yt_dlp/extractor/mediaklikk.py

   1 import urllib.parse
   2
   3 from .common import InfoExtractor
   4 from ..utils import (
   5     ExtractorError,
   6     traverse_obj,
   7     unified_strdate,
   8     url_or_none,
   9 )
  10
  11
  12 class MediaKlikkIE(InfoExtractor):
  13     _VALID_URL = r'''(?x)https?://(?:www\.)?
  14                         (?:mediaklikk|m4sport|hirado|petofilive)\.hu/.*?(?:videok?|cikk)/
  15                         (?:(?P<year>[0-9]{4})/(?P<month>[0-9]{1,2})/(?P<day>[0-9]{1,2})/)?
  16                         (?P<id>[^/#?_]+)'''
  17
  18     _TESTS = [{
  19         'url': 'https://mediaklikk.hu/filmajanlo/cikk/az-ajto/',
  20         'info_dict': {
  21             'id': '668177',
  22             'title': 'Az ajtó',
  23             'display_id': 'az-ajto',
  24             'ext': 'mp4',
  25             'thumbnail': 'https://cdn.cms.mtv.hu/wp-content/uploads/sites/4/2016/01/vlcsnap-2023-07-31-14h18m52s111.jpg',
  26         },
  27     }, {
  28         # (old) mediaklikk. date in html.
  29         'url': 'https://mediaklikk.hu/video/hazajaro-delnyugat-bacska-a-duna-menten-palankatol-doroszloig/',
  30         'info_dict': {
  31             'id': '4754129',
  32             'title': 'Hazajáró, DÉLNYUGAT-BÁCSKA – A Duna mentén Palánkától Doroszlóig',
  33             'ext': 'mp4',
  34             'upload_date': '20210901',
  35             'thumbnail': 'http://mediaklikk.hu/wp-content/uploads/sites/4/2014/02/hazajarouj_JO.jpg',
  36         },
  37         'skip': 'Webpage redirects to 404 page',
  38     }, {
  39         # mediaklikk. date in html.
  40         'url': 'https://mediaklikk.hu/video/hazajaro-fabova-hegyseg-kishont-koronaja/',
  41         'info_dict': {
  42             'id': '6696133',
  43             'title': 'Hazajáró, Fabova-hegység - Kishont koronája',
  44             'display_id': 'hazajaro-fabova-hegyseg-kishont-koronaja',
  45             'ext': 'mp4',
  46             'upload_date': '20230903',
  47             'thumbnail': 'https://mediaklikk.hu/wp-content/uploads/sites/4/2014/02/hazajarouj_JO.jpg',
  48         },
  49         'skip': 'Webpage redirects to 404 page',
  50     }, {
  51         # (old) m4sport
  52         'url': 'https://m4sport.hu/video/2021/08/30/gyemant-liga-parizs/',
  53         'info_dict': {
  54             'id': '4754999',
  55             'title': 'Gyémánt Liga, Párizs',
  56             'ext': 'mp4',
  57             'upload_date': '20210830',
  58             'thumbnail': 'http://m4sport.hu/wp-content/uploads/sites/4/2021/08/vlcsnap-2021-08-30-18h21m20s10-1024x576.jpg',
  59         },
  60         'skip': 'Webpage redirects to 404 page',
  61     }, {
  62         # m4sport
  63         'url': 'https://m4sport.hu/sportkozvetitesek/video/2023/09/08/atletika-gyemant-liga-brusszel/',
  64         'info_dict': {
  65             'id': '6711136',
  66             'title': 'Atlétika – Gyémánt Liga, Brüsszel',
  67             'display_id': 'atletika-gyemant-liga-brusszel',
  68             'ext': 'mp4',
  69             'upload_date': '20230908',
  70             'thumbnail': 'https://m4sport.hu/wp-content/uploads/sites/4/2023/09/vlcsnap-2023-09-08-22h43m18s691.jpg',
  71         },
  72         'skip': 'Webpage redirects to 404 page',
  73     }, {
  74         # m4sport with *video/ url and no date
  75         'url': 'https://m4sport.hu/bl-video/real-madrid-chelsea-1-1/',
  76         'info_dict': {
  77             'id': '4492099',
  78             'title': 'Real Madrid - Chelsea 1-1',
  79             'display_id': 'real-madrid-chelsea-1-1',
  80             'ext': 'mp4',
  81             'thumbnail': 'https://m4sport.hu/wp-content/uploads/sites/4/2021/04/Sequence-01.Still001-1024x576.png',
  82         },
  83         'skip': 'Webpage redirects to 404 page',
  84     }, {
  85         # (old) hirado
  86         'url': 'https://hirado.hu/videok/felteteleket-szabott-a-fovaros/',
  87         'info_dict': {
  88             'id': '4760120',
  89             'title': 'Feltételeket szabott a főváros',
  90             'ext': 'mp4',
  91             'thumbnail': 'http://hirado.hu/wp-content/uploads/sites/4/2021/09/vlcsnap-2021-09-01-20h20m37s165.jpg',
