================================

Custom template tags and filters

================================

Introduction

============

Django's template system comes with a wide variety of :doc:`built-in

tags and filters </ref/templates/builtins>` designed to address the

presentation logic needs of your application. Nevertheless, you may

find yourself needing functionality that is not covered by the core

set of template primitives. You can extend the template engine by

defining custom tags and filters using Python, and then make them

available to your templates using the ``{% load %}`` tag.

Code layout

-----------

Custom template tags and filters must live inside a Django app. If they relate

to an existing app it makes sense to bundle them there; otherwise, you should

create a new app to hold them.

The app should contain a ``templatetags`` directory, at the same level as

``models.py``, ``views.py``, etc. If this doesn't already exist, create it -

don't forget the ``__init__.py`` file to ensure the directory is treated as a

Python package.

Your custom tags and filters will live in a module inside the ``templatetags``

directory. The name of the module file is the name you'll use to load the tags

later, so be careful to pick a name that won't clash with custom tags and

filters in another app.

For example, if your custom tags/filters are in a file called

``poll_extras.py``, your app layout might look like this::

    polls/

        models.py

        templatetags/

            __init__.py

            poll_extras.py

        views.py

And in your template you would use the following:

.. code-block:: html+django

    {% load poll_extras %}

The app that contains the custom tags must be in :setting:`INSTALLED_APPS` in

order for the ``{% load %}`` tag to work. This is a security feature: It allows

you to host Python code for many template libraries on a single host machine

without enabling access to all of them for every Django installation.

There's no limit on how many modules you put in the ``templatetags`` package.

Just keep in mind that a ``{% load %}`` statement will load tags/filters for

the given Python module name, not the name of the app.

To be a valid tag library, the module must contain a module-level variable

named ``register`` that is a ``template.Library`` instance, in which all the

tags and filters are registered. So, near the top of your module, put the

following::

    from django import template

    register = template.Library()

.. admonition:: Behind the scenes

    For a ton of examples, read the source code for Django's default filters

    and tags. They're in ``django/template/defaultfilters.py`` and

    ``django/template/defaulttags.py``, respectively.

Writing custom template filters

-------------------------------

Custom filters are just Python functions that take one or two arguments:

    * The value of the variable (input) -- not necessarily a string.

    * The value of the argument -- this can have a default value, or be left

      out altogether.

For example, in the filter ``{{ var|foo:"bar" }}``, the filter ``foo`` would be

passed the variable ``var`` and the argument ``"bar"``.

Filter functions should always return something. They shouldn't raise

exceptions. They should fail silently. In case of error, they should return

either the original input or an empty string -- whichever makes more sense.

Here's an example filter definition::

    def cut(value, arg):

        "Removes all values of arg from the given string"

        return value.replace(arg, '')

And here's an example of how that filter would be used:

.. code-block:: html+django

    {{ somevariable|cut:"0" }}

Most filters don't take arguments. In this case, just leave the argument out of

your function. Example::

    def lower(value): # Only one argument.

        "Converts a string into all lowercase"

        return value.lower()

Template filters that expect strings

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

If you're writing a template filter that only expects a string as the first

argument, you should use the decorator ``stringfilter``. This will

convert an object to its string value before being passed to your function::

    from django.template.defaultfilters import stringfilter

    @stringfilter

    def lower(value):

        return value.lower()

This way, you'll be able to pass, say, an integer to this filter, and it

won't cause an ``AttributeError`` (because integers don't have ``lower()``

methods).

Registering custom filters

~~~~~~~~~~~~~~~~~~~~~~~~~~

Once you've written your filter definition, you need to register it with

your ``Library`` instance, to make it available to Django's template language::

    register.filter('cut', cut)

    register.filter('lower', lower)

The ``Library.filter()`` method takes two arguments:

    1. The name of the filter -- a string.

    2. The compilation function -- a Python function (not the name of the

       function as a string).

You can use ``register.filter()`` as a decorator instead::

    @register.filter(name='cut')

    @stringfilter

    def cut(value, arg):

        return value.replace(arg, '')

    @register.filter

    @stringfilter

    def lower(value):

        return value.lower()

If you leave off the ``name`` argument, as in the second example above, Django

will use the function's name as the filter name.

Filters and auto-escaping

~~~~~~~~~~~~~~~~~~~~~~~~~

.. versionadded:: 1.0

When writing a custom filter, give some thought to how the filter will interact

with Django's auto-escaping behavior. Note that three types of strings can be

passed around inside the template code:

    * **Raw strings** are the native Python ``str`` or ``unicode`` types. On

      output, they're escaped if auto-escaping is in effect and presented

      unchanged, otherwise.

