yt_dlp/extractor/tbs.py

   1 import re
   2
   3 from .turner import TurnerBaseIE
   4 from ..compat import (
   5     compat_urllib_parse_urlparse,
   6     compat_parse_qs,
   7 )
   8 from ..utils import (
   9     float_or_none,
  10     int_or_none,
  11     strip_or_none,
  12 )
  13
  14
  15 class TBSIE(TurnerBaseIE):
  16     _VALID_URL = r'https?://(?:www\.)?(?P<site>tbs|tntdrama)\.com(?P<path>/(?:movies|watchtnt|watchtbs|shows/[^/]+/(?:clips|season-\d+/episode-\d+))/(?P<id>[^/?#]+))'
  17     _TESTS = [{
  18         'url': 'http://www.tntdrama.com/shows/the-alienist/clips/monster',
  19         'info_dict': {
  20             'id': '8d384cde33b89f3a43ce5329de42903ed5099887',
  21             'ext': 'mp4',
  22             'title': 'Monster',
  23             'description': 'Get a first look at the theatrical trailer for TNT’s highly anticipated new psychological thriller The Alienist, which premieres January 22 on TNT.',
  24             'timestamp': 1508175329,
  25             'upload_date': '20171016',
  26         },
  27         'params': {
  28             # m3u8 download
  29             'skip_download': True,
  30         }
  31     }, {
  32         'url': 'http://www.tbs.com/shows/search-party/season-1/episode-1/explicit-the-mysterious-disappearance-of-the-girl-no-one-knew',
  33         'only_matching': True,
  34     }, {
  35         'url': 'http://www.tntdrama.com/movies/star-wars-a-new-hope',
  36         'only_matching': True,
  37     }]
  38
  39     def _real_extract(self, url):
  40         site, path, display_id = self._match_valid_url(url).groups()
  41         webpage = self._download_webpage(url, display_id)
  42         drupal_settings = self._parse_json(self._search_regex(
  43             r'<script[^>]+?data-drupal-selector="drupal-settings-json"[^>]*?>({.+?})</script>',
  44             webpage, 'drupal setting'), display_id)
  45         isLive = 'watchtnt' in path or 'watchtbs' in path
  46         video_data = next(v for v in drupal_settings['turner_playlist'] if isLive or v.get('url') == path)
  47
  48         media_id = video_data['mediaID']
  49         title = video_data['title']
  50         tokenizer_query = compat_parse_qs(compat_urllib_parse_urlparse(
  51             drupal_settings['ngtv_token_url']).query)
  52
  53         info = self._extract_ngtv_info(
  54             media_id, tokenizer_query, {
  55                 'url': url,
  56                 'site_name': site[:3].upper(),
  57                 'auth_required': video_data.get('authRequired') == '1' or isLive,
  58                 'is_live': isLive
  59             })
  60
  61         thumbnails = []
  62         for image_id, image in video_data.get('images', {}).items():
  63             image_url = image.get('url')
  64             if not image_url or image.get('type') != 'video':
  65                 continue
  66             i = {
  67                 'id': image_id,
  68                 'url': image_url,
  69             }
  70             mobj = re.search(r'(\d+)x(\d+)', image_url)
  71             if mobj:
  72                 i.update({
  73                     'width': int(mobj.group(1)),
  74                     'height': int(mobj.group(2)),
  75                 })
  76             thumbnails.append(i)
  77
  78         info.update({
  79             'id': media_id,
  80             'title': title,
  81             'description': strip_or_none(video_data.get('descriptionNoTags') or video_data.get('shortDescriptionNoTags')),
  82             'duration': float_or_none(video_data.get('duration')) or info.get('duration'),
  83             'timestamp': int_or_none(video_data.get('created')),
  84             'season_number': int_or_none(video_data.get('season')),
  85             'episode_number': int_or_none(video_data.get('episode')),
  86             'thumbnails': thumbnails,
  87             'is_live': isLive
  88         })
  89         return info