property-based-testing

Use the @given decorator with strategies like st.integers() or st.text() to generate inputs, then assert a property that must always hold, such as commutativity or round-trip equality. Hypothesis runs many generated examples and shrinks any failure to a minimal counterexample.

Unit tests check specific hand-written examples, while property-based tests define general properties and generate hundreds of random inputs to verify them. Property-based testing excels at finding edge cases but is not suited for simple happy-path or business-rule scenarios.

Subclass RuleBasedStateMachine and define @rule methods for operations plus @invariant methods for conditions that must always hold. Hypothesis then generates random sequences of operations and checks the invariants after each step.

Compose built-in strategies using st.builds, st.sampled_from, and .filter() to constrain generated values, such as emails containing '@' or dates within a range. Filters should not be too restrictive, or Hypothesis will struggle to generate valid examples.

Avoid property-based tests for simple happy-path checks, specific business rules, and integration tests against external dependencies. Use example-based unit tests for concrete scenarios and reserve property tests for complex logic and invariants.