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             'subtitles': {
  26                 'en': 'mincount:5',
  27             },
  28             'categories': [
  29                 'Stations\\Spoken Word\\KCBSTV',
  30                 'Syndication\\MSN',
  31                 'Syndication\\NDN',
  32                 'Syndication\\AOL',
  33                 'Syndication\\Yahoo',
  34                 'Syndication\\Tribune',
  35                 'Syndication\\Curb.tv',
  36                 'Content\\News'
  37             ],
  38         },
  39     }, {
  40         # SendtoNews embed
  41         'url': 'http://cleveland.cbslocal.com/2016/05/16/indians-score-season-high-15-runs-in-blowout-win-over-reds-rapid-reaction/',
  42         'info_dict': {
  43             'id': 'GxfCe0Zo7D-175909-5588',
  44         },
  45         'playlist_count': 9,
  46         'params': {
  47             # m3u8 download
  48             'skip_download': True,
  49         },
  50     }]
  51
  52     def _real_extract(self, url):
  53         display_id = self._match_id(url)
  54         webpage = self._download_webpage(url, display_id)
  55
  56         sendtonews_url = SendtoNewsIE._extract_url(webpage)
  57         if sendtonews_url:
  58             return self.url_result(
  59                 compat_urlparse.urljoin(url, sendtonews_url),
  60                 ie=SendtoNewsIE.ie_key())
  61
  62         info_dict = self._extract_anvato_videos(webpage, display_id)
  63
  64         time_str = self._html_search_regex(
  65             r'class="entry-date">([^<]+)<', webpage, 'released date', fatal=False)
  66         timestamp = unified_timestamp(time_str)
  67
  68         info_dict.update({
  69             'display_id': display_id,
  70             'timestamp': timestamp,
  71         })
  72
  73         return info_dict