jonathan-s/djantic2

Pydantic model support for Django

Djantic2

Pydantic model support for Django

---

Djantic2 is a fork of djantic which works with pydantic >2, it is a library that provides a configurable utility class for automatically creating a Pydantic model instance for any Django model class. It is intended to support all of the underlying Pydantic model functionality such as JSON schema generation and introduces custom behaviour for exporting Django model instance data.

Quickstart

Install using pip:

pip install djantic2

Create a model schema:

from users.models import User
from pydantic import ConfigDict

from djantic import ModelSchema

class UserSchema(ModelSchema):
    model_config = ConfigDict(model=User, include=["id", "first_name"])

print(UserSchema.schema())

Output:

{
    "description": "A user of the application.",
    "properties": {
        "id": {
            "anyOf": [{"type": "integer"}, {"type": "null"}],
            "default": None,
            "description": "id",
            "title": "Id",
        },
        "first_name": {
            "description": "first_name",
            "maxLength": 50,
            "title": "First Name",
            "type": "string",
        },
    },
    "required": ["first_name"],
    "title": "UserSchema",
    "type": "object",
}

See https://pydantic-docs.helpmanual.io/usage/models/ for more.

Loading and exporting model instances

Use the from_orm method on the model schema to load a Django model instance for export:

user = User.objects.create(
    first_name="Jordan",
    last_name="Eremieff",
    email="jordan@eremieff.com"
)

user_schema = UserSchema.from_orm(user)
print(user_schema.json(indent=2))

Output:

{
    "profile": null,
    "id": 1,
    "first_name": "Jordan",
    "last_name": "Eremieff",
    "email": "jordan@eremieff.com",
    "created_at": "2020-08-15T16:50:30.606345+00:00",
    "updated_at": "2020-08-15T16:50:30.606452+00:00"
}

Using multiple level relations

Djantic supports multiple level relations. This includes foreign keys, many-to-many, and one-to-one relationships.

Consider the following example Django model and Djantic model schema definitions for a number of related database records:

# models.py
from django.db import models

class OrderUser(models.Model):
    email = models.EmailField(unique=True)


class OrderUserProfile(models.Model):
    address = models.CharField(max_length=255)
    user = models.OneToOneField(OrderUser, on_delete=models.CASCADE, related_name='profile')


class Order(models.Model):
    total_price = models.DecimalField(max_digits=8, decimal_places=5, default=0)
    user = models.ForeignKey(
        OrderUser, on_delete=models.CASCADE, related_name="orders"
    )


class OrderItem(models.Model):
    price = models.DecimalField(max_digits=8, decimal_places=5, default=0)
    quantity = models.IntegerField(default=0)
    order = models.ForeignKey(
        Order, on_delete=models.CASCADE, related_name="items"
    )


class OrderItemDetail(models.Model):
    name = models.CharField(max_length=30)
    order_item = models.ForeignKey(
        OrderItem, on_delete=models.CASCADE, related_name="details"
    )

# schemas.py
from djantic import ModelSchema
from pydantic import ConfigDict

from orders.models import OrderItemDetail, OrderItem, Order, OrderUserProfile


class OrderItemDetailSchema(ModelSchema):
    model_config = ConfigDict(model=OrderItemDetail)


class OrderItemSchema(ModelSchema):
    details: List[OrderItemDetailSchema]
    model_config = ConfigDict(model=OrderItem)


class OrderSchema(ModelSchema):
    items: List[OrderItemSchema]
    model_config = ConfigDict(model=Order)


class OrderUserProfileSchema(ModelSchema):
    model_config = ConfigDict(model=OrderUserProfile)


class OrderUserSchema(ModelSchema):
    orders: List[OrderSchema]
    profile: OrderUserProfileSchema
    model_config = ConfigDict(model=OrderUser)

Now let's assume you're interested in exporting the order and profile information for a particular user into a JSON format that contains the details accross all of the related item objects:

user = OrderUser.objects.first()
print(OrderUserSchema.from_orm(user).json(ident=4))

Output:

{
    "profile": {
        "id": 1,
        "address": "",
        "user": 1
    },
    "orders": [
        {
            "items": [
                {
                    "details": [
                        {
                            "id": 1,
                            "name": "",
                            "order_item": 1
                        }
                    ],
                    "id": 1,
                    "price": 0.0,
                    "quantity": 0,
                    "order": 1
                }
            ],
            "id": 1,
            "total_price": 0.0,
            "user": 1
        }
    ],
    "id": 1,
    "email": ""
}

The model schema definitions are composable and support customization of the output according to the auto-generated fields and any additional annotations.

Including and excluding fields

The fields exposed in the model instance may be configured using two options: include and exclude. These represent iterables that should contain a list of field name strings. Only one of these options may be set at the same time, and if neither are set then the default behaviour is to include all of the fields from the Django model.

For example, to include all of the fields from a user model except a field named email_address, you would use the exclude option:

from pydantic import ConfigDict

class UserSchema(ModelSchema):
    model_config = ConfigDict(model=User, exclude=["email_address"])

In addition to this, you may also limit the fields to only include annotations from the model schema class by setting the include option to a special string value: "__annotations__".

from pydantic import ConfigDict

class ProfileSchema(ModelSchema):
        website: str
        model_config = ConfigDict(model=Profile, include="__annotations__")


    assert ProfileSchema.schema() == {
        "title": "ProfileSchema",
        "description": "A user's profile.",
        "type": "object",
        "properties": {
            "website": {
                "title": "Website",
                "type": "string"
            }
        },
        "required": [
            "website"
        ]
    }