How do (or can) I get a generator called from within a nose test to be processed by nose?

Question

I'm trying to create a genericized data driven test handler that I can call from my nose tests. My test file would look like:

import ScenarioHandler

def test_foo():
    scenario = ScenarioHandler(__test_foo, [1, 2])
    scenario.run()

def __test_foo(var):
    assert var % 2 == 0, 'Odd!'

ScenarioHandler would be something like this:

class ScenarioHandler(object):

    def __init__(self, test, args):
        self.test = test
        self.args = args

    def run(self):
        for arg in self.args:
            yield self.test, arg

The problem I'm running in to is that I can't figure out how to bubble the generator from ScenarioHandler.run() back up to nose. I've tried returning the generator from run() in test_foo(), and that was no good either. Is this even possible?

Answer 1

Why don't you use something like:

def test_foo():
    for var in [1, 2]:
        yield __test_foo, var

See also add_tests(generator) function from my answer.

Answer 2

I don't know how this nose stuff works exactly.

But your given example cannot work.

You have the call tree

test_foo() -> run().

As run() returns a generator which doesn't start until iteration starts, you should do so.

Either do

def test_foo():
    scenario = ScenarioHandler(__test_foo, [1, 2])
    for f, a in scenario.run():
        f(a)

or better, as nose provides this already by explicitly testing the flags of the given function with func.func_code.co_flags & CO_GENERATOR != 0,

def test_foo():
    scenario = ScenarioHandler(__test_foo, [1, 2])
    for f, a in scenario.run():
        yield f, a

in order to have nose make these calls.

(There could have been an even shorter thing, i. e. to use

test_meth = ScenarioHandler(__test_foo, [1, 2]).run

which, alas, doesn't work because methods are obviously not correctly detected as valid call targets.)

But you could add a "test maker" to your class.

# as method:
    def make_test(self):
        def test_generator():
            for item in self.run():
                yield item
        return test_generator

So you can do

test_foo = ScenarioHandler(__test_foo, [1, 2]).make_test()

But probably nose provides better ways to encapsulate your tests in classes by subclassing TestCase and TestSuite.