yt_dlp/extractor/trtcocuk.py

   1 from .common import InfoExtractor
   2 from ..utils import ExtractorError, int_or_none, parse_iso8601, traverse_obj
   3
   4
   5 class TrtCocukVideoIE(InfoExtractor):
   6     _VALID_URL = r'https?://www\.trtcocuk\.net\.tr/video/(?P<id>[\w-]+)'
   7     _TESTS = [{
   8         'url': 'https://www.trtcocuk.net.tr/video/kaptan-pengu-ve-arkadaslari-1',
   9         'info_dict': {
  10             'id': '3789738',
  11             'ext': 'mp4',
  12             'season_number': 1,
  13             'series': '"Kaptan Pengu ve Arkadaşları"',
  14             'season': 'Season 1',
  15             'title': 'Kaptan Pengu ve Arkadaşları 1 Bölüm İzle TRT Çocuk',
  16             'release_date': '20201209',
  17             'release_timestamp': 1607513774,
  18         },
  19     }, {
  20         'url': 'https://www.trtcocuk.net.tr/video/sef-rokanin-lezzet-dunyasi-17',
  21         'info_dict': {
  22             'id': '10260842',
  23             'ext': 'mp4',
  24             'series': '"Şef Roka\'nın Lezzet Dünyası"',
  25             'title': 'Şef Roka\'nın Lezzet Dünyası 17 Bölüm İzle TRT Çocuk',
  26         },
  27     }]
  28
  29     def _real_extract(self, url):
  30         display_id = self._match_id(url)
  31         webpage = self._download_webpage(url, display_id)
  32         nuxtjs_data = self._search_nuxt_data(webpage, display_id)['data']
  33
  34         try:
  35             video_url = self._parse_json(nuxtjs_data['video'], display_id)
  36         except ExtractorError:
  37             video_url = nuxtjs_data['video']
  38         formats, subtitles = self._extract_m3u8_formats_and_subtitles(video_url, display_id)
  39
  40         return {
  41             'id': str(nuxtjs_data['id']),
  42             'formats': formats,
  43             'subtitles': subtitles,
  44             'season_number': int_or_none(nuxtjs_data.get('season')),
  45             'release_timestamp': parse_iso8601(nuxtjs_data.get('publishedDate')),
  46             'series': traverse_obj(nuxtjs_data, ('show', 0, 'title')),
  47             'title': self._html_extract_title(webpage),  # TODO: get better title
  48         }