yt_dlp/extractor/tvplay.py

   1 import re
   2
   3 from .common import InfoExtractor
   4 from ..compat import (
   5     compat_HTTPError,
   6     compat_urlparse,
   7 )
   8 from ..utils import (
   9     determine_ext,
  10     ExtractorError,
  11     int_or_none,
  12     parse_iso8601,
  13     qualities,
  14     traverse_obj,
  15     try_get,
  16     update_url_query,
  17     url_or_none,
  18     urljoin,
  19 )
  20
  21
  22 class TVPlayIE(InfoExtractor):
  23     IE_NAME = 'mtg'
  24     IE_DESC = 'MTG services'
  25     _VALID_URL = r'''(?x)
  26                     (?:
  27                         mtg:|
  28                         https?://
  29                             (?:www\.)?
  30                             (?:
  31                                 tvplay(?:\.skaties)?\.lv(?:/parraides)?|
  32                                 (?:tv3play|play\.tv3)\.lt(?:/programos)?|
  33                                 tv3play(?:\.tv3)?\.ee/sisu
  34                             )
  35                             /(?:[^/]+/)+
  36                         )
  37                         (?P<id>\d+)
  38                     '''
  39     _TESTS = [
  40         {
  41             'url': 'http://www.tvplay.lv/parraides/vinas-melo-labak/418113?autostart=true',
  42             'md5': 'a1612fe0849455423ad8718fe049be21',
  43             'info_dict': {
  44                 'id': '418113',
  45                 'ext': 'mp4',
  46                 'title': 'Kādi ir īri? - Viņas melo labāk',
  47                 'description': 'Baiba apsmej īrus, kādi tie ir un ko viņi dara.',
  48                 'series': 'Viņas melo labāk',
  49                 'season': '2.sezona',
  50                 'season_number': 2,
  51                 'duration': 25,
  52                 'timestamp': 1406097056,
  53                 'upload_date': '20140723',
  54             },
  55         },
  56         {
  57             'url': 'http://play.tv3.lt/programos/moterys-meluoja-geriau/409229?autostart=true',
  58             'info_dict': {
  59                 'id': '409229',
  60                 'ext': 'flv',
  61                 'title': 'Moterys meluoja geriau',
  62                 'description': 'md5:9aec0fc68e2cbc992d2a140bd41fa89e',
  63                 'series': 'Moterys meluoja geriau',
  64                 'episode_number': 47,
  65                 'season': '1 sezonas',
  66                 'season_number': 1,
  67                 'duration': 1330,
  68                 'timestamp': 1403769181,
  69                 'upload_date': '20140626',
  70             },
  71             'params': {
  72                 # rtmp download
  73                 'skip_download': True,
  74             },
  75         },
  76         {
  77             'url': 'http://www.tv3play.ee/sisu/kodu-keset-linna/238551?autostart=true',
  78             'info_dict': {
  79                 'id': '238551',
  80                 'ext': 'flv',
  81                 'title': 'Kodu keset linna 398537',
  82                 'description': 'md5:7df175e3c94db9e47c0d81ffa5d68701',
  83                 'duration': 1257,
  84                 'timestamp': 1292449761,
  85                 'upload_date': '20101215',
  86             },
  87             'params': {
  88                 # rtmp download
  89                 'skip_download': True,
  90             },
  91         },
  92         {
  93             'url': 'http://tvplay.skaties.lv/parraides/vinas-melo-labak/418113?autostart=true',
  94             'only_matching': True,
  95         },
  96         {
  97             'url': 'https://tvplay.skaties.lv/vinas-melo-labak/418113/?autostart=true',
  98             'only_matching': True,
  99         },
 100         {
 101             # views is null
 102             'url': 'http://tvplay.skaties.lv/parraides/tv3-zinas/760183',
 103             'only_matching': True,
 104         },
 105         {
 106             'url': 'http://tv3play.tv3.ee/sisu/kodu-keset-linna/238551?autostart=true',
 107             'only_matching': True,
 108         },
 109         {
 110             'url': 'mtg:418113',
 111             'only_matching': True,
 112         }
 113     ]
 114
 115     def _real_extract(self, url):
 116         video_id = self._match_id(url)
 117         geo_country = self._search_regex(
 118             r'https?://[^/]+\.([a-z]{2})', url,
 119             'geo country', default=None)
 120         if geo_country:
 121             self._initialize_geo_bypass({'countries': [geo_country.upper()]})
