Using Pydantic with typing.Protocol

Question

I'm attempting to use typing protocols and pydantic together and am running into some issues. I expect probably my understanding of how to properly use the two of them together is lacking.

This pydantic code is producing an error:

from typing import TypeVar, Protocol, Generic, runtime_checkable
from pydantic import BaseModel

D = TypeVar('D')


@runtime_checkable
class SupportsDimensionality(Protocol[D]):
    dimensionality: D


class ValueModel(BaseModel, Generic[D]):
    value: SupportsDimensionality[D]

Error text:

pydantic.errors.PydanticSchemaGenerationError: Unable to generate pydantic-core schema for main.SupportsDimensionality[~D]. Set arbitrary_types_allowed=True in the model_config to ignore this error or implement __get_pydantic_core_schema__ on your type to fully support it.

I've been trying to figure out why this happens and how to fix it but need some help.

Answer 1

I figured it out. The docs say here you can use the arbitrary_types_allowed=True configuration option in a case like this. The problem seems to be that pydantic doesn't support validation of an arbitrary Protocol.

What this does is circumvent the detailed/thorough model checking for fields in cases they are just arbitrary types that are not directly supported by pydantic. Instead pydantic just does a simple instance check. Since it is doing an instance check of a Protocol at runtime, you do have to use the runtime_checkable decorator.

Here is the fix I went with:

from typing import TypeVar, Protocol, Generic, runtime_checkable
from pydantic import BaseModel, ConfigDict

D = TypeVar('D')


@runtime_checkable
class SupportsDimensionality(Protocol[D]):
    dimensionality: D


class Model(BaseModel):
    """A base model that allows protocols to be used for fields."""

    model_config = ConfigDict(arbitrary_types_allowed=True)


class ValueModel(Model, Generic[D]):
    value: SupportsDimensionality[D]

Also see this issue at GH with a short discussion of the problem.