yt_dlp/extractor/mediaklikk.py

   1 from ..utils import (
   2     ExtractorError,
   3     traverse_obj,
   4     unified_strdate,
   5     url_or_none,
   6 )
   7 from .common import InfoExtractor
   8 from ..compat import (
   9     compat_urllib_parse_unquote,
  10     compat_str
  11 )
  12
  13
  14 class MediaKlikkIE(InfoExtractor):
  15     _VALID_URL = r'''(?x)https?://(?:www\.)?
  16                         (?:mediaklikk|m4sport|hirado|petofilive)\.hu/.*?(?:videok?|cikk)/
  17                         (?:(?P<year>[0-9]{4})/(?P<month>[0-9]{1,2})/(?P<day>[0-9]{1,2})/)?
  18                         (?P<id>[^/#?_]+)'''
  19
  20     _TESTS = [{
  21         # (old) mediaklikk. date in html.
  22         'url': 'https://mediaklikk.hu/video/hazajaro-delnyugat-bacska-a-duna-menten-palankatol-doroszloig/',
  23         'info_dict': {
  24             'id': '4754129',
  25             'title': 'Hazajáró, DÉLNYUGAT-BÁCSKA – A Duna mentén Palánkától Doroszlóig',
  26             'ext': 'mp4',
  27             'upload_date': '20210901',
  28             'thumbnail': 'http://mediaklikk.hu/wp-content/uploads/sites/4/2014/02/hazajarouj_JO.jpg'
  29         },
  30         'skip': 'Webpage redirects to 404 page',
  31     }, {
  32         # mediaklikk. date in html.
  33         'url': 'https://mediaklikk.hu/video/hazajaro-fabova-hegyseg-kishont-koronaja/',
  34         'info_dict': {
  35             'id': '6696133',
  36             'title': 'Hazajáró, Fabova-hegység - Kishont koronája',
  37             'display_id': 'hazajaro-fabova-hegyseg-kishont-koronaja',
  38             'ext': 'mp4',
  39             'upload_date': '20230903',
  40             'thumbnail': 'https://mediaklikk.hu/wp-content/uploads/sites/4/2014/02/hazajarouj_JO.jpg'
  41         }
  42     }, {
  43         # (old) m4sport
  44         'url': 'https://m4sport.hu/video/2021/08/30/gyemant-liga-parizs/',
  45         'info_dict': {
  46             'id': '4754999',
  47             'title': 'Gyémánt Liga, Párizs',
  48             'ext': 'mp4',
  49             'upload_date': '20210830',
  50             'thumbnail': 'http://m4sport.hu/wp-content/uploads/sites/4/2021/08/vlcsnap-2021-08-30-18h21m20s10-1024x576.jpg'
  51         },
  52         'skip': 'Webpage redirects to 404 page',
  53     }, {
  54         # m4sport
  55         'url': 'https://m4sport.hu/sportkozvetitesek/video/2023/09/08/atletika-gyemant-liga-brusszel/',
  56         'info_dict': {
  57             'id': '6711136',
  58             'title': 'Atlétika – Gyémánt Liga, Brüsszel',
  59             'display_id': 'atletika-gyemant-liga-brusszel',
  60             'ext': 'mp4',
  61             'upload_date': '20230908',
  62             'thumbnail': 'https://m4sport.hu/wp-content/uploads/sites/4/2023/09/vlcsnap-2023-09-08-22h43m18s691.jpg'
  63         }
  64     }, {
  65         # m4sport with *video/ url and no date
  66         'url': 'https://m4sport.hu/bl-video/real-madrid-chelsea-1-1/',
  67         'info_dict': {
  68             'id': '4492099',
  69             'title': 'Real Madrid - Chelsea 1-1',
  70             'display_id': 'real-madrid-chelsea-1-1',
  71             'ext': 'mp4',
  72             'thumbnail': 'https://m4sport.hu/wp-content/uploads/sites/4/2021/04/Sequence-01.Still001-1024x576.png'
  73         }
  74     }, {
  75         # (old) hirado
  76         'url': 'https://hirado.hu/videok/felteteleket-szabott-a-fovaros/',
  77         'info_dict': {
  78             'id': '4760120',
  79             'title': 'Feltételeket szabott a főváros',
  80             'ext': 'mp4',
  81             'thumbnail': 'http://hirado.hu/wp-content/uploads/sites/4/2021/09/vlcsnap-2021-09-01-20h20m37s165.jpg'
  82         },
  83         'skip': 'Webpage redirects to video list page',
  84     }, {
  85         # hirado
