def test_attribute_during_dt_dt():
result = parse([
'during', ['get', 'attr'], '2000-01-01T00:00:00Z',
'2000-01-01T00:00:01Z'
])
assert result == ast.TimeDuring(
ast.Attribute('attr'),
values.Interval(
datetime(2000,
1,
1,
0,
0,
0,
tzinfo=StaticTzInfo('Z', timedelta(0))),
datetime(2000,
1,
1,
0,
0,
1,
tzinfo=StaticTzInfo('Z', timedelta(0))),
),
)
class TestStaticTzInfo(BaseTestCase):
def setUp(self):
super(TestStaticTzInfo, self).setUp()
@parameterized.expand([
param(given_date=datetime(2007, 1, 18, tzinfo=timezone('UTC'))),
param(given_date=datetime(2003, 3, 31, tzinfo=timezone('US/Arizona'))),
param(given_date=datetime(
2000, 2, 20, tzinfo=timezone('Pacific/Samoa'))),
])
def test_localize_raises_error_if_date_has_tzinfo(self, given_date):
self.timezone_info = StaticTzInfo('UTC\\+00:00', timedelta(0))
self.when_date_is_localized(given_date)
self.then_error_was_raised(
ValueError, ['Not naive datetime (tzinfo is already set)'])
def when_date_is_localized(self, given_date):
try:
self.localized_date = self.timezone_info.localize(given_date)
except Exception as error:
self.error = error
def then_localized_date_is(self, expected_date, expected_tzname):
self.assertEqual(self.localized_date.date(), expected_date.date())
self.assertEqual(self.localized_date.tzname(), expected_tzname)
def localize_timezone(date_time, tz_string):
if date_time.tzinfo:
return date_time
tz = None
try:
tz = timezone(tz_string)
except UnknownTimeZoneError as e:
for name, info in _tz_offsets:
if info['regex'].search(' %s' % tz_string):
tz = StaticTzInfo(name, info['offset'])
break
else:
raise e
return tz.localize(date_time)
def localize_timezone(date_time, tz_string):
if date_time.tzinfo:
return date_time
tz = None
try:
tz = timezone(tz_string)
except UnknownTimeZoneError as e:
for name, info in _tz_offsets:
if info['regex'].search(' %s' % tz_string):
tz = StaticTzInfo(name, info['offset'])
break
else:
raise e
return tz.localize(date_time)
def test_begins():
# using timePosition directly
result = parse('''
<fes:Filter xmlns:fes="http://www.opengis.net/fes/2.0"
xmlns:xsd="http://www.w3.org/2001/XMLSchema-datatypes"
xmlns:gml="http://www.opengis.net/gml">
<fes:Begins>
<fes:ValueReference>attr</fes:ValueReference>
<gml:TimePeriod>
<gml:begin>
<gml:TimeInstant>
<gml:timePosition>2000-01-01T00:00:00Z</gml:timePosition>
</gml:TimeInstant>
</gml:begin>
<gml:end>
<gml:TimeInstant>
<gml:timePosition>2001-01-01T00:00:00Z</gml:timePosition>
</gml:TimeInstant>
</gml:end>
</gml:TimePeriod>
</fes:Begins>
</fes:Filter>
''')
assert result == ast.TimeBegins(
ast.Attribute('attr'),
values.Interval(
datetime(2000,
1,
1,
0,
0,
0,
tzinfo=StaticTzInfo('Z', timedelta(0))),
datetime(2001,
1,
1,
0,
0,
0,
tzinfo=StaticTzInfo('Z', timedelta(0))),
),
)
def test_attribute_after_dt_dt():
result = parse({
"after": [
{"property": "attr"},
["2000-01-01T00:00:00Z", "2000-01-01T00:00:01Z"]
]
})
assert result == ast.TimeAfter(
ast.Attribute('attr'),
values.Interval(
datetime(
2000, 1, 1, 0, 0, 0,
tzinfo=StaticTzInfo('Z', timedelta(0))
),
datetime(
2000, 1, 1, 0, 0, 1,
tzinfo=StaticTzInfo('Z', timedelta(0))
),
),
)
def test_attribute_after_dt_dt():
result = parse(
{
"op": "t_after",
"args": [
{"property": "attr"},
{"interval": ["2000-01-01T00:00:00Z", "2000-01-01T00:00:01Z"]},
],
}
)
assert result == ast.TimeAfter(
ast.Attribute("attr"),
values.Interval(
datetime(
2000, 1, 1, 0, 0, 0, tzinfo=StaticTzInfo("Z", timedelta(0))
),
datetime(
2000, 1, 1, 0, 0, 1, tzinfo=StaticTzInfo("Z", timedelta(0))
),
),
)
def test_attribute_before_or_during_dt_dt():
result = parse(
'attr BEFORE OR DURING 2000-01-01T00:00:00Z / 2000-01-01T00:00:01Z')
assert result == ast.TimeBeforeOrDuring(
ast.Attribute('attr'),
values.Interval(
datetime(2000,
1,
1,
0,
0,
0,
tzinfo=StaticTzInfo('Z', timedelta(0))),
datetime(2000,
1,
1,
0,
0,
1,
tzinfo=StaticTzInfo('Z', timedelta(0))),
),
)
def test_attribute_before():
result = parse(
{
"op": "t_before",
"args": [
{"property": "attr"},
{"timestamp": "2000-01-01T00:00:01Z"},
],
}
)
assert result == ast.TimeBefore(
ast.Attribute("attr"),
datetime(2000, 1, 1, 0, 0, 1, tzinfo=StaticTzInfo("Z", timedelta(0))),
)
def test_attribute_before():
result = parse({
"before": [
{"property": "attr"},
"2000-01-01T00:00:01Z",
]
})
assert result == ast.TimeBefore(
ast.Attribute('attr'),
datetime(
2000, 1, 1, 0, 0, 1,
tzinfo=StaticTzInfo('Z', timedelta(0))
),
)
def test_attribute_overlappedby_dt_open():
result = parse({
"overlappedby": [
{"property": "attr"},
["2000-01-01T00:00:03Z", ".."]
]
})
assert result == ast.TimeOverlappedBy(
ast.Attribute('attr'),
values.Interval(
datetime(
2000, 1, 1, 0, 0, 3,
tzinfo=StaticTzInfo('Z', timedelta(0))
),
None,
),
)
def test_attribute_toverlaps_open_dt():
result = parse({
"toverlaps": [
{"property": "attr"},
["..", "2000-01-01T00:00:03Z"]
]
})
assert result == ast.TimeOverlaps(
ast.Attribute('attr'),
values.Interval(
None,
datetime(
2000, 1, 1, 0, 0, 3,
tzinfo=StaticTzInfo('Z', timedelta(0))
),
),
)
def test_attribute_metby_dr_dt():
result = parse({
"metby": [
{"property": "attr"},
["PT4S", "2000-01-01T00:00:03Z"]
]
})
assert result == ast.TimeMetBy(
ast.Attribute('attr'),
values.Interval(
timedelta(seconds=4),
datetime(
2000, 1, 1, 0, 0, 3,
tzinfo=StaticTzInfo('Z', timedelta(0))
),
),
)
def test_meets_dt_dr():
result = parse({
"meets": [
{"property": "attr"},
["2000-01-01T00:00:00Z", "PT4S"]
]
})
assert result == ast.TimeMeets(
ast.Attribute('attr'),
values.Interval(
datetime(
2000, 1, 1, 0, 0, 0,
tzinfo=StaticTzInfo('Z', timedelta(0))
),
timedelta(seconds=4),
),
)
def test_after():
result = parse('''
<fes:Filter xmlns:fes="http://www.opengis.net/fes/2.0"
xmlns:xsd="http://www.w3.org/2001/XMLSchema-datatypes"
xmlns:gml="http://www.opengis.net/gml">
<fes:After>
<fes:ValueReference>attr</fes:ValueReference>
<gml:TimeInstant>
<gml:timePosition>2000-01-01T00:00:00Z</gml:timePosition>
</gml:TimeInstant>
</fes:After>
</fes:Filter>
''')
assert result == ast.TimeAfter(
ast.Attribute('attr'),
datetime(2000, 1, 1, 0, 0, 0, tzinfo=StaticTzInfo('Z', timedelta(0))),
)
def test_meets_dt_dr():
result = parse(
{
"op": "t_meets",
"args": [
{"property": "attr"},
{"interval": ["2000-01-01T00:00:00Z", "PT4S"]},
],
}
)
assert result == ast.TimeMeets(
ast.Attribute("attr"),
values.Interval(
datetime(
2000, 1, 1, 0, 0, 0, tzinfo=StaticTzInfo("Z", timedelta(0))
),
timedelta(seconds=4),
),
)
def test_attribute_metby_dr_dt():
result = parse(
{
"op": "t_metby",
"args": [
{"property": "attr"},
{"interval": ["PT4S", "2000-01-01T00:00:03Z"]},
],
}
)
assert result == ast.TimeMetBy(
ast.Attribute("attr"),
values.Interval(
timedelta(seconds=4),
datetime(
2000, 1, 1, 0, 0, 3, tzinfo=StaticTzInfo("Z", timedelta(0))
),
),
)
def test_attribute_overlappedby_dt_open():
result = parse(
{
"op": "t_overlappedby",
"args": [
{"property": "attr"},
{"interval": ["2000-01-01T00:00:03Z", ".."]},
],
}
)
assert result == ast.TimeOverlappedBy(
ast.Attribute("attr"),
values.Interval(
datetime(
2000, 1, 1, 0, 0, 3, tzinfo=StaticTzInfo("Z", timedelta(0))
),
None,
),
)
def test_localize_raises_error_if_date_has_tzinfo(self, given_date):
self.timezone_info = StaticTzInfo('UTC\\+00:00', timedelta(0))
self.when_date_is_localized(given_date)
self.then_error_was_raised(
ValueError, ['Not naive datetime (tzinfo is already set)'])
def apply_dateparser_timezone(utc_datetime, offset_or_timezone_abb):
for name, info in _tz_offsets:
if info['regex'].search(' %s' % offset_or_timezone_abb):
tz = StaticTzInfo(name, info['offset'])
return utc_datetime.astimezone(tz)
class TestDateParser(BaseTestCase):
def setUp(self):
super(TestDateParser, self).setUp()
self.parser = NotImplemented
self.result = NotImplemented
self.date_parser = NotImplemented
self.date_result = NotImplemented
@parameterized.expand([
# English dates
param('[Sept] 04, 2014.', datetime(2014, 9, 4)),
param('Tuesday Jul 22, 2014', datetime(2014, 7, 22)),
param('Tues 9th Aug, 2015', datetime(2015, 8, 9)),
param('10:04am', datetime(2012, 11, 13, 10, 4)),
param('Friday', datetime(2012, 11, 9)),
param('November 19, 2014 at noon', datetime(2014, 11, 19, 12, 0)),
param('December 13, 2014 at midnight', datetime(2014, 12, 13, 0, 0)),
param('Nov 25 2014 10:17 pm', datetime(2014, 11, 25, 22, 17)),
param('Wed Aug 05 12:00:00 2015', datetime(2015, 8, 5, 12, 0)),
param('April 9, 2013 at 6:11 a.m.', datetime(2013, 4, 9, 6, 11)),
param('Aug. 9, 2012 at 2:57 p.m.', datetime(2012, 8, 9, 14, 57)),
param('December 10, 2014, 11:02:21 pm',
datetime(2014, 12, 10, 23, 2, 21)),
param('8:25 a.m. Dec. 12, 2014', datetime(2014, 12, 12, 8, 25)),
param('2:21 p.m., December 11, 2014', datetime(2014, 12, 11, 14, 21)),
param('Fri, 12 Dec 2014 10:55:50', datetime(2014, 12, 12, 10, 55, 50)),
