pytest-mypy-testing provides a pytest plugin to test that mypy produces a given output. As mypy can be told to display the type of an expression this allows us to check mypys type interference.

pip install pytest-mypy-testing

The Python distribution package contains an entry point so that the plugin is automatically discovered by pytest. To disable the plugin when it is installed , you can use the pytest command line option -p no:mypy-testing.

A mypy test case is a top-level functions decorated with @pytest.mark.mypy_testing in a file name test_*.py or test_*.mypy-testing. Note that we use the Python ast module to parse candidate files and do not import any file, i.e., the decorator must be exactly named @pytest.mark.mypy_testing.

In a test_*.py file you may combine both regular pytest test functions and mypy test functions. A single function can even be both.

Example: A simple mypy test case could look like this:

@pytest.mark.mypy_testing def mypy_test_invalid_assginment(): foo = "abc" foo = 123 # E: Incompatible types in assignment (expression has type "int", variable has type "str")

The plugin runs mypy for every file containing at least one mypy test case. The mypy output is then compared to special Python comments in the file:

• # N: <msg> - we expect a mypy note message
• # W: <msg> - we expect a mypy warning message
• # E: <msg> - we expect a mypy error message
• # R: <msg> - we expect a mypy note message Revealed type is '<msg>'. This is useful to easily check reveal_type output:

  @pytest.mark.mypy_testing def mypy_use_reveal_type(): reveal_type(123) # N: Revealed type is 'Literal[123]?' reveal_type(456) # R: Literal[456]?

Mypy test case functions can be decorated with @pytest.mark.skip and @pytest.mark.xfail to mark them as to-be-skipped and as expected-to-fail, respectively. As with the @pytest.mark.mypy_testing mark, the names must match exactly as the decorators are extracted from the ast.

• Create and activate a Python virtual environment.
• Install development dependencies by calling pip install -U -r requirements.txt.
• Start developing
• To run all tests with tox, Python 3.6, 3.7 and 3.8 must be available. You might want to look into using pyenv.