Skip to content

gersmann/django-mongo-backend

Repository files navigation

Django backend for MongoDB

Django backend for MongoDB.

Supports:

  • Column mappings to MongoDB documents
  • Single table (collection) inheritance and single table OneToOne relationships
  • Filters (filter/exclude)

Setup / Configuration

Not supported as primary database, as Django contrib apps rely on Integer primary keys in built in migrations (and because it is a use case that is not a priority at the moment).

# settings.py
DATABASES = {
    # or any other primary databse
    "default": {
        "ENGINE": "django.db.backends.sqlite3",
        "NAME": BASE_DIR / "db.sqlite3",
    },
    # mongodb database, Client constructor options are passe in 'CLIENT', the database name in 'NAME'
    "mongodb": {
        "ENGINE": "django_mongodb",
        "NAME": "django_mongodb",
        "CONN_MAX_AGE": 120,
        "CLIENT": {
            "host": os.environ.get("MONGODB_URL"),
        },
    },
}
# A database is required
DATABASE_ROUTERS = ["testproject.router.DatabaseRouter"]

Using the database in models requires a DatabaseRouter, which could look like this

class DatabaseRouter:
    def db_for_read(self, model, **hints):
        if model._meta.app_label == "mymongoapp":
            return "default"
        return "default"

    def db_for_write(self, model, **hints):
        if model._meta.app_label == "mymongoapp":
            return "default"
        return "default"

    def allow_relation(self, obj1, obj2, **hints):
        if obj1._meta.app_label == obj2._meta.app_label:
            return True
        return None

    def allow_migrate(self, db, app_label, model_name=None, **hints):
        if app_label == "mymongoapp":
            # we are disabling migrations, as MongoDB is schema-less. Alerts, such as renaming fields, etc. are not supported
            return False
        return None

Finally we are going to change the default primary key of the app using MongoDB (if that is the case, otherwise add ObjectIdAutoField to the models, where you need it).

# apps.py
class TestappConfig(AppConfig):
    default_auto_field = "django_mongodb.models.ObjectIdAutoField"
    name = "mymongoapp"

Defining Models

A simple model, in an app, which has ObjectIdAutoField as default_auto_field

class MyModel(models.Model):
    json_field = JSONField()
    name = models.CharField(max_length=100)
    datetime_field = models.DateTimeField(auto_now_add=True)
    time_field = models.TimeField(auto_now_add=True)
    date_field = models.DateField(auto_now_add=True)

Single table inheritance

class SameTableChild(MyModel):
    my_model_ptr = models.OneToOneField(
        MyModel,
        on_delete=models.CASCADE,
        parent_link=True,
        related_name="same_table_child",
        # pointer to the primary key of the parent model
        db_column="_id",
    )
    extended = models.CharField(max_length=100)

    class Meta:
        # We are using the parent collection as db_table
        db_table = "mymongoapp_mymodel"

Single table OneToOne relationships

class SameTableOneToOne(models.Model):
    dummy_model = models.OneToOneField(
        MyModel,
        primary_key=True,
        on_delete=models.CASCADE,
        related_name="extends",
        db_column="_id",
    )
    extra = models.CharField(max_length=100)

    class Meta:
        # we are using the same collection to persist one-to-one relationships
        db_table = "mymongoapp_mymodel"

Querying

# get all objects
MyModel.objects.all()

# get all objects, which have a name in list ["foo", "bar"]
MyModel.objects.filter(name_in=["foo", "bar"])

# select related with single table inheritance and one to one relationships
MyModel.objects.select_related("same_table_child", "extends").all()

# simple aggregations
MyModel.objects.filter(name_in=["foo", "bar"]).count()

Search

Using the prefer_search() extension of MongoQueryset, we can use the $search operator of MongoDB to query, if we have search indexes configured on the model.

MyModel.objects.prefer_search().filter(name="foo").all()

PostgreSQL search vectors map down to MongoDB search indexes, so we can use the same syntax as with PostgreSQL.

class MyModel(models.Model):
    name = models.CharField(max_length=100)
MyModel.objects.annotate(search=SearchVector('name')).filter(search=SearchQuery('foo')).all()