yt_dlp/extractor/mediastream.py

   1 import re
   2
   3 from .common import InfoExtractor
   4 from ..utils import (
   5     clean_html,
   6     filter_dict,
   7     parse_qs,
   8     remove_end,
   9     traverse_obj,
  10     update_url_query,
  11     urljoin,
  12 )
  13
  14
  15 class MediaStreamBaseIE(InfoExtractor):
  16     _EMBED_BASE_URL = 'https://mdstrm.com/embed'
  17     _BASE_URL_RE = r'https?://mdstrm\.com/(?:embed|live-stream)'
  18
  19     def _extract_mediastream_urls(self, webpage):
  20         yield from traverse_obj(list(self._yield_json_ld(webpage, None, default={})), (
  21             lambda _, v: v['@type'] == 'VideoObject', ('embedUrl', 'contentUrl'),
  22             {lambda x: x if re.match(rf'{self._BASE_URL_RE}/\w+', x) else None}))
  23
  24         for mobj in re.finditer(r'<script[^>]+>[^>]*playerMdStream\.mdstreamVideo\(\s*[\'"](?P<video_id>\w+)', webpage):
  25             yield f'{self._EMBED_BASE_URL}/{mobj.group("video_id")}'
  26
  27         yield from re.findall(
  28             rf'<iframe[^>]+\bsrc="({self._BASE_URL_RE}/\w+)', webpage)
  29
  30         for mobj in re.finditer(
  31             r'''(?x)
  32                 <(?:div|ps-mediastream)[^>]+
  33                 (class="[^"]*MediaStreamVideoPlayer)[^"]*"[^>]+
  34                 data-video-id="(?P<video_id>\w+)"
  35                 (?:\s*data-video-type="(?P<video_type>[^"]+))?
  36                 (?:[^>]*>\s*<div[^>]+\1[^"]*"[^>]+data-mediastream=["\'][^>]+
  37                     https://mdstrm\.com/(?P<live>live-stream))?
  38                 ''', webpage):
  39
  40             video_type = 'live-stream' if mobj.group('video_type') == 'live' or mobj.group('live') else 'embed'
  41             yield f'https://mdstrm.com/{video_type}/{mobj.group("video_id")}'
  42
  43
  44 class MediaStreamIE(MediaStreamBaseIE):
  45     _VALID_URL = MediaStreamBaseIE._BASE_URL_RE + r'/(?P<id>\w+)'
  46
  47     _TESTS = [{
  48         'url': 'https://mdstrm.com/embed/6318e3f1d1d316083ae48831',
  49         'md5': '97b4f2634b8e8612cc574dfcd504df05',
  50         'info_dict': {
  51             'id': '6318e3f1d1d316083ae48831',
  52             'title': 'Video: Así fue el despido de Thomas Tuchel del Chelsea',
  53             'description': 'md5:358ce1e1396010d50a1ece1be3633c95',
  54             'thumbnail': r're:^https?://[^?#]+6318e3f1d1d316083ae48831',
  55             'ext': 'mp4',
  56         },
  57         'params': {'skip_download': 'm3u8'},
  58     }]
  59
  60     _WEBPAGE_TESTS = [{
  61         'url': 'https://www.multimedios.com/video/costa-rica-tv-en-vivo/v2616',
  62         'info_dict': {
  63             'id': '5a7b1e63a8da282c34d65445',
  64             'title': 're:mmtv-costarica',
  65             'description': 'mmtv-costarica',
  66             'thumbnail': 're:^https?://[^?#]+5a7b1e63a8da282c34d65445',
  67             'ext': 'mp4',
  68             'live_status': 'is_live',
  69         },
  70         'params': {'skip_download': 'Livestream'},
  71     }, {
  72         'url': 'https://www.multimedios.com/television/clases-de-llaves-y-castigos-quien-sabe-mas',
  73         'md5': 'de31f0b1ecc321fb35bf22d58734ea40',
  74         'info_dict': {
  75             'id': '63731bab8ec9b308a2c9ed28',
  76             'title': 'Clases de llaves y castigos ¿Quién sabe más?',
  77             'description': 'md5:1b49aa1ee5a4b32fbd66104b2d629e9d',
  78             'thumbnail': 're:^https?://[^?#]+63731bab8ec9b308a2c9ed28',
  79             'ext': 'mp4',
  80         },
  81         'params': {'skip_download': 'm3u8'},
  82     }, {
  83         'url': 'https://www.americatv.com.pe/videos/esto-es-guerra/facundo-gonzalez-sufrio-fuerte-golpe-durante-competencia-frente-hugo-garcia-eeg-noticia-139120',
  84         'info_dict': {
