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

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import calendar
   5 import datetime
   6
   7 from .anvato import AnvatoIE
   8 from .sendtonews import SendtoNewsIE
   9 from ..compat import compat_urlparse
  10
  11
  12 class CBSLocalIE(AnvatoIE):
  13     _VALID_URL = r'https?://[a-z]+\.cbslocal\.com/\d+/\d+/\d+/(?P<id>[0-9a-z-]+)'
  14
  15     _TESTS = [{
  16         # Anvato backend
  17         'url': 'http://losangeles.cbslocal.com/2016/05/16/safety-advocates-say-fatal-car-seat-failures-are-public-health-crisis',
  18         'md5': 'f0ee3081e3843f575fccef901199b212',
  19         'info_dict': {
  20             'id': '3401037',
  21             'ext': 'mp4',
  22             'title': 'Safety Advocates Say Fatal Car Seat Failures Are \'Public Health Crisis\'',
  23             '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.',
  24             'thumbnail': 're:^https?://.*',
  25             'timestamp': 1463440500,
  26             'upload_date': '20160516',
  27             'subtitles': {
  28                 'en': 'mincount:5',
  29             },
  30             'categories': [
  31                 'Stations\\Spoken Word\\KCBSTV',
  32                 'Syndication\\MSN',
  33                 'Syndication\\NDN',
  34                 'Syndication\\AOL',
  35                 'Syndication\\Yahoo',
  36                 'Syndication\\Tribune',
  37                 'Syndication\\Curb.tv',
  38                 'Content\\News'
  39             ],
  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             'ext': 'mp4',
  47             'title': 'Recap: CLE 15, CIN 6',
  48             'description': '5/16/16: Indians\' bats explode for 15 runs in a win',
  49             'upload_date': '20160516',
  50             'timestamp': 1463433840,
  51             'duration': 49,
  52         },
  53         'params': {
  54             # m3u8 download
  55             'skip_download': True,
  56         },
  57     }]
  58
  59     def _real_extract(self, url):
  60         display_id = self._match_id(url)
  61         webpage = self._download_webpage(url, display_id)
  62
  63         sendtonews_url = SendtoNewsIE._extract_url(webpage)
  64         if sendtonews_url:
  65             info_dict = {
  66                 '_type': 'url_transparent',
  67                 'url': compat_urlparse.urljoin(url, sendtonews_url),
  68             }
  69         else:
  70             info_dict = self._extract_anvato_videos(webpage, display_id)
  71
  72         time_str = self._html_search_regex(
  73             r'class="entry-date">([^<]+)<', webpage, 'released date', fatal=False)
  74         timestamp = None
  75         if time_str:
  76             timestamp = calendar.timegm(datetime.datetime.strptime(
  77                 time_str, '%b %d, %Y %I:%M %p').timetuple())
  78
  79         info_dict.update({
  80             'display_id': display_id,
  81             'timestamp': timestamp,
  82         })
  83
  84         return info_dict