Tips for Debugging Python Nosetests

Apr 23, 2016 using tags python, nosetest

Running a single unit test

The incantation to run a single nose test is:

nosetests -v module.path:PythonClassName.method_name

For example, assuming your test directory structure looks something like:

├── __init__.pyc

And contains something along the lines of:

class TestUpdateUser(unittest.TestCase):

    def test_non_existent_user(self):

The following would only run the test named test_non_existent_user contained in the file.

nosetests -v tests.test_update_user:TestUpdateUser.test_non_existent_user

This is very useful if you need to, for example, strace a unit test in order to understand what’s happening under the hood.

strace -f -s 100000 nosetests -v tests.test_update_user:TestUpdateUser.test_non_existent_user

Triggering pdb on errors or failures

One very useful nose feature is its ability to drop you into pdb whenever it encounters failures or errors.

The relevant command line flags:

  --pdb                 Drop into debugger on failures or errors
  --pdb-failures        Drop into debugger on failures
  --pdb-errors          Drop into debugger on errors

Setting breakpoints inside unit tests

from import set_trace; set_trace()

Nose will break you out into the debugger whenever it encounters this statement. This works across all python source files (application or test code).

The reason import pdb; pdb.set_trace() does not work as you would expect is because nose by default captures stdout (which you can work around by supplying the -s flag)