    * **Safe strings** are strings that have been marked safe from further

      escaping at output time. Any necessary escaping has already been done.

      They're commonly used for output that contains raw HTML that is intended

      to be interpreted as-is on the client side.

      Internally, these strings are of type ``SafeString`` or ``SafeUnicode``.

      They share a common base class of ``SafeData``, so you can test

      for them using code like::

          if isinstance(value, SafeData):

              # Do something with the "safe" string.

    * **Strings marked as "needing escaping"** are *always* escaped on

      output, regardless of whether they are in an ``autoescape`` block or not.

      These strings are only escaped once, however, even if auto-escaping

      applies.

      Internally, these strings are of type ``EscapeString`` or

      ``EscapeUnicode``. Generally you don't have to worry about these; they

      exist for the implementation of the ``escape`` filter.

Template filter code falls into one of two situations:

    1. Your filter does not introduce any HTML-unsafe characters (``<``, ``>``,

       ``'``, ``"`` or ``&``) into the result that were not already present. In

       this case, you can let Django take care of all the auto-escaping

       handling for you. All you need to do is put the ``is_safe`` attribute on

       your filter function and set it to ``True``, like so::

           @register.filter

           def myfilter(value):

               return value

           myfilter.is_safe = True

       This attribute tells Django that if a "safe" string is passed into your

       filter, the result will still be "safe" and if a non-safe string is

       passed in, Django will automatically escape it, if necessary.

       You can think of this as meaning "this filter is safe -- it doesn't

       introduce any possibility of unsafe HTML."

       The reason ``is_safe`` is necessary is because there are plenty of

       normal string operations that will turn a ``SafeData`` object back into

       a normal ``str`` or ``unicode`` object and, rather than try to catch

       them all, which would be very difficult, Django repairs the damage after

       the filter has completed.

       For example, suppose you have a filter that adds the string ``xx`` to the

       end of any input. Since this introduces no dangerous HTML characters to

       the result (aside from any that were already present), you should mark

       your filter with ``is_safe``::

           @register.filter

           def add_xx(value):

               return '%sxx' % value

           add_xx.is_safe = True

       When this filter is used in a template where auto-escaping is enabled,

       Django will escape the output whenever the input is not already marked as

       "safe".

       By default, ``is_safe`` defaults to ``False``, and you can omit it from

       any filters where it isn't required.

       Be careful when deciding if your filter really does leave safe strings

       as safe. If you're *removing* characters, you might inadvertently leave

       unbalanced HTML tags or entities in the result. For example, removing a

       ``>`` from the input might turn ``<a>`` into ``<a``, which would need to

       be escaped on output to avoid causing problems. Similarly, removing a

       semicolon (``;``) can turn ``&`` into ``&amp``, which is no longer a

       valid entity and thus needs further escaping. Most cases won't be nearly

       this tricky, but keep an eye out for any problems like that when

       reviewing your code.

       Marking a filter ``is_safe`` will coerce the filter's return value to

       a string.  If your filter should return a boolean or other non-string

       value, marking it ``is_safe`` will probably have unintended

       consequences (such as converting a boolean False to the string

       'False').

    2. Alternatively, your filter code can manually take care of any necessary

       escaping. This is necessary when you're introducing new HTML markup into

       the result. You want to mark the output as safe from further

       escaping so that your HTML markup isn't escaped further, so you'll need

       to handle the input yourself.

       To mark the output as a safe string, use

       :func:`django.utils.safestring.mark_safe`.

       Be careful, though. You need to do more than just mark the output as

       safe. You need to ensure it really *is* safe, and what you do depends on

       whether auto-escaping is in effect. The idea is to write filters than

       can operate in templates where auto-escaping is either on or off in

       order to make things easier for your template authors.

       In order for your filter to know the current auto-escaping state, set

       the ``needs_autoescape`` attribute to ``True`` on your function. (If you

       don't specify this attribute, it defaults to ``False``). This attribute

       tells Django that your filter function wants to be passed an extra

       keyword argument, called ``autoescape``, that is ``True`` if

       auto-escaping is in effect and ``False`` otherwise.

