/:year/:month:/day', custom_predicates=(twenty_ten,))

  • The repoze.bfg.url.route_url API has changed. If a keyword _app_url is present in the arguments passed to route_url, this value will be used as the protocol/hostname/port/leading path prefix of the generated URL. For example, using an _app_url of http://example.com:8080/foo would cause the URL http://example.com:8080/foo/fleeb/flub to be returned from this function if the expansion of the route pattern associated with the route_name expanded to /fleeb/flub.

  • It is now possible to use a URL as the name argument fed to repoze.bfg.configuration.Configurator.add_static_view. When the name argument is a URL, the repoze.bfg.url.static_url API will generate join this URL (as a prefix) to a path including the static file name. This makes it more possible to put static media on a separate webserver for production, while keeping static media package-internal and served by the development webserver during development.

Documentation
  • The authorization chapter of the ZODB Wiki Tutorial (docs/tutorials/bfgwiki) was changed to demonstrate authorization via a group rather than via a direct username (thanks to Alex Marandon).
  • The authorization chapter of the SQLAlchemy Wiki Tutorial (docs/tutorials/bfgwiki2) was changed to demonstrate authorization via a group rather than via a direct username.
  • Redirect requests for tutorial sources to http://docs.repoze.org/bfgwiki-1.3 and http://docs.repoze.org/bfgwiki2-1.3/ respectively.
  • A section named Custom Route Predicates was added to the URL Dispatch narrative chapter.
  • The Static Resources chapter has been updated to mention using static_url to generate URLs to external webservers.
Internal
  • Removed repoze.bfg.static.StaticURLFactory in favor of a new abstraction revolving around the (still-internal) repoze.bfg.static.StaticURLInfo helper class.

1.3a3 (2010-05-01)

Paster Templates
  • The bfg_alchemy and bfg_routesalchemy templates no longer register a handle_teardown event listener which calls DBSession.remove. This was found by Chris Withers to be unnecessary.
Documentation
  • The “bfgwiki2” (URL dispatch wiki) tutorial code and documentation was changed to remove the handle_teardown event listener which calls DBSession.remove.
  • Any mention of the handle_teardown event listener as used by the paster templates was removed from the URL Dispatch narrative chapter.
  • A section entitled Detecting Available Languages was added to the i18n narrative docs chapter.

1.3a2 (2010-04-28)

Features
  • A locale negotiator no longer needs to be registered explicitly. The default locale negotiator at repoze.bfg.i18n.default_locale_negotiator is now used unconditionally as... um, the default locale negotiator.
  • The default locale negotiator has become more complex.
    • First, the negotiator looks for the _LOCALE_ attribute of the request object (possibly set by a view or an event listener).
    • Then it looks for the request.params['_LOCALE_'] value.
    • Then it looks for the request.cookies['_LOCALE_'] value.
Backwards Incompatibilities
  • The default locale negotiator now looks for the parameter named _LOCALE_ rather than a parameter named locale in request.params.
Behavior Changes
  • A locale negotiator may now return None, signifying that the default locale should be used.
Documentation
  • Documentation concerning locale negotiation in the Internationalizationa and Localization chapter was updated.
  • Expanded portion of i18n narrative chapter docs which discuss working with gettext files.

1.3a1 (2010-04-26)

Features

  • Added “exception views”. When you use an exception (anything that inherits from the Python Exception builtin) as view context argument, e.g.:

    from repoze.bfg.view import bfg_view
from repoze.bfg.exceptions import NotFound
from webob.exc import HTTPNotFound

@bfg_view(context=NotFound)
def notfound_view(request):
    return HTTPNotFound()

    For the above example, when the repoze.bfg.exceptions.NotFound exception is raised by any view or any root factory, the notfound_view view callable will be invoked and its response returned.

    Other normal view predicates can also be used in combination with an exception view registration:

    from repoze.bfg.view import bfg_view
from repoze.bfg.exceptions import NotFound
from webob.exc import HTTPNotFound

@bfg_view(context=NotFound, route_name='home')
def notfound_view(request):
    return HTTPNotFound()

    The above exception view names the route_name of home, meaning that it will only be called when the route matched has a name of home. You can therefore have more than one exception view for any given exception in the system: the “most specific” one will be called when the set of request circumstances which match the view registration. The only predicate that cannot be not be used successfully is name. The name used to look up an exception view is always the empty string.

    Existing (pre-1.3) normal views registered against objects inheriting from Exception will continue to work. Exception views used for user-defined exceptions and system exceptions used as contexts will also work.

    The feature can be used with any view registration mechanism (@bfg_view decorator, ZCML, or imperative config.add_view styles).

    This feature was kindly contributed by Andrey Popp.

  • Use “Venusian” (http://docs.repoze.org/venusian) to perform bfg_view decorator scanning rather than relying on a BFG-internal decorator scanner. (Truth be told, Venusian is really just a generalization of the BFG-internal decorator scanner).

  • Internationalization and localization features as documented in the narrative documentation chapter entitled Internationalization and Localization.

  • A new deployment setting named default_locale_name was added. If this string is present as a Paster .ini file option, it will be considered the default locale name. The default locale name is used during locale-related operations such as language translation.

  • It is now possible to turn on Chameleon template “debugging mode” for all Chameleon BFG templates by setting a BFG-related Paster .ini file setting named debug_templates. The exceptions raised by Chameleon templates when a rendering fails are sometimes less than helpful. debug_templates allows you to configure your application development environment so that exceptions generated by Chameleon during template compilation and execution will contain more helpful debugging information. This mode is on by default in all new projects.

  • Add a new method of the Configurator named derive_view which can be used to generate a BFG view callable from a user-supplied function, instance, or class. This useful for external framework and plugin authors wishing to wrap callables supplied by their users which follow the same calling conventions and response conventions as objects that can be supplied directly to BFG as a view callable. See the derive_view method in the repoze.bfg.configuration.Configurator docs.

ZCML
  • Add a translationdir ZCML directive to support localization.
  • Add a localenegotiator ZCML directive to support localization.
Deprecations
  • The exception views feature replaces the need for the set_notfound_view and set_forbidden_view methods of the Configurator as well as the notfound and forbidden ZCML directives. Those methods and directives will continue to work for the foreseeable future, but they are deprecated in the documentation.
Dependencies
  • A new install-time dependency on the venusian distribution was added.
  • A new install-time dependency on the translationstring distribution was added.
  • Chameleon 1.2.3 or better is now required (internationalization and per-template debug settings).
Internal
  • View registrations and lookups are now done with three “requires” arguments instead of two to accomodate orthogonality of exception views.
  • The repoze.bfg.interfaces.IForbiddenView and repoze.bfg.interfaces.INotFoundView interfaces were removed; they weren’t APIs and they became vestigial with the addition of exception views.
  • Remove repoze.bfg.compat.pkgutil_26.py and import alias repoze.bfg.compat.walk_packages. These were only required by internal scanning machinery; Venusian replaced the internal scanning machinery, so these are no longer required.
Documentation
  • Exception view documentation was added to the Hooks narrative chapter.
  • A new narrative chapter entitled Internationalization and Localization was added.
  • The “Environment Variables and ini File Settings” chapter was changed: documentation about the default_locale_name setting was added.
  • A new API chapter for the repoze.bfg.i18n module was added.
  • Documentation for the new translationdir and localenegotiator ZCML directives were added.
  • A section was added to the Templates chapter entitled “Nicer Exceptions in Templates” describing the result of setting debug_templates = true.
Paster Templates
  • All paster templates now create a setup.cfg which includes commands related to nose testing and Babel message catalog extraction/compilation.
  • A default_locale_name = en setting was added to each existing paster template.
  • A debug_templates = true setting was added to each existing paster template.
Licensing
  • The Edgewall (BSD) license was added to the LICENSES.txt file, as some code in the repoze.bfg.i18n derives from Babel source.

1.2 (2010-02-10)

  • No changes from 1.2b6.

1.2b6 (2010-02-06)

Backwards Incompatibilities
  • Remove magical feature of repoze.bfg.url.model_url which prepended a fully-expanded urldispatch route URL before a the model’s path if it was noticed that the request had matched a route. This feature was ill-conceived, and didn’t work in all scenarios.
Bug Fixes
  • More correct conversion of provided renderer values to resource specification values (internal).

1.2b5 (2010-02-04)

Bug Fixes
  • 1.2b4 introduced a bug whereby views added via a route configuration that named a view callable and also a view_attr became broken. Symptom: MyViewClass is not callable or the __call__ of a class was being called instead of the method named via view_attr.
  • Fix a bug whereby a renderer argument to the @bfg_view decorator that provided a package-relative template filename might not have been resolved properly. Symptom: inappropriate Missing template resource errors.

1.2b4 (2010-02-03)

Documentation
  • Update GAE tutorial to use Chameleon instead of Jinja2 (now that it’s possible).
Bug Fixes
  • Ensure that secure flag for AuthTktAuthenticationPolicy constructor does what it’s documented to do (merge Daniel Holth’s fancy-cookies-2 branch).
Features
  • Add path and http_only options to AuthTktAuthenticationPolicy constructor (merge Daniel Holth’s fancy-cookies-2 branch).
Backwards Incompatibilities
  • Remove view_header, view_accept, view_xhr, view_path_info, view_request_method, view_request_param, and view_containment predicate arguments from the Configurator.add_route argument list. These arguments were speculative. If you need the features exposed by these arguments, add a view associated with a route using the route_name argument to the add_view method instead.
  • Remove view_header, view_accept, view_xhr, view_path_info, view_request_method, view_request_param, and view_containment predicate arguments from the route ZCML directive attribute set. These attributes were speculative. If you need the features exposed by these attributes, add a view associated with a route using the route_name attribute of the view ZCML directive instead.
Dependencies
  • Remove dependency on sourcecodegen (not depended upon by Chameleon 1.1.1+).

1.2b3 (2010-01-24)

Bug Fixes
  • When “hybrid mode” (both traversal and urldispatch) is in use, default to finding route-related views even if a non-route-related view registration has been made with a more specific context. The default used to be to find views with a more specific context first. Use the new use_global_views argument to the route definition to get back the older behavior.
Features
  • Add use_global_views argument to add_route method of Configurator. When this argument is true, views registered for no route will be found if no more specific view related to the route is found.
  • Add use_global_views attribute to ZCML <route> directive (see above).
Internal
  • When registering a view, register the view adapter with the “requires” interfaces as (request_type, context_type) rather than (context_type, request_type). This provides for saner lookup, because the registration will always be made with a specific request interface, but registration may not be made with a specific context interface. In general, when creating multiadapters, you want to order the requires interfaces so that the the elements which are more likely to be registered using specific interfaces are ordered before those which are less likely.

1.2b2 (2010-01-21)

Bug Fixes
  • When the Configurator is passed an instance of zope.component.registry.Components as a registry constructor argument, fix the instance up to have the attributes we expect of an instance of repoze.bfg.registry.Registry when setup_registry is called. This makes it possible to use the global Zope component registry as a BFG application registry.
  • When WebOb 0.9.7.1 was used, a deprecation warning was issued for the class attribute named charset within repoze.bfg.request.Request. BFG now requires WebOb >= 0.9.7, and code was added so that this deprecation warning has disappeared.
  • Fix a view lookup ordering bug whereby a view with a larger number of predicates registered first (literally first, not “earlier”) for a triad would lose during view lookup to one registered with fewer.
  • Make sure views with exactly N custom predicates are always called before views with exactly N non-custom predicates given all else is equal in the view configuration.
Documentation
  • Change renderings of ZCML directive documentation.
  • Add a narrative documentation chapter: “Using the Zope Component Architecture in repoze.bfg”.
Dependencies
  • Require WebOb >= 0.9.7

1.2b1 (2010-01-18)

Bug Fixes
  • In bfg_routesalchemy, bfg_alchemy paster templates and the bfgwiki2 tutorial, clean up the SQLAlchemy connection by registering a repoze.tm.after_end callback instead of relying on a __del__ method of a Cleanup class added to the WSGI environment. The __del__ strategy was fragile and caused problems in the wild. Thanks to Daniel Holth for testing.
Features
  • Read logging configuration from PasteDeploy config file loggers section (and related) when paster bfgshell is invoked.
Documentation
  • Major rework in preparation for book publication.

1.2a11 (2010-01-05)

Bug Fixes
  • Make paster bfgshell and paster create -t bfg_xxx work on Jython (fix minor incompatibility with treatment of __doc__ at the class level).
  • Updated dependency on WebOb to require a version which supports features now used in tests.
Features
  • Jython compatibility (at least when repoze.bfg.jinja2 is used as the templating engine; Chameleon does not work under Jython).
  • Show the derived abspath of template resource specifications in the traceback when a renderer template cannot be found.
  • Show the original traceback when a Chameleon template cannot be rendered due to a platform incompatibility.

1.2a10 (2010-01-04)

Features
  • The Configurator.add_view method now accepts an argument named context. This is an alias for the older argument named for_; it is preferred over for_, but for_ will continue to be supported “forever”.
  • The view ZCML directive now accepts an attribute named context. This is an alias for the older attribute named for; it is preferred over for, but for will continue to be supported “forever”.
  • The Configurator.add_route method now accepts an argument named view_context. This is an alias for the older argument named view_for; it is preferred over view_for, but view_for will continue to be supported “forever”.
  • The route ZCML directive now accepts an attribute named view_context. This is an alias for the older attribute named view_for; it is preferred over view_for, but view_for will continue to be supported “forever”.
Documentation and Paster Templates
  • LaTeX rendering tweaks.
  • All uses of the Configurator.add_view method that used its for_ argument now use the context argument instead.
  • All uses of the Configurator.add_route method that used its view_for argument now use the view_context argument instead.
  • All uses of the view ZCML directive that used its for attribute now use the context attribute instead.
  • All uses of the route ZCML directive that used its view_for attribute now use the view_context attribute instead.
  • Add a (minimal) tutorial dealing with use of repoze.catalog in a repoze.bfg application.
Documentation Licensing

1.2a9 (2009-12-27)

Documentation Licensing
  • The documentation (the result of make <html|latex|htmlhelp> within the docs directory) in this release is now offered under the Creative Commons Attribution-Noncommercial-No Derivative Works 3.0 United States License as described by http://creativecommons.org/licenses/by-nc-nd/3.0/us/ . This is only a licensing change for the documentation; the repoze.bfg software continues to be offered under the Repoze Public License at http://repoze.org/license.html (BSD-like).
Documentation
  • Added manual index entries to generated index.
  • Document the previously existing (but non-API) repoze.bfg.configuration.Configurator.setup_registry method as an official API of a Configurator.
  • Fix syntax errors in various documentation code blocks.
  • Created new top-level documentation section: “ZCML Directives”. This section contains detailed ZCML directive information, some of which was removed from various narrative chapters.
  • The LaTeX rendering of the documentation has been improved.
  • Added a “Fore-Matter” section with author, copyright, and licensing information.

1.2a8 (2009-12-24)

Features
  • Add a **kw arg to the Configurator.add_settings API.
  • Add hook_zca and unhook_zca methods to the Configurator API.
  • The repoze.bfg.testing.setUp method now returns a Configurator instance which can be used to do further configuration during unit tests.
Bug Fixes
  • The json renderer failed to set the response content type to application/json. It now does, by setting request.response_content_type unless this attribute is already set.
  • The string renderer failed to set the response content type to text/plain. It now does, by setting request.response_content_type unless this attribute is already set.
Documentation
  • General documentation improvements by using better Sphinx roles such as “class”, “func”, “meth”, and so on. This means that there are many more hyperlinks pointing to API documentation for API definitions in all narrative, tutorial, and API documentation elements.
  • Added a description of imperative configuration in various places which only described ZCML configuration.
  • A syntactical refreshing of various tutorials.
  • Added the repoze.bfg.authentication, repoze.bfg.authorization, and repoze.bfg.interfaces modules to API documentation.
Deprecations
  • The repoze.bfg.testing.registerRoutesMapper API (added in an early 1.2 alpha) was deprecated. Its import now generates a deprecation warning.

1.2a7 (2009-12-20)

Features
  • Add four new testing-related APIs to the repoze.bfg.configuration.Configurator class: testing_securitypolicy, testing_models, testing_add_subscriber, and testing_add_template. These were added in order to provide more direct access to the functionality of the repoze.bfg.testing APIs named registerDummySecurityPolicy, registerModels, registerEventListener, and registerTemplateRenderer when a configurator is used. The testing APIs named are nominally deprecated (although they will likely remain around “forever”, as they are in heavy use in the wild).
  • Add a new API to the repoze.bfg.configuration.Configurator class: add_settings. This API can be used to add “settings” (information returned within via the repoze.bfg.settings.get_settings API) after the configurator has been initially set up. This is most useful for testing purposes.
  • Add a custom_predicates argument to the Configurator add_view method, the bfg_view decorator and the attribute list of the ZCML view directive. If custom_predicates is specified, it must be a sequence of predicate callables (a predicate callable accepts two arguments: context and request and returns True or False). The associated view callable will only be invoked if all custom predicates return True. Use one or more custom predicates when no existing predefined predicate is useful. Predefined and custom predicates can be mixed freely.
  • Add a custom_predicates argument to the Configurator add_route and the attribute list of the ZCML route directive. If custom_predicates is specified, it must be a sequence of predicate callables (a predicate callable accepts two arguments: context and request and returns True or False). The associated route will match will only be invoked if all custom predicates return True, else route matching continues. Note that the value context will always be None when passed to a custom route predicate. Use one or more custom predicates when no existing predefined predicate is useful. Predefined and custom predicates can be mixed freely.
Internal
  • Remove the repoze.bfg.testing.registerTraverser function. This function was never an API.
Documenation
  • Doc-deprecated most helper functions in the repoze.bfg.testing module. These helper functions likely won’t be removed any time soon, nor will they generate a warning any time soon, due to their heavy use in the wild, but equivalent behavior exists in methods of a Configurator.

1.2a6 (2009-12-18)

Features

  • The Configurator object now has two new methods: begin and end. The begin method is meant to be called before any “configuration” begins (e.g. before add_view, et. al are called). The end method is meant to be called after all “configuration” is complete.

    Previously, before there was imperative configuration at all (1.1 and prior), configuration begin and end was invariably implied by the process of loading a ZCML file. When a ZCML load happened, the threadlocal data structure containing the request and registry was modified before the load, and torn down after the load, making sure that all framework code that needed get_current_registry for the duration of the ZCML load was satisfied.

    Some API methods called during imperative configuration, (such as Configurator.add_view when a renderer is involved) end up for historical reasons calling get_current_registry. However, in 1.2a5 and below, the Configurator supplied no functionality that allowed people to make sure that get_current_registry returned the registry implied by the configurator being used. begin now serves this purpose. Inversely, end pops the thread local stack, undoing the actions of begin.

    We make this boundary explicit to reduce the potential for confusion when the configurator is used in different circumstances (e.g. in unit tests and app code vs. just in initial app setup).

    Existing code written for 1.2a1-1.2a5 which does not call begin or end continues to work in the same manner it did before. It is however suggested that this code be changed to call begin and end to reduce the potential for confusion in the future.

  • All paster templates which generate an application skeleton now make use of the new begin and end methods of the Configurator they use in their respective copies of run.py and tests.py.

Documentation
  • All documentation that makes use of a Configurator object to do application setup and test setup now makes use of the new begin and end methods of the configurator.
Bug Fixes
  • When a repoze.bfg.exceptions.NotFound or repoze.bfg.exceptions.Forbidden class (as opposed to instance) was raised as an exception within a root factory (or route root factory), the exception would not be caught properly by the repoze.bfg. Router and it would propagate to up the call stack, as opposed to rendering the not found view or the forbidden view as would have been expected.
  • When Chameleon page or text templates used as renderers were added imperatively (via Configurator.add_view or some derivative), they too-eagerly attempted to look up the reload_templates setting via get_settings, meaning they were always registered in non-auto-reload-mode (the default). Each now waits until its respective template attribute is accessed to look up the value.
  • When a route with the same name as a previously registered route was added, the old route was not removed from the mapper’s routelist. Symptom: the old registered route would be used (and possibly matched) during route lookup when it should not have had a chance to ever be used.

1.2a5 (2009-12-10)

Features
  • When the repoze.bfg.exceptions.NotFound or repoze.bfg.exceptions.Forbidden error is raised from within a custom root factory or the factory of a route, the appropriate response is now sent to the requesting user agent (the result of the notfound view or the forbidden view, respectively). When these errors are raised from within a root factory, the context passed to the notfound or forbidden view will be None. Also, the request will not be decorated with view_name, subpath, context, etc. as would normally be the case if traversal had been allowed to take place.
Internals
  • The exception class representing the error raised by various methods of a Configurator is now importable as repoze.bfg.exceptions.ConfigurationError.
Documentation
  • General documentation freshening which takes imperative configuration into account in more places and uses glossary references more liberally.
  • Remove explanation of changing the request type in a new request event subscriber, as other predicates are now usually an easier way to get this done.
  • Added “Thread Locals” narrative chapter to documentation, and added a API chapter documenting the repoze.bfg.threadlocals module.
  • Added a “Special Exceptions” section to the “Views” narrative documentation chapter explaining the effect of raising repoze.bfg.exceptions.NotFound and repoze.bfg.exceptions.Forbidden from within view code.
Dependencies
  • A new dependency on the twill package was added to the setup.py tests_require argument (Twill will only be downloaded when repoze.bfg setup.py test or setup.py nosetests is invoked).

1.2a4 (2009-12-07)

Features
  • repoze.bfg.testing.DummyModel now accepts a new constructor keyword argument: __provides__. If this constructor argument is provided, it should be an interface or a tuple of interfaces. The resulting model will then provide these interfaces (they will be attached to the constructed model via zope.interface.alsoProvides).
Bug Fixes
  • Operation on GAE was broken, presumably because the repoze.bfg.configuration module began to attempt to import the repoze.bfg.chameleon_zpt and repoze.bfg.chameleon_text modules, and these cannot be used on non-CPython platforms. It now tolerates startup time import failures for these modules, and only raise an import error when a template from one of these packages is actually used.

1.2a3 (2009-12-02)

Bug Fixes
  • The repoze.bfg.url.route_url function inappropriately passed along _query and/or _anchor arguments to the mapper.generate function, resulting in blowups.
  • When two views were registered with differering for interfaces or classes, and the for of first view registered was a superclass of the second, the repoze.bfg view machinery would incorrectly associate the two views with the same “multiview”. Multiviews are meant to be collections of views that have exactly the same for/request/viewname values, without taking inheritance into account. Symptom: wrong view callable found even when you had correctly specified a for_ interface/class during view configuration for one or both view configurations.
Backwards Incompatibilities
  • The repoze.bfg.templating module has been removed; it had been deprecated in 1.1 and never actually had any APIs in it.

1.2a2 (2009-11-29)

Bug Fixes
  • The the long description of this package (as shown on PyPI) was not valid reStructuredText, and so was not renderable.
  • Trying to use an HTTP method name string such as GET as a request_type predicate argument caused a startup time failure when it was encountered in imperative configuration or in a decorator (symptom: Type Error: Required specification must be a specification). This now works again, although request_method is now the preferred predicate argument for associating a view configuration with an HTTP request method.
Documentation
  • Fixed “Startup” narrative documentation chapter; it was explaining “the old way” an application constructor worked.

1.2a1 (2009-11-28)

Features

  • An imperative configuration mode.

    A repoze.bfg application can now begin its life as a single Python file. Later, the application might evolve into a set of Python files in a package. Even later, it might start making use of other configuration features, such as ZCML. But neither the use of a package nor the use of non-imperative configuration is required to create a simple repoze.bfg application any longer.

    Imperative configuration makes repoze.bfg competetive with “microframeworks” such as Bottle and Tornado. repoze.bfg has a good deal of functionality that most microframeworks lack, so this is hopefully a “best of both worlds” feature.

    The simplest possible repoze.bfg application is now:

    from webob import Response
from wsgiref import simple_server
from repoze.bfg.configuration import Configurator

def hello_world(request):
    return Response('Hello world!')

if __name__ == '__main__':
    config = Configurator()
    config.add_view(hello_world)
    app = config.make_wsgi_app()
    simple_server.make_server('', 8080, app).serve_forever()

  • A new class now exists: repoze.bfg.configuration.Configurator. This class forms the basis for sharing machinery between “imperatively” configured applications and traditional declaratively-configured applications.

  • The repoze.bfg.testing.setUp function now accepts three extra optional keyword arguments: registry, request and hook_zca.

    If the registry argument is not None, the argument will be treated as the registry that is set as the “current registry” (it will be returned by repoze.bfg.threadlocal.get_current_registry) for the duration of the test. If the registry argument is None (the default), a new registry is created and used for the duration of the test.

    The value of the request argument is used as the “current request” (it will be returned by repoze.bfg.threadlocal.get_current_request) for the duration of the test; it defaults to None.

    If hook_zca is True (the default), the zope.component.getSiteManager function will be hooked with a function that returns the value of registry (or the default-created registry if registry is None) instead of the registry returned by zope.component.getGlobalSiteManager, causing the Zope Component Architecture API (getSiteManager, getAdapter, getUtility, and so on) to use the testing registry instead of the global ZCA registry.

  • The repoze.bfg.testing.tearDown function now accepts an unhook_zca argument. If this argument is True (the default), zope.component.getSiteManager.reset() will be called. This will cause the result of the zope.component.getSiteManager function to be the global ZCA registry (the result of zope.component.getGlobalSiteManager) once again.

  • The run.py module in various repoze.bfg paster templates now use a repoze.bfg.configuration.Configurator class instead of the (now-legacy) repoze.bfg.router.make_app function to produce a WSGI application.

Documentation
  • The documentation now uses the “request-only” view calling convention in most examples (as opposed to the context, request convention). This is a documentation-only change; the context, request convention is also supported and documented, and will be “forever”.
  • repoze.bfg.configuration API documentation has been added.
  • A narrative documentation chapter entitled “Creating Your First repoze.bfg Application” has been added. This chapter details usage of the new repoze.bfg.configuration.Configurator class, and demonstrates a simplified “imperative-mode” configuration; doing repoze.bfg application configuration imperatively was previously much more difficult.
  • A narrative documentation chapter entitled “Configuration, Decorations and Code Scanning” explaining ZCML- vs. imperative- vs. decorator-based configuration equivalence.
  • The “ZCML Hooks” chapter has been renamed to “Hooks”; it documents how to override hooks now via imperative configuration and ZCML.
  • The explanation about how to supply an alternate “response factory” has been removed from the “Hooks” chapter. This feature may be removed in a later release (it still works now, it’s just not documented).
  • Add a section entitled “Test Set Up and Tear Down” to the unittesting chapter.
Bug Fixes
  • The ACL authorization policy debugging output when debug_authorization console debugging output was turned on wasn’t as clear as it could have been when a view execution was denied due to an authorization failure resulting from the set of principals passed never having matched any ACE in any ACL in the lineage. Now in this case, we report <default deny> as the ACE value and either the root ACL or <No ACL found on any object in model lineage> if no ACL was found.
  • When two views were registered with the same accept argument, but were otherwise registered with the same arguments, if a request entered the application which had an Accept header that accepted either of the media types defined by the set of views registered with predicates that otherwise matched, a more or less “random” one view would “win”. Now, we try harder to use the view callable associated with the view configuration that has the most specific accept argument. Thanks to Alberto Valverde for an initial patch.
Internals

  • The routes mapper is no longer a root factory wrapper. It is now consulted directly by the router.

  • The repoze.bfg.registry.make_registry callable has been removed.

  • The repoze.bfg.view.map_view callable has been removed.

  • The repoze.bfg.view.owrap_view callable has been removed.

  • The repoze.bfg.view.predicate_wrap callable has been removed.

  • The repoze.bfg.view.secure_view callable has been removed.

  • The repoze.bfg.view.authdebug_view callable has been removed.

  • The repoze.bfg.view.renderer_from_name callable has been removed. Use repoze.bfg.configuration.Configurator.renderer_from_name instead (still not an API, however).

  • The repoze.bfg.view.derive_view callable has been removed. Use repoze.bfg.configuration.Configurator.derive_view instead (still not an API, however).

  • The repoze.bfg.settings.get_options callable has been removed. Its job has been subsumed by the repoze.bfg.settings.Settings class constructor.

  • The repoze.bfg.view.requestonly function has been moved to repoze.bfg.configuration.requestonly.

  • The repoze.bfg.view.rendered_response function has been moved to repoze.bfg.configuration.rendered_response.

  • The repoze.bfg.view.decorate_view function has been moved to repoze.bfg.configuration.decorate_view.

  • The repoze.bfg.view.MultiView class has been moved to repoze.bfg.configuration.MultiView.

  • The repoze.bfg.zcml.Uncacheable class has been removed.

  • The repoze.bfg.resource.resource_spec function has been removed.

  • All ZCML directives which deal with attributes which are paths now use the path method of the ZCML context to resolve a relative name to an absolute one (imperative configuration requirement).

  • The repoze.bfg.scripting.get_root API now uses a ‘real’ WebOb request rather than a FakeRequest when it sets up the request as a threadlocal.

  • The repoze.bfg.traversal.traverse API now uses a ‘real’ WebOb request rather than a FakeRequest when it calls the traverser.

  • The repoze.bfg.request.FakeRequest class has been removed.

  • Most uses of the ZCA threadlocal API (the getSiteManager, getUtility, getAdapter, getMultiAdapter threadlocal API) have been removed from the core. Instead, when a threadlocal is necessary, the core uses the repoze.bfg.threadlocal.get_current_registry API to obtain the registry.

  • The internal ILogger utility named repoze.bfg.debug is now just an IDebugLogger unnamed utility. A named utility with the old name is registered for b/w compat.

  • The repoze.bfg.interfaces.ITemplateRendererFactory interface was removed; it has become unused.

  • Instead of depending on the martian package to do code scanning, we now just use our own scanning routines.

  • We now no longer have a dependency on repoze.zcml package; instead, the repoze.bfg package includes implementations of the adapter, subscriber and utility directives.

  • Relating to the following functions:

    repoze.bfg.view.render_view

    repoze.bfg.view.render_view_to_iterable

    repoze.bfg.view.render_view_to_response

    repoze.bfg.view.append_slash_notfound_view

    repoze.bfg.view.default_notfound_view

    repoze.bfg.view.default_forbidden_view

    repoze.bfg.configuration.rendered_response

    repoze.bfg.security.has_permission

    repoze.bfg.security.authenticated_userid

    repoze.bfg.security.effective_principals

    repoze.bfg.security.view_execution_permitted

    repoze.bfg.security.remember

    repoze.bfg.security.forget

    repoze.bfg.url.route_url

    repoze.bfg.url.model_url

    repoze.bfg.url.static_url

    repoze.bfg.traversal.virtual_root

    Each of these functions now expects to be called with a request object that has a registry attribute which represents the current repoze.bfg registry. They fall back to obtaining the registry from the threadlocal API.

Backwards Incompatibilites

  • Unit tests which use zope.testing.cleanup.cleanUp for the purpose of isolating tests from one another may now begin to fail due to lack of isolation between tests.

    Here’s why: In repoze.bfg 1.1 and prior, the registry returned by repoze.bfg.threadlocal.get_current_registry when no other registry had been pushed on to the threadlocal stack was the zope.component.globalregistry.base global registry (aka the result of zope.component.getGlobalSiteManager()). In repoze.bfg 1.2+, however, the registry returned in this situation is the new module-scope repoze.bfg.registry.global_registry object. The zope.testing.cleanup.cleanUp function clears the zope.component.globalregistry.base global registry unconditionally. However, it does not know about the repoze.bfg.registry.global_registry object, so it does not clear it.

