def test3get_match_element_with_different_date_formats(self):
"""Test if different date_formats can be used to match data."""
tz_gmt10 = pytz.timezone("Etc/GMT+10")
en_gb_utf8 = "en_GB.utf8"
en_us_utf8 = "en_US.utf8"
de_at_utf8 = "de_AT.utf8"
multi_locale_dtme = MultiLocaleDateTimeModelElement(
self.id_, [(b"%d.%m.%Y %H:%M:%S.%f", None, None),
(b"%d.%m.%Y %H:%M:%S%z", None, None),
(b"%d.%m.%Y %H:%M:%S", None, None),
(b"%d.%m.%YT%H:%M:%S", None, None),
(b"%d.%m.%Y", None, None),
(b"%H:%M:%S:%f", None, de_at_utf8),
(b"%H:%M:%S", None, None),
(b"%b %d", tz_gmt10, de_at_utf8),
(b"%d %b %Y", None, en_gb_utf8),
(b"%dth %b %Y", None, en_gb_utf8),
(b"%d/%m/%Y", None, en_gb_utf8),
(b"%m-%d-%Y", None, en_us_utf8),
(b"%d.%m. %H:%M:%S:%f", None, de_at_utf8)])
# test normal date
data = b"07.02.2019 11:40:00: it still works"
date = b"07.02.2019 11:40:00"
match_context = DummyMatchContext(data)
match_element = multi_locale_dtme.get_match_element(
self.path, match_context)
self.compare_match_results(data, match_element, match_context,
self.id_ + "/format1", self.path, date,
1549539600, None)
# test leap year date
data = b"29.02.2020 11:40:00: it still works"
date = b"29.02.2020 11:40:00"
match_context = DummyMatchContext(data)
match_element = multi_locale_dtme.get_match_element(
self.path, match_context)
self.compare_match_results(data, match_element, match_context,
self.id_ + "/format1", self.path, date,
1582976400, None)
# test normal date with T
data = b"07.02.2019T11:40:00: it still works"
date = b"07.02.2019T11:40:00"
match_context = DummyMatchContext(data)
match_element = multi_locale_dtme.get_match_element(
self.path, match_context)
self.compare_match_results(data, match_element, match_context,
self.id_ + "/format3", self.path, date,
1549539600, None)
# test normal date with fractions
data = b"07.02.2019 11:40:00.123456: it still works"
date = b"07.02.2019 11:40:00.123456"
match_context = DummyMatchContext(data)
match_element = multi_locale_dtme.get_match_element(
self.path, match_context)
self.compare_match_results(data, match_element, match_context,
self.id_ + "/format0", self.path, date,
1549539600.123456, None)
# test normal date with z
data = b"07.02.2019 11:40:00+0000: it still works"
date = b"07.02.2019 11:40:00+0000"
match_context = DummyMatchContext(data)
match_element = multi_locale_dtme.get_match_element(
self.path, match_context)
self.compare_match_results(data, match_element, match_context,
self.id_ + "/format1", self.path, date,
1549539600, None)
# test with only date defined
data = b"07.02.2019: it still works"
date = b"07.02.2019"
match_context = DummyMatchContext(data)
match_element = multi_locale_dtme.get_match_element(
self.path, match_context)
self.compare_match_results(data, match_element, match_context,
self.id_ + "/format4", self.path, date,
1549497600, None)
# test with only time defined. Here obviously the seconds can not be tested.
data = b"11:40:23: it still works"
date = b"11:40:23"
match_context = DummyMatchContext(data)
match_element = multi_locale_dtme.get_match_element(
self.path, match_context)
self.compare_match_results(data, match_element, match_context,
self.id_ + "/format6", self.path, date,
match_element.match_object, None)
data = b"Feb 25 something happened"
date = b"Feb 25"
match_context = DummyMatchContext(data)
match_element = multi_locale_dtme.get_match_element(
self.path, match_context)
dtm = datetime(datetime.now().year, 2, 25, tzinfo=tz_gmt10)
