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

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import re
   5
   6 from .theplatform import ThePlatformIE
   7 from ..utils import int_or_none
   8
   9
  10 class CBSInteractiveIE(ThePlatformIE):
  11     _VALID_URL = r'https?://(?:www\.)?(?P<site>cnet|zdnet)\.com/(?:videos|video/share)/(?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': 'flv',
  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             'duration': 70,
  22             'timestamp': 1396479627,
  23             'upload_date': '20140402',
  24         },
  25     }, {
  26         'url': 'http://www.cnet.com/videos/whiny-pothole-tweets-at-local-government-when-hit-by-cars-tomorrow-daily-187/',
  27         'info_dict': {
  28             'id': '56527b93-d25d-44e3-b738-f989ce2e49ba',
  29             'ext': 'flv',
  30             'title': 'Whiny potholes tweet at local government when hit by cars (Tomorrow Daily 187)',
  31             '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',
  32             'uploader_id': 'b163284d-6b73-44fc-b3e6-3da66c392d40',
  33             'uploader': 'Ashley Esqueda',
  34             'duration': 1482,
  35             'timestamp': 1433289889,
  36             'upload_date': '20150603',
  37         },
  38     }, {
  39         'url': 'http://www.zdnet.com/video/share/video-keeping-android-smartphones-and-tablets-secure/',
  40         'info_dict': {
  41             'id': 'bc1af9f0-a2b5-4e54-880d-0d95525781c0',
  42             'ext': 'mp4',
  43             'title': 'Video: Keeping Android smartphones and tablets secure',
  44             'description': 'Here\'s the best way to keep Android devices secure, and what you do when they\'ve come to the end of their lives.',
  45             'uploader_id': 'f2d97ea2-8175-11e2-9d12-0018fe8a00b0',
  46             'uploader': 'Adrian Kingsley-Hughes',
  47             'timestamp': 1448961720,
  48             'upload_date': '20151201',
  49         },
  50         'params': {
  51             # m3u8 download
  52             'skip_download': True,
  53         }
  54     }]
  55     TP_RELEASE_URL_TEMPLATE = 'http://link.theplatform.com/s/kYEXFC/%s?mbr=true'
  56     MPX_ACCOUNTS = {
  57         'cnet': 2288573011,
  58         'zdnet': 2387448114,
  59     }
  60
  61     def _real_extract(self, url):
  62         site, display_id = re.match(self._VALID_URL, url).groups()
  63         webpage = self._download_webpage(url, display_id)
  64
  65         data_json = self._html_search_regex(
  66             r"data-(?:cnet|zdnet)-video(?:-uvp(?:js)?)?-options='([^']+)'",
  67             webpage, 'data json')
  68         data = self._parse_json(data_json, display_id)
  69         vdata = data.get('video') or data['videos'][0]
  70
  71         video_id = vdata['id']
  72         title = vdata['title']
  73         author = vdata.get('author')
  74         if author:
  75             uploader = '%s %s' % (author['firstName'], author['lastName'])
  76             uploader_id = author.get('id')
  77         else:
  78             uploader = None
  79             uploader_id = None
  80
  81         media_guid_path = 'media/guid/%d/%s' % (self.MPX_ACCOUNTS[site], vdata['mpxRefId'])
  82         formats, subtitles = [], {}
  83         for (fkey, vid) in vdata['files'].items():
  84             if fkey == 'hls_phone' and 'hls_tablet' in vdata['files']:
  85                 continue
  86             release_url = self.TP_RELEASE_URL_TEMPLATE % vid
  87             if fkey == 'hds':
  88                 release_url += '&manifest=f4m'
  89             tp_formats, tp_subtitles = self._extract_theplatform_smil(release_url, video_id, 'Downloading %s SMIL data' % fkey)
  90             formats.extend(tp_formats)
  91             subtitles = self._merge_subtitles(subtitles, tp_subtitles)
  92         self._sort_formats(formats)
  93
  94         info = self._extract_theplatform_metadata('kYEXFC/%s' % media_guid_path, video_id)
  95         info.update({
  96             'id': video_id,
  97             'display_id': display_id,
  98             'title': title,
  99             'duration': int_or_none(vdata.get('duration')),
 100             'uploader': uploader,
 101             'uploader_id': uploader_id,
 102             'subtitles': subtitles,
 103             'formats': formats,
 104         })
 105         return info