The problem is that the middleware and synchronization interface are not in the same thread

[hanzhenbjm]

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

my_fast_api.py

import uuid
import asyncio
import threading
from hypercorn.config import Config
from hypercorn.asyncio import serve
from fastapi import FastAPI, Request
from log_new import logger, g

app = FastAPI()
config = Config()
config.bind = ["127.0.0.1:9999"]


@app.middleware("http")
async def cal_time(req: Request, call_next):
    g.transaction_id = str(uuid.uuid4()).replace('-', '')
    logger.warn("threading id = %s, func=%s ", threading.current_thread(), "middleware")
    response = await call_next(req)
    return response


@app.get("/test1")
def test1():
    print("test1 logger=", logger)  # test1 logger= <LoggerPrinter log_test_p2956 (NOTSET)>
    logger.warn("threading id = %s, func=%s ", threading.current_thread(), "test1")
    return "test1"


@app.get("/test2")
async def test2():
    print("test2 logger=", logger)  # test2 logger= <LoggerPrinter log_test_p2956 (NOTSET)>
    logger.warn("threading id = %s, func=%s ", threading.current_thread(), "test2")
    return "test2"


async def run_server():
    await serve(app, config)


if __name__ == '__main__':
    asyncio.run(run_server(), debug=True)

log_new.py

import logging
import platform
from gevent.local import local
from MDCLogger import LoggerPrintImpl

reset_error_logs = ["kafka", "tornado.access", "tornado.application", "tornado.general"]
APP_NAME = "log_test"

LOG_BASE_PATH = ""
if platform.system() == 'Windows':
    LOG_BASE_PATH = r'D:/LOG/fastapi/%s' % APP_NAME
else:
    pass
logger = LoggerPrintImpl.get_logger(APP_NAME, False, reset_error_logs, LOG_BASE_PATH)

g = local()


class CustomFilter(logging.Filter):
    def filter(self, record):
        record.transaction_id = getattr(g, 'transaction_id', None)
        return True



custom_filter = CustomFilter()


logger.addFilter(custom_filter)

### Description

Synchronous middleware log
```{"log_time": "2024-04-24 11:54:58.311", "app_name": "log_test_code", "log_type": "desc", "level": "WARN", "code_message": "threading id = <_MainThread(MainThread, started 13668)>, func=middleware ", "traceId": null, "logger": "MDCLoggerImpl.py", "HOSTNAME": "DESKTOP-FKJ40RC", "module": "MDCLoggerImpl", "lineno": 186, "funcName": "warning", "created": 1713930898.311691, "thread": 13668, "threadName": "MainThread", "processName": "MainProcess", "process": 11108, "transaction_id": "f639a3a248164adaac2c1214726a9b76"}```
Synchronous  Interface log:
```{"log_time": "2024-04-24 11:54:58.326", "app_name": "log_test_code", "log_type": "desc", "level": "WARN", "code_message": "threading id = <WorkerThread(AnyIO worker thread, started 11056)>, func=test1 ", "traceId": null, "logger": "MDCLoggerImpl.py", "HOSTNAME": "DESKTOP-FKJ40RC", "module": "MDCLoggerImpl", "lineno": 186, "funcName": "warning", "created": 1713930898.3266118, "thread": 11056, "threadName": "AnyIO worker thread", "processName": "MainProcess", "process": 11108, "transaction_id": null}```
I found that transaction_id is null in the interface log

asynchronous middleware log
```{"log_time": "2024-04-24 11:54:56.476", "app_name": "log_test_code", "log_type": "desc", "level": "WARN", "code_message": "threading id = <_MainThread(MainThread, started 13668)>, func=middleware ", "traceId": null, "logger": "MDCLoggerImpl.py", "HOSTNAME": "DESKTOP-FKJ40RC", "module": "MDCLoggerImpl", "lineno": 186, "funcName": "warning", "created": 1713930896.4759488, "thread": 13668, "threadName": "MainThread", "processName": "MainProcess", "process": 11108, "transaction_id": "307f06bd2ca04feb8b622e39ec3b0ab8"}```,
asynchronous interface log
```{"log_time": "2024-04-24 11:54:56.483", "app_name": "log_test_code", "log_type": "desc", "level": "WARN", "code_message": "threading id = <_MainThread(MainThread, started 13668)>, func=test2 ", "traceId": null, "logger": "MDCLoggerImpl.py", "HOSTNAME": "DESKTOP-FKJ40RC", "module": "MDCLoggerImpl", "lineno": 186, "funcName": "warning", "created": 1713930896.4839623, "thread": 13668, "threadName": "MainThread", "processName": "MainProcess", "process": 11108, "transaction_id": "307f06bd2ca04feb8b622e39ec3b0ab8"}```

transaction_id is valid when asynchronous. Is this due to different threads?
Because I use a lot of synchro libraries, I want to use a synchro interface to do this


How can transaction_id in the log take effect on the synchronization interface and the transaction_id in the middleware log and the transaction_id in the synchronization interface be consistent so that you can query logs later
transaction_id is used for log link tracing
thank you



### Operating System

Windows

### Operating System Details

_No response_

### FastAPI Version

0.110.1

### Pydantic Version

2.7.0

### Python Version

Python 3.11.1

### Additional Context

![image](https://github.com/tiangolo/fastapi/assets/33510155/a0f36685-13e6-4dec-b466-638977433bf9)

[hanzhenbjm]

The contextvars library works