Fields

While most built-in fields work out of the box, it’s recommended to use the fields from cratedb_django.fields. All django fields are available in cratedb_django.fields. These fields enhance compatibility and provide unique CrateDB features.

Shared functionality

All fields share functionalities regarding CrateDB column definitions.

• db_index=False disable index creation.

• column_store=False disable columnstore, not recommended in versions >=5.10.

Example:

from cratedb_django.fields import TextField
TextField(db_index=False, column_store=False)

CrateDB-specific fields

Fields that have a different implementation than their native counterparts.

AutoUUIDField

This field automatically generates an elasticflake (unique id), it’s recommended to be used as a primary key. It is the recommended AUTO field for the id column.

The value is generated by the database and returned to Django upon creation.

class UniquePerson(CrateDBModel):
    name = fields.TextField()
    uid = fields.AutoUUIDField(primary_key=True)
    class Meta:
        app_label = "test_app"
        
person = UniquePerson.objects.create(name='edwards')
print(person.uid)
# S2EsXJsBYKmOR0jb5NL5

GeneratedField

Generated fields are fields where the value is computed in the database, virtual or db_persist are ignored, every value is always persisted to the database.

class UniquePerson(CrateDBModel):
    name = fields.TextField()
    user_uid = fields.GeneratedField(
        # concat("name", gen_random_text_uuid())
        expression=RawSQL("concat(%s, %s)", [F("name"), UUID()]),
        output_field=fields.TextField()
    )
    huge_cardinality = fields.IntegerField()
    partition_value = fields.GeneratedField(expression=F("some_value") % 10)
    last_modified = fields.GeneratedField(functions.CURRENT_TIMESTAMP)

    class Meta:
        app_label = "test_app"

ObjectField

Objects support three different column policies, strict, dynamic and ignored. The default is dynamic.

If the policy is strict, a schema needs to be provided, composed by a dictionary of fields describing the desired data model for the object. The schema can be nested.

class SomeModel(CrateDBModel):
    f = fields.ObjectField()
    f1 = fields.ObjectField(policy="ignored")
    
    # Translates to: f2 OBJECT(strict) as (name varchar,obj OBJECT(strict) as (age integer)) NOT NULL
    f2 = fields.ObjectField(
        policy="strict",
        schema={
            "name": CharField(max_length=None),
            "obj": fields.ObjectField(
                policy="strict", schema={"age": fields.IntegerField()}
            ),
        },
    )

    class Meta:
        app_label = "_crate_test"

ArrayField

Array fields can only be of one type and they can be arbitrarily nested.

class SomeModel(CrateDBModel):
    f1 = fields.ArrayField(fields.IntegerField())
    f2 = fields.ArrayField(
        fields.ArrayField(fields.CharField(max_length=120))
    )
    f3 = fields.ArrayField(fields.ArrayField(fields.ObjectField()))

    class Meta:
        app_label = "_crate_test"