Can't get proper captured log call

[kiddten]

Miniamal example to reproduce issue:

mycode.py

import asyncio

async def run():
    await asyncio.sleep(2)


def dumb(a, b):
    pass


def work():
    task = asyncio.ensure_future(run)
    task.add_done_callback(dumb)

test_code.py

from mycode import work
import pytest


@pytest.mark.asyncio
async def test_code():
    await work()
    # raise

pytest -s test_code.py
run it with commented raise and get successful result w/o any report

but there is an obvious issue with add_done_callback it will call that function with only one parameter

so if we uncomment raise statement we will get additional detail called Captured log call

__________________________________________________________________________________________________ test_code __________________________________________________________________________________________________

    @pytest.mark.asyncio
    async def test_code():
        await work()
>       raise
E       RuntimeError: No active exception to reraise

test_code.py:8: RuntimeError
---------------------------------------------------------------------------------------------- Captured log call ----------------------------------------------------------------------------------------------
base_events.py            1266 ERROR    Exception in callback dumb(<Task finishe...> result=None>) at /home/prj/mycode.py:8
handle: <Handle dumb(<Task finishe...> result=None>) at /home/prj/mycode.py:8>
Traceback (most recent call last):
  File "/usr/lib/python3.6/asyncio/events.py", line 145, in _run
    self._callback(*self._args)
TypeError: dumb() missing 1 required positional argument: 'b'

so is it possible to get that output with "successful" run?

[pytestbot]

GitMate.io thinks possibly related issues are #2647 (call to pytest_runtest_logreport is not proper ), #225 (finalizer function does not get called), #447 (Fixture params not accessible inside the fixture, not getting called multiple times.), #1210 (msg in pytest.exit(msg) doesn't get printed when called in pytest hook), and #2616 (The plugin hooks pytest_cmdline_parse and pytest_load_initial_conftests don't appear to get called).

[nicoddemus]

Hi @kiddick,

Currently the captured logs are only shown if a test fails.

I suppose you want your test to fail instead of just showing up an warning, so I think you can write a fixture which fails a test if it captures an specific logging message. Untested:

@pytest.fixture(autouse=True)
def error_on_async_callback(caplog):
    for r in caplog.records:
        if 'Exception in callback' in r.message:
            assert False, f'Error logged in callback: {r.message}'

[kiddten]

HI @nicoddemus! Thank you for reply, I'll test this approach. May be it is reasonable to consider some cli option to enable log capturing even with successful run.

[nicoddemus]

@kiddick I agree, currently we have -rP which prints captured output even for passing tests, I guess it makes sense to add -rW which does the same for captured warnings.

[nicoddemus]

Created #3633 to follow up the feature request so I'm closing this one. Thanks again @kiddick!