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\.|{_SUBDOMAIN_RE}){_DOMAIN_RE}/embed/[^/?#]+\.js\?(?:[^#]+&)?video='
  10     _ID_RE = r'[\da-f]{32}|eyJ[\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).)*)\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/embed/sdk-iframe-integration.fla9.latest.js?video=0e8e040aec776862e1d632a699edf59e',
  29         'info_dict': {
  30             'id': '0e8e040aec776862e1d632a699edf59e',
  31             'ext': 'mp4',
  32             'title': '0e8e040aec776862e1d632a699edf59e',
  33             'thumbnail': 'https://videodelivery.net/0e8e040aec776862e1d632a699edf59e/thumbnails/thumbnail.jpg',
  34         },
  35     }, {
  36         'url': 'https://watch.cloudflarestream.com/9df17203414fd1db3e3ed74abbe936c1',
  37         'only_matching': True,
  38     }, {
  39         'url': 'https://cloudflarestream.com/31c9291ab41fac05471db4e73aa11717/manifest/video.mpd',
  40         'only_matching': True,
  41     }, {
  42         'url': 'https://embed.videodelivery.net/embed/r4xu.fla9.latest.js?video=81d80727f3022488598f68d323c1ad5e',
  43         'only_matching': True,
  44     }, {
  45         'url': 'https://customer-aw5py76sw8wyqzmh.cloudflarestream.com/2463f6d3e06fa29710a337f5f5389fd8/iframe',
  46         'only_matching': True,
  47     }, {
  48         'url': 'https://watch.cloudflarestream.com/eyJhbGciOiJSUzI1NiIsInR5cCI6IkpXVCJ9.eyJraWQiOiJmYTA0YjViMzQ2NDkwYTM5NWJiNzQ1NWFhZTA2YzYwZSIsInN1YiI6Ijg4ZDQxMDhhMzY0MjA3M2VhYmFhZjg3ZGExODJkMjYzIiwiZXhwIjoxNjAwNjA5MzE5fQ.xkRJwLGkt0nZ%5F0BlPiwU7iW4pqb4lKkznbKfAhGg0tGcxSS6ZBA3lcTUwu7W%2DyCFbnAl%2Dhqk3Fn%5FqeQS%5FQydP27qTHpB9iIFFsMtk1tqzGZV5v4yrYDnwLSKzEKvVd6QwJnfABtxH2JdpSNuWlMUiVXFxGWgjOw6QeTNDDklTQYXV%5FNLV7sErSn5CeOPeRRkdXb%2D8ip%5FVOcfk1nDsFoOo4fctFtGP0wYMyY5ae8nhhatydHwevuvJCcEvEfh%2D4qjq9mCZOodevmtSQ4YWmggf4BxtWnDWYrGW8Otp6oqezrR8oY4%2DbKdV6PaqBj49aJdcls6xK7PmM8%5Fvjy3xfm0Mg',
  49         'only_matching': True,
  50     }]
  51     _WEBPAGE_TESTS = [{
  52         'url': 'https://upride.cc/incident/shoulder-pass-at-light/',
  53         'info_dict': {
  54             'id': 'eaef9dea5159cf968be84241b5cedfe7',
  55             'ext': 'mp4',
  56             'title': 'eaef9dea5159cf968be84241b5cedfe7',
  57             'thumbnail': 'https://videodelivery.net/eaef9dea5159cf968be84241b5cedfe7/thumbnails/thumbnail.jpg',
  58         },
  59         'params': {
  60             'skip_download': 'm3u8',
  61         },
  62     }]
  63
  64     def _real_extract(self, url):
  65         video_id = self._match_id(url)
  66         domain = 'bytehighway.net' if 'bytehighway.net/' in url else 'videodelivery.net'
  67         base_url = f'https://{domain}/{video_id}/'
  68         if '.' in video_id:
  69             video_id = self._parse_json(base64.urlsafe_b64decode(
  70                 video_id.split('.')[1] + '==='), video_id)['sub']
  71         manifest_base_url = base_url + 'manifest/video.'
  72
  73         formats, subtitles = self._extract_m3u8_formats_and_subtitles(
  74             manifest_base_url + 'm3u8', video_id, 'mp4',
  75             'm3u8_native', m3u8_id='hls', fatal=False)
  76         fmts, subs = self._extract_mpd_formats_and_subtitles(
  77             manifest_base_url + 'mpd', video_id, mpd_id='dash', fatal=False)
  78         formats.extend(fmts)
  79         self._merge_subtitles(subs, target=subtitles)
  80
  81         return {
  82             'id': video_id,
  83             'title': video_id,
  84             'thumbnail': base_url + 'thumbnails/thumbnail.jpg',
  85             'formats': formats,
  86             'subtitles': subtitles,
  87         }