 122         video = self._download_json(
 123             'http://playapi.mtgx.tv/v3/videos/%s' % video_id, video_id, 'Downloading video JSON')
 124
 125         title = video['title']
 126
 127         try:
 128             streams = self._download_json(
 129                 'http://playapi.mtgx.tv/v3/videos/stream/%s' % video_id,
 130                 video_id, 'Downloading streams JSON')
 131         except ExtractorError as e:
 132             if isinstance(e.cause, compat_HTTPError) and e.cause.code == 403:
 133                 msg = self._parse_json(e.cause.read().decode('utf-8'), video_id)
 134                 raise ExtractorError(msg['msg'], expected=True)
 135             raise
 136
 137         quality = qualities(['hls', 'medium', 'high'])
 138         formats = []
 139         for format_id, video_url in streams.get('streams', {}).items():
 140             video_url = url_or_none(video_url)
 141             if not video_url:
 142                 continue
 143             ext = determine_ext(video_url)
 144             if ext == 'f4m':
 145                 formats.extend(self._extract_f4m_formats(
 146                     update_url_query(video_url, {
 147                         'hdcore': '3.5.0',
 148                         'plugin': 'aasp-3.5.0.151.81'
 149                     }), video_id, f4m_id='hds', fatal=False))
 150             elif ext == 'm3u8':
 151                 formats.extend(self._extract_m3u8_formats(
 152                     video_url, video_id, 'mp4', 'm3u8_native',
 153                     m3u8_id='hls', fatal=False))
 154             else:
 155                 fmt = {
 156                     'format_id': format_id,
 157                     'quality': quality(format_id),
 158                     'ext': ext,
 159                 }
 160                 if video_url.startswith('rtmp'):
 161                     m = re.search(
 162                         r'^(?P<url>rtmp://[^/]+/(?P<app>[^/]+))/(?P<playpath>.+)$', video_url)
 163                     if not m:
 164                         continue
 165                     fmt.update({
 166                         'ext': 'flv',
 167                         'url': m.group('url'),
 168                         'app': m.group('app'),
 169                         'play_path': m.group('playpath'),
 170                         'preference': -1,
 171                     })
 172                 else:
 173                     fmt.update({
 174                         'url': video_url,
 175                     })
 176                 formats.append(fmt)
 177
 178         if not formats and video.get('is_geo_blocked'):
 179             self.raise_geo_restricted(
 180                 'This content might not be available in your country due to copyright reasons',
 181                 metadata_available=True)
 182
 183         # TODO: webvtt in m3u8
 184         subtitles = {}
 185         sami_path = video.get('sami_path')
 186         if sami_path:
 187             lang = self._search_regex(
 188                 r'_([a-z]{2})\.xml', sami_path, 'lang',
 189                 default=compat_urlparse.urlparse(url).netloc.rsplit('.', 1)[-1])
 190             subtitles[lang] = [{
 191                 'url': sami_path,
 192             }]
 193
 194         series = video.get('format_title')
 195         episode_number = int_or_none(video.get('format_position', {}).get('episode'))
 196         season = video.get('_embedded', {}).get('season', {}).get('title')
 197         season_number = int_or_none(video.get('format_position', {}).get('season'))
 198
 199         return {
 200             'id': video_id,
 201             'title': title,
 202             'description': video.get('description'),
 203             'series': series,
 204             'episode_number': episode_number,
 205             'season': season,
 206             'season_number': season_number,
 207             'duration': int_or_none(video.get('duration')),
 208             'timestamp': parse_iso8601(video.get('created_at')),
 209             'view_count': try_get(video, lambda x: x['views']['total'], int),
 210             'age_limit': int_or_none(video.get('age_limit', 0)),
 211             'formats': formats,
 212             'subtitles': subtitles,
 213         }
 214
 215
 216 class TVPlayHomeIE(InfoExtractor):
 217     _VALID_URL = r'''(?x)
 218             https?://
 219             (?:tv3?)?
 220             play\.(?:tv3|skaties)\.(?P<country>lv|lt|ee)/
 221             (?P<live>lives/)?
