浏览代码

[krasview] Fix extraction

Sergey M․ 10 年之前
父节点
当前提交
1e10802990
共有 1 个文件被更改,包括 9 次插入13 次删除
  1. 9 13
      youtube_dl/extractor/krasview.py

+ 9 - 13
youtube_dl/extractor/krasview.py

@@ -2,18 +2,17 @@
 from __future__ import unicode_literals
 from __future__ import unicode_literals
 
 
 import json
 import json
-import re
 
 
 from .common import InfoExtractor
 from .common import InfoExtractor
 from ..utils import (
 from ..utils import (
     int_or_none,
     int_or_none,
-    unescapeHTML,
+    js_to_json,
 )
 )
 
 
 
 
 class KrasViewIE(InfoExtractor):
 class KrasViewIE(InfoExtractor):
     IE_DESC = 'Красвью'
     IE_DESC = 'Красвью'
-    _VALID_URL = r'https?://krasview\.ru/video/(?P<id>\d+)'
+    _VALID_URL = r'https?://krasview\.ru/(?:video|embed)/(?P<id>\d+)'
 
 
     _TEST = {
     _TEST = {
         'url': 'http://krasview.ru/video/512228',
         'url': 'http://krasview.ru/video/512228',
@@ -29,20 +28,18 @@ class KrasViewIE(InfoExtractor):
     }
     }
 
 
     def _real_extract(self, url):
     def _real_extract(self, url):
-        mobj = re.match(self._VALID_URL, url)
-        video_id = mobj.group('id')
+        video_id = self._match_id(url)
 
 
         webpage = self._download_webpage(url, video_id)
         webpage = self._download_webpage(url, video_id)
 
 
-        flashvars = json.loads(self._search_regex(
-            r'flashvars\s*:\s*({.+?})\s*}\);', webpage, 'flashvars'))
+        flashvars = json.loads(js_to_json(self._search_regex(
+            r'video_Init\(({.+?})', webpage, 'flashvars')))
 
 
         video_url = flashvars['url']
         video_url = flashvars['url']
-        title = unescapeHTML(flashvars['title'])
-        description = unescapeHTML(flashvars.get('subtitle') or self._og_search_description(webpage, default=None))
-        thumbnail = flashvars['image']
-        duration = int(flashvars['duration'])
-        filesize = int(flashvars['size'])
+        title = self._og_search_title(webpage)
+        description = self._og_search_description(webpage, default=None)
+        thumbnail = flashvars.get('image') or self._og_search_thumbnail(webpage)
+        duration = int_or_none(flashvars.get('duration'))
         width = int_or_none(self._og_search_property('video:width', webpage, 'video width'))
         width = int_or_none(self._og_search_property('video:width', webpage, 'video width'))
         height = int_or_none(self._og_search_property('video:height', webpage, 'video height'))
         height = int_or_none(self._og_search_property('video:height', webpage, 'video height'))
 
 
@@ -53,7 +50,6 @@ class KrasViewIE(InfoExtractor):
             'description': description,
             'description': description,
             'thumbnail': thumbnail,
             'thumbnail': thumbnail,
             'duration': duration,
             'duration': duration,
-            'filesize': filesize,
             'width': width,
             'width': width,
             'height': height,
             'height': height,
         }
         }