[TIP] py.test: How to define a test function with two or more parametrized arguments?

[Benjamin Peterson]

2010/3/20 Stefano Lattarini <stefano.lattarini at gmail.com>:
> Hello everybody.
>
> I want to make it clear first that I'm a newbie when it comes to py.test,
> and I still don't understand test parametrization and funcargs very well,
> so please forgive me if this is a stupid or badly formulated question.
>
> Also, you might found it useful to know that I'm using python 2.5 and
> python 3.1 on Debian GNU/Linux, and that my py.test's version is 1.2.1.
>
> My question is: there is a way to define a test function with two or more
> parametrized arguments (provided by a `pytest_generate_tests' hook), in a
> way that forces the test function to run with *all possible combinations*
> of those parametrized tests?
>
> An example might help to clarify the matter.
>
> I tried this:
>
>  def pytest_generate_tests(metafunc):
>      if "an_even_number" in metafunc.funcargnames:
>          for e in (0, 2, 4, 6):
>              metafunc.addcall(funcargs={'an_even_number': e})
>      if "an_odd_number" in metafunc.funcargnames:
>          for o in (1, 3, 5, 7):
>              metafunc.addcall(funcargs={'an_odd_number': o})
>
>  def test_with_multiple_parametrized_args(an_even_number, an_odd_number):
>      assert (an_even_number + an_odd_number) % 2 == 1
>
> expecting 16 tests to be run, one for each couple (x, y) with  x = 0, 2, 4, 6
> and  y = 1, 3, 5, 7.

The is problem you only provided one funcarg at a time, so the call
ended up trying to request the other one! A nested loop should work
better. :)



-- 
Regards,
Benjamin