    If you use the zope.testing.cleanup.cleanUp function in the setUp of test cases in your unit test suite instead of using the (more correct as of 1.1) repoze.bfg.testing.setUp, you will need to replace all calls to zope.testing.cleanup.cleanUp with a call to repoze.bfg.testing.setUp.

    If replacing all calls to zope.testing.cleanup.cleanUp with a call to repoze.bfg.testing.setUp is infeasible, you can put this bit of code somewhere that is executed exactly once (not for each test in a test suite; in the `` __init__.py`` of your package or your package’s tests subpackage would be a reasonable place):

    import zope.testing.cleanup
from repoze.bfg.testing import setUp
zope.testing.cleanup.addCleanUp(setUp)

  • When there is no “current registry” in the repoze.bfg.threadlocal.manager threadlocal data structure (this is the case when there is no “current request” or we’re not in the midst of a r.b.testing.setUp-bounded unit test), the .get method of the manager returns a data structure containing a global registry. In previous releases, this function returned the global Zope “base” registry: the result of zope.component.getGlobalSiteManager, which is an instance of the zope.component.registry.Component class. In this release, however, the global registry returns a globally importable instance of the repoze.bfg.registry.Registry class. This registry instance can always be imported as repoze.bfg.registry.global_registry.

    Effectively, this means that when you call repoze.bfg.threadlocal.get_current_registry when no request or setUp bounded unit test is in effect, you will always get back the global registry that lives in repoze.bfg.registry.global_registry. It also means that repoze.bfg APIs that call get_current_registry will use this registry.

    This change was made because repoze.bfg now expects the registry it uses to have a slightly different API than a bare instance of zope.component.registry.Components.

  • View registration no longer registers a repoze.bfg.interfaces.IViewPermission adapter (it is no longer checked by the framework; since 1.1, views have been responsible for providing their own security).

  • The repoze.bfg.router.make_app callable no longer accepts the authentication_policy nor the authorization_policy arguments. This feature was deprecated in version 1.0 and has been removed.

  • Obscure: the machinery which configured views with a request_type and a route_name would ignore the request interface implied by route_name registering a view only for the interface implied by request_type. In the unlikely event that you were trying to use these two features together, the symptom would have been that views that named a request_type but which were also associated with routes were not found when the route matched. Now if a view is configured with both a request_type and a route_name, an error is raised.

  • The route ZCML directive now no longer accepts the request_type or view_request_type attributes. These attributes didn’t actually work in any useful way (see entry above this one).

  • Because the repoze.bfg package now includes implementations of the adapter, subscriber and utility ZCML directives, it is now an error to have <include package="repoze.zcml" file="meta.zcml"/> in the ZCML of a repoze.bfg application. A ZCML conflict error will be raised if your ZCML does so. This shouldn’t be an issue for “normal” installations; it has always been the responsibility of the repoze.bfg.includes ZCML to include this file in the past; it now just doesn’t.

  • The repoze.bfg.testing.zcml_configure API was removed. Use the Configurator.load_zcml API instead.

Deprecations

  • The repoze.bfg.router.make_app function is now nominally deprecated. Its import and usage does not throw a warning, nor will it probably ever disappear. However, using a repoze.bfg.configuration.Configurator class is now the preferred way to generate a WSGI application.

    Note that make_app calls zope.component.getSiteManager.sethook( repoze.bfg.threadlocal.get_current_registry) on the caller’s behalf, hooking ZCA global API lookups, for backwards compatibility purposes. If you disuse make_app, your calling code will need to perform this call itself, at least if your application uses the ZCA global API (getSiteManager, getAdapter, etc).

Dependencies
  • A dependency on the martian package has been removed (its functionality is replaced internally).
  • A dependency on the repoze.zcml package has been removed (its functionality is replaced internally).

1.1.1 (2009-11-21)

Bug Fixes
  • “Hybrid mode” applications (applications which explicitly used traversal after url dispatch via <route> paths containing the *traverse element) were broken in 1.1-final and all 1.1 alpha and beta releases. Views registered without a route_name route shadowed views registered with a route_name inappropriately.

1.1 (2009-11-15)

Internals
  • Remove dead IRouteRequirement interface from repoze.bfg.zcml module.
Documentation
  • Improve the “Extending an Existing Application” narrative chapter.
  • Add more sections to the “Defending Design” chapter.

1.1b4 (2009-11-12)

Bug Fixes
  • Use alsoProvides in the urldispatch module to attach an interface to the request rather than directlyProvides to avoid disturbing interfaces set in a NewRequest event handler.
Documentation
  • Move 1.0.1 and previous changelog to HISTORY.txt.
  • Add examples to repoze.bfg.url.model_url docstring.
  • Add “Defending BFG Design” chapter to frontpage docs.
Templates
  • Remove ez_setup.py and its import from all paster templates, samples, and tutorials for distribute compatibility. The documentation already explains how to install virtualenv (which will include some setuptools package), so these files, imports and usages were superfluous.
Deprecations
  • The options kw arg to the repoze.bfg.router.make_app function is deprecated. In its place is the keyword argument settings. The options keyword continues to work, and a deprecation warning is not emitted when it is detected. However, the paster templates, code samples, and documentation now make reference to settings rather than options. This change/deprecation was mainly made for purposes of clarity and symmetry with the get_settings() API and dicussions of “settings” in various places in the docs: we want to use the same name to refer to the same thing everywhere.

1.1b3 (2009-11-06)

Features
  • repoze.bfg.testing.registerRoutesMapper testing facility added. This testing function registers a routes “mapper” object in the registry, for tests which require its presence. This function is documented in the repoze.bfg.testing API documentation.
Bug Fixes

  • Compound statements that used an assignment entered into in an interactive IPython session invoked via paster bfgshell no longer fail to mutate the shell namespace correctly. For example, this set of statements used to fail:

    In [2]: def bar(x): return x
  ...:
In [3]: list(bar(x) for x in 'abc')
Out[3]: NameError: 'bar'

    In this release, the bar function is found and the correct output is now sent to the console. Thanks to Daniel Holth for the patch.

  • The bfgshell command did not function properly; it was still expecting to be able to call the root factory with a bare environ rather than a request object.

Backwards Incompatibilities
  • The repoze.bfg.scripting.get_root function now expects a request object as its second argument rather than an environ.

1.1b2 (2009-11-02)

Bug Fixes
  • Prevent PyPI installation failure due to easy_install trying way too hard to guess the best version of Paste. When easy_install pulls from PyPI it reads links off various pages to determine “more up to date” versions. It incorrectly picks up a link for an ancient version of a package named “Paste-Deploy-0.1” (note the dash) when trying to find the “Paste” distribution and somehow believes it’s the latest version of “Paste”. It also somehow “helpfully” decides to check out a version of this package from SVN. We pin the Paste dependency version to a version greater than 1.7 to work around this easy_install bug.
Documentation
  • Fix “Hybrid” narrative chapter: stop claiming that <view> statements that mention a route_name need to come afer (in XML order) the <route> statement which creates the route. This hasn’t been true since 1.1a1.
  • “What’s New in repoze.bfg 1.1” document added to narrative documentation.
Features

  • Add a new event type: repoze.bfg.events.AfterTraversal. Events of this type will be sent after traversal is completed, but before any view code is invoked. Like repoze.bfg.events.NewRequest, This event will have a single attribute: request representing the current request. Unlike the request attribute of repoze.bfg.events.NewRequest however, during an AfterTraversal event, the request object will possess attributes set by the traverser, most notably context, which will be the context used when a view is found and invoked. The interface repoze.bfg.events.IAfterTraversal can be used to subscribe to the event. For example:

    <subscriber for="repoze.bfg.interfaces.IAfterTraversal"
            handler="my.app.handle_after_traverse"/>

    Like any framework event, a subscriber function should expect one parameter: event.

Dependencies
  • Rather than depending on chameleon.core and chameleon.zpt distributions individually, depend on Malthe’s repackaged Chameleon distribution (which includes both chameleon.core and chameleon.zpt).

1.1b1 (2009-11-01)

Bug Fixes
  • The routes root factory called route factories and the default route factory with an environ rather than a request. One of the symptoms of this bug: applications generated using the bfg_zodb paster template in 1.1a9 did not work properly.
  • Reinstate renderer alias for view_renderer in the <route> ZCML directive (in-the-wild 1.1a bw compat).
  • bfg_routesalchemy paster template: change <route> declarations: rename renderer attribute to view_renderer.
  • Header values returned by the authtktauthenticationpolicy remember and forget methods would be of type unicode. This violated the WSGI spec, causing a TypeError to be raised when these headers were used under mod_wsgi.
  • If a BFG app that had a route matching the root URL was mounted under a path in modwsgi, ala WSGIScriptAlias /myapp /Users/chrism/projects/modwsgi/env/bfg.wsgi, the home route (a route with the path of '/' or '') would not match when the path /myapp was visited (only when the path /myapp/ was visited). This is now fixed: if the urldispatch root factory notes that the PATH_INFO is empty, it converts it to a single slash before trying to do matching.
Documentation
  • In <route> declarations in tutorial ZCML, rename renderer attribute to view_renderer (fwd compat).
  • Fix various tutorials broken by 1.1a9 <route> directive changes.
Internal
  • Deal with a potential circref in the traversal module.

1.1a9 (2009-10-31)

Bug Fixes
  • An incorrect ZCML conflict would be encountered when the request_param predicate attribute was used on the ZCML view directive if any two otherwise same-predicated views had the combination of a predicate value with an = sign and one without (e.g. a vs. a=123).
Features
  • In previous versions of BFG, the “root factory” (the get_root callable passed to make_app or a function pointed to by the factory attribute of a route) was called with a “bare” WSGI environment. In this version, and going forward, it will be called with a request object. The request object passed to the factory implements dictionary-like methods in such a way that existing root factory code which expects to be passed an environ will continue to work.
  • The __call__ of a plugin “traverser” implementation (registered as an adapter for ITraverser or ITraverserFactory) will now receive a request as the single argument to its __call__ method. In previous versions it was passed a WSGI environ object. The request object passed to the factory implements dictionary-like methods in such a way that existing traverser code which expects to be passed an environ will continue to work.
  • The ZCML route directive’s attributes xhr, request_method, path_info, request_param, header and accept are now route predicates rather than view predicates. If one or more of these predicates is specified in the route configuration, all of the predicates must return true for the route to match a request. If one or more of the route predicates associated with a route returns False when checked during a request, the route match fails, and the next match in the routelist is tried. This differs from the previous behavior, where no route predicates existed and all predicates were considered view predicates, because in that scenario, the next route was not tried.
Documentation
  • Various changes were made to narrative and API documentation supporting the change from passing a request rather than an environ to root factories and traversers.
Internal
  • The request implements dictionary-like methods that mutate and query the WSGI environ. This is only for the purpose of backwards compatibility with root factories which expect an environ rather than a request.
  • The repoze.bfg.request.create_route_request_factory function, which returned a request factory was removed in favor of a repoze.bfg.request.route_request_interface function, which returns an interface.
  • The repoze.bfg.request.Request class, which is a subclass of webob.Request now defines its own __setattr__, __getattr__ and __delattr__ methods, which override the default WebOb behavior. The default WebOb behavior stores attributes of the request in self.environ['webob.adhoc_attrs'], and retrieves them from that dictionary during a __getattr__. This behavior was undesirable for speed and “expectation” reasons. Now attributes of the request are stored in request.__dict__ (as you otherwise might expect from an object that did not override these methods).
  • The router no longer calls repoze.bfg.traversal._traverse and does its work “inline” (speed).
  • Reverse the order in which the router calls the request factory and the root factory. The request factory is now called first; the resulting request is passed to the root factory.
  • The repoze.bfg.request.request_factory function has been removed. Its functionality is no longer required.
  • The “routes root factory” that wraps the default root factory when there are routes mentioned in the configuration now attaches an interface to the request via zope.interface.directlyProvides. This replaces logic in the (now-gone) repoze.bfg.request.request_factory function.
  • The route and view ZCML directives now register an interface as a named utility (retrieved from repoze.bfg.request.route_request_interface) rather than a request factory (the previous return value of the now-missing repoze.bfg.request.create_route_request_factory.
  • The repoze.bfg.functional module was renamed to repoze.bfg.compat.
Backwards Incompatibilities
  • Explicitly revert the feature introduced in 1.1a8: where the name root is available as an attribute of the request before a NewRequest event is emitted. This makes some potential future features impossible, or at least awkward (such as grouping traversal and view lookup into a single adapter lookup).
  • The containment, attr and renderer attributes of the route ZCML directive were removed.

1.1a8 (2009-10-27)

Features
  • Add path_info view configuration predicate.
  • paster bfgshell now supports IPython if it’s available for import. Thanks to Daniel Holth for the initial patch.
  • Add repoze.bfg.testing.registerSettings API, which is documented in the “repoze.bfg.testing” API chapter. This allows for registration of “settings” values obtained via repoze.bfg.settings.get_settings() for use in unit tests.
  • The name root is available as an attribute of the request slightly earlier now (before a NewRequest event is emitted). root is the result of the application “root factory”.
  • Added max_age parameter to authtktauthenticationpolicy ZCML directive. If this value is set, it must be an integer representing the number of seconds which the auth tkt cookie will survive. Mainly, its existence allows the auth_tkt cookie to survive across browser sessions.
Bug Fixes
  • Fix bug encountered during “scan” (when <scan ..> directive is used in ZCML) introduced in 1.1a7. Symptom: AttributeError: object has no attribute __provides__ raised at startup time.
  • The reissue_time argument to the authtktauthenticationpolicy ZCML directive now actually works. When it is set to an integer value, an authticket set-cookie header is appended to the response whenever a request requires authentication and ‘now’ minus the authticket’s timestamp is greater than reissue_time seconds.
Documentation
  • Add a chapter titled “Request and Response” to the narrative documentation, content cribbed from the WebOb documentation.
  • Call out predicate attributes of ZCML directive within “Views” chapter.
  • Fix route_url documentation (_query argument documented as query and _anchor argument documented as anchor).
Backwards Incompatibilities
  • The authtkt authentication policy remember method now no longer honors token or userdata keyword arguments.
Internal
  • Change how bfg_view decorator works when used as a class method decorator. In 1.1a7, the``scan``directive actually tried to grope every class in scanned package at startup time, calling dir against each found class, and subsequently invoking getattr against each thing found by dir to see if it was a method. This led to some strange symptoms (e.g. AttributeError: object has no attribute __provides__), and was generally just a bad idea. Now, instead of groping classes for methods at startup time, we just cause the bfg_view decorator itself to populate the method’s class’ __dict__ when it is used as a method decorator. This also requires a nasty _getframe thing but it’s slightly less nasty than the startup time groping behavior. This is essentially a reversion back to 1.1a6 “grokking” behavior plus some special magic for using the bfg_view decorator as method decorator inside the bfg_view class itself.
  • The router now checks for a global_response_headers attribute of the request object before returning a response. If this value exists, it is presumed to be a sequence of two-tuples, representing a set of headers to append to the ‘normal’ response headers. This feature is internal, rather than exposed externally, because it’s unclear whether it will stay around in the long term. It was added to support the reissue_time feature of the authtkt authentication policy.
  • The interface ITraverserFactory is now just an alias for ITraverser.

1.1a7 (2009-10-18)

Features

  • More than one @bfg_view decorator may now be stacked on top of any number of others. Each invocation of the decorator registers a single view configuration. For instance, the following combination of decorators and a function will register two view configurations for the same view callable:

    from repoze.bfg.view import bfg_view

@bfg_view(name='edit')
@bfg_view(name='change')
def edit(context, request):
    pass

    This makes it possible to associate more than one view configuration with a single callable without requiring any ZCML.

  • The @bfg_view decorator can now be used against a class method:

    from webob import Response
from repoze.bfg.view import bfg_view

class MyView(object):
    def __init__(self, context, request):
        self.context = context
        self.request = request

    @bfg_view(name='hello')
    def amethod(self):
        return Response('hello from %s!' % self.context)

    When the bfg_view decorator is used against a class method, a view is registered for the class (it’s a “class view” where the “attr” happens to be the name of the method it is attached to), so the class it’s defined within must have a suitable constructor: one that accepts context, request or just request.

Documentation
  • Added Changing the Traverser and Changing How :mod:`repoze.bfg.url.model_url` Generates a URL to the “Hooks” narrative chapter of the docs.
Internal
  • Remove ez_setup.py and imports of it within setup.py. In the new world, and as per virtualenv setup instructions, people will already have either setuptools or distribute.

1.1a6 (2009-10-15)

Features
  • Add xhr, accept, and header view configuration predicates to ZCML view declaration, ZCML route declaration, and bfg_view decorator. See the Views narrative documentation chapter for more information about these predicates.
  • Add setUp and tearDown functions to the repoze.bfg.testing module. Using setUp in a test setup and tearDown in a test teardown is now the recommended way to do component registry setup and teardown. Previously, it was recommended that a single function named repoze.bfg.testing.cleanUp be called in both the test setup and tear down. repoze.bfg.testing.cleanUp still exists (and will exist “forever” due to its widespread use); it is now just an alias for repoze.bfg.testing.setUp and is nominally deprecated.
  • The BFG component registry is now available in view and event subscriber code as an attribute of the request ie. request.registry. This fact is currently undocumented except for this note, because BFG developers never need to interact with the registry directly anywhere else.
  • The BFG component registry now inherits from dict, meaning that it can optionally be used as a simple dictionary. Component registrations performed against it via e.g. registerUtility, registerAdapter, and similar API methods are kept in a completely separate namespace than its dict members, so using the its component API methods won’t effect the keys and values in the dictionary namespace. Likewise, though the component registry “happens to be” a dictionary, use of mutating dictionary methods such as __setitem__ will have no influence on any component registrations made against it. In other words, the registry object you obtain via e.g. repoze.bfg.threadlocal.get_current_registry or request.registry happens to be both a component registry and a dictionary, but using its component-registry API won’t impact data added to it via its dictionary API and vice versa. This is a forward compatibility move based on the goals of “marco”.
  • Expose and document repoze.bfg.testing.zcml_configure API. This function populates a component registry from a ZCML file for testing purposes. It is documented in the “Unit and Integration Testing” chapter.
Documentation
  • Virtual hosting narrative docs chapter updated with info about mod_wsgi.
  • Point all index URLs at the literal 1.1 index (this alpha cycle may go on a while).
  • Various tutorial test modules updated to use repoze.bfg.testing.setUp and repoze.bfg.testing.tearDown methods in order to encourage this as best practice going forward.
  • Added “Creating Integration Tests” section to unit testing narrative documentation chapter. As a result, the name of the unittesting chapter is now “Unit and Integration Testing”.
Backwards Incompatibilities
  • Importing getSiteManager and get_registry from repoze.bfg.registry is no longer supported. These imports were deprecated in repoze.bfg 1.0. Import of getSiteManager should be done as from zope.component import getSiteManager. Import of get_registry should be done as from repoze.bfg.threadlocal import get_current_registry. This was done to prevent a circular import dependency.
  • Code bases which alternately invoke both zope.testing.cleanup.cleanUp and repoze.bfg.testing.cleanUp (treating them equivalently, using them interchangeably) in the setUp/tearDown of unit tests will begin to experience test failures due to lack of test isolation. The “right” mechanism is repoze.bfg.testing.cleanUp (or the combination of repoze.bfg.testing.setUp and repoze.bfg.testing.tearDown). but a good number of legacy codebases will use zope.testing.cleanup.cleanUp instead. We support zope.testing.cleanup.cleanUp but not in combination with repoze.bfg.testing.cleanUp in the same codebase. You should use one or the other test cleanup function in a single codebase, but not both.
Internal
  • Created new repoze.bfg.configuration module which assumes responsibilities previously held by the repoze.bfg.registry and repoze.bfg.router modules (avoid a circular import dependency).
  • The result of the zope.component.getSiteManager function in unit tests set up with repoze.bfg.testing.cleanUp or repoze.bfg.testing.setUp will be an instance of repoze.bfg.registry.Registry instead of the global zope.component.globalregistry.base registry. This also means that the threadlocal ZCA API functions such as getAdapter and getUtility as well as internal BFG machinery (such as model_url and route_url) will consult this registry within unit tests. This is a forward compatibility move based on the goals of “marco”.
  • Removed repoze.bfg.testing.addCleanUp function and associated module-scope globals. This was never an API.

1.1a5 (2009-10-10)

Documentation
  • Change “Traversal + ZODB” and “URL Dispatch + SQLAlchemy” Wiki tutorials to make use of the new-to-1.1 “renderer” feature (return dictionaries from all views).
  • Add tests to the “URL Dispatch + SQLAlchemy” tutorial after the “view” step.
  • Added a diagram of model graph traversal to the “Traversal” narrative chapter of the documentation.
  • An exceptions API chapter was added, documenting the new repoze.bfg.exceptions module.
  • Describe “request-only” view calling conventions inside the urldispatch narrative chapter, where it’s most helpful.
  • Add a diagram which explains the operation of the BFG router to the “Router” narrative chapter.
Features
  • Add a new repoze.bfg.testing API: registerRoute, for registering routes to satisfy calls to e.g. repoze.bfg.url.route_url in unit tests.
  • The notfound and forbidden ZCML directives now accept the following addtional attributes: attr, renderer, and wrapper. These have the same meaning as they do in the context of a ZCML view directive.
  • For behavior like Django’s APPEND_SLASH=True, use the repoze.bfg.view.append_slash_notfound_view view as the Not Found view in your application. When this view is the Not Found view (indicating that no view was found), and any routes have been defined in the configuration of your application, if the value of PATH_INFO does not already end in a slash, and if the value of PATH_INFO plus a slash matches any route’s path, do an HTTP redirect to the slash-appended PATH_INFO. Note that this will lose POST data information (turning it into a GET), so you shouldn’t rely on this to redirect POST requests.
  • Speed up repoze.bfg.location.lineage slightly.
  • Speed up repoze.bfg.encode.urlencode (nee’ repoze.bfg.url.urlencode) slightly.
  • Speed up repoze.bfg.traversal.model_path.
  • Speed up repoze.bfg.traversal.model_path_tuple slightly.
  • Speed up repoze.bfg.traversal.traverse slightly.
  • Speed up repoze.bfg.url.model_url slightly.
  • Speed up repoze.bfg.url.route_url slightly.
  • Sped up repoze.bfg.traversal.ModelGraphTraverser:__call__ slightly.
  • Minor speedup of repoze.bfg.router.Router.__call__.
  • New repoze.bfg.exceptions module was created to house exceptions that were previously sprinkled through various modules.
Internal
  • Move repoze.bfg.traversal._url_quote into repoze.bfg.encode as url_quote.
Deprecations
  • The import of repoze.bfg.view.NotFound is deprecated in favor of repoze.bfg.exceptions.NotFound. The old location still functions, but emits a deprecation warning.
  • The import of repoze.bfg.security.Unauthorized is deprecated in favor of repoze.bfg.exceptions.Forbidden. The old location still functions but emits a deprecation warning. The rename from Unauthorized to Forbidden brings parity to the the name of the exception and the system view it invokes when raised.
Backwards Incompatibilities

  • We previously had a Unicode-aware wrapper for the urllib.urlencode function named repoze.bfg.url.urlencode which delegated to the stdlib function, but which marshalled all unicode values to utf-8 strings before calling the stdlib version. A newer replacement now lives in repoze.bfg.encode The replacement does not delegate to the stdlib.

    The replacement diverges from the stdlib implementation and the previous repoze.bfg.url url implementation inasmuch as its doseq argument is now a decoy: it always behaves in the doseq=True way (which is the only sane behavior) for speed purposes.

    The old import location (repoze.bfg.url.urlencode) still functions and has not been deprecated.

  • In 0.8a7, the return value expected from an object implementing ITraverserFactory was changed from a sequence of values to a dictionary containing the keys context, view_name, subpath, traversed, virtual_root, virtual_root_path, and root. Until now, old-style traversers which returned a sequence have continued to work but have generated a deprecation warning. In this release, traversers which return a sequence instead of a dictionary will no longer work.

1.1a4 (2009-09-23)

Bug Fixes
  • On 64-bit Linux systems, views that were members of a multiview (orderings of views with predicates) were not evaluated in the proper order. Symptom: in a configuration that had two views with the same name but one with a request_method=POST predicate and one without, the one without the predicate would be called unconditionally (even if the request was a POST request). Thanks much to Sebastien Douche for providing the buildbots that pointed this out.
Documentation
  • Added a tutorial which explains how to use repoze.session (ZODB-based sessions) in a ZODB-based repoze.bfg app.
  • Added a tutorial which explains how to add ZEO to a ZODB-based repoze.bfg application.
  • Added a tutorial which explains how to run a repoze.bfg application under mod_wsgi. See “Running a repoze.bfg Application under mod_wsgi” in the tutorials section of the documentation.
Features
  • Add a repoze.bfg.url.static_url API which is capable of generating URLs to static resources defined by the <static> ZCML directive. See the “Views” narrative chapter’s section titled “Generating Static Resource URLs” for more information.
  • Add a string renderer. This renderer converts a non-Response return value of any view callble into a string. It is documented in the “Views” narrative chapter.
  • Give the route ZCML directive the view_attr and view_renderer parameters (bring up to speed with 1.1a3 features). These can also be spelled as attr and renderer.
Backwards Incompatibilities
  • An object implementing the IRenderer interface (and ITemplateRenderer`, which is a subclass of ``IRenderer) must now accept an extra system argument in its __call__ method implementation. Values computed by the system (as opposed to by the view) are passed by the system in the system parameter, which will always be a dictionary. Keys in the dictionary include: view (the view object that returned the value), renderer_name (the template name or simple name of the renderer), context (the context object passed to the view), and request (the request object passed to the view). Previously only ITemplateRenderers received system arguments as elements inside the main value dictionary.
Internal
  • The way bfg_view declarations are scanned for has been modified. This should have no external effects.
  • Speed: do not register an ITraverserFactory in configure.zcml; instead rely on queryAdapter and a manual default to ModelGraphTraverser.
  • Speed: do not register an IContextURL in configure.zcml; instead rely on queryAdapter and a manual default to TraversalContextURL.
  • General speed microimprovements for helloworld benchmark: replace try/excepts with statements which use ‘in’ keyword.

1.1a3 (2009-09-16)

Documentation
  • The “Views” narrative chapter in the documentation has been updated extensively to discuss “renderers”.
Features
  • A renderer attribute has been added to view configurations, replacing the previous (1.1a2) version’s template attribute. A “renderer” is an object which accepts the return value of a view and converts it to a string. This includes, but is not limited to, templating systems.
  • A new interface named IRenderer was added. The existing interface, ITemplateRenderer now derives from this new interface. This interface is internal.
  • A new interface named IRendererFactory was added. An existing interface named ITemplateRendererFactory now derives from this interface. This interface is internal.
  • The view attribute of the view ZCML directive is no longer required if the ZCML directive also has a renderer attribute. This is useful when the renderer is a template renderer and no names need be passed to the template at render time.
  • A new zcml directive renderer has been added. It is documented in the “Views” narrative chapter of the documentation.
  • A ZCML view directive (and the associated bfg_view decorator) can now accept a “wrapper” value. If a “wrapper” value is supplied, it is the value of a separate view’s name attribute. When a view with a wrapper attribute is rendered, the “inner” view is first rendered normally. Its body is then attached to the request as “wrapped_body”, and then a wrapper view name is looked up and rendered (using repoze.bfg.render_view_to_response), passed the request and the context. The wrapper view is assumed to do something sensible with request.wrapped_body, usually inserting its structure into some other rendered template. This feature makes it possible to specify (potentially nested) “owrap” relationships between views using only ZCML or decorators (as opposed always using ZPT METAL and analogues to wrap view renderings in outer wrappers).
Dependencies
  • When used under Python < 2.6, BFG now has an installation time dependency on the simplejson package.
Deprecations
  • The repoze.bfg.testing.registerDummyRenderer API has been deprecated in favor of repoze.bfg.testing.registerTemplateRenderer. A deprecation warning is not issued at import time for the former name; it will exist “forever”; its existence has been removed from the documentation, however.
  • The repoze.bfg.templating.renderer_from_cache function has been moved to repoze.bfg.renderer.template_renderer_factory. This was never an API, but code in the wild was spotted that used it. A deprecation warning is issued at import time for the former.
Backwards Incompatibilities
  • The ITemplateRenderer interface has been changed. Previously its __call__ method accepted **kw. It now accepts a single positional parameter named kw (REVISED: it accepts two positional parameters as of 1.1a4: value and system). This is mostly an internal change, but it was exposed in APIs in one place: if you’ve used the repoze.bfg.testing.registerDummyRenderer API in your tests with a custom “renderer” argument with your own renderer implementation, you will need to change that renderer implementation to accept kw instead of **kw in its __call__ method (REVISED: make it accept value and system positional arguments as of 1.1a4).
  • The ITemplateRendererFactory interface has been changed. Previously its __call__ method accepted an auto_reload keyword parameter. Now its __call__ method accepts no keyword parameters. Renderers are now themselves responsible for determining details of auto-reload. This is purely an internal change. This interface was never external.
  • The template_renderer ZCML directive introduced in 1.1a2 has been removed. It has been replaced by the renderer directive.
  • The previous release (1.1a2) added a view configuration attribute named template. In this release, the attribute has been renamed to renderer. This signifies that the attribute is more generic: it can now be not just a template name but any renderer name (ala json).
  • In the previous release (1.1a2), the Chameleon text template renderer was used if the system didn’t associate the template view configuration value with a filename with a “known” extension. In this release, you must use a renderer attribute which is a path that ends with a .txt extension (e.g. templates/foo.txt) to use the Chameleon text renderer.

1.1a2 (2009-09-14)

Features
  • A ZCML view directive (and the associated bfg_view decorator) can now accept an “attr” value. If an “attr” value is supplied, it is considered a method named of the view object to be called when the response is required. This is typically only good for views that are classes or instances (not so useful for functions, as functions typically have no methods other than __call__).
  • A ZCML view directive (and the associated bfg_view decorator) can now accept a “template” value. If a “template” value is supplied, and the view callable returns a dictionary, the associated template is rendered with the dictionary as keyword arguments. See the section named “Views That Have a template” in the “Views” narrative documentation chapter for more information.

