[comedycentral] Match more URLs

Looks like they only offer clips instead of full episodes now. We'll need to add new parsing code as well.
This commit is contained in:
Philipp Hagemeister 2014-04-09 11:43:13 +02:00
parent f983c44199
commit 29e3e682af
2 changed files with 11 additions and 2 deletions

View File

@ -156,6 +156,15 @@ class TestAllURLsMatching(unittest.TestCase):
self.assertMatch( self.assertMatch(
'http://thedailyshow.cc.com/guests/michael-lewis/3efna8/exclusive---michael-lewis-extended-interview-pt--3', 'http://thedailyshow.cc.com/guests/michael-lewis/3efna8/exclusive---michael-lewis-extended-interview-pt--3',
['ComedyCentralShows']) ['ComedyCentralShows'])
self.assertMatch(
'http://thedailyshow.cc.com/episodes/sy7yv0/april-8--2014---denis-leary',
['ComedyCentralShows'])
self.assertMatch(
'http://thecolbertreport.cc.com/episodes/8ase07/april-8--2014---jane-goodall',
['ComedyCentralShows'])
self.assertMatch(
'http://thedailyshow.cc.com/video-playlists/npde3s/the-daily-show-19088-highlights',
['ComedyCentralShows'])
def test_yahoo_https(self): def test_yahoo_https(self):
# https://github.com/rg3/youtube-dl/issues/2701 # https://github.com/rg3/youtube-dl/issues/2701

View File

@ -41,9 +41,9 @@ class ComedyCentralShowsIE(InfoExtractor):
_VALID_URL = r'''(?x)^(:(?P<shortname>tds|thedailyshow|cr|colbert|colbertnation|colbertreport) _VALID_URL = r'''(?x)^(:(?P<shortname>tds|thedailyshow|cr|colbert|colbertnation|colbertreport)
|https?://(:www\.)? |https?://(:www\.)?
(?P<showname>thedailyshow|thecolbertreport)\.(?:cc\.)?com/ (?P<showname>thedailyshow|thecolbertreport)\.(?:cc\.)?com/
(full-episodes/(?:[0-9a-z]{6}/)?(?P<episode>.*)| ((?:full-)?episodes/(?:[0-9a-z]{6}/)?(?P<episode>.*)|
(?P<clip> (?P<clip>
(?:(?:guests/[^/]+|videos)/[^/]+/(?P<videotitle>[^/?#]+)) (?:(?:guests/[^/]+|videos|video-playlists)/[^/]+/(?P<videotitle>[^/?#]+))
|(the-colbert-report-(videos|collections)/(?P<clipID>[0-9]+)/[^/]*/(?P<cntitle>.*?)) |(the-colbert-report-(videos|collections)/(?P<clipID>[0-9]+)/[^/]*/(?P<cntitle>.*?))
|(watch/(?P<date>[^/]*)/(?P<tdstitle>.*)) |(watch/(?P<date>[^/]*)/(?P<tdstitle>.*))
)| )|