Unify coding cookie

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

diff --git a/youtube_dl/extractor/vk.py b/youtube_dl/extractor/vk.py
index 9f7a593efed4a3851fe81cf4f184dfa57950307a..ac77bc6235c89d3b3b841948460718fee8aa4392 100644 (file)
--- a/youtube_dl/extractor/vk.py
+++ b/youtube_dl/extractor/vk.py
@@ -1,4 +1,4 @@
-# encoding: utf-8
+# coding: utf-8
 from __future__ import unicode_literals
 
 import collections
@@ -20,7 +20,7 @@ from ..utils import (
     remove_start,
     str_to_int,
     unescapeHTML,
-    unified_strdate,
+    unified_timestamp,
     urlencode_postdata,
 )
 from .dailymotion import DailymotionIE
@@ -106,6 +106,7 @@ class VKIE(VKBaseIE):
                 'title': 'ProtivoGunz - Хуёвая песня',
                 'uploader': 're:(?:Noize MC|Alexander Ilyashenko).*',
                 'duration': 195,
+                'timestamp': 1329060660,
                 'upload_date': '20120212',
                 'view_count': int,
             },
@@ -119,6 +120,7 @@ class VKIE(VKBaseIE):
                 'uploader': 'Tom Cruise',
                 'title': 'No name',
                 'duration': 9,
+                'timestamp': 1374374880,
                 'upload_date': '20130721',
                 'view_count': int,
             }
@@ -195,6 +197,7 @@ class VKIE(VKBaseIE):
                 'upload_date': '20150709',
                 'view_count': int,
             },
+            'skip': 'Removed',
         },
         {
             # youtube embed
@@ -237,6 +240,7 @@ class VKIE(VKBaseIE):
                 'ext': 'mp4',
                 'title': 'S-Dance, репетиции к The way show',
                 'uploader': 'THE WAY SHOW | 17 апреля',
+                'timestamp': 1454870100,
                 'upload_date': '20160207',
                 'view_count': int,
             },
@@ -251,6 +255,7 @@ class VKIE(VKBaseIE):
                 'title': 'ИгроМир 2016 — день 1',
                 'uploader': 'Игромания',
                 'duration': 5239,
+                'view_count': int,
             },
         },
         {
@@ -372,20 +377,13 @@ class VKIE(VKBaseIE):
         if data.get('live') == 2:
             title = self._live_title(title)
 
-        # Extract upload date
-        upload_date = None
-        mobj = re.search(r'id="mv_date(?:_views)?_wrap"[^>]*>([a-zA-Z]+ [0-9]+), ([0-9]+) at', info_page)
-        if mobj is not None:
-            mobj.group(1) + ' ' + mobj.group(2)
-            upload_date = unified_strdate(mobj.group(1) + ' ' + mobj.group(2))
-
-        view_count = None
-        views = self._html_search_regex(
-            r'"mv_views_count_number"[^>]*>(.+?\bviews?)<',
-            info_page, 'view count', default=None)
-        if views:
-            view_count = str_to_int(self._search_regex(
-                r'([\d,.]+)', views, 'view count', fatal=False))
+        timestamp = unified_timestamp(self._html_search_regex(
+            r'class=["\']mv_info_date[^>]+>([^<]+)(?:<|from)', info_page,
+            'upload date', fatal=False))
+
+        view_count = str_to_int(self._search_regex(
+            r'class=["\']mv_views_count[^>]+>\s*([\d,.]+)',
+            info_page, 'view count', fatal=False))
 
         formats = []
         for format_id, format_url in data.items():
@@ -418,7 +416,7 @@ class VKIE(VKBaseIE):
             'thumbnail': data.get('jpg'),
             'uploader': data.get('md_author'),
             'duration': data.get('duration'),
-            'upload_date': upload_date,
+            'timestamp': timestamp,
             'view_count': view_count,
         }