There are many possible points of failure, so we need to have a test that covers all of it, including the shell-scripts.

The file to be tested is i https://pagure.io/mirrors-countme/blob/main/f/countme-totals.py
There might be some refactoring to be able to execute the equivalent of main() from within the test.

You should be able to use https://pagure.io/mirrors-countme/blob/main/f/test_data/test_result_cmp.tar.xz as the COUNTME_RAW_DB argument.

The test should be data-driven, so that we can easily add reproducers.

You can use the other integration test that focuses on the parsing as a template: https://pagure.io/mirrors-countme/blob/main/f/tests/test_parse.py