_ Git - youtube-dl/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             'timestamp': 1396479627,
  21             'upload_date': '20140402',
  22         },
  23     }, {
  24         'url': 'http://www.cnet.com/videos/whiny-pothole-tweets-at-local-government-when-hit-by-cars-tomorrow-daily-187/',
  25         'info_dict': {
  26             'id': '56527b93-d25d-44e3-b738-f989ce2e49ba',
  27             'ext': 'flv',
  28             'title': 'Whiny potholes tweet at local government when hit by cars (Tomorrow Daily 187)',
  29             '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',
  30             'uploader_id': 'b163284d-6b73-44fc-b3e6-3da66c392d40',
  31             'uploader': 'Ashley Esqueda',
  32             'duration': 1482,
  33             'timestamp': 1433289889,
  34             'upload_date': '20150603',
  35         },
  36     }]
  37     TP_RELEASE_URL_TEMPLATE = 'http://link.theplatform.com/s/kYEXFC/%s?mbr=true'
  38
  39     def _real_extract(self, url):
  40         display_id = self._match_id(url)
  41         webpage = self._download_webpage(url, display_id)
  42
  43         data_json = self._html_search_regex(
  44             r"data-cnet-video(?:-uvp)?-options='([^']+)'",
  45             webpage, 'data json')
  46         data = self._parse_json(data_json, display_id)
  47         vdata = data.get('video') or data['videos'][0]
  48
  49         video_id = vdata['id']
  50         title = vdata['title']
  51         author = vdata.get('author')
  52         if author:
  53             uploader = '%s %s' % (author['firstName'], author['lastName'])
  54             uploader_id = author.get('id')
  55         else:
  56             uploader = None
  57             uploader_id = None
  58
  59         media_guid_path = 'media/guid/2288573011/%s' % vdata['mpxRefId']
  60         formats, subtitles = self._extract_theplatform_smil(self.TP_RELEASE_URL_TEMPLATE % media_guid_path, video_id)
  61         for (fkey, vid) in vdata['files'].items():
  62             if fkey == 'hls_phone' and 'hls_tablet' in vdata['files']:
  63                 continue
  64             release_url = self.TP_RELEASE_URL_TEMPLATE % vid
  65             if fkey == 'hds':
  66                 release_url += '&manifest=f4m'
  67             tp_formats, tp_subtitles = self._extract_theplatform_smil(release_url, video_id, 'Downloading %s SMIL data' % fkey)
  68             formats.extend(tp_formats)
  69             subtitles = self._merge_subtitles(subtitles, tp_subtitles)
  70         self._sort_formats(formats)
  71
  72         info = self.get_metadata('kYEXFC/%s' % media_guid_path, video_id)
  73         info.update({
  74             'id': video_id,
  75             'display_id': display_id,
  76             'title': title,
  77             'duration': int_or_none(vdata.get('duration')),
  78             'uploader': uploader,
  79             'uploader_id': uploader_id,
  80             'subtitles': subtitles,
  81             'formats': formats,
  82         })
  83         return info