pytest: How to get a list of all failed tests at the end of the session? (and while using xdist)

Question

I would like to have a list of all the tests that have failed to be used at the end of session.

Pytest lets you define a hook pytest_sessionfinish(session, exitstatus), that is called at the end of the session, where I wish to have that list.

session is a _pytest.main.Session instance that has the attribute items (type list), but I couldn't find whether the each item in that list passed of failed.

1. How can a list of all failed tests could be retrieved at the end of the session?

2. How can it be done while using pytest-xdist plugin, where I would like to get that list in the master process. Using this plugin, session does not even have items attribute in the master:

   def pytest_sessionfinish(session, exitstatus):
       if os.environ.get("PYTEST_XDIST_WORKER", "master") == "master":
            print(hasattr(session, "items"))  # False
   

Solution 1

If you want results of the tests you can use hook runtest_makereport:

@pytest.hookimpl(tryfirst=True, hookwrapper=True)
def pytest_runtest_makereport(item, call):
    outcome = yield
    rep = outcome.get_result()
    if rep.when == 'call' and rep.failed:
        mode = 'a' if os.path.exists('failures') else 'w'
        try:  # Just to not crash py.test reporting
          pass  # the test 'item' failed
        except Exception as e:
            pass

Solution 2

Run pytest with -rf to get it to print a list of failed tests at the end.

From py.test --help:

  -r chars              show extra test summary info as specified by chars
                        (f)ailed, (E)error, (s)skipped, (x)failed, (X)passed,
                        (p)passed, (P)passed with output, (a)all except pP.
                        Warnings are displayed at all times except when
                        --disable-warnings is set

Here's what you get:

$ py.test -rf
================= test session starts =================
platform darwin -- Python 3.7.2, pytest-4.3.1, py-1.6.0, pluggy-0.7.1
[...]
=============== short test summary info ===============
FAILED test_foo.py::test_foo_is_flar
FAILED test_spam.py::test_spam_is_mostly_pork
FAILED test_eggs.py::test_eggs_are_also_spam
=== 3 failed, 222 passed, 8 warnings in 12.52 seconds ==

Solution 3

--result-log is deprecated. You can instead use -v to output the test case names as they run. If you pipe that into a file, you can query it. So if you're running your tests from a script you can do something like:

pytest -v | tee log.txt
grep -E '::.*(FAILURE|ERROR)' log.txt

Solution 4

You can use command line option --result-log:

test_dummy.py:

def test_dummy_success():
    return


def test_dummy_fail():
    raise Exception('Dummy fail')

Command line:

$ py.test --result-log=test_result.txt

Content of the test_result.txt

. test_dummy.py::test_dummy_success
F test_dummy.py::test_dummy_fail
 def test_dummy_fail():
 >       raise Exception('Dummy fail')
 E       Exception: Dummy fail

 test_dummy.py:6: Exception

Just search 'F' in first column and after that would be [file]::[test]

Solution 5

I wanted a concise report of failed tests and parametrized variations so went with pytest_terminal_summary in conftest.py:

def pytest_terminal_summary(terminalreporter, exitstatus, config):
    terminalreporter.section('Failed tests')
    failures = [report.nodeid.split('::')[-1]
                for report in terminalreporter.stats.get('failed', [])]
    terminalreporter.write('\n'.join(failures) + '\n')

If you inspect terminalreporter._session.items, there's more information you can add to the report, this is just what I wanted.

View more solutions

Comments

[Itay]

Thanks! That seems to work well when xdist is not in use, and I can save the failed tests to a global variable to be used by pytest_sessionfinish at the end. When using xdist, pytest_runtest_makereport is called only for the workers and not for master, so they can communicate through file, but a better solution is using pytest_report_teststatus and pytest_runtest_logreport that are called in the master and can be used to collect failed test.

[Itay]

I would like to use this information in pytest_sessionfinish and this method means failed tests list is available only after the pytest process has exited.

[Arindam Roychowdhury]

Could have given the whole function...I tried your solution..did not work for me..... perhaps mention the pytest version you have used....and also a print of the result....and the test case that was run......

[Aleks Amirkhanov]

What exactly does not work? Do you have skipped tests in your setup?

[Arindam Roychowdhury]

Says...item does not have rep_call

[Aleks Amirkhanov]

I guess it means that the failing test was not started. What if you replace item.rep_call.outcome with item.rep_call.outcome if hasattr(session_item, 'rep_call') else item.rep_setup.outcome

[cryanbhu]

I use pytest ... -rA and it gives failures output before passed output, is there a way to make it give failures output at the end? So less scrolling up