Passing python script arguments to test modules

Question

I have several test modules that are all invoked together via a driver script that can take a variety of arguments. The tests themselves are written using the python unittest module.

import optparse
import unittest
import sys
import os

from tests import testvalidator
from tests import testmodifier
from tests import testimporter

#modify the path so that the test modules under /tests have access to the project root
sys.path.insert(0, os.path.dirname(__file__))

def run(verbosity):
    if verbosity == "0":
            sys.stdout = open(os.devnull, 'w')

    test_suite = unittest.TestSuite()
    test_suite.addTest(unittest.TestLoader().loadTestsFromTestCase(testvalidator.TestValidator))
    test_suite.addTest(unittest.TestLoader().loadTestsFromTestCase(testmodifier.TestModifier))
    test_suite.addTest(unittest.TestLoader().loadTestsFromTestCase(testimporter.TestDataImporter))

    unittest.TextTestRunner(verbosity=int(verbosity)).run(test_suite)

if __name__ == "__main__":

    #a simple way to control output verbosity
    parser = optparse.OptionParser()
    parser.add_option("--verbosity", "--verbosity", dest="verbosity", default="0")
    (options, args) = parser.parse_args()

    run(options.verbosity)

My issue is that, within these test modules, I have certain tests I'd like to skip based on different parameters passed to the driver. I'm aware that unittest provides a family of decorators meant to do this, but I don't know the best way to pass this information on to the individual modules. If I had a --skip-slow argument, for example, how could I then annotate tests as slow, and have them skipped?

Thank you for your time.

Answer 1

I had in fact been wondering this myself, and finally found the solution.

main file...

...
if __name__ == '__main__':
    args = argparser()

    from tests import *

    ...

And in your test modules, just do:

from __main__ import args

print args

I tested this out, and it worked rather nicely. Nice thing is how simple it is, and it's not too much of a hack at all.

Answer 2

You can use nose test runner with the attrib plugin that lets you select test cases based on attributes. In particular, the example in the plugin documentation uses @attr(slow) to mark slow test cases.

After that, from the command line:

• To select all the test cases marked as slow:

  $ nosetests -a slow

• To select all the test cases not marked as slow:

  $ nosetests -a '!slow'