param('20 Mar 2013 10h11', datetime(2013, 3, 20, 10, 11)),
param('10:06am Dec 11, 2014', datetime(2014, 12, 11, 10, 6)),
param('19 February 2013 year 09:10', datetime(2013, 2, 19, 9, 10)),
param('21 January 2012 13:11:23.678',
datetime(2012, 1, 21, 13, 11, 23, 678000)),
param('1/1/16 9:02:43.1', datetime(2016, 1, 1, 9, 2, 43, 100000)),
# French dates
param('11 Mai 2014', datetime(2014, 5, 11)),
param('dimanche, 11 Mai 2014', datetime(2014, 5, 11)),
param('22 janvier 2015 à 14h40', datetime(2015, 1, 22, 14, 40)),
param('Dimanche 1er Février à 21:24', datetime(2012, 2, 1, 21, 24)),
param('vendredi, décembre 5 2014.', datetime(2014, 12, 5, 0, 0)),
param('le 08 Déc 2014 15:11', datetime(2014, 12, 8, 15, 11)),
param('Le 11 Décembre 2014 à 09:00', datetime(2014, 12, 11, 9, 0)),
param('fév 15, 2013', datetime(2013, 2, 15, 0, 0)),
param('Jeu 15:12', datetime(2012, 11, 8, 15, 12)),
# Spanish dates
param('Martes 21 de Octubre de 2014', datetime(2014, 10, 21)),
param('Miércoles 20 de Noviembre de 2013', datetime(2013, 11, 20)),
param('12 de junio del 2012', datetime(2012, 6, 12)),
param('13 Ago, 2014', datetime(2014, 8, 13)),
param('13 Septiembre, 2014', datetime(2014, 9, 13)),
param('11 Marzo, 2014', datetime(2014, 3, 11)),
param('julio 5, 2015 en 1:04 pm', datetime(2015, 7, 5, 13, 4)),
param('Vi 17:15', datetime(2012, 11, 9, 17, 15)),
# Dutch dates
param('11 augustus 2014', datetime(2014, 8, 11)),
param('14 januari 2014', datetime(2014, 1, 14)),
param('vr jan 24, 2014 12:49', datetime(2014, 1, 24, 12, 49)),
# Italian dates
param('16 giu 2014', datetime(2014, 6, 16)),
param('26 gennaio 2014', datetime(2014, 1, 26)),
param('Ven 18:23', datetime(2012, 11, 9, 18, 23)),
# Portuguese dates
param('sexta-feira, 10 de junho de 2014 14:52',
datetime(2014, 6, 10, 14, 52)),
param('13 Setembro, 2014', datetime(2014, 9, 13)),
param('Sab 3:03', datetime(2012, 11, 10, 3, 3)),
# Russian dates
param('10 мая', datetime(2012, 5, 10)), # forum.codenet.ru
param('26 апреля', datetime(2012, 4, 26)),
param('20 ноября 2013', datetime(2013, 11, 20)),
param('28 октября 2014 в 07:54', datetime(2014, 10, 28, 7, 54)),
param('13 января 2015 г. в 13:34', datetime(2015, 1, 13, 13, 34)),
param('09 августа 2012', datetime(2012, 8, 9, 0, 0)),
param('Авг 26, 2015 15:12', datetime(2015, 8, 26, 15, 12)),
param('2 Декабрь 95 11:15', datetime(1995, 12, 2, 11, 15)),
param('13 янв. 2005 19:13', datetime(2005, 1, 13, 19, 13)),
param('13 авг. 2005 19:13', datetime(2005, 8, 13, 19, 13)),
param('13 авг. 2005г. 19:13', datetime(2005, 8, 13, 19, 13)),
param('13 авг. 2005 г. 19:13', datetime(2005, 8, 13, 19, 13)),
# Turkish dates
param('11 Ağustos, 2014', datetime(2014, 8, 11)),
param('08.Haziran.2014, 11:07', datetime(2014, 6, 8, 11,
7)), # forum.andronova.net
param('17.Şubat.2014, 17:51', datetime(2014, 2, 17, 17, 51)),
param('14-Aralık-2012, 20:56', datetime(2012, 12, 14, 20,
56)), # forum.ceviz.net
# Romanian dates
param('13 iunie 2013', datetime(2013, 6, 13)),
param('14 aprilie 2014', datetime(2014, 4, 14)),
param('18 martie 2012', datetime(2012, 3, 18)),
param('12-Iun-2013', datetime(2013, 6, 12)),
# German dates
param('21. Dezember 2013', datetime(2013, 12, 21)),
param('19. Februar 2012', datetime(2012, 2, 19)),
param('26. Juli 2014', datetime(2014, 7, 26)),
param('18.10.14 um 22:56 Uhr', datetime(2014, 10, 18, 22, 56)),
param('12-Mär-2014', datetime(2014, 3, 12)),
param('Mit 13:14', datetime(2012, 11, 7, 13, 14)),
# Czech dates
param('pon 16. čer 2014 10:07:43', datetime(2014, 6, 16, 10, 7, 43)),
param('13 Srpen, 2014', datetime(2014, 8, 13)),
param('čtv 14. lis 2013 12:38:43', datetime(2013, 11, 14, 12, 38, 43)),
# Thai dates
param('ธันวาคม 11, 2014, 08:55:08 PM',
datetime(2014, 12, 11, 20, 55, 8)),
param('22 พฤษภาคม 2012, 22:12', datetime(2012, 5, 22, 22, 12)),
param('11 กุมภา 2020, 8:13 AM', datetime(2020, 2, 11, 8, 13)),
param('1 เดือนตุลาคม 2005, 1:00 AM', datetime(2005, 10, 1, 1, 0)),
param('11 ก.พ. 2020, 1:13 pm', datetime(2020, 2, 11, 13, 13)),
# Vietnamese dates
param('Thứ năm', datetime(2012, 11, 8)), # Thursday
param('Thứ sáu', datetime(2012, 11, 9)), # Friday
param('Tháng Mười Hai 29, 2013, 14:14',
datetime(2013, 12, 29, 14,
14)), # bpsosrcs.wordpress.com # NOQA
param('05 Tháng một 2015 - 03:54 AM', datetime(2015, 1, 5, 3, 54)),
# Belarusian dates
param('11 траўня', datetime(2012, 5, 11)),
param('4 мая', datetime(2012, 5, 4)),
param('Чацвер 06 жніўня 2015', datetime(2015, 8, 6)),
param('Нд 14 сакавіка 2015 у 7 гадзін 10 хвілін',
datetime(2015, 3, 14, 7, 10)),
param('5 жніўня 2015 года у 13:34', datetime(2015, 8, 5, 13, 34)),
# Ukrainian dates
param('2015-кві-12', datetime(2015, 4, 12)),
param('21 чер 2013 3:13', datetime(2013, 6, 21, 3, 13)),
param('12 лютого 2012, 13:12:23', datetime(2012, 2, 12, 13, 12, 23)),
param('вів о 14:04', datetime(2012, 11, 13, 14, 4)),
# Tagalog dates
param('12 Hulyo 2003 13:01', datetime(2003, 7, 12, 13, 1)),
param('1978, 1 Peb, 7:05 PM', datetime(1978, 2, 1, 19, 5)),
param('2 hun', datetime(2012, 6, 2)),
param('Lin 16:16', datetime(2012, 11, 11, 16, 16)),
# Japanese dates
param('2016年3月20日(日) 21時40分', datetime(2016, 3, 20, 21, 40)),
param("2016年3月20日 21時40分", datetime(2016, 3, 20, 21, 40)),
# Numeric dates
param('06-17-2014', datetime(2014, 6, 17)),
param('13/03/2014', datetime(2014, 3, 13)),
param('11. 12. 2014, 08:45:39', datetime(2014, 11, 12, 8, 45, 39)),
# Miscellaneous dates
param('1 Ni 2015', datetime(2015, 4, 1, 0, 0)),
param('1 Mar 2015', datetime(2015, 3, 1, 0, 0)),
param('1 сер 2015', datetime(2015, 8, 1, 0, 0)),
param('2016020417:10', datetime(2016, 2, 4, 17, 10)),
# Chinese dates
param('2015年04月08日10:05', datetime(2015, 4, 8, 10, 5)),
param('2012年12月20日10:35', datetime(2012, 12, 20, 10, 35)),
param('2016年06月30日09时30分', datetime(2016, 6, 30, 9, 30)),
param('2016年6月2911:30', datetime(2016, 6, 29, 11, 30)),
param('2016年6月29', datetime(2016, 6, 29, 0, 0)),
param('2016年 2月 5日', datetime(2016, 2, 5, 0, 0)),
param('2016年9月14日晚8:00', datetime(2016, 9, 14, 20, 0)),
# Bulgarian
param('25 ян 2016', datetime(2016, 1, 25, 0, 0)),
param('23 декември 2013 15:10:01', datetime(2013, 12, 23, 15, 10, 1)),
# Bangla dates
param('[সেপ্টেম্বর] 04, 2014.', datetime(2014, 9, 4)),
param('মঙ্গলবার জুলাই 22, 2014', datetime(2014, 7, 22)),
param('শুক্রবার', datetime(2012, 11, 9)),
param('শুক্র, 12 ডিসেম্বর 2014 10:55:50',
datetime(2014, 12, 12, 10, 55, 50)),
param('1লা জানুয়ারী 2015', datetime(2015, 1, 1)),
param('25শে মার্চ 1971', datetime(1971, 3, 25)),
param('8ই মে 2002', datetime(2002, 5, 8)),
param('10:06am ডিসেম্বর 11, 2014', datetime(2014, 12, 11, 10, 6)),
param('19 ফেব্রুয়ারী 2013 সাল 09:10', datetime(2013, 2, 19, 9, 10)),
# Hindi dates
param('11 जुलाई 1994, 11:12', datetime(1994, 7, 11, 11, 12)),
param('१७ अक्टूबर २०१८', datetime(2018, 10, 17, 0, 0)),
param('12 जनवरी 1997 11:08 अपराह्न', datetime(1997, 1, 12, 23, 8)),
# Georgian dates
param('2011 წლის 17 მარტი, ოთხშაბათი', datetime(2011, 3, 17, 0, 0)),
param('2015 წ. 12 ივნ, 15:34', datetime(2015, 6, 12, 15, 34))
])
def test_dates_parsing(self, date_string, expected):
self.given_parser(settings={
'NORMALIZE': False,
'RELATIVE_BASE': datetime(2012, 11, 13)
})
self.when_date_is_parsed(date_string)
self.then_date_was_parsed_by_date_parser()
self.then_period_is('day')
self.then_date_obj_exactly_is(expected)
def test_stringified_datetime_should_parse_fine(self):
expected_date = datetime(2012, 11, 13, 10, 15, 5, 330256)
self.given_parser(settings={'RELATIVE_BASE': expected_date})
date_string = str(self.parser.get_date_data('today')['date_obj'])
self.when_date_is_parsed(date_string)
self.then_date_was_parsed_by_date_parser()
self.then_period_is('day')
self.then_date_obj_exactly_is(expected_date)
@parameterized.expand([
# English dates
param('[Sept] 04, 2014.', datetime(2014, 9, 4)),
param('Tuesday Jul 22, 2014', datetime(2014, 7, 22)),
param('10:04am', datetime(2012, 11, 13, 10, 4)),
param('Friday', datetime(2012, 11, 9)),
param('November 19, 2014 at noon', datetime(2014, 11, 19, 12, 0)),
param('December 13, 2014 at midnight', datetime(2014, 12, 13, 0, 0)),
param('Nov 25 2014 10:17 pm', datetime(2014, 11, 25, 22, 17)),
param('Wed Aug 05 12:00:00 2015', datetime(2015, 8, 5, 12, 0)),
param('April 9, 2013 at 6:11 a.m.', datetime(2013, 4, 9, 6, 11)),
param('Aug. 9, 2012 at 2:57 p.m.', datetime(2012, 8, 9, 14, 57)),
param('December 10, 2014, 11:02:21 pm',
datetime(2014, 12, 10, 23, 2, 21)),
param('8:25 a.m. Dec. 12, 2014', datetime(2014, 12, 12, 8, 25)),
param('2:21 p.m., December 11, 2014', datetime(2014, 12, 11, 14, 21)),
param('Fri, 12 Dec 2014 10:55:50', datetime(2014, 12, 12, 10, 55, 50)),
param('20 Mar 2013 10h11', datetime(2013, 3, 20, 10, 11)),
