Pytest Patterns

Pytest Patterns

Patterns and conventions for writing Python tests with pytest. Follow the principle of plain test functions over class-based tests, leverage fixtures for setup and teardown, and prefer integration tests over excessive mocking.

Core Rules

• Plain test functions -- Never use TestCase classes. Write standalone test_ functions
• Fixtures for setup -- Use fixtures for all test setup; shared fixtures go in conftest.py
• Run with poetry -- Execute tests using workspace/.venv environment and poetry: poetry run pytest
• | None over Optional -- Use modern union syntax in all test code
• 100-character line limit -- Applies to test files as well
• No docstrings unless necessary -- Test function names should be self-explanatory

Running Tests

# Run all tests
poetry run pytest

# Run with verbose output
poetry run pytest -v

# Run a specific file
poetry run pytest tests/test_orders.py

# Run a specific test
poetry run pytest tests/test_orders.py::test_create_order_with_valid_data

# Run tests matching a keyword
poetry run pytest -k "order and not cancel"

# Run with coverage
poetry run pytest --cov=src --cov-report=term-missing

Key Patterns Overview

Fixtures

Fixtures provide reusable test setup with automatic cleanup. Define fixtures in conftest.py for sharing across test modules. Use fixture scopes (function, module, session) to control lifecycle. Fixture factories return callables for parameterized object creation.

Parametrize

@pytest.mark.parametrize runs the same test with different inputs. Combine multiple parameter sets, use indirect to pass parameters through fixtures, and use pytest.param with id for readable test IDs.

Mocking

Use pytest-mock's mocker fixture over raw unittest.mock.patch. Patch at the import site, not the definition site. Prefer spec=True for type-safe mocks. Use side_effect for conditional returns or exceptions.

Async Testing

Use pytest-asyncio with asyncio_mode = "auto" in pyproject.toml to avoid decorating every async test. Use @pytest_asyncio.fixture for async fixtures. Test FastAPI with httpx.AsyncClient and ASGITransport.

Factory Boy

Use factory_boy with DjangoModelFactory for test data generation. Define SubFactory for related models, LazyAttribute for computed fields, and Trait for common variations.

Reference Documents

• fixtures-parametrize.md -- Fixture scopes, factories, conftest hierarchy, parametrize patterns, factory_boy, monkeypatch, markers, and configuration
• mocking.md -- unittest.mock, pytest-mock, patching strategies, spec-based mocks, external API mocking, Django/FastAPI test clients
• async-testing.md -- pytest-asyncio, async fixtures, FastAPI async testing, Celery task testing, WebSocket testing, anyio, event loop configuration

When to Consult References

• Setting up test fixtures: Read fixtures-parametrize.md for scope management, conftest hierarchy, and fixture factories
• Writing parametrized tests: Read fixtures-parametrize.md for single/multiple params, indirect parametrize, and factory_boy
• Mocking dependencies: Read mocking.md for patch targets, mock types, and framework-specific test clients
• Testing async code: Read async-testing.md for pytest-asyncio setup, async fixtures, and framework integration
• Testing external APIs: Read mocking.md for responses library, httpx mock, and when to prefer integration tests

Source: weorbitant/compound-engineering-feat-python-plugin — distributed by TomeVault.