yt_dlp/extractor/peekvids.py

   1 import re
   2
   3 from .common import InfoExtractor
   4 from ..utils import (
   5     ExtractorError,
   6     get_element_by_class,
   7     int_or_none,
   8     merge_dicts,
   9     url_or_none,
  10 )
  11
  12
  13 class PeekVidsBaseIE(InfoExtractor):
  14     def _real_extract(self, url):
  15         domain, video_id = self._match_valid_url(url).group('domain', 'id')
  16         webpage = self._download_webpage(url, video_id, expected_status=429)
  17         if '>Rate Limit Exceeded' in webpage:
  18             raise ExtractorError(
  19                 f'You are suspected as a bot. Wait, or pass the captcha on the site and provide cookies. {self._login_hint()}',
  20                 video_id=video_id, expected=True)
  21
  22         title = self._html_search_regex(r'(?s)<h1\b[^>]*>(.+?)</h1>', webpage, 'title')
  23
  24         display_id = video_id
  25         video_id = self._search_regex(r'(?s)<video\b[^>]+\bdata-id\s*=\s*["\']?([\w-]+)', webpage, 'short video ID')
  26         srcs = self._download_json(
  27             f'https://www.{domain}/v-alt/{video_id}', video_id,
  28             note='Downloading list of source files')
  29
  30         formats = []
  31         for k, v in srcs.items():
  32             f_url = url_or_none(v)
  33             if not f_url:
  34                 continue
  35
  36             height = self._search_regex(r'^data-src(\d{3,})$', k, 'height', default=None)
  37             if not height:
  38                 continue
  39
  40             formats.append({
  41                 'url': f_url,
  42                 'format_id': height,
  43                 'height': int_or_none(height),
  44             })
  45
  46         if not formats:
  47             formats = [{'url': url} for url in srcs.values()]
  48
  49         info = self._search_json_ld(webpage, video_id, expected_type='VideoObject', default={})
  50         info.pop('url', None)
  51
  52         # may not have found the thumbnail if it was in a list in the ld+json
  53         info.setdefault('thumbnail', self._og_search_thumbnail(webpage))
  54         detail = (get_element_by_class('detail-video-block', webpage)
  55                   or get_element_by_class('detail-block', webpage) or '')
  56         info['description'] = self._html_search_regex(
  57             rf'(?s)(.+?)(?:{re.escape(info.get("description", ""))}\s*<|<ul\b)',
  58             detail, 'description', default=None) or None
  59         info['title'] = re.sub(r'\s*[,-][^,-]+$', '', info.get('title') or title) or self._generic_title(url)
  60
  61         def cat_tags(name, html):
  62             l = self._html_search_regex(
  63                 rf'(?s)<span\b[^>]*>\s*{re.escape(name)}\s*:\s*</span>(.+?)</li>',
  64                 html, name, default='')
  65             return list(filter(None, re.split(r'\s+', l)))
  66
  67         return merge_dicts({
  68             'id': video_id,
  69             'display_id': display_id,
  70             'age_limit': 18,
  71             'formats': formats,
  72             'categories': cat_tags('Categories', detail),
  73             'tags': cat_tags('Tags', detail),
  74             'uploader': self._html_search_regex(r'[Uu]ploaded\s+by\s(.+?)"', webpage, 'uploader', default=None),
  75         }, info)
  76
  77
  78 class PeekVidsIE(PeekVidsBaseIE):
  79     _VALID_URL = r'''(?x)
  80         https?://(?:www\.)?(?P<domain>peekvids\.com)/
  81         (?:(?:[^/?#]+/){2}|embed/?\?(?:[^#]*&)?v=)
  82         (?P<id>[^/?&#]*)
  83     '''
  84     _TESTS = [{
  85         'url': 'https://peekvids.com/pc/dane-jones-cute-redhead-with-perfect-tits-with-mini-vamp/BSyLMbN0YCd',
  86         'md5': '2ff6a357a9717dc9dc9894b51307e9a2',
  87         'info_dict': {
  88             'id': '1262717',
  89             'display_id': 'BSyLMbN0YCd',
  90             'title': ' Dane Jones - Cute redhead with perfect tits with Mini Vamp',
  91             'ext': 'mp4',
  92             'thumbnail': r're:^https?://.*\.jpg$',
