[Python] Unable to create a nested DefaultDict in a pydantic BaseModel

Consider:

import pydantic
from typing import Annotated
from collections import defaultdict
from uuid import UUID

class Repro(pydantic.BaseModel):
value: Annotated[
defaultdict[UUID, defaultdict[UUID, dict]],
pydantic.Field(
default_factory=lambda: defaultdict[UUID, defaultdict[UUID, dict]](lambda: defaultdict(dict))
),
]

# to validate that the typechecker thinks the above are valid:
instance = Repro(value=defaultdict[UUID, defaultdict[UUID, dict]](lambda: defaultdict(dict)))


The above passes type checking with pyright, but at runtime, the class definition fails while pydantic is trying to introspect the object's schema:

pydantic.errors.PydanticSchemaGenerationError: Unable to infer a default factory for keys of type <class 'dict'>. Only str, int, bool, list, dict, frozenset, tuple, float, set are supported, other types require an explicit default factory set using `DefaultDict[..., Annotated[..., Field(default_factory=...)]]
​

Thinking it might be introspection of the inner defaultdict triggering the error message, I tried adding a pydantic.Field to its type via an inner annotation:

class Repro(pydantic.BaseModel):
value: Annotated[
defaultdict[
UUID, Annotated[defaultdict[UUID, dict], pydantic.Field(default_factory=dict)]
],
pydantic.Field(default_factory=lambda: defaultdict[UUID, defaultdict[UUID, dict]](lambda: defaultdict(dict))),
]


...but with identical effect.


The above is observed with Python 3.12 and Pydantic 2.11.0, type-checked with pyright 1.1.382.

Continue reading...