dataclasses-jsonschema 2.5.0

JSON schema generation from dataclasses

A library to generate JSON Schema from python 3.7 dataclasses. Python 3.6 is supported through the dataclasses backport. Aims to be a more lightweight alternative to similar projects such as marshmallow & pydantic.

Feature Overview

• Support for draft-04, draft-06, Swagger 2.0 & OpenAPI 3 schema types

• Serialisation and deserialisation

• Data validation against the generated schema

• APISpec support. Example below:

Installation

~$ pip install dataclasses-jsonschema

For improved validation performance using fastjsonschema, install with:

~$ pip install dataclasses-jsonschema[fast-validation]

Examples

from dataclasses import dataclass

from dataclasses_jsonschema import JsonSchemaMixin


@dataclass
class Point(JsonSchemaMixin):
    "A 2D point"
    x: float
    y: float

Generate the schema:

>>> pprint(Point.json_schema())
{
    'description': 'A 2D point',
    'type': 'object',
    'properties': {
        'x': {'format': 'float', 'type': 'number'},
        'y': {'format': 'float', 'type': 'number'}
    },
    'required': ['x', 'y']
}

Serialise data:

>>> Point(x=3.5, y=10.1).to_dict()
{'x': 3.5, 'y': 10.1}

Deserialise data:

>>> Point.from_dict({'x': 3.14, 'y': 1.5})
Point(x=3.14, y=1.5)
>>> Point.from_dict({'x': 3.14, y: 'wrong'})
dataclasses_jsonschema.ValidationError: 'wrong' is not of type 'number'

Generate a schema for embedding into an API spec:

from dataclasses_jsonschema import JsonSchemaMixin, SchemaType

@dataclass
class Address(JsonSchemaMixin):
    """Postal Address"""
    building: str
    street: str
    city: str

@dataclass
class Company(JsonSchemaMixin):
    """Company Details"""
    name: str
    address: Address

>>> pprint(JsonSchemaMixin.all_json_schemas(schema_type=SchemaType.SWAGGER_V3))
{'Address': {'description': 'Postal Address',
             'properties': {'building': {'type': 'string'},
                            'city': {'type': 'string'},
                            'street': {'type': 'string'}},
             'required': ['building', 'street', 'city'],
             'type': 'object'},
 'Company': {'description': 'Company Details',
             'properties': {'address': {'$ref': '#/components/schemas/Address'},
                            'name': {'type': 'string'}},
             'required': ['name', 'address'],
             'type': 'object'}}

Custom validation rules can be added using NewType:

from dataclasses_jsonschema import JsonSchemaMixin, FieldEncoder

PhoneNumber = NewType('PhoneNumber', str)

class PhoneNumberField(FieldEncoder):

    @property
    def json_schema(self):
        return {'type': 'string', 'pattern': r'^(\([0-9]{3}\))?[0-9]{3}-[0-9]{4}$'}

JsonSchemaMixin.register_field_encoders({PhoneNumber: PhoneNumberField()})

@dataclass
class Person(JsonSchemaMixin):
    name: str
    phone_number: PhoneNumber

For more examples see the tests

APISpec Plugin

New in v2.5.0

OpenAPI & Swagger specs can be generated using the apispec plugin:

from typing import Optional, List
from dataclasses import dataclass

from apispec import APISpec
from apispec_webframeworks.flask import FlaskPlugin
from flask import Flask, jsonify
import pytest

from dataclasses_jsonschema.apispec import DataclassesPlugin
from dataclasses_jsonschema import JsonSchemaMixin


# Create an APISpec
spec = APISpec(
    title="Swagger Petstore",
    version="1.0.0",
    openapi_version="3.0.2",
    plugins=[FlaskPlugin(), DataclassesPlugin()],
)


@dataclass
class Category(JsonSchemaMixin):
    """Pet category"""
    name: str
    id: Optional[int]

@dataclass
class Pet(JsonSchemaMixin):
    """A pet"""
    categories: List[Category]
    name: str


app = Flask(__name__)


@app.route("/random")
def random_pet():
    """A cute furry animal endpoint.
    ---
    get:
      description: Get a random pet
      responses:
        200:
          content:
            application/json:
              schema: Pet
    """
    pet = get_random_pet()
    return jsonify(pet.to_dict())

# Dependant schemas (e.g. 'Category') are added automatically
spec.components.schema("Pet", schema=Pet)
with app.test_request_context():
    spec.path(view=random_pet)

TODO

• Add benchmarks against alternatives such as pydantic and marshmallow