yt_dlp/extractor/mitele.py

   1 from .telecinco import TelecincoBaseIE
   2 from ..utils import (
   3     int_or_none,
   4     parse_iso8601,
   5 )
   6
   7
   8 class MiTeleIE(TelecincoBaseIE):
   9     IE_DESC = 'mitele.es'
  10     _VALID_URL = r'https?://(?:www\.)?mitele\.es/(?:[^/]+/)+(?P<id>[^/]+)/player'
  11     _TESTS = [{
  12         'url': 'http://www.mitele.es/programas-tv/diario-de/57b0dfb9c715da65618b4afa/player',
  13         'info_dict': {
  14             'id': 'FhYW1iNTE6J6H7NkQRIEzfne6t2quqPg',
  15             'ext': 'mp4',
  16             'title': 'Diario de La redacción Programa 144',
  17             'description': 'md5:07c35a7b11abb05876a6a79185b58d27',
  18             'series': 'Diario de',
  19             'season': 'Season 14',
  20             'season_number': 14,
  21             'episode': 'Tor, la web invisible',
  22             'episode_number': 3,
  23             'thumbnail': r're:(?i)^https?://.*\.jpg$',
  24             'duration': 2913,
  25             'age_limit': 16,
  26             'timestamp': 1471209401,
  27             'upload_date': '20160814',
  28         },
  29         'skip': 'HTTP Error 404 Not Found',
  30     }, {
  31         # no explicit title
  32         'url': 'http://www.mitele.es/programas-tv/cuarto-milenio/57b0de3dc915da14058b4876/player',
  33         'info_dict': {
  34             'id': 'oyNG1iNTE6TAPP-JmCjbwfwJqqMMX3Vq',
  35             'ext': 'mp4',
  36             'title': 'Cuarto Milenio Temporada 6 Programa 226',
  37             'description': 'md5:5ff132013f0cd968ffbf1f5f3538a65f',
  38             'series': 'Cuarto Milenio',
  39             'season': 'Season 6',
  40             'season_number': 6,
  41             'episode': 'Episode 24',
  42             'episode_number': 24,
  43             'thumbnail': r're:(?i)^https?://.*\.jpg$',
  44             'duration': 7313,
  45             'age_limit': 12,
  46             'timestamp': 1471209021,
  47             'upload_date': '20160814',
  48         },
  49         'params': {
  50             'skip_download': True,
  51         },
  52         'skip': 'HTTP Error 404 Not Found',
  53     }, {
  54         'url': 'https://www.mitele.es/programas-tv/horizonte/temporada-5/programa-171-40_013480051/player/',
  55         'info_dict': {
  56             'id': '7adbe22e-cd41-4787-afa4-36f3da7c2c6f',
  57             'ext': 'mp4',
  58             'title': 'Horizonte Temporada 5 Programa 171',
  59             'description': 'md5:97f1fb712c5ac27e5693a8b3c5c0c6e3',
  60             'episode': 'Las Zonas de Bajas Emisiones, a debate',
  61             'episode_number': 171,
  62             'season': 'Season 5',
  63             'season_number': 5,
  64             'series': 'Horizonte',
  65             'duration': 7012,
  66             'upload_date': '20240927',
  67             'timestamp': 1727416450,
  68             'thumbnail': 'https://album.mediaset.es/eimg/2024/09/27/horizonte-171_9f02.jpg',
  69             'age_limit': 12,
  70         },
  71         'params': {'geo_bypass_country': 'ES'},
  72     }, {
  73         'url': 'http://www.mitele.es/series-online/la-que-se-avecina/57aac5c1c915da951a8b45ed/player',
  74         'only_matching': True,
  75     }, {
  76         'url': 'https://www.mitele.es/programas-tv/diario-de/la-redaccion/programa-144-40_1006364575251/player/',
  77         'only_matching': True,
  78     }]
  79
  80     def _real_extract(self, url):
  81         display_id = self._match_id(url)
  82         webpage = self._download_webpage(url, display_id)
  83         pre_player = self._search_json(
  84             r'window\.\$REACTBASE_STATE\.prePlayer_mtweb\s*=',
  85             webpage, 'Pre Player', display_id)['prePlayer']
  86         title = pre_player['title']
  87         video_info = self._parse_content(pre_player['video'], url)
  88         content = pre_player.get('content') or {}
  89         info = content.get('info') or {}
  90
  91         video_info.update({
  92             'title': title,
  93             'description': info.get('synopsis'),
  94             'series': content.get('title'),
  95             'season_number': int_or_none(info.get('season_number')),
  96             'episode': content.get('subtitle'),
  97             'episode_number': int_or_none(info.get('episode_number')),
  98             'duration': int_or_none(info.get('duration')),
  99             'age_limit': int_or_none(info.get('rating')),
 100             'timestamp': parse_iso8601(pre_player.get('publishedTime')),
 101         })
 102         return video_info