       For example, let's write a filter that emphasizes the first character of

       a string::

           from django.utils.html import conditional_escape

           from django.utils.safestring import mark_safe

           def initial_letter_filter(text, autoescape=None):

               first, other = text[0], text[1:]

               if autoescape:

                   esc = conditional_escape

               else:

                   esc = lambda x: x

               result = '<strong>%s</strong>%s' % (esc(first), esc(other))

               return mark_safe(result)

           initial_letter_filter.needs_autoescape = True

       The ``needs_autoescape`` attribute on the filter function and the

       ``autoescape`` keyword argument mean that our function will know whether

       automatic escaping is in effect when the filter is called. We use

       ``autoescape`` to decide whether the input data needs to be passed

       through ``django.utils.html.conditional_escape`` or not. (In the latter

       case, we just use the identity function as the "escape" function.) The

       ``conditional_escape()`` function is like ``escape()`` except it only

       escapes input that is **not** a ``SafeData`` instance. If a ``SafeData``

       instance is passed to ``conditional_escape()``, the data is returned

       unchanged.

       Finally, in the above example, we remember to mark the result as safe

       so that our HTML is inserted directly into the template without further

       escaping.

       There's no need to worry about the ``is_safe`` attribute in this case

       (although including it wouldn't hurt anything). Whenever you manually

       handle the auto-escaping issues and return a safe string, the

       ``is_safe`` attribute won't change anything either way.

Writing custom template tags

----------------------------

Tags are more complex than filters, because tags can do anything.

A quick overview

~~~~~~~~~~~~~~~~

Above, this document explained that the template system works in a two-step

process: compiling and rendering. To define a custom template tag, you specify

how the compilation works and how the rendering works.

When Django compiles a template, it splits the raw template text into

''nodes''. Each node is an instance of ``django.template.Node`` and has

a ``render()`` method. A compiled template is, simply, a list of ``Node``

objects. When you call ``render()`` on a compiled template object, the template

calls ``render()`` on each ``Node`` in its node list, with the given context.

The results are all concatenated together to form the output of the template.

Thus, to define a custom template tag, you specify how the raw template tag is

converted into a ``Node`` (the compilation function), and what the node's

``render()`` method does.

Writing the compilation function

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

For each template tag the template parser encounters, it calls a Python

function with the tag contents and the parser object itself. This function is

responsible for returning a ``Node`` instance based on the contents of the tag.

For example, let's write a template tag, ``{% current_time %}``, that displays

the current date/time, formatted according to a parameter given in the tag, in

`strftime syntax`_. It's a good idea to decide the tag syntax before anything

else. In our case, let's say the tag should be used like this:

.. code-block:: html+django

    <p>The time is {% current_time "%Y-%m-%d %I:%M %p" %}.</p>

.. _`strftime syntax`: http://docs.python.org/library/time.html#time.strftime

The parser for this function should grab the parameter and create a ``Node``

object::

    from django import template

    def do_current_time(parser, token):

        try:

            # split_contents() knows not to split quoted strings.

            tag_name, format_string = token.split_contents()

        except ValueError:

            raise template.TemplateSyntaxError, "%r tag requires a single argument" % token.contents.split()[0]

        if not (format_string[0] == format_string[-1] and format_string[0] in ('"', "'")):

            raise template.TemplateSyntaxError, "%r tag's argument should be in quotes" % tag_name

        return CurrentTimeNode(format_string[1:-1])

Notes:

    * ``parser`` is the template parser object. We don't need it in this

      example.

    * ``token.contents`` is a string of the raw contents of the tag. In our

      example, it's ``'current_time "%Y-%m-%d %I:%M %p"'``.

    * The ``token.split_contents()`` method separates the arguments on spaces

      while keeping quoted strings together. The more straightforward

      ``token.contents.split()`` wouldn't be as robust, as it would naively

      split on *all* spaces, including those within quoted strings. It's a good

      idea to always use ``token.split_contents()``.

    * This function is responsible for raising

      ``django.template.TemplateSyntaxError``, with helpful messages, for

      any syntax error.

    * The ``TemplateSyntaxError`` exceptions use the ``tag_name`` variable.

      Don't hard-code the tag's name in your error messages, because that

      couples the tag's name to your function. ``token.contents.split()[0]``

      will ''always'' be the name of your tag -- even when the tag has no

      arguments.

    * The function returns a ``CurrentTimeNode`` with everything the node needs

      to know about this tag. In this case, it just passes the argument --

      ``"%Y-%m-%d %I:%M %p"``. The leading and trailing quotes from the

      template tag are removed in ``format_string[1:-1]``.

    * The parsing is very low-level. The Django developers have experimented

      with writing small frameworks on top of this parsing system, using

      techniques such as EBNF grammars, but those experiments made the template

      engine too slow. It's low-level because that's fastest.

Writing the renderer

~~~~~~~~~~~~~~~~~~~~

The second step in writing custom tags is to define a ``Node`` subclass that

has a ``render()`` method.

