yt_dlp/extractor/zenporn.py

   1 import base64
   2 import binascii
   3
   4 from .common import InfoExtractor
   5 from ..utils import ExtractorError, determine_ext, unified_strdate, url_or_none
   6 from ..utils.traversal import traverse_obj
   7
   8
   9 class ZenPornIE(InfoExtractor):
  10     _VALID_URL = r'https?://(?:www\.)?zenporn\.com/video/(?P<id>\d+)'
  11     _TESTS = [{
  12         'url': 'https://zenporn.com/video/15627016/desi-bhabi-ki-chudai',
  13         'md5': '07bd576b5920714d74975c054ca28dee',
  14         'info_dict': {
  15             'id': '9563799',
  16             'display_id': '15627016',
  17             'ext': 'mp4',
  18             'title': 'md5:669eafd3bbc688aa29770553b738ada2',
  19             'description': '',
  20             'thumbnail': 'md5:2fc044a19bab450fef8f1931e7920a18',
  21             'upload_date': '20230925',
  22             'uploader': 'md5:9fae59847f1f58d1da8f2772016c12f3',
  23             'age_limit': 18,
  24         },
  25     }, {
  26         'url': 'https://zenporn.com/video/15570701',
  27         'md5': 'acba0d080d692664fcc8c4e5502b1a67',
  28         'info_dict': {
  29             'id': '2297875',
  30             'display_id': '15570701',
  31             'ext': 'mp4',
  32             'title': 'md5:47aebdf87644ec91e8b1a844bc832451',
  33             'description': '',
  34             'thumbnail': 'https://mstn.nv7s.com/contents/videos_screenshots/2297000/2297875/480x270/1.jpg',
  35             'upload_date': '20230921',
  36             'uploader': 'Lois Clarke',
  37             'age_limit': 18,
  38         },
  39     }, {
  40         'url': 'https://zenporn.com/video/8531117/amateur-students-having-a-fuck-fest-at-club/',
  41         'md5': '67411256aa9451449e4d29f3be525541',
  42         'info_dict': {
  43             'id': '12791908',
  44             'display_id': '8531117',
  45             'ext': 'mp4',
  46             'title': 'Amateur students having a fuck fest at club',
  47             'description': '',
  48             'thumbnail': 'https://tn.txxx.tube/contents/videos_screenshots/12791000/12791908/288x162/1.jpg',
  49             'upload_date': '20191005',
  50             'uploader': 'Jackopenass',
  51             'age_limit': 18,
  52         },
  53     }, {
  54         'url': 'https://zenporn.com/video/15872038/glad-you-came/',
  55         'md5': '296ccab437f5bac6099433768449d8e1',
  56         'info_dict': {
  57             'id': '111585',
  58             'display_id': '15872038',
  59             'ext': 'mp4',
  60             'title': 'Glad You Came',
  61             'description': '',
  62             'thumbnail': 'https://vpim.m3pd.com/contents/videos_screenshots/111000/111585/480x270/1.jpg',
  63             'upload_date': '20231024',
  64             'uploader': 'Martin Rudenko',
  65             'age_limit': 18,
  66         },
  67     }]
  68
  69     def _gen_info_url(self, ext_domain, extr_id, lifetime=86400):
  70         """ This function is a reverse engineering from the website javascript """
  71         result = '/'.join(str(int(extr_id) // i * i) for i in (1_000_000, 1_000, 1))
  72         return f'https://{ext_domain}/api/json/video/{lifetime}/{result}.json'
  73
  74     @staticmethod
  75     def _decode_video_url(encoded_url):
  76         """ This function is a reverse engineering from the website javascript """
  77         # Replace lookalike characters and standardize map
  78         translation = str.maketrans('АВСЕМ.,~', 'ABCEM+/=')
  79         try:
  80             return base64.b64decode(encoded_url.translate(translation), validate=True).decode()
  81         except (binascii.Error, ValueError):
  82             return None
  83
  84     def _real_extract(self, url):
  85         display_id = self._match_id(url)
  86         webpage = self._download_webpage(url, display_id)
  87
  88         ext_domain, video_id = self._search_regex(
  89             r'https://(?P<ext_domain>[\w.-]+\.\w{3})/embed/(?P<extr_id>\d+)/',
  90             webpage, 'embed info', group=('ext_domain', 'extr_id'))
  91
  92         info_json = self._download_json(
  93             self._gen_info_url(ext_domain, video_id), video_id, fatal=False)
  94
  95         video_json = self._download_json(
  96             f'https://{ext_domain}/api/videofile.php', video_id, query={
  97                 'video_id': video_id,
  98                 'lifetime': 8640000,
  99             }, note='Downloading video file JSON', errnote='Failed to download video file JSON')
 100
 101         decoded_url = self._decode_video_url(video_json[0]['video_url'])
 102         if not decoded_url:
 103             raise ExtractorError('Unable to decode the video url')
 104
 105         return {
 106             'id': video_id,
 107             'display_id': display_id,
 108             'ext': traverse_obj(video_json, (0, 'format', {determine_ext})),
 109             'url': f'https://{ext_domain}{decoded_url}',
 110             'age_limit': 18,
 111             **traverse_obj(info_json, ('video', {
 112                 'title': ('title', {str}),
 113                 'description': ('description', {str}),
 114                 'thumbnail': ('thumb', {url_or_none}),
 115                 'upload_date': ('post_date', {unified_strdate}),
 116                 'uploader': ('user', 'username', {str}),
 117             })),
 118         }