  93             'description': 'md5:0a61df3620de26c0af8963b1a730cd69',
  94             'timestamp': 1642579329,
  95             'upload_date': '20220119',
  96             'duration': 416,
  97             'view_count': int,
  98             'age_limit': 18,
  99             'uploader': 'SEXYhub.com',
 100             'categories': list,
 101             'tags': list,
 102         },
 103     }]
 104
 105
 106 class PlayVidsIE(PeekVidsBaseIE):
 107     _VALID_URL = r'https?://(?:www\.)?(?P<domain>playvids\.com)/(?:embed/|\w\w?/)?(?P<id>[^/?#]*)'
 108     _TESTS = [{
 109         'url': 'https://www.playvids.com/U3pBrYhsjXM/pc/dane-jones-cute-redhead-with-perfect-tits-with-mini-vamp',
 110         'md5': '2f12e50213dd65f142175da633c4564c',
 111         'info_dict': {
 112             'id': '1978030',
 113             'display_id': 'U3pBrYhsjXM',
 114             'title': ' Dane Jones - Cute redhead with perfect tits with Mini Vamp',
 115             'ext': 'mp4',
 116             'thumbnail': r're:^https?://.*\.jpg$',
 117             'description': 'md5:0a61df3620de26c0af8963b1a730cd69',
 118             'timestamp': 1640435839,
 119             'upload_date': '20211225',
 120             'duration': 416,
 121             'view_count': int,
 122             'age_limit': 18,
 123             'uploader': 'SEXYhub.com',
 124             'categories': list,
 125             'tags': list,
 126         },
 127     }, {
 128         'url': 'https://www.playvids.com/es/U3pBrYhsjXM/pc/dane-jones-cute-redhead-with-perfect-tits-with-mini-vamp',
 129         'only_matching': True,
 130     }, {
 131         'url': 'https://www.playvids.com/embed/U3pBrYhsjXM',
 132         'only_matching': True,
 133     }, {
 134         'url': 'https://www.playvids.com/bKmGLe3IwjZ/sv/brazzers-800-phone-sex-madison-ivy-always-on-the-line',
 135         'md5': 'e783986e596cafbf46411a174ab42ba6',
 136         'info_dict': {
 137             'id': '762385',
 138             'display_id': 'bKmGLe3IwjZ',
 139             'ext': 'mp4',
 140             'title': 'Brazzers - 1 800 Phone Sex: Madison Ivy Always On The Line 6',
 141             'description': 'md5:bdcd2db2b8ad85831a491d7c8605dcef',
 142             'timestamp': 1516958544,
 143             'upload_date': '20180126',
 144             'thumbnail': r're:^https?://.*\.jpg$',
 145             'duration': 480,
 146             'uploader': 'Brazzers',
 147             'age_limit': 18,
 148             'view_count': int,
 149             'categories': list,
 150             'tags': list,
 151         },
 152     }, {
 153         'url': 'https://www.playvids.com/v/47iUho33toY',
 154         'md5': 'b056b5049d34b648c1e86497cf4febce',
 155         'info_dict': {
 156             'id': '700621',
 157             'display_id': '47iUho33toY',
 158             'ext': 'mp4',
 159             'title': 'KATEE OWEN STRIPTIASE IN SEXY RED LINGERIE',
 160             'timestamp': 1507052209,
 161             'upload_date': '20171003',
 162             'thumbnail': r're:^https?://.*\.jpg$',
 163             'duration': 332,
 164             'uploader': 'Cacerenele',
 165             'age_limit': 18,
 166             'view_count': int,
 167             'categories': list,
 168             'tags': list,
 169         },
 170     }, {
 171         'url': 'https://www.playvids.com/z3_7iwWCmqt/sexy-teen-filipina-striptease-beautiful-pinay-bargirl-strips-and-dances',
 172         'md5': 'efa09be9f031314b7b7e3bc6510cd0df',
 173         'info_dict': {
 174             'id': '1523518',
 175             'display_id': 'z3_7iwWCmqt',
 176             'ext': 'mp4',
 177             'title': 'SEXY TEEN FILIPINA STRIPTEASE - Beautiful Pinay Bargirl Strips and Dances',
 178             'timestamp': 1607470323,
 179             'upload_date': '20201208',
 180             'thumbnail': r're:^https?://.*\.jpg$',
 181             'duration': 593,
 182             'uploader': 'yorours',
 183             'age_limit': 18,
 184             'view_count': int,
 185             'categories': list,
 186             'tags': list,
 187         },
 188     }]