param('10:06am Dec 11, 2014', datetime(2014, 12, 11, 10, 6)),
param('19 February 2013 year 09:10', datetime(2013, 2, 19, 9, 10)),
# French dates
param('11 Mai 2014', datetime(2014, 5, 11)),
param('dimanche, 11 Mai 2014', datetime(2014, 5, 11)),
param('22 janvier 2015 à 14h40', datetime(2015, 1, 22, 14,
40)), # wrong
param('Dimanche 1er Février à 21:24', datetime(2012, 2, 1, 21, 24)),
param('vendredi, décembre 5 2014.', datetime(2014, 12, 5, 0, 0)),
param('le 08 Déc 2014 15:11', datetime(2014, 12, 8, 15, 11)),
param('Le 11 Décembre 2014 à 09:00', datetime(2014, 12, 11, 9, 0)),
param('fév 15, 2013', datetime(2013, 2, 15, 0, 0)),
param('Jeu 15:12', datetime(2012, 11, 8, 15, 12)),
# Spanish dates
param('Martes 21 de Octubre de 2014', datetime(2014, 10, 21)),
param('Miércoles 20 de Noviembre de 2013', datetime(2013, 11, 20)),
param('12 de junio del 2012', datetime(2012, 6, 12)),
param('13 Ago, 2014', datetime(2014, 8, 13)),
param('13 Septiembre, 2014', datetime(2014, 9, 13)),
param('11 Marzo, 2014', datetime(2014, 3, 11)),
param('julio 5, 2015 en 1:04 pm', datetime(2015, 7, 5, 13, 4)),
param('Vi 17:15', datetime(2012, 11, 9, 17, 15)),
# Dutch dates
param('11 augustus 2014', datetime(2014, 8, 11)),
param('14 januari 2014', datetime(2014, 1, 14)),
param('vr jan 24, 2014 12:49', datetime(2014, 1, 24, 12, 49)),
# Italian dates
param('16 giu 2014', datetime(2014, 6, 16)),
param('26 gennaio 2014', datetime(2014, 1, 26)),
param('Ven 18:23', datetime(2012, 11, 9, 18, 23)),
# Portuguese dates
param('sexta-feira, 10 de junho de 2014 14:52',
datetime(2014, 6, 10, 14, 52)),
param('13 Setembro, 2014', datetime(2014, 9, 13)),
param('Sab 3:03', datetime(2012, 11, 10, 3, 3)),
# Russian dates
param('10 мая', datetime(2012, 5, 10)), # forum.codenet.ru
param('26 апреля', datetime(2012, 4, 26)),
param('20 ноября 2013', datetime(2013, 11, 20)),
param('28 октября 2014 в 07:54', datetime(2014, 10, 28, 7, 54)),
param('13 января 2015 г. в 13:34', datetime(2015, 1, 13, 13, 34)),
param('09 августа 2012', datetime(2012, 8, 9, 0, 0)),
param('Авг 26, 2015 15:12', datetime(2015, 8, 26, 15, 12)),
param('2 Декабрь 95 11:15', datetime(1995, 12, 2, 11, 15)),
param('13 янв. 2005 19:13', datetime(2005, 1, 13, 19, 13)),
param('13 авг. 2005 19:13', datetime(2005, 8, 13, 19, 13)),
param('13 авг. 2005г. 19:13', datetime(2005, 8, 13, 19, 13)),
param('13 авг. 2005 г. 19:13', datetime(2005, 8, 13, 19, 13)),
# Turkish dates
param('11 Ağustos, 2014', datetime(2014, 8, 11)),
param('08.Haziran.2014, 11:07', datetime(2014, 6, 8, 11,
7)), # forum.andronova.net
param('17.Şubat.2014, 17:51', datetime(2014, 2, 17, 17, 51)),
param('14-Aralık-2012, 20:56', datetime(2012, 12, 14, 20,
56)), # forum.ceviz.net
# Romanian dates
param('13 iunie 2013', datetime(2013, 6, 13)),
param('14 aprilie 2014', datetime(2014, 4, 14)),
param('18 martie 2012', datetime(2012, 3, 18)),
param('S 14:14', datetime(2012, 11, 10, 14, 14)),
param('12-Iun-2013', datetime(2013, 6, 12)),
# German dates
param('21. Dezember 2013', datetime(2013, 12, 21)),
param('19. Februar 2012', datetime(2012, 2, 19)),
param('26. Juli 2014', datetime(2014, 7, 26)),
param('18.10.14 um 22:56 Uhr', datetime(2014, 10, 18, 22, 56)),
param('12-Mär-2014', datetime(2014, 3, 12)),
param('Mit 13:14', datetime(2012, 11, 7, 13, 14)),
# Czech dates
param('pon 16. čer 2014 10:07:43', datetime(2014, 6, 16, 10, 7, 43)),
param('13 Srpen, 2014', datetime(2014, 8, 13)),
param('čtv 14. lis 2013 12:38:43', datetime(2013, 11, 14, 12, 38, 43)),
# Thai dates
param('ธันวาคม 11, 2014, 08:55:08 PM',
datetime(2014, 12, 11, 20, 55, 8)),
param('22 พฤษภาคม 2012, 22:12', datetime(2012, 5, 22, 22, 12)),
param('11 กุมภา 2020, 8:13 AM', datetime(2020, 2, 11, 8, 13)),
param('1 เดือนตุลาคม 2005, 1:00 AM', datetime(2005, 10, 1, 1, 0)),
param('11 ก.พ. 2020, 1:13 pm', datetime(2020, 2, 11, 13, 13)),
# Vietnamese dates
param('Thứ năm', datetime(2012, 11, 8)), # Thursday
param('Thứ sáu', datetime(2012, 11, 9)), # Friday
param('Tháng Mười Hai 29, 2013, 14:14',
datetime(2013, 12, 29, 14,
14)), # bpsosrcs.wordpress.com # NOQA
param('05 Tháng một 2015 - 03:54 AM', datetime(2015, 1, 5, 3, 54)),
# Belarusian dates
param('11 траўня', datetime(2012, 5, 11)),
param('4 мая', datetime(2012, 5, 4)),
param('Чацвер 06 жніўня 2015', datetime(2015, 8, 6)),
param('Нд 14 сакавіка 2015 у 7 гадзін 10 хвілін',
datetime(2015, 3, 14, 7, 10)),
param('5 жніўня 2015 года у 13:34', datetime(2015, 8, 5, 13, 34)),
# Ukrainian dates
param('2015-кві-12', datetime(2015, 4, 12)),
param('21 чер 2013 3:13', datetime(2013, 6, 21, 3, 13)),
param('12 лютого 2012, 13:12:23', datetime(2012, 2, 12, 13, 12, 23)),
param('вів о 14:04', datetime(2012, 11, 13, 14, 4)),
# Filipino dates
param('12 Hulyo 2003 13:01', datetime(2003, 7, 12, 13, 1)),
param('1978, 1 Peb, 7:05 PM', datetime(1978, 2, 1, 19, 5)),
param('2 hun', datetime(2012, 6, 2)),
param('Lin 16:16', datetime(2012, 11, 11, 16, 16)),
# Japanese dates
param('2016年3月20日(日) 21時40分', datetime(2016, 3, 20, 21, 40)),
param("2016年3月20日 21時40分", datetime(2016, 3, 20, 21, 40)),
# Bangla dates
param('[সেপ্টেম্বর] 04, 2014.', datetime(2014, 9, 4)),
param('মঙ্গলবার জুলাই 22, 2014', datetime(2014, 7, 22)),
param('শুক্রবার', datetime(2012, 11, 9)),
param('শুক্র, 12 ডিসেম্বর 2014 10:55:50',
datetime(2014, 12, 12, 10, 55, 50)),
param('1লা জানুয়ারী 2015', datetime(2015, 1, 1)),
param('25শে মার্চ 1971', datetime(1971, 3, 25)),
param('8ই মে 2002', datetime(2002, 5, 8)),
param('10:06am ডিসেম্বর 11, 2014', datetime(2014, 12, 11, 10, 6)),
param('19 ফেব্রুয়ারী 2013 সাল 09:10', datetime(2013, 2, 19, 9, 10)),
# Numeric dates
param('06-17-2014', datetime(2014, 6, 17)),
param('13/03/2014', datetime(2014, 3, 13)),
param('11. 12. 2014, 08:45:39', datetime(2014, 11, 12, 8, 45, 39)),
# Miscellaneous dates
param('1 Ni 2015', datetime(2015, 4, 1, 0, 0)),
param('1 Mar 2015', datetime(2015, 3, 1, 0, 0)),
param('1 сер 2015', datetime(2015, 8, 1, 0, 0)),
# Bulgarian
param('24 ян 2015г.', datetime(2015, 1, 24, 0, 0)),
# Hindi dates
param('बुधवार 24 मई 1997 12:09', datetime(1997, 5, 24, 12, 9)),
param('28 दिसम्बर 2000 , 01:09:08', datetime(2000, 12, 28, 1, 9, 8)),
param('१६ दिसम्बर १९७१', datetime(1971, 12, 16, 0, 0)),
param('सन् 1989 11 फ़रवरी 09:43', datetime(1989, 2, 11, 9, 43)),
])
def test_dates_parsing_with_normalization(self, date_string, expected):
self.given_local_tz_offset(0)
self.given_parser(settings={
'NORMALIZE': True,
'RELATIVE_BASE': datetime(2012, 11, 13)
})
self.when_date_is_parsed(normalize_unicode(date_string))
self.then_date_was_parsed_by_date_parser()
self.then_period_is('day')
self.then_date_obj_exactly_is(expected)
@parameterized.expand([
param('Sep 03 2014 | 4:32 pm EDT', datetime(2014, 9, 3, 20, 32)),
param('17th October, 2034 @ 01:08 am PDT',
datetime(2034, 10, 17, 8, 8)),
param('15 May 2004 23:24 EDT', datetime(2004, 5, 16, 3, 24)),
param('08/17/14 17:00 (PDT)', datetime(2014, 8, 18, 0, 0)),
])
def test_parsing_with_time_zones_and_converting_to_UTC(
self, date_string, expected):
self.given_parser(settings={'TO_TIMEZONE': 'UTC'})
self.when_date_is_parsed(date_string)
self.then_date_was_parsed_by_date_parser()
self.then_period_is('day')
self.then_timezone_parsed_is('UTC')
self.then_date_obj_exactly_is(expected)
@parameterized.expand([
param('Sep 03 2014 | 4:32 pm EDT', 'EDT', datetime(2014, 9, 3, 16,
32)),
param('17th October, 2034 @ 01:08 am PDT', 'PDT',
datetime(2034, 10, 17, 1, 8)),
param('15 May 2004 23:24 EDT', 'EDT', datetime(2004, 5, 15, 23, 24)),
param('08/17/14 17:00 (PDT)', 'PDT', datetime(2014, 8, 17, 17, 0)),
param('15 May 2004 16:10 -0400', '-04:00',
datetime(2004, 5, 15, 16, 10)),
param('1999-12-31 19:00:00 -0500', '-05:00',
datetime(1999, 12, 31, 19, 0)),
param('1999-12-31 19:00:00 +0500', '+05:00',
datetime(1999, 12, 31, 19, 0)),
param('Fri, 09 Sep 2005 13:51:39 -0700', '-07:00',
datetime(2005, 9, 9, 13, 51, 39)),
param('Fri, 09 Sep 2005 13:51:39 +0000', '+00:00',
datetime(2005, 9, 9, 13, 51, 39)),
])
def test_dateparser_should_return_tzaware_date_when_tz_info_present_in_date_string(
self, date_string, timezone_str, expected):
self.given_parser()
self.when_date_is_parsed(date_string)
self.then_date_was_parsed_by_date_parser()
self.then_period_is('day')
self.then_timezone_parsed_is(timezone_str)
self.then_date_obj_exactly_is(expected)
@parameterized.expand([
param('15 May 2004 16:10 -0400', 'UTC', datetime(2004, 5, 15, 20, 10)),
param('1999-12-31 19:00:00 -0500', 'UTC', datetime(2000, 1, 1, 0, 0)),
param('1999-12-31 19:00:00 +0500', 'UTC',
datetime(1999, 12, 31, 14, 0)),
param('Fri, 09 Sep 2005 13:51:39 -0700', 'GMT',
datetime(2005, 9, 9, 20, 51, 39)),
param('Fri, 09 Sep 2005 13:51:39 +0000', 'GMT',
datetime(2005, 9, 9, 13, 51, 39)),
