]> Raphaƫl G. Git Repositories - youtubedl/blobdiff - youtube_dl/extractor/redtube.py
debian/README.source: Adapt according to changes in git-builpackage.
[youtubedl] / youtube_dl / extractor / redtube.py
index ebc4e23261bfe029e6e30150c2ca3b264d24ab81..846b76c81528431c0faf8ea3fc9bbd6b017db099 100644 (file)
@@ -1,29 +1,40 @@
-import re
+from __future__ import unicode_literals
 
 from .common import InfoExtractor
 
 
 class RedTubeIE(InfoExtractor):
-    _VALID_URL = r'(?:http://)?(?:www\.)?redtube\.com/(?P<id>[0-9]+)'
-
-    def _real_extract(self,url):
-        mobj = re.match(self._VALID_URL, url)
-
-        video_id = mobj.group('id')
-        video_extension = 'mp4'        
+    _VALID_URL = r'http://(?:www\.)?redtube\.com/(?P<id>[0-9]+)'
+    _TEST = {
+        'url': 'http://www.redtube.com/66418',
+        'info_dict': {
+            'id': '66418',
+            'ext': 'mp4',
+            "title": "Sucked on a toilet",
+            "age_limit": 18,
+        }
+    }
+
+    def _real_extract(self, url):
+        video_id = self._match_id(url)
         webpage = self._download_webpage(url, video_id)
 
-        self.report_extraction(video_id)
-
-        video_url = self._html_search_regex(r'<source src="(.+?)" type="video/mp4">',
-            webpage, u'video URL')
-
-        video_title = self._html_search_regex('<h1 class="videoTitle slidePanelMovable">(.+?)</h1>',
-            webpage, u'title')
-
-        return [{
-            'id':       video_id,
-            'url':      video_url,
-            'ext':      video_extension,
-            'title':    video_title,
-        }]
+        video_url = self._html_search_regex(
+            r'<source src="(.+?)" type="video/mp4">', webpage, 'video URL')
+        video_title = self._html_search_regex(
+            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': 'mp4',
+            'title': video_title,
+            'thumbnail': video_thumbnail,
+            'age_limit': age_limit,
+        }