Grokking the test fixtures

When creating tests using this infrastructure you write the test as if it was aimed at a single device. We then leverege the pytest framework to write test fixtures that make the test run for each device in the specified site.

The fixtures are defined in the conftest.py script, in the root of the repository, the fixture is defined, at the time of writing, as below. This documentation might not keep up with changes, but it aims to provide a bit of understanding of the principle in play.

class DeviceFixture:
    def __init__(self, dev: BCDevice):
        self._device = dev

    @property
    def device(self) -> BCDevice:
        return self._device

    @property
    def kalman_active(self) -> bool:
        kalman_decks = ['bcLighthouse4', 'bcFlow', 'bcFlow2', 'bcDWM1000']
        if self._device.decks:
            return all(deck in kalman_decks for deck in self._device.decks)
        else:
            return False


@pytest.fixture
def test_setup(request):
    ''' This code will run before (and after) a test '''
    fix = DeviceFixture(request.param)
    yield fix  # code after this point will run as teardown after test
    fix.device.cf.close_link()

And in the the tests, for instance in test_param.py, we make use of the fixture like:

@pytest.mark.parametrize(
    'test_setup',
    conftest.get_devices(),
    indirect=True,
    ids=lambda d: d.name
)
class TestParameters:
    def test_param_ronly(self, test_setup):

[...]

The fixture concept of parametrize basicly means run the test for all the combinations we create here. So, what this translates to is that we want to run the tests in the class TestParameters for all the devices returned by conftest.get_devices(). The indirect=True parameter means that we want to preprocess the device before we pass it as an argument to test. We pass it to the test_setup function above. The steps outlined are, for each test:

1. Get the list of devices defined in a site
2. Get the properties and decks required for the test and check if the deck has them. If not, do not generate a test
3. For each device in the list …
   • … pass device as an argument to test_setup and get a DeviceFixure class …
   • … run the test methods with DeviceFixture as a fixture …
   • … which means we will get the class as an argument and that the code after yield will be run when each test is complete