django-filtering 0.4.0

Allows users to filter Django models

Django Filtering

A library for filtering Django Models.

The original usecase for this project required the following:

• provides a means of allowing users to filter modeled data
• provides the ability to group filters by AND, OR and NOT operators
• serializes, validates, etc.

A user interface (UI) is available for this package in the django-filtering-ui package.

State of development

This package is very much a work-in-progress. All APIs are completely unstable.

Installation

Install via pip or the preferred package manager:

pip install django-filtering

Add to the Django project's INSTALLED_APPS:

INSTALLED_APPS = [
    # ...
    'django_filtering',
    # ...
]

Usage

Say you have a Post model that you want users to be able to filter. We'd start by creating a FilterSet.

import django_filtering as filtering

class PostFilterSet(filtering.FilterSet):
    title = filtering.Filter(
        filtering.InputLookup('icontains', label="contains"),
        label="Title",
    )
    author = filtering.Filter(
        filtering.InputLookup('fullname__iexact', label="fullname is"),
        filtering.InputLookup('email__iexact', label="email is"),
        label="Author",
    )
    content = filtering.Filter(
        filtering.InputLookup('icontains', label="contains"),
        label="Content",
    )

    class Meta:
        model = Post

This can also be expressed using the declarative style:

class PostFilterSet(filtering.FilterSet):
    class Meta:
        model = Post
        fields = {
            'title': ['icontains'],
            'author': ['fullname__iexact', 'email__iexact'],
            'content': ['icontains'],
        }

Note, this package does not come with an interface for user filtering. The django-filtering-ui package does provide an interface.

The filters can be posted in a Form. For example, we'll say we have a form that has a single q JSON field.

q = [
    'and',
    [
        ['title', {'lookup': 'icontains', 'value': 'foo'}],
        ['content', {'lookup': 'icontains', 'value': 'bar'}],
    ]
]

The basic structure is an array with an operator and array of further criteria, where that can be a filter array or another operator grouping.

An example of a user posting filters could look like the following url:

/posts/?q=["and",[["title",{"lookup":"icontains","value":"foo"}],["content",{"lookup":"icontains","value":"bar"}]]

In this case we have a q query string value with JSON content. This query data structure is documented in more detail later in this document.

Let's say this url is a listing view for Post objects, something that looks like:

def posts_list(request):
    query_data = json.loads(request.GET.get('q', '[]'))
    filterset = PostFilterSet(query_data)
    queryset = filterset.filter_queryset()
    return HttpResponse('\n'.join([o.get_absolute_url() for o in queryset]))

In this example view we use the PostFilterSet with the query string value. We get the fitlered results by calling the <FilterSet>.filter_queryset method.

About the query data structure

The JSON serialiable query data is a loosely lisp'ish data structure that looks something like:

query-data := [<operator>, [<filter|operator>,...]]
operator := 'and' | 'or' | 'not' | 'xor'
filter := [<field-name>, {"lookup": <lookup>, "value": <value>}]
field-name := string
lookup := string
value := any

Note, the value can be of any JSON serialiable type.

Testing

Note, I'm testing within a docker container, because I never run anything locally. For the moment the container is simply run with:

docker run --rm --name django-filtering --workdir /code -v $PWD:/code -d python:3.12 sleep infinity

Then I execute commands on the shell within it:

docker exec django-filtering pip install -e '.[tests]'
docker exec -it django-filtering bash

Within the container's shell you can now execute pytest.

License

GPL v3 (see LICENSE file)

Copyright

© 2025 The Shadowserver Foundation