])
def test_dateparser_should_return_date_in_setting_timezone_if_timezone_info_present_both_in_datestring_and_given_in_settings(
self, date_string, setting_timezone, expected):
self.given_parser(settings={'TIMEZONE': setting_timezone})
self.when_date_is_parsed(date_string)
self.then_date_was_parsed_by_date_parser()
self.then_period_is('day')
self.then_timezone_parsed_is(setting_timezone)
self.then_date_obj_exactly_is(expected)
@parameterized.expand([
param('15 May 2004 16:10 -0400', datetime(2004, 5, 15, 20, 10)),
param('1999-12-31 19:00:00 -0500', datetime(2000, 1, 1, 0, 0)),
param('1999-12-31 19:00:00 +0500', datetime(1999, 12, 31, 14, 0)),
param('Fri, 09 Sep 2005 13:51:39 -0700',
datetime(2005, 9, 9, 20, 51, 39)),
param('Fri, 09 Sep 2005 13:51:39 +0000',
datetime(2005, 9, 9, 13, 51, 39)),
param('Fri Sep 23 2016 10:34:51 GMT+0800 (CST)',
datetime(2016, 9, 23, 2, 34, 51)),
])
def test_parsing_with_utc_offsets(self, date_string, expected):
self.given_parser(settings={'TO_TIMEZONE': 'utc'})
self.when_date_is_parsed(date_string)
self.then_date_was_parsed_by_date_parser()
self.then_period_is('day')
self.then_timezone_parsed_is('UTC')
self.then_date_obj_exactly_is(expected)
def test_empty_dates_string_is_not_parsed(self):
self.when_date_is_parsed_by_date_parser('')
self.then_error_was_raised(ValueError, ["Empty string"])
@parameterized.expand([
param('invalid date string', 'Unable to parse: invalid'),
param('Aug 7, 2014Aug 7, 2014', 'Unable to parse: Aug'),
param('24h ago', 'Unable to parse: h'),
param('2015-03-17t16:37:51+00:002015-03-17t15:24:37+00:00',
'Unable to parse: 00:002015'),
param(
'8 enero 2013 martes 7:03 AM EST 8 enero 2013 martes 7:03 AM EST',
'Unable to parse: 8'),
param('12/09/18567', 'Unable to parse: 18567'),
])
def test_dates_not_parsed(self, date_string, message):
self.when_date_is_parsed_by_date_parser(date_string)
self.then_error_was_raised(ValueError, message)
@parameterized.expand([
param('10 December', datetime(2014, 12, 10)),
param('March', datetime(2014, 3, 15)),
param('Friday', datetime(2015, 2, 13)),
param('Monday', datetime(2015, 2, 9)),
param('Sunday', datetime(2015, 2, 8)), # current day
param('10:00PM', datetime(2015, 2, 14, 22, 0)),
param('16:10', datetime(2015, 2, 14, 16, 10)),
param('14:05', datetime(2015, 2, 15, 14, 5)),
param('15 february 15:00', datetime(2015, 2, 15, 15, 0)),
param('3/3/50', datetime(1950, 3, 3)),
param('3/3/94', datetime(1994, 3, 3)),
])
def test_preferably_past_dates(self, date_string, expected):
self.given_parser(
settings={
'PREFER_DATES_FROM': 'past',
'RELATIVE_BASE': datetime(2015, 2, 15, 15, 30)
})
self.when_date_is_parsed(date_string)
self.then_date_was_parsed_by_date_parser()
self.then_date_obj_exactly_is(expected)
@parameterized.expand([
param('10 December', datetime(2015, 12, 10)),
param('March', datetime(2015, 3, 15)),
param('Friday', datetime(2015, 2, 20)),
param('Sunday', datetime(2015, 2, 22)), # current day
param('Monday', datetime(2015, 2, 16)),
param('10:00PM', datetime(2015, 2, 15, 22, 0)),
param('16:10', datetime(2015, 2, 15, 16, 10)),
param('14:05', datetime(2015, 2, 16, 14, 5)),
param('3/3/50', datetime(2050, 3, 3)),
param('3/3/94', datetime(2094, 3, 3)),
])
def test_preferably_future_dates(self, date_string, expected):
self.given_local_tz_offset(0)
self.given_parser(
settings={
'PREFER_DATES_FROM': 'future',
'RELATIVE_BASE': datetime(2015, 2, 15, 15, 30)
})
self.when_date_is_parsed(date_string)
self.then_date_was_parsed_by_date_parser()
self.then_date_obj_exactly_is(expected)
@parameterized.expand([
param('10 December', datetime(2015, 12, 10)),
param('March', datetime(2015, 3, 15)),
param('Friday', datetime(2015, 2, 13)),
param('Sunday', datetime(2015, 2, 15)), # current weekday
param('10:00PM', datetime(2015, 2, 15, 22, 00)),
param('16:10', datetime(2015, 2, 15, 16, 10)),
param('14:05', datetime(2015, 2, 15, 14, 5)),
])
def test_dates_without_preference(self, date_string, expected):
self.given_local_tz_offset(0)
self.given_parser(
settings={
'PREFER_DATES_FROM': 'current_period',
'RELATIVE_BASE': datetime(2015, 2, 15, 15, 30)
})
self.when_date_is_parsed(date_string)
self.then_date_was_parsed_by_date_parser()
self.then_date_obj_exactly_is(expected)
@parameterized.expand([
param('February 2015',
today=datetime(2015, 1, 31),
expected=datetime(2015, 2, 28)),
param('February 2012',
today=datetime(2015, 1, 31),
expected=datetime(2012, 2, 29)),
param('March 2015',
today=datetime(2015, 1, 25),
expected=datetime(2015, 3, 25)),
param('April 2015',
today=datetime(2015, 1, 31),
expected=datetime(2015, 4, 30)),
param('April 2015',
today=datetime(2015, 2, 28),
expected=datetime(2015, 4, 28)),
param('December 2014',
today=datetime(2015, 2, 15),
expected=datetime(2014, 12, 15)),
])
def test_dates_with_day_missing_preferring_current_day_of_month(
self, date_string, today=None, expected=None):
self.given_parser(settings={
'PREFER_DAY_OF_MONTH': 'current',
'RELATIVE_BASE': today
})
self.when_date_is_parsed(date_string)
self.then_date_was_parsed_by_date_parser()
self.then_date_obj_exactly_is(expected)
@parameterized.expand([
param('February 2015',
today=datetime(2015, 1, 1),
expected=datetime(2015, 2, 28)),
param('February 2012',
today=datetime(2015, 1, 1),
expected=datetime(2012, 2, 29)),
param('March 2015',
today=datetime(2015, 1, 25),
expected=datetime(2015, 3, 31)),
param('April 2015',
today=datetime(2015, 1, 15),
expected=datetime(2015, 4, 30)),
param('April 2015',
today=datetime(2015, 2, 28),
expected=datetime(2015, 4, 30)),
param('December 2014',
today=datetime(2015, 2, 15),
expected=datetime(2014, 12, 31)),
])
def test_dates_with_day_missing_preferring_last_day_of_month(
self, date_string, today=None, expected=None):
self.given_parser(settings={
'PREFER_DAY_OF_MONTH': 'last',
'RELATIVE_BASE': today
})
self.when_date_is_parsed(date_string)
self.then_date_was_parsed_by_date_parser()
self.then_date_obj_exactly_is(expected)
@parameterized.expand([
param('February 2015',
today=datetime(2015, 1, 8),
expected=datetime(2015, 2, 1)),
param('February 2012',
today=datetime(2015, 1, 7),
expected=datetime(2012, 2, 1)),
param('March 2015',
today=datetime(2015, 1, 25),
expected=datetime(2015, 3, 1)),
param('April 2015',
today=datetime(2015, 1, 15),
expected=datetime(2015, 4, 1)),
param('April 2015',
today=datetime(2015, 2, 28),
expected=datetime(2015, 4, 1)),
param('December 2014',
today=datetime(2015, 2, 15),
expected=datetime(2014, 12, 1)),
])
def test_dates_with_day_missing_preferring_first_day_of_month(
self, date_string, today=None, expected=None):
self.given_parser(settings={
'PREFER_DAY_OF_MONTH': 'first',
'RELATIVE_BASE': today
})
self.when_date_is_parsed(date_string)
self.then_date_was_parsed_by_date_parser()
self.then_date_obj_exactly_is(expected)
@parameterized.expand([
param(prefer_day_of_month='current'),
param(prefer_day_of_month='last'),
param(prefer_day_of_month='first'),
])
def test_that_day_preference_does_not_affect_dates_with_explicit_day(
self, prefer_day_of_month=None):
self.given_parser(
settings={
'PREFER_DAY_OF_MONTH': prefer_day_of_month,
'RELATIVE_BASE': datetime(2015, 2, 12)
})
self.when_date_is_parsed('24 April 2012')
self.then_date_was_parsed_by_date_parser()
self.then_date_obj_exactly_is(datetime(2012, 4, 24))
def test_date_is_parsed_when_skip_tokens_are_supplied(self):
self.given_parser(settings={
'SKIP_TOKENS': ['de'],
'RELATIVE_BASE': datetime(2015, 2, 12)
})
self.when_date_is_parsed('24 April 2012 de')
self.then_date_was_parsed_by_date_parser()
self.then_date_obj_exactly_is(datetime(2012, 4, 24))
@parameterized.expand([
param('29 February 2015', 'day must be in 1..28'),
param('32 January 2015', 'day must be in 1..31'),
param('31 April 2015', 'day must be in 1..30'),
param('31 June 2015', 'day must be in 1..30'),
param('31 September 2015', 'day must be in 1..30'),
])
def test_error_should_be_raised_for_invalid_dates_with_too_large_day_number(
self, date_string, message):
self.when_date_is_parsed_by_date_parser(date_string)
self.then_error_was_raised(ValueError,
['day is out of range for month', message])
@parameterized.expand([
param('2015-05-02T10:20:19+0000',
languages=['fr'],
expected=datetime(2015, 5, 2, 10, 20, 19)),
param('2015-05-02T10:20:19+0000',
languages=['en'],
expected=datetime(2015, 5, 2, 10, 20, 19)),
])
def test_iso_datestamp_format_should_always_parse(self, date_string,
languages, expected):
self.given_local_tz_offset(0)
self.given_parser(languages=languages,
settings={'PREFER_LOCALE_DATE_ORDER': False})
self.when_date_is_parsed(date_string)
self.then_date_was_parsed_by_date_parser()
self.result['date_obj'] = self.result['date_obj'].replace(tzinfo=None)
self.then_date_obj_exactly_is(expected)
