refactor(python): Return correct temporal type from Rust in to_numpy

[stinodego]

Ref #14334

... as opposed to returning an integer array and then creating a view afterwards.

[ritchie46]

Couldn't we reuse the integer code branches? It feels a bit wasteful (code bloat wise).

[stinodego]

Couldn't we reuse the integer code branches? It feels a bit wasteful (code bloat wise).

We cannot reuse those because they return floats.

This could be a little bit less wasteful if we had a way to create a Datetime view of an existing integer array, similar to what we have in Python right now. But I don't think the difference will be significant.

I think you were planning to add a view-type functionality to our Rust to_numpy though, right? So maybe we wait for that and see if we can do this more efficiently.

[ritchie46]

Alright. @stinodego can you refactor this one with the new array views? We could probably remove the old view completely.

[stinodego]

Alright. @stinodego can you refactor this one with the new array views? We could probably remove the old view completely.

The old view has now been removed in a different PR. It would still be nice to support creating views with different dtypes in Rust, e.g. a datetime64[ns] view, which seems to be possible. But that can be done in a different PR.

This PR is about the path with nulls, which requires copying the data. Technically we can create an i64 array and then create a view afterwards, but just creating the correct array type right away seems more sensible.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 80.95%. Comparing base (b2291b9) to head (1515087).
Report is 4 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #14353      +/-   ##
==========================================
- Coverage   80.96%   80.95%   -0.02%     
==========================================
  Files        1386     1386              
  Lines      178492   178428      -64     
  Branches     2882     2877       -5     
==========================================
- Hits       144508   144438      -70     
- Misses      33492    33498       +6     
  Partials      492      492              

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