Edit me on GitHub

pyramid_debugtoolbar

Overview

pyramid_debugtoolbar provides a useful debug toolbar while you’re developing a Pyramid application.

The toolbar is a blatant rip-off of Michael van Tellingen’s flask-debugtoolbar (which itself was derived from Rob Hudson’s django-debugtoolbar). It also includes a lightly sanded down version of the Werkzeug debugger code by Armin Ronacher and team.

Warning

This package only works with Pyramid 1.2a1 and better.

Installation

Install using setuptools, e.g. (within a virtualenv):

$ easy_install pyramid_debugtoolbar

Setup

Once the pyramid_debugtoolbar is installed, you must use the config.include mechanism to include it into your Pyramid project’s configuration:

config = Configurator(.....)
config.include('pyramid_debugtoolbar')

Alternately, you may activate the toolbar by changing your application’s .ini file by adding it to the pyramid.includes list:

pyramid.includes = pyramid_debugtoolbar

Warning

The debug toolbar should never be enabled in a production environment or on a machine with its Pyramid HTTP port exposed directly to the internet; it allows arbitrary code execution from only semi-trusted sources when configured poorly.

Usage

Once Pyramid is restarted, the toolbar will be available to inspect requests and responses by the application by visiting the /_debug_toolbar/ URL (note the trailing slash). For example, if your application is available at http://localhost:6543/ then you may visit http://localhost:6543/_debug_toolbar/ to inspect the requests.

For any HTML responses generated by the application, a link to the toolbar for the current page will be available in the upper right corner, provided the response contains a closing </body> tag.

Debugging Unhandled Exceptions

If an exception is unhandled by the Pyramid application, the toolbar will catch it and render an HTML page with a traceback and an interactive debugger that can be used to dive into the stack and execute arbitrary Python expressions to inspect the state of the system.

A URL leading to a debugging page for each exception raised by your application will additionally be logged to the console.

Settings

Settings can be used to control the operation of the toolbar. These settings are typically specified in the Pyramid “app” section of the Pyramid .ini file.

debugtoolbar.hosts

If the request’s REMOTE_ADDR is not in this list, the toolbar will not be displayed and the exception handler will not be active. Default value is ['127.0.0.1', '::1']. Note that each of the values in the list can be a hostmask, e.g., 192.168.1.0/24.

This should be a list if setup is done in Python or, if defined in a Paste .ini file, a single-line list of IP addresses/hostmasks separated by spaces. For example:

debugtoolbar.hosts = 192.168.1.1 192.168.2.0/24

To enable access from any host, use the hostmask 0.0.0.0/0.

debugtoolbar.enabled

true if the toolbar is enabled; false if the toolbar is disabled. Default is true. This disables both the exception handler and the toolbar overlay.

debugtoolbar.intercept_exc

This setting can have one of three values: display, debug or false. Default is debug. If this value is display, the toolbar will display a “pretty” traceback page which allows source viewing and when an exception happens. If this value is debug, the “pretty” traceback page will be shown, but it will also contain interactive debugging controls which allow you to evaluate arbitrary Python expressions in the context of a portion of the traceback, which is useful when attempting to track down the cause of the exception. If this value is false, the “pretty” traceback will be disabled and all exceptions will be raised to the caller of the Pyramid application (usually a WSGI server). Default is debug. This setting differs from debugtoolbar.enabled: it only enables or disables the exception handler. Note that, for backwards compatibility purposes, the value true provided to this setting is interpreted as debug.

debugtoolbar.eval_exc

true if real-time exception debugging is enabled when intercept_exc is true; false if real-time exception debugging is disabled. Default is true. This differs from debugtoolbar.intercept_exc: it only controls whether the pretty exception rendering displays real-time in-browser debugging controls. The real-time in-browser debugging controls allow you to evaluate arbitrary Python expresssions in the context of a stack frame via a browser control.

debugtoolbar.show_on_exc_only

Default is false. If set to true the debugtoolbar will only be injected into the response in case a exception is raised. If the response is processed without exception, the returned html code is not changed by the debugtoolbar at all. This option allows the developer to use the toolbar for debugging purposes without interfering with successful responses.