Continuing the above example, we need to define ``CurrentTimeNode``::

    from django import template

    import datetime

    class CurrentTimeNode(template.Node):

        def __init__(self, format_string):

            self.format_string = format_string

        def render(self, context):

            return datetime.datetime.now().strftime(self.format_string)

Notes:

    * ``__init__()`` gets the ``format_string`` from ``do_current_time()``.

      Always pass any options/parameters/arguments to a ``Node`` via its

      ``__init__()``.

    * The ``render()`` method is where the work actually happens.

    * ``render()`` should never raise ``TemplateSyntaxError`` or any other

      exception. It should fail silently, just as template filters should.

Ultimately, this decoupling of compilation and rendering results in an

efficient template system, because a template can render multiple contexts

without having to be parsed multiple times.

Auto-escaping considerations

~~~~~~~~~~~~~~~~~~~~~~~~~~~~

.. versionadded:: 1.0

The output from template tags is **not** automatically run through the

auto-escaping filters. However, there are still a couple of things you should

keep in mind when writing a template tag.

If the ``render()`` function of your template stores the result in a context

variable (rather than returning the result in a string), it should take care

to call ``mark_safe()`` if appropriate. When the variable is ultimately

rendered, it will be affected by the auto-escape setting in effect at the

time, so content that should be safe from further escaping needs to be marked

as such.

Also, if your template tag creates a new context for performing some

sub-rendering, set the auto-escape attribute to the current context's value.

The ``__init__`` method for the ``Context`` class takes a parameter called

``autoescape`` that you can use for this purpose. For example::

    def render(self, context):

        # ...

        new_context = Context({'var': obj}, autoescape=context.autoescape)

        # ... Do something with new_context ...

This is not a very common situation, but it's useful if you're rendering a

template yourself. For example::

    def render(self, context):

        t = template.loader.get_template('small_fragment.html')

        return t.render(Context({'var': obj}, autoescape=context.autoescape))

If we had neglected to pass in the current ``context.autoescape`` value to our

new ``Context`` in this example, the results would have *always* been

automatically escaped, which may not be the desired behavior if the template

tag is used inside a ``{% autoescape off %}`` block.

.. _template_tag_thread_safety:

Thread-safety considerations

~~~~~~~~~~~~~~~~~~~~~~~~~~~~

.. versionadded:: 1.2

Once a node is parsed, its ``render`` method may be called any number of times.

Since Django is sometimes run in multi-threaded environments, a single node may

be simultaneously rendering with different contexts in response to two separate

requests. Therefore, it's important to make sure your template tags are thread

safe.

To make sure your template tags are thread safe, you should never store state

information on the node itself. For example, Django provides a builtin ``cycle``

template tag that cycles among a list of given strings each time it's rendered::

    {% for o in some_list %}

        <tr class="{% cycle 'row1' 'row2' %}>

            ...

        </tr>

    {% endfor %}

A naive implementation of ``CycleNode`` might look something like this::

    class CycleNode(Node):

        def __init__(self, cyclevars):

            self.cycle_iter = itertools.cycle(cyclevars)

        def render(self, context):

            return self.cycle_iter.next()

But, suppose we have two templates rendering the template snippet from above at

the same time:

    1. Thread 1 performs its first loop iteration, ``CycleNode.render()``

       returns 'row1'

    2. Thread 2 performs its first loop iteration, ``CycleNode.render()``

       returns 'row2'

    3. Thread 1 performs its second loop iteration, ``CycleNode.render()``

       returns 'row1'

    4. Thread 2 performs its second loop iteration, ``CycleNode.render()``

       returns 'row2'

The CycleNode is iterating, but it's iterating globally. As far as Thread 1

and Thread 2 are concerned, it's always returning the same value. This is

obviously not what we want!

To address this problem, Django provides a ``render_context`` that's associated

with the ``context`` of the template that is currently being rendered. The

``render_context`` behaves like a Python dictionary, and should be used to store

``Node`` state between invocations of the ``render`` method.

Let's refactor our ``CycleNode`` implementation to use the ``render_context``::

    class CycleNode(Node):

        def __init__(self, cyclevars):

            self.cyclevars = cyclevars

        def render(self, context):

            if self not in context.render_context:

                context.render_context[self] = itertools.cycle(self.cyclevars)

            cycle_iter = context.render_context[self]

            return cycle_iter.next()

Note that it's perfectly safe to store global information that will not change

throughout the life of the ``Node`` as an attribute. In the case of

``CycleNode``, the ``cyclevars`` argument doesn't change after the ``Node`` is