isolate settings context before first test

[rababerladuseladim]

PR Context

• blocked by: feature/mx-1502 wikidata results refactor mex-extractors#39 as testing is only possible after the mex-common dependency in mex-extractors is updated to v0.24.0
• Bug: the first test that is executed does not receive isolated settings. It therefore receives the production data as they are in mex-assets. If the test uses data from mex-assets, it will get a different result than if it is not executed first.
• Reproduce: in mex-extractors, execute pdm run pytest .\tests\sumo\test_extract.py::test_extract_cc1_data_model_nokeda
  • passes if not executed as first test
  • fails if executed in isolation / as first test

Fixed

• isolate settings context before first test

[erichesse]

I can confirm the bug, I couldn't confirm the fix. when I install mex-extractors witht "mex-common@git+https://github.com/robert-koch-institut/[email protected]@hotfix/isolate-settings-context-before-first-test" in pyproject.toml, the error seems to persist. How did you test the fix?

Also, should there also be a hotfix version be bumped for mex-common?

[rababerladuseladim]

I bumped the version and will create a new release once this is merged into main.

Regarding testing, I do not really know how to address this. Running the reproduction step on main of mex-extractors will results in this bug. Updating mex-common to 0.24.0 introduces a whole lot of breaking changes, which Kamran is currently addressing in his PR. Hence, testing this is only possible after Kamran finishes his PR. I tested this manually by applying the changed to the installed version of mex-common in the mex-extractors venv.

[erichesse]

test these changes, after tests are fixed in /pull/157 and PR is merged

[rababerladuseladim]

@erichesse Could you review again? I created a branch using this version of mex-common in mex-extractors for testing: robert-koch-institut/mex-extractors#68

[erichesse]

Great! Looks good now