Inspection of requests is still possible by visiting the toolbar manually.

debugtoolbar.intercept_redirects

true if the redirection handler is enabled; false if the handler is disabled. Default is false. This differs from debugtoolbar.enabled: it only enables or disables the redirection handler.

debugtoolbar.panels

A list of dotted Python global names to panel classes. Defaults to a list of all panel types known by pyramid_debugtoolbar, as documented in pyramid_debugtoolbar API. If this is spelled in an .ini file, it should be a space- or newline-separated sequence of dotted Python names. For example:

debugtoolbar.panels =
    pyramid_debugtoolbar.panels.versions.VersionDebugPanel
    pyramid_debugtoolbar.panels.settings.SettingsDebugPanel
    pyramid_debugtoolbar.panels.headers.HeaderDebugPanel
    pyramid_debugtoolbar.panels.request_vars.RequestVarsDebugPanel
    pyramid_debugtoolbar.panels.renderings.RenderingsDebugPanel
    pyramid_debugtoolbar.panels.logger.LoggingPanel
    pyramid_debugtoolbar.panels.performance.PerformanceDebugPanel
    pyramid_debugtoolbar.panels.routes.RoutesDebugPanel
    pyramid_debugtoolbar.panels.sqla.SQLADebugPanel
    pyramid_debugtoolbar.panels.tweens.TweensDebugPanel
    pyramid_debugtoolbar.panels.introspection.IntrospectionDebugPanel

debugtoolbar.extra_panels

A list of dotted Python global names to panel classes. This list of panels is appended to the panels defined in debugtoolbar.panels. If you’d like to maintain the default panels and add on some extra ones, this should help:

debugtoolbar.extra_panels =
    myapp.debugtoolbar.panels.MyCustomPanel

debugtoolbar.button_style

Any inline css styles you want to apply to the toolbar button. This will override the default style (top:30px;) set by toolbar.css. If, for example, you want the toolbar button to show up at the bottom off the screen, just set debugtoolbar.button_style to top:auto;bottom:30px;. If your browser supports the zoom property, you can even control the magnification level of the toolbar button, e.g., zoom:50%;.

debugtoolbar.exclude_prefixes

The debug toolbar won’t be shown and no data will be recorded if the PATH_INFO variable starts with any of the prefixes listed in this setting. If configuration is done via an .ini file, the prefixes should be separated by carriage returns. For example:

debugtoolbar.exclude_prefixes =
    /settings
    /static

If configuration is done via Python, the setting should be a list. This setting was added in debugtoolbar version 1.0.4.

Useful settings for debugging panels/debugtoolbar

When developing custom panels for an application, the following settings may be used to influence how debugtoolbar itself behaves and what information it logs.

debugtoolbar.debug_notfound

Print view-related NotFound debug messages to stderr when this value is true.

debugtoolbar.debug_routematch

Print debugging messages related to URL dispatch route matching when this value is true.

debugtoolbar.reload_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.

debugtoolbar.reload_resources

Alias for debugtoolbar.reload_assets.

debugtoolbar.reload_assets

Don’t cache any asset file data when this value is true.

debugtoolbar.prevent_http_cache

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.

Custom authorization

Since version 1.0.5 pyramid_debugtoolbar offers custom authorization mechanism to control toolbar feature on per-request basis. Using the config.set_debugtoolbar_request_authorization(callback) directive, you can specify your own function to control whether toolbar functionality is enabled or not.

Note

Custom authorization is performed after a successful IP address check when the debugtoolbar.hosts settings option is used.

Note

Custom authorization does not have an effect on the pyramid_debugtoolbar static route and /_debug_toolbar/static/* contents will still be accessible.

from pyramid.security import authenticated_userid
from pyramid.settings import aslist

def admin_only_debugtoolbar(request):
    """
    Enable toolbar for administrators only.
    Returns True when it should be enabled.
    """
    admins = aslist(request.registry.settings.get('admins', ''))
    userid = authenticated_userid(request)
    toolbar_enabled = userid and userid in admins
    return toolbar_enabled

config = Configurator(.....)
config.include('pyramid_debugtoolbar')
config.set_debugtoolbar_request_authorization(admin_only_debugtoolbar)

The Toolbar

When you include the toolbar in your application, a floating toolbar logo will appear over your application’s HTML:

_images/toolbar-closed.png

If you click on the “Pyramid DT” logo, a new target window will open with your current request highlighted and all of your configured panels loaded.

