An interesting note about how dateparser parses date strings with tzinfo in them; then some expected interpretations of date strings with tzinfo

test_dateparser_tzstring_conversions.py

import re
import pytz
from datetime import datetime
import dateparser
import pytest

def expected_tz_conversion(datetime_obj, pytz_tzinfo_offset):
    # keep the day and time, just give it tzinfo
    return pytz_tzinfo_offset.localize(datetime_obj)

def another_expected_tz_conversion(datetime_obj, pytz_tzinfo_offset):
    # use the subtraction operator to calc the tzinfo offset
    datetime_obj_as_gmt = pytz.timezone('GMT').localize(datetime_obj)
    return datetime_obj_as_gmt.astimezone(pytz_tzinfo_offset)

@pytest.mark.parametrize('date_string, dateparser_expected_datetime, expected_datetime, another_expected_datetime',[
    # assumes your /etc/localtime -> /usr/share/zoneinfo/America/Los_Angeles
    [
    '13/03/2014 MST',
    datetime(2014, 3, 12, 23, 0),
    datetime(2014, 3, 13, 0, 0).replace(tzinfo=pytz.timezone('MST')),
    datetime(2014, 3, 12, 17, 0).replace(tzinfo=pytz.timezone('MST'))
    ]
])
def test_dateparser_and_two_alternatives(date_string, dateparser_expected_datetime, \
                                                 expected_datetime, another_expected_datetime):
    '''
    parsing a date string such as "13/03/2014 MST" has many potential meanings
    '''

    # the default dateparser interpretation is do a tzinfo shift plus a local conversion
    # https://github.com/scrapinghub/dateparser/blob/master/dateparser/timezone_parser.py#L18-L19
    datetime_obj = dateparser.parse(date_string)
    assert datetime_obj == dateparser_expected_datetime

    # one reasonable expectation would be to keep the day and time the same
    # and just add the tzinfo to the datetime object
    date_string_only, tzinfo_string = re.findall(r'[0-9\/]+',date_string)[0], re.findall(r'[a-zA-Z]+',date_string)[0]
    datetime_obj = dateparser.parse(date_string_only)
    assert expected_tz_conversion(datetime_obj, pytz.timezone(tzinfo_string)) == expected_datetime

    # another reasonable expectation would be to calculate only the tzinfo offset and not try to convert to local tz
    datetime_obj = dateparser.parse(date_string_only)
    assert another_expected_tz_conversion(datetime_obj, pytz.timezone(tzinfo_string)) == another_expected_datetime