ARROW-6652: [Python] Fix ChunkedArray.to_pandas to retain timezone #5471
Conversation
python/pyarrow/table.pxi
Outdated
| if tz is not None: | ||
| tz = string_to_tzinfo(tz) | ||
| result = (result.dt.tz_localize('utc') | ||
| .dt.tz_convert(tz)) |
There was a problem hiding this comment.
I am not sure those 6 lines are worth a separate helper function (now this is duplicated from Array.to_pandas), but if others prefer I can certainly do that
There was a problem hiding this comment.
It's small but carries domain knowledge, so I'd rather see a helper function (perhaps in the pandas_compat module).
There was a problem hiding this comment.
OK, I made a small helper function (see new push).
For now I still put it behind the if check to only need to import it from pandas_compat if it would actually be needed. But that made that it is not much shorter as before in lines of code .. If find this clearer in the code to read, but I can also always call the helper function and move the if check inside the helper function.
Codecov Report
@@ Coverage Diff @@
## master #5471 +/- ##
===========================================
- Coverage 88.64% 66.39% -22.25%
===========================================
Files 958 505 -453
Lines 127522 69887 -57635
Branches 1498 0 -1498
===========================================
- Hits 113039 46403 -66636
- Misses 14118 23484 +9366
+ Partials 365 0 -365
Continue to review full report at Codecov.
|
There was a problem hiding this comment.
LGTM, thanks @jorisvandenbossche ! I confirmed that with this change and #5465 I can pass Spark integration tests locally.
jorisvandenbossche
force-pushed
the
ARROW-6652-chunked-array-timezone
branch
from
dbf788a to
89d0044
Compare
Follow-up on #5462 to also apply this fix for ChunkedArray.