_images/toolbar-open.png

Toolbar Panels

These are the default toolbar panels:

Versions

Displays versions of all installed Python software as well as the Python version and platform itself.

_images/versions.png

Settings

Displays Pyramid deployment settings, i.e., registry.settings.

_images/settings.png

HTTP Headers

Displays HTTP request and response headers for the current page.

_images/headers.png

Request Vars

Displays objects attached to the request of the current page and the WSGI environment.

_images/requestvars.png

Renderings

Displays the renderings performed by Pyramid for the current page.

_images/renderings.png

Logging

Displays messages logged by the current page.

_images/logging.png

Performance

Displays timing information, and, if enabled, Python profiling information for the current page. When it is red, only timing will be done and no profiling information.

Note

An internal profiler can be enabled through the “performance” checkmark in the “Settings” tab in the navigation bar. When the checkbox is green, the request will be profiled and profiling information will be gathered and displayed on the “Performance” panel output.

_images/performance.png

Routes

Displays the routes currently configured in your application.

_images/routes.png

Tweens

Displays the tween chain for your application, and whether they were defined explicitly or implicitly.

_images/tweens.png

SQLAlchemy

Displays SQL queries made by SQLAlchemy by the current page along with timing information.

_images/sqla.png

Provides the ability to re-run the query using the “SELECT” link.

_images/sqla-select.png

Provides the ability to get more detail about the query using the “EXPLAIN” link.

_images/sqla-explain.png

Introspection

Displays a rendering of the data available in Pyramid’s configuration introspection system (available in Pyramid 1.3+ only).

_images/introspection.png

Exception Handling

When an exception is raised and the debugtoolbar.intercept_exc setting is display or debug, Pyramid presents a pretty traceback page. If the setting value is debug, you will be able to examine locals in each frame in the traceback and execute code in the context of each frame. Read the instructions on the exception page for more information.

_images/exc.png

Redirect Handling

When a response is returned to Pyramid that has a redirect status code (301, 302, etc.) and the debugtoolbar.intercept_redirect setting is true, Pyramid presents an interim page with a link to the target of the redirect. You can use the toolbar on the redirect source page, then when finished, use the link to continue to the target page.

_images/redirect.png

Adding custom panels

In some cases it can be desirable to add a custom panel to the toolbar to display some application specific data. There are two steps for adding such a panel to an application: writing the panel, and adding it to your application settings.

Understanding how debugtoolbar works

Before writing the panel, you should understand how pyramid_debugtoolbar interacts with your application in its two phase process.

pyramid_debugtoolbar wraps every request within a Pyramid “tween” via toolbar_tween_factory. This tween allows the toolbar to record data during the original request (Phase 1) and injects a link to the toolbar interface into the rendered Pyramid web pages. The data is displayed during a secondary request to the toolbar (Phase 2).

Phase 1 - The original request

When pyramid_debugtoolbar is enabled, it can start tracking data on the original request. This involves calling the following panel methods in-band with the original request:

These methods are used to store and manipulate a self.data variable on each panel during this original request. Typically self.data is first generated on the __init__ method. It is important to note that the request and event variables available to these methods refer to the original request.

Phase 2 - The debugtoolbar request

When the “/_debug_toolbar/{request_id}” is accessed, the history of the original request_id and its associated panels are accessed. Variables such as data that were generated during the original request are made available for further processing. The data variable is injected into the template for display.

The following panel methods are called or accessed on the debugtoolbar request:

Writing the panel

The panel can be created as part of your application or as a standalone package. The easiest way to write a panel is to subclass from the pyramid_debugtoolbar.panels.DebugPanel class. Here is the code for a sample panel:

from pyramid_debugtoolbar.panels import DebugPanel

_ = lambda x: x

class SampleDebugPanel(DebugPanel):
    """
    Sample debug panel
    """
    name = 'Sample'
    has_content = True
    template = 'myapp.lib.debugtoolbar_custom.panels:templates/sample.dbtmako'

    def __init__(self, request):
        self.data = { 'request_path' : request.path_info }

    def nav_title(self):
        return _('Sample')

    def title(self):
        return _('Sample')

def includeme(config):
    config.registry.settings['debugtoolbar.panels'].append(SampleDebugPanel)

