Bug: beanie.PydanticObjectId cannot be a path parameter anymore

[gsakkis]

Description

I tried upgrading from v2.0.0beta2 to the latest main and found that using a beanie.PydanticObjectId as path parameter raises validation error. Using git bisect I traced it down to b32088f.

URL to code causing the issue

No response

MCVE

from beanie import PydanticObjectId
from litestar import Litestar, get


@get("/test/{id:str}")
async def test(id: PydanticObjectId) -> dict[str, PydanticObjectId]:
    return {"id": id}


app = Litestar(route_handlers=[test], type_encoders={PydanticObjectId: str})

Steps to reproduce

Right before b32088f (66700b7):

curl -s 'localhost:8000/test/64d6088586da7ecbfe53340f' | jq .
{
  "id": "64d6088586da7ecbfe53340f"
}

At b32088f:

curl -s 'localhost:8000/test/64d6088586da7ecbfe53340f' | jq .
{
  "status_code": 400,
  "detail": "Validation failed for GET http://localhost:8000/test/64d6088586da7ecbfe53340f",
  "extra": [
    {
      "message": "Unsupported type: <class 'str'>",
      "key": "id"
    }
  ]
}

### Screenshots

```bash
"![SCREENSHOT_DESCRIPTION](SCREENSHOT_LINK.png)"

Logs

No response

Litestar Version

b32088f

Platform

• Linux
• Mac
• Windows
• Other (Please specify in the description above)

---

Funding

• If you would like to see an issue prioritized, make a pledge towards it!
• We receive the pledge once the issue is completed & verified

[Goldziher]

Hi, have you tried defining a type decoder for the beanie type?

[gsakkis]

No, how? I don't see a type_decoders parameter.

Also why I have to do something extra? The beanie type validates strings already:

In [1]: from beanie import PydanticObjectId

In [2]: PydanticObjectId("64d6088586da7ecbfe53340f")
Out[2]: ObjectId('64d6088586da7ecbfe53340f')

In [3]: PydanticObjectId("123")
...
InvalidId: '123' is not a valid ObjectId, it must be a 12-byte input or a 24-character hex string

[Goldziher]

No, how? I don't see a type_decoders parameter.

Also why I have to do something extra? The beanie type validates strings already:

In [1]: from beanie import PydanticObjectId

In [2]: PydanticObjectId("64d6088586da7ecbfe53340f")
Out[2]: ObjectId('64d6088586da7ecbfe53340f')

In [3]: PydanticObjectId("123")
...
InvalidId: '123' is not a valid ObjectId, it must be a 12-byte input or a 24-character hex string

Msgspec doesn't know how to handle this type.

There is a type_decoders and type_encoders kwargs on all levels of the app.

[Goldziher]

• You need encoders for request and decoders for marshaling json

[gsakkis]

Ah I was at the breaking commit, type_decoders was added later. I'll dig in the code as there are no docs for it afaict.

[provinzkraut]

I'll dig in the code as there are no docs for it afaict.

Yeah, we need to add those. We're lacking docs for a lot of serialisation intricacies in general.

[gsakkis]

This seems to do the trick:

app = Litestar(
    route_handlers=[test],
    type_encoders={PydanticObjectId: str},
    type_decoders=[
        (
            lambda target_type: issubclass(target_type, PydanticObjectId),
            lambda target_type, value: target_type.validate(value),
        )
    ],
)

Wondering if it would be possible to support automatically all Pydantic custom data types with __get_validators__ and/or the equivalent for Pydantic v2.

[provinzkraut]

Wondering if it would be possible to support automatically all Pydantic custom data types with __get_validators__ and/or the equivalent for Pydantic v2.

Should be. We'd just need to add logic for those to the Pydantic plugin.

[gsakkis]

Thanks, closing this.