Django includes a contenttypes
application that can
track all of the models installed in your Django-powered project, providing a
high-level, generic interface for working with your models.
At the heart of the contenttypes application is the
ContentType
model, which lives at
django.contrib.contenttypes.models.ContentType
. Instances of
ContentType
represent and store
information about the models installed in your project, and new instances of
ContentType
are automatically
created whenever new models are installed.
Instances of ContentType
have
methods for returning the model classes they represent and for querying objects
from those models. ContentType
also has a custom manager that adds methods for
working with ContentType
and for
obtaining instances of ContentType
for a particular model.
Relations between your models and
ContentType
can also be used to
enable “generic” relationships between an instance of one of your
models and instances of any model you have installed.
The contenttypes framework is included in the default
INSTALLED_APPS
list created by django-admin startproject
,
but if you’ve removed it or if you manually set up your
INSTALLED_APPS
list, you can enable it by adding
'django.contrib.contenttypes'
to your INSTALLED_APPS
setting.
It’s generally a good idea to have the contenttypes framework installed; several of Django’s other bundled applications require it:
The admin application uses it to log the history of each object added or changed through the admin interface.
Django’s authentication framework
uses it
to tie user permissions to specific models.
ContentType
model¶Each instance of ContentType
has two fields which, taken together, uniquely describe an installed
model:
The name of the application the model is part of. This is taken from
the app_label
attribute of the model, and includes only the
last part of the application’s Python import path;
django.contrib.contenttypes
, for example, becomes an
app_label
of contenttypes
.
The name of the model class.
Additionally, the following property is available:
The human-readable name of the content type. This is taken from the
verbose_name
attribute of the model.
Let’s look at an example to see how this works. If you already have
the contenttypes
application installed, and then add
the sites application
to your
INSTALLED_APPS
setting and run manage.py migrate
to install it,
the model django.contrib.sites.models.Site
will be installed into
your database. Along with it a new instance of
ContentType
will be
created with the following values:
ContentType
instances¶Each ContentType
instance has
methods that allow you to get from a
ContentType
instance to the
model it represents, or to retrieve objects from that model:
Takes a set of valid lookup arguments for the
model the ContentType
represents, and does
a get() lookup
on that model, returning the corresponding object.
Returns the model class represented by this
ContentType
instance.
For example, we could look up the
ContentType
for the
User
model:
>>> from django.contrib.contenttypes.models import ContentType
>>> user_type = ContentType.objects.get(app_label="auth", model="user")
>>> user_type
<ContentType: user>
And then use it to query for a particular
User
, or to get access
to the User
model class:
>>> user_type.model_class()
<class 'django.contrib.auth.models.User'>
>>> user_type.get_object_for_this_type(username="Guido")
<User: Guido>
Together,
get_object_for_this_type()
and model_class()
enable
two extremely important use cases:
Using these methods, you can write high-level generic code that
performs queries on any installed model – instead of importing and
using a single specific model class, you can pass an app_label
and
model
into a
ContentType
lookup at
runtime, and then work with the model class or retrieve objects from it.
You can relate another model to
ContentType
as a way of
tying instances of it to particular model classes, and use these methods
to get access to those model classes.
Several of Django’s bundled applications make use of the latter technique.
For example,
the permissions system
in
Django’s authentication framework uses a
Permission
model with a foreign
key to ContentType
; this lets
Permission
represent concepts like
“can add blog entry” or “can delete news story”.
ContentTypeManager
¶ContentType
also has a custom
manager, ContentTypeManager
,
which adds the following methods:
Clears an internal cache used by
ContentType
to keep track
of models for which it has created
ContentType
instances. You
probably won’t ever need to call this method yourself; Django will call
it automatically when it’s needed.
Lookup a ContentType
by ID.
Since this method uses the same shared cache as
get_for_model()
,
it’s preferred to use this method over the usual
ContentType.objects.get(pk=id)
Takes either a model class or an instance of a model, and returns the
ContentType
instance
representing that model. for_concrete_model=False
allows fetching
the ContentType
of a proxy
model.
Takes a variadic number of model classes, and returns a dictionary
mapping the model classes to the
ContentType
instances
representing them. for_concrete_models=False
allows fetching the
ContentType
of proxy
models.
Returns the ContentType
instance uniquely identified by the given application label and model
name. The primary purpose of this method is to allow
ContentType
objects to be
referenced via a natural key
during deserialization.
The get_for_model()
method is especially
useful when you know you need to work with a
ContentType
but don’t
want to go to the trouble of obtaining the model’s metadata to perform a manual
lookup:
>>> from django.contrib.auth.models import User
>>> ContentType.objects.get_for_model(User)
<ContentType: user>
Adding a foreign key from one of your own models to
ContentType
allows your model to
effectively tie itself to another model class, as in the example of the
Permission
model above. But it’s possible
to go one step further and use
ContentType
to enable truly
generic (sometimes called “polymorphic”) relationships between models.
