[vk] Extract video URL from extra_data (Closes #8646)
authorSergey M․ <dstftw@gmail.com>
Tue, 23 Feb 2016 12:47:13 +0000 (18:47 +0600)
committerSergey M․ <dstftw@gmail.com>
Tue, 23 Feb 2016 12:47:13 +0000 (18:47 +0600)
youtube_dl/extractor/vk.py

index 0805e3c083937d7f58acbe4b872d41c19f8d1c02..670a438afff624b5c9f09d0b79fb34fc4e237a7a 100644 (file)
@@ -11,6 +11,7 @@ from ..compat import (
 )
 from ..utils import (
     ExtractorError,
+    int_or_none,
     orderedSet,
     sanitized_Request,
     str_to_int,
@@ -151,6 +152,19 @@ class VKIE(InfoExtractor):
                 'view_count': int,
             },
         },
+        {
+            # video key is extra_data not url\d+
+            'url': 'http://vk.com/video-110305615_171782105',
+            'md5': 'e13fcda136f99764872e739d13fac1d1',
+            'info_dict': {
+                'id': '171782105',
+                'ext': 'mp4',
+                'title': 'S-Dance, репетиции к The way show',
+                'uploader': 'THE WAY SHOW | 17 апреля',
+                'upload_date': '20160207',
+                'view_count': int,
+            },
+        },
         {
             # removed video, just testing that we match the pattern
             'url': 'http://vk.com/feed?z=video-43215063_166094326%2Fbb50cacd3177146d7a',
@@ -298,12 +312,17 @@ class VKIE(InfoExtractor):
             view_count = str_to_int(self._search_regex(
                 r'([\d,.]+)', views, 'view count', fatal=False))
 
-        formats = [{
-            'format_id': k,
-            'url': v,
-            'width': int(k[len('url'):]),
-        } for k, v in data.items()
-            if k.startswith('url')]
+        formats = []
+        for k, v in data.items():
+            if not k.startswith('url') and k != 'extra_data' or not v:
+                continue
+            height = int_or_none(self._search_regex(
+                r'^url(\d+)', k, 'height', default=None))
+            formats.append({
+                'format_id': k,
+                'url': v,
+                'height': height,
+            })
         self._sort_formats(formats)
 
         return {