  86         'url': 'https://hirado.hu/belfold/video/2023/09/11/marad-az-eves-elszamolas-a-napelemekre-beruhazo-csaladoknal',
  87         'info_dict': {
  88             'id': '6716068',
  89             'title': 'Marad az éves elszámolás a napelemekre beruházó családoknál',
  90             'display_id': 'marad-az-eves-elszamolas-a-napelemekre-beruhazo-csaladoknal',
  91             'ext': 'mp4',
  92             'upload_date': '20230911',
  93             'thumbnail': 'https://hirado.hu/wp-content/uploads/sites/4/2023/09/vlcsnap-2023-09-11-09h16m09s882.jpg'
  94         }
  95     }, {
  96         # (old) petofilive
  97         'url': 'https://petofilive.hu/video/2021/06/07/tha-shudras-az-akusztikban/',
  98         'info_dict': {
  99             'id': '4571948',
 100             'title': 'Tha Shudras az Akusztikban',
 101             'ext': 'mp4',
 102             'upload_date': '20210607',
 103             'thumbnail': 'http://petofilive.hu/wp-content/uploads/sites/4/2021/06/vlcsnap-2021-06-07-22h14m23s915-1024x576.jpg'
 104         },
 105         'skip': 'Webpage redirects to empty page',
 106     }, {
 107         # petofilive
 108         'url': 'https://petofilive.hu/video/2023/09/09/futball-fesztival-a-margitszigeten/',
 109         'info_dict': {
 110             'id': '6713233',
 111             'title': 'Futball Fesztivál a Margitszigeten',
 112             'display_id': 'futball-fesztival-a-margitszigeten',
 113             'ext': 'mp4',
 114             'upload_date': '20230909',
 115             'thumbnail': 'https://petofilive.hu/wp-content/uploads/sites/4/2023/09/Clipboard11-2.jpg'
 116         }
 117     }]
 118
 119     def _real_extract(self, url):
 120         mobj = self._match_valid_url(url)
 121         display_id = mobj.group('id')
 122         webpage = self._download_webpage(url, display_id)
 123
 124         player_data_str = self._html_search_regex(
 125             r'mtva_player_manager\.player\(document.getElementById\(.*\),\s?(\{.*\}).*\);', webpage, 'player data')
 126         player_data = self._parse_json(player_data_str, display_id, compat_urllib_parse_unquote)
 127         video_id = compat_str(player_data['contentId'])
 128         title = player_data.get('title') or self._og_search_title(webpage, fatal=False) or \
 129             self._html_search_regex(r'<h\d+\b[^>]+\bclass="article_title">([^<]+)<', webpage, 'title')
 130
 131         upload_date = unified_strdate(
 132             '%s-%s-%s' % (mobj.group('year'), mobj.group('month'), mobj.group('day')))
 133         if not upload_date:
 134             upload_date = unified_strdate(self._html_search_regex(
 135                 r'<p+\b[^>]+\bclass="article_date">([^<]+)<', webpage, 'upload date', default=None))
 136
 137         player_data['video'] = player_data.pop('token')
 138         player_page = self._download_webpage('https://player.mediaklikk.hu/playernew/player.php', video_id, query=player_data)
 139         player_json = self._search_json(
 140             r'\bpl\.setup\s*\(', player_page, 'player json', video_id, end_pattern=r'\);')
 141         playlist_url = traverse_obj(
 142             player_json, ('playlist', lambda _, v: v['type'] == 'hls', 'file', {url_or_none}), get_all=False)
 143         if not playlist_url:
 144             raise ExtractorError('Unable to extract playlist url')
 145
 146         formats = self._extract_wowza_formats(
 147             playlist_url, video_id, skip_protocols=['f4m', 'smil', 'dash'])
 148
 149         return {
 150             'id': video_id,
 151             'title': title,
 152             'display_id': display_id,
 153             'formats': formats,
 154             'upload_date': upload_date,
 155             'thumbnail': player_data.get('bgImage') or self._og_search_thumbnail(webpage)
 156         }