For example, it could be used for a tagging system like so:
from django.contrib.contenttypes.fields import GenericForeignKey
from django.contrib.contenttypes.models import ContentType
from django.db import models
class TaggedItem(models.Model):
tag = models.SlugField()
content_type = models.ForeignKey(ContentType, on_delete=models.CASCADE)
object_id = models.PositiveIntegerField()
content_object = GenericForeignKey("content_type", "object_id")
def __str__(self):
return self.tag
class Meta:
indexes = [
models.Index(fields=["content_type", "object_id"]),
]
A normal ForeignKey
can only “point
to” one other model, which means that if the TaggedItem
model used a
ForeignKey
it would have to
choose one and only one model to store tags for. The contenttypes
application provides a special field type (GenericForeignKey
) which
works around this and allows the relationship to be with any
model:
There are three parts to setting up a
GenericForeignKey
:
Give your model a ForeignKey
to ContentType
. The usual
name for this field is “content_type”.
Give your model a field that can store primary key values from the
models you’ll be relating to. For most models, this means a
PositiveIntegerField
. The usual name
for this field is “object_id”.
Give your model a
GenericForeignKey
, and
pass it the names of the two fields described above. If these fields
are named “content_type” and “object_id”, you can omit this – those
are the default field names
GenericForeignKey
will
look for.
Unlike for the ForeignKey
, a database index is
not automatically created on the
GenericForeignKey
, so it’s
recommended that you use
Meta.indexes
to add your own
multiple column index. This behavior may change in the
future.
If False
, the field will be able to reference proxy models. Default
is True
. This mirrors the for_concrete_model
argument to
get_for_model()
.
Primary key type compatibility
The “object_id” field doesn’t have to be the same type as the
primary key fields on the related models, but their primary key values
must be coercible to the same type as the “object_id” field by its
get_db_prep_value()
method.
For example, if you want to allow generic relations to models with either
IntegerField
or
CharField
primary key fields, you
can use CharField
for the
“object_id” field on your model since integers can be coerced to
strings by get_db_prep_value()
.
For maximum flexibility you can use a
TextField
which doesn’t have a
maximum length defined, however this may incur significant performance
penalties depending on your database backend.
There is no one-size-fits-all solution for which field type is best. You should evaluate the models you expect to be pointing to and determine which solution will be most effective for your use case.
Serializing references to ContentType
objects
If you’re serializing data (for example, when generating
fixtures
) from a model that implements
generic relations, you should probably be using a natural key to uniquely
identify related ContentType
objects. See natural keys and
dumpdata --natural-foreign
for more information.
This will enable an API similar to the one used for a normal
ForeignKey
;
each TaggedItem
will have a content_object
field that returns the
object it’s related to, and you can also assign to that field or use it when
creating a TaggedItem
:
>>> from django.contrib.auth.models import User
>>> guido = User.objects.get(username="Guido")
>>> t = TaggedItem(content_object=guido, tag="bdfl")
>>> t.save()
>>> t.content_object
<User: Guido>
If the related object is deleted, the content_type
and object_id
fields
remain set to their original values and the GenericForeignKey
returns
None
:
>>> guido.delete()
>>> t.content_object # returns None
Due to the way GenericForeignKey
is implemented, you cannot use such fields directly with filters (filter()
and exclude()
, for example) via the database API. Because a
GenericForeignKey
isn’t a
normal field object, these examples will not work:
# This will fail
>>> TaggedItem.objects.filter(content_object=guido)
# This will also fail
>>> TaggedItem.objects.get(content_object=guido)
Likewise, GenericForeignKey
s
does not appear in ModelForm
s.
The relation on the related object back to this object doesn’t exist by
default. Setting related_query_name
creates a relation from the
related object back to this one. This allows querying and filtering
from the related object.
