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

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .anvato import AnvatoIE
   5 from .sendtonews import SendtoNewsIE
   6 from ..compat import compat_urlparse
   7 from ..utils import unified_timestamp
   8
   9
  10 class CBSLocalIE(AnvatoIE):
  11     _VALID_URL = r'https?://[a-z]+\.cbslocal\.com/\d+/\d+/\d+/(?P<id>[0-9a-z-]+)'
  12
  13     _TESTS = [{
  14         # Anvato backend
  15         'url': 'http://losangeles.cbslocal.com/2016/05/16/safety-advocates-say-fatal-car-seat-failures-are-public-health-crisis',
  16         'md5': 'f0ee3081e3843f575fccef901199b212',
  17         'info_dict': {
  18             'id': '3401037',
  19             'ext': 'mp4',
  20             'title': 'Safety Advocates Say Fatal Car Seat Failures Are \'Public Health Crisis\'',
  21             'description': 'Collapsing seats have been the focus of scrutiny for decades, though experts say remarkably little has been done to address the issue. Randy Paige reports.',
  22             'thumbnail': 're:^https?://.*',
  23             'timestamp': 1463440500,
  24             'upload_date': '20160516',
  25             'uploader': 'CBS',
  26             'subtitles': {
  27                 'en': 'mincount:5',
  28             },
  29             'categories': [
  30                 'Stations\\Spoken Word\\KCBSTV',
  31                 'Syndication\\MSN',
  32                 'Syndication\\NDN',
  33                 'Syndication\\AOL',
  34                 'Syndication\\Yahoo',
  35                 'Syndication\\Tribune',
  36                 'Syndication\\Curb.tv',
  37                 'Content\\News'
  38             ],
  39             'tags': ['CBS 2 News Evening'],
  40         },
  41     }, {
  42         # SendtoNews embed
  43         'url': 'http://cleveland.cbslocal.com/2016/05/16/indians-score-season-high-15-runs-in-blowout-win-over-reds-rapid-reaction/',
  44         'info_dict': {
  45             'id': 'GxfCe0Zo7D-175909-5588',
  46         },
  47         'playlist_count': 9,
  48         'params': {
  49             # m3u8 download
  50             'skip_download': True,
  51         },
  52     }]
  53
  54     def _real_extract(self, url):
  55         display_id = self._match_id(url)
  56         webpage = self._download_webpage(url, display_id)
  57
  58         sendtonews_url = SendtoNewsIE._extract_url(webpage)
  59         if sendtonews_url:
  60             return self.url_result(
  61                 compat_urlparse.urljoin(url, sendtonews_url),
  62                 ie=SendtoNewsIE.ie_key())
  63
  64         info_dict = self._extract_anvato_videos(webpage, display_id)
  65
  66         time_str = self._html_search_regex(
  67             r'class="entry-date">([^<]+)<', webpage, 'released date', fatal=False)
  68         timestamp = unified_timestamp(time_str)
  69
  70         info_dict.update({
  71             'display_id': display_id,
  72             'timestamp': timestamp,
  73         })
  74
  75         return info_dict