pytest-github

Plugin for py.test that integrates with github using markers. Integration allows tests to xfail (or skip) based on the status of linked github issues.

Installation

Install the plugin using pip

pip install pytest-github

Usage

1. Once installed, the following py.test command-line parameters are available.

py.test \
    [--github-cfg=GITHUB_CFG] \
    [--github-username=GITHUB_USERNAME] \
    [--github-token=GITHUB_TOKEN] \
    [--github-completed=GITHUB_COMPLETED] \
    [--github-summary]

2. Next, create a configure file called github.yml that contains your GitHub username and personal api token. A sample file is included below.

github:
    username: j.doe
    token: XXXXXXXXXXXXX

Marker

The following py.test marker is available:

@pytest.mark.github(*args): GitHub issue integration

The marker can be used to influence the outcome of tests. See the examples below for guidance.

Example: xfail

Often, when a test fails, one might file a GitHub issue to track the resolution of the problem. Alternatively, you could use the built-in xfail marker. This is where pytest-github can be of use. To avoid having to review known failures with each test run, and to avoid always using xfail, consider the github marker to dynamically influence the test outcome based on the state of the GitHub issue.

The following example demonstrates using the github marker to influence the outcome of a known failing test.

@pytest.mark.github('https://github.com/some/open/issues/1')
def test_will_xfail():
    assert False

Running this test with py.test will produce the following output:

test.py::test_will_xfail xfail

Example: Anticipating specific exceptions with the 'raises' keyword

To avoid masking additional failures that might be uncovered by a test while a github issue is being resolved, you can restrict expected failures to specific exceptions using the raises keyword argument:

@pytest.mark.github('https://github.com/some/open/issues/1', raises=ZeroDivisionError)
def test_will_xfail():
    foo = 1/0

@pytest.mark.github('https://github.com/some/open/issues/1', raises=ValueError)
def test_will_fail():
    # This test has been marked with an open issue but it will still fail
    # because the exception raised is different from the one indicated by
    # the 'raises' keyword.
    foo = 1/0

Running this test with py.test will produce the following output:

collected 2 items
collected 1 github issues

test.py::test_will_xfail xfail
test.py::test_will_fail FAILED

Example: XPASS

The following example demonstrates a test that succeeds, despite being associated with an open GitHub issue.

@pytest.mark.github('https://github.com/some/open/issues/1')
def test_will_xpass():
    assert True

In this example, the XPASS outcome (a.k.a. unexpected pass) is used.

test.py::test_will_xpass XPASS

Example: PASSED

The following example demonstrates a test that succeeds, while it is associated with a closed GitHub issue. When a test associated with a GitHub

@pytest.mark.github('https://github.com/some/closed/issues/2')
def test_will_pass():
    assert True

In this example, the PASSED outcome is used.

test.py::test_will_pass PASSED

Example: FAILED

The following example demonstrates a test that fails, while it is associated with a closed GitHub issue.

@pytest.mark.github('https://github.com/some/closed/issues/2')
def test_will_fail():
    assert False

In this example, the FAILED outcome is used.

test.py::test_will_fail FAILED

Example: SKIPPED

The following example demonstrates a test that fails, while it is associated with an open GitHub issue.

@pytest.mark.github('https://github.com/some/open/issues/1', skip=True)
def test_will_skip():
    assert False

In this example, the SKIPPED outcome is used.

test.py::test_will_skip SKIPPED

In this example, the SKIPPED outcome is used.

test.py::test_will_skip SKIPPED

Example: IDS

The following example demonstrates a parametrize test that uses the github marker to influence the outcome of a subset of the known failing test.

@pytest.mark.github('https://github.com/some/open/issues/1', ids=['even2', 'even4'])
@pytest.mark.parametrize("count", [1, 2, 3, 4], ids=["odd1", "even2", "odd3", "even4"])
def test_will_xfail(count):
    assert count % 2

Summary of GitHub markers and their associated tests

The --github-summary option lists all GitHub issues referenced by a github marker. The list is divided into two sections, Resolved Issues and Unresolved Issues, where an issue is considered resolved if it has one of the GITHUB_COMPLETED labels. Beneath each issue is a listing of all tests that reference the issue.

Sample output:

Unresolved Issues
https://github.com/repo/open/issues/1
 - test_suite.py:test_foo
https://github.com/repo/open/issues/2
 - test_suite.py:test_bar

Resolved Issues
https://github.com/repo/open/issues/3
 - test_suite.py:test_baz
 - test_suite.py:test_bah