After inheriting from the DebugPanel class, you have to define a few methods and attributes on your panel:

name
Attribute. String value. A unique identifier for the name of the panel. This must be defined by a subclass.
has_content
Attribute. Boolean value. Default is True This attribute determines if the tab is enabled or not. If False then the panel’s tab will be disabled and .render_content will not be invoked. Most subclasses will want to set this to True by default. An example of this panel’s dynamic utility is the SQLA panel; if no SqlAlchemy statements were executed in the request, this value is set to False and the tab is simply disabled.
user_activate
Attribute. Boolean value. If the client is able to activate/de-activate the panel then this should be True.
is_active
Attribute. Boolean value. This property will be set by the toolbar, indicating the user’s decision to activate or deactivate the panel. If user_activate is False then is_active will always be set to True.
template
Attribute. String value. Must be overridden. A mako asset specification. The default implementation of render_content in the base class (DebugPanel) will attempt to render template. If template is not defined, and render_content is not overridden, a NotImplemented exception will be raised.
nav_title
Method. Returns a string. Called to get the title to be used on the toolbar’s navigation bar for this panel.
url
Method. Returns a string. Can be overridden to point the panel at any arbitrary URL when the tab is clicked.
title
Method. Returns a string. Called to get the title to be used on the panel’s display page.
__init__
Method. This method should defines a data attribute, which is used when rendering the template. This is the first (and often most appropriate) opportunity to initialize data with values that can be derived from the request object itself.
render_content
Method. Return a string containing the HTML to be rendered for the panel. By default this will render the template defined by the template attribute with a rendering context defined by the data attribute combined with the dict returned from render_vars. The request here is the active request in the toolbar. Not the original request that this panel represents.
render_vars
Method. Invoked by the default implementation of render_content as an opportunity to enhance the rendering context. This method is expected to return a dict of values to use when rendering the panel’s HTML content. This value is usually injected into templates as the rendering context. This is a useful hook for adding any data you need in the templates, which was not already added into the panel’s.`data`. The default SQLA panel is a good example of this functionality in use. The request here is the active request in the toolbar. Not the original request that this panel represents.
wrap_handler
Method. This method is a hook available to the panel in order to track the lifecycle of the original request. A handler accepts a request and returns a response; it is essentially the same as a Pyramid tween. This can be used to update the data dict with values that are wanted for rendering. The main toolbar routine works by wrapping each request in a handler (tween). Before generating a response, the main toolbar routine will call the`wrap_handler` method of each panel. This functionality is often used for decorating the handlers with timing or performance metrics.
process_beforerender
Method. Arguments: self, event. This method is a hook available to the panel in order to track the lifecycle of the original request. The debugtoolbar uses a subscriber event (pyramid.events.BeforeRender) to call the process_beforerender method of each enabled panel. This can be used to update the data dict with values that are wanted for rendering or track properties of the rendering events.
process_response
Method. Arguments: self, response. This method is a hook available to the panel in order to track the lifecycle of the original request. The main toolbar routine works by wrapping each request in a tween. The process_response method of each panel is called within the tween, after the original request has generated a response.

When creating a new panel, some of these methods must be subclassed, while others can rely on the base class.

Once you define the panel it has to be added to the debugtoolbar.panels setting of the configuration. A good way to do this is to use an includeme method in the panel’s __init__.py.

The source code for the standard debugpanel request_vars.py is a good starting point for inspiration.

Configuring an application to use the panel

Once your panel is ready, you can simply add its package name to the pyramid.includes setting on your application configuration file:

pyramid.includes =
    pyramid_debugtoolbar
    samplepanel

Panel and UI Extras

The following is a listing of panels and user interface extras for pyramid_debugtoolbar created by its users. These extras are unofficial and not supported by the Pylons Project. To add your contribution, please submit a pull request to update this documentation.

Page Up
For tabs that have content which requires lots of scrolling down or to the right, clicking the Page Up icon resets the window to 0,0.
pyramid_debugtoolbar_ajax
Adds an “AJAX” panel to the pyramid_debugtoolbar. This panel contains a button to replay the request in a new window – allowing you to spawn a debugger window for errors encountered on background ajax requests.
pyramid_debugtoolbar_dogpile
dogpile caching support for pyramid_debugtoolbar.

Development Versions

Visit http://github.com/Pylons/pyramid_debugtoolbar to download development or tagged versions.

Reporting Bugs

Visit http://github.com/Pylons/pyramid_debugtoolbar/issues to report bugs.