python-testing-standards

The AAA pattern organizes pytest tests into three phases: Arrange (set up test data and mocks), Act (execute the code being tested), and Assert (verify results). This structure makes tests readable, maintainable, and easier to debug by clearly separating setup, execution, and validation logic.

Pytest fixtures provide reusable, isolated test dependencies by creating fresh instances for each test. Use function-scoped fixtures by default for isolation, session-scoped for expensive setup, and parametrized fixtures to test multiple inputs. This approach prevents test interdependence and ensures deterministic behavior.

Use unittest.mock to replace external dependencies with controlled test doubles. Mock at the point of use, verify mock interactions, and avoid over-mocking internal logic. This isolates the code under test and prevents flaky tests caused by external service calls or side effects.

Aim for 80–90% line coverage using pytest-cov, focusing on critical paths and edge cases rather than arbitrary percentages. Coverage measures test reach, not quality; prioritize testing business logic, error handling, and boundary conditions over trivial code paths.

Avoid test interdependence, hardcoded data, missing isolation through fixtures, over-mocking, unclear test names, and tests that verify implementation rather than behavior. Anti-patterns lead to flaky, slow, and hard-to-maintain test suites; follow AAA structure and fixture discipline instead.

Pytest runs both pytest-native and unittest.TestCase styles, but pytest conventions—fixtures, parametrization, clear naming—are incompatible with unittest class-based setup/teardown. Migrate to pytest for cleaner syntax, better isolation, and access to advanced features like fixtures and mocking integration.