Testing MGET

MGET relies on the pytest testing framework to implement tests that verify that MGET is working as designed. Ultimately, we aim to use GitHub’s automation features to run the tests automatically as part of a continuous integration process. This is nontrivial, though, because a number of tests require ArcGIS, which is non-free software that may not be easily runnable in GitHub’s own environment. Until we sort that out, we are only running tests manually. You should run all tests prior to pushing any commit to the GitHub repo.

Prepare your Python environment

After building MGET as a Python wheel and installing it, you should run the tests from the same Python virtual environment that you installed MGET into. First, you need to install the Python packages needed to run pytest. If you have done this already for this virtual environment, you don’t need to do it again. On Linux:

python3 -m pip install -U pytest python-dotenv

On Windows, replace python3 with python.

Next, if you have not done so already, create a file test/.env in your copy of the MGET repository. This file will contain values of environment variables needed to log in to remote services that MGET interacts with. Open the file with your text editor and paste in the following text and replace ******** with your own credentials:

CMEMS_USERNAME=********
CMEMS_PASSWORD=********
NASA_EARTHDATA_USERNAME=********
NASA_EARTHDATA_PASSWORD=********

  • CMEMS_USERNAME and CMEMS_PASSWORD - your username and password for Copernicus Marine Service, a.k.a. CMEMS.

  • NASA_EARTHDATA_USERNAME and NASA_EARTHDATA_PASSWORD - your username and password for NASA Earthdata.

Warning

Your credentials must be entered in plaintext. To keep them secret, be careful not to allow anyone else to access this file. Keep the file local to your machine. Do not try to commit this file to the repo. .gitignore at the root level of the MGET repository is configured to ignore the file if it exists.

If you don’t have credentials for a service, delete the lines associated with that service. MGET’s test code will detect that no credentials are available and skip tests associated with that service.

Run pytest

On Linux, when you’re ready, from the same virtual environment, execute:

cd test/GeoEco
pytest

On Windows, execute:

cd test/GeoEco
pytest -p no:faulthandler

The text scripts will skip relevant tests when needed credentials are not supplied in the .env file or certain software is not installed, including:

  • ArcGIS Pro

  • MATLAB Runtime R2026a or the full version of MATLAB R2026a

The -p no:faulthandler disables the faulthandler plugin. If you do not disable this plugin, then when pytest first imports the ArcGIS arcpy module, it will report a very large number of messages similar to Windows fatal exception: code 0xc0000139, each followed by a roughly 50 line stack trace. These appear to be first-chance exceptions raised but successfully handled within arcpy when it is imported. Apparently arcpy expects these exceptions under normal operation. They do not represent problems that need to be fixed, but having them in your output window is very distracting and usually overwhelms your Command Prompt window’s buffer. Disalbing the faulthandler plugin prevents it from logging these exceptions.

Please see the pytest documentation for instructions on how to run specific tests and for other information about using pytest.

Writing tests

Eventually we hope to provide guidance on writing tests for MGET. For now we only have this placeholder reminding us it needs to be done.