py.test passing results of one test to another

问题

Currently I have test looking like this:

@pytest.mark.parametrize("param", [1,2,3])
def test_two_services(param):
    id = check_service_one(param)
    check_service_two(id)

Is there any way to split this test in two, where a second test depends on a first?

回答1:

Remember to test at the boundary. So if the the values of id depend solely on param and if id isn't some implementation detail, but a part of the defined behaviour of the system under test, split up your tests like so:

def test_service_one(param, id):
    assert check_service_one(param) == id

def test_service_two(id):
    check_service_two(id)  # I'm assuming this does some assertion of its own.

@pytest.fixture
def param(param_and_id):
    param, _ = param_and_id
    return param

@pytest.fixture
def id(param_and_id):
    _, id = param_and_id
    return id

@pytest.fixture(
    params=[
        (1, EXPECTED_ID_FOR_PARAM_1),
        (2, EXPECTED_ID_FOR_PARAM_2),
        (3, EXPECTED_ID_FOR_PARAM_3),
    ],
)
def param_and_id(request):
    return request.param

Like this, the tests are loosely coupled by the inputs of check_service_two matching the expected (and checked by assertion in test_service_one) results of check_service_one, rather than test_service_two depending hard on test_service_one. Thus, the tests can be run in arbitrary order and any one test can be run isolated (without having to run another test first).

来源：https://stackoverflow.com/questions/37613883/py-test-passing-results-of-one-test-to-another