added rerun_except option

[dheepan2]

Added option support to rerun-except a particular Error/Exception.
Similar to rerun_only.

#179 (comment)

[sallner]

Hello @dheepan2, thank you for your contribution.

This looks like a nice addition to the options, although I think, this can already be accomplished in most cases with the --only-run argument.

To get this PR forward it still needs some tests like

pytest-rerunfailures/test_pytest_rerunfailures.py

Lines 522 to 552 in 409822b

	@pytest.mark.parametrize(
	"only_rerun_texts, should_rerun",
	[
	(["AssertionError"], True),
	(["Assertion*"], True),
	(["Assertion"], True),
	(["ValueError"], False),
	([""], True),
	(["AssertionError: "], True),
	(["AssertionError: ERR"], True),
	(["ERR"], True),
	(["AssertionError,ValueError"], False),
	(["AssertionError ValueError"], False),
	(["AssertionError", "ValueError"], True),
	],
	)
	def test_only_rerun_flag(testdir, only_rerun_texts, should_rerun):
	testdir.makepyfile('def test_only_rerun(): raise AssertionError("ERR")')
	num_failed = 1
	num_passed = 0
	num_reruns = 1
	num_reruns_actual = num_reruns if should_rerun else 0
	pytest_args = ["--reruns", str(num_reruns)]
	for only_rerun_text in only_rerun_texts:
	pytest_args.extend(["--only-rerun", only_rerun_text])
	result = testdir.runpytest(*pytest_args)
	assert_outcomes(
	result, passed=num_passed, failed=num_failed, rerun=num_reruns_actual
	)

but probably also for different errors raised. Additionally a changelog entry is needed.

[dheepan2]

Hello @dheepan2, thank you for your contribution.

This looks like a nice addition to the options, although I think, this can already be accomplished in most cases with the --only-run argument.

To get this PR forward it still needs some tests like

pytest-rerunfailures/test_pytest_rerunfailures.py

Lines 522 to 552 in 409822b

	@pytest.mark.parametrize(
	"only_rerun_texts, should_rerun",
	[
	(["AssertionError"], True),
	(["Assertion*"], True),
	(["Assertion"], True),
	(["ValueError"], False),
	([""], True),
	(["AssertionError: "], True),
	(["AssertionError: ERR"], True),
	(["ERR"], True),
	(["AssertionError,ValueError"], False),
	(["AssertionError ValueError"], False),
	(["AssertionError", "ValueError"], True),
	],
	)
	def test_only_rerun_flag(testdir, only_rerun_texts, should_rerun):
	testdir.makepyfile('def test_only_rerun(): raise AssertionError("ERR")')
	num_failed = 1
	num_passed = 0
	num_reruns = 1
	num_reruns_actual = num_reruns if should_rerun else 0
	pytest_args = ["--reruns", str(num_reruns)]
	for only_rerun_text in only_rerun_texts:
	pytest_args.extend(["--only-rerun", only_rerun_text])
	result = testdir.runpytest(*pytest_args)
	assert_outcomes(
	result, passed=num_passed, failed=num_failed, rerun=num_reruns_actual
	)

but probably also for different errors raised. Additionally a changelog entry is needed.

[dheepan2]

Hello @dheepan2, thank you for your contribution.
This looks like a nice addition to the options, although I think, this can already be accomplished in most cases with the --only-run argument.
To get this PR forward it still needs some tests like

pytest-rerunfailures/test_pytest_rerunfailures.py

Lines 522 to 552 in 409822b

	@pytest.mark.parametrize(
	"only_rerun_texts, should_rerun",
	[
	(["AssertionError"], True),
	(["Assertion*"], True),
	(["Assertion"], True),
	(["ValueError"], False),
	([""], True),
	(["AssertionError: "], True),
	(["AssertionError: ERR"], True),
	(["ERR"], True),
	(["AssertionError,ValueError"], False),
	(["AssertionError ValueError"], False),
	(["AssertionError", "ValueError"], True),
	],
	)
	def test_only_rerun_flag(testdir, only_rerun_texts, should_rerun):
	testdir.makepyfile('def test_only_rerun(): raise AssertionError("ERR")')
	num_failed = 1
	num_passed = 0
	num_reruns = 1
	num_reruns_actual = num_reruns if should_rerun else 0
	pytest_args = ["--reruns", str(num_reruns)]
	for only_rerun_text in only_rerun_texts:
	pytest_args.extend(["--only-rerun", only_rerun_text])
	result = testdir.runpytest(*pytest_args)
	assert_outcomes(
	result, passed=num_passed, failed=num_failed, rerun=num_reruns_actual
	)

but probably also for different errors raised. Additionally a changelog entry is needed.

Yeah, you are right, but in certain cases with infra issues or system issues(we will not be aware of the run time Error/Exception type), during those scenarios I want the pytest to rerun any Errors other than an AssertionError.

I have added the test and updated the changelog too.

[sallner]

@dheepan2 Thank you for also putting in the effort and clean up the readme.

[sallner]

👍