yt_dlp/extractor/tele13.py

   1 from .common import InfoExtractor
   2 from .youtube import YoutubeIE
   3 from ..utils import (
   4     determine_ext,
   5     js_to_json,
   6     qualities,
   7 )
   8
   9
  10 class Tele13IE(InfoExtractor):
  11     _VALID_URL = r'https?://(?:www\.)?t13\.cl/videos(?:/[^/]+)+/(?P<id>[\w-]+)'
  12     _TESTS = [
  13         {
  14             'url': 'http://www.t13.cl/videos/actualidad/el-circulo-de-hierro-de-michelle-bachelet-en-su-regreso-a-la-moneda',
  15             'md5': '4cb1fa38adcad8fea88487a078831755',
  16             'info_dict': {
  17                 'id': 'el-circulo-de-hierro-de-michelle-bachelet-en-su-regreso-a-la-moneda',
  18                 'ext': 'mp4',
  19                 'title': 'El círculo de hierro de Michelle Bachelet en su regreso a La Moneda',
  20             },
  21             'params': {
  22                 # HTTP Error 404: Not Found
  23                 'skip_download': True,
  24             },
  25         },
  26         {
  27             'url': 'http://www.t13.cl/videos/mundo/tendencias/video-captan-misteriosa-bola-fuego-cielos-bangkok',
  28             'md5': '867adf6a3b3fef932c68a71d70b70946',
  29             'info_dict': {
  30                 'id': 'rOoKv2OMpOw',
  31                 'ext': 'mp4',
  32                 'title': 'Shooting star seen on 7-Sep-2015',
  33                 'description': 'md5:7292ff2a34b2f673da77da222ae77e1e',
  34                 'uploader': 'Porjai Jaturongkhakun',
  35                 'upload_date': '20150906',
  36                 'uploader_id': 'UCnLY_3ezwNcDSC_Wc6suZxw',
  37             },
  38             'add_ie': ['Youtube'],
  39         },
  40     ]
  41
  42     def _real_extract(self, url):
  43         display_id = self._match_id(url)
  44         webpage = self._download_webpage(url, display_id)
  45
  46         setup_js = self._search_regex(
  47             r"(?s)jwplayer\('player-vivo'\).setup\((\{.*?\})\)",
  48             webpage, 'setup code')
  49         sources = self._parse_json(self._search_regex(
  50             r'sources\s*:\s*(\[[^\]]+\])', setup_js, 'sources'),
  51             display_id, js_to_json)
  52
  53         preference = qualities(['Móvil', 'SD', 'HD'])
  54         formats = []
  55         urls = []
  56         for f in sources:
  57             format_url = f['file']
  58             if format_url and format_url not in urls:
  59                 ext = determine_ext(format_url)
  60                 if ext == 'm3u8':
  61                     formats.extend(self._extract_m3u8_formats(
  62                         format_url, display_id, 'mp4', 'm3u8_native',
  63                         m3u8_id='hls', fatal=False))
  64                 elif YoutubeIE.suitable(format_url):
  65                     return self.url_result(format_url, 'Youtube')
  66                 else:
  67                     formats.append({
  68                         'url': format_url,
  69                         'format_id': f.get('label'),
  70                         'quality': preference(f.get('label')),
  71                         'ext': ext,
  72                     })
  73                 urls.append(format_url)
  74
  75         return {
  76             'id': display_id,
  77             'title': self._search_regex(
  78                 r'title\s*:\s*"([^"]+)"', setup_js, 'title'),
  79             'description': self._html_search_meta(
  80                 'description', webpage, 'description'),
  81             'thumbnail': self._search_regex(
  82                 r'image\s*:\s*"([^"]+)"', setup_js, 'thumbnail', default=None),
  83             'formats': formats,
  84         }