Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/cnet.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .theplatform import ThePlatformIE
   5 from ..utils import int_or_none
   6
   7
   8 class CNETIE(ThePlatformIE):
   9     _VALID_URL = r'https?://(?:www\.)?cnet\.com/videos/(?P<id>[^/]+)/'
  10     _TESTS = [{
  11         'url': 'http://www.cnet.com/videos/hands-on-with-microsofts-windows-8-1-update/',
  12         'info_dict': {
  13             'id': '56f4ea68-bd21-4852-b08c-4de5b8354c60',
  14             'ext': 'flv',
  15             'title': 'Hands-on with Microsoft Windows 8.1 Update',
  16             'description': 'The new update to the Windows 8 OS brings improved performance for mouse and keyboard users.',
  17             'uploader_id': '6085384d-619e-11e3-b231-14feb5ca9861',
  18             'uploader': 'Sarah Mitroff',
  19             'duration': 70,
  20         },
  21     }, {
  22         'url': 'http://www.cnet.com/videos/whiny-pothole-tweets-at-local-government-when-hit-by-cars-tomorrow-daily-187/',
  23         'info_dict': {
  24             'id': '56527b93-d25d-44e3-b738-f989ce2e49ba',
  25             'ext': 'flv',
  26             'title': 'Whiny potholes tweet at local government when hit by cars (Tomorrow Daily 187)',
  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             'duration': 1482,
  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"data-cnet-video(?:-uvp)?-options='([^']+)'",
  40             webpage, 'data json')
  41         data = self._parse_json(data_json, display_id)
  42         vdata = data.get('video') or 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
  56         metadata = self.get_metadata('%s/%s' % (mpx_account, list(vdata['files'].values())[0]), video_id)
  57         description = vdata.get('description') or metadata.get('description')
  58         duration = int_or_none(vdata.get('duration')) or metadata.get('duration')
  59
  60         formats = []
  61         subtitles = {}
  62         for (fkey, vid) in vdata['files'].items():
  63             if fkey == 'hls_phone' and 'hls_tablet' in vdata['files']:
  64                 continue
  65             release_url = 'http://link.theplatform.com/s/%s/%s?format=SMIL&mbr=true' % (mpx_account, vid)
  66             if fkey == 'hds':
  67                 release_url += '&manifest=f4m'
  68             tp_formats, tp_subtitles = self._extract_theplatform_smil(release_url, video_id, 'Downloading %s SMIL data' % fkey)
  69             formats.extend(tp_formats)
  70             subtitles = self._merge_subtitles(subtitles, tp_subtitles)
  71         self._sort_formats(formats)
  72
  73         return {
  74             'id': video_id,
  75             'display_id': display_id,
  76             'title': title,
  77             'description': description,
  78             'thumbnail': metadata.get('thumbnail'),
  79             'duration': duration,
  80             'uploader': uploader,
  81             'uploader_id': uploader_id,
  82             'subtitles': subtitles,
  83             'formats': formats,
  84         }