# total_seconds should be in UTC, so the timezones are parsed out.
total_seconds = (dtm - datetime(1970, 1, 1, tzinfo=tz_gmt10)
).days * 86400 - dtm.utcoffset().total_seconds()
self.compare_match_results(data, match_element, match_context,
self.id_ + "/format7", self.path, date,
total_seconds, None)
# British date
data = b"13 Apr 2019 something happened"
date = b"13 Apr 2019"
match_context = DummyMatchContext(data)
match_element = multi_locale_dtme.get_match_element(
self.path, match_context)
self.compare_match_results(data, match_element, match_context,
self.id_ + "/format8", self.path, date,
1555113600, None)
# British date 2
data = b"13th Apr 2019 something happened"
date = b"13th Apr 2019"
match_context = DummyMatchContext(data)
match_element = multi_locale_dtme.get_match_element(
self.path, match_context)
self.compare_match_results(data, match_element, match_context,
self.id_ + "/format9", self.path, date,
1555113600, None)
# British date 3
data = b"13/04/2019 something happened"
date = b"13/04/2019"
match_context = DummyMatchContext(data)
match_element = multi_locale_dtme.get_match_element(
self.path, match_context)
self.compare_match_results(data, match_element, match_context,
self.id_ + "/format10", self.path, date,
1555113600, None)
# US date
data = b"04-13-2019 something happened"
date = b"04-13-2019"
match_context = DummyMatchContext(data)
match_element = multi_locale_dtme.get_match_element(
self.path, match_context)
self.compare_match_results(data, match_element, match_context,
self.id_ + "/format11", self.path, date,
1555113600, None)
# Austrian date no year - year should already be learnt.
# start year has to be 2021, because all other formats have defined years.
data = b"13.04. 15:12:54:201 something happened"
date = b"13.04. 15:12:54:201"
match_context = DummyMatchContext(data)
match_element = multi_locale_dtme.get_match_element(
self.path, match_context)
self.compare_match_results(data, match_element, match_context,
self.id_ + "/format12", self.path, date,
1618326774.201, None)
multi_locale_dtme.latest_parsed_timestamp = None
# Austrian time no date
data = b"15:12:54:201 something happened"
date = b"15:12:54:201"
match_context = DummyMatchContext(data)
match_element = multi_locale_dtme.get_match_element(
self.path, match_context)
dtm = datetime(datetime.now().year,
datetime.now().month,
datetime.now().day,
15,
12,
54,
201,
tzinfo=timezone.utc)
# total_seconds should be in UTC, so the timezones are parsed out.
delta = (dtm - datetime(1970, 1, 1, tzinfo=dtm.tzinfo))
total_seconds = delta.days * 86400 + delta.seconds + delta.microseconds / 1000
self.compare_match_results(data, match_element, match_context,
self.id_ + "/format5", self.path, date,
total_seconds, None)
def test3get_match_element_with_different_date_formats(self):
"""Test if different date_formats can be used to match data."""
tz_gmt10 = pytz.timezone("Etc/GMT+10")
en_gb_utf8 = "en_GB.utf8"
en_us_utf8 = "en_US.utf8"
de_at_utf8 = "de_AT.utf8"
multi_locale_dtme = MultiLocaleDateTimeModelElement("path", [
(b"%d.%m.%Y %H:%M:%S.%f", None, None), (b"%d.%m.%Y %H:%M:%S%z", None, None), (b"%d.%m.%Y %H:%M:%S", None, None),
(b"%d.%m.%YT%H:%M:%S", None, None), (b"%d.%m.%Y", None, None), (b"%H:%M:%S:%f", None, de_at_utf8),
(b"%H:%M:%S", None, None), (b"%b %d", tz_gmt10, de_at_utf8), (b"%d %b %Y", None, en_gb_utf8),
(b"%dth %b %Y", None, en_gb_utf8), (b"%d/%m/%Y", None, en_gb_utf8), (b"%m-%d-%Y", None, en_us_utf8),
(b"%d.%m. %H:%M:%S:%f", None, de_at_utf8)])
# test normal date
match_context = DummyMatchContext(b"07.02.2019 11:40:00: it still works")
match_element = multi_locale_dtme.get_match_element("match1", match_context)
self.assertEqual(match_element.match_string, b"07.02.2019 11:40:00")
self.assertEqual(match_element.match_object, 1549539600)
self.assertEqual(match_context.match_string, b"07.02.2019 11:40:00")
# test leap year date
match_context = DummyMatchContext(b"29.02.2020 11:40:00: it still works")
match_element = multi_locale_dtme.get_match_element("match1", match_context)
self.assertEqual(match_element.match_string, b"29.02.2020 11:40:00")
self.assertEqual(match_element.match_object, 1582976400)
self.assertEqual(match_context.match_string, b"29.02.2020 11:40:00")
# test normal date with T
match_context = DummyMatchContext(b"07.02.2019T11:40:00: it still works")
match_element = multi_locale_dtme.get_match_element("match1", match_context)
self.assertEqual(match_element.match_string, b"07.02.2019T11:40:00")
self.assertEqual(match_element.match_object, 1549539600)
self.assertEqual(match_context.match_string, b"07.02.2019T11:40:00")
# test normal date with fractions
match_context = DummyMatchContext(b"07.02.2019 11:40:00.123456: it still works")
match_element = multi_locale_dtme.get_match_element("match1", match_context)
self.assertEqual(match_element.match_string, b"07.02.2019 11:40:00.123456")
self.assertEqual(match_element.match_object, 1549539600.123456)
self.assertEqual(match_context.match_string, b"07.02.2019 11:40:00.123456")
# test normal date with z
match_context = DummyMatchContext(b"07.02.2019 11:40:00+0000: it still works")
match_element = multi_locale_dtme.get_match_element("match1", match_context)
self.assertEqual(match_element.match_string, b"07.02.2019 11:40:00+0000")
self.assertEqual(match_element.match_object, 1549539600)
self.assertEqual(match_context.match_string, b"07.02.2019 11:40:00+0000")
match_context = DummyMatchContext(b"07.02.2019 11:40:00: it still works")
match_element = multi_locale_dtme.get_match_element("match1", match_context)
self.assertEqual(match_element.match_string, b"07.02.2019 11:40:00")
self.assertEqual(match_element.match_object, 1549539600)
self.assertEqual(match_context.match_string, b"07.02.2019 11:40:00")
# test with only date defined
match_context = DummyMatchContext(b"07.02.2019: it still works")
match_element = multi_locale_dtme.get_match_element("match1", match_context)
self.assertEqual(match_element.match_string, b"07.02.2019")
self.assertEqual(match_element.match_object, 1549497600)
self.assertEqual(match_context.match_string, b"07.02.2019")
