[vube] Add support for new data format (Fixes #3325)

youtube_dl/extractor/vube.py

@@ -1,5 +1,6 @@
 from __future__ import unicode_literals
 
+import json
 import re
 
 from .common import InfoExtractor
@@ -20,9 +21,8 @@ class VubeIE(InfoExtractor):
                 'ext': 'mp4',
                 'title': 'Chiara Grispo - Price Tag by Jessie J',
                 'description': 'md5:8ea652a1f36818352428cb5134933313',
-                'thumbnail': 'http://frame.thestaticvube.com/snap/228x128/102e7e63057-5ebc-4f5c-4065-6ce4ebde131f.jpg',
+                'thumbnail': 're:^http://frame\.thestaticvube\.com/snap/[0-9x]+/102e7e63057-5ebc-4f5c-4065-6ce4ebde131f\.jpg$',
                 'uploader': 'Chiara.Grispo',
-                'uploader_id': '1u3hX0znhP',
                 'timestamp': 1388743358,
                 'upload_date': '20140103',
                 'duration': 170.56
@@ -36,13 +36,27 @@ class VubeIE(InfoExtractor):
                 'ext': 'mp4',
                 'title': 'My 7 year old Sister and I singing "Alive" by Krewella',
                 'description': 'md5:40bcacb97796339f1690642c21d56f4a',
-                'thumbnail': 'http://frame.thestaticvube.com/snap/228x128/102265d5a9f-0f17-4f6b-5753-adf08484ee1e.jpg',
+                'thumbnail': 're:^http://frame\.thestaticvube\.com/snap/[0-9x]+/102265d5a9f-0f17-4f6b-5753-adf08484ee1e\.jpg$',
                 'uploader': 'Seraina',
-                'uploader_id': 'XU9VE2BQ2q',
                 'timestamp': 1396492438,
                 'upload_date': '20140403',
                 'duration': 240.107
             }
+        }, {
+            'url': 'http://vube.com/vote/Siren+Gene/0nmsMY5vEq?n=2&t=s',
+            'md5': '0584fc13b50f887127d9d1007589d27f',
+            'info_dict': {
+                'id': '0nmsMY5vEq',
+                'ext': 'mp4',
+                'title': 'Frozen - Let It Go Cover by Siren Gene',
+                'description': 'My rendition of "Let It Go" originally sung by Idina Menzel.',
+                'uploader': 'Siren Gene',
+                'uploader_id': 'Siren',
+                'thumbnail': 're:^http://frame\.thestaticvube\.com/snap/[0-9x]+/10283ab622a-86c9-4681-51f2-30d1f65774af\.jpg$',
+                'duration': 221.788,
+                'like_count': int,
+                'dislike_count': int,
+            }
         }
     ]
 
@@ -50,8 +64,17 @@ class VubeIE(InfoExtractor):
         mobj = re.match(self._VALID_URL, url)
         video_id = mobj.group('id')
 
-        video = self._download_json(
-            'http://vube.com/api/v2/video/%s' % video_id, video_id, 'Downloading video JSON')
+        webpage = self._download_webpage(url, video_id)
+        data_json = self._search_regex(
+            r'(?s)window\["(?:tapiVideoData|vubeOriginalVideoData)"\]\s*=\s*(\{.*?\n});\n',
+            webpage, 'video data'
+        )
+        data = json.loads(data_json)
+        open('/dev/shm/f', 'w').write(json.dumps(data, indent=2))
+        video = (
+            data.get('video') or
+            data)
+        assert isinstance(video, dict)
 
         public_id = video['public_id']
 
@@ -69,16 +92,16 @@ class VubeIE(InfoExtractor):
 
         title = video['title']
         description = video.get('description')
-        thumbnail = video['thumbnail_src']
-        if thumbnail.startswith('//'):
-            thumbnail = 'http:' + thumbnail
-        uploader = video['user_alias']
-        uploader_id = video['user_url_id']
-        timestamp = int(video['upload_time'])
+        thumbnail = self._proto_relative_url(
+            video.get('thumbnail') or video.get('thumbnail_src'),
+            scheme='http:')
+        uploader = data.get('user', {}).get('channel', {}).get('name') or video.get('user_alias')
+        uploader_id = data.get('user', {}).get('name')
+        timestamp = int_or_none(video.get('upload_time'))
         duration = video['duration']
         view_count = video.get('raw_view_count')
-        like_count = video.get('total_likes')
-        dislike_count= video.get('total_hates')
+        like_count = video.get('rlikes')
+        dislike_count = video.get('rhates')
 
         comment = self._download_json(
             'http://vube.com/api/video/%s/comment' % video_id, video_id, 'Downloading video comment JSON')