yt_dlp/extractor/wimtv.py

   1 from .common import InfoExtractor
   2 from ..utils import (
   3     ExtractorError,
   4     determine_ext,
   5     parse_duration,
   6     urlencode_postdata,
   7 )
   8
   9
  10 class WimTVIE(InfoExtractor):
  11     _player = None
  12     _UUID_RE = r'[\da-f]{8}-[\da-f]{4}-[\da-f]{4}-[\da-f]{4}-[\da-f]{12}'
  13     _VALID_URL = r'''(?x:
  14         https?://platform\.wim\.tv/
  15         (?:
  16             (?:embed/)?\?
  17             |\#/webtv/.+?/
  18         )
  19         (?P<type>vod|live|cast)[=/]
  20         (?P<id>%s).*?)''' % _UUID_RE
  21     _EMBED_REGEX = [rf'<iframe[^>]+src=["\'](?P<url>{_VALID_URL})']
  22     _TESTS = [{
  23         # vod stream
  24         'url': 'https://platform.wim.tv/embed/?vod=db29fb32-bade-47b6-a3a6-cb69fe80267a',
  25         'md5': 'db29fb32-bade-47b6-a3a6-cb69fe80267a',
  26         'info_dict': {
  27             'id': 'db29fb32-bade-47b6-a3a6-cb69fe80267a',
  28             'ext': 'mp4',
  29             'title': 'AMA SUPERCROSS 2020 - R2 ST. LOUIS',
  30             'duration': 6481,
  31             'thumbnail': r're:https?://.+?/thumbnail/.+?/720$'
  32         },
  33         'params': {
  34             'skip_download': True,
  35         },
  36     }, {
  37         # live stream
  38         'url': 'https://platform.wim.tv/embed/?live=28e22c22-49db-40f3-8c37-8cbb0ff44556&autostart=true',
  39         'info_dict': {
  40             'id': '28e22c22-49db-40f3-8c37-8cbb0ff44556',
  41             'ext': 'mp4',
  42             'title': 'Streaming MSmotorTV',
  43             'is_live': True,
  44         },
  45         'params': {
  46             'skip_download': True,
  47         },
  48     }, {
  49         'url': 'https://platform.wim.tv/#/webtv/automotornews/vod/422492b6-539e-474d-9c6b-68c9d5893365',
  50         'only_matching': True,
  51     }, {
  52         'url': 'https://platform.wim.tv/#/webtv/renzoarborechannel/cast/f47e0d15-5b45-455e-bf0d-dba8ffa96365',
  53         'only_matching': True,
  54     }]
  55
  56     def _real_initialize(self):
  57         if not self._player:
  58             self._get_player_data()
  59
  60     def _get_player_data(self):
  61         msg_id = 'Player data'
  62         self._player = {}
  63
  64         datas = [{
  65             'url': 'https://platform.wim.tv/common/libs/player/wimtv/wim-rest.js',
  66             'vars': [{
  67                 'regex': r'appAuth = "(.+?)"',
  68                 'variable': 'app_auth',
  69             }]
  70         }, {
  71             'url': 'https://platform.wim.tv/common/config/endpointconfig.js',
  72             'vars': [{
  73                 'regex': r'PRODUCTION_HOSTNAME_THUMB = "(.+?)"',
  74                 'variable': 'thumb_server',
  75             }, {
  76                 'regex': r'PRODUCTION_HOSTNAME_THUMB\s*\+\s*"(.+?)"',
  77                 'variable': 'thumb_server_path',
  78             }]
  79         }]
  80
  81         for data in datas:
  82             temp = self._download_webpage(data['url'], msg_id)
  83             for var in data['vars']:
  84                 val = self._search_regex(var['regex'], temp, msg_id)
  85                 if not val:
  86                     raise ExtractorError('%s not found' % var['variable'])
  87                 self._player[var['variable']] = val
  88
  89     def _generate_token(self):
  90         json = self._download_json(
  91             'https://platform.wim.tv/wimtv-server/oauth/token', 'Token generation',
  92             headers={'Authorization': 'Basic %s' % self._player['app_auth']},
  93             data=urlencode_postdata({'grant_type': 'client_credentials'}))
  94         token = json.get('access_token')
  95         if not token:
  96             raise ExtractorError('access token not generated')
  97         return token
  98
  99     def _generate_thumbnail(self, thumb_id, width='720'):
 100         if not thumb_id or not self._player.get('thumb_server'):
 101             return None
 102         if not self._player.get('thumb_server_path'):
 103             self._player['thumb_server_path'] = ''
 104         return '%s%s/asset/thumbnail/%s/%s' % (
 105             self._player['thumb_server'],
 106             self._player['thumb_server_path'],
 107             thumb_id, width)
 108
 109     def _real_extract(self, url):
 110         urlc = self._match_valid_url(url).groupdict()
 111         video_id = urlc['id']
 112         stream_type = is_live = None
 113         if urlc['type'] in {'live', 'cast'}:
 114             stream_type = urlc['type'] + '/channel'
 115             is_live = True
 116         else:
 117             stream_type = 'vod'
 118             is_live = False
 119         token = self._generate_token()
 120         json = self._download_json(
 121             'https://platform.wim.tv/wimtv-server/api/public/%s/%s/play' % (
 122                 stream_type, video_id), video_id,
 123             headers={'Authorization': 'Bearer %s' % token,
 124                      'Content-Type': 'application/json'},
 125             data=bytes('{}', 'utf-8'))
 126
 127         formats = []
 128         for src in json.get('srcs') or []:
 129             if src.get('mimeType') == 'application/x-mpegurl':
 130                 formats.extend(
 131                     self._extract_m3u8_formats(
 132                         src.get('uniqueStreamer'), video_id, 'mp4'))
 133             if src.get('mimeType') == 'video/flash':
 134                 formats.append({
 135                     'format_id': 'rtmp',
 136                     'url': src.get('uniqueStreamer'),
 137                     'ext': determine_ext(src.get('uniqueStreamer'), 'flv'),
 138                     'rtmp_live': is_live,
 139                 })
 140         json = json.get('resource')
 141         thumb = self._generate_thumbnail(json.get('thumbnailId'))
 142
 143         return {
 144             'id': video_id,
 145             'title': json.get('title') or json.get('name'),
 146             'duration': parse_duration(json.get('duration')),
 147             'formats': formats,
 148             'thumbnail': thumb,
 149             'is_live': is_live,
 150         }