Add logic to handle globalns and localns in pydantic_model_creator

[alistairmaclean]

Description

When using typing.get_type_hints in Python 3.12, the get_type_hints method will not evaluate forward references unless the referenced objects are part of the objects module scope, or the references are specified in the globalns/localns argument of get_type_hints.

Motivation and Context

When the models are not in the same module, forward references can't be evaluated.

My project structure resembles:

# models/__init__.py
from .event import Event
from .tournament import Tournament

__all__ = [
    Event,
    Tournament,
]

# models/event.py
from tortoise import Model, fields

from .tournament import Tournament


class Event(Model):
    id = fields.IntField(primary_key=True)
    name = fields.TextField()
    tournament: fields.ForeignKeyRelation[Tournament] = fields.ForeignKeyField(
        "models.Tournament", related_name="events"
    )

    def __str__(self):
        return self.name

# models/tournament.py
from tortoise import Model, fields


class Tournament(Model):
    id = fields.IntField(primary_key=True)
    name = fields.TextField()

    events: fields.ReverseRelation["Event"]

    def __str__(self):
        return self.name

# main.py
from models import Tournament

from tortoise.contrib.pydantic.creator import pydantic_model_creator

if __name__ == "__main__":
    a = pydantic_model_creator(Tournament)

Output:

NameError: name 'Event' is not defined

As a temporary fix, I've added this logic in models/__init__.py, which is not ideal:

for m in __all__:
    m.__module__ = __name__

I propose a solution where users specify the globalns for typing.get_type_hints:

# main.py
import models

from tortoise.contrib.pydantic.creator import pydantic_model_creator


globalns = {
    "Tournament": models.Tournament,
    "Event": models.Event,
}

if __name__ == "__main__":
    a = pydantic_model_creator(models.Tournament, globalns=globalns)

I might be missing something, or maybe there's a more elegant solution. I'm open to discussion and feedback.

How Has This Been Tested?

I have been able to generate pydantic models by specifying globalns Dict[str, Type[Model]] when using pydantic_model_creator of my fork.

Checklist:

• My code follows the code style of this project.
• My change requires a change to the documentation.
• I have updated the documentation accordingly.
• I have added the changelog accordingly.
• I have read the CONTRIBUTING document.
• I have added tests to cover my changes.
• All new and existing tests passed.