@parameterized.expand([
# Epoch timestamps.
param('1484823450', expected=datetime(2017, 1, 19, 10, 57, 30)),
param('1436745600000', expected=datetime(2015, 7, 13, 0, 0)),
param('1015673450', expected=datetime(2002, 3, 9, 11, 30, 50)),
param('2016-09-23T02:54:32.845Z',
expected=datetime(2016,
9,
23,
2,
54,
32,
845000,
tzinfo=StaticTzInfo('Z', timedelta(0))))
])
def test_parse_timestamp(self, date_string, expected):
self.given_local_tz_offset(0)
self.given_parser(settings={'TO_TIMEZONE': 'UTC'})
self.when_date_is_parsed(date_string)
self.then_date_obj_exactly_is(expected)
@parameterized.expand([
param('10 December', expected=datetime(2015, 12, 10), period='day'),
param('March', expected=datetime(2015, 3, 15), period='month'),
param('April', expected=datetime(2015, 4, 15), period='month'),
param('December', expected=datetime(2015, 12, 15), period='month'),
param('Friday', expected=datetime(2015, 2, 13), period='day'),
param('Monday', expected=datetime(2015, 2, 9), period='day'),
param('10:00PM', expected=datetime(2015, 2, 15, 22, 00), period='day'),
param('16:10', expected=datetime(2015, 2, 15, 16, 10), period='day'),
param('2014', expected=datetime(2014, 2, 15), period='year'),
param('2008', expected=datetime(2008, 2, 15), period='year'),
])
def test_extracted_period(self, date_string, expected=None, period=None):
self.given_local_tz_offset(0)
self.given_parser(
settings={'RELATIVE_BASE': datetime(2015, 2, 15, 15, 30)})
self.when_date_is_parsed(date_string)
self.then_date_was_parsed_by_date_parser()
self.then_date_obj_exactly_is(expected)
self.then_period_is(period)
@parameterized.expand([
param('12th December 2019 19:00',
expected=datetime(2019, 12, 12, 19, 0),
period='time'),
param('9 Jan 11 0:00',
expected=datetime(2011, 1, 9, 0, 0),
period='time'),
])
def test_period_is_time_if_return_time_as_period_setting_applied_and_time_component_present(
self, date_string, expected=None, period=None):
self.given_parser(settings={'RETURN_TIME_AS_PERIOD': True})
self.when_date_is_parsed(date_string)
self.then_date_was_parsed_by_date_parser()
self.then_date_obj_exactly_is(expected)
self.then_period_is(period)
@parameterized.expand([
param('16:00', expected=datetime(2018, 12, 13, 16, 0), period='time'),
param('Monday 7:15 AM',
expected=datetime(2018, 12, 10, 7, 15),
period='time'),
])
def test_period_is_time_if_return_time_as_period_and_relative_base_settings_applied_and_time_component_present(
self, date_string, expected=None, period=None):
self.given_parser(
settings={
'RETURN_TIME_AS_PERIOD': True,
'RELATIVE_BASE': datetime(2018, 12, 13, 15, 15)
})
self.when_date_is_parsed(date_string)
self.then_date_was_parsed_by_date_parser()
self.then_date_obj_exactly_is(expected)
self.then_period_is(period)
@parameterized.expand([
param('12th March 2010',
expected=datetime(2010, 3, 12, 0, 0),
period='day'),
param('21-12-19', expected=datetime(2019, 12, 21, 0, 0), period='day'),
])
def test_period_is_day_if_return_time_as_period_setting_applied_and_time_component_is_not_present(
self, date_string, expected=None, period=None):
self.given_parser(settings={'RETURN_TIME_AS_PERIOD': True})
self.when_date_is_parsed(date_string)
self.then_date_was_parsed_by_date_parser()
self.then_date_obj_exactly_is(expected)
self.then_period_is(period)
@parameterized.expand([
param('16:00', expected=datetime(2017, 1, 10, 16, 0), period='day'),
param('Monday 7:15 AM',
expected=datetime(2017, 1, 9, 7, 15),
period='day'),
])
def test_period_is_day_if_return_time_as_period_setting_not_applied(
self, date_string, expected=None, period=None):
self.given_parser(
settings={
'RETURN_TIME_AS_PERIOD': False,
'RELATIVE_BASE': datetime(2017, 1, 10, 15, 15)
})
self.when_date_is_parsed(date_string)
self.then_date_was_parsed_by_date_parser()
self.then_date_obj_exactly_is(expected)
self.then_period_is(period)
@parameterized.expand([
param('15-12-18 06:00',
expected=datetime(2015, 12, 18, 6, 0),
order='YMD'),
param('15-18-12 06:00',
expected=datetime(2015, 12, 18, 6, 0),
order='YDM'),
param('10-11-12 06:00',
expected=datetime(2012, 10, 11, 6, 0),
order='MDY'),
param('10-11-12 06:00',
expected=datetime(2011, 10, 12, 6, 0),
order='MYD'),
param('10-11-12 06:00',
expected=datetime(2011, 12, 10, 6, 0),
order='DYM'),
param('15-12-18 06:00',
expected=datetime(2018, 12, 15, 6, 0),
order='DMY'),
param('12/09/08 04:23:15.567',
expected=datetime(2008, 9, 12, 4, 23, 15, 567000),
order='DMY'),
param('10/9/1914 03:07:09.788888 pm',
expected=datetime(1914, 10, 9, 15, 7, 9, 788888),
order='MDY'),
param('1-8-09 07:12:49 AM',
expected=datetime(2009, 1, 8, 7, 12, 49),
order='MDY'),
param('201508', expected=datetime(2015, 8, 20, 0, 0), order='DYM'),
param('201508', expected=datetime(2020, 8, 15, 0, 0), order='YDM'),
param('201108', expected=datetime(2008, 11, 20, 0, 0), order='DMY'),
param('2016 july 13.',
expected=datetime(2016, 7, 13, 0, 0),
order='YMD'),
param('16 july 13.', expected=datetime(2016, 7, 13, 0, 0),
order='YMD'),
param('Sunday 23 May 1856 12:09:08 AM',
expected=datetime(1856, 5, 23, 0, 9, 8),
order='DMY'),
])
def test_order(self, date_string, expected=None, order=None):
self.given_parser(settings={'DATE_ORDER': order})
self.when_date_is_parsed(date_string)
self.then_date_was_parsed_by_date_parser()
self.then_date_obj_exactly_is(expected)
@parameterized.expand([
param('10.1.2019',
expected=datetime(2019, 1, 10, 0, 0),
languages=['de'],
settings={'PREFER_DAY_OF_MONTH': 'first'}),
param('10.1.2019',
expected=datetime(2019, 1, 10, 0, 0),
languages=['de']),
param('10.1.2019',
expected=datetime(2019, 10, 1, 0, 0),
settings={'DATE_ORDER': 'MDY'}),
param('03/11/2559 05:13',
datetime(2559, 3, 11, 5, 13),
languages=["th"],
settings={"DATE_ORDER": "MDY"}),
param('03/15/2559 05:13',
datetime(2559, 3, 15, 5, 13),
languages=["th"],
settings={"DATE_ORDER": "MDY"})
])
def test_if_settings_provided_date_order_is_retained(
self, date_string, expected=None, languages=None, settings=None):
self.given_parser(languages=languages, settings=settings)
self.when_date_is_parsed(date_string)
self.then_date_was_parsed_by_date_parser()
self.then_date_obj_exactly_is(expected)
@parameterized.expand([
param('::', None),
param('..', None),
param(' ', None),
param('--', None),
param('//', None),
param('++', None),
])
def test_parsing_strings_containing_only_separator_tokens(
self, date_string, expected):
self.given_parser()
self.when_date_is_parsed(date_string)
self.then_period_is('day')
self.then_date_obj_exactly_is(expected)
def given_local_tz_offset(self, offset):
self.add_patch(
patch.object(dateparser.timezone_parser,
'local_tz_offset',
new=timedelta(seconds=3600 * offset)))
def given_parser(self, *args, **kwds):
def collecting_get_date_data(parse):
@wraps(parse)
def wrapped(*args, **kwargs):
self.date_result = parse(*args, **kwargs)
return self.date_result
return wrapped
self.add_patch(
patch.object(date_parser, 'parse',
collecting_get_date_data(date_parser.parse)))
self.date_parser = Mock(wraps=date_parser)
self.add_patch(
patch('dateparser.date.date_parser', new=self.date_parser))
self.parser = DateDataParser(*args, **kwds)
def when_date_is_parsed(self, date_string):
self.result = self.parser.get_date_data(date_string)
def when_date_is_parsed_by_date_parser(self, date_string):
try:
self.result = DateParser().parse(date_string)
except Exception as error:
self.error = error
def then_period_is(self, period):
self.assertEqual(period, self.result['period'])
def then_date_obj_exactly_is(self, expected):
self.assertEqual(expected, self.result['date_obj'])
def then_date_was_parsed_by_date_parser(self):
self.assertNotEqual(NotImplemented, self.date_result,
"Date was not parsed")
self.assertEqual(self.result['date_obj'], self.date_result[0])
def then_timezone_parsed_is(self, tzstr):
self.assertTrue(tzstr in repr(self.result['date_obj'].tzinfo))
self.result['date_obj'] = self.result['date_obj'].replace(tzinfo=None)
def test_attribute_before():
result = parse('attr BEFORE 2000-01-01T00:00:01Z')
assert result == ast.TimeBefore(
ast.Attribute('attr'),
datetime(2000, 1, 1, 0, 0, 1, tzinfo=StaticTzInfo('Z', timedelta(0))),
)
class TestTranslateSearch(BaseTestCase):
def setUp(self):
super(TestTranslateSearch, self).setUp()
self.search_with_detection = DateSearchWithDetection()
self.exact_language_search = self.search_with_detection.search
@staticmethod
def make_python3_msg(text):
text = text.replace('unicode', 'str')
text = text.replace('u\'', '\'')
text = text.replace('type', 'class')
return text
def run_search_dates_function_invalid_languages(self, text, languages,
error_type):
try:
search_dates(text=text, languages=languages)
except Exception as error:
self.error = error
self.assertIsInstance(self.error, error_type)
def check_error_message(self, message):
self.assertEqual(self.make_python3_msg(str(self.error)), message)
