[BUG] int64 behave unexpectedly about beanie HOT 11 CLOSED

Just to clarify, this is only happening after upgrading to the Pydantic v2 release.

from beanie.

I have narrowed it down to the parse_model function in beanie/odm/utils/pydantic.py line 42

def parse_model(model_type: Type[BaseModel], data: Any):
    if IS_PYDANTIC_V2:
        return model_type.model_validate(data)
    else:
        return model_type.parse_obj(data)

The data that goes in there looks good. But then comes out wrong. (referring to the int64s)

from beanie.

With some even more digging, it seems that I was able to make the output match what is expected with the following bson_encoders

class TestCollection(Document):
    test: int

    class Settings:
        bson_encoders = {int: lambda x: str(x)}

from beanie.

Thank you for the cath!

from beanie.

No problem, I love this library so I wanted to help as much as I could.

from beanie.

Woah. It looks like it happens somewhere inside Pydantic. But at the same time I can not reproduce it with pure Pydantic. What a black magic! :)

from beanie.

Ah, no. I can.

from bson import Int64
from pydantic import BaseModel


class A(BaseModel):
    i: int


a = A.model_validate({"i": Int64(1136885225876639845)})

print(a.i)

from beanie.

I've opened an issue: pydantic/pydantic#7201

from beanie.

I believe it got fixed on the Pydantic side

from beanie.

This issue can be closed if or when Beanies dependencies are up to date.

from beanie.

This issue is stale because it has been open 30 days with no activity.

from beanie.