yt_dlp/extractor/cloudflarestream.py

   1 import base64
   2
   3 from .common import InfoExtractor
   4
   5
   6 class CloudflareStreamIE(InfoExtractor):
   7     _SUBDOMAIN_RE = r'(?:(?:watch|iframe|customer-\w+)\.)?'
   8     _DOMAIN_RE = r'(?:cloudflarestream\.com|(?:videodelivery|bytehighway)\.net)'
   9     _EMBED_RE = rf'embed\.{_DOMAIN_RE}/embed/[^/]+\.js\?.*?\bvideo='
  10     _ID_RE = r'[\da-f]{32}|[\w-]+\.[\w-]+\.[\w-]+'
  11     _VALID_URL = rf'https?://(?:{_SUBDOMAIN_RE}{_DOMAIN_RE}/|{_EMBED_RE})(?P<id>{_ID_RE})'
  12     _EMBED_REGEX = [
  13         rf'<script[^>]+\bsrc=(["\'])(?P<url>(?:https?:)?//{_EMBED_RE}(?:{_ID_RE}).*?)\1',
  14         rf'<iframe[^>]+\bsrc=["\'](?P<url>https?://{_SUBDOMAIN_RE}{_DOMAIN_RE}/[\da-f]{{32}})',
  15     ]
  16     _TESTS = [{
  17         'url': 'https://embed.cloudflarestream.com/embed/we4g.fla9.latest.js?video=31c9291ab41fac05471db4e73aa11717',
  18         'info_dict': {
  19             'id': '31c9291ab41fac05471db4e73aa11717',
  20             'ext': 'mp4',
  21             'title': '31c9291ab41fac05471db4e73aa11717',
  22             'thumbnail': 'https://videodelivery.net/31c9291ab41fac05471db4e73aa11717/thumbnails/thumbnail.jpg',
  23         },
  24         'params': {
  25             'skip_download': 'm3u8',
  26         },
  27     }, {
  28         'url': 'https://watch.cloudflarestream.com/9df17203414fd1db3e3ed74abbe936c1',
  29         'only_matching': True,
  30     }, {
  31         'url': 'https://cloudflarestream.com/31c9291ab41fac05471db4e73aa11717/manifest/video.mpd',
  32         'only_matching': True,
  33     }, {
  34         'url': 'https://embed.videodelivery.net/embed/r4xu.fla9.latest.js?video=81d80727f3022488598f68d323c1ad5e',
  35         'only_matching': True,
  36     }, {
  37         'url': 'https://customer-aw5py76sw8wyqzmh.cloudflarestream.com/2463f6d3e06fa29710a337f5f5389fd8/iframe',
  38         'only_matching': True,
  39     }]
  40     _WEBPAGE_TESTS = [{
  41         'url': 'https://upride.cc/incident/shoulder-pass-at-light/',
  42         'info_dict': {
  43             'id': 'eaef9dea5159cf968be84241b5cedfe7',
  44             'ext': 'mp4',
  45             'title': 'eaef9dea5159cf968be84241b5cedfe7',
  46             'thumbnail': 'https://videodelivery.net/eaef9dea5159cf968be84241b5cedfe7/thumbnails/thumbnail.jpg',
  47         },
  48         'params': {
  49             'skip_download': 'm3u8',
  50         },
  51     }]
  52
  53     def _real_extract(self, url):
  54         video_id = self._match_id(url)
  55         domain = 'bytehighway.net' if 'bytehighway.net/' in url else 'videodelivery.net'
  56         base_url = 'https://%s/%s/' % (domain, video_id)
  57         if '.' in video_id:
  58             video_id = self._parse_json(base64.urlsafe_b64decode(
  59                 video_id.split('.')[1] + '==='), video_id)['sub']
  60         manifest_base_url = base_url + 'manifest/video.'
  61
  62         formats, subtitles = self._extract_m3u8_formats_and_subtitles(
  63             manifest_base_url + 'm3u8', video_id, 'mp4',
  64             'm3u8_native', m3u8_id='hls', fatal=False)
  65         fmts, subs = self._extract_mpd_formats_and_subtitles(
  66             manifest_base_url + 'mpd', video_id, mpd_id='dash', fatal=False)
  67         formats.extend(fmts)
  68         self._merge_subtitles(subs, target=subtitles)
  69
  70         return {
  71             'id': video_id,
  72             'title': video_id,
  73             'thumbnail': base_url + 'thumbnails/thumbnail.jpg',
  74             'formats': formats,
  75             'subtitles': subtitles,
  76         }