pydantic-jsonschema 0.0.13

JSON Schema to Pydantic model conversion library

Pydantic JSON Schema

Pydantic turns models into JSON Schema. This library does the reverse — it turns a JSON Schema into a Pydantic model, so you can validate data against a schema you already have, with all of Pydantic's runtime checks and editor support.

Reach for it when the schema comes first: API contracts, config files, tool definitions, or validating LLM output against a fixed shape.

Installation

uv add pydantic-jsonschema

Requires Python 3.12+. See the installation guide for optional validator libraries.

Quick start

from pydantic_jsonschema import Schema, to_model

schema = Schema.model_validate({
    "type": "object",
    "properties": {
        "name": {"type": "string"},
        "age": {"type": "integer", "minimum": 0},
    },
    "required": ["name"],
})

User = to_model(schema, model_name="User")

user = User(name="Alice", age=30)
print(user.model_dump())
#> {'name': 'Alice', 'age': 30}

What's inside

Three building blocks — and a fourth on the way.

1. The schema model — Schema & Reference

A Pydantic model for JSON Schema itself: parse, inspect, and serialize schemas with full type safety. $refs are parsed as Reference objects and resolved during conversion.

from pydantic_jsonschema import DataType, Schema

schema = Schema(
    type=DataType.OBJECT,
    properties={"name": Schema(type=DataType.STRING)},
    required=["name"],
)

print(schema.model_dump_json())
#> {"type":"object","properties":{"name":{"type":"string"}},"required":["name"]}

2. The converter — to_model / SchemaConverter

Turns a Schema into a Pydantic model (see Quick start). It resolves $ref / $defs, maps anyOf / oneOf (including discriminated unions) / allOf to Python types, and applies string, number, array, and object constraints as Pydantic validation.

3. Formats

Built-in types for every format in the JSON Schema spec (email, uri, uuid, date-time, hostname, json-pointer, regex, and more) — with zero extra dependencies. Map your own Pydantic type for custom formats.

from pydantic_jsonschema import Schema, to_model
from pydantic_jsonschema.formats import Email

schema = Schema.model_validate({
    "type": "object",
    "properties": {"email": {"type": "string", "format": "email"}},
    "required": ["email"],
})

User = to_model(schema, formats={"email": Email})

print(User(email="[e-mail]").email)
#> [e-mail]

Planned: configurable loading by type

Per-object control over how each type is loaded, inspired by adaptix.

Documentation

https://danipulok.github.io/pydantic-jsonschema/

Acknowledgments

• vgavro — initial library idea and early guidance
• bodlan — early documentation review and feedback

License

MIT License - see LICENSE for details.