1.1a1 (2009-09-06)

Bug Fixes
  • “tests” module removed from the bfg_alchemy paster template; these tests didn’t work.
  • Bugfix: the discriminator for the ZCML “route” directive was incorrect. It was possible to register two routes that collided without the system spitting out a ConfigurationConflictError at startup time.
Features
  • Feature addition: view predicates. These are exposed as the request_method, request_param, and containment attributes of a ZCML view declaration, or the respective arguments to a @bfg_view decorator. View predicates can be used to register a view for a more precise set of environment parameters than was previously possible. For example, you can register two views with the same name with different request_param attributes. If the request.params dict contains ‘foo’ (request_param=”foo”), one view might be called; if it contains ‘bar’ (request_param=”bar”), another view might be called. request_param can also name a key/value pair ala foo=123. This will match only when the foo key is in the request.params dict and it has the value ‘123’. This particular example makes it possible to write separate view functions for different form submissions. The other predicates, containment and request_method work similarly. containment is a view predicate that will match only when the context’s graph lineage has an object possessing a particular class or interface, for example. request_method is a view predicate that will match when the HTTP REQUEST_METHOD equals some string (eg. ‘POST’).
  • The @bfg_view decorator now accepts three additional arguments: request_method, request_param, and containment. request_method is used when you’d like the view to match only a request with a particular HTTP REQUEST_METHOD; a string naming the REQUEST_METHOD can also be supplied as request_type for backwards compatibility. request_param is used when you’d like a view to match only a request that contains a particular request.params key (with or without a value). containment is used when you’d like to match a request that has a context that has some class or interface in its graph lineage. These are collectively known as “view predicates”.
  • The route ZCML directive now honors view_request_method, view_request_param and view_containment attributes, which pass along these values to the associated view if any is provided. Additionally, the request_type attribute can now be spelled as view_request_type, and permission can be spelled as view_permission. Any attribute which starts with view_ can now be spelled without the view_ prefix, so view_for can be spelled as for now, etc. Both forms are documented in the urldispatch narraitve documentation chapter.
  • The request_param ZCML view directive attribute (and its bfg_view decorator cousin) can now specify both a key and a value. For example, request_param="foo=123" means that the foo key must have a value of 123 for the view to “match”.
  • Allow repoze.bfg.traversal.find_interface API to use a class object as the argument to compare against the model passed in. This means you can now do find_interface(model, SomeClass) and the first object which is found in the lineage which has SomeClass as its class (or the first object found which has SomeClass as any of its superclasses) will be returned.
  • Added static ZCML directive which registers a route for a view that serves up files in a directory. See the “Views” narrative documentation chapter’s “Serving Static Resources Using a ZCML Directive” section for more information.
  • The repoze.bfg.view.static class now accepts a string as its first argument (“root_dir”) that represents a package-relative name e.g. somepackage:foo/bar/static. This is now the preferred mechanism for spelling package-relative static paths using this class. A package_name keyword argument has been left around for backwards compatibility. If it is supplied, it will be honored.
  • The API repoze.bfg.testing.registerView now takes a permission argument. Use this instead of using repoze.bfg.testing.registerViewPermission.
  • The ordering of route declarations vs. the ordering of view declarations that use a “route_name” in ZCML no longer matters. Previously it had been impossible to use a route_name from a route that had not yet been defined in ZCML (order-wise) within a “view” declaration.
  • The repoze.bfg router now catches both repoze.bfg.security.Unauthorized and repoze.bfg.view.NotFound exceptions while rendering a view. When the router catches an Unauthorized, it returns the registered forbidden view. When the router catches a NotFound, it returns the registered notfound view.
Internal
  • Change urldispatch internals: Route object is now constructed using a path, a name, and a factory instead of a name, a matcher, a generator, and a factory.
  • Move (non-API) default_view, default_forbidden_view, and default_notfound_view functions into the repoze.bfg.view module (moved from repoze.bfg.router).
  • Removed ViewPermissionFactory from repoze.bfg.security. View permission checking is now done by registering and looking up an ISecuredView.
  • The static ZCML directive now uses a custom root factory when constructing a route.
  • The interface IRequestFactories was removed from the repoze.bfg.interfaces module. This interface was never an API.
  • The function named named_request_factories and the data structure named DEFAULT_REQUEST_FACTORIES have been removed from the repoze.bfg.request module. These were never APIs.
  • The IViewPermissionFactory interface has been removed. This was never an API.
Documentation
  • Request-only-convention examples in the “Views” narrative documentation were broken.
  • Fixed documentation bugs related to forget and remember in security API docs.
  • Fixed documentation for repoze.bfg.view.static (in narrative Views chapter).
Deprecations
  • The API repoze.bfg.testing.registerViewPermission has been deprecated.
Backwards Incompatibilities

  • The interfaces IPOSTRequest, IGETRequest, IPUTRequest, IDELETERequest, and IHEADRequest have been removed from the repoze.bfg.interfaces module. These were not documented as APIs post-1.0. Instead of using one of these, use a request_method ZCML attribute or request_method bfg_view decorator parameter containing an HTTP method name (one of GET, POST, HEAD, PUT, DELETE) instead of one of these interfaces if you were using one explicitly. Passing a string in the set (GET, HEAD, PUT, POST, DELETE) as a request_type argument will work too. Rationale: instead of relying on interfaces attached to the request object, BFG now uses a “view predicate” to determine the request type.

  • Views registered without the help of the ZCML view directive are now responsible for performing their own authorization checking.

  • The registry_manager backwards compatibility alias importable from “repoze.bfg.registry”, deprecated since repoze.bfg 0.9 has been removed. If you are tring to use the registry manager within a debug script of your own, use a combination of the “repoze.bfg.paster.get_app” and “repoze.bfg.scripting.get_root” APIs instead.

  • The INotFoundAppFactory interface has been removed; it has been deprecated since repoze.bfg 0.9. If you have something like the following in your configure.zcml:

    <utility provides="repoze.bfg.interfaces.INotFoundAppFactory"
         component="helloworld.factories.notfound_app_factory"/>

    Replace it with something like:

    <notfound
    view="helloworld.views.notfound_view"/>

    See “Changing the Not Found View” in the “Hooks” chapter of the documentation for more information.

  • The IUnauthorizedAppFactory interface has been removed; it has been deprecated since repoze.bfg 0.9. If you have something like the following in your configure.zcml:

    <utility provides="repoze.bfg.interfaces.IUnauthorizedAppFactory"
         component="helloworld.factories.unauthorized_app_factory"/>

    Replace it with something like:

    <forbidden
    view="helloworld.views.forbidden_view"/>

    See “Changing the Forbidden View” in the “Hooks” chapter of the documentation for more information.

  • ISecurityPolicy-based security policies, deprecated since repoze.bfg 0.9, have been removed. If you have something like this in your configure.zcml, it will no longer work:

    <utility
  provides="repoze.bfg.interfaces.ISecurityPolicy"
  factory="repoze.bfg.security.RemoteUserInheritingACLSecurityPolicy"
 />

    If ZCML like the above exists in your application, you will receive an error at startup time. Instead of the above, you’ll need something like:

    <remoteuserauthenticationpolicy/>
<aclauthorizationpolicy/>

    This is just an example. See the “Security” chapter of the repoze.bfg documentation for more information about configuring security policies.

  • Custom ZCML directives which register an authentication or authorization policy (ala “authtktauthenticationpolicy” or “aclauthorizationpolicy”) should register the policy “eagerly” in the ZCML directive instead of from within a ZCML action. If an authentication or authorization policy is not found in the component registry by the view machinery during deferred ZCML processing, view security will not work as expected.

1.0.1 (2009-07-22)

  • Added support for has_resource, resource_isdir, and resource_listdir to the resource “OverrideProvider”; this fixes a bug with a symptom that a file could not be overridden in a resource directory unless a file with the same name existed in the original directory being overridden.
  • Fixed documentation bug showing invalid test for values from the matchdict: they are stored as attributes of the Article, rather than subitems.
  • Fixed documentation bug showing wrong environment key for the matchdict produced by the matching route.
  • Added a workaround for a bug in Python 2.6, 2.6.1, and 2.6.2 having to do with a recursion error in the mimetypes module when trying to serve static files from Paste’s FileApp: http://bugs.python.org/issue5853. Symptom: File “/usr/lib/python2.6/mimetypes.py”, line 244, in guess_type return guess_type(url, strict) RuntimeError: maximum recursion depth exceeded. Thanks to Armin Ronacher for identifying the symptom and pointing out a fix.
  • Minor edits to tutorials for accuracy based on feedback.
  • Declared Paste and PasteDeploy dependencies.

1.0 (2009-07-05)

  • Retested and added some content to GAE tutorial.
  • Edited “Extending” narrative docs chapter.
  • Added “Deleting the Database” section to the “Defining Models” chapter of the traversal wiki tutorial.
  • Spell checking of narratives and tutorials.

1.0b2 (2009-07-03)

  • remoteuserauthenticationpolicy ZCML directive didn’t work without an environ_key directive (didn’t match docs).
  • Fix configure_zcml filespec check on Windows. Previously if an absolute filesystem path including a drive letter was passed as filename (or as configure_zcml in the options dict) to repoze.bfg.router.make_app, it would be treated as a package:resource_name specification.
  • Fix inaccuracies and import errors in bfgwiki (traversal+ZODB) and bfgwiki2 (urldispatch+SA) tutorials.
  • Use bfgsite index for all tutorial setup.cfg files.
  • Full documentation grammar/style/spelling audit.

1.0b1 (2009-07-02)

Features
  • Allow a Paste config file (configure_zcml) value or an environment variable (BFG_CONFIGURE_ZCML) to name a ZCML file (optionally package-relative) that will be used to bootstrap the application. Previously, the integrator could not influence which ZCML file was used to do the boostrapping (only the original application developer could do so).
Documentation
  • Added a “Resources” chapter to the narrative documentation which explains how to override resources within one package from another package.
  • Added an “Extending” chapter to the narrative documentation which explains how to extend or modify an existing BFG application using another Python package and ZCML.

1.0a9 (2009-07-01)

Features
  • Make it possible to pass strings in the form “package_name:relative/path” to APIs like render_template, render_template_to_response, and get_template. Sometimes the package in which a caller lives is a direct namespace package, so the module which is returned is semi-useless for navigating from. In this way, the caller can control the horizontal and vertical of where things get looked up from.

1.0a8 (2009-07-01)

Deprecations
  • Deprecate the authentication_policy and authorization_policy arguments to repoze.bfg.router.make_app. Instead, developers should use the various authentication policy ZCML directives (repozewho1authenticationpolicy, remoteuserauthenticationpolicy and authtktauthenticationpolicy) and the aclauthorizationpolicy` authorization policy directive as described in the changes to the “Security” narrative documenation chapter and the wiki tutorials.
Features
  • Add three new ZCML directives which configure authentication policies:
    • repozewho1authenticationpolicy
    • remoteuserauthenticationpolicy
    • authtktauthenticationpolicy
  • Add a new ZCML directive which configures an ACL authorization policy named aclauthorizationpolicy.
Bug Fixes
  • Bug fix: when a repoze.bfg.resource.PackageOverrides class was instantiated, and the package it was overriding already had a __loader__ attribute, it would fail at startup time, even if the __loader__ attribute was another PackageOverrides instance. We now replace any __loader__ that is also a PackageOverrides instance. Symptom: ConfigurationExecutionError: <type 'exceptions.TypeError'>: Package <module 'karl.views' from '/Users/chrism/projects/osi/bfgenv/src/karl/karl/views/__init__.pyc'> already has a __loader__ (probably a module in a zipped egg).

1.0a7 (2009-06-30)

Features
  • Add a reload_resources configuration file setting (aka the BFG_RELOAD_RESOURCES environment variable). When this is set to true, the server never needs to be restarted when moving files between directory resource overrides (esp. for templates currently).
  • Add a reload_all configuration file setting (aka the BFG_RELOAD_ALL environment variable) that implies both reload_resources and reload_templates.
  • The static helper view class now uses a PackageURLParser in order to allow for the overriding of static resources (CSS / logo files, etc) using the resource ZCML directive. The PackageURLParser class was added to a (new) static module in BFG; it is a subclass of the StaticURLParser class in paste.urlparser.
  • The repoze.bfg.templating.renderer_from_cache function now checks for the reload_resources setting; if it’s true, it does not register a template renderer (it won’t use the registry as a template renderer cache).
Documentation
  • Add pkg_resources to the glossary.
  • Update the “Environment” docs to note the existence of reload_resources and reload_all.
  • Updated the bfg_alchemy paster template to include two views: the view on the root shows a list of links to records; the view on a record shows the details for that object.
Internal
  • Use a colon instead of a tab as the separator between package name and relpath to form the “spec” when register a ITemplateRenderer.
  • Register a repoze.bfg.resource.OverrideProvider as a pkg_resources provider only for modules which are known to have overrides, instead of globally, when a <resource> directive is used (performance).

1.0a6 (2009-06-29)

Bug Fixes
  • Use caller_package function instead of caller_module function within templating to avoid needing to name the caller module in resource overrides (actually match docs).
  • Make it possible to override templates stored directly in a module with templates in a subdirectory of the same module, stored directly within another module, or stored in a subdirectory of another module (actually match docs).

1.0a5 (2009-06-28)

Features

  • A new ZCML directive exists named “resource”. This ZCML directive allows you to override Chameleon templates within a package (both directories full of templates and individual template files) with other templates in the same package or within another package. This allows you to “fake out” a view’s use of a template, causing it to retrieve a different template than the one actually named by a relative path to a call like render_template_to_response('templates/mytemplate.pt'). For example, you can override a template file by doing:

    <resource
  to_override="some.package:templates/mytemplate.pt"
  override_with="another.package:othertemplates/anothertemplate.pt"
 />

    The string passed to “to_override” and “override_with” is named a “specification”. The colon separator in a specification separates the package name from a package-relative directory name. The colon and the following relative path are optional. If they are not specified, the override attempts to resolve every lookup into a package from the directory of another package. For example:

    <resource
  to_override="some.package"
  override_with="another.package"
 />

    Individual subdirectories within a package can also be overridden:

    <resource
  to_override="some.package:templates/"
  override_with="another.package:othertemplates/"
 />

    If you wish to override a directory with another directory, you must make sure to attach the slash to the end of both the to_override specification and the override_with specification. If you fail to attach a slash to the end of a specification that points a directory, you will get unexpected results. You cannot override a directory specification with a file specification, and vice versa (a startup error will occur if you try).

    You cannot override a resource with itself (a startup error will occur if you try).

    Only individual package resources may be overridden. Overrides will not traverse through subpackages within an overridden package. This means that if you want to override resources for both some.package:templates, and some.package.views:templates, you will need to register two overrides.

    The package name in a specification may start with a dot, meaning that the package is relative to the package in which the ZCML file resides. For example:

    <resource
  to_override=".subpackage:templates/"
  override_with="another.package:templates/"
 />

    Overrides for the same to_overrides specification can be named multiple times within ZCML. Each override_with path will be consulted in the order defined within ZCML, forming an override search path.

    Resource overrides can actually override resources other than templates. Any software which uses the pkg_resources get_resource_filename, get_resource_stream or get_resource_string APIs will obtain an overridden file when an override is used. However, the only built-in facility which uses the pkg_resources API within BFG is the templating stuff, so we only call out template overrides here.

  • Use the pkg_resources API to locate template filenames instead of dead-reckoning using the os.path module.

  • The repoze.bfg.templating module now uses pkg_resources to locate and register template files instead of using an absolute path name.

1.0a4 (2009-06-25)

Features
  • Cause :segment matches in route paths to put a Unicode-decoded and URL-dequoted value in the matchdict for the value matched. Previously a non-decoded non-URL-dequoted string was placed in the matchdict as the value.
  • Cause *remainder matches in route paths to put a tuple in the matchdict dictionary in order to be able to present Unicode-decoded and URL-dequoted values for the traversal path. Previously a non-decoded non-URL-dequoted string was placed in the matchdict as the value.
  • Add optional max_age keyword value to the remember method of repoze.bfg.authentication.AuthTktAuthenticationPolicy; if this value is passed to remember, the generated cookie will have a corresponding Max-Age value.
Documentation
  • Add information to the URL Dispatch narrative documentation about path pattern matching syntax.
Bug Fixes
  • Make route_url URL-quote segment replacements during generation. Remainder segments are not quoted.

1.0a3 (2009-06-24)

Implementation Changes
  • repoze.bfg no longer relies on the Routes package to interpret URL paths. All known existing path patterns will continue to work with the reimplemented logic, which lives in repoze.bfg.urldispatch. <route> ZCML directives which use certain attributes (uncommon ones) may not work (see “Backwards Incompatibilities” below).
Bug Fixes
  • model_url when passed a request that was generated as a result of a route match would fail in a call to route.generate.
  • BFG-on-GAE didn’t work due to a corner case bug in the fallback Python implementation of threading.local (symptom: “Initialization arguments are not supported”). Thanks to Michael Bernstein for the bug report.
Documentation
  • Added a “corner case” explanation to the “Hybrid Apps” chapter explaining what to do when “the wrong” view is matched.
  • Use repoze.bfg.url.route_url API in tutorials rather than Routes url_for API.
Features
  • Added the repoze.bfg.url.route_url API. This API allows you to generate URLs based on <route> declarations. See the URL Dispatch narrative chapter and the “repoze.bfg.url” module API documentation for more information.
Backwards Incompatibilities
  • As a result of disusing Routes, using the Routes url_for API inside a BFG application (as was suggested by previous iterations of tutorials) will no longer work. Use the repoze.bfg.url.route_url method instead.
  • The following attributes on the <route> ZCML directive no longer work: encoding, static, filter, condition_method, condition_subdomain, condition_function, explicit, or subdomains. These were all Routes features.
  • The <route> ZCML directive no longer supports the <requirement> subdirective. This was a Routes feature.

1.0a2 (2009-06-23)

Bug Fixes
  • The bfg_routesalchemy paster template app tests failed due to a mismatch between test and view signatures.
Features
  • Add a view_for attribute to the route ZCML directive. This attribute should refer to an interface or a class (ala the for attribute of the view ZCML directive).
Documentation
  • Conditional documentation in installation section (“how to install a Python interpreter”).
Backwards Incompatibilities
  • The callback argument of the repoze.bfg.authentication authentication policies named RepozeWho1AuthenticationPolicy, RemoteUserAuthenticationPolicy, and AuthTktAuthenticationPolicy now must accept two positional arguments: the orginal argument accepted by each (userid or identity) plus a second argument, which will be the current request. Apologies, this is required to service finding groups when there is no “global” database connection.

1.0a1 (2009-06-22)

Features

  • A new ZCML directive was added named notfound. This ZCML directive can be used to name a view that should be invoked when the request can’t otherwise be resolved to a view callable. For example:

    <notfound
    view="helloworld.views.notfound_view"/>

  • A new ZCML directive was added named forbidden. This ZCML directive can be used to name a view that should be invoked when a view callable for a request is found, but cannot be invoked due to an authorization failure. For example:

    <forbidden
    view="helloworld.views.forbidden_view"/>

  • Allow views to be optionally defined as callables that accept only a request object, instead of both a context and a request (which still works, and always will). The following types work as views in this style:

    • functions that accept a single argument request, e.g.:

      def aview(request):
    pass

    • new and old-style classes that have an __init__ method that accepts self, request, e.g.:

      def View(object):
    __init__(self, request):
       pass

    • Arbitrary callables that have a __call__ method that accepts self, request, e.g.:

      def AView(object):
    def __call__(self, request):
       pass
view = AView()

    This likely should have been the calling convention all along, as the request has context as an attribute already, and with views called as a result of URL dispatch, having the context in the arguments is not very useful. C’est la vie.

  • Cache the absolute path in the caller’s package globals within repoze.bfg.path to get rid of repeated (expensive) calls to os.path.abspath.

  • Add reissue_time and timeout parameters to repoze.bfg.authentication.AuthTktAuthenticationPolicy constructor. If these are passed, cookies will be reset every so often (cadged from the same change to repoze.who lately).

  • The matchdict related to the matching of a Routes route is available on the request as the matchdict attribute: request.matchdict. If no route matched, this attribute will be None.

  • Make 404 responses slightly cheaper by showing environ["PATH_INFO"] on the notfound result page rather than the fullly computed URL.

  • Move LRU cache implementation into a separate package (repoze.lru).

  • The concepts of traversal and URL dispatch have been unified. It is now possible to use the same sort of factory as both a traversal “root factory” and what used to be referred to as a urldispatch “context factory”.

  • When the root factory argument (as a first argument) passed to repoze.bfg.router.make_app is None, a default root factory is used. This is in support of using routes as “root finders”; it supplants the idea that there is a default IRoutesContextFactory.

  • The view` ZCML statement and the repoze.bfg.view.bfg_view decorator now accept an extra argument: route_name. If a route_name is specified, it must match the name of a previously defined route statement. When it is specified, the view will only be called when that route matches during a request.

  • It is now possible to perfom traversal after a route has matched. Use the pattern *traverse in a <route> path attribute within ZCML, and the path remainder which it matches will be used as a traversal path.

  • When any route defined matches, the WSGI environment will now contain a key bfg.routes.route (the Route object which matched), and a key bfg.routes.matchdict (the result of calling route.match).

Deprecations
  • Utility registrations against repoze.bfg.interfaces.INotFoundView and repoze.bfg.interfaces.IForbiddenView are now deprecated. Use the notfound and forbidden ZCML directives instead (see the “Hooks” chapter for more information). Such registrations will continue to work, but the notfound and forbidden directives do “extra work” to ensure that the callable named by the directive can be called by the router even if it’s a class or request-argument-only view.
Removals
  • The IRoutesContext, IRoutesContextFactory, and IContextNotFound interfaces were removed from repoze.bfg.interfaces. These were never APIs.
  • The repoze.bfg.urldispatch.RoutesContextNotFound, repoze.bfg.urldispatch.RoutesModelTraverser and repoze.bfg.urldispatch.RoutesContextURL classes were removed. These were also never APIs.
Backwards Incompatibilities
  • Moved the repoze.bfg.push module, which implemented the pushpage decorator, into a separate distribution, repoze.bfg.pushpage. Applications which used this decorator should continue to work after adding that distribution to their installation requirements.
  • Changing the default request factory via an IRequestFactory utility registration (as used to be documented in the “Hooks” chapter’s “Changing the request factory” section) is no longer supported. The dance to manufacture a request is complicated as a result of unifying traversal and url dispatch, making it highly unlikely for anyone to be able to override it properly. For those who just want to decorate or modify a request, use a NewRequestEvent subscriber (see the Events chapter in the documentation).
  • The repoze.bfg.IRequestFactory interface was removed. See the bullet above for why.
  • Routes “context factories” (spelled as the factory argument to a route statement in ZCML) must now expect the WSGI environ as a single argument rather than a set of keyword arguments. They can obtain the match dictionary by asking for environ[‘bfg.routes.matchdict’]. This is the same set of keywords that used to be passed to urldispatch “context factories” in BFG 0.9 and below.
  • Using the @zope.component.adapter decorator on a bfg view function no longer works. Use the @repoze.bfg.view.bfg_view decorator instead to mark a function (or a class) as a view.
  • The name under which the matching route object is found in the environ was changed from bfg.route to bfg.routes.route.
  • Finding the root is now done before manufacturing a request object (and sending a new request event) within the router (it used to be performed afterwards).
  • Adding *path_info to a route no longer changes the PATH_INFO for a request that matches using URL dispatch. This feature was only there to service the repoze.bfg.wsgi.wsgiapp2 decorator and it did it wrong; use *subpath instead now.
  • The values of subpath, traversed, and virtual_root_path attached to the request object are always now tuples instead of lists (performance).
Bug Fixes
  • The bfg_alchemy Paster template named “repoze.tm” in its pipeline rather than “repoze.tm2”, causing the startup to fail.
  • Move BBB logic for registering an IAuthenticationPolicy/IForbiddenView/INotFoundView based on older concepts from the router module’s make_app function into the repoze.bfg.zcml.zcml_configure callable, to service compatibility with scripts that use “zope.configuration.xmlconfig” (replace with repoze.bfg.zml.zcml_configure as necessary to get BBB logic)
Documentation
  • Add interface docs related to how to create authentication policies and authorization policies to the “Security” narrative chapter.
  • Added a (fairly sad) “Combining Traversal and URL Dispatch” chapter to the narrative documentation. This explains the usage of *traverse and *subpath in routes URL patters.
  • A “router” chapter explaining the request/response lifecycle at a high level was added.
  • Replaced all mentions and explanations of a routes “context factory” with equivalent explanations of a “root factory” (context factories have been disused).
  • Updated Routes bfgwiki2 tutorial to reflect the fact that context factories are now no longer used.

0.9.1 (2009-06-02)

Features
  • Add API named repoze.bfg.settings.get_settings which retrieves a derivation of values passed as the options value of repoze.bfg.router.make_app. This API should be preferred instead of using getUtility(ISettings). I added a new repoze.bfg.settings API document as well.
Bug Fixes
  • Restored missing entry point declaration for bfg_alchemy paster template, which was accidentally removed in 0.9.
Documentation
  • Fix a reference to wsgiapp in the wsgiapp2 API documentation within the repoze.bfg.wsgi module.
API Removals
  • The repoze.bfg.location.locate API was removed: it didn’t do enough to be very helpful and had a misleading name.

0.9 (2009-06-01)

Bug Fixes
  • It was not possible to register a custom IRoutesContextFactory for use as a default context factory as documented in the “Hooks” chapter.
Features
  • The request_type argument of ZCML view declarations and bfg_view decorators can now be one of the strings GET, POST, PUT, DELETE, or HEAD instead of a reference to the respective interface type imported from repoze.bfg.interfaces.
  • The route ZCML directive now accepts request_type as an alias for its condition_method argument for symmetry with the view directive.
  • The bfg_routesalchemy paster template now provides a unit test and actually uses the database during a view rendering.
Removals
  • Remove repoze.bfg.threadlocal.setManager. It was only used in unit tests.
  • Remove repoze.bfg.wsgi.HTTPException, repoze.bfg.wsgi.NotFound, and repoze.bfg.wsgi.Unauthorized. These classes were disused with the introduction of the IUnauthorizedView and INotFoundView machinery.
Documentation
  • Add description to narrative templating chapter about how to use Chameleon text templates.
  • Changed Views narrative chapter to use method strings rather than interface types, and moved advanced interface type usage to Events narrative chapter.
  • Added a Routes+SQLAlchemy wiki tutorial.

0.9a8 (2009-05-31)

Features
  • It is now possible to register a custom repoze.bfg.interfaces.INotFoundView for a given application. This feature replaces the repoze.bfg.interfaces.INotFoundAppFactory feature previously described in the Hooks chapter. The INotFoundView will be called when the framework detects that a view lookup done as a result of a request fails; it should accept a context object and a request object; it should return an IResponse object (a webob response, basically). See the Hooks narrative chapter of the BFG docs for more info.
  • The error presented when a view invoked by the router returns a non-response object now includes the view’s name for troubleshooting purposes.
Bug Fixes
  • A “new response” event is emitted for forbidden and notfound views.
Deprecations
  • The repoze.bfg.interfaces.INotFoundAppFactory interface has been deprecated in favor of using the new repoze.bfg.interfaces.INotFoundView mechanism.
Renames
  • Renamed repoze.bfg.interfaces.IForbiddenResponseFactory to repoze.bfg.interfaces.IForbiddenView.

0.9a7 (2009-05-30)

Features
  • Remove “context” argument from effective_principals and authenticated_userid function APIs in repoze.bfg.security, effectively a doing reversion to 0.8 and before behavior. Both functions now again accept only the request parameter.

0.9a6 (2009-05-29)

Documentation
  • Changed “BFG Wiki” tutorial to use AuthTktAuthenticationPolicy rather than repoze.who.
Features
  • Add an AuthTktAuthenticationPolicy. This policy retrieves credentials from an auth_tkt cookie managed by the application itself (instead of relying on an upstream data source for authentication data). See the Security API chapter of the documentation for more info.
  • Allow RemoteUserAuthenticationPolicy and RepozeWho1AuthenticationPolicy to accept various constructor arguments. See the Security API chapter of the documentation for more info.

0.9a5 (2009-05-28)

Features
  • Add a get_app API functions to the paster module. This obtains a WSGI application from a config file given a config file name and a section name. See the repoze.bfg.paster API docs for more information.
  • Add a new module named scripting. It contains a get_root API function, which, provided a Router instance, returns a traversal root object and a “closer”. See the repoze.bfg.scripting API docs for more info.

0.9a4 (2009-05-27)

Bug Fixes
  • Try checking for an “old style” security policy after we parse ZCML (thinko).

0.9a3 (2009-05-27)

Features
  • Allow IAuthenticationPolicy and IAuthorizationPolicy to be overridden via ZCML registrations (do ZCML parsing after registering these in router.py).
Documentation
  • Added “BFG Wiki” tutorial to documentation; it describes step-by-step how to create a traversal-based ZODB application with authentication.
Deprecations
  • Added deprecations for imports of ACLSecurityPolicy, InheritingACLSecurityPolicy, RemoteUserACLSecurityPolicy, RemoteUserInheritingACLSecurityPolicy, WhoACLSecurityPolicy, and WhoInheritingACLSecurityPolicy from the repoze.bfg.security module; for the meantime (for backwards compatibility purposes) these live in the repoze.bfg.secpols module. Note however, that the entire concept of a “security policy” is deprecated in BFG in favor of separate authentication and authorization policies, so any use of a security policy will generate additional deprecation warnings even if you do start using repoze.bfg.secpols. repoze.bfg.secpols will disappear in a future release of repoze.bfg.
Deprecated Import Alias Removals
  • Remove repoze.bfg.template module. All imports from this package have been deprecated since 0.3.8. Instead, import get_template, render_template, and render_template_to_response from the repoze.bfg.chameleon_zpt module.
  • Remove backwards compatibility import alias for repoze.bfg.traversal.split_path (deprecated since 0.6.5). This must now be imported as repoze.bfg.traversal.traversal_path).
  • Remove backwards compatibility import alias for repoze.bfg.urldispatch.RoutesContext (deprecated since 0.6.5). This must now be imported as repoze.bfg.urldispatch.DefaultRoutesContext.
  • Removed backwards compatibility import aliases for repoze.bfg.router.get_options and repoze.bfg.router.Settings (deprecated since 0.6.2). These both must now be imported from repoze.bfg.settings.
  • Removed backwards compatibility import alias for repoze.bfg.interfaces.IRootPolicy (deprecated since 0.6.2). It must be imported as repoze.bfg.interfaces.IRootFactory now.
  • Removed backwards compatibility import alias for repoze.bfg.interfaces.ITemplate (deprecated since 0.4.4). It must be imported as repoze.bfg.interfaces.ITemplateRenderer now.
  • Removed backwards compatibility import alias for repoze.bfg.interfaces.ITemplateFactory (deprecated since 0.4.4). It must be imported as repoze.bfg.interfaces.ITemplateRendererFactory now.
  • Removed backwards compatibility import alias for repoze.bfg.chameleon_zpt.ZPTTemplateFactory (deprecated since 0.4.4). This must be imported as repoze.bfg.ZPTTemplateRenderer now.

0.9a2 (2009-05-27)

Features

  • A paster command has been added named “bfgshell”. This command can be used to get an interactive prompt with your BFG root object in the global namespace. E.g.:

    bin/paster bfgshell /path/to/myapp.ini myapp

    See the Project chapter in the BFG documentation for more information.

Deprecations
  • The name repoze.bfg.registry.registry_manager was never an API, but scripts in the wild were using it to set up an environment for use under a debug shell. A backwards compatibility shim has been added for this purpose, but the feature is deprecated.

0.9a1 (2009-5-27)

Features

  • New API functions named forget and remember are available in the security module. The forget function returns headers which will cause the currently authenticated user to be logged out when set in a response. The remember function (when passed the proper arguments) will return headers which will cause a principal to be “logged in” when set in a response. See the Security API chapter of the docs for more info.

  • New keyword arguments to the repoze.bfg.router.make_app call have been added: authentication_policy and authorization_policy. These should, respectively, be an implementation of an authentication policy (an object implementing the repoze.bfg.interfaces.IAuthenticationPolicy interface) and an implementation of an authorization policy (an object implementing repoze.bfg.interfaces.IAuthorizationPolicy). Concrete implementations of authentication policies exist in repoze.bfg.authentication. Concrete implementations of authorization policies exist in repoze.bfg.authorization.

    Both authentication_policy and authorization_policy default to None.

    If authentication_policy is None, but authorization_policy is not None, then authorization_policy is ignored (the ability to do authorization depends on authentication).

    If the authentication_policy argument is not None, and the authorization_policy argument is None, the authorization policy defaults to an authorization implementation that uses ACLs (repoze.bfg.authorization.ACLAuthorizationPolicy).

    We no longer encourage configuration of “security policies” using ZCML, as previously we did for ISecurityPolicy. This is because it’s not uncommon to need to configure settings for concrete authorization or authentication policies using paste .ini parameters; the app entry point for your application is the natural place to do this.

