_ Git - youtube-dl/blob - youtube_dl/extractor/cnet.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import json
   5
   6 from .common import InfoExtractor
   7 from .theplatform import ThePlatformIE
   8
   9
  10 class CNETIE(InfoExtractor):
  11     _VALID_URL = r'https?://(?:www\.)?cnet\.com/videos/(?P<id>[^/]+)/'
  12     _TESTS = [{
  13         'url': 'http://www.cnet.com/videos/hands-on-with-microsofts-windows-8-1-update/',
  14         'info_dict': {
  15             'id': '56f4ea68-bd21-4852-b08c-4de5b8354c60',
  16             'ext': 'mp4',
  17             'title': 'Hands-on with Microsoft Windows 8.1 Update',
  18             'description': 'The new update to the Windows 8 OS brings improved performance for mouse and keyboard users.',
  19             'uploader_id': '6085384d-619e-11e3-b231-14feb5ca9861',
  20             'uploader': 'Sarah Mitroff',
  21         },
  22     }, {
  23         'url': 'http://www.cnet.com/videos/whiny-pothole-tweets-at-local-government-when-hit-by-cars-tomorrow-daily-187/',
  24         'info_dict': {
  25             'id': '56527b93-d25d-44e3-b738-f989ce2e49ba',
  26             'ext': 'mp4',
  27             'description': 'Khail and Ashley wonder what other civic woes can be solved by self-tweeting objects, investigate a new kind of VR camera and watch an origami robot self-assemble, walk, climb, dig and dissolve. #TDPothole',
  28             'uploader_id': 'b163284d-6b73-44fc-b3e6-3da66c392d40',
  29             'uploader': 'Ashley Esqueda',
  30             'title': 'Whiny potholes tweet at local government when hit by cars (Tomorrow Daily 187)',
  31         },
  32     }]
  33
  34     def _real_extract(self, url):
  35         display_id = self._match_id(url)
  36         webpage = self._download_webpage(url, display_id)
  37
  38         data_json = self._html_search_regex(
  39             r"<div class=\"videoPlayer\"\s+.*?data-cnet-video-uvp-options='([^']+)'",
  40             webpage, 'data json')
  41         data = json.loads(data_json)
  42         vdata = data['videos'][0]
  43
  44         video_id = vdata['id']
  45         title = vdata['title']
  46         author = vdata.get('author')
  47         if author:
  48             uploader = '%s %s' % (author['firstName'], author['lastName'])
  49             uploader_id = author.get('id')
  50         else:
  51             uploader = None
  52             uploader_id = None
  53
  54         mpx_account = data['config']['uvpConfig']['default']['mpx_account']
  55         tp = ThePlatformIE(self._downloader)
  56         formats = []
  57         subtitles = {}
  58         description = vdata.get('description')
  59
  60         for (fkey, vid) in vdata['files'].items():
  61             if fkey == 'hls_phone' and 'hls_tablet' in vdata['files']:
  62                 continue
  63             result = tp.extract(('http://link.theplatform.com/s/%s/%s' % (mpx_account, vid)))
  64             formats.extend(result['formats'])
  65             subtitles = self._merge_subtitles(subtitles, result['subtitles'])
  66             description = description or result.get('description')
  67
  68         self._sort_formats(formats)
  69
  70         return {
  71             'id': video_id,
  72             'display_id': display_id,
  73             'title': title,
  74             'description': description,
  75             'uploader': uploader,
  76             'uploader_id': uploader_id,
  77             'subtitles': subtitles,
  78             'formats': formats,
  79         }