def _get_case_dates(self):
case_dates = []
for e in self.html.xpath(self.base_path):
s = ' '.join(e.xpath(".//text()"))
case_date = parse_dates(s)[0]
case_dates.extend([case_date] * int(e.xpath("count(./following::ul[1]//li)")))
return case_dates
def _get_case_dates(self):
case_dates = []
for e in self.html.xpath(self.base_path):
s = ' '.join(e.xpath(".//text()"))
case_date = parse_dates(s)[0]
case_dates.extend([case_date] *
int(e.xpath("count(./following::ul[1]//li)")))
return case_dates
def test_various_date_extractions(self):
test_pairs = (
# Dates separated by semicolons and JUMP words
('February 5, 1980; March 14, 1980 and May 28, 1980.', [
datetime.datetime(1980, 2, 5, 0, 0),
datetime.datetime(1980, 3, 14, 0, 0),
datetime.datetime(1980, 5, 28, 0, 0)
]),
# Misspelled month value.
('Febraury 17, 1945', [datetime.datetime(1945, 2, 17, 0, 0)]),
('Sepetmber 19 1924', [datetime.datetime(1924, 9, 19)]),
# Using 'Term' as an indicator.
('November Term 2004.', [datetime.datetime(2004, 11, 01)]),
(u'April 26, 1961.[†]', [datetime.datetime(1961, 4, 26)]),
)
for pair in test_pairs:
dates = parse_dates(pair[0])
self.assertEqual(dates, pair[1])
def test_various_date_extractions(self):
test_pairs = (
# Dates separated by semicolons and JUMP words
('February 5, 1980; March 14, 1980 and May 28, 1980.',
[datetime.datetime(1980, 2, 5, 0, 0),
datetime.datetime(1980, 3, 14, 0, 0),
datetime.datetime(1980, 5, 28, 0, 0)]),
# Misspelled month value.
('Febraury 17, 1945',
[datetime.datetime(1945, 2, 17, 0, 0)]),
('Sepetmber 19 1924',
[datetime.datetime(1924, 9, 19)]),
# Using 'Term' as an indicator.
('November Term 2004.',
[datetime.datetime(2004, 11, 01)]),
(u'April 26, 1961.[†]',
[datetime.datetime(1961, 4, 26)]),
)
for pair in test_pairs:
dates = parse_dates(pair[0])
self.assertEqual(dates, pair[1])