  85             'id': '63756df1c638b008a5659dec',
  86             'title': 'Facundo González sufrió fuerte golpe durante competencia frente a Hugo García en EEG',
  87             'description': 'md5:9490c034264afd756eef7b2c3adee69e',
  88             'thumbnail': 're:^https?://[^?#]+63756df1c638b008a5659dec',
  89             'ext': 'mp4',
  90         },
  91         'params': {'skip_download': 'm3u8'},
  92     }, {
  93         'url': 'https://www.americatv.com.pe/videos/al-fondo-hay-sitio/nuevas-lomas-town-bernardo-mata-se-enfrento-sujeto-luchar-amor-macarena-noticia-139083',
  94         'info_dict': {
  95             'id': '637307669609130f74cd3a6e',
  96             'title': 'Las Nuevas Lomas Town: Bernardo De La Mata se enfrentó a sujeto para luchar por el amor de Macarena',
  97             'description': 'md5:60d71772f1e1496923539ae58aa17124',
  98             'thumbnail': 're:^https?://[^?#]+637307669609130f74cd3a6e',
  99             'ext': 'mp4',
 100         },
 101         'params': {'skip_download': 'm3u8'},
 102     }]
 103
 104     def _extract_from_webpage(self, url, webpage):
 105         for embed_url in self._extract_mediastream_urls(webpage):
 106             yield self.url_result(embed_url, MediaStreamIE, None)
 107
 108     def _real_extract(self, url):
 109         video_id = self._match_id(url)
 110         webpage = self._download_webpage(url, video_id)
 111
 112         for message in [
 113             'Debido a tu ubicación no puedes ver el contenido',
 114             'You are not allowed to watch this video: Geo Fencing Restriction',
 115             'Este contenido no está disponible en tu zona geográfica.',
 116             'El contenido sólo está disponible dentro de',
 117         ]:
 118             if message in webpage:
 119                 self.raise_geo_restricted()
 120
 121         player_config = self._search_json(r'window\.MDSTRM\.OPTIONS\s*=', webpage, 'metadata', video_id)
 122
 123         formats, subtitles = [], {}
 124         for video_format in player_config['src']:
 125             if video_format == 'hls':
 126                 params = {
 127                     'at': 'web-app',
 128                     'access_token': traverse_obj(parse_qs(url), ('access_token', 0)),
 129                 }
 130                 for name, key in (('MDSTRMUID', 'uid'), ('MDSTRMSID', 'sid'), ('MDSTRMPID', 'pid'), ('VERSION', 'av')):
 131                     params[key] = self._search_regex(
 132                         rf'window\.{name}\s*=\s*["\']([^"\']+)["\'];', webpage, key, default=None)
 133
 134                 fmts, subs = self._extract_m3u8_formats_and_subtitles(
 135                     update_url_query(player_config['src'][video_format], filter_dict(params)), video_id)
 136                 formats.extend(fmts)
 137                 self._merge_subtitles(subs, target=subtitles)
 138             elif video_format == 'mpd':
 139                 fmts, subs = self._extract_mpd_formats_and_subtitles(player_config['src'][video_format], video_id)
 140                 formats.extend(fmts)
 141                 self._merge_subtitles(subs, target=subtitles)
 142             else:
 143                 formats.append({
 144                     'url': player_config['src'][video_format],
 145                 })
 146
 147         return {
 148             'id': video_id,
 149             'title': self._og_search_title(webpage) or player_config.get('title'),
 150             'description': self._og_search_description(webpage),
 151             'formats': formats,
 152             'subtitles': subtitles,
 153             'is_live': player_config.get('type') == 'live',
 154             'thumbnail': self._og_search_thumbnail(webpage),
 155         }
 156
 157
 158 class WinSportsVideoIE(MediaStreamBaseIE):
