7. Testing

Test scripts are found in the GitHub repository in the internal/tests directory.

7.1. Unit Tests

Unit tests are run with pytest. They are found in the internal/tests/pytests directory under the wrappers and util directories. Each tool has its own subdirectory containing its test files.

7.1.1. Pytest Requirements

The following Python packages are required to run the tests.

  • pytest: Runs the tests

  • python-dateutil: Required to run METplus wrappers

  • netCDF4: Required for some METplus wrapper functionality

  • pytest-cov (optional): Only if generating code coverage stats

  • pillow (optional): Only used if running diff utility tests

  • pdf2image (optional): Only used if running diff utility tests

7.1.2. Running

To run the unit tests, set the environment variable METPLUS_TEST_OUTPUT_BASE to a path where the user running has write permissions, navigate to the METplus directory, then call pytest:

export METPLUS_TEST_OUTPUT_BASE=/d1/personal/${USER}/pytest
cd METplus
pytest internal/tests/pytests

A report will be output showing which pytest categories failed. To view verbose test output, add the -vv argument:

pytest internal/tests/pytests -vv

7.1.3. Code Coverage

If the pytest-cov package is installed, the code coverage report can be generated from the tests by running:

pytest internal/tests/pytests --cov=metplus --cov-report=term-missing

In addition to the pass/fail report, the code coverage information will be displayed including line numbers that are not covered by any test.

7.1.4. Subsetting Tests by Directory

A subset of the unit tests can be run by adjusting the path. Be sure to include the –cov-append argument so the results of the run are appended to the full code coverage results. To run only the GridStat unit tests:

pytest internal/tests/pytests/wrappers/grid_stat --cov=metplus --cov-report=term-missing --cov-append

7.1.5. Subsetting Tests by Marker

Pytest allows contributors to use markers on test functions. Markers are used to set various features/attributes to test functions so that users can easily exclude or include some tests from the test execution. Pytest provides many inbuilt markers such as xfail, skip and parametrize. Apart from the inbuilt marker names, users can create their own custom marker names.

Creating Pytest Markers

The METplus team has defined various custom marker names for the unit tests and contributors can add to these markers. These custom markers are listed with a brief description, in the internal/tests/pytests/pytest.ini file. Some examples include:

  • diff (custom marker for diff util tests that require additional packages)

  • run_metplus (custom marker for testing run_metplus.py script)

  • util (custom marker for testing metplus/util logic)

  • long (custom marker for tests that take a long time to run)

  • plotting (custom marker for tests that involve plotting)

To set up unit test functions so that they can be easily excluded or included from text execution, contributors can add the following on the line before the function definition in the Python test file:


where <MARKER-NAME> is one of the custom marker strings listed in the internal/tests/pytests/pytest.ini file.

To see an example of the “util” custom marker (@pytest.mark.util), look in the file internal/tests/pytests/util/time_util/test_time_util.py.

New pytest markers should be added to the internal/tests/pytests/pytest.ini file with a brief description. If they are not added to the markers list, then a warning will be output when running the tests.

Using Pytest Markers

The tests that have associated marker names can be now run to include or exclude tests based on the given name.

To run only tests with a given marker, run:

pytest internal/tests/pytests -m <MARKER-NAME>

To run all tests that do not have a given marker, run:

pytest internal/tests/pytests -m "not <MARKER-NAME>"

For example, if you are running on a system that does not have the additional dependencies required to run the diff utility tests, you can run all of the tests except those by running:

pytest internal/tests/pytests -m "not diff"

Multiple marker groups can be run by using the or keyword:

pytest internal/tests/pytests -m "<MARKER-NAME1> or <MARKER-NAME2>"

7.1.6. Writing Unit Tests metplus_config fixture

Many unit tests utilize a pytest fixture named metplus_config. This is defined in the conftest.py file in internal/tests/pytests. This is used to create a METplusConfig object that contains the minimum configurations needed to run METplus, like OUTPUT_BASE. Using this fixture in a pytest will initialize the METplusConfig object to use in the tests.

This also creates a unique output directory for each test where logs and output files are written. This directory is created under $METPLUS_TEST_OUTPUT_BASE/test_output and is named with the run ID. If the test passes, then the output directory is automatically removed. If the test fails, the output directory will not be removed so the content can be reviewed to debug the issue.

To use it, add metplus_config as an argument to the test function:

def test_something(metplus_config)

then set a variable called config using the fixture name:

config = metplus_config

Additional configuration variables can be set by using the set method:

config.set('config', key, value)