django-filter-autotools

Installation

Either using pip

python -m pip install django-filter-autotools

or directly copying the file django_filters_autotools/mixins.py into your project.

DefaultLookupsMixin

By default django-filter FilterSet class will only generate exact lookups if given a list of fields. This mixin can be used to customize this behaviour:

class MyFilterSet(DefaultLookupsMixin, FilterSet):

    DEFAULT_LOOKUPS = {
        models.CharField:      [ 'exact', 'icontains' ],
        models.IntegerField:   [ 'exact', 'lt', 'lte', 'gt', 'gte', 'range' ]
    }

MyFilterSet will automatically generate filters for all these lookups for any field included in fields if fields is an array. In the case that fields is a dictionary, defaults will be generated for any key having None as value.

PseudoLookupsMixin

We can use this mixin to patch the FilterSet filter creation algorithm to understand some lookups which are not registered into Django. Two common scenarios are checking whether a string is empty, adding "exclusion" filters (not) and adding "conjoined" filters (and for m2m):

class EmptyStringFilter(filters.BooleanFilter):
    def filter(self, qs, value):
        if value in EMPTY_VALUES:
            return qs

        exclude = self.exclude ^ (value is False)
        method = qs.exclude if exclude else qs.filter

        return method(**{self.field_name: ""})


class MyFilterSet(PseudoLookupsMixin, FilterSet):

    PSEUDO_LOOKUPS = { 
        'isempty': {
            'behaves_like': 'isnull',
            'filter_class': EmptyStringFilter,
        },
        'not_icontains': {
            'behaves_like': 'icontains',
            'filter_class': None,
            'replace_with': 'icontains',
            'extra': lambda f: {'exclude': True}
        },
        'exact_all': {
            'behaves_like': 'exact',
            'filter_class': None,
            'replace_with': 'exact',
            'extra': lambda f: {'conjoined': True}
        }
    }

Under the hood this works by replacing the lookup with the one specified in the behaves_like key, running the filter creation algorithm and then patching the filters if necessary. This way any transformer or any extra lookups registered into Django will also be available.

PSEUDO_LOOKUPS is a dictionary whose keys are the new lookups to be supported and the values are dictionary with the following keys:

behaves_like: should be a lookup registered in Django with similar semantics.
filter_class: filter class to be used for this lookup. If not specified or set to None, the one chosen django-filter for the behaves_like lookup will be used. Note that these may be fine-tuned by overriding the FILTER_FOR_DBFIELD_DEFAULTS dictionary.
replace_with: if present the filter class is patched so that its __init__ method replaces the new lookup with this value.
extra: if present the object returned from applying this function to the field will be merged into the filter class kwargs.

Override the filter_for_pseudolookup class method on the FilterSet if you need to choose different filter classes depending on both the field type and the lookup.

Integrating everything into DRF by default

Complete example making available several lookups by default for all fields of type CharField, and conjoined filters for all fields of type ManyToManyField:

filters.py:

from django_filters import rest_framework as filters
from django.db import models
from django.core.validators import EMPTY_VALUES
from django_filters_autotools.mixins import *


class EmptyStringFilter(filters.BooleanFilter):
    def filter(self, qs, value):
        if value in EMPTY_VALUES:
            return qs

        exclude = self.exclude ^ (value is False)
        method = qs.exclude if exclude else qs.filter

        return method(**{self.field_name: ""})


class MyFilterSet(DefaultLookupsMixin, PseudoLookupsMixin, filters.FilterSet):

    PSEUDO_LOOKUPS = { 
        'isempty': {
            'behaves_like': 'isnull',
            'filter_class': EmptyStringFilter,
        },
        'not_icontains': {
            'behaves_like': 'icontains',
            'filter_class': None,
            'replace_with': 'icontains',
            'extra': lambda f: {'exclude': True}
        },
        'exact_all': {
            'behaves_like': 'exact',
            'filter_class': None,
            'replace_with': 'exact',
            'extra': lambda f: {'conjoined': True}
        }
    }

    DEFAULT_LOOKUPS = {
        models.CharField:        [ 'exact', 'icontains', 'isempty', 'not_icontains' ],
        models.ManyToManyField:  [ 'exact', 'exact_all' ],
    }

    
class MyFilterBackend(filters.DjangoFilterBackend):
    filterset_base = MyFilterSet

settings.py:

REST_FRAMEWORK = {
    'DEFAULT_FILTER_BACKENDS': [
        'path.to.filters.MyFilterBackend',
    ],
}

More info

PyPI project