Test parametrization

Enables running a single test function against multiple named input-output pairs so each case is individually identifiable in CI reports and failure diagnostics.

Unit testing: verifying a pure function against several input-output pairs with descriptive names. CI pipelines: needing human-readable test node names to quickly locate which parametrized case failed. Regression testing: adding a specifically-named edge case to an existing parametrized test suite.

pytest test suites, Python unit testing, CI/CD test reporting, TDD workflows

@pytest.mark.parametrize(argnames, [pytest.param(argval1, argval2, id=test_id)])

@pytest.mark.parametrize(..., [pytest.param(..., id=...)])

argnames: comma-separated string of parameter names, argval1: first test input value, argval2: second test input value (add more positional args for more parameters), test_id: string identifier for the individual test case

• pytest.fixture
• pytest.raises
• pytest.approx
• assert
• pytest.mark.xfail

• Plain tuple list without pytest.param: simpler but no custom IDs — failure output shows opaque indices. ids= keyword on parametrize: assigns names to all cases at once but couples name list to value list ordering. hypothesis.strategies: property-based alternative generating many cases automatically without explicit enumeration.

1. Mismatching number of values in pytest.param with argnames count: causes ValueError at collection time. 2. Forgetting the list brackets around pytest.param calls: decorator expects an iterable, causes TypeError. 3. Duplicate id strings across cases: pytest silently overwrites earlier cases, only the last duplicate runs. 4. Passing id as a positional argument instead of keyword: shifts value mapping and causes confusing test failures.

pytest.mark.parametrize with plain tuples: no per-case ID control. unittest.TestCase.subTest: stdlib alternative where subtests do not produce independent test nodes. pytest.mark.parametrize with indirect=True: delegates argument resolution to fixtures rather than injecting raw values.

Coming from unittest: may try to use subTest for parametrization — pytest.param creates fully independent test nodes with separate pass/fail status, unlike subTest which shares a single node. Coming from JUnit: may expect @ParameterizedTest method naming — pytest uses the id string directly in the node name.

• pytest.mark.parametrize
• pytest.param
• pytest.mark.fixture
• pytest.mark.xfail
• pytest.mark.skip
• pytest.raises

1. When NOT to use: single-case tests, quick ad-hoc scripts, or when test IDs add no diagnostic value. 2. Each pytest.param creates a separate test item at collection time; very large param sets (thousands) noticeably slow collection. 3. pytest.param also accepts a marks keyword for per-case markers like skip or xfail, which plain tuples cannot do.

Produces individually named test cases in CI reports, making it trivial to identify which specific input combination caused a failure without digging through assertion output or log files.

Like giving each guest at a dinner party a name tag — pytest.param lets you label every test case so you know exactly who misbehaved when something goes wrong.

pytest.param() also accepts a marks keyword argument to apply markers such as pytest.mark.skip or pytest.mark.xfail to individual cases without affecting the rest of the parametrized set.

pytest.param with marks= keyword for per-case skip/xfail, then hypothesis for property-based testing with automatic case generation