def test_four_digit_is_not_year_with_timestamp(self):
"""A 4-digit number should not be mistaken as year (w/ timestamp)."""
self.assertEqual(
self.ts.timestripper(
'2000 people will attend. --12:12, 14 December 2015 (UTC)'),
datetime.datetime(
2015, 12, 14, 12, 12, tzinfo=tzoneFixedOffset(0, 'UTC')))
def test_timestripper_match(self, key):
"""Test that correct date is matched."""
self.ts = TimeStripper(self.get_site(key))
tzone = tzoneFixedOffset(self.ts.site.siteinfo['timeoffset'],
self.ts.site.siteinfo['timezone'])
txtMatch = self.sites[key]['match']
res = datetime.datetime(2010, 2, 7, 19, 48, tzinfo=tzone)
self.assertEqual(self.ts.timestripper(txtMatch), res)
if 'match2' not in self.sites[key]:
return
txtMatch = self.sites[key]['match2']
res = datetime.datetime(2008, 9, 12, 16, 41, tzinfo=tzone)
self.assertEqual(self.ts.timestripper(txtMatch), res)
if 'match3' not in self.sites[key]:
return
txtMatch = self.sites[key]['match3']
res = datetime.datetime(2014, 8, 14, 21, 18, tzinfo=tzone)
self.assertEqual(self.ts.timestripper(txtMatch), res)
def test_timestripper_match(self, key):
"""Test that correct date is matched."""
self.ts = TimeStripper(self.get_site(key))
tzone = tzoneFixedOffset(self.ts.site.siteinfo['timeoffset'],
self.ts.site.siteinfo['timezone'])
txt_match = self.sites[key]['match']
res = datetime.datetime(2010, 2, 7, 19, 48, tzinfo=tzone)
self.assertEqual(self.ts.timestripper(txt_match), res)
if 'match2' not in self.sites[key]:
return
txt_match = self.sites[key]['match2']
res = datetime.datetime(2008, 9, 12, 16, 41, tzinfo=tzone)
self.assertEqual(self.ts.timestripper(txt_match), res)
if 'match3' not in self.sites[key]:
return
txt_match = self.sites[key]['match3']
res = datetime.datetime(2014, 8, 14, 21, 18, tzinfo=tzone)
self.assertEqual(self.ts.timestripper(txt_match), res)
def test_four_digit_is_not_year_with_timestamp(self):
"""A 4-digit number should not be mistaken as year (w/ timestamp)."""
self.assertEqual(
self.ts.timestripper(
'2000 people will attend. --12:12, 14 December 2015 (UTC)'),
datetime.datetime(
2015, 12, 14, 12, 12, tzinfo=tzoneFixedOffset(0, 'UTC')))
def test_timestripper(self):
"""Test that correct date is matched"""
txtMatch = u'3 février 2010 à 19:48 (CET) 7 février 2010 à 19:48 (CET)'
txtNoMatch = u'3 March 2010 19:48 (CET) 7 March 2010 19:48 (CET)'
res = datetime.datetime(2010, 2, 7, 19, 48,
tzinfo=tzoneFixedOffset(60, 'Europe/Paris'))
self.assertEqual(self.ts.timestripper(txtMatch), res)
self.assertEqual(self.ts.timestripper(txtNoMatch), None)
def test_timestripper(self):
"""Test that correct date is matched"""
txtMatch = u'3 février 2010 à 19:48 (CET) 7 février 2010 à 19:48 (CET)'
txtNoMatch = u'3 March 2010 19:48 (CET) 7 March 2010 19:48 (CET)'
tzone = tzoneFixedOffset(self.ts.site.siteinfo['timeoffset'],
self.ts.site.siteinfo['timezone'])
res = datetime.datetime(2010, 2, 7, 19, 48, tzinfo=tzone)
self.assertEqual(self.ts.timestripper(txtMatch), res)
self.assertEqual(self.ts.timestripper(txtNoMatch), None)
def test_timestripper(self):
"""Test that correct date is matched"""
txtMatch = u'3 février 2010 à 19:48 (CET) 7 février 2010 à 19:48 (CET)'
txtNoMatch = u'3 March 2010 19:48 (CET) 7 March 2010 19:48 (CET)'
tzone = tzoneFixedOffset(self.ts.site.siteinfo['timeoffset'],
self.ts.site.siteinfo['timezone'])
res = datetime.datetime(2010, 2, 7, 19, 48, tzinfo=tzone)
self.assertEqual(self.ts.timestripper(txtMatch), res)
self.assertEqual(self.ts.timestripper(txtNoMatch), None)
class TestTimeStripperDoNotArchiveUntil(TestCase):
"""Test cases for Do Not Archive Until templates.
See https://commons.wikimedia.org/wiki/Template:DNAU and
https://en.wikipedia.org/wiki/Template:Do_not_archive_until.
