pyramid_jinja2 is a set of bindings that make templates written for the Jinja2 templating system work under the Pyramid web framework.


Install using pip, where $VENV is the path to a virtual environment.

$ $VENV/bin/pip install pyramid_jinja2



If you start a project from scratch, consider using the project template which comes with a working setup and sensible defaults.

There are multiple ways to make sure that pyramid_jinja2 is active. All are completely equivalent:

  1. Use the includeme() function via include().

    config = Configurator()
  2. Add pyramid_jinja2 to the list of your pyramid.includes in your .ini settings file.

    pyramid.includes =
  3. If you use pyramid_zcml instead of imperative configuration, ensure that some ZCML file with an analogue of the following contents is executed by your Pyramid application:

    <include package="pyramid_jinja2"/>

Once activated in any of these ways, the following happens:

  1. Files with the .jinja2 extension are considered to be Jinja2 templates and a jinja2.Environment is registered to handle this extension.

  2. The pyramid_jinja2.add_jinja2_renderer() directive is added to the Configurator instance.

  3. The pyramid_jinja2.add_jinja2_search_path() directive is added to the Configurator instance.

  4. The pyramid_jinja2.add_jinja2_extension() directive is added to the Configurator instance.

  5. The pyramid_jinja2.get_jinja2_environment() directive is added to the Configurator instance.

Preparing for distribution

If you want to make sure your .jinja2 template files are included in your package's source distribution (e.g., when using python sdist), add *.jinja2 to your

recursive-include yourapp *.ico *.png *.css *.gif *.jpg *.pt *.txt *.mak *.mako *.jinja2 *.js *.html *.xml


Once pyramid_jinja2 has been activated, .jinja2 templates can be used by the Pyramid rendering system.

When used as the renderer argument of a view, the view must return a Python dict which will be passed into the template as the set of available variables.

Template Lookup Mechanisms

There are several ways to configure pyramid_jinja2 to find your templates.

Asset Specifications

Templates may always be defined using an asset specification. These are strings which define an absolute location of the template relative to some Python package. For example myapp.views:templates/home.jinja2. These specifications are supported throughout Pyramid and provide a fool-proof way to find any supporting assets bundled with your application.

Here's an example view configuration which uses an asset specification:

2def hello_world(request):
3    return {"a": 1}

Asset specifications have some significant benefits in Pyramid, as they can be fully overridden. An add-on package can ship with code that renders using asset specifications. Later, another package can externally override the templates without having to actually modify the add-on in any way. See Overriding Assets for more information.

Caller-Relative Template Lookup

By default, templates are discovered relative to the caller's package. This means that if you define a view in a Python module, the templates would be found relative to the module's directory on the filesystem.

Let's look at an example:

2def my_view(request):
3    return {"foo": 1, "bar": 2}

Imagine that the above code is in a myapp.admin.views module. The template would be relative to that module on the filesystem, as shown below:

`- admin
   `- templates
      |- base.jinja2
      `- mytemplate.jinja2

Caller-relative lookup avoids naming collisions which can be common in a search path-based approach.

A caller-relative template lookup is converted to a asset specification underneath the hood. This means that it's almost always possible to override the actual template in an add-on package without having to fork the add-on itself. For example, the full asset specification for the view above would be myapp.admin.views:templates/mytemplate.jinja2. This template, or the entire templates folder, may be overridden.


See Overriding Assets for more information.

Search Path-Based Template Lookup

When used outside of Pyramid, Jinja2's default lookup mechanism is a search path. To use a search path within Pyramid, simply define the jinja2.directories configuration setting, or use the add_jinja2_search_path() configurator directive.

Rendering Jinja2 templates with a search path is typically done as follows:

def my_view(request):
    return {"foo": 1, "bar": 2}

If mytemplate.jinja2 is not found in the same directory as the module, then it will be searched for on the search path. We are now dependent on our configuration settings to tell us where the template may be located. Commonly a templates directory is created at the base of the package and the configuration file will include the following directive:

jinja2.directories = mypkg:templates


It is possible to specify a relative path to the templates folder, such as jinja2.directories = templates. This folder will be found relative to the first package that includes pyramid_jinja2, which will normally be the root of your application. It is always better to be explicit when in doubt.