@parameterized.expand([
param('en', "Sep 03 2014"),
param('en', "friday, 03 september 2014"),
param('en', 'Aug 06, 2018 05:05 PM CDT'),
# Chinese
param('zh', "1年11个月"),
param('zh', "1年11個月"),
param('zh', "2015年04月08日10点05"),
param('zh', "2015年04月08日10:05"),
param('zh', "2013年04月08日"),
param('zh', "周一"),
param('zh', "礼拜一"),
param('zh', "周二"),
param('zh', "礼拜二"),
param('zh', "周三"),
param('zh', "礼拜三"),
param('zh', "星期日 2015年04月08日10:05"),
param('zh', "周六 2013年04月08日"),
param('zh', "下午3:30"),
param('zh', "凌晨3:30"),
param('zh', "中午"),
# French
param('fr', "20 Février 2012"),
param('fr', "Mercredi 19 Novembre 2013"),
param('fr', "18 octobre 2012 à 19 h 21 min"),
# German
param('de', "29. Juni 2007"),
param('de', "Montag 5 Januar, 2015"),
# Hungarian
param('hu', '2016 augusztus 11'),
param('hu', '2016-08-13 szombat 10:21'),
param('hu', '2016. augusztus 14. vasárnap 10:21'),
param('hu', 'hétfő'),
param('hu', 'tegnapelőtt'),
param('hu', 'ma'),
param('hu', '2 hónappal ezelőtt'),
param('hu', '2016-08-13 szombat 10:21 GMT'),
# Spanish
param('es', "Miércoles 31 Diciembre 2014"),
# Italian
param('it', "Giovedi Maggio 29 2013"),
param('it', "19 Luglio 2013"),
# Portuguese
param('pt', "22 de dezembro de 2014 às 02:38"),
# Russian
param('ru', "5 августа 2014 г в 12:00"),
# Real: param('ru', "5 августа 2014 г. в 12:00"),
# Turkish
param('tr', "2 Ocak 2015 Cuma, 16:49"),
# Czech
param('cs', "22. prosinec 2014 v 2:38"),
# Dutch
param('nl', "maandag 22 december 2014 om 2:38"),
# Romanian
param('ro', "22 Decembrie 2014 la 02:38"),
# Polish
param('pl', "4 stycznia o 13:50"),
param('pl', "29 listopada 2014 o 08:40"),
# Ukrainian
param('uk', "30 листопада 2013 о 04:27"),
# Belarusian
param('be', "5 снежня 2015 г у 12:00"),
# Real: param('be', "5 снежня 2015 г. у 12:00"), Issue: Abbreviation segmentation.
param('be', "11 верасня 2015 г у 12:11"),
# Real: param('be', "11 верасня 2015 г. у 12:11"),
param('be', "3 стд 2015 г у 10:33"),
# Real: param('be', "3 стд 2015 г. у 10:33"),
# Arabic
param('ar', "6 يناير، 2015، الساعة 05:16 مساءً"),
param('ar', "7 يناير، 2015، الساعة 11:00 صباحاً"),
# Vietnamese
# Disabled - wrong segmentation at "Thứ Năm"
# param('vi', "Thứ Năm, ngày 8 tháng 1 năm 2015"),
# Disabled - wrong segmentation at "Thứ Tư"
# param('vi', "Thứ Tư, 07/01/2015 | 22:34"),
param('vi', "9 Tháng 1 2015 lúc 15:08"),
# Thai
# Disabled - spacing differences
# param('th', "เมื่อ กุมภาพันธ์ 09, 2015, 09:27:57 AM"),
# param('th', "เมื่อ กรกฎาคม 05, 2012, 01:18:06 AM"),
# Tagalog
param('tl', "Biyernes Hulyo 3, 2015"),
param('tl', "Pebrero 5, 2015 7:00 pm"),
# Indonesian
param('id', "06 Sep 2015"),
param('id', "07 Feb 2015 20:15"),
# Miscellaneous
param('en', "2014-12-12T12:33:39-08:00"),
param('en', "2014-10-15T16:12:20+00:00"),
param('en', "28 Oct 2014 16:39:01 +0000"),
# Disabled - wrong split at "a las".
# param('es', "13 Febrero 2015 a las 23:00"),
# Danish
param('da', "Sep 03 2014"),
param('da', "fredag, 03 september 2014"),
param('da', "fredag d. 3 september 2014"),
# Finnish
param('fi', "maanantai tammikuu 16, 2015"),
param('fi', "ma tammi 16, 2015"),
param('fi', "tiistai helmikuu 16, 2015"),
param('fi', "ti helmi 16, 2015"),
param('fi', "keskiviikko maaliskuu 16, 2015"),
param('fi', "ke maalis 16, 2015"),
param('fi', "torstai huhtikuu 16, 2015"),
param('fi', "to huhti 16, 2015"),
param('fi', "perjantai toukokuu 16, 2015"),
param('fi', "pe touko 16, 2015"),
param('fi', "lauantai kesäkuu 16, 2015"),
param('fi', "la kesä 16, 2015"),
param('fi', "sunnuntai heinäkuu 16, 2015"),
param('fi', "su heinä 16, 2015"),
param('fi', "su elokuu 16, 2015"),
param('fi', "su elo 16, 2015"),
param('fi', "su syyskuu 16, 2015"),
param('fi', "su syys 16, 2015"),
param('fi', "su lokakuu 16, 2015"),
param('fi', "su loka 16, 2015"),
param('fi', "su marraskuu 16, 2015"),
param('fi', "su marras 16, 2015"),
param('fi', "su joulukuu 16, 2015"),
param('fi', "su joulu 16, 2015"),
param('fi', "1. tammikuuta, 2016"),
param('fi', "tiistaina, 27. lokakuuta 2015"),
# Japanese
param('ja', "午後3時"),
param('ja', "2時"),
param('ja', "11時42分"),
param('ja', "3ヶ月"),
param('ja', "約53か月前"),
param('ja', "3月"),
param('ja', "十二月"),
param('ja', "2月10日"),
param('ja', "2013年2月"),
param('ja', "2013年04月08日"),
param('ja', "2016年03月24日 木曜日 10時05分"),
param('ja', "2016年3月20日 21時40分"),
param('ja', "2016年03月21日 23時05分11秒"),
param('ja', "2016年3月21日(月) 14時48分"),
param('ja', "2016年3月20日(日) 21時40分"),
param('ja', "2016年3月20日 (日) 21時40分"),
# Hebrew
param('he', "20 לאפריל 2012"),
param('he', "יום רביעי ה-19 בנובמבר 2013"),
param('he', "18 לאוקטובר 2012 בשעה 19:21"),
# Disabled - wrong split at "יום ה'".
# param('he', "יום ה' 6/10/2016"),
param('he', "חצות"),
param('he', "1 אחר חצות"),
param('he', "3 לפנות בוקר"),
param('he', "3 בבוקר"),
param('he', "3 בצהריים"),
param('he', "6 לפנות ערב"),
param('he', "6 אחרי הצהריים"),
param('he', "6 אחרי הצהרים"),
# Bangla
param('bn', "সেপ্টেম্বর 03 2014"),
param('bn', "শুক্রবার, 03 সেপ্টেম্বর 2014"),
# Hindi
param('hi', 'सोमवार 13 जून 1998'),
param('hi', 'मंगल 16 1786 12:18'),
param('hi', 'शनि 11 अप्रैल 2002 03:09'),
# Swedish
param('sv', "Sept 03 2014"),
param('sv', "fredag, 03 september 2014"),
])
def test_search_date_string(self, shortname, datetime_string):
result = self.exact_language_search.search(shortname,
datetime_string,
settings=Settings())[1][0]
self.assertEqual(result, datetime_string)
@parameterized.expand([
# Arabic
param(
'ar', 'في 29 يوليو 1938 غزت القوات اليابانية الاتحاد'
' السوفييتي ووقعت أولى المعارك والتي انتصر فيها السوفييت، وعلى الرغم من ذلك رفضت'
' اليابان الاعتراف بذلك وقررت في 11 مايو 1939 تحريك الحدود المنغولية حتى نهر غول،'
' حيث وقعت معركة خالخين غول والتي انتصر فيها الجيش الأحمر على جيش كوانتونغ',
[('في 29 يوليو 1938', datetime.datetime(1938, 7, 29, 0, 0)),
('في 11 مايو 1939', datetime.datetime(1939, 5, 11, 0, 0))],
settings={'RELATIVE_BASE': datetime.datetime(2000, 1, 1)}),
# Belarusian
param(
'be',
'Пасля апублікавання Патсдамскай дэкларацыі 26 ліпеня 1945 года і адмовы Японіі капітуляваць '
'на яе ўмовах ЗША скінулі атамныя бомбы.',
[('26 ліпеня 1945 года і', datetime.datetime(1945, 7, 26, 0, 0))],
settings={'RELATIVE_BASE': datetime.datetime(2000, 1, 1)}),
# Bulgarian
param(
'bg', 'На 16 юни 1944 г. започват въздушни '
'бомбардировки срещу Япония, използувайки новозавладените острови като бази.',
[('На 16 юни 1944 г', datetime.datetime(1944, 6, 16, 0, 0))],
settings={'RELATIVE_BASE': datetime.datetime(2000, 1, 1)}),
# Chinese
param('zh',
'不過大多數人仍多把第二次世界大戰的爆發定為1939年9月1日德國入侵波蘭開始，這次入侵行動隨即導致英國與法國向德國宣戰。',
[('1939年9月1', datetime.datetime(1939, 9, 1, 0, 0))],
settings={'RELATIVE_BASE': datetime.datetime(2000, 1, 1)}),
# Czech
param(
'cs',
'V roce 1920 byla proto vytvořena Společnost národů, jež měla fungovat jako fórum, '
'na němž měly národy mírovým způsobem urovnávat svoje spory.',
[('1920', datetime.datetime(1920, 1, 1, 0, 0))],
settings={'RELATIVE_BASE': datetime.datetime(2000, 1, 1)}),
# Danish
param(
'da',
'Krigen i Europa begyndte den 1. september 1939, da Nazi-Tyskland invaderede Polen, '
'og endte med Nazi-Tysklands betingelsesløse overgivelse den 8. maj 1945.',
[('1. september 1939', datetime.datetime(1939, 9, 1, 0, 0)),
('8. maj 1945', datetime.datetime(1945, 5, 8, 0, 0))],
settings={'RELATIVE_BASE': datetime.datetime(2000, 1, 1)}),
# Dutch
param(
'nl',
' De meest dramatische uitbreiding van het conflict vond plaats op 22 juni 1941 met de '
'Duitse aanval op de Sovjet-Unie.',
[('22 juni 1941', datetime.datetime(1941, 6, 22, 0, 0))],
settings={'RELATIVE_BASE': datetime.datetime(2000, 1, 1)}),
# English
param('en',
'I will meet you tomorrow at noon',
[('tomorrow at noon', datetime.datetime(2000, 1, 2, 12, 0))],
settings={'RELATIVE_BASE': datetime.datetime(2000, 1, 1)}),
param('en',
'in a minute',
[('in a minute', datetime.datetime(2000, 1, 1, 0, 1))],
settings={'RELATIVE_BASE': datetime.datetime(2000, 1, 1)}),
param('en',
'July 13th.\r\n July 14th',
[('July 13th', datetime.datetime(2000, 7, 13, 0, 0)),
('July 14th', datetime.datetime(2000, 7, 14, 0, 0))],
settings={'RELATIVE_BASE': datetime.datetime(2000, 1, 1)}),
param('en',
'last updated Aug 06, 2018 05:05 PM CDT',
[('Aug 06, 2018 05:05 PM CDT',
datetime.datetime(2018,
8,
6,
17,
5,
tzinfo=StaticTzInfo(
'CDT',
datetime.timedelta(seconds=-18000))))],
settings={'RELATIVE_BASE': datetime.datetime(2000, 1, 1)}),