  • Two new abstractions have been added in the way of adapters used by the system: an IAuthorizationPolicy and an IAuthenticationPolicy. A combination of these (as registered by the securitypolicy ZCML directive) take the place of the ISecurityPolicy abstraction in previous releases of repoze.who. The API functions in repoze.who.security (such as authentication_userid, effective_principals, has_permission, and so on) have been changed to try to make use of these new adapters. If you’re using an older ISecurityPolicy adapter, the system will still work, but it will print deprecation warnings when such a policy is used.

  • The way the (internal) IViewPermission utilities registered via ZCML are invoked has changed. They are purely adapters now, returning a boolean result, rather than returning a callable. You shouldn’t have been using these anyway. ;-)

  • New concrete implementations of IAuthenticationPolicy have been added to the repoze.bfg.authentication module: RepozeWho1AuthenticationPolicy which uses repoze.who identity to retrieve authentication data from and RemoteUserAuthenticationPolicy, which uses the REMOTE_USER value in the WSGI environment to retrieve authentication data.

  • A new concrete implementation of IAuthorizationPolicy has been added to the repoze.bfg.authorization module: ACLAuthorizationPolicy which uses ACL inheritance to do authorization.

  • It is now possible to register a custom repoze.bfg.interfaces.IForbiddenResponseFactory for a given application. This feature replaces the repoze.bfg.interfaces.IUnauthorizedAppFactory feature previously described in the Hooks chapter. The IForbiddenResponseFactory will be called when the framework detects an authorization failure; it should accept a context object and a request object; it should return an IResponse object (a webob response, basically). Read the below point for more info and see the Hooks narrative chapter of the BFG docs for more info.

Backwards Incompatibilities
  • Custom NotFound and Forbidden (nee’ Unauthorized) WSGI applications (registered as a utility for INotFoundAppFactory and IUnauthorizedAppFactory) could rely on an environment key named message describing the circumstance of the response. This key has been renamed to repoze.bfg.message (as per the WSGI spec, which requires environment extensions to contain dots).
Deprecations
  • The repoze.bfg.interfaces.IUnauthorizedAppFactory interface has been deprecated in favor of using the new repoze.bfg.interfaces.IForbiddenResponseFactory mechanism.
  • The view_execution_permitted API should now be imported from the repoze.bfg.security module instead of the repoze.bfg.view module.
  • The authenticated_userid and effective_principals APIs in repoze.bfg.security used to only take a single argument (request). They now accept two arguments (context and request). Calling them with a single argument is still supported but issues a deprecation warning. (NOTE: this change was reverted in 0.9a7; meaning the 0.9 versions of these functions again accept request only, just like 0.8 and before).
  • Use of “old-style” security policies (those base on ISecurityPolicy) is now deprecated. See the “Security” chapter of the docs for info about activating an authorization policy and an authentication poicy.

0.8.1 (2009-05-21)

Features

  • Class objects may now be used as view callables (both via ZCML and via use of the bfg_view decorator in Python 2.6 as a class decorator). The calling semantics when using a class as a view callable is similar to that of using a class as a Zope “browser view”: the class’ __init__ must accept two positional parameters (conventionally named context, and request). The resulting instance must be callable (it must have a __call__ method). When called, the instance should return a response. For example:

     from webob import Response

 class MyView(object):
     def __init__(self, context, request):
         self.context = context
         self.request = request

     def __call__(self):
         return Response('hello from %s!' % self.context)

See the "Views" chapter in the documentation and the
``repoze.bfg.view`` API documentation for more information.

  • Removed the pickling of ZCML actions (the code that wrote configure.zcml.cache next to configure.zcml files in projects). The code which managed writing and reading of the cache file was a source of subtle bugs when users switched between imperative (e.g. @bfg_view) registrations and declarative registrations (e.g. the view directive in ZCML) on the same project. On a moderately-sized project (535 ZCML actions and 15 ZCML files), executing actions read from the pickle was saving us only about 200ms (2.5 sec vs 2.7 sec average). On very small projects (1 ZCML file and 4 actions), startup time was comparable, and sometimes even slower when reading from the pickle, and both ways were so fast that it really just didn’t matter anyway.

0.8 (2009-05-18)

Features
  • Added a traverse function to the repoze.bfg.traversal module. This function may be used to retrieve certain values computed during path resolution. See the Traversal API chapter of the documentation for more information about this function.
Deprecations
  • Internal: ITraverser callables should now return a dictionary rather than a tuple. Up until 0.7.0, all ITraversers were assumed to return a 3-tuple. In 0.7.1, ITraversers were assumed to return a 6-tuple. As (by evidence) it’s likely we’ll need to add further information to the return value of an ITraverser callable, 0.8 assumes that an ITraverser return a dictionary with certain elements in it. See the repoze.bfg.interfaces.ITraverser interface for the list of keys that should be present in the dictionary. ITraversers which return tuples will still work, although a deprecation warning will be issued.
Backwards Incompatibilities
  • If your code used the ITraverser interface directly (not via an API function such as find_model) via an adapter lookup, you’ll need to change your code to expect a dictionary rather than a 3- or 6-tuple if your code ever gets return values from the default ModelGraphTraverser or RoutesModelTraverser adapters.

0.8a7 (2009-05-16)

Backwards Incompatibilities

  • The RoutesMapper class in repoze.bfg.urldispatch has been removed, as well as its documentation. It had been deprecated since 0.6.3. Code in repoze.bfg.urldispatch.RoutesModelTraverser which catered to it has also been removed.

  • The semantics of the route ZCML directive have been simplified. Previously, it was assumed that to use a route, you wanted to map a route to an externally registered view. The new route directive instead has a view attribute which is required, specifying the dotted path to a view callable. When a route directive is processed, a view is registered using the name attribute of the route directive as its name and the callable as its value. The view_name and provides attributes of the route directive are therefore no longer used. Effectively, if you were previously using the route directive, it means you must change a pair of ZCML directives that look like this:

    <route
   name="home"
   path=""
   view_name="login"
   factory=".models.root.Root"
 />

<view
   for=".models.root.Root"
   name="login"
   view=".views.login_view"
 />

    To a ZCML directive that looks like this:

    <route
   name="home"
   path=""
   view=".views.login_view"
   factory=".models.root.Root"
 />

    In other words, to make old code work, remove the view directives that were only there to serve the purpose of backing route directives, and move their view= attribute into the route directive itself.

    This change also necessitated that the name attribute of the route directive is now required. If you were previously using route directives without a name attribute, you’ll need to add one (the name is arbitrary, but must be unique among all route and view statements).

    The provides attribute of the route directive has also been removed. This directive specified a sequence of interface types that the generated context would be decorated with. Since route views are always generated now for a single interface (repoze.bfg.IRoutesContext) as opposed to being looked up arbitrarily, there is no need to decorate any context to ensure a view is found.

Documentation
  • Added API docs for the repoze.bfg.testing methods registerAdapter, registerUtiity, registerSubscriber, and cleanUp.
  • Added glossary entry for “root factory”.
  • Noted existence of repoze.bfg.pagetemplate template bindings in “Available Add On Template System Bindings” in Templates chapter in narrative docs.
  • Update “Templates” narrative chapter in docs (expand to show a sample template and correct macro example).
Features
  • Courtesty Carlos de la Guardia, added an alchemy Paster template. This paster template sets up a BFG project that uses SQAlchemy (with SQLite) and uses traversal to resolve URLs. (no Routes areused). This template can be used via paster create -t bfg_alchemy.
  • The Routes Route object used to resolve the match is now put into the environment as bfg.route when URL dispatch is used.
  • You can now change the default Routes “context factory” globally. See the “ZCML Hooks” chapter of the documentation (in the “Changing the Default Routes Context Factory” section).

0.8a6 (2009-05-11)

Features
  • Added a routesalchemy Paster template. This paster template sets up a BFG project that uses SQAlchemy (with SQLite) and uses Routes exclusively to resolve URLs (no traversal root factory is used). This template can be used via paster create -t bfg_routesalchemy.
Documentation
  • Added documentation to the URL Dispatch chapter about how to catch the root URL using a ZCML route directive.
  • Added documentation to the URL Dispatch chapter about how to perform a cleanup function at the end of a request (e.g. close the SQL connection).
Bug Fixes

  • In version 0.6.3, passing a get_root callback (a “root factory”) to repoze.bfg.router.make_app became optional if any route declaration was made in ZCML. The intent was to make it possible to disuse traversal entirely, instead relying entirely on URL dispatch (Routes) to resolve all contexts. However a compound set of bugs prevented usage of a Routes-based root view (a view which responds to “/”). One bug existed in repoze.bfg.urldispatch`, another existed in Routes itself.

    To resolve this issue, the urldispatch module was fixed, and a fork of the Routes trunk was put into the “dev” index named Routes-1.11dev-chrism-home. The source for the fork exists at http://bitbucket.org/chrism/routes-home/; its contents have been merged into the Routes trunk (what will be Routes 1.11).

0.8a5 (2009-05-08)

Features
  • Two new security policies were added: RemoteUserInheritingACLSecurityPolicy and WhoInheritingACLSecurityPolicy. These are security policies which take into account all ACLs defined in the lineage of a context rather than stopping at the first ACL found in a lineage. See the “Security” chapter of the API documentation for more information.
  • The API and narrative documentation dealing with security was changed to introduce the new “inheriting” security policy variants.
  • Added glossary entry for “lineage”.
Deprecations

  • The security policy previously named RepozeWhoIdentityACLSecurityPolicy now has the slightly saner name of WhoACLSecurityPolicy. A deprecation warning is emitted when this policy is imported under the “old” name; usually this is due to its use in ZCML within your application. If you’re getting this deprecation warning, change your ZCML to use the new name, e.g. change:

    <utility
  provides="repoze.bfg.interfaces.ISecurityPolicy"
  factory="repoze.bfg.security.RepozeWhoIdentityACLSecurityPolicy"
  />

    To:

    <utility
  provides="repoze.bfg.interfaces.ISecurityPolicy"
  factory="repoze.bfg.security.WhoACLSecurityPolicy"
  />

0.8a4 (2009-05-04)

Features
  • zope.testing is no longer a direct dependency, although our dependencies (such as zope.interface, repoze.zcml, etc) still depend on it.
  • Tested on Google App Engine. Added a tutorial to the documentation explaining how to deploy a BFG app to GAE.
Backwards Incompatibilities
  • Applications which rely on zope.testing.cleanup.cleanUp in unit tests can still use that function indefinitely. However, for maximum forward compatibility, they should import cleanUp from repoze.bfg.testing instead of from zope.testing.cleanup. The BFG paster templates and docs have been changed to use this function instead of the zope.testing.cleanup version.

0.8a3 (2009-05-03)

Features
  • Don’t require a successful import of zope.testing at BFG application runtime. This allows us to get rid of zope.testing on platforms like GAE which have file limits.

0.8a2 (2009-05-02)

Features
  • We no longer include the configure.zcml of the chameleon.zpt package within the configure.zcml of the “repoze.bfg.includes” package. This has been a no-op for some time now.
  • The repoze.bfg.chameleon_zpt package no longer imports from chameleon.zpt at module scope, deferring the import until later within a method call. The chameleon.zpt package can’t be imported on platforms like GAE.

0.8a1 (2009-05-02)

Deprecation Warning and Import Alias Removals
  • Since version 0.6.1, a deprecation warning has been emitted when the name model_url is imported from the repoze.bfg.traversal module. This import alias (and the deprecation warning) has been removed. Any import of the model_url function will now need to be done from repoze.bfg.url; any import of the name model_url from repoze.bfg.traversal will now fail. This was done to remove a dependency on zope.deferredimport.
  • Since version 0.6.5, a deprecation warning has been emitted when the name RoutesModelTraverser is imported from the repoze.bfg.traversal module. This import alias (and the deprecation warning) has been removed. Any import of the RoutesModelTraverser class will now need to be done from repoze.bfg.urldispatch; any import of the name RoutesModelTraverser from repoze.bfg.traversal will now fail. This was done to remove a dependency on zope.deferredimport.
Features

  • This release of repoze.bfg is “C-free”. This means it has no hard dependencies on any software that must be compiled from C source at installation time. In particular, repoze.bfg no longer depends on the lxml package.

    This change has introduced some backwards incompatibilities, described in the “Backwards Incompatibilities” section below.

  • This release was tested on Windows XP. It appears to work fine and all the tests pass.

Backwards Incompatibilities

Incompatibilities related to making repoze.bfg “C-free”:

  • Removed the repoze.bfg.chameleon_genshi module, and thus support for Genshi-style chameleon templates. Genshi-style Chameleon templates depend upon lxml, which is implemented in C (as opposed to pure Python) and the repoze.bfg core is “C-free” as of this release. You may get Genshi-style Chameleon support back by installing the repoze.bfg.chameleon_genshi package availalable from http://svn.repoze.org/repoze.bfg.chameleon_genshi (also available in the index at http://dist.repoze.org/bfg/0.8/simple). All existing code that depended on the chameleon_genshi module prior to this release of repoze.bfg should work without change after this addon is installed.
  • Removed the repoze.bfg.xslt module and thus support for XSL templates. The repoze.bfg.xslt module depended upon lxml, which is implemented in C, and the repoze.bfg core is “C-free” as of this release. You bay get XSL templating back by installing the repoze.bfg.xslt package available from http://svn.repoze.org/repoze.bfg.xslt/ (also available in the index at http://dist.repoze.org/bfg/0.8/simple). All existing code that depended upon the xslt module prior to this release of repoze.bfg should work without modification after this addon is installed.
  • Removed the repoze.bfg.interfaces.INodeTemplateRenderer interface and the an old b/w compat aliases from that interface to repoze.bfg.interfaces.INodeTemplate. This interface must now be imported from the repoze.bfg.xslt.interfaces package after installation of the repoze.bfg.xslt addon package described above as repoze.bfg.interfaces.INodeTemplateRenderer. This interface was never part of any public API.

Other backwards incompatibilities:

  • The render_template function in repoze.bfg.chameleon_zpt returns Unicode instead of a string. Likewise, the individual values returned by the iterable created by the render_template_to_iterable function are also each Unicode. This is actually a backwards incompatibility inherited from our new use of the combination of chameleon.core 1.0b32 (the non-lxml-depending version) and chameleon.zpt 1.0b16+ ; the chameleon.zpt PageTemplateFile implementation used to return a string, but now returns Unicode.

0.7.1 (2009-05-01)

Features

  • The “paster create” templates have been modified to use links to the new “bfg.repoze.org” and “docs.repoze.org” websites.

  • Added better documentation for virtual hosting at a URL prefix within the virtual hosting docs chapter.

  • The interface for repoze.bfg.interfaces.ITraverser and the built-in implementations that implement the interface (repoze.bfg.traversal.ModelGraphTraverser, and repoze.bfg.urldispatch.RoutesModelTraverser) now expect the __call__ method of an ITraverser to return 3 additional arguments: traversed, virtual_root, and virtual_root_path (the old contract was that the __call__ method of an ITraverser returned; three arguments, the contract new is that it returns six). traversed will be a sequence of Unicode names that were traversed (including the virtual root path, if any) or None if no traversal was performed, virtual_root will be a model object representing the virtual root (or the physical root if traversal was not performed), and virtual_root_path will be a sequence representing the virtual root path (a sequence of Unicode names) or None if traversal was not performed.

    Six arguments are now returned from BFG ITraversers. They are returned in this order: context, view_name, subpath, traversed, virtual_root, and virtual_root_path.

    Places in the BFG code which called an ITraverser continue to accept a 3-argument return value, although BFG will generate and log a warning when one is encountered.

  • The request object now has the following attributes: traversed (the sequence of names traversed or None if traversal was not performed), virtual_root (the model object representing the virtual root, including the virtual root path if any), and virtual_root_path (the seuquence of names representing the virtual root path or None if traversal was not performed).

  • A new decorator named wsgiapp2 was added to the repoze.bfg.wsgi module. This decorator performs the same function as repoze.bfg.wsgi.wsgiapp except it fixes up the SCRIPT_NAME, and PATH_INFO environment values before invoking the WSGI subapplication.

  • The repoze.bfg.testing.DummyRequest object now has default attributes for traversed, virtual_root, and virtual_root_path.

  • The RoutesModelTraverser now behaves more like the Routes “RoutesMiddleware” object when an element in the match dict is named path_info (usually when there’s a pattern like http://foo/*path_info). When this is the case, the PATH_INFO environment variable is set to the value in the match dict, and the SCRIPT_NAME is appended to with the prefix of the original PATH_INFO not including the value of the new variable.

  • The notfound debug now shows the traversed path, the virtual root, and the virtual root path too.

  • Speed up / clarify ‘traversal’ module’s ‘model_path’, ‘model_path_tuple’, and ‘_model_path_list’ functions.

Backwards Incompatibilities

  • In previous releases, the repoze.bfg.url.model_url, repoze.bfg.traversal.model_path and repoze.bfg.traversal.model_path_tuple functions always ignored the __name__ argument of the root object in a model graph ( effectively replacing it with a leading / in the returned value) when a path or URL was generated. The code required to perform this operation was not efficient. As of this release, the root object in a model graph must have a __name__ attribute that is either None or the empty string ('') for URLs and paths to be generated properly from these APIs. If your root model object has a __name__ argument that is not one of these values, you will need to change your code for URLs and paths to be generated properly. If your model graph has a root node with a string __name__ that is not null, the value of __name__ will be prepended to every path and URL generated.

  • The repoze.bfg.location.LocationProxy class and the repoze.bfg.location.ClassAndInstanceDescr class have both been removed in order to be able to eventually shed a dependency on zope.proxy. Neither of these classes was ever an API.

  • In all previous releases, the repoze.bfg.location.locate function worked like so: if a model did not explicitly provide the repoze.bfg.interfaces.ILocation interface, locate returned a LocationProxy object representing model with its __parent__ attribute assigned to parent and a __name__ attribute assigned to __name__. In this release, the repoze.bfg.location.locate function simply jams the __name__ and __parent__ attributes on to the supplied model unconditionally, no matter if the object implements ILocation or not, and it never returns a proxy. This was done because the LocationProxy behavior has now moved into an add-on package (repoze.bfg.traversalwrapper), in order to eventually be able to shed a dependency on zope.proxy.

  • In all previous releases, by default, if traversal was used (as opposed to URL-dispatch), and the root object supplied the``repoze.bfg.interfaces.ILocation`` interface, but the children returned via its __getitem__ returned an object that did not implement the same interface, repoze.bfg provided some implicit help during traversal. This traversal feature wrapped subobjects from the root (and thereafter) that did not implement ILocation in proxies which automatically provided them with a __name__ and __parent__ attribute based on the name being traversed and the previous object traversed. This feature has now been removed from the base repoze.bfg package for purposes of eventually shedding a dependency on zope.proxy.

    In order to re-enable the wrapper behavior for older applications which cannot be changed, register the “traversalwrapper” ModelGraphTraverser as the traversal policy, rather than the default ModelGraphTraverser. To use this feature, you will need to install the repoze.bfg.traversalwrapper package (an add-on package, available at http://svn.repoze.org/repoze.bfg.traversalwrapper) Then change your application’s configure.zcml to include the following stanza:

    <adapter

    factory=”repoze.bfg.traversalwrapper.ModelGraphTraverser” provides=”repoze.bfg.interfaces.ITraverserFactory” for=”*” />

    When this ITraverserFactory is used instead of the default, no object in the graph (even the root object) must supply a __name__ or __parent__ attribute. Even if subobjects returned from the root do implement the ILocation interface, these will still be wrapped in proxies that override the object’s “real” __parent__ and __name__ attributes.

    See also changes to the “Models” chapter of the documentation (in the “Location-Aware Model Instances”) section.

0.7.0 (2009-04-11)

Bug Fixes
  • Fix a bug in repoze.bfg.wsgi.HTTPException: the content length was returned as an int rather than as a string.
  • Add explicit dependencies on zope.deferredimport, zope.deprecation, and zope.proxy for forward compatibility reasons (zope.component will stop relying on zope.deferredimport soon and although we use it directly, it’s only a transitive dependency, and ‘’zope.deprecation`` and zope.proxy are used directly even though they’re only transitive dependencies as well).
  • Using model_url or model_path against a broken model graph (one with models that had a non-root model with a __name__ of None) caused an inscrutable error to be thrown: ( if not _must_quote[cachekey].search(s): TypeError: expected string or buffer). Now URLs and paths generated against graphs that have None names in intermediate nodes will replace the None with the empty string, and, as a result, the error won’t be raised. Of course the URL or path will still be bogus.
Features
  • Make it possible to have testing.DummyTemplateRenderer return some nondefault string representation.
  • Added a new anchor keyword argument to model_url. If anchor is present, its string representation will be used as a named anchor in the generated URL (e.g. if anchor is passed as foo and the model URL is http://example.com/model/url, the generated URL will be http://example.com/model/url#foo).
Backwards Incompatibilities

  • The default request charset encoding is now utf-8. As a result, the request machinery will attempt to decode values from the utf-8 encoding to Unicode automatically when they are obtained via request.params, request.GET, and request.POST. The previous behavior of BFG was to return a bytestring when a value was accessed in this manner. This change will break form handling code in apps that rely on values from those APIs being considered bytestrings. If you are manually decoding values from form submissions in your application, you’ll either need to change the code that does that to expect Unicode values from request.params, request.GET and request.POST, or you’ll need to explicitly reenable the previous behavior. To reenable the previous behavior, add the following to your application’s configure.zcml:

    <subscriber for="repoze.bfg.interfaces.INewRequest"
            handler="repoze.bfg.request.make_request_ascii"/>

    See also the documentation in the “Views” chapter of the BFG docs entitled “Using Views to Handle Form Submissions (Unicode and Character Set Issues)”.

Documentation
  • Add a section to the narrative Views chapter entitled “Using Views to Handle Form Submissions (Unicode and Character Set Issues)” explaining implicit decoding of form data values.

0.6.9 (2009-02-16)

Bug Fixes
  • lru cache was unstable under concurrency (big surprise!) when it tried to redelete a key in the cache that had already been deleted. Symptom: line 64 in put:del data[oldkey]:KeyError: ‘/some/path’. Now we just ignore the key error if we can’t delete the key (it has already been deleted).
  • Empty location names in model paths when generating a URL using repoze.bfg.model_url based on a model obtained via traversal are no longer ignored in the generated URL. This means that if a non-root model object has a __name__ of '', the URL will reflect it (e.g. model_url will generate http://foo/bar//baz if an object with the __name__ of '' is a child of bar and the parent of baz). URLs generated with empty path segments are, however, still irresolveable by the model graph traverser on request ingress (the traverser strips empty path segment names).
Features
  • Microspeedups of repoze.bfg.traversal.model_path, repoze.bfg.traversal.model_path_tuple, repoze.bfg.traversal.quote_path_segment, and repoze.bfg.url.urlencode.
  • add zip_safe = false to setup.cfg.
Documentation
  • Add a note to the repoze.bfg.traversal.quote_path_segment API docs about caching of computed values.
Implementation Changes
  • Simplification of repoze.bfg.traversal.TraversalContextURL.__call__ (it now uses repoze.bfg.traversal.model_path instead of rolling its own path-generation).

0.6.8 (2009-02-05)

Backwards Incompatibilities

  • The repoze.bfg.traversal.model_path API now returns a quoted string rather than a string represented by series of unquoted elements joined via / characters. Previously it returned a string or unicode object representing the model path, with each segment name in the path joined together via / characters, e.g. /foo /bar. Now it returns a string, where each segment is a UTF-8 encoded and URL-quoted element e.g. /foo%20/bar. This change was (as discussed briefly on the repoze-dev maillist) necessary to accomodate model objects which themselves have __name__ attributes that contain the / character.

    For people that have no models that have high-order Unicode __name__ attributes or __name__ attributes with values that require URL-quoting with in their model graphs, this won’t cause any issue. However, if you have code that currently expects model_path to return an unquoted string, or you have an existing application with data generated via the old method, and you’re too lazy to change anything, you may wish replace the BFG-imported model_path in your code with this function (this is the code of the “old” model_path implementation):

    from repoze.bfg.location import lineage

def i_am_too_lazy_to_move_to_the_new_model_path(model, *elements):
    rpath = []
    for location in lineage(model):
        if location.__name__:
            rpath.append(location.__name__)
    path = '/' + '/'.join(reversed(rpath))
    if elements:
        suffix = '/'.join(elements)
        path = '/'.join([path, suffix])
    return path

  • The repoze.bfg.traversal.find_model API no longer implicitly converts unicode representations of a full path passed to it as a Unicode object into a UTF-8 string. Callers should either use prequoted path strings returned by repoze.bfg.traversal.model_path, or tuple values returned by the result of repoze.bfg.traversal.model_path_tuple or they should use the guidelines about passing a string path argument described in the find_model API documentation.

Bugfixes
  • Each argument contained in elements passed to repoze.bfg.traversal.model_path will now have any / characters contained within quoted to %2F in the returned string. Previously, / characters in elements were left unquoted (a bug).
Features
  • A repoze.bfg.traversal.model_path_tuple API was added. This API is an alternative to model_path (which returns a string); model_path_tuple returns a model path as a tuple (much like Zope’s getPhysicalPath).
  • A repoze.bfg.traversal.quote_path_segment API was added. This API will quote an individual path segment (string or unicode object). See the repoze.bfg.traversal API documentation for more information.
  • The repoze.bfg.traversal.find_model API now accepts “path tuples” (see the above note regarding model_path_tuple) as well as string path representations (from repoze.bfg.traversal.model_path) as a path argument.
  • Add ` renderer` argument (defaulting to None) to repoze.bfg.testing.registerDummyRenderer. This makes it possible, for instance, to register a custom renderer that raises an exception in a unit test.
Implementation Changes
  • Moved _url_quote function back to repoze.bfg.traversal from repoze.bfg.url. This is not an API.

0.6.7 (2009-01-27)

Features
  • The repoze.bfg.url.model_url API now works against contexts derived from Routes URL dispatch (Routes.util.url_for is called under the hood).
  • “Virtual root” support for traversal-based applications has been added. Virtual root support is useful when you’d like to host some model in a repoze.bfg model graph as an application under a URL pathname that does not include the model path itself. For more information, see the (new) “Virtual Hosting” chapter in the documentation.
  • A repoze.bfg.traversal.virtual_root API has been added. When called, it returns the virtual root object (or the physical root object if no virtual root has been specified).
Implementation Changes
  • repoze.bfg.traversal.RoutesModelTraverser has been moved to repoze.bfg.urldispatch.
  • model_url URL generation is now performed via an adapter lookup based on the context and the request.
  • ZCML which registers two adapters for the IContextURL interface has been added to the configure.zcml in repoze.bfg.includes.

0.6.6 (2009-01-26)

Implementation Changes
  • There is an indirection in repoze.bfg.url.model_url now that consults a utility to generate the base model url (without extra elements or a query string). Eventually this will service virtual hosting; for now it’s undocumented and should not be hooked.

0.6.5 (2009-01-26)

Features
  • You can now override the NotFound and Unauthorized responses that repoze.bfg generates when a view cannot be found or cannot be invoked due to lack of permission. See the “ZCML Hooks” chapter in the docs for more information.
  • Added Routes ZCML directive attribute explanations in documentation.
  • Added a traversal_path API to the traversal module; see the “traversal” API chapter in the docs. This was a function previously known as split_path that was not an API but people were using it anyway. Unlike split_path, it now returns a tuple instead of a list (as its values are cached).
Behavior Changes
  • The repoze.bfg.view.render_view_to_response API will no longer raise a ValueError if an object returned by a view function it calls does not possess certain attributes (headerlist, app_iter, status). This API used to attempt to perform a check using the is_response function in repoze.bfg.view, and raised a ValueError if the is_response check failed. The responsibility is now the caller’s to ensure that the return value from a view function is a “real” response.
  • WSGI environ dicts passed to repoze.bfg ‘s Router must now contain a REQUEST_METHOD key/value; if they do not, a KeyError will be raised (speed).
  • It is no longer permissible to pass a “nested” list of principals to repoze.bfg.ACLAuthorizer.permits (e.g. ['fred', ['larry', 'bob']]). The principals list must be fully expanded. This feature was never documented, and was never an API, so it’s not a backwards incompatibility.
  • It is no longer permissible for a security ACE to contain a “nested” list of permissions (e.g. (Allow, Everyone, ['read', ['view', ['write', 'manage']]])`)`.  The list must instead be fully expanded (e.g. ``(Allow, Everyone, ['read', 'view', 'write', 'manage])). This feature was never documented, and was never an API, so it’s not a backwards incompatibility.
  • The repoze.bfg.urldispatch.RoutesRootFactory now injects the wsgiorg.routing_args environment variable into the environ when a route matches. This is a tuple of ((), routing_args) where routing_args is the value that comes back from the routes mapper match (the “match dict”).
  • The repoze.bfg.traversal.RoutesModelTraverser class now wants to obtain the view_name and subpath from the wsgiorgs.routing_args environment variable. It falls back to obtaining these from the context for backwards compatibility.
Implementation Changes
  • Get rid of repoze.bfg.security.ACLAuthorizer: the ACLSecurityPolicy now does what it did inline.
  • Get rid of repoze.bfg.interfaces.NoAuthorizationInformation exception: it was used only by ACLAuthorizer.
  • Use a homegrown NotFound error instead of webob.exc.HTTPNotFound (the latter is slow).
  • Use a homegrown Unauthorized error instead of webob.exc.Unauthorized (the latter is slow).
  • the repoze.bfg.lru.lru_cached decorator now uses functools.wraps in order to make documentation of LRU-cached functions possible.
  • Various speed micro-tweaks.
Bug Fixes
  • repoze.bfg.testing.DummyModel did not have a get method; it now does.

0.6.4 (2009-01-23)

Backwards Incompatibilities
  • The unicode_path_segments configuration variable and the BFG_UNICODE_PATH_SEGMENTS configuration variable have been removed. Path segments are now always passed to model __getitem__ methods as unicode. “True” has been the default for this setting since 0.5.4, but changing this configuration setting to false allowed you to go back to passing raw path element strings to model __getitem__ methods. Removal of this knob services a speed goal (we get about +80 req/s by removing the check), and it’s clearer just to always expect unicode path segments in model __getitem__ methods.
Implementation Changes

  • repoze.bfg.traversal.split_path now also handles decoding path segments to unicode (for speed, because its results are cached).

  • repoze.bfg.traversal.step was made a method of the

    ModelGraphTraverser.

  • Use “precooked” Request subclasses (e.g. repoze.bfg.request.GETRequest) that correspond to HTTP request methods within router.py when constructing a request object rather than using alsoProvides to attach the proper interface to an unsubclassed webob.Request. This pattern is purely an optimization (e.g. preventing calls to alsoProvides means the difference between 590 r/s and 690 r/s on a MacBook 2GHz).

  • Tease out an extra 4% performance boost by changing the Router; instead of using imported ZCA APIs, use the same APIs directly against the registry that is an attribute of the Router.

  • The registry used by BFG is now a subclass of zope.component.registry.Components (defined as repoze.bfg.registry.Registry); it has a notify method, a registerSubscriptionAdapter and a registerHandler method. If no subscribers are registered via registerHandler or registerSubscriptionAdapter, notify is a noop for speed.

  • The Allowed and Denied classes in repoze.bfg.security now are lazier about constructing the representation of a reason message for speed; repoze.bfg.view_execution_permitted takes advantage of this.

  • The is_response check was sped up by about half at the expense of making its code slightly uglier.

New Modules
  • repoze.bfg.lru implements an LRU cache class and a decorator for internal use.

