python-testing

Use the @pytest.mark.parametrize decorator with a list of input and expected value tuples. Pytest runs the test function once per parameter set, reporting each case individually for clear failure diagnosis.

Add --cov=your_package and --cov-fail-under=85 to the addopts setting under [tool.pytest.ini_options] in pyproject.toml. The test run then fails automatically if coverage drops below the threshold.

Use the mocker fixture from pytest-mock to patch functions, for example mocker.patch("my_lib.client.fetch", return_value={"data": []}). This keeps tests deterministic and removes dependencies on external services.

Fixtures declared with @pytest.fixture are injected by name into test functions and can be shared across files via conftest.py. They provide cleaner dependency management and teardown compared to traditional xUnit-style setup methods.

Use property-based testing with Hypothesis when input spaces are large or edge cases are hard to enumerate manually. It generates hundreds of inputs automatically, complementing parametrized example tests for known critical values.