yt_dlp/extractor/cloudflarestream.py

   1 import base64
   2 import re
   3
   4 from .common import InfoExtractor
   5
   6
   7 class CloudflareStreamIE(InfoExtractor):
   8     _DOMAIN_RE = r'(?:cloudflarestream\.com|(?:videodelivery|bytehighway)\.net)'
   9     _EMBED_RE = r'embed\.%s/embed/[^/]+\.js\?.*?\bvideo=' % _DOMAIN_RE
  10     _ID_RE = r'[\da-f]{32}|[\w-]+\.[\w-]+\.[\w-]+'
  11     _VALID_URL = r'''(?x)
  12                     https?://
  13                         (?:
  14                             (?:watch\.)?%s/|
  15                             %s
  16                         )
  17                         (?P<id>%s)
  18                     ''' % (_DOMAIN_RE, _EMBED_RE, _ID_RE)
  19     _TESTS = [{
  20         'url': 'https://embed.cloudflarestream.com/embed/we4g.fla9.latest.js?video=31c9291ab41fac05471db4e73aa11717',
  21         'info_dict': {
  22             'id': '31c9291ab41fac05471db4e73aa11717',
  23             'ext': 'mp4',
  24             'title': '31c9291ab41fac05471db4e73aa11717',
  25         },
  26         'params': {
  27             'skip_download': True,
  28         },
  29     }, {
  30         'url': 'https://watch.cloudflarestream.com/9df17203414fd1db3e3ed74abbe936c1',
  31         'only_matching': True,
  32     }, {
  33         'url': 'https://cloudflarestream.com/31c9291ab41fac05471db4e73aa11717/manifest/video.mpd',
  34         'only_matching': True,
  35     }, {
  36         'url': 'https://embed.videodelivery.net/embed/r4xu.fla9.latest.js?video=81d80727f3022488598f68d323c1ad5e',
  37         'only_matching': True,
  38     }]
  39
  40     @staticmethod
  41     def _extract_urls(webpage):
  42         return [
  43             mobj.group('url')
  44             for mobj in re.finditer(
  45                 r'<script[^>]+\bsrc=(["\'])(?P<url>(?:https?:)?//%s(?:%s).*?)\1' % (CloudflareStreamIE._EMBED_RE, CloudflareStreamIE._ID_RE),
  46                 webpage)]
  47
  48     def _real_extract(self, url):
  49         video_id = self._match_id(url)
  50         domain = 'bytehighway.net' if 'bytehighway.net/' in url else 'videodelivery.net'
  51         base_url = 'https://%s/%s/' % (domain, video_id)
  52         if '.' in video_id:
  53             video_id = self._parse_json(base64.urlsafe_b64decode(
  54                 video_id.split('.')[1] + '==='), video_id)['sub']
  55         manifest_base_url = base_url + 'manifest/video.'
  56
  57         formats = self._extract_m3u8_formats(
  58             manifest_base_url + 'm3u8', video_id, 'mp4',
  59             'm3u8_native', m3u8_id='hls', fatal=False)
  60         formats.extend(self._extract_mpd_formats(
  61             manifest_base_url + 'mpd', video_id, mpd_id='dash', fatal=False))
  62         self._sort_formats(formats)
  63
  64         return {
  65             'id': video_id,
  66             'title': video_id,
  67             'thumbnail': base_url + 'thumbnails/thumbnail.jpg',
  68             'formats': formats,
  69         }