If you know which models you’ll be using most often, you can also add a “reverse” generic relationship to enable an additional API. For example:
from django.contrib.contenttypes.fields import GenericRelation
from django.db import models
class Bookmark(models.Model):
url = models.URLField()
tags = GenericRelation(TaggedItem)
Bookmark
instances will each have a tags
attribute, which can
be used to retrieve their associated TaggedItems
:
>>> b = Bookmark(url="https://www.djangoproject.com/")
>>> b.save()
>>> t1 = TaggedItem(content_object=b, tag="django")
>>> t1.save()
>>> t2 = TaggedItem(content_object=b, tag="python")
>>> t2.save()
>>> b.tags.all()
<QuerySet [<TaggedItem: django>, <TaggedItem: python>]>
You can also use add()
, create()
, or set()
to create
relationships:
>>> t3 = TaggedItem(tag="Web development")
>>> b.tags.add(t3, bulk=False)
>>> b.tags.create(tag="Web framework")
<TaggedItem: Web framework>
>>> b.tags.all()
<QuerySet [<TaggedItem: django>, <TaggedItem: python>, <TaggedItem: Web development>, <TaggedItem: Web framework>]>
>>> b.tags.set([t1, t3])
>>> b.tags.all()
<QuerySet [<TaggedItem: django>, <TaggedItem: Web development>]>
The remove()
call will bulk delete the specified model objects:
>>> b.tags.remove(t3)
>>> b.tags.all()
<QuerySet [<TaggedItem: django>]>
>>> TaggedItem.objects.all()
<QuerySet [<TaggedItem: django>]>
The clear()
method can be used to bulk delete all related objects for an
instance:
>>> b.tags.clear()
>>> b.tags.all()
<QuerySet []>
>>> TaggedItem.objects.all()
<QuerySet []>
Defining GenericRelation
with
related_query_name
set allows querying from the related object:
tags = GenericRelation(TaggedItem, related_query_name="bookmark")
This enables filtering, ordering, and other query operations on Bookmark
from TaggedItem
:
>>> # Get all tags belonging to bookmarks containing `django` in the url
>>> TaggedItem.objects.filter(bookmark__url__contains="django")
<QuerySet [<TaggedItem: django>, <TaggedItem: python>]>
If you don’t add the related_query_name
, you can do the same types of
lookups manually:
>>> bookmarks = Bookmark.objects.filter(url__contains="django")
>>> bookmark_type = ContentType.objects.get_for_model(Bookmark)
>>> TaggedItem.objects.filter(content_type__pk=bookmark_type.id, object_id__in=bookmarks)
<QuerySet [<TaggedItem: django>, <TaggedItem: python>]>
Just as GenericForeignKey
accepts the names of the content-type and object-ID fields as
arguments, so too does
GenericRelation
;
if the model which has the generic foreign key is using non-default names
for those fields, you must pass the names of the fields when setting up a
GenericRelation
to it. For example, if the TaggedItem
model
referred to above used fields named content_type_fk
and
object_primary_key
to create its generic foreign key, then a
GenericRelation
back to it would need to be defined like so:
tags = GenericRelation(
TaggedItem,
content_type_field="content_type_fk",
object_id_field="object_primary_key",
)
Note also, that if you delete an object that has a
GenericRelation
, any objects
which have a GenericForeignKey
pointing at it will be deleted as well. In the example above, this means that
if a Bookmark
object were deleted, any TaggedItem
objects pointing at
it would be deleted at the same time.
Unlike ForeignKey
,
GenericForeignKey
does not accept
an on_delete
argument to customize this
behavior; if desired, you can avoid the cascade-deletion by not using
GenericRelation
, and alternate
behavior can be provided via the pre_delete
signal.
Django’s database aggregation API works with a
GenericRelation
. For example, you
can find out how many tags all the bookmarks have:
>>> Bookmark.objects.aggregate(Count("tags"))
{'tags__count': 3}
The django.contrib.contenttypes.forms
module provides:
A formset factory, generic_inlineformset_factory()
, for use with
GenericForeignKey
.
Returns a GenericInlineFormSet
using
modelformset_factory()
.
You must provide ct_field
and fk_field
if they are different from
the defaults, content_type
and object_id
respectively. Other
parameters are similar to those documented in
modelformset_factory()
and
inlineformset_factory()
.
The for_concrete_model
argument corresponds to the
for_concrete_model
argument on GenericForeignKey
.
The django.contrib.contenttypes.admin
module provides
GenericTabularInline
and
GenericStackedInline
(subclasses of
GenericInlineModelAdmin
)
These classes and functions enable the use of generic relations in forms and the admin. See the model formset and admin documentation for more information.
The GenericInlineModelAdmin
class inherits all properties from an
InlineModelAdmin
class. However,
it adds a couple of its own for working with the generic relation:
The name of the
ContentType
foreign key
field on the model. Defaults to content_type
.
The name of the integer field that represents the ID of the related
object. Defaults to object_id
.
Subclasses of GenericInlineModelAdmin
with stacked and tabular
layouts, respectively.
GenericPrefetch()
¶This lookup is similar to Prefetch()
and it should only be used on
GenericForeignKey
. The querysets
argument accepts a list of querysets,
each for a different ContentType
. This is useful for GenericForeignKey
with non-homogeneous set of results.
>>> from django.contrib.contenttypes.prefetch import GenericPrefetch
>>> bookmark = Bookmark.objects.create(url="https://www.djangoproject.com/")
>>> animal = Animal.objects.create(name="lion", weight=100)
>>> TaggedItem.objects.create(tag="great", content_object=bookmark)
>>> TaggedItem.objects.create(tag="awesome", content_object=animal)
>>> prefetch = GenericPrefetch(
... "content_object", [Bookmark.objects.all(), Animal.objects.only("name")]
... )
>>> TaggedItem.objects.prefetch_related(prefetch).all()
<QuerySet [<TaggedItem: Great>, <TaggedItem: Awesome>]>
Dec 25, 2023