param('en',
'25th march 2015 , i need this report today.',
[('25th march 2015', datetime.datetime(2015, 3, 25))],
settings={
'PARSERS': [
parser for parser in default_parsers
if parser != 'relative-time'
]
}),
param('en',
'25th march 2015 , i need this report today.',
[('25th march 2015', datetime.datetime(2015, 3, 25)),
('today', datetime.datetime(2000, 1, 1))],
settings={'RELATIVE_BASE': datetime.datetime(2000, 1, 1)}),
# Filipino / Tagalog
param(
'tl',
'Maraming namatay sa mga Hapon hanggang sila\'y sumuko noong Agosto 15, 1945.',
[('noong Agosto 15, 1945', datetime.datetime(1945, 8, 15, 0, 0))],
settings={'RELATIVE_BASE': datetime.datetime(2000, 1, 1)}),
# Finnish
param(
'fi',
'Iso-Britannia ja Ranska julistivat sodan Saksalle 3. syyskuuta 1939.',
[('3. syyskuuta 1939', datetime.datetime(1939, 9, 3, 0, 0))],
settings={'RELATIVE_BASE': datetime.datetime(2000, 1, 1)}),
# French
param(
'fr',
'La 2e Guerre mondiale, ou Deuxième Guerre mondiale4, est un conflit armé à '
'l\'échelle planétaire qui dura du 1 septembre 1939 au 2 septembre 1945.',
[('1 septembre 1939', datetime.datetime(1939, 9, 1, 0, 0)),
('2 septembre 1945', datetime.datetime(1945, 9, 2, 0, 0))],
settings={'RELATIVE_BASE': datetime.datetime(2000, 1, 1)}),
# Hebrew
param('he',
'במרץ 1938 "אוחדה" אוסטריה עם גרמניה (אנשלוס). ',
[('במרץ 1938', datetime.datetime(1938, 3, 1, 0, 0))],
settings={'RELATIVE_BASE': datetime.datetime(2000, 1, 1)}),
# Hindi
param(
'hi',
'जुलाई 1937 में, मार्को-पोलो ब्रिज हादसे का बहाना लेकर जापान ने चीन पर हमला कर दिया और चीनी साम्राज्य की राजधानी बीजिंग '
'पर कब्जा कर लिया,',
[('जुलाई 1937 में', datetime.datetime(1937, 7, 1, 0, 0))],
settings={'RELATIVE_BASE': datetime.datetime(2000, 1, 1)}),
# Hungarian
param(
'hu',
'A háború Európában 1945. május 8-án Németország feltétel nélküli megadásával, '
'míg Ázsiában szeptember 2-án, Japán kapitulációjával fejeződött be.',
[('1945. május 8-án', datetime.datetime(1945, 5, 8, 0, 0)),
('szeptember 2-án', datetime.datetime(2000, 9, 2, 0, 0))],
settings={'RELATIVE_BASE': datetime.datetime(2000, 1, 1)}),
# Georgian
param('ka',
'1937 წელს დაიწყო იაპონია-ჩინეთის მეორე ომი.',
[('1937', datetime.datetime(1937, 1, 1, 0, 0))],
settings={'RELATIVE_BASE': datetime.datetime(2000, 1, 1)}),
# German
param('de', 'Die UdSSR blieb gemäß dem Neutralitätspakt '
'vom 13. April 1941 gegenüber Japan vorerst neutral.',
[('Die', datetime.datetime(1999, 12, 28, 0, 0)),
('13. April 1941', datetime.datetime(1941, 4, 13, 0, 0))],
settings={'RELATIVE_BASE': datetime.datetime(2000, 1, 1)}),
# Indonesian
param(
'id',
'Kekaisaran Jepang menyerah pada tanggal 15 Agustus 1945, sehingga mengakhiri perang '
'di Asia dan memperkuat kemenangan total Sekutu atas Poros.',
[('tanggal 15 Agustus 1945', datetime.datetime(1945, 8, 15, 0, 0))
],
settings={'RELATIVE_BASE': datetime.datetime(2000, 1, 1)}),
# Italian
param('it', ' Con questo il 2 ottobre 1935 prese il via la campagna '
'd\'Etiopia. Il 9 maggio 1936 venne proclamato l\'Impero. ',
[('2 ottobre 1935', datetime.datetime(1935, 10, 2, 0, 0)),
('9 maggio 1936', datetime.datetime(1936, 5, 9, 0, 0))],
settings={'RELATIVE_BASE': datetime.datetime(2000, 1, 1)}),
# Japanese
param('ja',
'1939年9月1日、ドイツ軍がポーランドへ侵攻したことが第二次世界大戦の始まりとされている。',
[('1939年9月1', datetime.datetime(1939, 9, 1, 0, 0))],
settings={'RELATIVE_BASE': datetime.datetime(2000, 1, 1)}),
# Persian
param('fa',
'نگ جهانی دوم جنگ جدی بین سپتامبر 1939 و 2 سپتامبر 1945 بود.',
[('سپتامبر 1939', datetime.datetime(1939, 9, 1, 0, 0)),
('2 سپتامبر 1945', datetime.datetime(1945, 9, 2, 0, 0))],
settings={'RELATIVE_BASE': datetime.datetime(2000, 1, 1)}),
# Polish
param(
'pl', 'II wojna światowa – największa wojna światowa w historii, '
'trwająca od 1 września 1939 do 2 września 1945 (w Europie do 8 maja 1945)',
[('1 września 1939', datetime.datetime(1939, 9, 1, 0, 0)),
('2 września 1945 (w', datetime.datetime(1945, 9, 2, 0, 0)),
('8 maja 1945', datetime.datetime(1945, 5, 8, 0, 0))],
settings={'RELATIVE_BASE': datetime.datetime(2000, 1, 1)}),
# Portuguese
param(
'pt',
'Em outubro de 1936, Alemanha e Itália formaram o Eixo Roma-Berlim.',
[('Em outubro de 1936', datetime.datetime(1936, 10, 1, 0, 0))],
settings={'RELATIVE_BASE': datetime.datetime(2000, 1, 1)}),
# Romanian
param(
'ro',
'Pe 17 septembrie 1939, după semnarea unui acord de încetare a focului cu Japonia, '
'sovieticii au invadat Polonia dinspre est.',
[('17 septembrie 1939', datetime.datetime(1939, 9, 17, 0, 0))],
settings={'RELATIVE_BASE': datetime.datetime(2000, 1, 1)}),
# Russian
param(
'ru',
'Втора́я мирова́я война́ (1 сентября 1939 — 2 сентября 1945) — '
'война двух мировых военно-политических коалиций, ставшая крупнейшим вооружённым '
'конфликтом в истории человечества.',
[('1 сентября 1939', datetime.datetime(1939, 9, 1, 0, 0)),
('2 сентября 1945', datetime.datetime(1945, 9, 2, 0, 0))],
settings={'RELATIVE_BASE': datetime.datetime(2000, 1, 1)}),
# Spanish
param(
'es',
'Desde finales de 1939 hasta inicios de 1941 Alemania conquistó o sometió '
'gran parte de la Europa continental.',
[('de 1939', datetime.datetime(1939, 1, 1, 0, 0)),
('de 1941', datetime.datetime(1941, 1, 1, 0, 0))],
settings={'RELATIVE_BASE': datetime.datetime(2000, 1, 1)}),
# Swedish
param(
'sv',
'Efter kommunisternas seger 1922 drog de allierade och Japan bort sina trupper.',
[('1922', datetime.datetime(1922, 1, 1, 0, 0))],
settings={'RELATIVE_BASE': datetime.datetime(2000, 1, 1)}),
# Thai
param(
'th',
'และเมื่อวันที่ 11 พฤษภาคม 1939 ญี่ปุ่นตัดสินใจขยายพรมแดนญี่ปุ่น-มองโกเลียขึ้นไปถึงแม่น้ำคัลคินกอลด้วยกำลัง',
[('11 พฤษภาคม 1939', datetime.datetime(1939, 5, 11, 0, 0))],
settings={'RELATIVE_BASE': datetime.datetime(2000, 1, 1)}),
# Turkish
param(
'tr',
'Almanya’nın Polonya’yı işgal ettiği 1 Eylül 1939 savaşın başladığı '
'tarih olarak genel kabul görür.',
[('1 Eylül 1939', datetime.datetime(1939, 9, 1, 0, 0))],
settings={'RELATIVE_BASE': datetime.datetime(2000, 1, 1)}),
# Ukrainian
param(
'uk',
'Інші дати, що розглядаються деякими авторами як дати початку війни: початок японської інтервенції '
'в Маньчжурію 13 вересня 1931, початок другої японсько-китайської війни 7 липня 1937 року та '
'початок угорсько-української війни 14 березня 1939 року.',
[('13 вересня 1931', datetime.datetime(1931, 9, 13, 0, 0)),
('7 липня 1937', datetime.datetime(1937, 7, 7, 0, 0)),
('14 березня 1939', datetime.datetime(1939, 3, 14, 0, 0))],
settings={'RELATIVE_BASE': datetime.datetime(2000, 1, 1)}),
# Vietnamese
param(
'vi',
'Ý theo gương Đức, đã tiến hành xâm lược Ethiopia năm 1935 và sát '
'nhập Albania vào ngày 12 tháng 4 năm 1939.',
[('năm 1935', datetime.datetime(1935, 1, 1, 0, 0)),
('ngày 12 tháng 4 năm 1939', datetime.datetime(1939, 4, 12, 0, 0))
],
settings={'RELATIVE_BASE': datetime.datetime(2000, 1, 1)}),
])
@apply_settings
def test_search_and_parse(self,
shortname,
string,
expected,
settings=None):
result = self.exact_language_search.search_parse(shortname,
string,
settings=settings)
self.assertEqual(result, expected)
@parameterized.expand([
# English
param('en', 'January 3, 2017 - February 1st',
[('January 3, 2017', datetime.datetime(2017, 1, 3, 0, 0)),
('February 1st', datetime.datetime(2017, 2, 1, 0, 0))]),
param(
'en', '2014 was good! October was excellent!'
' Friday, 21 was especially good!',
[('2014',
datetime.datetime(2014,
datetime.datetime.utcnow().month,
datetime.datetime.utcnow().day, 0, 0)),
('October',
datetime.datetime(2014, 10,
datetime.datetime.utcnow().day, 0, 0)),
('Friday, 21', datetime.datetime(2014, 10, 21, 0, 0))]),
# Russian
param(
'ru',
'19 марта 2001 был хороший день. 20 марта тоже был хороший день. 21 марта был отличный день.',
[('19 марта 2001', datetime.datetime(2001, 3, 19, 0, 0)),
('20 марта', datetime.datetime(2001, 3, 20, 0, 0)),
('21 марта', datetime.datetime(2001, 3, 21, 0, 0))]),
# relative dates
param(
'ru',
'19 марта 2001. Сегодня был хороший день. 2 дня назад был хороший день. '
'Вчера тоже был хороший день.',
[('19 марта 2001', datetime.datetime(2001, 3, 19, 0, 0)),
('Сегодня', datetime.datetime(2001, 3, 19, 0, 0)),
('2 дня назад', datetime.datetime(2001, 3, 17, 0, 0)),
('Вчера', datetime.datetime(2001, 3, 18, 0, 0))]),
# Hungarian
param(
'hu',
'1962 augusztus 11 Föld körüli pályára bocsátották a szovjet Vosztok-3 űrhajót, '
'mely páros űrrepülést hajtott végre a másnap föld körüli pályára bocsátott Vosztok-4-gyel.'
