Skip to main content

Testing in python

Test is integral part of sofware quality and should not be missed.​

  • Always use pytest for testing codes.
    • unittest provided by standard python can be used too if there is some blockage in using pytest.
  • tox and nox are vey good tools especially for CI and multiple version tests.
  • mock should be used for mocking data.
  • factory_boy and faker can be used for fixtures and fake data.
  • hypothesis can be used for property testing.
  • Testing should be broken to unit as well as functional.
  • Use coverage to alert yourself of test coverage. Keep a target of 80 % - 90 % coverage if 100% is not achieved.
  • Only test the changes you made or functionality you added when testing a codebase of well known frameworks.
  • selenium as well as webtest can be used for web based API testing.
  • jsonschema and genson like tool can be used for JSON validity.
  • Always confirm the schema when testing Web API response data.
  • Passing tests for merge should be priority for all projects.
  • Tests should always cover:
    • Unit: for your code units. Please use mock for external dependency and side effects.
    • Functional: Your program functionality.
    • Integration: Your whole program integration.

See tools for packages links.