WIP/ENH: Pass tzinfos to dateutil parser

[jbrockmendel]

Putting this up early because the set_option behavior deserves discussion. The alternative is to make/let users pass tzinfos in Timestamp, to_datetime, DatetimeIndex, and probably others.

• closes Incorrect tzinfo check in tslibs.parsing #22234, Timezones silently dropped in parsing #18702
• tests added / passed
• passes git diff upstream/master -u -- "*.py" | flake8 --diff
• whatsnew entry

Still need to:

• document usage, whether via set_option or passing tzinfos
• tests for array_to_datetime, to_datetime, Timestamp, DatetimeIndex
• test for the fix of Incorrect tzinfo check in tslibs.parsing #22234

[pep8speaks]

Hello @jbrockmendel! Thanks for submitting the PR.

• There are no PEP8 issues in the file pandas/core/config_init.py !

• There are no PEP8 issues in the file pandas/tests/tslibs/test_parsing.py !

[jreback]

huh? what is this supposed to accomplish from a user POV ?

[jbrockmendel]

huh? what is this supposed to accomplish from a user POV ?

At the moment we aren't correctly parsing any timezones aside from UTC offsets.

>>> ts = pd.Timestamp("2018-11-04 3:45 PM CST")
/usr/lib/python2.7/site-packages/dateutil/parser/_parser.py:1204: UnknownTimezoneWarning: tzname CST identified but not understood.  Pass `tzinfos` argument in order to correctly return a timezone-aware datetime.  In a future version, this will raise an exception.
  category=UnknownTimezoneWarning)
>>> 
>>> ts
Timestamp('2018-11-04 15:45:00')

(yes, yes, I know "CST" isn't a timezone. But PST, CST, EST, ... are not uncommon in the wild)

[jreback]

that’s not my point

how is this actually useful?

[jbrockmendel]

how is this actually useful?

Because it allows users to parse date-like strings correctly. Because at the moment we are silently returning incorrect results (well, dateutil is issuing a warning). Because before long dateutil will start raising in these cases and we should provide some way of letting users pass a tzinfos dict.

[jreback]

this is a really tiny edge case and adds lots of complexity

[jbrockmendel]

I agree that this adds complexity, but not that it is a tiny corner case. Let's put this on the back-burner until after 0.24.0.

[jreback]

ok, prob best to close this. and simply create a project for those issues.

[jbrockmendel]

Closing until after 0.24.0