yt_dlp/extractor/lnkgo.py

   1 from .common import InfoExtractor
   2 from ..compat import compat_str
   3 from ..utils import (
   4     clean_html,
   5     format_field,
   6     int_or_none,
   7     parse_iso8601,
   8     unified_strdate,
   9 )
  10
  11
  12 class LnkGoIE(InfoExtractor):
  13     _VALID_URL = r'https?://(?:www\.)?lnk(?:go)?\.(?:alfa\.)?lt/(?:visi-video/[^/]+|video)/(?P<id>[A-Za-z0-9-]+)(?:/(?P<episode_id>\d+))?'
  14     _TESTS = [{
  15         'url': 'http://www.lnkgo.lt/visi-video/aktualai-pratesimas/ziurek-putka-trys-klausimai',
  16         'info_dict': {
  17             'id': '10809',
  18             'ext': 'mp4',
  19             'title': "Put'ka: Trys Klausimai",
  20             'upload_date': '20161216',
  21             'description': 'Seniai matytas Put’ka užduoda tris klausimėlius. Pabandykime surasti atsakymus.',
  22             'age_limit': 18,
  23             'duration': 117,
  24             'thumbnail': r're:^https?://.*\.jpg$',
  25             'timestamp': 1481904000,
  26         },
  27         'params': {
  28             'skip_download': True,  # HLS download
  29         },
  30     }, {
  31         'url': 'http://lnkgo.alfa.lt/visi-video/aktualai-pratesimas/ziurek-nerdas-taiso-kompiuteri-2',
  32         'info_dict': {
  33             'id': '10467',
  34             'ext': 'mp4',
  35             'title': 'Nėrdas: Kompiuterio Valymas',
  36             'upload_date': '20150113',
  37             'description': 'md5:7352d113a242a808676ff17e69db6a69',
  38             'age_limit': 18,
  39             'duration': 346,
  40             'thumbnail': r're:^https?://.*\.jpg$',
  41             'timestamp': 1421164800,
  42         },
  43         'params': {
  44             'skip_download': True,  # HLS download
  45         },
  46     }, {
  47         'url': 'https://lnk.lt/video/neigalieji-tv-bokste/37413',
  48         'only_matching': True,
  49     }]
  50     _AGE_LIMITS = {
  51         'N-7': 7,
  52         'N-14': 14,
  53         'S': 18,
  54     }
  55     _M3U8_TEMPL = 'https://vod.lnk.lt/lnk_vod/lnk/lnk/%s:%s/playlist.m3u8%s'
  56
  57     def _real_extract(self, url):
  58         display_id, video_id = self._match_valid_url(url).groups()
  59
  60         video_info = self._download_json(
  61             'https://lnk.lt/api/main/video-page/%s/%s/false' % (display_id, video_id or '0'),
  62             display_id)['videoConfig']['videoInfo']
  63
  64         video_id = compat_str(video_info['id'])
  65         title = video_info['title']
  66         prefix = 'smil' if video_info.get('isQualityChangeAvailable') else 'mp4'
  67         formats = self._extract_m3u8_formats(
  68             self._M3U8_TEMPL % (prefix, video_info['videoUrl'], video_info.get('secureTokenParams') or ''),
  69             video_id, 'mp4', 'm3u8_native')
  70
  71         return {
  72             'id': video_id,
  73             'display_id': display_id,
  74             'title': title,
  75             'formats': formats,
  76             'thumbnail': format_field(video_info, 'posterImage', 'https://lnk.lt/all-images/%s'),
  77             'duration': int_or_none(video_info.get('duration')),
  78             'description': clean_html(video_info.get('htmlDescription')),
  79             'age_limit': self._AGE_LIMITS.get(video_info.get('pgRating'), 0),
  80             'timestamp': parse_iso8601(video_info.get('airDate')),
  81             'view_count': int_or_none(video_info.get('viewsCount')),
  82         }
  83
  84
  85 class LnkIE(InfoExtractor):
  86     _VALID_URL = r'https?://(?:www\.)?lnk\.lt/[^/]+/(?P<id>\d+)'
  87
  88     _TESTS = [{
  89         'url': 'https://lnk.lt/zinios/79791',
  90         'info_dict': {
  91             'id': '79791',
  92             'ext': 'mp4',
  93             'title': 'LNK.lt: Viešintų gyventojai sukilo prieš radijo bangų siųstuvą',
  94             'description': 'Svarbiausios naujienos trumpai, LNK žinios ir Info dienos pokalbiai.',
  95             'view_count': int,
  96             'duration': 233,
  97             'upload_date': '20191123',
  98             'thumbnail': r're:^https?://.*\.jpg$',
  99             'episode_number': 13431,
 100             'series': 'Naujausi žinių reportažai',
 101             'episode': 'Episode 13431'
 102         },
 103         'params': {'skip_download': True}
 104     }, {
 105         'url': 'https://lnk.lt/istorijos-trumpai/152546',
 106         'info_dict': {
 107             'id': '152546',
 108             'ext': 'mp4',
 109             'title': 'Radžio koncertas gaisre ',
 110             'description': 'md5:0666b5b85cb9fc7c1238dec96f71faba',
 111             'view_count': int,
 112             'duration': 54,
 113             'upload_date': '20220105',
 114             'thumbnail': r're:^https?://.*\.jpg$',
 115             'episode_number': 1036,
 116             'series': 'Istorijos trumpai',
 117             'episode': 'Episode 1036'
 118         },
 119         'params': {'skip_download': True}
 120     }, {
 121         'url': 'https://lnk.lt/gyvunu-pasaulis/151549',
 122         'info_dict': {
 123             'id': '151549',
 124             'ext': 'mp4',
 125             'title': 'Gyvūnų pasaulis',
 126             'description': '',
 127             'view_count': int,
 128             'duration': 1264,
 129             'upload_date': '20220108',
 130             'thumbnail': r're:^https?://.*\.jpg$',
 131             'episode_number': 16,
 132             'series': 'Gyvūnų pasaulis',
 133             'episode': 'Episode 16'
 134         },
 135         'params': {'skip_download': True}
 136     }]
 137
 138     def _real_extract(self, url):
 139         id = self._match_id(url)
 140         video_json = self._download_json(f'https://lnk.lt/api/video/video-config/{id}', id)['videoInfo']
 141         formats, subtitles = [], {}
 142         if video_json.get('videoUrl'):
 143             fmts, subs = self._extract_m3u8_formats_and_subtitles(video_json['videoUrl'], id)
 144             formats.extend(fmts)
 145             subtitles = self._merge_subtitles(subtitles, subs)
 146         if video_json.get('videoFairplayUrl') and not video_json.get('drm'):
 147             fmts, subs = self._extract_m3u8_formats_and_subtitles(video_json['videoFairplayUrl'], id)
 148             formats.extend(fmts)
 149             subtitles = self._merge_subtitles(subtitles, subs)
 150
 151         return {
 152             'id': id,
 153             'title': video_json.get('title'),
 154             'description': video_json.get('description'),
 155             'view_count': video_json.get('viewsCount'),
 156             'duration': video_json.get('duration'),
 157             'upload_date': unified_strdate(video_json.get('airDate')),
 158             'thumbnail': format_field(video_json, 'posterImage', 'https://lnk.lt/all-images/%s'),
 159             'episode_number': int_or_none(video_json.get('episodeNumber')),
 160             'series': video_json.get('programTitle'),
 161             'formats': formats,
 162             'subtitles': subtitles,
 163         }