'2 hónappal ezelőtt furcsa, nem forgó jellegű szédülést tapasztaltam.',
[('1962 augusztus 11', datetime.datetime(1962, 8, 11, 0, 0)),
('2 hónappal ezelőtt', datetime.datetime(1962, 6, 11, 0, 0))]),
# Vietnamese
param(
'vi',
'1/1/1940. Vào tháng 8 năm 1940, với lực lượng lớn của Pháp tại Bắc Phi chính thức trung lập trong '
'cuộc chiến, Ý mở một cuộc tấn công vào thuộc địa Somalia của Anh tại Đông Phi. '
'Đến tháng 9 quân Ý vào đến Ai Cập (cũng đang dưới sự kiểm soát của Anh). ',
[('1/1/1940', datetime.datetime(1940, 1, 1, 0, 0)),
('tháng 8 năm 1940', datetime.datetime(1940, 8, 1, 0, 0)),
('tháng 9', datetime.datetime(1940, 9, 1, 0, 0))])
])
@apply_settings
def test_relative_base_setting(self,
shortname,
string,
expected,
settings=None):
result = self.exact_language_search.search_parse(shortname,
string,
settings=settings)
self.assertEqual(result, expected)
@parameterized.expand([
param('en', 'July 12th, 2014. July 13th, July 14th',
[('July 12th, 2014', datetime.datetime(2014, 7, 12, 0, 0)),
('July 13th', datetime.datetime(2014, 7, 13, 0, 0)),
('July 14th', datetime.datetime(2014, 7, 14, 0, 0))]),
param('en', '2014. July 13th July 14th',
[('2014',
datetime.datetime(2014,
datetime.datetime.utcnow().month,
datetime.datetime.utcnow().day, 0, 0)),
('July 13th', datetime.datetime(2014, 7, 13, 0, 0)),
('July 14th', datetime.datetime(2014, 7, 14, 0, 0))]),
param('en', 'July 13th 2014 July 14th 2014',
[('July 13th 2014', datetime.datetime(2014, 7, 13, 0, 0)),
('July 14th 2014', datetime.datetime(2014, 7, 14, 0, 0))]),
param('en', 'July 13th 2014 July 14th',
[('July 13th 2014', datetime.datetime(2014, 7, 13, 0, 0)),
('July 14th', datetime.datetime(2014, 7, 14, 0, 0))]),
param('en', 'July 13th, 2014 July 14th, 2014',
[('July 13th, 2014', datetime.datetime(2014, 7, 13, 0, 0)),
('July 14th, 2014', datetime.datetime(2014, 7, 14, 0, 0))]),
param('en', '2014. July 12th, July 13th, July 14th',
[('2014',
datetime.datetime(2014,
datetime.datetime.utcnow().month,
datetime.datetime.utcnow().day, 0, 0)),
('July 12th', datetime.datetime(2014, 7, 12, 0, 0)),
('July 13th', datetime.datetime(2014, 7, 13, 0, 0)),
('July 14th', datetime.datetime(2014, 7, 14, 0, 0))]),
# Swedish
param(
'sv',
'1938–1939 marscherade tyska soldater i Österrike samtidigt som '
'österrikiska soldater marscherade i Berlin.',
[('1938',
datetime.datetime(1938,
datetime.datetime.utcnow().month,
datetime.datetime.utcnow().day, 0, 0)),
('1939',
datetime.datetime(1939,
datetime.datetime.utcnow().month,
datetime.datetime.utcnow().day, 0, 0))]),
# German
param(
'de',
'Verteidiger der Stadt kapitulierten am 2. Mai 1945. Am 8. Mai 1945 (VE-Day) trat '
'bedingungslose Kapitulation der Wehrmacht in Kraft',
[('am 2. Mai 1945', datetime.datetime(1945, 5, 2, 0, 0)),
('Am 8. Mai 1945', datetime.datetime(1945, 5, 8, 0, 0))]),
])
@apply_settings
def test_splitting_of_not_parsed(self,
shortname,
string,
expected,
settings=None):
result = self.exact_language_search.search_parse(shortname,
string,
settings=settings)
self.assertEqual(result, expected)
@parameterized.expand([
# Arabic
param(
'ar', 'في 29 يوليو 1938 غزت القوات اليابانية الاتحاد'
' السوفييتي ووقعت أولى المعارك والتي انتصر فيها السوفييت، وعلى الرغم من ذلك رفضت'
' اليابان الاعتراف بذلك وقررت في 11 مايو 1939 تحريك الحدود المنغولية حتى نهر غول،'
),
# Belarusian
param(
'be',
'Пасля апублікавання Патсдамскай дэкларацыі 26 ліпеня 1945 года і адмовы Японіі капітуляваць '
'на яе ўмовах ЗША скінулі атамныя бомбы.'),
# Bulgarian
param(
'bg', 'На 16 юни 1944 г. започват въздушни '
'бомбардировки срещу Япония, използувайки новозавладените острови като бази.'
),
# Chinese
param('zh',
'不過大多數人仍多把第二次世界大戰的爆發定為1939年9月1日德國入侵波蘭開始，2015年04月08日10点05。'),
# Czech
param(
'cs',
'V rok 1920 byla proto vytvořena Společnost národů, jež měla fungovat jako fórum, '
'na němž měly národy mírovým způsobem urovnávat svoje spory.'),
# Danish
param(
'da',
'Krigen i Europa begyndte den 1. september 1939, da Nazi-Tyskland invaderede Polen, '
'og endte med Nazi-Tysklands betingelsesløse overgivelse den 8. marts 1945.'
),
# Dutch
param(
'nl',
' De meest dramatische uitbreiding van het conflict vond plaats op Maandag 22 juni 1941 met de '
'Duitse aanval op de Sovjet-Unie.'),
# English
param('en', 'I will meet you tomorrow at noon'),
# Filipino / Tagalog
param(
'tl',
'Maraming namatay sa mga Hapon hanggang sila\'y sumuko noong Agosto 15, 1945.'
),
# Finnish
param(
'fi',
'Iso-Britannia ja Ranska julistivat sodan Saksalle 3. syyskuuta 1939.'
),
# French
param(
'fr',
'La Seconde Guerre mondiale, ou Deuxième Guerre mondiale4, est un conflit armé à '
'l\'échelle planétaire qui dura du 1 septembre 1939 au 2 septembre 1945.'
),
# Hebrew
param('he', 'במרץ 1938 "אוחדה" אוסטריה עם גרמניה (אנשלוס). '),
# Hindi
param(
'hi',
'जुलाई 1937 में, मार्को-पोलो ब्रिज हादसे का बहाना लेकर जापान ने चीन पर हमला कर दिया और चीनी साम्राज्य की राजधानी बीजिंग '
'पर कब्जा कर लिया,'),
# Hungarian
param(
'hu',
'A háború Európában 1945. május 8-án Németország feltétel nélküli megadásával, '
'míg Ázsiában szeptember 2-án, Japán kapitulációjával fejeződött be.'
),
# Georgian
param('ka', '1937 წელს დაიწყო იაპონია-ჩინეთის მეორე ომი.'),
# German
param(
'de', 'Die UdSSR blieb dem Neutralitätspakt '
'vom 13. April 1941 gegenüber Japan vorerst neutral.'),
# Indonesian
param(
'id',
'Kekaisaran Jepang menyerah pada tanggal 15 Agustus 1945, sehingga mengakhiri perang '
'di Asia dan memperkuat kemenangan total Sekutu atas Poros.'),
# Italian
param(
'it', ' Con questo il 2 ottobre 1935 prese il via la campagna '
'd\'Etiopia. Il 9 maggio 1936 venne proclamato l\'Impero. '),
# Japanese
param('ja', '1933年（昭和8年）12月23日午前6時39分、宮城（現：皇居）内の産殿にて誕生。'),
# Persian
param('fa',
'نگ جهانی دوم جنگ جدی بین سپتامبر 1939 و 2 سپتامبر 1945 بود.'),
# Polish
param(
'pl', 'II wojna światowa – największa wojna światowa w historii, '
'trwająca od 1 września 1939 do 2 września 1945 (w Europie do 8 maja 1945)'
),
# Portuguese
param(
'pt',
'Em outubro de 1936, Alemanha e Itália formaram o Eixo Roma-Berlim.'
),
# Romanian
param(
'ro',
'Pe 17 septembrie 1939, după semnarea unui acord de încetare a focului cu Japonia, '
'sovieticii au invadat Polonia dinspre est.'),
# Russian
param(
'ru',
'Втора́я мирова́я война́ (1 сентября 1939 — 2 сентября 1945) — '
'война двух мировых военно-политических коалиций, ставшая крупнейшим вооружённым '
'конфликтом в истории человечества.'),
# Spanish
param('es', '11 junio 2010'),
# Swedish
param('sv', ' den 15 augusti 1945 då Kejsardömet'),
# Thai
param(
'th',
'และเมื่อวันที่ 11 พฤษภาคม 1939 ญี่ปุ่นตัดสินใจขยายพรมแดนญี่ปุ่น-มองโกเลียขึ้นไปถึงแม่น้ำคัลคินกอลด้วยกำลัง'
),
# Turkish
param(
'tr',
'Almanya’nın Polonya’yı işgal ettiği 1 Eylül 1939 savaşın başladığı '
'tarih olarak genel kabul görür.'),
# Ukrainian
param(
'uk',
'Інші дати, що розглядаються деякими авторами як дати початку війни: початок японської інтервенції '
'в Маньчжурію 13 вересня 1931, початок другої японсько-китайської війни 7 липня 1937 року та '
'початок угорсько-української війни 14 березня 1939 року.'),
# Vietnamese
param(
'vi',
'Ý theo gương Đức, đã tiến hành xâm lược Ethiopia năm 1935 và sát '
'nhập Albania vào ngày 12 tháng 4 năm 1939.'),
# only digits
param('en', '2007'),
])
def test_detection(self, shortname, text):
result = self.search_with_detection.detect_language(text,
languages=None)
self.assertEqual(result, shortname)
@parameterized.expand([
param(
text=
'19 марта 2001 был хороший день. 20 марта тоже был хороший день. 21 марта был отличный день.',
languages=['en', 'ru'],
settings=None,
expected=[('19 марта 2001', datetime.datetime(2001, 3, 19, 0, 0)),
('20 марта', datetime.datetime(2001, 3, 20, 0, 0)),
('21 марта', datetime.datetime(2001, 3, 21, 0, 0))]),
param(
text=
'Em outubro de 1936, Alemanha e Itália formaram o Eixo Roma-Berlim.',
languages=None,
settings={'RELATIVE_BASE': datetime.datetime(2000, 1, 1)},
expected=[('Em outubro de 1936',
datetime.datetime(1936, 10, 1, 0, 0))]),
param(text='19 марта 2001, 20 марта, 21 марта был отличный день.',
languages=['en', 'ru'],
settings=None,
expected=[('19 марта 2001', datetime.datetime(2001, 3, 19, 0,
0)),
('20 марта', datetime.datetime(2001, 3, 20, 0, 0)),
('21 марта', datetime.datetime(2001, 3, 21, 0, 0))]),
# dates not found
param(text='', languages=None, settings=None, expected=None),
# language not detected
param(text='Привет', languages=['en'], settings=None, expected=None),
# ZeroDivisionError
param(text='01.09 – 03.09.2017',
languages=None,
settings={'STRICT_PARSING': True},
expected=[('03.09.2017', datetime.datetime(2017, 3, 9, 0, 0))]),
param(text="DECEMBER 21 19.87 87",
languages=None,
settings=None,
expected=[('DECEMBER 21 19',
datetime.datetime(2019, 12, 21, 0, 0))]),
param(
text=
'bonjour, pouvez vous me joindre svp par telephone 08 11 58 54 41',
languages=None,
settings={'STRICT_PARSING': True},
expected=None),
param(text="a Americ", languages=None, settings=None, expected=None),
])
def test_date_search_function(self, text, languages, settings, expected):
result = search_dates(text, languages=languages, settings=settings)
self.assertEqual(result, expected)
@parameterized.expand([
param(text="15 de outubro de 1936",
add_detected_language=True,
expected=[("15 de outubro de 1936",
datetime.datetime(1936, 10, 15, 0, 0), "pt")]),
param(text="15 de outubro de 1936",
add_detected_language=False,
expected=[("15 de outubro de 1936",
datetime.datetime(1936, 10, 15, 0, 0))]),
])
def test_search_dates_returning_detected_languages_if_requested(
self, text, add_detected_language, expected):
result = search_dates(text,
add_detected_language=add_detected_language)
self.assertEqual(result, expected)
@parameterized.expand([
param(text='19 марта 2001',
languages='wrong type: str instead of list'),
])
def test_date_search_function_invalid_languages_type(
self, text, languages):
self.run_search_dates_function_invalid_languages(text=text,
languages=languages,
error_type=TypeError)
self.check_error_message(
"languages argument must be a list (<class \'str\'> given)")
@parameterized.expand([
param(text='19 марта 2001', languages=['unknown language code']),
])
def test_date_search_function_invalid_language_code(self, text, languages):
self.run_search_dates_function_invalid_languages(text=text,
languages=languages,
error_type=ValueError)
self.check_error_message(
"Unknown language(s): 'unknown language code'")
