yt_dlp/extractor/jwplatform.py

   1 import re
   2
   3 from .common import InfoExtractor
   4 from ..utils import unsmuggle_url
   5
   6
   7 class JWPlatformIE(InfoExtractor):
   8     _VALID_URL = r'(?:https?://(?:content\.jwplatform|cdn\.jwplayer)\.com/(?:(?:feed|player|thumb|preview|manifest)s|jw6|v2/media)/|jwplatform:)(?P<id>[a-zA-Z0-9]{8})'
   9     _TESTS = [{
  10         'url': 'http://content.jwplatform.com/players/nPripu9l-ALJ3XQCI.js',
  11         'md5': '3aa16e4f6860e6e78b7df5829519aed3',
  12         'info_dict': {
  13             'id': 'nPripu9l',
  14             'ext': 'mp4',
  15             'title': 'Big Buck Bunny Trailer',
  16             'description': 'Big Buck Bunny is a short animated film by the Blender Institute. It is made using free and open source software.',
  17             'upload_date': '20081127',
  18             'timestamp': 1227796140,
  19             'duration': 32.0,
  20             'thumbnail': 'https://cdn.jwplayer.com/v2/media/nPripu9l/poster.jpg?width=720',
  21         },
  22     }, {
  23         'url': 'https://cdn.jwplayer.com/players/nPripu9l-ALJ3XQCI.js',
  24         'only_matching': True,
  25     }]
  26
  27     _WEBPAGE_TESTS = [{
  28         # JWPlatform iframe
  29         'url': 'https://www.covermagazine.co.uk/feature/2465255/business-protection-involved',
  30         'info_dict': {
  31             'id': 'AG26UQXM',
  32             'ext': 'mp4',
  33             'upload_date': '20160719',
  34             'timestamp': 1468923808,
  35             'title': '2016_05_18 Cover L&G Business Protection V1 FINAL.mp4',
  36             'thumbnail': 'https://cdn.jwplayer.com/v2/media/AG26UQXM/poster.jpg?width=720',
  37             'description': '',
  38             'duration': 294.0,
  39         },
  40     }, {
  41         # Player url not surrounded by quotes
  42         'url': 'https://www.deutsche-kinemathek.de/en/online/streaming/school-trip',
  43         'info_dict': {
  44             'id': 'jUxh5uin',
  45             'title': 'Klassenfahrt',
  46             'ext': 'mp4',
  47             'upload_date': '20230109',
  48             'thumbnail': 'https://cdn.jwplayer.com/v2/media/jUxh5uin/poster.jpg?width=720',
  49             'timestamp': 1673270298,
  50             'description': '',
  51             'duration': 5193.0,
  52         },
  53         'params': {'allowed_extractors': ['generic', 'jwplatform']},
  54     }, {
  55         # iframe src attribute includes backslash before URL string
  56         'url': 'https://www.elespectador.com/colombia/video-asi-se-evito-la-fuga-de-john-poulos-presunto-feminicida-de-valentina-trespalacios-explicacion',
  57         'info_dict': {
  58             'id': 'QD3gsexj',
  59             'title': 'Así se evitó la fuga de John Poulos, presunto feminicida de Valentina Trespalacios',
  60             'ext': 'mp4',
  61             'upload_date': '20230127',
  62             'thumbnail': 'https://cdn.jwplayer.com/v2/media/QD3gsexj/poster.jpg?width=720',
  63             'timestamp': 1674862986,
  64             'description': 'md5:128fd74591c4e1fc2da598c5cb6f5ce4',
  65             'duration': 263.0,
  66         },
  67     }]
  68
  69     @classmethod
  70     def _extract_embed_urls(cls, url, webpage):
  71         for tag, key in ((r'(?:script|iframe)', 'src'), ('input', 'value')):
  72             # <input value=URL> is used by hyland.com
  73             # if we find <iframe>, dont look for <input>
  74             ret = re.findall(
  75                 rf'<{tag}[^>]+?{key}=\\?["\']?((?:https?:)?//(?:content\.jwplatform|cdn\.jwplayer)\.com/players/[a-zA-Z0-9]{{8}})',
  76                 webpage)
  77             if ret:
  78                 return ret
  79         mobj = re.search(r'<div\b[^>]* data-video-jw-id="([a-zA-Z0-9]{8})"', webpage)
  80         if mobj:
  81             return [f'jwplatform:{mobj.group(1)}']
  82
  83     def _real_extract(self, url):
  84         url, smuggled_data = unsmuggle_url(url, {})
  85         self._initialize_geo_bypass({
  86             'countries': smuggled_data.get('geo_countries'),
  87         })
  88         video_id = self._match_id(url)
  89         json_data = self._download_json('https://cdn.jwplayer.com/v2/media/' + video_id, video_id)
  90         return self._parse_jwplayer_data(json_data, video_id)