yt_dlp/extractor/mediastream.py

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