Subclass of pydantic.BaseModel is not recognized as Body (json) parameter

[rigens]

First Check

• I added a very descriptive title here.
• I used the GitHub search to find a similar question and didn't find it.
• I searched the FastAPI documentation, with the integrated search.
• I already searched in Google "How to X in FastAPI" and didn't find any information.
• I already read and followed all the tutorial in the docs and didn't find an answer.
• I already checked if it is not related to FastAPI but to Pydantic.
• I already checked if it is not related to FastAPI but to Swagger UI.
• I already checked if it is not related to FastAPI but to ReDoc.

Commit to Help

• I commit to help with one of those options 👆

Example Code

import collections

import uvicorn
from fastapi import FastAPI
from pydantic import RootModel


class Strings(RootModel[list[str]], collections.abc.Sequence):
    def __getitem__(self, index):
        return self.root[index]

    def __len__(self):
        return len(self.root)

    def __iter__(self):
        yield from self.root


app = FastAPI()


@app.post('/post-strings')
async def post_strings(strings: Strings):
    return strings


uvicorn.run(
    app,  # type: ignore
    host='127.0.0.1',
    port=8088,
)

Description

Strings is a subclass of pydantic.BaseModel and should be automatically recognized as Body parameter. But FastAPI treats it as Form parameter

curl -X POST http://127.0.0.1:8088/post-strings -d '["foo","bar","baz"]' -H "Content-Type: application/json"
...
< HTTP/1.1 422 Unprocessable Entity
{"detail":[{"type":"missing","loc":["body","strings"],"msg":"Field required","input":null}]}

The reason for this behavior is incorrect implementation of the _compat.is_uploadfile_sequence_annotation function.

Related PR: #10393

Operating System

Linux

Operating System Details

No response

FastAPI Version

0.111.0

Pydantic Version

2.8.2

Python Version

3.12.3

Additional Context

No response

[ecly]

@rigens while I'm not certain if this is intended behavior, this seems to be a result of the collections.abc.Sequence subclassing, causing the FieldInfo to become an instance of fastapi.params.File instead of fastapi.params.Body.

Bit of a partial solution, but I'd omit that sequence subclasses and if you need to for type checking, simply pass around the root object instead of the entire RootModel for the rest of your application.

[rigens]

@ecly, this is a BUG as described here #10393

Bit of a partial solution, but I'd omit that sequence subclasses and if you need to for type checking, simply pass around the root object instead of the entire RootModel for the rest of your application

thanks, but we need this subclassing

[rigens]

Clearly this is a BUG.
@tiangolo, please create an issue from this discussion.

[tiangolo]

Hello @rigens, you can achieve the same with:

from fastapi import FastAPI

app = FastAPI()


@app.post("/post-strings")
async def post_strings(strings: list[str]):
    return strings

As FastAPI allows you to use types used for Pydantic fields, there's no need to use RootModel, it's not officially supported, because you can achieve the same in better and simpler ways.

If you need to modify things further, you can also use additional Pydantic validators: https://docs.pydantic.dev/latest/concepts/validators/

[rigens]

...you can achieve the same with...

It's not the same... We need the custom pydantic model (not list[str]) for some other purposes dictated by our application logic (not only for FastAPI request validation)

...RootModel, it's not officially supported...

Really? RootModel is just a subclass of BaseModel, and should be treated the same way as any other pydantic model

@tiangolo, I have clearly pointed out the BUG that is causing the incorrect behavior, and instead of fixing it, you suggest that I use some workarounds