"""
family = 'wikisource'
code = 'en'
cached = True
username = '******'
date = '06:57 06 June 2015 (UTC)'
user_and_date = username + ' ' + date
tzone = tzoneFixedOffset(0, 'UTC')
def test_timestripper_match(self):
"""Test that dates in comments are correctly recognised."""
ts = TimeStripper(self.get_site())
txt_match = '<!-- [[User:Do___ArchiveUntil]] ' + self.date + ' -->'
res = datetime.datetime(2015, 6, 6, 6, 57, tzinfo=self.tzone)
self.assertEqual(ts.timestripper(txt_match), res)
txt_match = '<!-- --> <!-- ' + self.user_and_date + ' <!-- -->'
res = datetime.datetime(2015, 6, 6, 6, 57, tzinfo=self.tzone)
self.assertEqual(ts.timestripper(txt_match), res)
txt_match = '<!-- ' + self.user_and_date + ' -->'
res = datetime.datetime(2015, 6, 6, 6, 57, tzinfo=self.tzone)
self.assertEqual(ts.timestripper(txt_match), res)
def test_timestripper_match_only(self):
"""Test that latest date is used instead of other dates."""
ts = TimeStripper(self.get_site())
later_date = '10:57 06 June 2015 (UTC)'
txt_match = '<!-- --> ' + self.user_and_date + ' <!-- -->' + later_date
res = datetime.datetime(2015, 6, 6, 10, 57, tzinfo=self.tzone)
self.assertEqual(ts.timestripper(txt_match), res)
earlier_date = '02:57 06 June 2015 (UTC)'
txt_match = '<!-- ' + self.user_and_date + ' --> ' + earlier_date
res = datetime.datetime(2015, 6, 6, 6, 57, tzinfo=self.tzone)
self.assertEqual(ts.timestripper(txt_match), res)
class TestTimeStripperTreatSpecialText(TestTimeStripperCase):
"""Test special text behaviour (comments, hyperlinks, wikilinks)."""
family = 'wikisource'
code = 'en'
date = '06:57 06 June 2015 (UTC)'
fake_date = '05:57 06 June 2015 (UTC)'
tzone = tzoneFixedOffset(0, 'UTC')
expected_date = datetime.datetime(2015, 6, 6, 6, 57, tzinfo=tzone)
def test_timestripper_match_comment(self):
"""Test that comments are correctly matched."""
ts = self.ts
txt_match = self.date + '<!--a test comment-->'
exp_match = 'a test comment'
self.assertEqual(ts._comment_pat.search(txt_match).group(1), exp_match)
def test_timestripper_match_hyperlink(self):
"""Test that hyperlinks are correctly matched."""
ts = self.ts
txt_match = '[http://test.org | a link]'
exp_match = '[http://test.org | a link]'
self.assertEqual(
ts._hyperlink_pat.search(txt_match).group(), exp_match)
def test_timestripper_match_wikilink(self):
"""Test that wikilinks are correctly matched."""
ts = self.ts
txt_match = '[[wikilink|a wikilink with no date]]'
exp_match_link = 'wikilink'
exp_match_anchor = '|a wikilink with no date'
self.assertEqual(
ts._wikilink_pat.search(txt_match).group('link'), exp_match_link)
self.assertEqual(
ts._wikilink_pat.search(txt_match).group('anchor'),
exp_match_anchor)
def test_timestripper_match_comment_with_date(self):
"""Test that dates in comments are correctly matched."""
ts = self.ts.timestripper
txt_match = self.date + '<!--' + self.fake_date + '-->'
self.assertEqual(ts(txt_match), self.expected_date)
txt_match = '<!--' + self.fake_date + '-->' + self.date
self.assertEqual(ts(txt_match), self.expected_date)
txt_match = '<!--' + self.date + '-->' + self.fake_date
self.assertEqual(ts(txt_match), self.expected_date)
txt_match = '<!--comment|' + self.date + '-->' + self.fake_date
self.assertEqual(ts(txt_match), self.expected_date)
def test_timestripper_skip_hyperlink(self):
"""Test that dates in hyperlinks are correctly skipped."""
ts = self.ts.timestripper
txt_match = self.date + '[http://' + self.fake_date + ']'
self.assertEqual(ts(txt_match), self.expected_date)
txt_match = '[http://' + self.fake_date + ']' + self.date
self.assertEqual(ts(txt_match), self.expected_date)
txt_match = ('{} [http://www.org | link with date {}]'.format(
self.date, self.fake_date))
self.assertEqual(ts(txt_match), self.expected_date)
txt_match = '[http://' + self.fake_date + ']' + self.date
self.assertEqual(ts(txt_match), self.expected_date)
def test_timestripper_skip_hyperlink_and_do_not_connect(self):
"""Test that skipping hyperlinks will not make gaps shorter."""
ts = self.ts.timestripper
txt_match = (
'{}[http://example.com Here is long enough text]{}'.format(
self.date[:9], self.date[9:]))
self.assertEqual(ts(txt_match), None)
def test_timestripper_match_wikilink_with_date(self):
"""Test that dates in wikilinks are correctly matched."""
ts = self.ts.timestripper
txt_match = self.date + '[[' + self.fake_date + ']]'
self.assertEqual(ts(txt_match), self.expected_date)
txt_match = '[[' + self.fake_date + ']]' + self.date
self.assertEqual(ts(txt_match), self.expected_date)
txt_match = '[[' + self.date + ']]' + self.fake_date
self.assertEqual(ts(txt_match), self.expected_date)
txt_match = '[[wikilink|' + self.date + ']]' + self.fake_date
self.assertEqual(ts(txt_match), self.expected_date)
def test_timestripper_skip_wikilink_and_do_not_connect(self):
"""Test that skipping wikilinks will not make gaps shorter."""
ts = self.ts.timestripper
txt_match = ('{}[[Here is long enough text]]{}'.format(
self.date[:9], self.date[9:]))
self.assertEqual(ts(txt_match), None)
txt_match = self.date[:9] + '[[foo]]' + self.date[9:]
self.assertEqual(ts(txt_match), self.expected_date)
