yt_dlp/extractor/stv.py

   1 from .common import InfoExtractor
   2 from ..compat import compat_str
   3 from ..utils import (
   4     float_or_none,
   5     int_or_none,
   6     smuggle_url,
   7     str_or_none,
   8     try_get,
   9 )
  10
  11
  12 class STVPlayerIE(InfoExtractor):
  13     IE_NAME = 'stv:player'
  14     _VALID_URL = r'https?://player\.stv\.tv/(?P<type>episode|video)/(?P<id>[a-z0-9]{4})'
  15     _TESTS = [{
  16         # shortform
  17         'url': 'https://player.stv.tv/video/4gwd/emmerdale/60-seconds-on-set-with-laura-norton/',
  18         'md5': '5adf9439c31d554f8be0707c7abe7e0a',
  19         'info_dict': {
  20             'id': '5333973339001',
  21             'ext': 'mp4',
  22             'upload_date': '20170301',
  23             'title': '60 seconds on set with Laura Norton',
  24             'description': "How many questions can Laura - a.k.a Kerry Wyatt - answer in 60 seconds? Let\'s find out!",
  25             'timestamp': 1488388054,
  26             'uploader_id': '1486976045',
  27         },
  28         'skip': 'this resource is unavailable outside of the UK',
  29     }, {
  30         # episodes
  31         'url': 'https://player.stv.tv/episode/4125/jennifer-saunders-memory-lane',
  32         'only_matching': True,
  33     }]
  34     BRIGHTCOVE_URL_TEMPLATE = 'http://players.brightcove.net/1486976045/default_default/index.html?videoId=%s'
  35     _PTYPE_MAP = {
  36         'episode': 'episodes',
  37         'video': 'shortform',
  38     }
  39
  40     def _real_extract(self, url):
  41         ptype, video_id = self._match_valid_url(url).groups()
  42
  43         webpage = self._download_webpage(url, video_id, fatal=False) or ''
  44         props = self._search_nextjs_data(webpage, video_id, default='{}').get('props') or {}
  45         player_api_cache = try_get(
  46             props, lambda x: x['initialReduxState']['playerApiCache']) or {}
  47
  48         api_path, resp = None, {}
  49         for k, v in player_api_cache.items():
  50             if k.startswith('/episodes/') or k.startswith('/shortform/'):
  51                 api_path, resp = k, v
  52                 break
  53         else:
  54             episode_id = str_or_none(try_get(
  55                 props, lambda x: x['pageProps']['episodeId']))
  56             api_path = '/%s/%s' % (self._PTYPE_MAP[ptype], episode_id or video_id)
  57
  58         result = resp.get('results')
  59         if not result:
  60             resp = self._download_json(
  61                 'https://player.api.stv.tv/v1' + api_path, video_id)
  62             result = resp['results']
  63
  64         video = result['video']
  65         video_id = compat_str(video['id'])
  66
  67         subtitles = {}
  68         _subtitles = result.get('_subtitles') or {}
  69         for ext, sub_url in _subtitles.items():
  70             subtitles.setdefault('en', []).append({
  71                 'ext': 'vtt' if ext == 'webvtt' else ext,
  72                 'url': sub_url,
  73             })
  74
  75         programme = result.get('programme') or {}
  76         if programme.get('drmEnabled'):
  77             self.report_drm(video_id)
  78
  79         return {
  80             '_type': 'url_transparent',
  81             'id': video_id,
  82             'url': smuggle_url(self.BRIGHTCOVE_URL_TEMPLATE % video_id, {'geo_countries': ['GB']}),
  83             'description': result.get('summary'),
  84             'duration': float_or_none(video.get('length'), 1000),
  85             'subtitles': subtitles,
  86             'view_count': int_or_none(result.get('views')),
  87             'series': programme.get('name') or programme.get('shortName'),
  88             'ie_key': 'BrightcoveNew',
  89         }