youtube_dl/extractor/groupon.py

   1 from __future__ import unicode_literals
   2
   3 from .common import InfoExtractor
   4
   5
   6 class GrouponIE(InfoExtractor):
   7     _VALID_URL = r'https?://(?:www\.)?groupon\.com/deals/(?P<id>[^/?#&]+)'
   8
   9     _TEST = {
  10         'url': 'https://www.groupon.com/deals/bikram-yoga-huntington-beach-2#ooid=tubGNycTo_9Uxg82uESj4i61EYX8nyuf',
  11         'info_dict': {
  12             'id': 'bikram-yoga-huntington-beach-2',
  13             'title': '$49 for 10 Yoga Classes or One Month of Unlimited Classes at Bikram Yoga Huntington Beach ($180 Value)',
  14             'description': 'Studio kept at 105 degrees and 40% humidity with anti-microbial and anti-slip Flotex flooring; certified instructors',
  15         },
  16         'playlist': [{
  17             'md5': '42428ce8a00585f9bc36e49226eae7a1',
  18             'info_dict': {
  19                 'id': 'fk6OhWpXgIQ',
  20                 'ext': 'mp4',
  21                 'title': 'Bikram Yoga Huntington Beach | Orange County !tubGNycTo@9Uxg82uESj4i61EYX8nyuf',
  22                 'description': 'md5:d41d8cd98f00b204e9800998ecf8427e',
  23                 'duration': 45,
  24                 'upload_date': '20160405',
  25                 'uploader_id': 'groupon',
  26                 'uploader': 'Groupon',
  27             },
  28             'add_ie': ['Youtube'],
  29         }],
  30         'params': {
  31             'skip_download': True,
  32         },
  33     }
  34
  35     _PROVIDERS = {
  36         'ooyala': ('ooyala:%s', 'Ooyala'),
  37         'youtube': ('%s', 'Youtube'),
  38     }
  39
  40     def _real_extract(self, url):
  41         playlist_id = self._match_id(url)
  42         webpage = self._download_webpage(url, playlist_id)
  43
  44         payload = self._parse_json(self._search_regex(
  45             r'(?:var\s+|window\.)payload\s*=\s*(.*?);\n', webpage, 'payload'), playlist_id)
  46         videos = payload['carousel'].get('dealVideos', [])
  47         entries = []
  48         for v in videos:
  49             provider = v.get('provider')
  50             video_id = v.get('media') or v.get('id') or v.get('baseURL')
  51             if not provider or not video_id:
  52                 continue
  53             url_pattern, ie_key = self._PROVIDERS.get(provider.lower())
  54             if not url_pattern:
  55                 self.report_warning(
  56                     '%s: Unsupported video provider %s, skipping video' %
  57                     (playlist_id, provider))
  58                 continue
  59             entries.append(self.url_result(url_pattern % video_id, ie_key))
  60
  61         return {
  62             '_type': 'playlist',
  63             'id': playlist_id,
  64             'entries': entries,
  65             'title': self._og_search_title(webpage),
  66             'description': self._og_search_description(webpage),
  67         }