yt_dlp/extractor/lnk.py

   1 from .common import InfoExtractor
   2 from ..utils import (
   3     format_field,
   4     int_or_none,
   5     unified_strdate,
   6 )
   7
   8
   9 class LnkIE(InfoExtractor):
  10     _VALID_URL = r'https?://(?:www\.)?lnk\.lt/[^/]+/(?P<id>\d+)'
  11
  12     _TESTS = [{
  13         'url': 'https://lnk.lt/zinios/79791',
  14         'info_dict': {
  15             'id': '79791',
  16             'ext': 'mp4',
  17             'title': 'LNK.lt: Viešintų gyventojai sukilo prieš radijo bangų siųstuvą',
  18             'description': 'Svarbiausios naujienos trumpai, LNK žinios ir Info dienos pokalbiai.',
  19             'view_count': int,
  20             'duration': 233,
  21             'upload_date': '20191123',
  22             'thumbnail': r're:^https?://.*\.jpg$',
  23             'episode_number': 13431,
  24             'series': 'Naujausi žinių reportažai',
  25             'episode': 'Episode 13431',
  26         },
  27         'params': {'skip_download': True},
  28     }, {
  29         'url': 'https://lnk.lt/istorijos-trumpai/152546',
  30         'info_dict': {
  31             'id': '152546',
  32             'ext': 'mp4',
  33             'title': 'Radžio koncertas gaisre ',
  34             'description': 'md5:0666b5b85cb9fc7c1238dec96f71faba',
  35             'view_count': int,
  36             'duration': 54,
  37             'upload_date': '20220105',
  38             'thumbnail': r're:^https?://.*\.jpg$',
  39             'episode_number': 1036,
  40             'series': 'Istorijos trumpai',
  41             'episode': 'Episode 1036',
  42         },
  43         'params': {'skip_download': True},
  44     }, {
  45         'url': 'https://lnk.lt/gyvunu-pasaulis/151549',
  46         'info_dict': {
  47             'id': '151549',
  48             'ext': 'mp4',
  49             'title': 'Gyvūnų pasaulis',
  50             'description': '',
  51             'view_count': int,
  52             'duration': 1264,
  53             'upload_date': '20220108',
  54             'thumbnail': r're:^https?://.*\.jpg$',
  55             'episode_number': 16,
  56             'series': 'Gyvūnų pasaulis',
  57             'episode': 'Episode 16',
  58         },
  59         'params': {'skip_download': True},
  60     }]
  61
  62     def _real_extract(self, url):
  63         video_id = self._match_id(url)
  64         video_json = self._download_json(f'https://lnk.lt/api/video/video-config/{video_id}', video_id)['videoInfo']
  65         formats, subtitles = [], {}
  66         if video_json.get('videoUrl'):
  67             fmts, subs = self._extract_m3u8_formats_and_subtitles(video_json['videoUrl'], video_id)
  68             formats.extend(fmts)
  69             subtitles = self._merge_subtitles(subtitles, subs)
  70         if video_json.get('videoFairplayUrl') and not video_json.get('drm'):
  71             fmts, subs = self._extract_m3u8_formats_and_subtitles(video_json['videoFairplayUrl'], video_id)
  72             formats.extend(fmts)
  73             subtitles = self._merge_subtitles(subtitles, subs)
  74
  75         return {
  76             'id': video_id,
  77             'title': video_json.get('title'),
  78             'description': video_json.get('description'),
  79             'view_count': video_json.get('viewsCount'),
  80             'duration': video_json.get('duration'),
  81             'upload_date': unified_strdate(video_json.get('airDate')),
  82             'thumbnail': format_field(video_json, 'posterImage', 'https://lnk.lt/all-images/%s'),
  83             'episode_number': int_or_none(video_json.get('episodeNumber')),
  84             'series': video_json.get('programTitle'),
  85             'formats': formats,
  86             'subtitles': subtitles,
  87         }