fix(python): Convert date and datetime in literal construction

[mcrumiller]

Fixes #16012.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 80.73%. Comparing base (120ec7f) to head (7fe2e2d).
Report is 2 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main   #16018   +/-   ##
=======================================
  Coverage   80.72%   80.73%           
=======================================
  Files        1475     1475           
  Lines      193337   193351   +14     
  Branches     2760     2764    +4     
=======================================
+ Hits       156079   156101   +22     
+ Misses      36748    36740    -8     
  Partials      510      510           

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

[alexander-beedie]

Could you tweak the unit tests to include a wider range of dates/datetimes (including pre-epoch, etc)? 👍

[mcrumiller]

@alexander-beedie thanks, working on parametrizing and indeed running into cases where we're outside the Datetime casting ability, but I'm not sure how to deal with it. It looks like Date.cast(Datetime) fails (may open an issue) in general (@MarcoGorelli) (opened #16039):

from datetime import date
import polars as pl

pl.Series([date(1677, 9, 22)]).cast(pl.Datetime("ns")) # 1677-09-22 00:00:00
pl.Series([date(1677, 9, 21)]).cast(pl.Datetime("ns")) # 2262-04-11 23:34:33.709551616

[mcrumiller]

@alexander-beedie fixed up a few things and expanded the tests. Let me know if you think it needs more testing or changes.

I restricted the times in the test to be within the Date -> Nanosecond allowance, as that can be addressed in the other issue I opened.

[alexander-beedie]

Sorry, let this one sit; can you rebase/update and I'll take another look :)

[mcrumiller]

@alexander-beedie should be good once tests complete.

[mcrumiller]

Hey @alexander-beedie wanna merge this one before it goes stale again?

[alexander-beedie]

Hey @alexander-beedie wanna merge this one before it goes stale again?

Oops! Will look shortly 👍

[mcrumiller]

@alexander-beedie rebased again.

[mcrumiller]

Hey @alexander-beedie this one's gone stale again and it's not too complex.

[alexander-beedie]

Hey @alexander-beedie this one's gone stale again and it's not too complex.

Shame on me! 😅
A few comments, then it's likely good to go.

[mcrumiller]

Test failure due to #18211, unrelated to this PR.

@alexander-beedie I think this is ready to go now.