0.6.3 (2009-01-19)

Bug Fixes
  • Readd root_policy attribute on Router object (as a property which returns the IRootFactory utility). It was inadvertently removed in 0.6.2. Code in the wild depended upon its presence (esp. scripts and “debug” helpers).
Features
  • URL-dispatch has been overhauled: it is no longer necessary to manually create a RoutesMapper in your application’s entry point callable in order to use URL-dispatch (aka Routes). A new route directive has been added to the available list of ZCML directives. Each route directive inserted into your application’s configure.zcml establishes a Routes mapper connection. If any route declarations are made via ZCML within a particular application, the get_root callable passed in to repoze.bfg.router.make_app will automatically be wrapped in the equivalent of a RoutesMapper. Additionally, the new route directive allows the specification of a context_interfaces attribute for a route, this will be used to tag the manufactured routes context with specific interfaces when a route specifying a context_interfaces attribute is matched.
  • A new interface repoze.bfg.interfaces.IContextNotFound was added. This interface is attached to a “dummy” context generated when Routes cannot find a match and there is no “fallback” get_root callable that uses traversal.
  • The bfg_starter and bfg_zodb “paster create” templates now contain images and CSS which are displayed when the default page is displayed after initial project generation.
  • Allow the repoze.bfg.view.static helper to be passed a relative root_path name; it will be considered relative to the file in which it was called.
  • The functionality of repoze.bfg.convention has been merged into the core. Applications which make use of repoze.bfg.convention will continue to work indefinitely, but it is recommended that apps stop depending upon it. To do so, substitute imports of repoze.bfg.convention.bfg_view with imports of repoze.bfg.view.bfg_view, and change the stanza in ZCML from <convention package="."> to <scan package=".">. As a result of the merge, bfg has grown a new dependency: martian.
  • View functions which use the pushpage decorator are now pickleable (meaning their use won’t prevent a configure.zcml.cache file from being written to disk).
  • Instead of invariably using webob.Request as the “request factory” (e.g. in the Router class) and webob.Response and the “response factory” (e.g. in render_template_to_response), allow both to be overridden via a ZCML utility hook. See the “Using ZCML Hooks” chapter of the documentation for more information.
Deprecations
  • The class repoze.bfg.urldispatch.RoutesContext has been renamed to repoze.bfg.urldispatch.DefaultRoutesContext. The class should be imported by the new name as necessary (although in reality it probably shouldn’t be imported from anywhere except internally within BFG, as it’s not part of the API).
Implementation Changes
  • The repoze.bfg.wsgi.wsgiapp decorator now uses webob.Request.get_response to do its work rather than relying on homegrown WSGI code.
  • The repoze.bfg.view.static helper now uses webob.Request.get_response to do its work rather than relying on homegrown WSGI code.
  • The repoze.bfg.urldispatch.RoutesModelTraverser class has been moved to repoze.bfg.traversal.RoutesModelTraverser.
  • The repoze.bfg.registry.makeRegistry function was renamed to repoze.bfg.registry.populateRegistry and now accepts a registry argument (which should be an instance of zope.component.registry.Components).
Documentation Additions
  • Updated narrative urldispatch chapter with changes required by <route..> ZCML directive.
  • Add a section on “Using BFG Security With URL Dispatch” into the urldispatch chapter of the documentation.
  • Better documentation of security policy implementations that ship with repoze.bfg.
  • Added a “Using ZPT Macros in repoze.bfg” section to the narrative templating chapter.

0.6.2 (2009-01-13)

Features
  • Tests can be run with coverage output if you’ve got nose installed in the interpreter which you use to run tests. Using an interpreter with nose installed, do python setup.py nosetests within a checkout of the repoze.bfg package to see test coverage output.
  • Added a post argument to the repoze.bfg.testing:DummyRequest constructor.
  • Added __len__ and __nonzero__ to repoze.bfg.testing:DummyModel.
  • The repoze.bfg.registry.get_options callable (now renamed to repoze.bfg.setings.get_options) used to return only framework-specific keys and values in the dictionary it returned. It now returns all the keys and values in the dictionary it is passed plus any framework-specific settings culled from the environment. As a side effect, all PasteDeploy application-specific config file settings are made available as attributes of the ISettings utility from within BFG.
  • Renamed the existing BFG paster template to bfg_starter. Added another template (bfg_zodb) showing default ZODB setup using repoze.zodbconn.
  • Add a method named assert_ to the DummyTemplateRenderer. This method accepts keyword arguments. Each key/value pair in the keyword arguments causes an assertion to be made that the renderer received this key with a value equal to the asserted value.
  • Projects generated by the paster templates now use the DummyTemplateRenderer.assert_ method in their view tests.
  • Make the (internal) thread local registry manager maintain a stack of registries in order to make it possible to call one BFG application from inside another.
  • An interface specific to the HTTP verb (GET/PUT/POST/DELETE/HEAD) is attached to each request object on ingress. The HTTP-verb-related interfaces are defined in repoze.bfg.interfaces and are IGETRequest, IPOSTRequest, IPUTRequest, IDELETERequest and IHEADRequest. These interfaces can be specified as the request_type attribute of a bfg view declaration. A view naming a specific HTTP-verb-matching interface will be found only if the view is defined with a request_type that matches the HTTP verb in the incoming request. The more general IRequest interface can be used as the request_type to catch all requests (and this is indeed the default). All requests implement IRequest. The HTTP-verb-matching idea was pioneered by repoze.bfg.restrequest . That package is no longer required, but still functions fine.
Bug Fixes
  • Fix a bug where the Paste configuration’s unicode_path_segments (and os.environ’s BFG_UNICODE_PATH_SEGMENTS) may have been defaulting to false in some circumstances. It now always defaults to true, matching the documentation and intent.
  • The repoze.bfg.traversal.find_model API did not work properly when passed a path argument which was unicode and contained high-order bytes when the unicode_path_segments or BFG_UNICODE_PATH_SEGMENTS configuration variables were “true”.
  • A new module was added: repoze.bfg.settings. This contains deployment-settings-related code.
Implementation Changes
  • The make_app callable within repoze.bfg.router now registers the root_policy argument as a utility (unnamed, using the new repoze.bfg.interfaces.IRootFactory as a provides interface) rather than passing it as the first argument to the repoze.bfg.router.Router class. As a result, the repoze.bfg.router.Router router class only accepts a single argument: registry. The repoze.bfg.router.Router class retrieves the root policy via a utility lookup now. The repoze.bfg.router.make_app API also now performs some important application registrations that were previously handled inside repoze.bfg.registry.makeRegistry.
New Modules
  • A repoze.bfg.settings module was added. It contains code related to deployment settings. Most of the code it contains was moved to it from the repoze.bfg.registry module.
Behavior Changes
  • The repoze.bfg.settings.Settings class (an instance of which is registered as a utility providing repoze.bfg.interfaces.ISettings when any application is started) now automatically calls repoze.bfg.settings.get_options on the options passed to its constructor. This means that usage of get_options within an application’s make_app function is no longer required (the “raw” options dict or None may be passed).
  • Remove old cold which attempts to recover from trying to unpickle a z3c.pt template; Chameleon has been the templating engine for a good long time now. Running repoze.bfg against a sandbox that has pickled z3c.pt templates it will now just fail with an unpickling error, but can be fixed by deleting the template cache files.
Deprecations
  • Moved the repoze.bfg.registry.Settings class. This has been moved to repoze.bfg.settings.Settings. A deprecation warning is issued when it is imported from the older location.
  • Moved the repoze.bfg.registry.get_options function This has been moved to repoze.bfg.settings.get_options. A deprecation warning is issued when it is imported from the older location.
  • The repoze.bfg.interfaces.IRootPolicy interface was renamed within the interfaces package. It has been renamed to IRootFactory. A deprecation warning is issued when it is imported from the older location.

0.6.1 (2009-01-06)

New Modules
  • A new module repoze.bfg.url has been added. It contains the model_url API (moved from repoze.bfg.traversal) and an implementation of urlencode (like Python’s urllib.urlencode) which can handle Unicode keys and values in parameters to the query argument.
Deprecations
  • The model_url function has been moved from repoze.bfg.traversal into repoze.bfg.url. It can still be imported from repoze.bfg.traversal but an import from repoze.bfg.traversal will emit a DeprecationWarning.
Features
  • A static helper class was added to the repoze.bfg.views module. Instances of this class are willing to act as BFG views which return static resources using files on disk. See the repoze.bfg.view docs for more info.
  • The repoze.bfg.url.model_url API (nee’ repoze.bfg.traversal.model_url) now accepts and honors a keyword argument named query. The value of this argument will be used to compose a query string, which will be attached to the generated URL before it is returned. See the API docs (in the docs directory or on the web) for more information.

0.6 (2008-12-26)

Backwards Incompatibilities

  • Rather than prepare the “stock” implementations of the ZCML directives from the zope.configuration package for use under repoze.bfg, repoze.bfg now makes available the implementations of directives from the repoze.zcml package (see http://static.repoze.org/zcmldocs). As a result, the repoze.bfg package now depends on the repoze.zcml package, and no longer depends directly on the zope.component, zope.configuration, zope.interface, or zope.proxy packages.

    The primary reason for this change is to enable us to eventually reduce the number of inappropriate repoze.bfg Zope package dependencies, as well as to shed features of dependent package directives that don’t make sense for repoze.bfg.

    Note that currently the set of requirements necessary to use bfg has not changed. This is due to inappropriate Zope package requirements in chameleon.zpt, which will hopefully be remedied soon. NOTE: in lemonade index a 1.0b8-repozezcml0 package exists which does away with these requirements.

  • BFG applications written prior to this release which expect the “stock” zope.component ZCML directive implementations (e.g. adapter, subscriber, or utility) to function now must either 1) include the meta.zcml file from zope.component manually (e.g. <include package="zope.component" file="meta.zcml">) and include the zope.security package as an install_requires dependency or 2) change the ZCML in their applications to use the declarations from repoze.zcml instead of the stock declarations. repoze.zcml only makes available the adapter, subscriber and utility directives.

    In short, if you’ve got an existing BFG application, after this update, if your application won’t start due to an import error for “zope.security”, the fastest way to get it working again is to add zope.security to the “install_requires” of your BFG application’s setup.py, then add the following ZCML anywhere in your application’s configure.zcml:

    <include package="zope.component" file="meta.zcml">

    Then re-setup.py develop or reinstall your application.

  • The http://namespaces.repoze.org/bfg XML namespace is now the default XML namespace in ZCML for paster-generated applications. The docs have been updated to reflect this.

  • The copies of BFG’s meta.zcml and configure.zcml were removed from the root of the repoze.bfg package. In 0.3.6, a new package named repoze.bfg.includes was added, which contains the “correct” copies of these ZCML files; the ones that were removed were for backwards compatibility purposes.

  • The BFG view ZCML directive no longer calls zope.component.interface.provideInterface for the for interface. We don’t support provideInterface in BFG because it mutates the global registry.

Other
  • The minimum requirement for chameleon.core is now 1.0b13. The minimum requirement for chameleon.zpt is now 1.0b8. The minimum requirement for chameleon.genshi is now 1.0b2.
  • Updated paster template “ez_setup.py” to one that requires setuptools 0.6c9.
  • Turn view_execution_permitted from the repoze.bfg.view module into a documented API.
  • Doc cleanups.
  • Documented how to create a view capable of serving static resources.

0.5.6 (2008-12-18)

  • Speed up traversal.model_url execution by using a custom url quoting function instead of Python’s urllib.quote, by caching URL path segment quoting and encoding results, by disusing Python’s urlparse.urljoin in favor of a simple string concatenation, and by using ob.__class__ is unicode rather than isinstance(ob, unicode) in one strategic place.

0.5.5 (2008-12-17)

Backwards Incompatibilities

  • In the past, during traversal, the ModelGraphTraverser (the default traverser) always passed each URL path segment to any __getitem__ method of a model object as a byte string (a str object). Now, by default the ModelGraphTraverser attempts to decode the path segment to Unicode (a unicode object) using the UTF-8 encoding before passing it to the __getitem__ method of a model object. This makes it possible for model objects to be dumber in __getitem__ when trying to resolve a subobject, as model objects themselves no longer need to try to divine whether or not to try to decode the path segment passed by the traverser.

    Note that since 0.5.4, URLs generated by repoze.bfg’s model_url API will contain UTF-8 encoded path segments as necessary, so any URL generated by BFG itself will be decodeable by the traverser. If another application generates URLs to a BFG application, to be resolved successully, it should generate the URL with UTF-8 encoded path segments to be successfully resolved. The decoder is not at all magical: if a non-UTF-8-decodeable path segment (e.g. one encoded using UTF-16 or some other insanity) is passed in the URL, BFG will raise a TypeError with a message indicating it could not decode the path segment.

    To turn on the older behavior, where path segments were not decoded to Unicode before being passed to model object __getitem__ by the traverser, and were passed as a raw byte string, set the unicode_path_segments configuration setting to a false value in your BFG application’s section of the paste .ini file, for example:

    unicode_path_segments = False

    Or start the application using the BFG_UNICODE_PATH_SEGMENT envvar set to a false value:

    BFG_UNICODE_PATH_SEGMENTS=0

0.5.4 (2008-12-13)

Backwards Incompatibilities
  • URL-quote “extra” element names passed in as **elements to the traversal.model_url API. If any of these names is a Unicode string, encode it to UTF-8 before URL-quoting. This is a slight backwards incompatibility that will impact you if you were already UTF-8 encoding or URL-quoting the values you passed in as elements to this API.
Bugfixes
  • UTF-8 encode each segment in the model path used to generate a URL before url-quoting it within the traversal.model_url API. This is a bugfix, as Unicode cannot always be successfully URL-quoted.
Features
  • Make it possible to run unit tests using a buildout-generated Python “interpreter”.
  • Add request.root to router.Router in order to have easy access to the application root.

0.5.3 (2008-12-07)

  • Remove the ITestingTemplateRenderer interface. When testing.registerDummyRenderer is used, it instead registers a dummy implementation using ITemplateRenderer interface, which is checked for when the built-in templating facilities do rendering. This change also allows developers to make explcit named utility registrations in the ZCML registry against ITemplateRenderer; these will be found before any on-disk template is looked up.

0.5.2 (2008-12-05)

  • The component registration handler for views (functions or class instances) now observes component adaptation annotations (see zope.component.adaptedBy) and uses them before the fallback values for for_ and request_type. This change does not affect existing code insomuch as the code does not rely on these defaults when an annotation is set on the view (unlikely). This means that for a new-style class you can do zope.component.adapts(ISomeContext, ISomeRequest) at class scope or at module scope as a decorator to a bfg view function you can do @zope.component.adapter(ISomeContext, ISomeRequest). This differs from r.bfg.convention inasmuch as you still need to put something in ZCML for the registrations to get done; it’s only the defaults that will change if these declarations exist.
  • Strip all slashes from end and beginning of path in clean_path within traversal machinery.

0.5.1 (2008-11-25)

  • Add keys, items, and values methods to testing.DummyModel.
  • Add __delitem__ method to testing.DummyModel.

0.5.0 (2008-11-18)

  • Fix ModelGraphTraverser; don’t try to change the __name__ or __parent__ of an object that claims it implements ILocation during traversal even if the __name__ or __parent__ of the object traversed does not match the name used in the traversal step or the or the traversal parent . Rationale: it was insane to do so. This bug was only found due to a misconfiguration in an application that mistakenly had intermediate persistent non-ILocation objects; traversal was causing a persistent write on every request under this setup.
  • repoze.bfg.location.locate now unconditionally sets __name__ and __parent__ on objects which provide ILocation (it previously only set them conditionally if they didn’t match attributes already present on the object via equality).

0.4.9 (2008-11-17)

  • Add chameleon text template API (chameleon ${name} renderings where the template does not need to be wrapped in any containing XML).
  • Change docs to explain install in terms of a virtualenv (unconditionally).
  • Make pushpage decorator compatible with repoze.bfg.convention’s bfg_view decorator when they’re stacked.
  • Add content_length attribute to testing.DummyRequest.
  • Change paster template tests.py to include a true unit test. Retain old test as an integration test. Update documentation.
  • Document view registrations against classes and repoze.bfg.convention in context.
  • Change the default paster template to register its single view against a class rather than an interface.
  • Document adding a request type interface to the request via a subscriber function in the events narrative documentation.

0.4.8 (2008-11-12)

Backwards Incompatibilities
  • repoze.bfg.traversal.model_url now always appends a slash to all generated URLs unless further elements are passed in as the third and following arguments. Rationale: views often use model_url without the third-and-following arguments in order to generate a URL for a model in order to point at the default view of a model. The URL that points to the default view of the root model is technically http://mysite/ as opposed to http://mysite (browsers happen to ask for ‘/’ implicitly in the GET request). Because URLs are never automatically generated for anything except models by model_url, and because the root model is not really special, we continue this pattern. The impact of this change is minimal (at most you will have too many slashes in your URL, which BFG deals with gracefully anyway).

0.4.7 (2008-11-11)

Features
  • Allow testing.registerEventListener to be used with Zope 3 style “object events” (subscribers accept more than a single event argument). We extend the list with the arguments, rather than append.

0.4.6 (2008-11-10)

Bug Fixes
  • The model_path and model_url traversal APIs returned the wrong value for the root object (e.g. model_path returned '' for the root object, while it should have been returning '/').

0.4.5 (2008-11-09)

Features
  • Added a clone method and a __contains__ method to the DummyModel testing object.
  • Allow DummyModel objects to receive extra keyword arguments, which will be attached as attributes.
  • The DummyTemplateRenderer now returns self as its implementation.

0.4.4 (2008-11-08)

Features
  • Added a repoze.bfg.testing module to attempt to make it slightly easier to write unittest-based automated tests of BFG applications. Information about this module is in the documentation.
  • The default template renderer now supports testing better by looking for ITestingTemplateRenderer using a relative pathname. This is exposed indirectly through the API named registerTemplateRenderer in repoze.bfg.testing.
Deprecations
  • The names repoze.bfg.interfaces.ITemplate , repoze.bfg.interfaces.ITemplateFactory and repoze.bfg.interfaces.INodeTemplate have been deprecated. These should now be imported as repoze.bfg.interfaces.ITemplateRenderer and repoze.bfg.interfaces.ITemplateRendererFactory, and INodeTemplateRenderer respectively.
  • The name repoze.bfg.chameleon_zpt.ZPTTemplateFactory is deprecated. Use repoze.bfg.chameleon_zpt.ZPTTemplateRenderer.
  • The name repoze.bfg.chameleon_genshi.GenshiTemplateFactory is deprecated. Use repoze.bfg.chameleon_genshi.GenshiTemplateRenderer.
  • The name repoze.bfg.xslt.XSLTemplateFactory is deprecated. Use repoze.bfg.xslt.XSLTemplateRenderer.

0.4.3 (2008-11-02)

Bug Fixes
  • Not passing the result of “get_options” as the second argument of make_app could cause attribute errors when attempting to look up settings against the ISettings object (internal). Fixed by giving the Settings objects defaults for debug_authorization and debug_notfound.
  • Return an instance of Allowed (rather than True) from has_permission when no security policy is in use.
  • Fix bug where default deny in authorization check would throw a TypeError (use ACLDenied instead of Denied).

0.4.2 (2008-11-02)

Features
  • Expose a single ILogger named “repoze.bfg.debug” as a utility; this logger is registered unconditionally and is used by the authorization debug machinery. Applications may also make use of it as necessary rather than inventing their own logger, for convenience.
  • The BFG_DEBUG_AUTHORIZATION envvar and the debug_authorization config file value now only imply debugging of view-invoked security checks. Previously, information was printed for every call to has_permission as well, which made output confusing. To debug has_permission checks and other manual permission checks, use the debugger and print statements in your own code.
  • Authorization debugging info is now only present in the HTTP response body oif debug_authorization is true.
  • The format of authorization debug messages was improved.
  • A new BFG_DEBUG_NOTFOUND envvar was added and a symmetric debug_notfound config file value was added. When either is true, and a NotFound response is returned by the BFG router (because a view could not be found), debugging information is printed to stderr. When this value is set true, the body of HTTPNotFound responses will also contain the same debugging information.
  • Allowed and Denied responses from the security machinery are now specialized into two types: ACL types, and non-ACL types. The ACL-related responses are instances of repoze.bfg.security.ACLAllowed and repoze.bfg.security.ACLDenied. The non-ACL-related responses are repoze.bfg.security.Allowed and repoze.bfg.security.Denied. The allowed-type responses continue to evaluate equal to things that themselves evaluate equal to the True boolean, while the denied-type responses continue to evaluate equal to things that themselves evaluate equal to the False boolean. The only difference between the two types is the information attached to them for debugging purposes.
  • Added a new BFG_DEBUG_ALL envvar and a symmetric debug_all config file value. When either is true, all other debug-related flags are set true unconditionally (e.g. debug_notfound and debug_authorization).
Documentation
  • Added info about debug flag changes.
  • Added a section to the security chapter named “Debugging Imperative Authorization Failures” (for e.g. has_permssion).
Bug Fixes
  • Change default paster template generator to use Paste#http server rather than PasteScript#cherrpy server. The cherrypy server has a security risk in it when REMOTE_USER is trusted by the downstream application.

0.4.1 (2008-10-28)

Bug Fixes
  • If the render_view_to_response function was called, if the view was found and called, but it returned something that did not implement IResponse, the error would pass by unflagged. This was noticed when I created a view function that essentially returned None, but received a NotFound error rather than a ValueError when the view was rendered. This was fixed.

0.4.0 (2008-10-03)

Docs
  • An “Environment and Configuration” chapter was added to the narrative portion of the documentation.
Features
  • Ensure bfg doesn’t generate warnings when running under Python 2.6.
  • The environment variable BFG_RELOAD_TEMPLATES is now available (serves the same purpose as reload_templates in the config file).
  • A new configuration file option debug_authorization was added. This turns on printing of security authorization debug statements to sys.stderr. The BFG_DEBUG_AUTHORIZATION environment variable was also added; this performs the same duty.
Bug Fixes
  • The environment variable BFG_SECURITY_DEBUG did not always work. It has been renamed to BFG_DEBUG_AUTHORIZATION and fixed.
Deprecations
  • A deprecation warning is now issued when old API names from the repoze.bfg.templates module are imported.
Backwards incompatibilities
  • The BFG_SECURITY_DEBUG environment variable was renamed to BFG_DEBUG_AUTHORIZATION.

0.3.9 (2008-08-27)

Features
  • A repoze.bfg.location API module was added.
Backwards incompatibilities

  • Applications must now use the repoze.bfg.interfaces.ILocation interface rather than zope.location.interfaces.ILocation to represent that a model object is “location-aware”. We’ve removed a dependency on zope.location for cleanliness purposes: as new versions of zope libraries are released which have improved dependency information, getting rid of our dependence on zope.location will prevent a newly installed repoze.bfg application from requiring the zope.security, egg, which not truly used at all in a “stock” repoze.bfg setup. These dependencies are still required by the stack at this time; this is purely a futureproofing move.

    The security and model documentation for previous versions of repoze.bfg recommended using the zope.location.interfaces.ILocation interface to represent that a model object is “location-aware”. This documentation has been changed to reflect that this interface should now be imported from repoze.bfg.interfaces.ILocation instead.

0.3.8 (2008-08-26)

Docs
  • Documented URL dispatch better in narrative form.
Bug fixes
  • Routes URL dispatch did not have access to the WSGI environment, so conditions such as method=GET did not work.
Features
  • Add principals_allowed_by_permission API to security module.
  • Replace z3c.pt support with support for chameleon.zpt. Chameleon is the new name for the package that used to be named z3c.pt. NOTE: If you update a repoze.bfg SVN checkout that you’re using for development, you will need to run “setup.py install” or “setup.py develop” again in order to obtain the proper Chameleon packages. z3c.pt is no longer supported by repoze.bfg. All API functions that used to render z3c.pt templates will work fine with the new packages, and your templates should render almost identically.
  • Add a repoze.bfg.chameleon_zpt module. This module provides Chameleon ZPT support.
  • Add a repoze.bfg.xslt module. This module provides XSLT support.
  • Add a repoze.bfg.chameleon_genshi module. This provides direct Genshi support, which did not exist previously.
Deprecations
  • Importing API functions directly from repoze.bfg.template is now deprecated. The get_template, render_template, render_template_to_response functions should now be imported from repoze.chameleon_zpt. The render_transform, and render_transform_to_response functions should now be imported from repoze.bfg.xslt. The repoze.bfg.template module will remain around “forever” to support backwards compatibility.

0.3.7 (2008-09-09)

Features
  • Add compatibility with z3c.pt 1.0a7+ (z3c.pt became a namespace package).
Bug fixes
  • repoze.bfg.traversal.find_model function did not function properly.

0.3.6 (2008-09-04)

Features
  • Add startup process docs.
  • Allow configuration cache to be bypassed by actions which include special “uncacheable” discriminators (for actions that have variable results).
Bug Fixes
  • Move core repoze.bfg ZCML into a repoze.bfg.includes package so we can use repoze.bfg better as a namespace package. Adjust the code generator to use it. We’ve left around the configure.zcml in the repoze.bfg package directly so as not to break older apps.
  • When a zcml application registry cache was unpickled, and it contained a reference to an object that no longer existed (such as a view), bfg would not start properly.

0.3.5 (2008-09-01)

Features
  • Event notification is issued after application is created and configured (IWSGIApplicationCreatedEvent).
  • New API module: repoze.bfg.view. This module contains the functions named render_view_to_response, render_view_to_iterable, render_view and is_response, which are documented in the API docs. These features aid programmatic (non-server-driven) view execution.

0.3.4 (2008-08-28)

Backwards incompatibilities
  • Make repoze.bfg a namespace package so we can allow folks to create subpackages (e.g. repoze.bfg.otherthing) within separate eggs. This is a backwards incompatible change which makes it impossible to import “make_app” and “get_options” from the repoze.bfg module directly. This change will break all existing apps generated by the paster code generator. Instead, you need to import these functions as repoze.bfg.router:make_app and repoze.bfg.registry:get_options, respectively. Sorry folks, it has to be done now or never, and definitely better now.
Features
  • Add model_path API function to traversal module.

Bugfixes

  • Normalize path returned by repoze.bfg.caller_path.

0.3.3 (2008-08-23)

  • Fix generated test.py module to use project name rather than package name.

0.3.2 (2008-08-23)

  • Remove sampleapp sample application from bfg package itself.
  • Remove dependency on FormEncode (only needed by sampleapp).
  • Fix paster template generation so that case-sensitivity is preserved for project vs. package name.
  • Depend on z3c.pt version 1.0a1 (which requires the [lxml] extra currently).
  • Read and write a pickled ZCML actions list, stored as configure.zcml.cache next to the applications’s “normal” configuration file. A given bfg app will usually start faster if it’s able to read the pickle data. It fails gracefully to reading the real ZCML file if it cannot read the pickle.

0.3.1 (2008-08-20)

  • Generated application differences: make_app entry point renamed to app in order to have a different name than the bfg function of the same name, to prevent confusion.
  • Add “options” processing to bfg’s make_app to support runtime options. A new API function named get_options was added to the registry module. This function is typically used in an application’s app entry point. The Paste config file section for the app can now supply the reload_templates option, which, if true, will prevent the need to restart the appserver in order for z3c.pt or XSLT template changes to be detected.
  • Use only the module name in generated project’s “test_suite” (run all tests found in the package).
  • Default port for generated apps changed from 5432 to 6543 (Postgres default port is 6543).

0.3.0 (2008-08-16)

  • Add get_template API to template module.

0.2.9 (2008-08-11)

  • 0.2.8 was “brown bag” release. It didn’t work at all. Symptom: ComponentLookupError when trying to render a page.

0.2.8 (2008-08-11)

  • Add find_model and find_root traversal APIs. In the process, make ITraverser a uni-adapter (on context) rather than a multiadapter (on context and request).

0.2.7 (2008-08-05)

  • Add a request_type attribute to the available attributes of a bfg:view configure.zcml element. This attribute will have a value which is a dotted Python path, pointing at an interface. If the request object implements this interface when the view lookup is performed, the appropriate view will be called. This is meant to allow for simple “skinning” of sites based on request type. An event subscriber should attach the interface to the request on ingress to support skins.
  • Remove “template only” views. These were just confusing and were never documented.
  • Small url dispatch overhaul: the connect method of the urldispatch.RoutesMapper object now accepts a keyword parameter named context_factory. If this parameter is supplied, it must be a callable which returns an instance. This instance is used as the context for the request when a route is matched.
  • The registration of a RoutesModelTraverser no longer needs to be performed by the application; it’s in the bfg ZCML now.

0.2.6 (2008-07-31)

  • Add event sends for INewRequest and INewResponse. See the events.rst chapter in the documentation’s api directory.

0.2.5 (2008-07-28)

  • Add model_url API.

0.2.4 (2008-07-27)

  • Added url-based dispatch.

0.2.3 (2008-07-20)

  • Add API functions for authenticated_userid and effective_principals.

0.2.2 (2008-07-20)

  • Add authenticated_userid and effective_principals API to security policy.

0.2.1 (2008-07-20)

  • Add find_interface API.

0.2 (2008-07-19)

  • Add wsgiapp decorator.
  • The concept of “view factories” was removed in favor of always calling a view, which is a callable that returns a response directly (as opposed to returning a view). As a result, the factory attribute in the bfg:view ZCML statement has been renamed to view. Various interface names were changed also.
  • render_template and render_transform no longer return a Response object. Instead, these return strings. The old behavior can be obtained by using render_template_to_response and render_transform_to_response.
  • Added ‘repoze.bfg.push:pushpage’ decorator, which creates BFG views from callables which take (context, request) and return a mapping of top-level names.
  • Added ACL-based security.
  • Support for XSLT templates via a render_transform method

0.1 (2008-07-08)

  • Initial release.

Design Documentation

Defending Pyramid’s Design

From time to time, challenges to various aspects of Pyramid design are lodged. To give context to discussions that follow, we detail some of the design decisions and trade-offs here. In some cases, we acknowledge that the framework can be made better and we describe future steps which will be taken to improve it; in some cases we just file the challenge as noted, as obviously you can’t please everyone all of the time.

Pyramid Provides More Than One Way to Do It

A canon of Python popular culture is “TIOOWTDI” (“there is only one way to do it”, a slighting, tongue-in-cheek reference to Perl’s “TIMTOWTDI”, which is an acronym for “there is more than one way to do it”).

Pyramid is, for better or worse, a “TIMTOWTDI” system. For example, it includes more than one way to resolve a URL to a view callable: via url dispatch or traversal. Multiple methods of configuration exist: imperative configuration, configuration decoration, and ZCML (optionally via pyramid_zcml). It works with multiple different kinds of persistence and templating systems. And so on. However, the existence of most of these overlapping ways to do things are not without reason and purpose: we have a number of audiences to serve, and we believe that TIMTOWTI at the web framework level actually prevents a much more insidious and harmful set of duplication at higher levels in the Python web community.

Pyramid began its life as repoze.bfg, written by a team of people with many years of prior Zope experience. The idea of traversal and the way view lookup works was stolen entirely from Zope. The authorization subsystem provided by Pyramid is a derivative of Zope’s. The idea that an application can be extended without forking is also a Zope derivative.

Implementations of these features were required to allow the Pyramid authors to build the bread-and-butter CMS-type systems for customers in the way they were accustomed to building them. No other system save Zope itself had such features. And Zope itself was beginning to show signs of its age. We were becoming hampered by consequences of its early design mistakes. Zope’s lack of documentation was also difficult to work around: it was hard to hire smart people to work on Zope applications, because there was no comprehensive documentation set to point them at which explained “it all” in one consumble place, and it was too large and self-inconsistent to document properly. Before repoze.bfg went under development, its authors obviously looked around for other frameworks that fit the bill. But no non-Zope framework did. So we embarked on building repoze.bfg.

