[youtube] Fix extraction.

[youtube-dl] / youtube_dl / extractor / thesun.py

diff --git a/youtube_dl/extractor/thesun.py b/youtube_dl/extractor/thesun.py
index 7f96bf8c9488de5f1222f72e1fa8eb01066e3dd1..15d4a693271f064b183106a77b824e29b878ebcc 100644 (file)
--- a/youtube_dl/extractor/thesun.py
+++ b/youtube_dl/extractor/thesun.py
@@ -1,27 +1,38 @@
 from __future__ import unicode_literals
 
+import re
+
 from .common import InfoExtractor
-from .ooyala import OoyalaIE
+from ..utils import extract_attributes
 
 
 class TheSunIE(InfoExtractor):
-    _VALID_URL = r'https://(?:www\.)?thesun\.co\.uk/\w+/(?P<id>\d+)/[\w-]'
+    _VALID_URL = r'https://(?:www\.)?thesun\.co\.uk/[^/]+/(?P<id>\d+)'
     _TEST = {
         'url': 'https://www.thesun.co.uk/tvandshowbiz/2261604/orlando-bloom-and-katy-perry-post-adorable-instagram-video-together-celebrating-thanksgiving-after-split-rumours/',
-        'md5': '5667123b24f25f43f4c4f381ef34c5c2',
         'info_dict': {
-            'id': 'h4OXN0NzE6rv6ObkEifKcNA-gYUw4xFf',
-            'ext': 'mp4',
-            'title': 'Katy Perry and Orlando Bloom shut down split rumours with cute Thanksgiving video',
-            'description': 'Still going strong',
-            'duration': 31.28,
-        }
+            'id': '2261604',
+            'title': 'md5:cba22f48bad9218b64d5bbe0e16afddf',
+        },
+        'playlist_count': 2,
     }
+    BRIGHTCOVE_URL_TEMPLATE = 'http://players.brightcove.net/%s/default_default/index.html?videoId=%s'
 
     def _real_extract(self, url):
-        video_id = self._match_id(url)
+        article_id = self._match_id(url)
+
+        webpage = self._download_webpage(url, article_id)
 
-        webpage = self._download_webpage(url, video_id)
-        ooyala_id = self._search_regex(r'id\s*=\s*"thesun-ooyala-player-([^"]+)"', webpage, 'ooyala id')
+        entries = []
+        for video in re.findall(
+                r'<video[^>]+data-video-id-pending=[^>]+>',
+                webpage):
+            attrs = extract_attributes(video)
+            video_id = attrs['data-video-id-pending']
+            account_id = attrs.get('data-account', '5067014667001')
+            entries.append(self.url_result(
+                self.BRIGHTCOVE_URL_TEMPLATE % (account_id, video_id),
+                'BrightcoveNew', video_id))
 
-        return OoyalaIE._build_url_result(ooyala_id)
+        return self.playlist_result(
+            entries, article_id, self._og_search_title(webpage, fatal=False))