Automatically and lazily generate three versions of your pydantic models: one with Extra.forbid, one with Extra.ignore, and one with all fields optional

---

Installation

pip install pydantic-duality

Quickstart

Given the following models:

from pydantic_duality import DualBaseModel


class User(DualBaseModel):
    id: UUID
    name: str

class Auth(DualBaseModel):
    some_field: str
    user: User

Using pydantic-duality is roughly equivalent to making all of the following models by hand:

from pydantic import BaseModel

# Equivalent to User and User.__request__
class UserRequest(BaseModel, extra=Extra.forbid):
    id: UUID
    name: str

# Rougly equivalent to Auth and Auth.__request__
class AuthRequest(BaseModel, extra=Extra.forbid):
    some_field: str
    user: UserRequest


# Rougly equivalent to User.__response__
class UserResponse(BaseModel, extra=Extra.ignore):
    id: UUID
    name: str

# Rougly equivalent to Auth.__response__
class AuthResponse(BaseModel, extra=Extra.ignore):
    some_field: str
    user: UserResponse


# Rougly equivalent to User.__patch_request__
class UserPatchRequest(BaseModel, extra=Extra.forbid):
    id: UUID | None
    name: str | None

# Rougly equivalent to Auth.__patch_request__
class AuthPatchRequest(BaseModel, extra=Extra.forbid):
    some_field: str | None
    user: UserPatchRequest | None

So it takes you up to 3 times less code to write the same thing. Note also that pydantic-duality does everything lazily so you will not notice any significant performance or memory usage difference when using it instead of writing everything by hand. Think of it as using all the customized models as cached properties.

Inheritance, inner models, custom configs, custom names, config kwargs, isinstance and subclass checks work intuitively and in the same manner as they would work if you were not using pydantic-duality.

Help

See documentation for more details