Pass revision in path to AbstractBufferedFile init

[albertvillanova]

Fix #1947.

[HuggingFaceDocBuilderDev]

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (c3b2c03) 82.41% compared to head (cd4e2df) 82.39%.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1948      +/-   ##
==========================================
- Coverage   82.41%   82.39%   -0.03%     
==========================================
  Files          66       66              
  Lines        8133     8133              
==========================================
- Hits         6703     6701       -2     
- Misses       1430     1432       +2     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[Wauplin]

Thanks for opening a PR to fix the issue @albertvillanova! I'm not sure though to understand what could be the side effects of initializing with path instead of resolved_path.unresolve() (@mariosasko could you double-check? 🙏).

Since it's fixing an issue, @albertvillanova could you add a regression test in test_hf_file_system.py? Thanks in advance!

[albertvillanova]

@Wauplin, as requested, I have added a regression test.

In relation with your question above, note that contrary to path, the unresolved resolved_path has the revision integrated in itself.

[Wauplin]

Thanks for adding the test @albertvillanova!

I also tried it myself and looks good. Let's wait for a quick review from @mariosasko or @lhoestq before merging (if it's not a hurry on your side). Thanks for the fix!

[lhoestq]

Thanks for the fix !

[Wauplin]

Thanks for the review, will merge it now :)