yt_dlp/extractor/eltrecetv.py

   1 from .common import InfoExtractor
   2
   3
   4 class ElTreceTVIE(InfoExtractor):
   5     IE_DESC = 'El Trece TV (Argentina)'
   6     _VALID_URL = r'https?://(?:www\.)?eltrecetv\.com\.ar/[\w-]+/capitulos/temporada-\d+/(?P<id>[\w-]+)'
   7     _TESTS = [
   8         {
   9             'url': 'https://www.eltrecetv.com.ar/ahora-caigo/capitulos/temporada-2023/programa-del-061023/',
  10             'md5': '71a66673dc63f9a5939d97bfe4b311ba',
  11             'info_dict': {
  12                 'id': 'AHCA05102023145553329621094',
  13                 'ext': 'mp4',
  14                 'title': 'AHORA CAIGO - Programa 06/10/23',
  15                 'thumbnail': 'https://thumbs.vodgc.net/AHCA05102023145553329621094.JPG?649339',
  16             },
  17         },
  18         {
  19             'url': 'https://www.eltrecetv.com.ar/poco-correctos/capitulos/temporada-2023/programa-del-250923-invitada-dalia-gutmann/',
  20             'only_matching': True,
  21         },
  22         {
  23             'url': 'https://www.eltrecetv.com.ar/argentina-tierra-de-amor-y-venganza/capitulos/temporada-2023/atav-2-capitulo-121-del-250923/',
  24             'only_matching': True,
  25         },
  26         {
  27             'url': 'https://www.eltrecetv.com.ar/ahora-caigo/capitulos/temporada-2023/programa-del-250923/',
  28             'only_matching': True,
  29         },
  30         {
  31             'url': 'https://www.eltrecetv.com.ar/pasaplatos/capitulos/temporada-2023/pasaplatos-el-restaurante-del-250923/',
  32             'only_matching': True,
  33         },
  34         {
  35             'url': 'https://www.eltrecetv.com.ar/el-galpon/capitulos/temporada-2023/programa-del-160923-invitado-raul-lavie/',
  36             'only_matching': True,
  37         },
  38     ]
  39
  40     def _real_extract(self, url):
  41         slug = self._match_id(url)
  42         webpage = self._download_webpage(url, slug)
  43         config = self._search_json(
  44             r'Fusion.globalContent\s*=', webpage, 'content', slug)['promo_items']['basic']['embed']['config']
  45         video_url = config['m3u8']
  46         video_id = self._search_regex(r'/(\w+)\.m3u8', video_url, 'video id', default=slug)
  47
  48         formats, subtitles = self._extract_m3u8_formats_and_subtitles(video_url, video_id, 'mp4', m3u8_id='hls')
  49         formats.extend([{
  50             'url': f['url'][:-23],
  51             'format_id': f['format_id'].replace('hls', 'http'),
  52             'width': f.get('width'),
  53             'height': f.get('height'),
  54         } for f in formats if f['url'].endswith('/tracks-v1a1/index.m3u8') and f.get('height') != 1080])
  55
  56         return {
  57             'id': video_id,
  58             'title': config.get('title'),
  59             'thumbnail': config.get('thumbnail'),
  60             'formats': formats,
  61             'subtitles': subtitles,
  62         }