rest-framework-generic-relations

Rest Framework Generic Relations

This library implements Django REST Framework serializers to handle generic foreign keys.

Maintenance status

Originally by Lily Foote. now maintained by Craig de Stigter.

I don't intend to make any major changes to this library, but I will attempt to fix reported bugs and keep it up to date with recent versions of Python / Django / DRF.

Requirements

• At least Python 3.8
• A supported version of Django REST Framework (at least 3.12+)
• A supported version of Django (at least 3.2+)

Installation

Install using pip...

pip install  rest-framework-generic-relations

Add 'generic_relations' to your INSTALLED_APPS setting.

INSTALLED_APPS = (
    ...
    'generic_relations',
)

API Reference

GenericRelatedField

This field serializes generic foreign keys. For a primer on generic foreign keys, first see: https://docs.djangoproject.com/en/dev/ref/contrib/contenttypes/

Let's assume a TaggedItem model which has a generic relationship with other arbitrary models:

class TaggedItem(models.Model):
    tag_name = models.SlugField()
    content_type = models.ForeignKey(ContentType, on_delete=models.CASCADE)
    object_id = models.PositiveIntegerField()
    tagged_object = GenericForeignKey('content_type', 'object_id')

And the following two models, which may have associated tags:

class Bookmark(models.Model):
    """
    A bookmark consists of a URL, and 0 or more descriptive tags.
    """
    url = models.URLField()
    tags = GenericRelation(TaggedItem)

class Note(models.Model):
    """
    A note consists of some text, and 0 or more descriptive tags.
    """
    text = models.CharField(max_length=1000)
    tags = GenericRelation(TaggedItem)

Now we define serializers for each model that may get associated with tags.

class BookmarkSerializer(serializers.ModelSerializer):
    class Meta:
        model = Bookmark
        fields = ('url',)

class NoteSerializer(serializers.ModelSerializer):
    class Meta:
        model = Note
        fields = ('text',)

The model serializer for the TaggedItem model could look like this:

from generic_relations.relations import GenericRelatedField

class TagSerializer(serializers.ModelSerializer):
    """
    A `TaggedItem` serializer with a `GenericRelatedField` mapping all possible
    models to their respective serializers.
    """
    tagged_object = GenericRelatedField({
        Bookmark: BookmarkSerializer(),
        Note: NoteSerializer()
    })

    class Meta:
        model = TaggedItem
        fields = ('tag_name', 'tagged_object')

The JSON representation of a TaggedItem object with name='django' and its generic foreign key pointing at a Bookmark object with url='https://www.djangoproject.com/' would look like this:

{
    "tagged_object": {
        "url": "https://www.djangoproject.com/"
    },
    "tag_name": "django"
}

If you want to have your generic foreign key represented as hyperlink, simply use HyperlinkedRelatedField objects:

class TagSerializer(serializers.ModelSerializer):
    """
    A `Tag` serializer with a `GenericRelatedField` mapping all possible
    models to properly set up `HyperlinkedRelatedField`s.
    """
    tagged_object = GenericRelatedField({
        Bookmark: serializers.HyperlinkedRelatedField(
            queryset = Bookmark.objects.all(),
            view_name='bookmark-detail',
        ),
        Note: serializers.HyperlinkedRelatedField(
            queryset = Note.objects.all(),
            view_name='note-detail',
        ),
    })

    class Meta:
        model = TaggedItem
        fields = ('tag_name', 'tagged_object')

The JSON representation of the same TaggedItem example object could now look something like this:

{
    "tagged_object": "/bookmark/1/",
    "tag_name": "django"
}

Writing to generic foreign keys

The above TagSerializer is also writable. By default, a GenericRelatedField iterates over its nested serializers and returns the value of the first serializer that is actually able to perform to_internal_value() without any errors. Note, that (at the moment) only HyperlinkedRelatedField is able to serialize model objects out of the box.

The following operations would create a TaggedItem object with it's tagged_object property pointing at the Bookmark object found at the given detail end point.

tag_serializer = TagSerializer(data={
    'tag_name': 'python',
    'tagged_object': '/bookmark/1/'
})

tag_serializer.is_valid()
tag_serializer.save()

If you feel that this default behavior doesn't suit your needs, you can subclass GenericRelatedField and override its get_serializer_for_instance or get_deserializer_for_data respectively to implement your own way of decision-making.

GenericModelSerializer

Sometimes you may want to serialize a single list of different top-level things. For instance, suppose I have an API view that returns what items are on my bookshelf. Let's define some models:

from django.core.validators import MaxValueValidator

class Book(models.Model):
    title = models.CharField(max_length=255)
    author = models.CharField(max_length=255)

class Bluray(models.Model):
    title = models.CharField(max_length=255)
    rating = models.PositiveSmallIntegerField(
        validators=[MaxValueValidator(5)],
    )

Then we could have a serializer for each type of object:

class BookSerializer(serializers.ModelSerializer):
    class Meta:
        model = Book
        fields = ('title', 'author')

class BluraySerializer(serializers.ModelSerializer):
    class Meta:
        model = Bluray
        fields = ('title', 'rating')

Now we can create a generic list serializer, which delegates to the above serializers based on the type of model it's serializing:

bookshelf_item_serializer = GenericModelSerializer(
    {
        Book: BookSerializer(),
        Bluray: BluraySerializer(),
    },
    many=True,
)

Then we can serialize a mixed list of items:

>>> bookshelf_item_serializer.to_representation([
    Book.objects.get(title='War and Peace'),
    Bluray.objects.get(title='Die Hard'),
    Bluray.objects.get(title='Shawshank Redemption'),
    Book.objects.get(title='To Kill a Mockingbird'),
])

[
    {'title': 'War and Peace', 'author': 'Leo Tolstoy'},
    {'title': 'Die Hard', 'rating': 5},
    {'title': 'Shawshank Redemption', 'rating': 5},
    {'title': 'To Kill a Mockingbird', 'author': 'Harper Lee'}
]

A few things you should note:

• Although GenericForeignKey fields can be set to any model object, the GenericRelatedField only handles models explicitly defined in its configuration dictionary.
• Reverse generic keys, expressed using the GenericRelation field, can be serialized using the regular relational field types, since the type of the target in the relationship is always known.
• The order in which you register serializers matters as far as write operations are concerned.
• Unless you provide a custom get_deserializer_for_data() method, only HyperlinkedRelatedField provides write access to generic model relations.