Depends does not work with callable class when from __future__ import annotations is enabled

[Pandede]

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

from __future__ import annotations

from typing import Annotated
from fastapi import Depends, FastAPI, Request

app = FastAPI()

# Function-based dependency (works fine)
def fetch_ip(request: Request) -> str:
    return request.client.host

IP = Annotated[str, Depends(fetch_ip)]

# Class-based dependency (does not work)
class IPFetcher:
    def __call__(self, request: Request) -> str:
        return request.client.host

IP = Annotated[str, Depends(IPFetcher())]  # This raises an error

@app.get("/ip")
async def ask_ip(ip: IP):
    return {"your IP address is": ip}

if __name__ == "__main__":
    from fastapi.testclient import TestClient

    with TestClient(app) as client:
        response = client.get("/ip")
        response.raise_for_status()
        print(response.json())

Description

Observation

When using Depends with a callable class and from __future__ import annotations enabled, FastAPI fails to resolve the dependency. However, function-based dependencies work as expected. Removing from __future__ import annotations resolves the issue.

Expected behaviour

Both function-based and class-based dependencies should work consistently, regardless of from __future__ import annotations.

Workaround

Removing from __future__ import annotations allows both implementations to work correctly.

Operating System

Linux

Operating System Details

No response

FastAPI Version

0.115.8

Pydantic Version

2.10.6

Python Version

Python 3.11.11

Additional Context

No response

[yvallois]

When using from __future__ import annotations, all annotations are stored as strings at runtime.
In this context, FastAPI resolves dependencies by accessing the function's __globals__ attribute.
However, instance methods do not have this attribute and instead FastApi use an empty dictionary.
A potential solution would be for FastAPI to trace back to the associated class method to retrieve globals, for example, using IPFetcher().__call__.__func__.__globals__.

As Workaround, You could also add a class attribute __globals__ = globals(). I'm not sure if this is a good idea, and I don't fully understand its implications :)

[Pandede]

You're right, adding class variable __globals__ = globals() may resolve this issue:

class IPFetcher(object):
    __globals__ = globals()

    def __call__(self, request: Request) -> str:
        return request.client.host

This workaround works fine regradless of the line from __future__ import annotations.