yt_dlp/extractor/mojevideo.py

   1 from .common import InfoExtractor
   2 from ..utils import js_to_json, remove_end, update_url_query
   3
   4
   5 class MojevideoIE(InfoExtractor):
   6     IE_DESC = 'mojevideo.sk'
   7     _VALID_URL = r'https?://(?:www\.)?mojevideo\.sk/video/(?P<id>\w+)/(?P<display_id>[\w()]+?)\.html'
   8
   9     _TESTS = [{
  10         'url': 'https://www.mojevideo.sk/video/3d17c/chlapci_dobetonovali_sme_mame_hotovo.html',
  11         'md5': '384a4628bd2bbd261c5206cf77c38c17',
  12         'info_dict': {
  13             'id': '3d17c',
  14             'ext': 'mp4',
  15             'title': 'Chlapci dobetónovali sme, máme hotovo!',
  16             'display_id': 'chlapci_dobetonovali_sme_mame_hotovo',
  17             'description': 'md5:a0822126044050d304a9ef58c92ddb34',
  18             'thumbnail': 'https://fs5.mojevideo.sk/imgfb/250236.jpg',
  19             'duration': 21.0,
  20             'upload_date': '20230919',
  21             'timestamp': 1695129706,
  22             'like_count': int,
  23             'dislike_count': int,
  24             'view_count': int,
  25             'comment_count': int,
  26         },
  27     }, {
  28         # 720p
  29         'url': 'https://www.mojevideo.sk/video/14677/den_blbec.html',
  30         'md5': '517c3e111c53a67d10b429c1f344ba2f',
  31         'info_dict': {
  32             'id': '14677',
  33             'ext': 'mp4',
  34             'title': 'Deň blbec?',
  35             'display_id': 'den_blbec',
  36             'description': 'I maličkosť vám môže zmeniť celý deň. Nikdy nezahadzujte žuvačky na zem!',
  37             'thumbnail': 'https://fs5.mojevideo.sk/imgfb/83575.jpg',
  38             'duration': 100.0,
  39             'upload_date': '20120515',
  40             'timestamp': 1337076481,
  41             'like_count': int,
  42             'dislike_count': int,
  43             'view_count': int,
  44             'comment_count': int,
  45         },
  46     }, {
  47         # 1080p
  48         'url': 'https://www.mojevideo.sk/video/2feb2/band_maid_onset_(instrumental)_live_zepp_tokyo_(full_hd).html',
  49         'md5': '64599a23d3ac31cf2fe069e4353d8162',
  50         'info_dict': {
  51             'id': '2feb2',
  52             'ext': 'mp4',
  53             'title': 'BAND-MAID - onset (Instrumental) Live - Zepp Tokyo (Full HD)',
  54             'display_id': 'band_maid_onset_(instrumental)_live_zepp_tokyo_(full_hd)',
  55             'description': 'Výborná inštrumentálna skladba od skupiny BAND-MAID.',
  56             'thumbnail': 'https://fs5.mojevideo.sk/imgfb/196274.jpg',
  57             'duration': 240.0,
  58             'upload_date': '20190708',
  59             'timestamp': 1562576592,
  60             'like_count': int,
  61             'dislike_count': int,
  62             'view_count': int,
  63             'comment_count': int,
  64         },
  65     }, {
  66         # 720p
  67         'url': 'https://www.mojevideo.sk/video/358c8/dva_nissany_skyline_strielaju_v_londyne.html',
  68         'only_matching': True,
  69     }, {
  70         # 720p
  71         'url': 'https://www.mojevideo.sk/video/2455d/gopro_hero4_session_nova_sportova_vodotesna_kamera.html',
  72         'only_matching': True,
  73     }, {
  74         # 1080p
  75         'url': 'https://www.mojevideo.sk/video/352ee/amd_rx_6800_xt_vs_nvidia_rtx_3080_(test_v_9_hrach).html',
  76         'only_matching': True,
  77     }, {
  78         # 1080p
  79         'url': 'https://www.mojevideo.sk/video/2cbeb/trailer_z_avengers_infinity_war.html',
  80         'only_matching': True,
  81     }]
  82
  83     def _real_extract(self, url):
  84         video_id, display_id = self._match_valid_url(url).groups()
  85         webpage = self._download_webpage(url, video_id)
  86
  87         video_id_dec = self._search_regex(
  88             r'\bvId\s*=\s*(\d+)', webpage, 'video id', fatal=False) or str(int(video_id, 16))
  89         video_exp = self._search_regex(r'\bvEx\s*=\s*["\'](\d+)', webpage, 'video expiry')
  90         video_hashes = self._search_json(
  91             r'\bvHash\s*=', webpage, 'video hashes', video_id,
  92             contains_pattern=r'\[(?s:.+)\]', transform_source=js_to_json)
  93
  94         formats = []
  95         for video_hash, (suffix, quality, format_note) in zip(video_hashes, [
  96             ('', 1, 'normálna kvalita'),
  97             ('_lq', 0, 'nízka kvalita'),
  98             ('_hd', 2, 'HD-720p'),
  99             ('_fhd', 3, 'FULL HD-1080p'),
 100             ('_2k', 4, '2K-1440p'),
 101         ]):
 102             formats.append({
 103                 'format_id': f'mp4-{quality}',
 104                 'quality': quality,
 105                 'format_note': format_note,
 106                 'url': update_url_query(
 107                     f'https://cache01.mojevideo.sk/securevideos69/{video_id_dec}{suffix}.mp4', {
 108                         'md5': video_hash,
 109                         'expires': video_exp,
 110                     }),
 111             })
 112
 113         return {
 114             'id': video_id,
 115             'display_id': display_id,
 116             'formats': formats,
 117             'title': (self._og_search_title(webpage, default=None)
 118                       or remove_end(self._html_extract_title(webpage, 'title'), ' - Mojevideo')),
 119             'description': self._og_search_description(webpage),
 120             **self._search_json_ld(webpage, video_id, default={}),
 121         }