  92         },
  93         'skip': 'Webpage redirects to video list page',
  94     }, {
  95         # hirado
  96         'url': 'https://hirado.hu/belfold/video/2023/09/11/marad-az-eves-elszamolas-a-napelemekre-beruhazo-csaladoknal',
  97         'info_dict': {
  98             'id': '6716068',
  99             'title': 'Marad az éves elszámolás a napelemekre beruházó családoknál',
 100             'display_id': 'marad-az-eves-elszamolas-a-napelemekre-beruhazo-csaladoknal',
 101             'ext': 'mp4',
 102             'upload_date': '20230911',
 103             'thumbnail': 'https://hirado.hu/wp-content/uploads/sites/4/2023/09/vlcsnap-2023-09-11-09h16m09s882.jpg',
 104         },
 105         'skip': 'Webpage redirects to video list page',
 106     }, {
 107         # (old) petofilive
 108         'url': 'https://petofilive.hu/video/2021/06/07/tha-shudras-az-akusztikban/',
 109         'info_dict': {
 110             'id': '4571948',
 111             'title': 'Tha Shudras az Akusztikban',
 112             'ext': 'mp4',
 113             'upload_date': '20210607',
 114             'thumbnail': 'http://petofilive.hu/wp-content/uploads/sites/4/2021/06/vlcsnap-2021-06-07-22h14m23s915-1024x576.jpg',
 115         },
 116         'skip': 'Webpage redirects to empty page',
 117     }, {
 118         # petofilive
 119         'url': 'https://petofilive.hu/video/2023/09/09/futball-fesztival-a-margitszigeten/',
 120         'info_dict': {
 121             'id': '6713233',
 122             'title': 'Futball Fesztivál a Margitszigeten',
 123             'display_id': 'futball-fesztival-a-margitszigeten',
 124             'ext': 'mp4',
 125             'upload_date': '20230909',
 126             'thumbnail': 'https://petofilive.hu/wp-content/uploads/sites/4/2023/09/Clipboard11-2.jpg',
 127         },
 128         'skip': 'Webpage redirects to video list page',
 129     }]
 130
 131     def _real_extract(self, url):
 132         mobj = self._match_valid_url(url)
 133         display_id = mobj.group('id')
 134         webpage = self._download_webpage(url, display_id)
 135
 136         player_data_str = self._html_search_regex(
 137             r'mtva_player_manager\.player\(document.getElementById\(.*\),\s?(\{.*\}).*\);', webpage, 'player data')
 138         player_data = self._parse_json(player_data_str, display_id, urllib.parse.unquote)
 139         video_id = str(player_data['contentId'])
 140         title = player_data.get('title') or self._og_search_title(webpage, fatal=False) or \
 141             self._html_search_regex(r'<h\d+\b[^>]+\bclass="article_title">([^<]+)<', webpage, 'title')
 142
 143         upload_date = unified_strdate(
 144             '{}-{}-{}'.format(mobj.group('year'), mobj.group('month'), mobj.group('day')))
 145         if not upload_date:
 146             upload_date = unified_strdate(self._html_search_regex(
 147                 r'<p+\b[^>]+\bclass="article_date">([^<]+)<', webpage, 'upload date', default=None))
 148
 149         player_data['video'] = player_data.pop('token')
 150         player_page = self._download_webpage(
 151             'https://player.mediaklikk.hu/playernew/player.php', video_id,
 152             query=player_data, headers={'Referer': url})
 153         player_json = self._search_json(
 154             r'\bpl\.setup\s*\(', player_page, 'player json', video_id, end_pattern=r'\);')
 155         playlist_url = traverse_obj(
 156             player_json, ('playlist', lambda _, v: v['type'] == 'hls', 'file', {url_or_none}), get_all=False)
 157         if not playlist_url:
 158             raise ExtractorError('Unable to extract playlist url')
 159
 160         formats, subtitles = self._extract_m3u8_formats_and_subtitles(playlist_url, video_id)
 161
 162         return {
 163             'id': video_id,
 164             'title': title,
 165             'display_id': display_id,
 166             'formats': formats,
 167             'subtitles': subtitles,
 168             'upload_date': upload_date,
 169             'thumbnail': player_data.get('bgImage') or self._og_search_thumbnail(webpage),
 170         }