 222             [^?#&]+(?:episode|programme|clip)-(?P<id>\d+)
 223     '''
 224     _TESTS = [{
 225         'url': 'https://play.tv3.lt/series/gauju-karai-karveliai,serial-2343791/serija-8,episode-2343828',
 226         'info_dict': {
 227             'id': '2343828',
 228             'ext': 'mp4',
 229             'title': 'Gaujų karai. Karveliai (2021) | S01E08: Serija 8',
 230             'description': 'md5:f6fcfbb236429f05531131640dfa7c81',
 231             'duration': 2710,
 232             'season': 'Gaujų karai. Karveliai',
 233             'season_number': 1,
 234             'release_year': 2021,
 235             'episode': 'Serija 8',
 236             'episode_number': 8,
 237         },
 238         'params': {
 239             'skip_download': 'm3u8',
 240         },
 241     }, {
 242         'url': 'https://play.tv3.lt/series/moterys-meluoja-geriau-n-7,serial-2574652/serija-25,episode-3284937',
 243         'info_dict': {
 244             'id': '3284937',
 245             'ext': 'mp4',
 246             'season': 'Moterys meluoja geriau [N-7]',
 247             'season_number': 14,
 248             'release_year': 2021,
 249             'episode': 'Serija 25',
 250             'episode_number': 25,
 251             'title': 'Moterys meluoja geriau [N-7] (2021) | S14|E25: Serija 25',
 252             'description': 'md5:c6926e9710f1a126f028fbe121eddb79',
 253             'duration': 2440,
 254         },
 255         'skip': '404'
 256     }, {
 257         'url': 'https://play.tv3.lt/lives/tv6-lt,live-2838694/optibet-a-lygos-rungtynes-marijampoles-suduva--vilniaus-riteriai,programme-3422014',
 258         'only_matching': True,
 259     }, {
 260         'url': 'https://tv3play.skaties.lv/series/women-lie-better-lv,serial-1024464/women-lie-better-lv,episode-1038762',
 261         'only_matching': True,
 262     }, {
 263         'url': 'https://play.tv3.ee/series/_,serial-2654462/_,episode-2654474',
 264         'only_matching': True,
 265     }, {
 266         'url': 'https://tv3play.skaties.lv/clips/tv3-zinas-valsti-lidz-15novembrim-bus-majsede,clip-3464509',
 267         'only_matching': True,
 268     }]
 269
 270     def _real_extract(self, url):
 271         country, is_live, video_id = self._match_valid_url(url).groups()
 272
 273         api_path = 'lives/programmes' if is_live else 'vods'
 274         data = self._download_json(
 275             urljoin(url, f'/api/products/{api_path}/{video_id}?platform=BROWSER&lang={country.upper()}'),
 276             video_id)
 277
 278         video_type = 'CATCHUP' if is_live else 'MOVIE'
 279         stream_id = data['programRecordingId'] if is_live else video_id
 280         stream = self._download_json(
 281             urljoin(url, f'/api/products/{stream_id}/videos/playlist?videoType={video_type}&platform=BROWSER'), video_id)
 282         formats, subtitles = self._extract_m3u8_formats_and_subtitles(
 283             stream['sources']['HLS'][0]['src'], video_id, 'mp4', 'm3u8_native', m3u8_id='hls')
 284
 285         thumbnails = set(traverse_obj(
 286             data, (('galary', 'images', 'artworks'), ..., ..., ('miniUrl', 'mainUrl')), expected_type=url_or_none))
 287
 288         return {
 289             'id': video_id,
 290             'title': self._resolve_title(data),
 291             'description': traverse_obj(data, 'description', 'lead'),
 292             'duration': int_or_none(data.get('duration')),
 293             'season': traverse_obj(data, ('season', 'serial', 'title')),
 294             'season_number': int_or_none(traverse_obj(data, ('season', 'number'))),
 295             'episode': data.get('title'),
 296             'episode_number': int_or_none(data.get('episode')),
 297             'release_year': int_or_none(traverse_obj(data, ('season', 'serial', 'year'))),
 298             'thumbnails': [{'url': url, 'ext': 'jpg'} for url in thumbnails],
 299             'formats': formats,
 300             'subtitles': subtitles,
 301         }
 302
 303     @staticmethod
 304     def _resolve_title(data):
 305         return try_get(data, lambda x: (
 306             f'{data["season"]["serial"]["title"]} ({data["season"]["serial"]["year"]}) | '
 307             f'S{data["season"]["number"]:02d}E{data["episode"]:02d}: {data["title"]}'
 308         )) or data.get('title')