The package that includes pyramid_jinja2 will always be added to the search path (in most cases this is the top-level package in your application). This behavior may be deprecated or removed in the future. It is always better to specify your search path explicitly.

Templates Including Templates

Jinja2 allows template inheritance as well as other mechanisms for templates to load each other. The lookup mechanisms supported in these cases include asset specifications, template-relative names, and normal template names found on the search path. The search path will always be consulted if a template cannot be found relative to the parent template. For example, if you had a template named templates/child.jinja2 that wanted to extend templates/base.jinja2, then it could use {% extends "base.jinja2" %} and locate the file relative to itself. Alternatively it could use {% extends "templates/base.jinja2" %} to find the template in a templates sub-folder rooted on the search path. The template-relative option will always override the search path.

An example:

 1<!-- templates/layout.jinja2 -->
 2<!DOCTYPE html>
 3<html lang="en">
 5  <meta charset="utf-8">
 6  <title>Hello World!</title>
 7  <link rel="stylesheet" href="style.css">
10  <div id="content">{% block content %}{% endblock %}</div>
1<!-- templates/root.jinja2 -->
2{% extends "templates/layout.jinja2" %}
3{% block content %}
6  Some random paragraph.
8{% endblock %}

For further information on Template Inheritance in Jinja2 templates please see Template Inheritance in Jinja2 documentation.

Adding or Overriding a Renderer

By default, only templates ending in the .jinja2 file extension are supported. However, it is very easy to add support for alternative file extensions using the pyramid_jinja2.add_jinja2_renderer() directive.


It would now be possible to use templates named foo.html and foo.jinja2. Each renderer extension will use its own jinja2.Environment. These alternative renderers can be extended at runtime using the name parameter to the other directives such as pyramid_jinja2.get_jinja2_environment().

config.add_jinja2_search_path("myapp:templates", name=".html")

It is also possible to set up different renderers that use different search paths, configuration settings, and environments if necessary. This technique can come in handy when different defaults are required for rendering templates with different content types. For example, a plain text email body versus an HTML page. For this reason, pyramid_jinja2.add_jinja2_renderer() accepts an optional parameter settings_prefix which can point a renderer at a different group of settings.

settings = {
    "jinja2.directories": "myapp:html_templates",
    "mail.jinja2.directories": "myapp:email_templates",

config = Configurator(settings=settings)
config.add_jinja2_renderer(".email", settings_prefix="mail.jinja2.")

Now will be rendered using the mail.jinja2.* settings.

Internalization (i18n)

When pyramid_jinja2 is included in a Pyramid application, either jinja2.ext.i18n or the extension configured by jinja2.i18n_extension is automatically activated.

Be sure to configure jinja2.i18n.domain according to setup.cfg domain settings. By default, jinja2.i18n.domain is set to the name of the package that included pyramid_jinja2. If no package was found, it will use messages.


Jinja2 derives additional settings to configure its template renderer. Many of these settings are optional and only need to be set if they should be different from the default. The below values can be present in the .ini file used to configure the Pyramid application (in the app section representing your Pyramid app) or they can be passed directly within the settings argument passed to a Pyramid Configurator.

Generic Settings

These settings correspond to the ones documented in Jinja2. Set them accordingly.

For reference please see:


For the boolean settings, use true or false.














Jinja2 autoescape setting.

Possible values: true or false.


By default Jinja2 sets autoescape to False.

pyramid_jinja2 sets it to True as it is considered a good security practice in a web setting where we want to prevent XSS attacks from rendering unsanitized user-generated content. To turn off escaping on a case-by-case basis, you may use the safe filter such as {{ html_blob|safe }}.


For usage see Pyramid: Automatically Reloading Templates.

True or False represent whether Jinja2 templates should be reloaded when they change on disk. In development, it is useful to set it to True. This setting sets the Jinja2 auto_reload setting.



Deprecated as of version 1.5, use pyramid.reload_templates instead.


Use Pyramid pyramid.reload_templates setting.


A list of directory names, or a newline-delimited string, where each line represents a directory name. These locations are where Jinja2 will search for templates. Each can optionally be an absolute resource specification (e.g., package:subdirectory/).


The input encoding of templates. Defaults to utf-8.


