Unit tests

An unit tests verified that a modules, a class or a method is working as expected. As opposed to integration and functional tests, they should be as specific as possible, attempting to isolate small blocks of code –i.e. one function at a time. The main goals for unit tests are:

• Confirm that cosipy was succesfully installed.

• Catch pull request that break previous code and prevent maintainer from merging them.

We are working towards having 100% unit test coverage. That means that all lines of code are run at least by one unit test. New pull request that reduce the coverge percentage should not be merge until this is remedied by adding new tests.

Running unit tests

Install pytest and pytest-cov with:

pip pytest pytest-cov

Go to the root folder of your local working repository, the one containing the folder tests. You need to clone the repository, as opposed to installing cosipy with pip.

Run:

pytest --cov=cosipy --cov-report term --cov-report html:tests/cov

The last line of the report will tell you how many test failed (if any), how many passed, and how many warnings were generated.

Open tests/cov/index.html in a browser and check the coverage. This is the percentage of lines that were executed during the tests. The goal is to have a 100% coverage!

Adding a test

The pytest library automatically looks inside the tests folder for function starting with the word test inside files starting with the word test. Subfolder inside the tests folder need to contain a file called __init__.py (can be empty) in order to be picked up.

Each test* function constitutes a test. If they are run without any errors or exception, the pytest considers that the test was suscessful, independently of the return value. Use the builtin asssert keyword to compare a result to an expected value:

from cosipy import test_data
from cosipy.response import FullDetectorResponse

def test_get_effective_area():

    response_path = test_data.path/"test_full_detector_response.h5"

    with FullDetectorResponse.open(response_path) as response:

        assert response[0].ndim == 6

The assert method will raise an exception if the result from the operation on the right return False.

For some tests you might need a dataset. When that’s the case, add your files to the cosipy/test_data folder, which the tests can find by calling cosipy.test_data.path, as shown in the example above. Use the smallest file size possible required by your test. Remember that for an unit test you do not need a full-fledge data file. Instead, mock data and small subsets are enough to test the algorithms, even if they don’t result in realistic outputs or with good scientific quality. Full-fledge data might be needed for integration and functional test, but those will be handled separetly using data outside the cosipy respository.