X-Git-Url: http://git.cielonegro.org/gitweb.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fredtube.py;h=d6054d7175fd49a22117dd357bea7905f6e739be;hb=838b93405bc31654b07e120795115aec6e883855;hp=bb19b898a4e76a8fe5e7f694a1399d94d6731d2a;hpb=cd214418f611f7071417df0063d115ea911705a3;p=youtube-dl.git

diff --git a/youtube_dl/extractor/redtube.py b/youtube_dl/extractor/redtube.py
index bb19b898a..d6054d717 100644
--- a/youtube_dl/extractor/redtube.py
+++ b/youtube_dl/extractor/redtube.py
@@ -1,38 +1,45 @@
-import re
+from __future__ import unicode_literals
 
 from .common import InfoExtractor
+from ..utils import ExtractorError
 
 
 class RedTubeIE(InfoExtractor):
-    _VALID_URL = r'(?:http://)?(?:www\.)?redtube\.com/(?P<id>[0-9]+)'
+    _VALID_URL = r'http://(?:www\.)?redtube\.com/(?P<id>[0-9]+)'
     _TEST = {
-        u'url': u'http://www.redtube.com/66418',
-        u'file': u'66418.mp4',
-        u'md5': u'7b8c22b5e7098a3e1c09709df1126d2d',
-        u'info_dict': {
-            u"title": u"Sucked on a toilet"
+        'url': 'http://www.redtube.com/66418',
+        'md5': '7b8c22b5e7098a3e1c09709df1126d2d',
+        'info_dict': {
+            'id': '66418',
+            'ext': 'mp4',
+            'title': 'Sucked on a toilet',
+            'age_limit': 18,
         }
     }
 
     def _real_extract(self, url):
-        mobj = re.match(self._VALID_URL, url)
-
-        video_id = mobj.group('id')
-        video_extension = 'mp4'
+        video_id = self._match_id(url)
         webpage = self._download_webpage(url, video_id)
 
-        self.report_extraction(video_id)
+        if any(s in webpage for s in ['video-deleted-info', '>This video has been removed']):
+            raise ExtractorError('Video %s has been removed' % video_id, expected=True)
 
         video_url = self._html_search_regex(
-            r'<source src="(.+?)" type="video/mp4">', webpage, u'video URL')
-
+            r'<source src="(.+?)" type="video/mp4">', webpage, 'video URL')
         video_title = self._html_search_regex(
-            r'<h1 class="videoTitle slidePanelMovable">(.+?)</h1>',
-            webpage, u'title')
+            r'<h1 class="videoTitle[^"]*">(.+?)</h1>',
+            webpage, 'title')
+        video_thumbnail = self._og_search_thumbnail(webpage)
+
+        # No self-labeling, but they describe themselves as
+        # "Home of Videos Porno"
+        age_limit = 18
 
         return {
-            'id':       video_id,
-            'url':      video_url,
-            'ext':      video_extension,
-            'title':    video_title,
+            'id': video_id,
+            'url': video_url,
+            'ext': 'mp4',
+            'title': video_title,
+            'thumbnail': video_thumbnail,
+            'age_limit': age_limit,
         }