[comedycentral/mtv] Add support for HLS videos (fixes #11600)

[youtube-dl.git] / youtube_dl / extractor / comedycentral.py

diff --git a/youtube_dl/extractor/comedycentral.py b/youtube_dl/extractor/comedycentral.py
index c76909e48dfdb7ff7c490c22971be8462ec8a59d..8bd58977493bf2445e768604d8953b0474287918 100644 (file)
--- a/youtube_dl/extractor/comedycentral.py
+++ b/youtube_dl/extractor/comedycentral.py
@@ -1,11 +1,12 @@
 from __future__ import unicode_literals
 
 from .mtv import MTVServicesInfoExtractor
+from .common import InfoExtractor
 
 
 class ComedyCentralIE(MTVServicesInfoExtractor):
     _VALID_URL = r'''(?x)https?://(?:www\.)?cc\.com/
-        (video-clips|episodes|cc-studios|video-collections|full-episodes|shows)
+        (video-clips|episodes|cc-studios|video-collections|shows(?=/[^/]+/(?!full-episodes)))
         /(?P<title>.*)'''
     _FEED_URL = 'http://comedycentral.com/feeds/mrss/'
 
@@ -26,6 +27,41 @@ class ComedyCentralIE(MTVServicesInfoExtractor):
     }]
 
 
+class ComedyCentralFullEpisodesIE(MTVServicesInfoExtractor):
+    _VALID_URL = r'''(?x)https?://(?:www\.)?cc\.com/
+        (?:full-episodes|shows(?=/[^/]+/full-episodes))
+        /(?P<id>[^?]+)'''
+    _FEED_URL = 'http://comedycentral.com/feeds/mrss/'
+
+    _TESTS = [{
+        'url': 'http://www.cc.com/full-episodes/pv391a/the-daily-show-with-trevor-noah-november-28--2016---ryan-speedo-green-season-22-ep-22028',
+        'info_dict': {
+            'description': 'Donald Trump is accused of exploiting his president-elect status for personal gain, Cuban leader Fidel Castro dies, and Ryan Speedo Green discusses "Sing for Your Life."',
+            'title': 'November 28, 2016 - Ryan Speedo Green',
+        },
+        'playlist_count': 4,
+    }, {
+        'url': 'http://www.cc.com/shows/the-daily-show-with-trevor-noah/full-episodes',
+        'only_matching': True,
+    }]
+
+    def _real_extract(self, url):
+        playlist_id = self._match_id(url)
+        webpage = self._download_webpage(url, playlist_id)
+
+        feed_json = self._search_regex(r'var triforceManifestFeed\s*=\s*(\{.+?\});\n', webpage, 'triforce feeed')
+        feed = self._parse_json(feed_json, playlist_id)
+        zones = feed['manifest']['zones']
+
+        video_zone = zones['t2_lc_promo1']
+        feed = self._download_json(video_zone['feed'], playlist_id)
+        mgid = feed['result']['data']['id']
+
+        videos_info = self._get_videos_info(mgid, use_hls=True)
+
+        return videos_info
+
+
 class ToshIE(MTVServicesInfoExtractor):
     IE_DESC = 'Tosh.0'
     _VALID_URL = r'^https?://tosh\.cc\.com/video-(?:clips|collections)/[^/]+/(?P<videotitle>[^/?#]+)'
@@ -44,7 +80,7 @@ class ToshIE(MTVServicesInfoExtractor):
                 'ext': 'mp4',
                 'title': 'Tosh.0|June 9, 2077|2|211|Twitter Users Share Summer Plans',
                 'description': 'Tosh asked fans to share their summer plans.',
-                'thumbnail': 're:^https?://.*\.jpg',
+                'thumbnail': r're:^https?://.*\.jpg',
                 # It's really reported to be published on year 2077
                 'upload_date': '20770610',
                 'timestamp': 3390510600,
@@ -96,3 +132,22 @@ class ComedyCentralTVIE(MTVServicesInfoExtractor):
             webpage, 'mrss url', group='url')
 
         return self._get_videos_info_from_url(mrss_url, video_id)
+
+
+class ComedyCentralShortnameIE(InfoExtractor):
+    _VALID_URL = r'^:(?P<id>tds|thedailyshow)$'
+    _TESTS = [{
+        'url': ':tds',
+        'only_matching': True,
+    }, {
+        'url': ':thedailyshow',
+        'only_matching': True,
+    }]
+
+    def _real_extract(self, url):
+        video_id = self._match_id(url)
+        shortcut_map = {
+            'tds': 'http://www.cc.com/shows/the-daily-show-with-trevor-noah/full-episodes',
+            'thedailyshow': 'http://www.cc.com/shows/the-daily-show-with-trevor-noah/full-episodes',
+        }
+        return self.url_result(shortcut_map[video_id])