fix: Fix decompress_impl for csv with n_rows set

[Mottl]

This commit truncates the out buffer in decompress_impl to exactly n_rows (if set).

Returning all the out buffer from decompress_impl yields to a corrupted last row and, as a result, to further parse errors like this:

ComputeError(ErrString("could not parse `-` as dtype `f64` at column 'b' (column number 2)...

[ritchie46]

Can we create the compressed file on the fly (in-memory). That saves adding a file to the repo.

[Mottl]

Current compressed.csv.gz is randomly generated and the n_rows which yields to ComputeError is found manually. I'd leave compressed.csv.gz as-is or remove test_read_compressed completely, since it's mostly here for illustrating the initial issue.

[ritchie46]

Maybe we can create a in memory random file that illustrates this. Otherwise removing the test and file is also fine. 👍

[Mottl]

A test_read_compressed and a correspondent csv.gz file are removed 👍

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 80.88%. Comparing base (8a6bf4b) to head (6fb4b51).
Report is 15 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #17118      +/-   ##
==========================================
+ Coverage   80.86%   80.88%   +0.02%     
==========================================
  Files        1456     1456              
  Lines      191141   191340     +199     
  Branches     2728     2739      +11     
==========================================
+ Hits       154562   154769     +207     
+ Misses      36073    36064       -9     
- Partials      506      507       +1     

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

[ritchie46]

Thanks @Mottl