A tiny tool whose input is the output of a django test run, and whose output is the failing test cases, formatted as input to the django test-runner.
That's all very circular, so how about an example that silently tests my_app and then immediately re-runs any tests that failed:
./manage.py test my_app 2>&1 | test_coprophagia.py | xargs ./manage.py test
Or, more usefully (and how my aliases are set up to use the script), run the tests normally:
./manage.py test my_app 2>&1 | tee test_output.log
Then, later re-run the tests that failed:
cat test_output.log | test_coprophagia.py | xargs ./manage.py test
A git pre-commit hook to pass any edits to python files through pyflakes and search for stray occurrences of pdb. This will likely save you a good deal of mockery at the hands of your peers.
This uses ack over grep due to my own personal habits, but you can change this in pre-commit if you do not have ack installed.
Move find_pdbs_from_diff.py and pre-commit into the .git/hooks directory of your local git repository.