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 (
   8     parse_iso8601,
   9     unified_timestamp,
  10 )
  11
  12
  13 class CBSLocalIE(AnvatoIE):
  14     _VALID_URL = r'https?://[a-z]+\.cbslocal\.com/(?:\d+/\d+/\d+|video)/(?P<id>[0-9a-z-]+)'
  15
  16     _TESTS = [{
  17         # Anvato backend
  18         'url': 'http://losangeles.cbslocal.com/2016/05/16/safety-advocates-say-fatal-car-seat-failures-are-public-health-crisis',
  19         'md5': 'f0ee3081e3843f575fccef901199b212',
  20         'info_dict': {
  21             'id': '3401037',
  22             'ext': 'mp4',
  23             'title': 'Safety Advocates Say Fatal Car Seat Failures Are \'Public Health Crisis\'',
  24             '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.',
  25             'thumbnail': 're:^https?://.*',
  26             'timestamp': 1463440500,
  27             'upload_date': '20160516',
  28             'uploader': 'CBS',
  29             'subtitles': {
  30                 'en': 'mincount:5',
  31             },
  32             'categories': [
  33                 'Stations\\Spoken Word\\KCBSTV',
  34                 'Syndication\\MSN',
  35                 'Syndication\\NDN',
  36                 'Syndication\\AOL',
  37                 'Syndication\\Yahoo',
  38                 'Syndication\\Tribune',
  39                 'Syndication\\Curb.tv',
  40                 'Content\\News'
  41             ],
  42             'tags': ['CBS 2 News Evening'],
  43         },
  44     }, {
  45         # SendtoNews embed
  46         'url': 'http://cleveland.cbslocal.com/2016/05/16/indians-score-season-high-15-runs-in-blowout-win-over-reds-rapid-reaction/',
  47         'info_dict': {
  48             'id': 'GxfCe0Zo7D-175909-5588',
  49         },
  50         'playlist_count': 9,
  51         'params': {
  52             # m3u8 download
  53             'skip_download': True,
  54         },
  55     }, {
  56         'url': 'http://newyork.cbslocal.com/video/3580809-a-very-blue-anniversary/',
  57         'info_dict': {
  58             'id': '3580809',
  59             'ext': 'mp4',
  60             'title': 'A Very Blue Anniversary',
  61             'description': 'CBS2’s Cindy Hsu has more.',
  62             'thumbnail': 're:^https?://.*',
  63             'timestamp': 1479962220,
  64             'upload_date': '20161124',
  65             'uploader': 'CBS',
  66             'subtitles': {
  67                 'en': 'mincount:5',
  68             },
  69             'categories': [
  70                 'Stations\\Spoken Word\\WCBSTV',
  71                 'Syndication\\AOL',
  72                 'Syndication\\MSN',
  73                 'Syndication\\NDN',
  74                 'Syndication\\Yahoo',
  75                 'Content\\News',
  76                 'Content\\News\\Local News',
  77             ],
  78             'tags': ['CBS 2 News Weekends', 'Cindy Hsu', 'Blue Man Group'],
  79         },
  80     }]
  81
  82     def _real_extract(self, url):
  83         display_id = self._match_id(url)
  84         webpage = self._download_webpage(url, display_id)
  85
  86         sendtonews_url = SendtoNewsIE._extract_url(webpage)
  87         if sendtonews_url:
  88             return self.url_result(
  89                 compat_urlparse.urljoin(url, sendtonews_url),
  90                 ie=SendtoNewsIE.ie_key())
  91
  92         info_dict = self._extract_anvato_videos(webpage, display_id)
  93
  94         time_str = self._html_search_regex(
  95             r'class="entry-date">([^<]+)<', webpage, 'released date', default=None)
  96         if time_str:
  97             timestamp = unified_timestamp(time_str)
  98         else:
  99             timestamp = parse_iso8601(self._html_search_meta('uploadDate', webpage))
 100
 101         info_dict.update({
 102             'display_id': display_id,
 103             'timestamp': timestamp,
 104         })
 105
 106         return info_dict