 159     _VALID_URL = r'https?://www\.winsports\.co/videos/(?P<id>[\w-]+)'
 160
 161     _TESTS = [{
 162         'url': 'https://www.winsports.co/videos/siempre-castellanos-gran-atajada-del-portero-cardenal-para-evitar-la-caida-de-su-arco-60536',
 163         'info_dict': {
 164             'id': '62dc8357162c4b0821fcfb3c',
 165             'display_id': 'siempre-castellanos-gran-atajada-del-portero-cardenal-para-evitar-la-caida-de-su-arco-60536',
 166             'title': '¡Siempre Castellanos! Gran atajada del portero \'cardenal\' para evitar la caída de su arco',
 167             'description': 'md5:eb811b2b2882bdc59431732c06b905f2',
 168             'thumbnail': r're:^https?://[^?#]+62dc8357162c4b0821fcfb3c',
 169             'ext': 'mp4',
 170         },
 171         'params': {'skip_download': 'm3u8'},
 172     }, {
 173         'url': 'https://www.winsports.co/videos/observa-aqui-los-goles-del-empate-entre-tolima-y-nacional-60548',
 174         'info_dict': {
 175             'id': '62dcb875ef12a5526790b552',
 176             'display_id': 'observa-aqui-los-goles-del-empate-entre-tolima-y-nacional-60548',
 177             'title': 'Observa aquí los goles del empate entre Tolima y Nacional',
 178             'description': 'md5:b19402ba6e46558b93fd24b873eea9c9',
 179             'thumbnail': r're:^https?://[^?#]+62dcb875ef12a5526790b552',
 180             'ext': 'mp4',
 181         },
 182         'params': {'skip_download': 'm3u8'},
 183     }, {
 184         'url': 'https://www.winsports.co/videos/equidad-vuelve-defender-su-arco-de-remates-de-junior',
 185         'info_dict': {
 186             'id': '63fa7eca72f1741ad3a4d515',
 187             'display_id': 'equidad-vuelve-defender-su-arco-de-remates-de-junior',
 188             'title': '⚽ Equidad vuelve a defender su arco de remates de Junior',
 189             'description': 'Remate de Sierra',
 190             'thumbnail': r're:^https?://[^?#]+63fa7eca72f1741ad3a4d515',
 191             'ext': 'mp4',
 192         },
 193         'params': {'skip_download': 'm3u8'},
 194     }, {
 195         'url': 'https://www.winsports.co/videos/bucaramanga-se-quedo-con-el-grito-de-gol-en-la-garganta',
 196         'info_dict': {
 197             'id': '6402adb62bbf3b18d454e1b0',
 198             'display_id': 'bucaramanga-se-quedo-con-el-grito-de-gol-en-la-garganta',
 199             'title': '⚽Bucaramanga se quedó con el grito de gol en la garganta',
 200             'description': 'Gol anulado Bucaramanga',
 201             'thumbnail': r're:^https?://[^?#]+6402adb62bbf3b18d454e1b0',
 202             'ext': 'mp4',
 203         },
 204         'params': {'skip_download': 'm3u8'},
 205     }]
 206
 207     def _real_extract(self, url):
 208         display_id = self._match_id(url)
 209         webpage = self._download_webpage(url, display_id)
 210         data = self._search_json(
 211             r'<script\s*[^>]+data-drupal-selector="drupal-settings-json">', webpage, 'data', display_id)
 212
 213         mediastream_url = urljoin(f'{self._EMBED_BASE_URL}/', (
 214             traverse_obj(data, (
 215                 (('settings', 'mediastream_formatter', ..., 'mediastream_id'), 'url'), {str}), get_all=False)
 216             or next(self._extract_mediastream_urls(webpage), None)))
 217
 218         if not mediastream_url:
 219             self.raise_no_formats('No MediaStream embed found in webpage')
 220
 221         title = clean_html(remove_end(
 222             self._search_json_ld(webpage, display_id, expected_type='VideoObject', default={}).get('title')
 223             or self._og_search_title(webpage), '| Win Sports'))
 224
 225         return self.url_result(
 226             mediastream_url, MediaStreamIE, display_id, url_transparent=True, display_id=display_id, video_title=title)