Nose doesn't just support doctest, it actually enhances it. When you're using Nose, you can write test fixtures for your doctest files.

If you pass --doctest-fixtures=_fixture on the command line, Nose will go looking for a fixture file whenever it finds a doctest file. The name of the fixture file is based on the name of the doctest file, and is calculated by appending the doctest fixture suffix (in other words, the value of doctest-fixtures) to the main part of the doctest filename, and then adding .py to the end. For example, if Nose found a doctest file called PID.txt, and had been told to find doctest‑fixtures=_fixture, it would try to find the test fixture in a file called PID_fixture.py.

The test fixture file for a doctest is very simple: it's just a Python module that contains a setup() or setUp() function, and a teardown() or tearDown() function. The setup function is executed before the doctest file, and the teardown function is executed after the doctest file.

The fixture operates in a different namespace from the doctest file, so none of the variables that get defined in the fixture module are visible in the actual tests. If you want to share the variables between the fixture and the test, you'll probably want to do it by making a simple little module to hold the variables, which you can import into both the fixture and the test.