Caching related objects in serializer context.

[dharani7998]

When using the serializer with a list of items, for each item the related object is fetched from the database even though the related instance is the same in multiple or all the items in the list.

If the related instances can be cached in the serializer context, it would save a lot of DB calls and save a lot of time when trying to load 1000s of items in one shot.

############## Models ##############
class TypeGroup(models.Model):
    name = CaseCharField(max_length=150,  unique=True)


class Type(models.Model):
    name = CaseCharField(max_length=150, titlecase=True, unique=True)
    type_group = models.ForeignKey(TypeGroup, on_delete=models.CASCADE, related_name="types")


############## Serializers ##############
class TypeGroupSerializer(ModelSerializer):
    class Meta:
        model = TypeGroup
        fields = ALL_FIELDS


class TypeSerializer(ModelSerializer):
    class Meta:
        model = Type
        fields = ALL_FIELDS


#######################################

data = [
    {"name": "type 1", "type_group": 1},
    {"name": "type 2", "type_group": 1},
    {"name": "type 3", "type_group": 1},
]

ser = TypeSerializer(data=data, many=True)
ser.is_valid()

Queries executed for this operation
(0.016) SELECT 1 AS "a" FROM "products_type" WHERE "products_type"."name" = 'Type 1' LIMIT 1; args=(1, 'Type 1'); alias=default
(0.000) SELECT "products_typegroup"."id", "products_typegroup"."created_at", "products_typegroup"."updated_at", "products_typegroup"."is_active", "products_typegroup"."created_by_id", "products_typegroup"."updated_by_id", "products_typegroup"."name" FROM "products_typegroup" WHERE "products_typegroup"."id" = 1 LIMIT 21; args=(1,); alias=default
(0.000) SELECT 1 AS "a" FROM "products_type" WHERE "products_type"."name" = 'Type 2' LIMIT 1; args=(1, 'Type 2'); alias=default
(0.000) SELECT "products_typegroup"."id", "products_typegroup"."created_at", "products_typegroup"."updated_at", "products_typegroup"."is_active", "products_typegroup"."created_by_id", "products_typegroup"."updated_by_id", "products_typegroup"."name" FROM "products_typegroup" WHERE "products_typegroup"."id" = 1 LIMIT 21; args=(1,); alias=default
(0.000) SELECT 1 AS "a" FROM "products_type" WHERE "products_type"."name" = 'Type 3' LIMIT 1; args=(1, 'Type 3'); alias=default
(0.000) SELECT "products_typegroup"."id", "products_typegroup"."created_at", "products_typegroup"."updated_at", "products_typegroup"."is_active", "products_typegroup"."created_by_id", "products_typegroup"."updated_by_id", "products_typegroup"."name" FROM "products_typegroup" WHERE "products_typegroup"."id" = 1 LIMIT 21; args=(1,); alias=default

As we can see here for each item in the list, a new type_group instance is fetched and created. This can be avoided if the related instances are cached in the serializer context.

[auvipy]

can you please share code snippets or examples please??

[dharani7998]

Done

[dharani7998]

@auvipy any thoughts on this?

[auvipy]

I'm going to check