Type stubs: Incorrect return type on validate_assignment

[hassec]

Is it possible that the below type stub is incorrect?

pydantic-core/python/pydantic_core/_pydantic_core.pyi

Lines 174 to 183 in 6a13975

	def validate_assignment(
	self,
	obj: Any,
	field_name: str,
	field_value: Any,
	*,
	strict: bool | None = None,
	from_attributes: bool | None = None,
	context: 'dict[str, Any] | None' = None,
	) -> dict[str, Any] | tuple[dict[str, Any], dict[str, Any] | None, set[str]]:

In the below example the returned value seems to be a of type A.

from pydantic import BaseModel

class A(BaseModel, validate_assignment=True):
    i: int

a = A(i=1)
x = a.__pydantic_validator__.validate_assignment(a, "i", 3)
print(type(x))  # <class '__main__.A'>
print(x is a)  # True

[davidhewitt]

Yes, the type stub looks pretty wonky. In general this should return the model / dataclass / TypedDict instance which has been validated. There is some complexity on the Rust side which is probably what led to this stub. @adriangb maybe you have an idea how we might be able to improve this?

[davidhewitt]

Maybe it's better to just make the type stub here -> Any?

[sydney-runkle]

@Viicos, maybe a good issue for you, as our resident typing guru.

[Viicos]

If the return type changes depending on the provided core schema, SchemaValidator should ideally be made generic. But if not possible, then Any is probably the way to go. But I'm not sure in which scenarios the current return type is valid?

[sydney-runkle]

If the return type changes depending on the provided core schema, SchemaValidator should ideally be made generic. But if not possible, then Any is probably the way to go.

Agreed re generic, and agreed re Any for now.