]>
Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/mailru.py
   2 from __future__ 
import unicode_literals
 
   6 from .common 
import InfoExtractor
 
   9 class MailRuIE(InfoExtractor
): 
  11     IE_DESC 
= 'Видео@Mail.Ru' 
  12     _VALID_URL 
= r
'http://(?:www\.)?my\.mail\.ru/(?:video/.*#video=/?(?P<idv1>(?:[^/]+/){3}\d+)|(?:(?P<idv2prefix>(?:[^/]+/){2})video/(?P<idv2suffix>[^/]+/\d+))\.html)' 
  16             'url': 'http://my.mail.ru/video/top#video=/mail/sonypicturesrus/75/76', 
  17             'md5': 'dea205f03120046894db4ebb6159879a', 
  21                 'title': 'Новый Человек-Паук. Высокое напряжение. Восстание Электро', 
  22                 'timestamp': 1393232740, 
  23                 'upload_date': '20140224', 
  24                 'uploader': 'sonypicturesrus', 
  25                 'uploader_id': 'sonypicturesrus@mail.ru', 
  30             'url': 'http://my.mail.ru/corp/hitech/video/news_hi-tech_mail_ru/1263.html', 
  31             'md5': '00a91a58c3402204dcced523777b475f', 
  33                 'id': '46843144_1263', 
  35                 'title': 'Samsung Galaxy S5 Hammer Smash Fail Battery Explosion', 
  36                 'timestamp': 1397217632, 
  37                 'upload_date': '20140411', 
  39                 'uploader_id': 'hitech@corp.mail.ru', 
  45     def _real_extract(self
, url
): 
  46         mobj 
= re
.match(self
._VALID
_URL
, url
) 
  47         video_id 
= mobj
.group('idv1') 
  50             video_id 
= mobj
.group('idv2prefix') + mobj
.group('idv2suffix') 
  52         video_data 
= self
._download
_json
( 
  53             'http://api.video.mail.ru/videos/%s.json?new=1' % video_id
, video_id
, 'Downloading video JSON') 
  55         author 
= video_data
['author'] 
  56         uploader 
= author
['name'] 
  57         uploader_id 
= author
.get('id') or author
.get('email') 
  58         view_count 
= video_data
.get('views_count') 
  60         meta_data 
= video_data
['meta'] 
  61         content_id 
= '%s_%s' % ( 
  62             meta_data
.get('accId', ''), meta_data
['itemId']) 
  63         title 
= meta_data
['title'] 
  64         if title
.endswith('.mp4'): 
  66         thumbnail 
= meta_data
['poster'] 
  67         duration 
= meta_data
['duration'] 
  68         timestamp 
= meta_data
['timestamp'] 
  73                 'format_id': video
['key'], 
  74                 'height': int(video
['key'].rstrip('p')) 
  75             } for video 
in video_data
['videos'] 
  77         self
._sort
_formats
(formats
) 
  82             'thumbnail': thumbnail
, 
  83             'timestamp': timestamp
, 
  85             'uploader_id': uploader_id
, 
  87             'view_count': view_count
,