SourceXtractor++ test suite

• tests contains the tests for SourceXtractor. Just execute pytest (Python 3 version!) on the project root directory to run them.
• bin contains a set of utilities required to prepare the tests. For instance, to gather some statistics from a sextractor run that can be used to verify that SourceXtractor++ is, at least, as good.

Markers

Tests are marked so different subsets can be executed separately. For instance, tests that take a long time to run SourceXtractor are marked as pytest.mark.slow, so it would be a bad idea to run them on each commit, for instance. They can be filtered out like

py.test-3 -m "not slow"

Markers used in the test suite:

• pytest.mark.slow
• pytest.mark.regression
• pytest.mark.report

Parametrized tests

Some tests are parametrized based on a column, error, etc. This avoids writing two separate tests that do basically the same thing but based on different parameters.

For instance, when testing the photometry, we may be using the isophotal flux, the aperture, the auto aperture, etc. But, ultimately, they are all compared in the same manner with the original 'stuff' simulation.

For these cases, you will see something like

@pytest.mark.parametrize(
    ['flux_column', 'sum_squared_errors'], [
        ['auto_flux', 16400],
        ['isophotal_flux', 23000],
    ]
)

That means the test is going to run twice, once for auto_flux, checking the sum of the squared errors is less than 16400, and another for isophotal_flux, where the error limit is 23000.

Where do these acceptable errors come from, anyway?

From equivalent runs from SExtractor 2. SourceXtractor++ should do at least as good, so these tests try to make sure not only that it runs, but that the emitted values are good enough.