yt_dlp/extractor/cbslocal.py

   1 from .anvato import AnvatoIE
   2 from .sendtonews import SendtoNewsIE
   3 from ..compat import compat_urlparse
   4 from ..utils import (
   5     parse_iso8601,
   6     unified_timestamp,
   7 )
   8
   9
  10 class CBSLocalIE(AnvatoIE):  # XXX: Do not subclass from concrete IE
  11     _VALID_URL_BASE = r'https?://[a-z]+\.cbslocal\.com/'
  12     _VALID_URL = _VALID_URL_BASE + r'video/(?P<id>\d+)'
  13
  14     _TESTS = [{
  15         'url': 'http://newyork.cbslocal.com/video/3580809-a-very-blue-anniversary/',
  16         'info_dict': {
  17             'id': '3580809',
  18             'ext': 'mp4',
  19             'title': 'A Very Blue Anniversary',
  20             'description': 'CBS2’s Cindy Hsu has more.',
  21             'thumbnail': 're:^https?://.*',
  22             'timestamp': int,
  23             'upload_date': r're:^\d{8}$',
  24             'uploader': 'CBS',
  25             'subtitles': {
  26                 'en': 'mincount:5',
  27             },
  28             'categories': [
  29                 'Stations\\Spoken Word\\WCBSTV',
  30                 'Syndication\\AOL',
  31                 'Syndication\\MSN',
  32                 'Syndication\\NDN',
  33                 'Syndication\\Yahoo',
  34                 'Content\\News',
  35                 'Content\\News\\Local News',
  36             ],
  37             'tags': ['CBS 2 News Weekends', 'Cindy Hsu', 'Blue Man Group'],
  38         },
  39         'params': {
  40             'skip_download': True,
  41         },
  42     }]
  43
  44     def _real_extract(self, url):
  45         mcp_id = self._match_id(url)
  46         return self.url_result(
  47             'anvato:anvato_cbslocal_app_web_prod_547f3e49241ef0e5d30c79b2efbca5d92c698f67:' + mcp_id, 'Anvato', mcp_id)
  48
  49
  50 class CBSLocalArticleIE(AnvatoIE):  # XXX: Do not subclass from concrete IE
  51     _VALID_URL = CBSLocalIE._VALID_URL_BASE + r'\d+/\d+/\d+/(?P<id>[0-9a-z-]+)'
  52
  53     _TESTS = [{
  54         # Anvato backend
  55         'url': 'http://losangeles.cbslocal.com/2016/05/16/safety-advocates-say-fatal-car-seat-failures-are-public-health-crisis',
  56         'md5': 'f0ee3081e3843f575fccef901199b212',
  57         'info_dict': {
  58             'id': '3401037',
  59             'ext': 'mp4',
  60             'title': 'Safety Advocates Say Fatal Car Seat Failures Are \'Public Health Crisis\'',
  61             '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.',
  62             'thumbnail': 're:^https?://.*',
  63             'timestamp': 1463440500,
  64             'upload_date': '20160516',
  65             'uploader': 'CBS',
  66             'subtitles': {
  67                 'en': 'mincount:5',
  68             },
  69             'categories': [
  70                 'Stations\\Spoken Word\\KCBSTV',
  71                 'Syndication\\MSN',
  72                 'Syndication\\NDN',
  73                 'Syndication\\AOL',
  74                 'Syndication\\Yahoo',
  75                 'Syndication\\Tribune',
  76                 'Syndication\\Curb.tv',
  77                 'Content\\News'
  78             ],
  79             'tags': ['CBS 2 News Evening'],
  80         },
  81     }, {
  82         # SendtoNews embed
  83         'url': 'http://cleveland.cbslocal.com/2016/05/16/indians-score-season-high-15-runs-in-blowout-win-over-reds-rapid-reaction/',
  84         'info_dict': {
  85             'id': 'GxfCe0Zo7D-175909-5588',
  86         },
  87         'playlist_count': 9,
  88         'params': {
  89             # m3u8 download
  90             'skip_download': True,
  91         },
  92     }]
  93
  94     def _real_extract(self, url):
  95         display_id = self._match_id(url)
  96         webpage = self._download_webpage(url, display_id)
  97
  98         sendtonews_url = SendtoNewsIE._extract_url(webpage)
  99         if sendtonews_url:
 100             return self.url_result(
 101                 compat_urlparse.urljoin(url, sendtonews_url),
 102                 ie=SendtoNewsIE.ie_key())
 103
 104         info_dict = self._extract_anvato_videos(webpage, display_id)
 105
 106         timestamp = unified_timestamp(self._html_search_regex(
 107             r'class="(?:entry|post)-date"[^>]*>([^<]+)', webpage,
 108             'released date', default=None)) or parse_iso8601(
 109             self._html_search_meta('uploadDate', webpage))
 110
 111         info_dict.update({
 112             'display_id': display_id,
 113             'timestamp': timestamp,
 114         })
 115
 116         return info_dict