如何使用pytest编写集成测试以及如何重复集成测试

Question

I am new to this so please do not mind if the question is not specific enough.

I want to know how to club unit tests into a single integration test in pytest. Furthermore, I would like to repeat the integration test in a single test session a couple of times. Please let me know if there is a way to do this in pytest.

Scenario: I have two unit tests name test_start_call and test_end_call that are invoked by pytest in that order.

Now I wanted to repeat the process a couple of times so I did this:

for i in range(0,c): pytest.main(some command)

which works fine which will start the test session and tear down the test session as many times as I want with one call being made in each test session.

But I want to make several calls in a single test session and by far I have not found any way to do this since last two days. I tried looking into xdist but I don't want to start new processes in parallel. The integration tests should serially execute unit tests (start call and end call) as many times as I want in a single test session.

I am stuck. So any help would be great. Thank you!

Answer 1

review https://docs.pytest.org/en/latest/parametrize.html

Then add mult parameter to each test and implement --mult in hook pytest_generate_tests to provide the fixture value.

# conftest
def pytest_addoptions(parser):
    parser.addoption('--mult', default=1, help="run many tests")

def pytest_generate_tests(metafunc):
    mult = metafunc.config.getoption('--mult')
    if 'mult' in metafunc.fixturenames:
        for x in xrange(1, mult):
            metafunc.parametrize("mult", mult)

# testfoo
def test_start_call(mult):
    ...

Answer 2

From what you're saying, I'm not quite sure that you are using the right toolset. It sounds like you are either trying to load test something ( run it multiple times and see if it falls over ), or trying to do something more "data driven" - aka given input values x through y, see how it behaves.

If you are trying to do something like load testing, I'd suggest looking into something like locust .

Here is a reasonable blog with different examples on driving unit tests via different data.

Again, not sure if either of these are actually what you're looking for.