Changes the undefined types that are used when a variable name lookup fails. If unset, defaults to Undefined (silent ignore). Setting it to strict will trigger StrictUndefined behavior (which raises an error, and is recommended for development). Setting it to debug will trigger DebugUndefined, which outputs debug information in some cases. See Undefined Types.


A list of extension objects, or a newline-delimited set of dotted import locations, where each line represents an extension. Either jinja2.ext.i18n or the i18n extension configured using jinja2.i18n_extension is automatically activated.


The name of the i18n extension to activate. Defaults to jinja2.ext.i18n.


Pyramid domain for translations. See Translation Domain in the Pyramid documentation. Defaults to the name of the package that activated pyramid_jinja2 or if that fails it will use messages as the domain.


A subclass of pyramid_jinja2.i18n.GetTextWrapper to override gettext and ngettext methods in Jinja i18n extension. The Subclass can be either a dotted name or the subclass itself.


A dictionary mapping a filter name to a filter object, or a newline-delimited string with each line in the format:

name =

representing Jinja2 filters.


A dictionary mapping a global name to a global template object, or a newline-delimited string with each line in the format:

name =

representing Jinja2 globals


A dictionary mapping a test name to a test object, or a newline-delimited string with each line in the format:

name =

representing Jinja2 tests.


If set to true, a file system bytecode cache will be configured in a directory determined by jinja2.bytecode_caching_directory. To configure other types of bytecode caching, jinja2.bytecode_caching may also be set directly to an instance of jinja2.BytecodeCache. However doing so cannot be done in a paste .ini file and it must be done programmatically. By default, no bytecode cache is configured.

Changed in version 1.10: Previously, jinja2.bytecode_caching defaulted to true.

Note that configuring a filesystem bytecode cache will (not surprisingly) generate files in the cache directory. As templates are changed, some of these will become stale, pointless wastes of disk space. You are advised to consider a clean up strategy (such as a cron job) to check for and remove such files.

See the Jinja2 Documentation for more information on bytecode caching.

Changed in version 1.10: Previously, an atexit callback which called jinja2.BytecodeCache.clear() was registered in an effort to delete the cache files. This is no longer done.


Absolute path to directory to store bytecode cache files. Defaults to the system temporary directory. This is only used if jinja2.bytecode_caching is set to true.


true or false to enable the use of newstyle gettext calls. Defaults to false.

See New Style Gettext.


A callable or a dotted-import string.

Jinja2 Filters

pyramid_jinja2 comes with Pyramid routing specific filters. All Jinja2 built-in filters are enabled in templates. Read how Filters work in Jinja2.

Installing filters

To use these filters, configure the settings of jinja2.filters:

2# ... other stuff ...
3jinja2.filters =
4    model_url = pyramid_jinja2.filters:model_url_filter
5    route_url = pyramid_jinja2.filters:route_url_filter
6    static_url = pyramid_jinja2.filters:static_url_filter

Filter reference

pyramid_jinja2.filters.resource_url_filter(ctx, model, *elements, **kw)

A filter from model to a string representing the absolute URL. This filter calls pyramid.url.resource_url().


<a href="{{ "my_traversable_object"|resource_url }}">
    See my object

You can also specify optional view name attached at the end of a path:

<a href="{{ "my_traversable_object"|resource_url("edit") }}">
    Edit my object
pyramid_jinja2.filters.model_url_filter(ctx, model, *elements, **kw)

A filter from model to a string representing the absolute URL. This filter calls pyramid.url.resource_url().


This is being deprecated. See pyramid_jinja2.filters.resource_url()

pyramid_jinja2.filters.route_url_filter(ctx, route_name, *elements, **kw)

A filter from route_name to a string representing the absolute URL. This filter calls pyramid.url.route_url().


<a href="{{ "login"|route_url }}">
    Sign in
pyramid_jinja2.filters.static_url_filter(ctx, path, **kw)

A filter from path to a string representing the absolute URL. This filter calls pyramid.url.static_url().