As the result of our research, however, it became apparent that, despite the fact that no one framework had all the features we required, lots of existing frameworks had good, and sometimes very compelling ideas. In particular, URL dispatch is a more direct mechanism to map URLs to code.

So although we couldn’t find a framework save for Zope that fit our needs, and while we incorporated a lot of Zope ideas into BFG, we also emulated the features we found compelling in other frameworks (such as url dispatch). After the initial public release of BFG, as time went on, features were added to support people allergic to various Zope-isms in the system, such as the ability to configure the application using imperative configuration and configuration decoration rather than solely using ZCML, and the elimination of the required use of interface objects. It soon became clear that we had a system that was very generic, and was beginning to appeal to non-Zope users as well as ex-Zope users.

As the result of this generalization, it became obvious BFG shared 90% of its featureset with the featureset of Pylons 1, and thus had a very similar target market. Because they were so similar, choosing between the two systems was an exercise in frustration for an otherwise non-partisan developer. It was also strange for the Pylons and BFG development communities to be in competition for the same set of users, given how similar the two frameworks were. So the Pylons and BFG teams began to work together to form a plan to merge. The features missing from BFG (notably view handler classes, flash messaging, and other minor missing bits), were added, to provide familiarity to ex-Pylons users. The result is Pyramid.

The Python web framework space is currently notoriously balkanized. We’re truly hoping that the amalgamation of components in Pyramid will appeal to at least two currently very distinct sets of users: Pylons and BFG users. By unifying the best concepts from Pylons and BFG into a single codebase and leaving the bad concepts from their ancestors behind, we’ll be able to consolidate our efforts better, share more code, and promote our efforts as a unit rather than competing pointlessly. We hope to be able to shortcut the pack mentality which results in a much larger duplication of effort, represented by competing but incredibly similar applications and libraries, each built upon a specific low level stack that is incompatible with the other. We’ll also shrink the choice of credible Python web frameworks down by at least one. We’re also hoping to attract users from other communities (such as Zope’s and TurboGears’) by providing the features they require, while allowing enough flexibility to do things in a familiar fashion. Some overlap of functionality to achieve these goals is expected and unavoidable, at least if we aim to prevent pointless duplication at higher levels. If we’ve done our job well enough, the various audiences will be able to coexist and cooperate rather than firing at each other across some imaginary web framework DMZ.

Pyramid Uses A Zope Component Architecture (“ZCA”) Registry

Pyramid uses a Zope Component Architecture (ZCA) “component registry” as its application registry under the hood. This is a point of some contention. Pyramid is of a Zope pedigree, so it was natural for its developers to use a ZCA registry at its inception. However, we understand that using a ZCA registry has issues and consequences, which we’ve attempted to address as best we can. Here’s an introspection about Pyramid use of a ZCA registry, and the trade-offs its usage involves.

Problems

The global API that may be used to access data in a ZCA component registry is not particularly pretty or intuitive, and sometimes it’s just plain obtuse. Likewise, the conceptual load on a casual source code reader of code that uses the ZCA global API is somewhat high. Consider a ZCA neophyte reading the code that performs a typical “unnamed utility” lookup using the zope.component.getUtility() global API:

1
2
3
from pyramid.interfaces import ISettings
from zope.component import getUtility
settings = getUtility(ISettings)

After this code runs, settings will be a Python dictionary. But it’s unlikely that any civilian would know that just by reading the code. There are a number of comprehension issues with the bit of code above that are obvious.

First, what’s a “utility”? Well, for the purposes of this discussion, and for the purpose of the code above, it’s just not very important. If you really want to know, you can read this. However, still, readers of such code need to understand the concept in order to parse it. This is problem number one.

Second, what’s this ISettings thing? It’s an interface. Is that important here? Not really, we’re just using it as a key for some lookup based on its identity as a marker: it represents an object that has the dictionary API, but that’s not very important in this context. That’s problem number two.

Third of all, what does the getUtility function do? It’s performing a lookup for the ISettings “utility” that should return.. well, a utility. Note how we’ve already built up a dependency on the understanding of an interface and the concept of “utility” to answer this question: a bad sign so far. Note also that the answer is circular, a really bad sign.

Fourth, where does getUtility look to get the data? Well, the “component registry” of course. What’s a component registry? Problem number four.

Fifth, assuming you buy that there’s some magical registry hanging around, where is this registry? Homina homina... “around”? That’s sort of the best answer in this context (a more specific answer would require knowledge of internals). Can there be more than one registry? Yes. So which registry does it find the registration in? Well, the “current” registry of course. In terms of Pyramid, the current registry is a thread local variable. Using an API that consults a thread local makes understanding how it works non-local.

You’ve now bought in to the fact that there’s a registry that is just hanging around. But how does the registry get populated? Why, via code that calls directives like config.add_view. In this particular case, however, the registration of ISettings is made by the framework itself under the hood: it’s not present in any user configuration. This is extremely hard to comprehend. Problem number six.

Clearly there’s some amount of cognitive load here that needs to be borne by a reader of code that extends the Pyramid framework due to its use of the ZCA, even if he or she is already an expert Python programmer and whom is an expert in the domain of web applications. This is suboptimal.

Ameliorations

First, the primary amelioration: Pyramid does not expect application developers to understand ZCA concepts or any of its APIs. If an application developer needs to understand a ZCA concept or API during the creation of a Pyramid application, we’ve failed on some axis.

Instead, the framework hides the presence of the ZCA registry behind special-purpose API functions that do use ZCA APIs. Take for example the pyramid.security.authenticated_userid function, which returns the userid present in the current request or None if no userid is present in the current request. The application developer calls it like so:

1
2
from pyramid.security import authenticated_userid
userid = authenticated_userid(request)

He now has the current user id.

Under its hood however, the implementation of authenticated_userid is this:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
def authenticated_userid(request):
    """ Return the userid of the currently authenticated user or
    ``None`` if there is no authentication policy in effect or there
    is no currently authenticated user. """

    registry = request.registry # the ZCA component registry
    policy = registry.queryUtility(IAuthenticationPolicy)
    if policy is None:
        return None
    return policy.authenticated_userid(request)

Using such wrappers, we strive to always hide the ZCA API from application developers. Application developers should just never know about the ZCA API: they should call a Python function with some object germane to the domain as an argument, and it should returns a result. A corollary that follows is that any reader of an application that has been written using Pyramid needn’t understand the ZCA API either.

Hiding the ZCA API from application developers and code readers is a form of enhancing domain specificity. No application developer wants to need to understand the small, detailed mechanics of how a web framework does its thing. People want to deal in concepts that are closer to the domain they’re working in: for example, web developers want to know about users, not utilities. Pyramid uses the ZCA as an implementation detail, not as a feature which is exposed to end users.

However, unlike application developers, framework developers, including people who want to override Pyramid functionality via preordained framework plugpoints like traversal or view lookup must understand the ZCA registry API.

Pyramid framework developers were so concerned about conceptual load issues of the ZCA registry API for framework developers that a replacement registry implementation named repoze.component was actually developed. Though this package has a registry implementation which is fully functional and well-tested, and its API is much nicer than the ZCA registry API, work on it was largely abandoned and it is not used in Pyramid. We continued to use a ZCA registry within Pyramid because it ultimately proved a better fit.

Note

We continued using ZCA registry rather than disusing it in favor of using the registry implementation in repoze.component largely because the ZCA concept of interfaces provides for use of an interface hierarchy, which is useful in a lot of scenarios (such as context type inheritance). Coming up with a marker type that was something like an interface that allowed for this functionality seemed like it was just reinventing the wheel.

Making framework developers and extenders understand the ZCA registry API is a trade-off. We (the Pyramid developers) like the features that the ZCA registry gives us, and we have long-ago borne the weight of understanding what it does and how it works. The authors of Pyramid understand the ZCA deeply and can read code that uses it as easily as any other code.

But we recognize that developers who might want to extend the framework are not as comfortable with the ZCA registry API as the original developers are with it. So, for the purposes of being kind to third-party Pyramid framework developers in, we’ve drawn some lines in the sand.

In all core code, We’ve made use of ZCA global API functions such as zope.component.getUtility and zope.component.getAdapter the exception instead of the rule. So instead of:

1
2
3
from pyramid.interfaces import IAuthenticationPolicy
from zope.component import getUtility
policy = getUtility(IAuthenticationPolicy)

Pyramid code will usually do:

1
2
3
4
from pyramid.interfaces import IAuthenticationPolicy
from pyramid.threadlocal import get_current_registry
registry = get_current_registry()
policy = registry.getUtility(IAuthenticationPolicy)

While the latter is more verbose, it also arguably makes it more obvious what’s going on. All of the Pyramid core code uses this pattern rather than the ZCA global API.

Rationale

Here are the main rationales involved in the Pyramid decision to use the ZCA registry:

  • History. A nontrivial part of the answer to this question is “history”. Much of the design of Pyramid is stolen directly from Zope. Zope uses the ZCA registry to do a number of tricks. Pyramid mimics these tricks, and, because the ZCA registry works well for that set of tricks, Pyramid uses it for the same purposes. For example, the way that Pyramid maps a request to a view callable using traversal is lifted almost entirely from Zope. The ZCA registry plays an important role in the particulars of how this request to view mapping is done.
  • Features. The ZCA component registry essentially provides what can be considered something like a superdictionary, which allows for more complex lookups than retrieving a value based on a single key. Some of this lookup capability is very useful for end users, such as being able to register a view that is only found when the context is some class of object, or when the context implements some interface.
  • Singularity. There’s only one place where “application configuration” lives in a Pyramid application: in a component registry. The component registry answers questions made to it by the framework at runtime based on the configuration of an application. Note: “an application” is not the same as “a process”, multiple independently configured copies of the same Pyramid application are capable of running in the same process space.
  • Composability. A ZCA component registry can be populated imperatively, or there’s an existing mechanism to populate a registry via the use of a configuration file (ZCML, via the optional pyramid_zcml package). We didn’t need to write a frontend from scratch to make use of configuration-file-driven registry population.
  • Pluggability. Use of the ZCA registry allows for framework extensibility via a well-defined and widely understood plugin architecture. As long as framework developers and extenders understand the ZCA registry, it’s possible to extend Pyramid almost arbitrarily. For example, it’s relatively easy to build a directive that registers several views all at once, allowing app developers to use that directive as a “macro” in code that they write. This is somewhat of a differentiating feature from other (non-Zope) frameworks.
  • Testability. Judicious use of the ZCA registry in framework code makes testing that code slightly easier. Instead of using monkeypatching or other facilities to register mock objects for testing, we inject dependencies via ZCA registrations and then use lookups in the code find our mock objects.
  • Speed. The ZCA registry is very fast for a specific set of complex lookup scenarios that Pyramid uses, having been optimized through the years for just these purposes. The ZCA registry contains optional C code for this purpose which demonstrably has no (or very few) bugs.
  • Ecosystem. Many existing Zope packages can be used in Pyramid with few (or no) changes due to our use of the ZCA registry.
Conclusion

If you only develop applications using Pyramid, there’s not much to complain about here. You just should never need to understand the ZCA registry API: use documented Pyramid APIs instead. However, you may be an application developer who doesn’t read API documentation because it’s unmanly. Instead you read the raw source code, and because you haven’t read the documentation, you don’t know what functions, classes, and methods even form the Pyramid API. As a result, you’ve now written code that uses internals and you’ve painted yourself into a conceptual corner as a result of needing to wrestle with some ZCA-using implementation detail. If this is you, it’s extremely hard to have a lot of sympathy for you. You’ll either need to get familiar with how we’re using the ZCA registry or you’ll need to use only the documented APIs; that’s why we document them as APIs.

If you extend or develop Pyramid (create new directives, use some of the more obscure hooks as described in Using Hooks, or work on the Pyramid core code), you will be faced with needing to understand at least some ZCA concepts. In some places it’s used unabashedly, and will be forever. We know it’s quirky, but it’s also useful and fundamentally understandable if you take the time to do some reading about it.

Pyramid Uses Interfaces Too Liberally

In this TOPP Engineering blog entry, Ian Bicking asserts that the way repoze.bfg used a Zope interface to represent an HTTP request method added too much indirection for not enough gain. We agreed in general, and for this reason, repoze.bfg version 1.1 (and subsequent versions including Pyramid 1.0+) added view predicate and route predicate modifiers to view configuration. Predicates are request-specific (or context -specific) matching narrowers which don’t use interfaces. Instead, each predicate uses a domain-specific string as a match value.

For example, to write a view configuration which matches only requests with the POST HTTP request method, you might write a @view_config decorator which mentioned the request_method predicate:

1
2
3
4
from pyramid.view import view_config
@view_config(name='post_view', request_method='POST', renderer='json')
def post_view(request):
    return 'POSTed'

You might further narrow the matching scenario by adding an accept predicate that narrows matching to something that accepts a JSON response:

1
2
3
4
5
from pyramid.view import view_config
@view_config(name='post_view', request_method='POST',
             accept='application/json', renderer='json')
def post_view(request):
    return 'POSTed'

Such a view would only match when the request indicated that HTTP request method was POST and that the remote user agent passed application/json (or, for that matter, application/*) in its Accept request header.

Under the hood, these features make no use of interfaces.

Many prebaked predicates exist. However, use of only prebaked predicates, however, doesn’t entirely meet Ian’s criterion. He would like to be able to match a request using a lambda or another function which interrogates the request imperatively. In repoze.bfg version 1.2, we acommodate this by allowing people to define custom view predicates:

1
2
3
4
5
6
7
8
9
from pyramid.view import view_config
from pyramid.response import Response

def subpath(context, request):
    return request.subpath and request.subpath[0] == 'abc'

@view_config(custom_predicates=(subpath,))
def aview(request):
    return Response('OK')

The above view will only match when the first element of the request’s subpath is abc.

Pyramid “Encourages Use of ZCML”

ZCML is a configuration language that can be used to configure the Zope Component Architecture registry that Pyramid uses for application configuration. Often people claim that Pyramid “needs ZCML”.

It doesn’t. In Pyramid 1.0, ZCML doesn’t ship as part of the core; instead it ships in the pyramid_zcml add-on package, which is completely optional. No ZCML is required at all to use Pyramid, nor any other sort of frameworky declarative frontend to application configuration.

Pyramid Uses “Model” To Represent A Node In The Graph of Objects Traversed

The repoze.bfg documentation used to refer to the graph being traversed when traversal is used as a “model graph”. A terminology overlap confused people who wrote applications that always use ORM packages such as SQLAlchemy, which has a different notion of the definition of a “model”. As a result, in Pyramid 1.0a7, the tree of objects traversed is now renamed to resource tree and its components are now named resource objects. Associated APIs have been changed. This hopefully alleviates the terminology confusion caused by overriding the term “model”.

Pyramid Does Traversal, And I Don’t Like Traversal

In Pyramid, traversal is the act of resolving a URL path to a resource object in a resource tree. Some people are uncomfortable with this notion, and believe it is wrong. Thankfully, if you use Pyramid, and you don’t want to model your application in terms of a resource tree, you needn’t use it at all. Instead, use URL dispatch to map URL paths to views.

The idea that some folks believe traversal is unilaterally wrong is understandable. The people who believe it is wrong almost invariably have all of their data in a relational database. Relational databases aren’t naturally hierarchical, so traversing one like a tree is not possible.

However, folks who deem traversal unilaterally wrong are neglecting to take into account that many persistence mechanisms are hierarchical. Examples include a filesystem, an LDAP database, a ZODB (or another type of graph) database, an XML document, and the Python module namespace. It is often convenient to model the frontend to a hierarchical data store as a graph, using traversal to apply views to objects that either are the resources in the tree being traversed (such as in the case of ZODB) or at least ones which stand in for them (such as in the case of wrappers for files from the filesystem).

Also, many website structures are naturally hierarchical, even if the data which drives them isn’t. For example, newspaper websites are often extremely hierarchical: sections within sections within sections, ad infinitum. If you want your URLs to indicate this structure, and the structure is indefinite (the number of nested sections can be “N” instead of some fixed number), a resource tree is an excellent way to model this, even if the backend is a relational database. In this situation, the resource tree a just a site structure.

Traversal also offers better composability of applications than URL dispatch, because it doesn’t rely on a fixed ordering of URL matching. You can compose a set of disparate functionality (and add to it later) around a mapping of view to resource more predictably than trying to get the right ordering of URL pattern matching.

But the point is ultimately moot. If you don’t want to use traversal, you needn’t. Use URL dispatch instead.

Pyramid Does URL Dispatch, And I Don’t Like URL Dispatch

In Pyramid, url dispatch is the act of resolving a URL path to a view callable by performing pattern matching against some set of ordered route definitions. The route definitions are examined in order: the first pattern which matches is used to associate the URL with a view callable.

Some people are uncomfortable with this notion, and believe it is wrong. These are usually people who are steeped deeply in Zope. Zope does not provide any mechanism except traversal to map code to URLs. This is mainly because Zope effectively requires use of ZODB, which is a hierarchical object store. Zope also supports relational databases, but typically the code that calls into the database lives somewhere in the ZODB object graph (or at least is a view related to a node in the object graph), and traversal is required to reach this code.

I’ll argue that URL dispatch is ultimately useful, even if you want to use traversal as well. You can actually combine URL dispatch and traversal in Pyramid (see Combining Traversal and URL Dispatch). One example of such a usage: if you want to emulate something like Zope 2’s “Zope Management Interface” UI on top of your object graph (or any administrative interface), you can register a route like config.add_route('manage', '/manage/*traverse') and then associate “management” views in your code by using the route_name argument to a view configuration, e.g. config.add_view('.some.callable', context=".some.Resource", route_name='manage'). If you wire things up this way someone then walks up to for example, /manage/ob1/ob2, they might be presented with a management interface, but walking up to /ob1/ob2 would present them with the default object view. There are other tricks you can pull in these hybrid configurations if you’re clever (and maybe masochistic) too.

Also, if you are a URL dispatch hater, if you should ever be asked to write an application that must use some legacy relational database structure, you might find that using URL dispatch comes in handy for one-off associations between views and URL paths. Sometimes it’s just pointless to add a node to the object graph that effectively represents the entry point for some bit of code. You can just use a route and be done with it. If a route matches, a view associated with the route will be called; if no route matches, Pyramid falls back to using traversal.

But the point is ultimately moot. If you use Pyramid, and you really don’t want to use URL dispatch, you needn’t use it at all. Instead, use traversal exclusively to map URL paths to views, just like you do in Zope.

Pyramid Views Do Not Accept Arbitrary Keyword Arguments

Many web frameworks (Zope, TurboGears, Pylons 1.X, Django) allow for their variant of a view callable to accept arbitrary keyword or positional arguments, which are filled in using values present in the request.POST or request.GET dictionaries or by values present in the route match dictionary. For example, a Django view will accept positional arguments which match information in an associated “urlconf” such as r'^polls/(?P<poll_id>\d+)/$:

1
2
def aview(request, poll_id):
    return HttpResponse(poll_id)

Zope, likewise allows you to add arbitrary keyword and positional arguments to any method of a resource object found via traversal:

1
2
3
4
5
from persistent import Persistent

class MyZopeObject(Persistent):
     def aview(self, a, b, c=None):
         return '%s %s %c' % (a, b, c)

When this method is called as the result of being the published callable, the Zope request object’s GET and POST namespaces are searched for keys which match the names of the positional and keyword arguments in the request, and the method is called (if possible) with its argument list filled with values mentioned therein. TurboGears and Pylons 1.X operate similarly.

Out of the box, Pyramid is configured to have none of these features. By default, pyramid view callables always accept only request and no other arguments. The rationale: this argument specification matching done aggressively can be costly, and Pyramid has performance as one of its main goals, so we’ve decided to make people, by default, obtain information by interrogating the request object within the view callable body instead of providing magic to do unpacking into the view argument list.

However, as of Pyramid 1.0a9, user code can influence the way view callables are expected to be called, making it possible to compose a system out of view callables which are called with arbitrary arguments. See Using a View Mapper.

Pyramid Provides Too Few “Rails”

By design, Pyramid is not a particularly opinionated web framework. It has a relatively parsimonious feature set. It contains no built in ORM nor any particular database bindings. It contains no form generation framework. It has no administrative web user interface. It has no built in text indexing. It does not dictate how you arrange your code.

Such opinionated functionality exists in applications and frameworks built on top of Pyramid. It’s intended that higher-level systems emerge built using Pyramid as a base. See also Pyramid Applications are Extensible; I Don’t Believe In Application Extensibility.

Pyramid Provides Too Many “Rails”

Pyramid provides some features that other web frameworks do not. These are features meant for use cases that might not make sense to you if you’re building a simple bespoke web application:

These features are important to the authors of Pyramid. The Pyramid authors are often commissioned to build CMS-style applications. Such applications are often frameworky because they have more than one deployment. Each deployment requires a slightly different composition of sub-applications, and the framework and sub-applications often need to be extensible. Because the application has more than one deployment, pluggability and extensibility is important, as maintaining multiple forks of the application, one per deployment, is extremely undesirable. Because it’s easier to extend a system that uses traversal from the outside than it is to do the same in a system that uses URL dispatch, each deployment uses a resource tree composed of a persistent tree of domain model objects, and uses traversal to map view callable code to resources in the tree. The resource tree contains very granular security declarations, as resources are owned and accessible by different sets of users. Interfaces are used to make unit testing and implementation substitutability easier.

In a bespoke web application, usually there’s a single canonical deployment, and therefore no possibility of multiple code forks. Extensibility is not required; the code is just changed in-place. Security requirements are often less granular. Using the features listed above will often be overkill for such an application.

If you don’t like these features, it doesn’t mean you can’t or shouldn’t use Pyramid. They are all optional, and a lot of time has been spent making sure you don’t need to know about them up-front. You can build “Pylons-1.X-style” applications using Pyramid that are purely bespoke by ignoring the features above. You may find these features handy later after building a bespoke web application that suddenly becomes popular and requires extensibility because it must be deployed in multiple locations.

Pyramid Is Too Big

“The Pyramid compressed tarball is almost 2MB. It must be enormous!”

No. We just ship it with test code and helper templates. Here’s a breakdown of what’s included in subdirectories of the package tree:

docs/

3.0MB

pyramid/tests/

1.1MB

pyramid/paster_templates/

804KB

pyramid/ (except for pyramd/tests and pyramid/paster_templates)

539K

The actual Pyramid runtime code is about 10% of the total size of the tarball omitting docs, helper templates used for package generation, and test code. Of the approximately 19K lines of Python code in the package, the code that actually has a chance of executing during normal operation, excluding tests and paster template Python files, accounts for approximately 5K lines of Python code. This is comparable to Pylons 1.X, which ships with a little over 2K lines of Python code, excluding tests.

Pyramid Has Too Many Dependencies

This is true. At the time of this writing, the total number of Python package distributions that Pyramid depends upon transitively is 15 if you use Python 2.7, or 17 if you use Python 2.5 or 2.6. This is a lot more than zero package distribution dependencies: a metric which various Python microframeworks and Django boast.

The zope.component, package on which Pyramid depends has transitive dependencies on several other packages (zope.event, and zope.interface). Pyramid also has its own direct dependencies, such as Paste, Chameleon, Mako WebOb, zope.deprecation and some of these in turn have their own transitive dependencies.

We try not to reinvent too many wheels (at least the ones that don’t need reinventing), and this comes at the cost of some number of dependencies. However, “number of package distributions” is just not a terribly great metric to measure complexity. For example, the zope.event distribution on which Pyramid depends has a grand total of four lines of runtime code.

In the meantime, Pyramid has a number of package distribution dependencies comparable to similarly-targeted frameworks such as Pylons 1.X. It may be in the future that we shed more dependencies as the result of a port to Python 3 (the less code we need to port, the better). In the future, we may also move templating system dependencies out of the core and place them in add-on packages, to be included by developers instead of by the framework. This would reduce the number of core dependencies by about five.

Pyramid “Cheats” To Obtain Speed

Complaints have been lodged by other web framework authors at various times that Pyramid “cheats” to gain performance. One claimed cheating mechanism is our use (transitively) of the C extensions provided by zope.interface to do fast lookups. Another claimed cheating mechanism is the religious avoidance of extraneous function calls.

If there’s such a thing as cheating to get better performance, we want to cheat as much as possible. We optimize Pyramid aggressively. This comes at a cost: the core code has sections that could be expressed more readably. As an amelioration, we’ve commented these sections liberally.

Pyramid Gets Its Terminology Wrong (“MVC”)

“I’m a MVC web framework user, and I’m confused. Pyramid calls the controller a view! And it doesn’t have any controllers.”

If you are in this camp, you might have come to expect things about how your existing “MVC” framework uses its terminology. For example, you probably expect that models are ORM models, controllers are classes that have methods that map to URLs, and views are templates. Pyramid indeed has each of these concepts, and each probably works almost exactly like your existing “MVC” web framework. We just don’t use the MVC terminology, as we can’t square its usage in the web framework space with historical reality.

People very much want to give web applications the same properties as common desktop GUI platforms by using similar terminology, and to provide some frame of reference for how various components in the common web framework might hang together. But in the opinion of the author, “MVC” doesn’t match the web very well in general. Quoting from the Model-View-Controller Wikipedia entry:

Though MVC comes in different flavors, control flow is generally as
follows:

  The user interacts with the user interface in some way (for
  example, presses a mouse button).

  The controller handles the input event from the user interface,
  often via a registered handler or callback and converts the event
  into appropriate user action, understandable for the model.

  The controller notifies the model of the user action, possibly
  resulting in a change in the model's state. (For example, the
  controller updates the user's shopping cart.)[5]

  A view queries the model in order to generate an appropriate
  user interface (for example, the view lists the shopping cart's
  contents). Note that the view gets its own data from the model.

  The controller may (in some implementations) issue a general
  instruction to the view to render itself. In others, the view is
  automatically notified by the model of changes in state
  (Observer) which require a screen update.

  The user interface waits for further user interactions, which
  restarts the cycle.

To the author, it seems as if someone edited this Wikipedia definition, tortuously couching concepts in the most generic terms possible in order to account for the use of the term “MVC” by current web frameworks. I doubt such a broad definition would ever be agreed to by the original authors of the MVC pattern. But even so, it seems most MVC web frameworks fail to meet even this falsely generic definition.

For example, do your templates (views) always query models directly as is claimed in “note that the view gets its own data from the model”? Probably not. My “controllers” tend to do this, massaging the data for easier use by the “view” (template). What do you do when your “controller” returns JSON? Do your controllers use a template to generate JSON? If not, what’s the “view” then? Most MVC-style GUI web frameworks have some sort of event system hooked up that lets the view detect when the model changes. The web just has no such facility in its current form: it’s effectively pull-only.

So, in the interest of not mistaking desire with reality, and instead of trying to jam the square peg that is the web into the round hole of “MVC”, we just punt and say there are two things: resources and views. The resource tree represents a site structure, the view presents a resource. The templates are really just an implementation detail of any given view: a view doesn’t need a template to return a response. There’s no “controller”: it just doesn’t exist. The “model” is either represented by the resource tree or by a “domain model” (like a SQLAlchemy model) that is separate from the framework entirely. This seems to us like more reasonable terminology, given the current constraints of the web.

Pyramid Applications are Extensible; I Don’t Believe In Application Extensibility

Any Pyramid application written obeying certain constraints is extensible. This feature is discussed in the Pyramid documentation chapters named Extending An Existing Pyramid Application and Advanced Configuration. It is made possible by the use of the Zope Component Architecture and within Pyramid.

“Extensible”, in this context, means:

  • The behavior of an application can be overridden or extended in a particular deployment of the application without requiring that the deployer modify the source of the original application.
  • The original developer is not required to anticipate any extensibility plugpoints at application creation time to allow fundamental application behavior to be overriden or extended.
  • The original developer may optionally choose to anticipate an application-specific set of plugpoints, which may be hooked by a deployer. If he chooses to use the facilities provided by the ZCA, the original developer does not need to think terribly hard about the mechanics of introducing such a plugpoint.

Many developers seem to believe that creating extensible applications is not worth it. They instead suggest that modifying the source of a given application for each deployment to override behavior is more reasonable. Much discussion about version control branching and merging typically ensues.

It’s clear that making every application extensible isn’t required. The majority of web applications only have a single deployment, and thus needn’t be extensible at all. However, some web applications have multiple deployments, and some have many deployments. For example, a generic content management system (CMS) may have basic functionality that needs to be extended for a particular deployment. That CMS system may be deployed for many organizations at many places. Some number of deployments of this CMS may be deployed centrally by a third party and managed as a group. It’s useful to be able to extend such a system for each deployment via preordained plugpoints than it is to continually keep each software branch of the system in sync with some upstream source: the upstream developers may change code in such a way that your changes to the same codebase conflict with theirs in fiddly, trivial ways. Merging such changes repeatedly over the lifetime of a deployment can be difficult and time consuming, and it’s often useful to be able to modify an application for a particular deployment in a less invasive way.

If you don’t want to think about Pyramid application extensibility at all, you needn’t. You can ignore extensibility entirely. However, if you follow the set of rules defined in Extending An Existing Pyramid Application, you don’t need to make your application extensible: any application you write in the framework just is automatically extensible at a basic level. The mechanisms that deployers use to extend it will be necessarily coarse: typically, views, routes, and resources will be capable of being overridden. But for most minor (and even some major) customizations, these are often the only override plugpoints necessary: if the application doesn’t do exactly what the deployment requires, it’s often possible for a deployer to override a view, route, or resource and quickly make it do what he or she wants it to do in ways not necessarily anticipated by the original developer. Here are some example scenarios demonstrating the benefits of such a feature.

  • If a deployment needs a different styling, the deployer may override the main template and the CSS in a separate Python package which defines overrides.
  • If a deployment needs an application page to do something differently needs it to expose more or different information, the deployer may override the view that renders the page within a separate Python package.
  • If a deployment needs an additional feature, the deployer may add a view to the override package.

As long as the fundamental design of the upstream package doesn’t change, these types of modifications often survive across many releases of the upstream package without needing to be revisited.

Extending an application externally is not a panacea, and carries a set of risks similar to branching and merging: sometimes major changes upstream will cause you to need to revisit and update some of your modifications. But you won’t regularly need to deal wth meaningless textual merge conflicts that trivial changes to upstream packages often entail when it comes time to update the upstream package, because if you extend an application externally, there just is no textual merge done. Your modifications will also, for whatever its worth, be contained in one, canonical, well-defined place.

Branching an application and continually merging in order to get new features and bugfixes is clearly useful. You can do that with a Pyramid application just as usefully as you can do it with any application. But deployment of an application written in Pyramid makes it possible to avoid the need for this even if the application doesn’t define any plugpoints ahead of time. It’s possible that promoters of competing web frameworks dismiss this feature in favor of branching and merging because applications written in their framework of choice aren’t extensible out of the box in a comparably fundamental way.

While Pyramid application are fundamentally extensible even if you don’t write them with specific extensibility in mind, if you’re moderately adventurous, you can also take it a step further. If you learn more about the Zope Component Architecture, you can optionally use it to expose other more domain-specific configuration plugpoints while developing an application. The plugpoints you expose needn’t be as coarse as the ones provided automatically by Pyramid itself. For example, you might compose your own directive that configures a set of views for a prebaked purpose (e.g. restview or somesuch) , allowing other people to refer to that directive when they make declarations in the includeme of their customization package. There is a cost for this: the developer of an application that defines custom plugpoints for its deployers will need to understand the ZCA or he will need to develop his own similar extensibility system.

