python-test-doubles

Create a Mock with spec set to the interface class inside a pytest fixture, then inject it into the system under test. Configure responses with return_value or side_effect in the Arrange step and verify calls in Assert.

Mock replaces synchronous callables, while AsyncMock replaces async functions and coroutines. AsyncMock supports await-specific assertions like assert_awaited_once_with, which Mock does not provide.

Use the patch decorator targeting the module path where the object is imported, such as @patch("boto3.client"). Configure the returned mock's methods inside the test to simulate API responses.

The spec argument restricts the mock to attributes existing on the specified class, so typos in method names raise AttributeError at test time instead of silently passing. This catches interface drift between tests and real code.

Assign an exception instance to the mock's side_effect attribute, for example a ClientError for AWS failures. When the code under test calls the mock, the exception is raised, letting you test error handling paths.

Use Mock(wraps=real_object) when you want to observe calls on a real implementation without changing its behavior. This creates a spy that executes the real logic while recording interactions for verification.