Unit testing in cerbere

This section is for cerbere developers, in particular when implementing a new reader class.

reader testing

Cerbere unit testing is based on pytest python package. This section describes the procedure to test new or existing reader classes.

test data

When providing a new reader, test data files should be provided too. These data files can not be stored on cerbere git server and must be made available on some server.

As an example, the test data files for the reader classes implemented by Ifremer are available at ftp://ftp.ifremer.fr/ifremer/cersat/projects/cerbere/test_data/

There should be one sub-folder for each reader module. For instance for the Sentinel-3 SLSTR reader classes (cerberecontrib-s3 package, safesl module), the test data files are to be found at:

ftp://ftp.ifremer.fr/ifremer/cersat/projects/cerbere/test_data/safesl/

There should be at least one test file per class in the reader module.

unit testing on reader classes

Generic test functions are provided in cerbere.test.pytest_reader module.

To unit test a reader class, you basically need to create a new test module, importing . This module should define the required fixtures for the cerbere.test.pytest_reader test functions to work, plus additional test functions you may want to add for a specific reader. A test module must be created for each class in your reader module.

This module file must be in the tests repository of cerbere or your contrib package.

For instance, for the GHRSST reader class available in cerberecontrib-sst <https://gitlab.ifremer.fr/cerbere/cerberecontrib-sst>, a simple test module would look as follow:

from .pytest_reader import *

from cerbere.feature.cswath import Swath

TEST_FILE = '/home/jfpiolle/git/naiad-index/tests/data/avhrr_metop' \
            '/20200330235803-OSISAF-L2P_GHRSST-SSTsubskin-AVHRR_SST_METOP_B' \
            '-sstmgr_metop01_20200330_235803-v02.0-fv01.0.nc'


@pytest.fixture(scope='module')
def test_file():
    return TEST_FILE


@pytest.fixture(scope='module')
def reader():
    return 'GHRSST'


@pytest.fixture(scope='module')
def feature_class():
    return Swath


def test_guess_backend(input_file):
    """Open file and create dataset object"""
    dst = xr.open_dataset(input_file).cb.cfdataset
    assert 'time' in dst
    assert 'id' in dst.attrs

Your test class must import the cerbere.test.pytest_reader that provides many predefine generic test functions to test a reader compliance to cerbere. Required pytest fixtures for these tests to work must be provided in the test class:

the reader class name and associated feature class name to be passed to cerbere.open_feature function, returned respectively by

reader() and feature_class() fixtures.

@pytest.fixture(scope='module')
def reader():
    return 'GHRSST'


@pytest.fixture(scope='module')
def feature_class():
    return Swath

which file to use for testing the class, returned by test_file() fixture.

@pytest.fixture(scope='module')
def test_file():
    return '/20200330235803-OSISAF-L2P_GHRSST-SSTsubskin-AVHRR_SST_METOP_B' \
            '-sstmgr_metop01_20200330_235803-v02.0-fv01.0.nc'

where to get this file, return by download_url() class method

@pytest.fixture(scope='module')
def download_url(cls):
    """Return the URL of the data test repository where to get the test
    files
    """
   return "ftp://ftp.ifremer.fr/ifremer/cersat/projects/cerbere/test_data/"

testing example

In order for the test to run, the location of your test data must be known to the testing procedures. To ensure this, set the environment variable CERBERE_TESTDATA_REPOSITORY to your root repository for your test data.

Set the environment variable to locate your test data:

export CERBERE_TESTDATA_REPOSITORY=/repo/where/you/put/your/test/data/

To run the test suite, go into the root folder of the tested contrib package:

pytest your_class_test.py