Ultimately, any argument about whether the extensibility features lent to applications by Pyramid are good or bad is mostly pointless. You needn’t take advantage of the extensibility features provided by a particular Pyramid application in order to affect a modification for a particular set of its deployments. You can ignore the application’s extensibility plugpoints entirely, and instead use version control branching and merging to manage application deployment modifications instead, as if you were deploying an application written using any other web framework.

Zope 3 Enforces “TTW” Authorization Checks By Default; Pyramid Does Not

Challenge

Pyramid performs automatic authorization checks only at view execution time. Zope 3 wraps context objects with a security proxy, which causes Zope 3 to do also security checks during attribute access. I like this, because it means:

  1. When I use the security proxy machinery, I can have a view that conditionally displays certain HTML elements (like form fields) or prevents certain attributes from being modified depending on the the permissions that the accessing user possesses with respect to a context object.
  2. I want to also expose my resources via a REST API using Twisted Web. If Pyramid performed authorization based on attribute access via Zope3’s security proxies, I could enforce my authorization policy in both Pyramid and in the Twisted-based system the same way.
Defense

Pyramid was developed by folks familiar with Zope 2, which has a “through the web” security model. This TTW security model was the precursor to Zope 3’s security proxies. Over time, as the Pyramid developers (working in Zope 2) created such sites, we found authorization checks during code interpretation extremely useful in a minority of projects. But much of the time, TTW authorization checks usually slowed down the development velocity of projects that had no delegation requirements. In particular, if we weren’t allowing untrusted users to write arbitrary Python code to be executed by our application, the burden of through the web security checks proved too costly to justify. We (collectively) haven’t written an application on top of which untrusted developers are allowed to write code in many years, so it seemed to make sense to drop this model by default in a new web framework.

And since we tend to use the same toolkit for all web applications, it’s just never been a concern to be able to use the same set of restricted-execution code under two web different frameworks.

Justifications for disabling security proxies by default notwithstanding, given that Zope 3 security proxies are viral by nature, the only requirement to use one is to make sure you wrap a single object in a security proxy and make sure to access that object normally when you want proxy security checks to happen. It is possible to override the Pyramid traverser for a given application (see Changing the Traverser). To get Zope3-like behavior, it is possible to plug in a different traverser which returns Zope3-security-proxy-wrapped objects for each traversed object (including the context and the root). This would have the effect of creating a more Zope3-like environment without much effort.

Pyramid Uses its Own HTTP Exception Class Hierarchy Rather Than webob.exc

Note

This defense is new as of Pyramid 1.1.

The HTTP exception classes defined in pyramid.httpexceptions are very much like the ones defined in webob.exc (e.g. HTTPNotFound, HTTPForbidden, etc). They have the same names and largely the same behavior and all have a very similar implementation, but not the same identity. Here’s why they have a separate identity:

  • Making them separate allows the HTTP exception classes to subclass pyramid.response.Response. This speeds up response generation slightly due to the way the Pyramid router works. The same speedup could be gained by monkeypatching webob.response.Response but it’s usually the case that monkeypatching turns out to be evil and wrong.
  • Making them separate allows them to provide alternate __call__ logic which also speeds up response generation.
  • Making them separate allows the exception classes to provide for the proper value of RequestClass (pyramid.request.Request).
  • Making them separate allows us freedom from having to think about backwards compatibility code present in webob.exc having to do with Python 2.4, which we no longer support in Pyramid 1.1+.
  • We change the behavior of two classes (HTTPNotFound and HTTPForbidden) in the module so that they can be used by Pyramid internally for notfound and forbidden exceptions.
  • Making them separate allows us to influence the docstrings of the exception classes to provide Pyramid-specific documentation.
  • Making them separate allows us to silence a stupid deprecation warning under Python 2.6 when the response objects are used as exceptions (related to self.message).

Pyramid has Simpler Traversal Machinery than Does Zope

Zope’s default traverser:

  • Allows developers to mutate the traversal name stack while traversing (they can add and remove path elements).
  • Attempts to use an adaptation to obtain the next element in the path from the currently traversed object, falling back to __bobo_traverse__, __getitem__ and eventually __getattr__.

Zope’s default traverser allows developers to mutate the traversal name stack during traversal by mutating REQUEST['TraversalNameStack']. Pyramid’s default traverser (pyramid.traversal.ResourceTreeTraverser) does not offer a way to do this; it does not maintain a stack as a request attribute and, even if it did, it does not pass the request to resource objects while it’s traversing. While it was handy at times, this feature was abused in frameworks built atop Zope (like CMF and Plone), often making it difficult to tell exactly what was happening when a traversal didn’t match a view. I felt it was better to make folks that wanted the feature replace the traverser rather than build that particular honey pot in to the default traverser.

Zope uses multiple mechanisms to attempt to obtain the next element in the resource tree based on a name. It first tries an adaptation of the current resource to ITraversable, and if that fails, it falls back to attempting number of magic methods on the resource (__bobo_traverse__, __getitem__, and __getattr__). My experience while both using Zope and attempting to reimplement its publisher in repoze.zope2 led me to believe the following:

  • The default traverser should be as simple as possible. Zope’s publisher is somewhat difficult to follow and replicate due to the fallbacks it tried when one traversal method failed. It is also slow.
  • The entire traverser should be replaceable, not just elements of the traversal machinery. Pyramid has a few big components rather than a plethora of small ones. If the entire traverser is replaceable, it’s an antipattern to make portions of the default traverser replaceable. Doing so is a “knobs on knobs” pattern, which is unfortunately somewhat endemic in Zope. In a “knobs on knobs” pattern, a replaceable subcomponent of a larger component is made configurable using the same configuration mechanism that can be used to replace the larger component. For example, in Zope, you can replace the default traverser by registering an adapter. But you can also (or alternately) control how the default traverser traverses by registering one or more adapters. As a result of being able to either replace the larger component entirely or turn knobs on the default implementation of the larger component, no one understands when (or whether) they should ever override the larger component entrirely. This results, over time, in a rusting together of the larger “replaceable” component and the framework itself, because people come to depend on the availability of the default component in order just to turn its knobs. The default component effectively becomes part of the framework, which entirely subverts the goal of making it replaceable. In Pyramid, typically if a component is replaceable, it will itself have no knobs (it will be solid state). If you want to influence behavior controlled by that component, you will replace the component instead of turning knobs attached to the component.

Microframeworks Have Smaller Hello World Programs

Self-described “microframeworks” exist: Bottle and Flask are two that are becoming popular. Bobo doesn’t describe itself as a microframework, but its intended userbase is much the same. Many others exist. We’ve actually even (only as a teaching tool, not as any sort of official project) created one using Pyramid (the videos use BFG, a precursor to Pyramid, but the resulting code is available for Pyramid too). Microframeworks are small frameworks with one common feature: each allows its users to create a fully functional application that lives in a single Python file.

Some developers and microframework authors point out that Pyramid’s “hello world” single-file program is longer (by about five lines) than the equivalent program in their favorite microframework. Guilty as charged.

This loss isn’t for lack of trying. Pyramid is useful in the same circumstance in which microframeworks claim dominance: single-file applications. But Pyramid doesn’t sacrifice its ability to credibly support larger applications in order to achieve hello-world LoC parity with the current crop of microframeworks. Pyramid’s design instead tries to avoid some common pitfalls associated with naive declarative configuration schemes. The subsections which follow explain the rationale.

Application Programmers Don’t Control The Module-Scope Codepath (Import-Time Side-Effects Are Evil)

Please imagine a directory structure with a set of Python files in it:

.
|-- app.py
|-- app2.py
`-- config.py

The contents of app.py:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
from config import decorator
from config import L
import pprint

@decorator
def foo():
    pass

if __name__ == '__main__':
    import app2
    pprint.pprint(L)

The contents of app2.py:

1
2
3
4
5
import app

@app.decorator
def bar():
    pass

The contents of config.py:

1
2
3
4
5
  L = []

  def decorator(func):
      L.append(func)
      return func

If we cd to the directory that holds these files and we run python app.py given the directory structure and code above, what happens? Presumably, our decorator decorator will be used twice, once by the decorated function foo in app.py and once by the decorated function bar in app2.py. Since each time the decorator is used, the list L in config.py is appended to, we’d expect a list with two elements to be printed, right? Sadly, no:

[chrism@thinko]$ python app.py
[<function foo at 0x7f4ea41ab1b8>,
 <function foo at 0x7f4ea41ab230>,
 <function bar at 0x7f4ea41ab2a8>]

By visual inspection, that outcome (three different functions in the list) seems impossible. We only defined two functions and we decorated each of those functions only once, so we believe that the decorator decorator will only run twice. However, what we believe is wrong because the code at module scope in our app.py module was executed twice. The code is executed once when the script is run as __main__ (via python app.py), and then it is executed again when app2.py imports the same file as app.

What does this have to do with our comparison to microframeworks? Many microframeworks in the current crop (e.g. Bottle, Flask) encourage you to attach configuration decorators to objects defined at module scope. These decorators execute arbitrarily complex registration code which populates a singleton registry that is a global defined in external Python module. This is analogous to the above example: the “global registry” in the above example is the list L.

Let’s see what happens when we use the same pattern with the Groundhog microframework. Replace the contents of app.py above with this:

1
2
3
4
5
6
7
8
9
from config import gh

@gh.route('/foo/')
def foo():
    return 'foo'

if __name__ == '__main__':
    import app2
    pprint.pprint(L)

Replace the contents of app2.py above with this:

1
2
3
4
5
import app

@app.gh.route('/bar/')
def bar():
    'return bar'

Replace the contents of config.py above with this:

1
2
from groundhog import Groundhog
gh = Groundhog('myapp', 'seekrit')

How many routes will be registered within the routing table of the “gh” Groundhog application? If you answered three, you are correct. How many would a casual reader (and any sane developer) expect to be registered? If you answered two, you are correct. Will the double registration be a problem? With our Groundhog framework’s route method backing this application, not really. It will slow the application down a little bit, because it will need to miss twice for a route when it does not match. Will it be a problem with another framework, another application, or another decorator? Who knows. You need to understand the application in its totality, the framework in its totality, and the chronology of execution to be able to predict what the impact of unintentional code double-execution will be.

The encouragement to use decorators which perform population of an external registry has an unintended consequence: the application developer now must assert ownership of every codepath that executes Python module scope code. Module-scope code is presumed by the current crop of decorator-based microframeworks to execute once and only once; if it executes more than once, weird things will start to happen. It is up to the application developer to maintain this invariant. Unfortunately, however, in reality, this is an impossible task, because, Python programmers do not own the module scope codepath, and never will. Anyone who tries to sell you on the idea that they do is simply mistaken. Test runners that you may want to use to run your code’s tests often perform imports of arbitrary code in strange orders that manifest bugs like the one demonstrated above. API documentation generation tools do the same. Some people even think it’s safe to use the Python reload command or delete objects from sys.modules, each of which has hilarious effects when used against code that has import-time side effects.

Global-registry-mutating microframework programmers therefore will at some point need to start reading the tea leaves about what might happen if module scope code gets executed more than once like we do in the previous paragraph. When Python programmers assume they can use the module-scope codepath to run arbitrary code (especially code which populates an external registry), and this assumption is challenged by reality, the application developer is often required to undergo a painful, meticulous debugging process to find the root cause of an inevitably obscure symptom. The solution is often to rearrange application import ordering or move an import statement from module-scope into a function body. The rationale for doing so can never be expressed adequately in the checkin message which accompanies the fix and can’t be documented succinctly enough for the benefit of the rest of the development team so that the problem never happens again. It will happen again, especially if you are working on a project with other people who haven’t yet internalized the lessons you learned while you stepped through module-scope code using pdb. This is a really pretty poor situation to find yourself in as an application developer: you probably didn’t even know your or your team signed up for the job, because the documentation offered by decorator-based microframeworks don’t warn you about it.

Folks who have a large investment in eager decorator-based configuration that populates an external data structure (such as microframework authors) may argue that the set of circumstances I outlined above is anomalous and contrived. They will argue that it just will never happen. If you never intend your application to grow beyond one or two or three modules, that’s probably true. However, as your codebase grows, and becomes spread across a greater number of modules, the circumstances in which module-scope code will be executed multiple times will become more and more likely to occur and less and less predictable. It’s not responsible to claim that double-execution of module-scope code will never happen. It will; it’s just a matter of luck, time, and application complexity.

If microframework authors do admit that the circumstance isn’t contrived, they might then argue that real damage will never happen as the result of the double-execution (or triple-execution, etc) of module scope code. You would be wise to disbelieve this assertion. The potential outcomes of multiple execution are too numerous to predict because they involve delicate relationships between application and framework code as well as chronology of code execution. It’s literally impossible for a framework author to know what will happen in all circumstances. But even if given the gift of omniscience for some limited set of circumstances, the framework author almost certainly does not have the double-execution anomaly in mind when coding new features. He’s thinking of adding a feature, not protecting against problems that might be caused by the 1% multiple execution case. However, any 1% case may cause 50% of your pain on a project, so it’d be nice if it never occured.

Responsible microframeworks actually offer a back-door way around the problem. They allow you to disuse decorator based configuration entirely. Instead of requiring you to do the following:

1
2
3
4
5
6
7
8
gh = Groundhog('myapp', 'seekrit')

@gh.route('/foo/')
def foo():
    return 'foo'

if __name__ == '__main__':
    gh.run()

They allow you to disuse the decorator syntax and go almost-all-imperative:

1
2
3
4
5
6
7
8
def foo():
    return 'foo'

gh = Groundhog('myapp', 'seekrit')

if __name__ == '__main__':
    gh.add_route(foo, '/foo/')
    gh.run()

This is a generic mode of operation that is encouraged in the Pyramid documentation. Some existing microframeworks (Flask, in particular) allow for it as well. None (other than Pyramid) encourage it. If you never expect your application to grow beyond two or three or four or ten modules, it probably doesn’t matter very much which mode you use. If your application grows large, however, imperative configuration can provide better predictability.

Note

Astute readers may notice that Pyramid has configuration decorators too. Aha! Don’t these decorators have the same problems? No. These decorators do not populate an external Python module when they are executed. They only mutate the functions (and classes and methods) they’re attached to. These mutations must later be found during a scan process that has a predictable and structured import phase. Module-localized mutation is actually the best-case circumstance for double-imports; if a module only mutates itself and its contents at import time, if it is imported twice, that’s OK, because each decorator invocation will always be mutating an independent copy of the object its attached to, not a shared resource like a registry in another module. This has the effect that double-registrations will never be performed.

Routes Need Relative Ordering

Consider the following simple Groundhog application:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
from groundhog import Groundhog
app = Groundhog('myapp', 'seekrit')

app.route('/admin')
def admin():
    return '<html>admin page</html>'

app.route('/:action')
def action():
    if action == 'add':
       return '<html>add</html>'
    if action == 'delete':
       return '<html>delete</html>'
    return app.abort(404)

if __name__ == '__main__':
    app.run()

If you run this application and visit the URL /admin, you will see the “admin” page. This is the intended result. However, what if you rearrange the order of the function definitions in the file?

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
from groundhog import Groundhog
app = Groundhog('myapp', 'seekrit')

app.route('/:action')
def action():
    if action == 'add':
       return '<html>add</html>'
    if action == 'delete':
       return '<html>delete</html>'
    return app.abort(404)

app.route('/admin')
def admin():
    return '<html>admin page</html>'

if __name__ == '__main__':
    app.run()

If you run this application and visit the URL /admin, you will now be returned a 404 error. This is probably not what you intended. The reason you see a 404 error when you rearrange function definition ordering is that routing declarations expressed via our microframework’s routing decorators have an ordering, and that ordering matters.

In the first case, where we achieved the expected result, we first added a route with the pattern /admin, then we added a route with the pattern /:action by virtue of adding routing patterns via decorators at module scope. When a request with a PATH_INFO of /admin enters our application, the web framework loops over each of our application’s route patterns in the order in which they were defined in our module. As a result, the view associated with the /admin routing pattern will be invoked: it matches first. All is right with the world.

In the second case, where we did not achieve the expected result, we first added a route with the pattern /:action, then we added a route with the pattern /admin. When a request with a PATH_INFO of /admin enters our application, the web framework loops over each of our application’s route patterns in the order in which they were defined in our module. As a result, the view associated with the /:action routing pattern will be invoked: it matches first. A 404 error is raised. This is not what we wanted; it just happened due to the order in which we defined our view functions.

This is because Groundhog routes are added to the routing map in import order, and matched in the same order when a request comes in. Bottle, like Groundhog, as of this writing, matches routes in the order in which they’re defined at Python execution time. Flask, on the other hand, does not order route matching based on import order; it reorders the routes you add to your application based on their “complexity”. Other microframeworks have varying strategies to do route ordering.

Your application may be small enough where route ordering will never cause an issue. If your application becomes large enough, however, being able to specify or predict that ordering as your application grows larger will be difficult. At some point, you will likely need to more explicitly start controlling route ordering, especially in applications that require extensibility.

If your microframework orders route matching based on complexity, you’ll need to understand what is meant by “complexity”, and you’ll need to attempt to inject a “less complex” route to have it get matched before any “more complex” one to ensure that it’s tried first.

If your microframework orders its route matching based on relative import/execution of function decorator definitions, you will need to ensure you execute all of these statements in the “right” order, and you’ll need to be cognizant of this import/execution ordering as you grow your application or try to extend it. This is a difficult invariant to maintain for all but the smallest applications.

In either case, your application must import the non-__main__ modules which contain configuration decorations somehow for their configuration to be executed. Does that make you a little uncomfortable? It should, because Application Programmers Don’t Control The Module-Scope Codepath (Import-Time Side-Effects Are Evil).

Pyramid uses neither decorator import time ordering nor does it attempt to divine the relative complexity of one route to another in order to define a route match ordering. In Pyramid, you have to maintain relative route ordering imperatively via the chronology of multiple executions of the pyramid.config.Configurator.add_route() method. The order in which you repeatedly call add_route becomes the order of route matching.

If needing to maintain this imperative ordering truly bugs you, you can use traversal instead of route matching, which is a completely declarative (and completely predictable) mechanism to map code to URLs. While URL dispatch is easier to understand for small non-extensible applications, traversal is a great fit for very large applications and applications that need to be arbitrarily extensible.

“Stacked Object Proxies” Are Too Clever / Thread Locals Are A Nuisance

Some microframeworks use the import statement to get a handle to an object which is not logically global:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
from flask import request

@app.route('/login', methods=['POST', 'GET'])
def login():
    error = None
    if request.method == 'POST':
        if valid_login(request.form['username'],
                       request.form['password']):
            return log_the_user_in(request.form['username'])
        else:
            error = 'Invalid username/password'
    # this is executed if the request method was GET or the
    # credentials were invalid

The Pylons 1.X web framework uses a similar strategy. It calls these things “Stacked Object Proxies”, so, for purposes of this discussion, I’ll do so as well.

Import statements in Python (import foo, from bar import baz) are most frequently performed to obtain a reference to an object defined globally within an external Python module. However, in normal programs, they are never used to obtain a reference to an object that has a lifetime measured by the scope of the body of a function. It would be absurd to try to import, for example, a variable named i representing a loop counter defined in the body of a function. For example, we’d never try to import i from the code below:

1
2
3
def afunc():
    for i in range(10):
        print i

By its nature, the request object created as the result of a WSGI server’s call into a long-lived web framework cannot be global, because the lifetime of a single request will be much shorter than the lifetime of the process running the framework. A request object created by a web framework actually has more similarity to the i loop counter in our example above than it has to any comparable importable object defined in the Python standard library or in normal library code.

However, systems which use stacked object proxies promote locally scoped objects such as request out to module scope, for the purpose of being able to offer users a nice spelling involving import. They, for what I consider dubious reasons, would rather present to their users the canonical way of getting at a request as from framework import request instead of a saner from myframework.threadlocals import get_request; request = get_request() even though the latter is more explicit.

It would be most explicit if the microframeworks did not use thread local variables at all. Pyramid view functions are passed a request object; many of Pyramid’s APIs require that an explicit request object be passed to them. It is possible to retrieve the current Pyramid request as a threadlocal variable but it is a “in case of emergency, break glass” type of activity. This explicitness makes Pyramid view functions more easily unit testable, as you don’t need to rely on the framework to manufacture suitable “dummy” request (and other similarly-scoped) objects during test setup. It also makes them more likely to work on arbitrary systems, such as async servers that do no monkeypatching.

Explicitly WSGI

Some microframeworks offer a run() method of an application object that executes a default server configuration for easy execution.

Pyramid doesn’t currently try to hide the fact that its router is a WSGI application behind a convenience run() API. It just tells people to import a WSGI server and use it to serve up their Pyramid application as per the documentation of that WSGI server.

The extra lines saved by abstracting away the serving step behind run() seem to have driven dubious second-order decisions related to API in some microframeworks. For example, Bottle contains a ServerAdapter subclass for each type of WSGI server it supports via its app.run() mechanism. This means that there exists code in bottle.py that depends on the following modules: wsgiref, flup, paste, cherrypy, fapws, tornado, google.appengine, twisted.web, diesel, gevent, gunicorn, eventlet, and rocket. You choose the kind of server you want to run by passing its name into the run method. In theory, this sounds great: I can try Bottle out on gunicorn just by passing in a name! However, to fully test Bottle, all of these third-party systems must be installed and functional; the Bottle developers must monitor changes to each of these packages and make sure their code still interfaces properly with them. This expands the packages required for testing greatly; this is a lot of requirements. It is likely difficult to fully automate these tests due to requirements conflicts and build issues.

As a result, for single-file apps, we currently don’t bother to offer a run() shortcut; we tell folks to import their WSGI server of choice and run it by hand. For the people who want a server abstraction layer, we suggest that they use PasteDeploy. In PasteDeploy-based systems, the onus for making sure that the server can interface with a WSGI application is placed on the server developer, not the web framework developer, making it more likely to be timely and correct.

Wrapping Up

Here’s a diagrammed version of the simplest pyramid application, where comments take into account what we’ve discussed in the Microframeworks Have Smaller Hello World Programs section.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
from pyramid.response import Response      # explicit response objects, no TL
from paste.httpserver import serve         # explicitly WSGI

def hello_world(request):  # accepts a request; no request thread local reqd
    # explicit response object means no response threadlocal
    return Response('Hello world!')

if __name__ == '__main__':
    from pyramid.config import Configurator
    config = Configurator()       # no global application object.
    config.add_view(hello_world)  # explicit non-decorator registration
    app = config.make_wsgi_app()  # explicitly WSGI
    serve(app, host='0.0.0.0')    # explicitly WSGI

Pyramid Doesn’t Offer Pluggable Apps

It is “Pyramidic” to compose multiple external sources into the same configuration using include(). Any number of includes can be done to compose an application; includes can even be done from within other includes. Any directive can be used within an include that can be used outside of one (such as add_view(), etc).

Pyramid has a conflict detection system that will throw an error if two included externals try to add the same configuration in a conflicting way (such as both externals trying to add a route using the same name, or both externals trying to add a view with the same set of predicates). It’s awful tempting to call this set of features something that can be used to compose a system out of “pluggable applications”. But in reality, there are a number of problems with claiming this:

  • The terminology is strained. Pyramid really has no notion of a plurality of “applications”, just a way to compose configuration from multiple sources to create a single WSGI application. That WSGI application may gain behavior by including or disincluding configuration, but once it’s all composed together, Pyramid doesn’t really provide any machinery which can be used to demarcate the boundaries of one “application” (in the sense of configuration from an external that adds routes, views, etc) from another.
  • Pyramid doesn’t provide enough “rails” to make it possible to integrate truly honest-to-god, download-an-app-from-a-random-place and-plug-it-in-to-create-a-system “pluggable” applications. Because Pyramid itself isn’t opinionated (it doesn’t mandate a particular kind of database, it offers multiple ways to map URLs to code, etc), it’s unlikely that someone who creates something application-like will be able to casually redistribute it to J. Random Pyramid User and have it just work by asking him to config.include a function from the package. This is particularly true of very high level components such as blogs, wikis, twitter clones, commenting systems, etc. The integrator (the Pyramid developer who has downloaded a package advertised as a “pluggable app”) will almost certainly have made different choices about e.g. what type of persistence system he’s using, and for the integrator to appease the requirements of the “pluggable application”, he may be required to set up a different database, make changes to his own code to prevent his application from shadowing the pluggable app (or vice versa), and any other number of arbitrary changes.

For this reason, we claim that Pyramid has “extensible” applications, not pluggable applications. Any Pyramid application can be extended without forking it as long as its configuration statements have been composed into things that can be pulled in via config.include.

It’s also perfectly reasonable for a single developer or team to create a set of interoperating components which can be enabled or disabled by using config.include. That developer or team will be able to provide the “rails” (by way of making high-level choices about the technology used to create the project, so there won’t be any issues with plugging all of the components together. The problem only rears its head when the components need to be distributed to arbitrary users. Note that Django has a similar problem with “pluggable applications” that need to work for arbitrary third parties, even though they provide many, many more rails than does Pyramid. Even the rails they provide are not enough to make the “pluggable application” story really work without local modification.

Truly pluggable applications need to be created at a much higher level than a web framework, as no web framework can offer enough constraints to really make them work out of the box. They really need to plug into an application, instead. It would be a noble goal to build an application with Pyramid that provides these constraints and which truly does offer a way to plug in applications (Joomla, Plone, Drupal come to mind).

Pyramid Has Zope Things In It, So It’s Too Complex

On occasion, someone will feel compelled to post a mailing list message that reads something like this:

had a quick look at pyramid ... too complex to me and not really
understand for which benefits.. I feel should consider whether it's time
for me to step back to django .. I always hated zope (useless ?)
complexity and I love simple way of thinking

(Paraphrased from a real email, actually.)

Let’s take this criticism point-by point.

Too Complex

If you can understand this hello world program, you can use Pyramid:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
from paste.httpserver import serve
from pyramid.config import Configurator
from pyramid.response import Response

def hello_world(request):
    return Response('Hello world!')

if __name__ == '__main__':
    config = Configurator()
    config.add_view(hello_world)
    app = config.make_wsgi_app()
    serve(app)

Pyramid has ~ 650 pages of documentation (printed), covering topics from the very basic to the most advanced. Nothing is left undocumented, quite literally. It also has an awesome, very helpful community. Visit the #pyramid IRC channel on freenode.net (irc://freenode.net#pyramid) and see.

Hate Zope

I’m sorry you feel that way. The Zope brand has certainly taken its share of lumps over the years, and has a reputation for being insular and mysterious. But the word “Zope” is literally quite meaningless without qualification. What part of Zope do you hate? “Zope” is a brand, not a technology.

If it’s Zope2-the-web-framework, Pyramid is not that. The primary designers and developers of Pyramid, if anyone, should know. We wrote Pyramid’s predecessor (repoze.bfg), in part, because we knew that Zope 2 had usability issues and limitations. repoze.bfg (and now Pyramid) was written to address these issues.

If it’s Zope3-the-web-framework, Pyramid is definitely not that. Making use of lots of Zope 3 technologies is territory already staked out by the Grok project. Save for the obvious fact that they’re both web frameworks, Pyramid is very, very different than Grok. Grok exposes lots of Zope technologies to end users. On the other hand, if you need to understand a Zope-only concept while using Pyramid, then we’ve failed on some very basic axis.

If it’s just the word Zope: this can only be guilt by association. Because a piece of software internally uses some package named zope.foo, it doesn’t turn the piece of software that uses it into “Zope”. There is a lot of great software written that has the word Zope in its name. Zope is not some sort of monolithic thing, and a lot of its software is usable externally. And while it’s not really the job of this document to defend it, Zope has been around for over 10 years and has an incredibly large, active community. If you don’t believe this, http://taichino.appspot.com/pypi_ranking/authors is an eye-opening reality check.

Love Simplicity

Years of effort have gone into honing this package and its documentation to make it as simple as humanly possible for developers to use. Everything is a tradeoff, of course, and people have their own ideas about what “simple” is. You may have a style difference if you believe Pyramid is complex. Its developers obviously disagree.

Other Challenges

Other challenges are encouraged to be sent to the Pylons-devel maillist. We’ll try to address them by considering a design change, or at very least via exposition here.

Sample Applications

cluegun is a simple pastebin application based on Rocky Burt’s ClueBin. It demonstrates form processing, security, and the use of ZODB within a Pyramid application. Check this application out via:

git clone git://github.com/Pylons/cluegun.git

virginia is a very simple dynamic file rendering application. It is willing to render structured text documents, HTML documents, and images from a filesystem directory. An earlier version of this application runs the repoze.org website. Check this application out via:

git clone git://github.com/Pylons/virginia.git

shootout is an example “idea competition” application by Carlos de la Guardia and Lukasz Fidosz. It demonstrates URL dispatch, simple authentication, integration with SQLAlchemy and pyramid_simpleform. Check this application out of version control via:

git clone git://github.com/Pylons/shootout.git

Older Sample Applications (repoze.bfg)

Note

These applications are for an older version of Pyramid, which was named repoze.bfg. They won’t work unmodified under Pyramid, but might provide useful clues.

bfgsite is the software which runs the bfg.repoze.org website. It demonstrates integration with Trac, and includes several mini-applications such as a pastebin and tutorial engine. Check a buildout for this application out of Subversion via:

svn co http://svn.repoze.org/buildouts/bfgsite/ bfgsite_buildout

KARL is a moderately-sized application (roughly 70K lines of Python code) built on top of repoze.bfg and other Repoze software. It is an open source web system for collaboration, organizational intranets, and knowledge management, It provides facilities for wikis, calendars, manuals, searching, tagging, commenting, and file uploads. See the KARL site for download and installation details.

Support and Development

The Pylons Project web site is the main online source of Pyramid support and development information.

To report bugs, use the issue tracker.

If you’ve got questions that aren’t answered by this documentation, contact the Pylons-devel maillist or join the #pyramid IRC channel.

Browse and check out tagged and trunk versions of Pyramid via the Pyramid GitHub repository. To check out the trunk via git, use this command:

git clone git@github.com:Pylons/pyramid.git

To find out how to become a contributor to Pyramid, please see the contributor’s section of the documentation.

