Jelajahi Sumber

[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.
Philipp Hagemeister 11 tahun lalu
induk
melakukan
29e3e682af
2 mengubah file dengan 11 tambahan dan 2 penghapusan
  1. 9 0
      test/test_all_urls.py
  2. 2 2
      youtube_dl/extractor/comedycentral.py

+ 9 - 0
test/test_all_urls.py

@@ -156,6 +156,15 @@ class TestAllURLsMatching(unittest.TestCase):
         self.assertMatch(
             'http://thedailyshow.cc.com/guests/michael-lewis/3efna8/exclusive---michael-lewis-extended-interview-pt--3',
             ['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):
         # https://github.com/rg3/youtube-dl/issues/2701

+ 2 - 2
youtube_dl/extractor/comedycentral.py

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