# test with only time defined. Here obviously the seconds can not be tested.
match_context = DummyMatchContext(b"11:40:23: it still works")
match_element = multi_locale_dtme.get_match_element("match1", match_context)
self.assertEqual(match_element.match_string, b"11:40:23")
self.assertEqual(match_context.match_string, b"11:40:23")
string = b"Feb 25"
match_context = DummyMatchContext(string)
match_element = multi_locale_dtme.get_match_element("match", match_context)
date = datetime(datetime.now().year, 2, 25, tzinfo=tz_gmt10)
# total_seconds should be in UTC, so the timezones are parsed out.
total_seconds = (date - datetime(1970, 1, 1, tzinfo=tz_gmt10)).days * 86400 - date.utcoffset().total_seconds()
self.assertEqual(match_element.match_string, b"Feb 25")
self.assertEqual(match_element.match_object, total_seconds)
self.assertEqual(match_context.match_string, b"Feb 25")
# British date
string = b"13 Apr 2019"
match_context = DummyMatchContext(string)
match_element = multi_locale_dtme.get_match_element("match", match_context)
self.assertEqual(match_element.match_string, b"13 Apr 2019")
self.assertEqual(match_element.match_object, 1555113600)
self.assertEqual(match_context.match_string, b"13 Apr 2019")
# British date 2
string = b"13th Apr 2019"
match_context = DummyMatchContext(string)
match_element = multi_locale_dtme.get_match_element("match", match_context)
self.assertEqual(match_element.match_string, b"13th Apr 2019")
self.assertEqual(match_element.match_object, 1555113600)
self.assertEqual(match_context.match_string, b"13th Apr 2019")
# British date 3
string = b"13/04/2019"
match_context = DummyMatchContext(string)
match_element = multi_locale_dtme.get_match_element("match", match_context)
self.assertEqual(match_element.match_string, b"13/04/2019")
self.assertEqual(match_element.match_object, 1555113600)
self.assertEqual(match_context.match_string, b"13/04/2019")
# US date
string = b"04-13-2019"
match_context = DummyMatchContext(string)
match_element = multi_locale_dtme.get_match_element("match", match_context)
self.assertEqual(match_element.match_string, b"04-13-2019")
self.assertEqual(match_element.match_object, 1555113600)
self.assertEqual(match_context.match_string, b"04-13-2019")
# Austrian date no year - year should already be learnt.
# start year has to be 2021, because all other formats have defined years.
string = b"13.04. 15:12:54:201"
match_context = DummyMatchContext(string)
match_element = multi_locale_dtme.get_match_element("match", match_context)
self.assertEqual(match_element.match_string, b"13.04. 15:12:54:201")
self.assertEqual(match_element.match_object, 1618326774.201)
self.assertEqual(match_context.match_string, b"13.04. 15:12:54:201")
multi_locale_dtme.latest_parsed_timestamp = None
# Austrian time no date
string = b"15:12:54:201"
match_context = DummyMatchContext(string)
match_element = multi_locale_dtme.get_match_element("match", match_context)
date = datetime(datetime.now().year, datetime.now().month, datetime.now().day, 15, 12, 54, 201, tzinfo=timezone.utc)
# total_seconds should be in UTC, so the timezones are parsed out.
delta = (date - datetime(1970, 1, 1, tzinfo=date.tzinfo))
total_seconds = delta.days * 86400 + delta.seconds + delta.microseconds / 1000
self.assertEqual(match_element.match_string, b"15:12:54:201")
self.assertEqual(match_element.match_object, total_seconds)
self.assertEqual(match_context.match_string, b"15:12:54:201")