Index and Glossary

Glossary

ACE
An access control entry. An access control entry is one element in an ACL. An access control entry is a three-tuple that describes three things: an action (one of either Allow or Deny), a principal (a string describing a user or group), and a permission. For example the ACE, (Allow, 'bob', 'read') is a member of an ACL that indicates that the principal bob is allowed the permission read against the resource the ACL is attached to.
ACL
An access control list. An ACL is a sequence of ACE tuples. An ACL is attached to a resource instance. An example of an ACL is [ (Allow, 'bob', 'read'), (Deny, 'fred', 'write')]. If an ACL is attached to a resource instance, and that resource is findable via the context resource, it will be consulted any active security policy to determine wither a particular request can be fulfilled given the authentication information in the request.
Agendaless Consulting
A consulting organization formed by Paul Everitt, Tres Seaver, and Chris McDonough. See also http://agendaless.com .
Akhet
Akhet is a Pyramid-based development environment which provides a Pylons-esque scaffold which sports support for view handler application development, SQLAlchemy support, Mako templating by default, and other Pylons-like features. See http://docs.pylonsproject.org/projects/akhet/dev/index.html for more information.
application registry
A registry of configuration information consulted by Pyramid while servicing an application. An application registry maps resource types to views, as well as housing other application-specific component registrations. Every Pyramid application has one (and only one) application registry.
asset
Any file contained within a Python package which is not a Python source code file.
asset specification
A colon-delimited identifier for an asset. The colon separates a Python package name from a package subpath. For example, the asset specification my.package:static/baz.css identifies the file named baz.css in the static subdirectory of the my.package Python package. See Understanding Asset Specifications for more info.
authentication
The act of determining that the credentials a user presents during a particular request are “good”. Authentication in Pyramid is performed via an authentication policy.
authentication policy
An authentication policy in Pyramid terms is a bit of code which has an API which determines the current principal (or principals) associated with a request.
authorization
The act of determining whether a user can perform a specific action. In pyramid terms, this means determining whether, for a given resource, any principal (or principals) associated with the request have the requisite permission to allow the request to continue. Authorization in Pyramid is performed via its authorization policy.
authorization policy
An authorization policy in Pyramid terms is a bit of code which has an API which determines whether or not the principals associated with the request can perform an action associated with a permission, based on the information found on the context resource.
Babel
A collection of tools for internationalizing Python applications. Pyramid does not depend on Babel to operate, but if Babel is installed, additional locale functionality becomes available to your application.
Chameleon
chameleon is an attribute language template compiler which supports both the ZPT and Genshi templating specifications. It is written and maintained by Malthe Borch. It has several extensions, such as the ability to use bracketed (Genshi-style) ${name} syntax, even within ZPT. It is also much faster than the reference implementations of both ZPT and Genshi. Pyramid offers Chameleon templating out of the box in ZPT and text flavors.
configuration declaration
An individual method call made to an instance of a Pyramid Configurator object which performs an arbitrary action, such as registering a view configuration (via the add_view() method of the configurator) or route configuration (via the add_route() method of the configurator). A set of configuration declarations is also implied by the configuration decoration detected by a scan of code in a package.
configuration decoration
Metadata implying one or more configuration declaration invocations. Often set by configuration Python decorator attributes, such as pyramid.view.view_config, aka @view_config.
configurator
An object used to do configuration declaration within an application. The most common configurator is an instance of the pyramid.config.Configurator class.
context
A resource in the resource tree that is found during traversal or URL dispatch based on URL data; if it’s found via traversal, it’s usually a resource object that is part of a resource tree; if it’s found via URL dispatch, it’s an object manufactured on behalf of the route’s “factory”. A context resource becomes the subject of a view, and often has security information attached to it. See the Traversal chapter and the URL Dispatch chapter for more information about how a URL is resolved to a context resource.
CPython
The C implementation of the Python language. This is the reference implementation that most people refer to as simply “Python”; Jython, Google’s App Engine, and PyPy are examples of non-C based Python implementations.
declarative configuration
The configuration mode in which you use the combination of configuration decoration and a scan to configure your Pyramid application.
decorator
A wrapper around a Python function or class which accepts the function or class as its first argument and which returns an arbitrary object. Pyramid provides several decorators, used for configuration and return value modification purposes. See also PEP 318.
Default Locale Name
The locale name used by an application when no explicit locale name is set. See Localization-Related Deployment Settings.
default permission
A permission which is registered as the default for an entire application. When a default permission is in effect, every view configuration registered with the system will be effectively amended with a permission argument that will require that the executing user possess the default permission in order to successfully execute the associated view callable See also Setting a Default Permission.
Default view
The default view of a resource is the view invoked when the view name is the empty string (''). This is the case when traversal exhausts the path elements in the PATH_INFO of a request before it returns a context resource.
Deployment settings
Deployment settings are settings passed to the Configurator as a settings argument. These are later accessible via a request.registry.settings dictionary in views or as config.registry.settings in configuration code. Deployment settings can be used as global application values.
distribution
(Setuptools/distutils terminology). A file representing an installable library or application. Distributions are usually files that have the suffix of .egg, .tar.gz, or .zip. Distributions are the target of Setuptools commands such as easy_install.
distutils
The standard system for packaging and distributing Python packages. See http://docs.python.org/distutils/index.html for more information. setuptools is actually an extension of the Distutils.
Django
A full-featured Python web framework.
domain model
Persistent data related to your application. For example, data stored in a relational database. In some applications, the resource tree acts as the domain model.
dotted Python name
A reference to a Python object by name using a string, in the form path.to.modulename:attributename. Often used in Paste and setuptools configurations. A variant is used in dotted names within configurator method arguments that name objects (such as the “add_view” method’s “view” and “context” attributes): the colon (:) is not used; in its place is a dot.
entry point
A setuptools indirection, defined within a setuptools distribution setup.py. It is usually a name which refers to a function somewhere in a package which is held by the distribution.
event
An object broadcast to zero or more subscriber callables during normal Pyramid system operations during the lifetime of an application. Application code can subscribe to these events by using the subscriber functionality described in Using Events.
exception response
A response that is generated as the result of a raised exception being caught by an exception view.
Exception view
An exception view is a view callable which may be invoked by Pyramid when an exception is raised during request processing. See Custom Exception Views for more information.
finished callback
A user-defined callback executed by the router unconditionally at the very end of request processing . See Using Finished Callbacks.
Forbidden view
An exception view invoked by Pyramid when the developer explicitly raises a pyramid.httpexceptions.HTTPForbidden exception from within view code or root factory code, or when the view configuration and authorization policy found for a request disallows a particular view invocation. Pyramid provides a default implementation of a forbidden view; it can be overridden. See Changing the Forbidden View.
Genshi
An XML templating language by Christopher Lenz.
Gettext
The GNU gettext library, used by the Pyramid translation machinery.
Google App Engine
Google App Engine (aka “GAE”) is a Python application hosting service offered by Google. Pyramid runs on GAE.
Grok
A web framework based on Zope 3.
HTTP Exception
The set of exception classes defined in pyramid.httpexceptions. These can be used to generate responses with various status codes when raised or returned from a view callable. See also HTTP Exceptions.
imperative configuration
The configuration mode in which you use Python to call methods on a Configurator in order to add each configuration declaration required by your application.
interface
A Zope interface object. In Pyramid, an interface may be attached to a resource object or a request object in order to identify that the object is “of a type”. Interfaces are used internally by Pyramid to perform view lookups and other policy lookups. The ability to make use of an interface is exposed to an application programmers during view configuration via the context argument, the request_type argument and the containment argument. Interfaces are also exposed to application developers when they make use of the event system. Fundamentally, Pyramid programmers can think of an interface as something that they can attach to an object that stamps it with a “type” unrelated to its underlying Python type. Interfaces can also be used to describe the behavior of an object (its methods and attributes), but unless they choose to, Pyramid programmers do not need to understand or use this feature of interfaces.
Internationalization
The act of creating software with a user interface that can potentially be displayed in more than one language or cultural context. Often shortened to “i18n” (because the word “internationalization” is I, 18 letters, then N). See also: Localization.
Jinja2
A text templating language by Armin Ronacher.
jQuery
A popular Javascript library.
JSON
JavaScript Object Notation is a data serialization format.
Jython
A Python implementation written for the Java Virtual Machine.
lineage
An ordered sequence of objects based on a “location -aware” resource. The lineage of any given resource is composed of itself, its parent, its parent’s parent, and so on. The order of the sequence is resource-first, then the parent of the resource, then its parent’s parent, and so on. The parent of a resource in a lineage is available as its __parent__ attribute.
Lingua
A package by Wichert Akkerman which provides Babel message extractors for Python source files and Chameleon ZPT template files.
Locale Name
A string like en, en_US, de, or de_AT which uniquely identifies a particular locale.
Locale Negotiator
An object supplying a policy determining which locale name best represents a given request. It is used by the pyramid.i18n.get_locale_name(), and pyramid.i18n.negotiate_locale_name() functions, and indirectly by pyramid.i18n.get_localizer(). The pyramid.i18n.default_locale_negotiator() function is an example of a locale negotiator.
Localization
The process of displaying the user interface of an internationalized application in a particular language or cultural context. Often shortened to “l10” (because the word “localization” is L, 10 letters, then N). See also: Internationalization.
Localizer
An instance of the class pyramid.i18n.Localizer which provides translation and pluralization services to an application. It is retrieved via the pyramid.i18n.get_localizer() function.
location
The path to an object in a resource tree. See Location-Aware Resources for more information about how to make a resource object location-aware.
Mako
Mako is a template language language which refines the familiar ideas of componentized layout and inheritance using Python with Python scoping and calling semantics.
matchdict
The dictionary attached to the request object as request.matchdict when a URL dispatch route has been matched. Its keys are names as identified within the route pattern; its values are the values matched by each pattern name.
Message Catalog
A gettext .mo file containing translations.
Message Identifier
A string used as a translation lookup key during localization. The msgid argument to a translation string is a message identifier. Message identifiers are also present in a message catalog.
METAL
Macro Expansion for TAL, a part of ZPT which makes it possible to share common look and feel between templates.
middleware
Middleware is a WSGI concept. It is a WSGI component that acts both as a server and an application. Interesting uses for middleware exist, such as caching, content-transport encoding, and other functions. See WSGI.org or PyPI to find middleware for your application.
mod_wsgi
mod_wsgi is an Apache module developed by Graham Dumpleton. It allows WSGI applications (such as applications developed using Pyramid) to be served using the Apache web server.
module
A Python source file; a file on the filesystem that typically ends with the extension .py or .pyc. Modules often live in a package.
multidict
An ordered dictionary that can have multiple values for each key. Adds the methods getall, getone, mixed, add and dict_of_lists to the normal dictionary interface. See Multidict and pyramid.interfaces.IMultiDict.
Not Found view
An exception view invoked by Pyramid when the developer explicitly raises a pyramid.httpexceptions.HTTPNotFound exception from within view code or root factory code, or when the current request doesn’t match any view configuration. Pyramid provides a default implementation of a not found view; it can be overridden. See Changing the Not Found View.
package
A directory on disk which contains an __init__.py file, making it recognizable to Python as a location which can be import -ed. A package exists to contain module files.
Paste
Paste is a WSGI development and deployment system developed by Ian Bicking.
PasteDeploy
PasteDeploy is a library used by Pyramid which makes it possible to configure WSGI components together declaratively within an .ini file. It was developed by Ian Bicking as part of Paste.
permission
A string or unicode object that represents an action being taken against a context resource. A permission is associated with a view name and a resource type by the developer. Resources are decorated with security declarations (e.g. an ACL), which reference these tokens also. Permissions are used by the active security policy to match the view permission against the resources’s statements about which permissions are granted to which principal in a context in order to answer the question “is this user allowed to do this”. Examples of permissions: read, or view_blog_entries.
pipeline
The Paste term for a single configuration of a WSGI server, a WSGI application, with a set of middleware in-between.
pkg_resources
A module which ships with setuptools that provides an API for addressing “asset files” within a Python package. Asset files are static files, template files, etc; basically anything non-Python-source that lives in a Python package can be considered a asset file. See also PkgResources
predicate
A test which returns True or False. Two different types of predicates exist in Pyramid: a view predicate and a route predicate. View predicates are attached to view configuration and route predicates are attached to route configuration.
pregenerator
A pregenerator is a function associated by a developer with a route. It is called by route_url() in order to adjust the set of arguments passed to it by the user for special purposes. It will influence the URL returned by route_url(). See pyramid.interfaces.IRoutePregenerator for more information.
principal
A principal is a string or unicode object representing a userid or a group id. It is provided by an authentication policy. For example, if a user had the user id “bob”, and Bob was part of two groups named “group foo” and “group bar”, the request might have information attached to it that would indicate that Bob was represented by three principals: “bob”, “group foo” and “group bar”.
project
(Setuptools/distutils terminology). A directory on disk which contains a setup.py file and one or more Python packages. The setup.py file contains code that allows the package(s) to be installed, distributed, and tested.
Pylons
A lightweight Python web framework and a predecessor of Pyramid.
PyPI
The Python Package Index, a collection of software available for Python.
PyPy
PyPy is an “alternative implementation of the Python language”:http://pypy.org/
Pyramid Cookbook
An additional documentation resource for Pyramid which presents topical, practical usages of Pyramid available via http://docs.pylonsproject.org/ .
pyramid_debugtoolbar
A Pyramid add on which displays a helpful debug toolbar “on top of” HTML pages rendered by your application, displaying request, routing, and database information. pyramid_debugtoolbar is configured into the development.ini of all applications which use a Pyramid scaffold. For more information, see http://docs.pylonsproject.org/projects/pyramid_debugtoolbar/dev/ .
pyramid_exclog
A package which logs Pyramid application exception (error) information to a standard Python logger. This add-on is most useful when used in production applications, because the logger can be configured to log to a file, to UNIX syslog, to the Windows Event Log, or even to email. See its documentation.
pyramid_handlers
An add-on package which allows Pyramid users to create classes that are analogues of Pylons 1 “controllers”. See http://docs.pylonsproject.org/projects/pyramid_handlers/dev/ .
pyramid_jinja2
Jinja2 templating system bindings for Pyramid, documented at http://docs.pylonsproject.org/projects/pyramid_jinja2/dev/ . This package also includes a scaffold named pyramid_jinja2_starter, which creates an application package based on the Jinja2 templating system.
pyramid_zcml
An add-on package to Pyramid which allows applications to be configured via ZCML. It is available on PyPI. If you use pyramid_zcml, you can use ZCML as an alternative to imperative configuration or configuration decoration.
Python
The programming language in which Pyramid is written.
renderer
A serializer that can be referred to via view configuration which converts a non-Response return values from a view into a string (and ultimately a response). Using a renderer can make writing views that require templating or other serialization less tedious. See Writing View Callables Which Use a Renderer for more information.
renderer factory
A factory which creates a renderer. See Adding and Changing Renderers for more information.
renderer globals
Values injected as names into a renderer based on application policy. See Adding Renderer Globals (Deprecated) for more information.
Repoze
“Repoze” is essentially a “brand” of software developed by Agendaless Consulting and a set of contributors. The term has no special intrinsic meaning. The project’s website has more information. The software developed “under the brand” is available in a Subversion repository. Pyramid was originally known as repoze.bfg.
repoze.catalog
An indexing and search facility (fielded and full-text) based on zope.index. See the documentation for more information.
repoze.lemonade
Zope2 CMF-like data structures and helper facilities for CA-and-ZODB-based applications useful within Pyramid applications.
repoze.who
Authentication middleware for WSGI applications. It can be used by Pyramid to provide authentication information.
repoze.workflow
Barebones workflow for Python apps . It can be used by Pyramid to form a workflow system.
request
An object that represents an HTTP request, usually an instance of the pyramid.request.Request class. See Request and Response Objects (narrative) and pyramid.request (API documentation) for information about request objects.
request factory
An object which, provided a WSGI environment as a single positional argument, returns a Pyramid-compatible request.
request type
An attribute of a request that allows for specialization of view invocation based on arbitrary categorization. The every request object that Pyramid generates and manipulates has one or more interface objects attached to it. The default interface attached to a request object is pyramid.interfaces.IRequest.
resource
An object representing a node in the resource tree of an application. If traversal is used, a resource is an element in the resource tree traversed by the system. When traversal is used, a resource becomes the context of a view. If url dispatch is used, a single resource is generated for each request and is used as the context resource of a view.
Resource Location
The act of locating a context resource given a request. Traversal and URL dispatch are the resource location subsystems used by Pyramid.
resource tree
A nested set of dictionary-like objects, each of which is a resource. The act of traversal uses the resource tree to find a context resource.
response
An object returned by a view callable that represents response data returned to the requesting user agent. It must implements the pyramid.interfaces.IResponse interface. A response object is typically an instance of the pyramid.response.Response class or a subclass such as pyramid.httpexceptions.HTTPFound. See Request and Response Objects for information about response objects.
response adapter
A callable which accepts an arbitrary object and “converts” it to a pyramid.response.Response object. See Changing How Pyramid Treats View Responses for more information.
response callback
A user-defined callback executed by the router at a point after a response object is successfully created. See Using Response Callbacks.
reStructuredText
A plain text format that is the defacto standard for descriptive text shipped in distribution files, and Python docstrings. This documentation is authored in ReStructuredText format.
root
The object at which traversal begins when Pyramid searches for a context resource (for URL Dispatch, the root is always the context resource unless the traverse= argument is used in route configuration).
root factory
The “root factory” of a Pyramid application is called on every request sent to the application. The root factory returns the traversal root of an application. It is conventionally named get_root. An application may supply a root factory to Pyramid during the construction of a Configurator. If a root factory is not supplied, the application uses a default root object. Use of the default root object is useful in application which use URL dispatch for all URL-to-view code mappings.
route
A single pattern matched by the url dispatch subsystem, which generally resolves to a root factory (and then ultimately a view). See also url dispatch.
route configuration
Route configuration is the act of associating request parameters with a particular route using pattern matching and route predicate statements. See URL Dispatch for more information about route configuration.
route predicate
An argument to a route configuration which implies a value that evaluates to True or False for a given request. All predicates attached to a route configuration must evaluate to True for the associated route to “match” the current request. If a route does not match the current request, the next route (in definition order) is attempted.
router
The WSGI application created when you start a Pyramid application. The router intercepts requests, invokes traversal and/or URL dispatch, calls view functions, and returns responses to the WSGI server on behalf of your Pyramid application.
Routes
A system by Ben Bangert which parses URLs and compares them against a number of user defined mappings. The URL pattern matching syntax in Pyramid is inspired by the Routes syntax (which was inspired by Ruby On Rails pattern syntax).
routes mapper
An object which compares path information from a request to an ordered set of route patterns. See URL Dispatch.
scaffold
A project template that helps users get started writing a Pyramid application quickly. Scaffolds are usually used via the paster create command.
scan
The term used by Pyramid to define the process of importing and examining all code in a Python package or module for configuration decoration.
session
A namespace that is valid for some period of continual activity that can be used to represent a user’s interaction with a web application.
session factory
A callable, which, when called with a single argument named request (a request object), returns a session object.
setuptools
Setuptools builds on Python’s distutils to provide easier building, distribution, and installation of libraries and applications.
SQLAlchemy
SQLAlchemy is an object relational mapper used in tutorials within this documentation.
subpath
A list of element “left over” after the router has performed a successful traversal to a view. The subpath is a sequence of strings, e.g. ['left', 'over', 'names']. Within Pyramid applications that use URL dispatch rather than traversal, you can use *subpath in the route pattern to influence the subpath. See Using *subpath in a Route Pattern for more information.
subscriber
A callable which receives an event. A callable becomes a subscriber via imperative configuration or via configuration decoration. See Using Events for more information.
template
A file with replaceable parts that is capable of representing some text, XML, or HTML when rendered.
thread local
A thread-local variable is one which is essentially a global variable in terms of how it is accessed and treated, however, each thread used by the application may have a different value for this same “global” variable. Pyramid uses a small number of thread local variables, as described in Thread Locals. See also the threading.local documentation for more information.
Translation Directory
A translation directory is a gettext translation directory. It contains language folders, which themselves contain LC_MESSAGES folders, which contain .mo files. Each .mo file represents a set of translations for a language in a translation domain. The name of the .mo file (minus the .mo extension) is the translation domain name.
Translation Domain
A string representing the “context” in which a translation was made. For example the word “java” might be translated differently if the translation domain is “programming-languages” than would be if the translation domain was “coffee”. A translation domain is represnted by a collection of .mo files within one or more translation directory directories.
Translation String
An instance of pyramid.i18n.TranslationString, which is a class that behaves like a Unicode string, but has several extra attributes such as domain, msgid, and mapping for use during translation. Translation strings are usually created by hand within software, but are sometimes created on the behalf of the system for automatic template translation. For more information, see Internationalization and Localization.
Translator
A callable which receives a translation string and returns a translated Unicode object for the purposes of internationalization. A localizer supplies a translator to a Pyramid application accessible via its translate method.
traversal
The act of descending “up” a tree of resource objects from a root resource in order to find a context resource. The Pyramid router performs traversal of resource objects when a root factory is specified. See the Traversal chapter for more information. Traversal can be performed instead of URL dispatch or can be combined with URL dispatch. See Combining Traversal and URL Dispatch for more information about combining traversal and URL dispatch (advanced).
tween
A bit of code that sits between the Pyramid router’s main request handling function and the upstream WSGI component that uses Pyramid as its ‘app’. The word “tween” is a contraction of “between”. A tween may be used by Pyramid framework extensions, to provide, for example, Pyramid-specific view timing support, bookkeeping code that examines exceptions before they are returned to the upstream WSGI application, or a variety of other features. Tweens behave a bit like WSGI ‘middleware’ but they have the benefit of running in a context in which they have access to the Pyramid application registry as well as the Pyramid rendering machinery. See Registering “Tweens”.
URL dispatch
An alternative to traversal as a mechanism for locating a context resource for a view. When you use a route in your Pyramid application via a route configuration, you are using URL dispatch. See the URL Dispatch for more information.
Venusian
Venusian is a library which allows framework authors to defer decorator actions. Instead of taking actions when a function (or class) decorator is executed at import time, the action usually taken by the decorator is deferred until a separate “scan” phase. Pyramid relies on Venusian to provide a basis for its scan feature.
view
Common vernacular for a view callable.
view callable
A “view callable” is a callable Python object which is associated with a view configuration; it returns a response object . A view callable accepts a single argument: request, which will be an instance of a request object. An alternate calling convention allows a view to be defined as a callable which accepts a pair of arguments: context and request: this calling convention is useful for traversal-based applications in which a context is always very important. A view callable is the primary mechanism by which a developer writes user interface code within Pyramid. See Views for more information about Pyramid view callables.
view configuration
View configuration is the act of associating a view callable with configuration information. This configuration information helps map a given request to a particular view callable and it can influence the response of a view callable. Pyramid views can be configured via imperative configuration, or by a special @view_config decorator coupled with a scan. See View Configuration for more information about view configuration.
View handler
A view handler ties together pyramid.config.Configurator.add_route() and pyramid.config.Configurator.add_view() to make it more convenient to register a collection of views as a single class when using url dispatch. View handlers ship as part of the pyramid_handlers add-on package.
View Lookup
The act of finding and invoking the “best” view callable given a request and a context resource.
view mapper
A view mapper is a class which implements the pyramid.interfaces.IViewMapperFactory interface, which performs view argument and return value mapping. This is a plug point for extension builders, not normally used by “civilians”.
view name
The “URL name” of a view, e.g index.html. If a view is configured without a name, its name is considered to be the empty string (which implies the default view).
view predicate
An argument to a view configuration which evaluates to True or False for a given request. All predicates attached to a view configuration must evaluate to true for the associated view to be considered as a possible callable for a given request.
virtual root
A resource object representing the “virtual” root of a request; this is typically the physical root object (the object returned by the application root factory) unless Virtual Hosting is in use.
virtualenv
An isolated Python environment. Allows you to control which packages are used on a particular project by cloning your main Python. virtualenv was created by Ian Bicking.
WebOb
WebOb is a WSGI request/response library created by Ian Bicking.
WebTest
WebTest is a package which can help you write functional tests for your WSGI application.
WSGI
Web Server Gateway Interface. This is a Python standard for connecting web applications to web servers, similar to the concept of Java Servlets. Pyramid requires that your application be served as a WSGI application.
ZCML
Zope Configuration Markup Language, an XML dialect used by Zope and pyramid_zcml for configuration tasks.
ZEO
Zope Enterprise Objects allows multiple simultaneous processes to access a single ZODB database.
ZODB
Zope Object Database, a persistent Python object store.
Zope
The Z Object Publishing Framework, a full-featured Python web framework.
Zope Component Architecture
The Zope Component Architecture (aka ZCA) is a system which allows for application pluggability and complex dispatching based on objects which implement an interface. Pyramid uses the ZCA “under the hood” to perform view dispatching and other application configuration tasks.
ZPT
The Zope Page Template templating language.

Foreword

A Foreword By Paul Everitt

Paul Everitt is a principal at Agendaless Consulting. Before his time at Agendaless, he was the co-founder of Digital Creations, which later became Zope Corporation. He has been a widely respected member of the Python community since 1994.

Some times amazing things can actually happen.

In the world of web frameworks, the rate of radioactive decay is very high. Projects are starting, splintering, folding, and clashing constantly. For Python, there are over 50 listed web frameworks. In some ways this shows health and experimentation. Yet others have started to ask: “Is this really good for Python developers?”

This book is the result of an event which bucked this trend, an event which Armin Ronacher wrote was “one of the greatest moves in Python’s web framework history.” Two projects merged and are bringing in a third. Consolidation won a victory over splintering.

As someone from the Zope world, I had a strong interest in repoze.bfg. I viewed it as the escape hatch for Zope, teleporting us into the modern world of Python development, permitting but not requiring Zope-style idioms. Chris McDonough established a great brand for repoze.bfg: small, documented, tested, fast, stable, friendly. As the project manager for a very large repoze.bfg application, I can strongly attest that it was a home run on those points.

But in a crowded web frameworks landscape, repoze.bfg was a long-shot to get critical mass. It had a lot to offer, but was missing critical pieces such as momentum and name recognition.

Pylons has long been viewed as holding the number two spot in Python web frameworks. It is one of (if not the) first “modern” web framework. With lots of users, and a “full-stack” framework atop it (TurboGears), Pylons had momentum and name recognition aplenty. But it needed more resources to accomplish its goals of an architectural transition, and Ben Bangert needed to share the load as architect during the transition.

Ben and Chris started talking during 2011 about architectural patterns and discovered Pylons and repoze.bfg covered almost exactly the same surface area. After some experiments, it became clear that, technically at least, the next version of Pylons could be the same as the next version of repoze.bfg.

But what about the non-technical parts? It was one thing to consolidate code. Consolidating projects was new territory.

I was fortunate to meet with the principals in Las Vegas and watch as they hashed out the idea. The projects would merge and keep the Pylons identity. repoze.bfg would sacrifice its identity, but provide the technical foundation. All the resources from the two projects would be combined.

I’ll confess, I had high hopes for the outcome. Now that the merge has happened and 1.0 released, I can honestly say it has done better than I could have imagined. The story of “consolidation” is catching on, and interest in working together is growing. Pyramid 1.0 is very, very high quality and ready to go for PyCon 2011. People interested in “simple, fast, documented, tested” have a strong framework and healthy project.

It took humility, patience, and pragmatism to reach this point of obvious success. Certainly by the project leaders, who each had to give up some of their sovereignty and sacred cows. But as well, each community had to discuss the challenges, the various alternatives for going forward, and the pros and cons of consolidation in general but also this particular consolidation. That such a conversation and change could happen in a responsible, adult fashion speaks volumes about the strength and maturity of each community.

What might happen in 2011? TurboGears is considering a move into the umbrella Pylons Project. As Armin writes in his post, there is fertile ground for consolidation at other layers. In my own interests, I hope the worlds of Zope and Plone view Pyramid as the base for the next decade of their ideas. But also, the Pylons Project as a vibrant home for such ideas.

Congratulations, Pylons Project. Not only have you accelerated your spot on the Python web frameworks chart, but you have injected the word “consolidation” into the lexicon of hot ideas for 2011.

The Pyramid Web Application Framework

Front Matter

Author Introduction

Welcome to “The Pyramid Web Application Framework”. In this introduction, I’ll describe the audience for this book, I’ll describe the book content, I’ll provide some context regarding the genesis of Pyramid, and I’ll thank some important people.

I hope you enjoy both this book and the software it documents. I’ve had a blast writing both.

Audience

This book is aimed primarily at a reader that has the following attributes:

  • At least a moderate amount of Python experience.
  • A familiarity with web protocols such as HTTP and CGI.

If you fit into both of these categories, you’re in the direct target audience for this book. But don’t worry, even if you have no experience with Python or the web, both are easy to pick up “on the fly”.

Python is an excellent language in which to write applications; becoming productive in Python is almost mind-blowingly easy. If you already have experience in another language such as Java, Visual Basic, Perl, Ruby, or even C/C++, learning Python will be a snap; it should take you no longer than a couple of days to become modestly productive. If you don’t have previous programming experience, it will be slightly harder, and it will take a little longer, but you’d be hard-pressed to find a better “first language.”

Web technology familiarity is assumed in various places within the book. For example, the book doesn’t try to define common web-related concepts like “URL” or “query string.” Likewise, the book describes various interactions in terms of the HTTP protocol, but it does not describe how the HTTP protocol works in detail. Like any good web framework, though, Pyramid shields you from needing to know most of the gory details of web protocols and low-level data structures. As a result, you can usually avoid becoming “blocked” while you read this book even if you don’t yet deeply understand web technologies.

Book Content

This book is divided into three major parts:

Narrative Documentation

This is documentation which describes Pyramid concepts in narrative form, written in a largely conversational tone. Each narrative documentation chapter describes an isolated Pyramid concept. You should be able to get useful information out of the narrative chapters if you read them out-of-order, or when you need only a reminder about a particular topic while you’re developing an application.

Tutorials

Each tutorial builds a sample application or implements a set of concepts with a sample; it then describes the application or concepts in terms of the sample. You should read the tutorials if you want a guided tour of Pyramid.

API Reference

Comprehensive reference material for every public API exposed by Pyramid. The API documentation is organized alphabetically by module name.
The Genesis of repoze.bfg

Before the end of 2010, Pyramid was known as repoze.bfg.

I wrote repoze.bfg after many years of writing applications using Zope. Zope provided me with a lot of mileage: it wasn’t until almost a decade of successfully creating applications using it that I decided to write a different web framework. Although repoze.bfg takes inspiration from a variety of web frameworks, it owes more of its core design to Zope than any other.

The Repoze “brand” existed before repoze.bfg was created. One of the first packages developed as part of the Repoze brand was a package named repoze.zope2. This was a package that allowed Zope 2 applications to run under a WSGI server without modification. Zope 2 did not have reasonable WSGI support at the time.

During the development of the repoze.zope2 package, I found that replicating the Zope 2 “publisher” – the machinery that maps URLs to code – was time-consuming and fiddly. Zope 2 had evolved over many years, and emulating all of its edge cases was extremely difficult. I finished the repoze.zope2 package, and it emulates the normal Zope 2 publisher pretty well. But during its development, it became clear that Zope 2 had simply begun to exceed my tolerance for complexity, and I began to look around for simpler options.

I considered using the Zope 3 application server machinery, but it turned out that it had become more indirect than the Zope 2 machinery it aimed to replace, which didn’t fulfill the goal of simplification. I also considered using Django and Pylons, but neither of those frameworks offer much along the axes of traversal, contextual declarative security, or application extensibility; these were features I had become accustomed to as a Zope developer.

I decided that in the long term, creating a simpler framework that retained features I had become accustomed to when developing Zope applications was a more reasonable idea than continuing to use any Zope publisher or living with the limitations and unfamiliarities of a different framework. The result is what is now Pyramid.

The Genesis of Pyramid

What was repoze.bfg has become Pyramid as the result of a coalition built between the Repoze and Pylons community throughout the year 2010. By merging technology, we’re able to reduce duplication of effort, and take advantage of more of each others’ technology.

Thanks

This book is dedicated to my grandmother, who gave me my first typewriter (a Royal), and my mother, who bought me my first computer (a VIC-20).

Thanks to the following people for providing expertise, resources, and software. Without the help of these folks, neither this book nor the software which it details would exist: Paul Everitt, Tres Seaver, Andrew Sawyers, Malthe Borch, Carlos de la Guardia, Chris Rossi, Shane Hathaway, Daniel Holth, Wichert Akkerman, Georg Brandl, Blaise Laflamme, Ben Bangert, Casey Duncan, Hugues Laflamme, Mike Orr, John Shipman, Chris Beelby, Patricio Paez, Simon Oram, Nat Hardwick, Ian Bicking, Jim Fulton, Michael Merickel, Tom Moroz of the Open Society Institute, and Todd Koym of Environmental Health Sciences.

Thanks to Guido van Rossum and Tim Peters for Python.

Special thanks to Tricia for putting up with me.

Narrative Documentation

Tutorials

API Reference

Glossary and Index