<link rel="stylesheet" href="{{ "yourapp:static/css/style.css"|static_url }}" />
pyramid_jinja2.filters.model_path_filter(ctx, model, *elements, **kw)

A filter from model to a string representing the relative URL. This filter calls pyramid.request.Request.resource_path().

pyramid_jinja2.filters.route_path_filter(ctx, route_name, *elements, **kw)

A filter from route_name to a string representing the relative URL. This filter calls pyramid.url.route_path().

pyramid_jinja2.filters.static_path_filter(ctx, path, **kw)

A filter from path to a string representing the relative URL. This filter calls pyramid.url.static_path().

Creating a Jinja2 Pyramid project

After you have installed pyramid_jinja2, you can invoke the following command to create a Jinja2-based Pyramid project from its included scaffold.

$ $VENV/bin/pcreate -s pyramid_jinja2_starter myproject

After it's created, you can visit the myproject directory and install the project in development mode.

$ cd myproject
$ $VENV/bin/pip install -e .

At this point you can start the application like any other Pyramid application.

$ $VENV/bin/pserve development.ini

This is a good way to see a working Pyramid application that uses Jinja2, even if you do not end up using the result.

See also

See also Creating a Pyramid Project.

Running tests for your application

The scaffold provides a convenience for the developer to install pytest and pytest-cov as the test runner and test coverage. To run unit tests for your application, you must first install the testing dependencies.

$ $VENV/bin/pip install -e ".[testing]"

Once the testing requirements are installed, then you can run the tests using the py.test command that was just installed in the bin directory of your virtual environment. The -q option means "quiet" output, and the --cov option includes test coverage.

$ $VENV/bin/py.test -q --cov

The scaffold includes configuration defaults for py.test and test coverage. These configuration files are pytest.ini and .coveragerc, located at the root of your package. Without these defaults, we would need to specify the path to the module on which we want to run tests and coverage.

$ $VENV/bin/py.test -q --cov=myproject myproject/

See also

See py.test's documentation for How to invoke pytest or invoke py.test -h to see its full set of options.

pcreate template i18n

The pcreate template automatically sets up pot/po/mo locale files for use with the generated project.

The usual pattern for working with i18n in pyramid_jinja2 is as follows:

# make sure Babel is installed
$ $VENV/bin/pip install Babel

# extract translatable strings from *.jinja2 / *.py
$ $VENV/bin/python extract_messages
$ $VENV/bin/python update_catalog

# Translate strings in <mypackage>/locale/<mylocale>/LC_MESSAGES/<myproject>.po
# and re-compile *.po files
$ $VENV/bin/python compile_catalog

If you see the following output:

running compile_catalog
1 of 1 messages (100%) translated in myproject/locale/de/LC_MESSAGES/myproject.po
catalog myproject/locale/de/LC_MESSAGES/myproject.po is marked as fuzzy, skipping
1 of 1 messages (100%) translated in myproject/locale/fr/LC_MESSAGES/myproject.po
catalog myproject/locale/fr/LC_MESSAGES/myproject.po is marked as fuzzy, skipping

When an item is marked as fuzzy, then you should review your .po files to make sure translations are correct. Fuzzy is not exact matching, but matches most of a word (its root) or phrase.

When you are satisfied that the translations are good, you can either remove the line marked with #, fuzzy immediately above its related msgid line (preferred) or force Babel to compile the message catalog with the -f flag.

$ $VENV/bin/python compile_catalog -f

Assuming you have already created a project following the instructions under Creating a Jinja2 Pyramid project, and started your application with pserve, then you should be able to view the various translations. Simply append a GET parameter, such as http://localhost:6543/?_LOCALE_=de for German, http://localhost:6543/?_LOCALE_=fr for French, or http://localhost:6543/?_LOCALE_=en for English. The default language does not require GET parameter.

The application could set the user's language preference with a cookie based on request parameters sent on the first request. Alternatively, and usually as a fallback, the application could read the web browser's Accept-Language header sent with each request and set the appropriate language. For example:

def prepare_env(event):
    request = event.request
    # set locale depending on browser settings
    settings = request.registry.settings
    locale = settings.get("pyramid.default_locale_name", "en")
    available = [loc["code"] for loc in AVAILABLE_LOCALES]
    if request.accept_language:
        accepted = request.accept_language
        locale = accepted.best_match(available, locale)
    request._LOCALE_ = locale

More Information

Reporting Bugs / Development Versions

Visit to download development or tagged versions.

Visit to report bugs.

Indices and tables