]>
Raphaƫl G. Git Repositories - youtubedl/blob - youtube_dl/extractor/ctvnews.py
   2 from __future__ 
import unicode_literals
 
   6 from .common 
import InfoExtractor
 
   7 from ..utils 
import orderedSet
 
  10 class CTVNewsIE(InfoExtractor
): 
  11     _VALID_URL 
= r
'https?://(?:.+?\.)?ctvnews\.ca/(?:video\?(?:clip|playlist|bin)Id=|.*?)(?P<id>[0-9.]+)' 
  13         'url': 'http://www.ctvnews.ca/video?clipId=901995', 
  14         'md5': '10deb320dc0ccb8d01d34d12fc2ea672', 
  18             'title': 'Extended: \'That person cannot be me\' Johnson says', 
  19             'description': 'md5:958dd3b4f5bbbf0ed4d045c790d89285', 
  20             'timestamp': 1467286284, 
  21             'upload_date': '20160630', 
  24         'url': 'http://www.ctvnews.ca/video?playlistId=1.2966224', 
  29         'playlist_mincount': 19, 
  31         'url': 'http://www.ctvnews.ca/video?binId=1.2876780', 
  36         'playlist_mincount': 100, 
  38         'url': 'http://www.ctvnews.ca/1.810401', 
  39         'only_matching': True, 
  41         'url': 'http://www.ctvnews.ca/canadiens-send-p-k-subban-to-nashville-in-blockbuster-trade-1.2967231', 
  42         'only_matching': True, 
  44         'url': 'http://vancouverisland.ctvnews.ca/video?clipId=761241', 
  45         'only_matching': True, 
  48     def _real_extract(self
, url
): 
  49         page_id 
= self
._match
_id
(url
) 
  51         def ninecninemedia_url_result(clip_id
): 
  53                 '_type': 'url_transparent', 
  55                 'url': '9c9media:ctvnews_web:%s' % clip_id
, 
  56                 'ie_key': 'NineCNineMedia', 
  60             return ninecninemedia_url_result(page_id
) 
  62             webpage 
= self
._download
_webpage
('http://www.ctvnews.ca/%s' % page_id
, page_id
, query
={ 
  63                 'ot': 'example.AjaxPageLayout.ot', 
  64                 'maxItemsPerPage': 1000000, 
  66             entries 
= [ninecninemedia_url_result(clip_id
) for clip_id 
in orderedSet( 
  67                 re
.findall(r
'clip\.id\s*=\s*(\d+);', webpage
))] 
  68             return self
.playlist_result(entries
, page_id
)