yt_dlp/extractor/xvideos.py

   1 import re
   2
   3 from .common import InfoExtractor
   4 from ..compat import compat_urllib_parse_unquote
   5 from ..utils import (
   6     clean_html,
   7     determine_ext,
   8     ExtractorError,
   9     int_or_none,
  10     parse_duration,
  11 )
  12
  13
  14 class XVideosIE(InfoExtractor):
  15     _VALID_URL = r'''(?x)
  16                     https?://
  17                         (?:
  18                             (?:[^/]+\.)?xvideos2?\.com/video\.?|
  19                             (?:www\.)?xvideos\.es/video\.?|
  20                             (?:www|flashservice)\.xvideos\.com/embedframe/|
  21                             static-hw\.xvideos\.com/swf/xv-player\.swf\?.*?\bid_video=
  22                         )
  23                         (?P<id>[0-9a-z]+)
  24                     '''
  25     _TESTS = [{
  26         'url': 'http://xvideos.com/video.ucuvbkfda4e/a_beautiful_red-haired_stranger_was_refused_but_still_came_to_my_room_for_sex',
  27         'md5': '396255a900a6bddb3e98985f0b86c3fd',
  28         'info_dict': {
  29             'id': 'ucuvbkfda4e',
  30             'ext': 'mp4',
  31             'title': 'A Beautiful Red-Haired Stranger Was Refused, But Still Came To My Room For Sex',
  32             'duration': 1238,
  33             'age_limit': 18,
  34             'thumbnail': r're:^https://cdn\d+-pic.xvideos-cdn.com/.+\.jpg',
  35         }
  36     }, {
  37         # Broken HLS formats
  38         'url': 'https://www.xvideos.com/video65982001/what_s_her_name',
  39         'md5': '56742808292c8fa1418e4538c262c58b',
  40         'info_dict': {
  41             'id': '65982001',
  42             'ext': 'mp4',
  43             'title': 'what\'s her name?',
  44             'duration': 120,
  45             'age_limit': 18,
  46             'thumbnail': r're:^https://cdn\d+-pic.xvideos-cdn.com/.+\.jpg',
  47         }
  48     }, {
  49         'url': 'https://flashservice.xvideos.com/embedframe/4588838',
  50         'only_matching': True,
  51     }, {
  52         'url': 'https://www.xvideos.com/embedframe/4588838',
  53         'only_matching': True,
  54     }, {
  55         'url': 'http://static-hw.xvideos.com/swf/xv-player.swf?id_video=4588838',
  56         'only_matching': True,
  57     }, {
  58         'url': 'http://xvideos.com/video4588838/biker_takes_his_girl',
  59         'only_matching': True
  60     }, {
  61         'url': 'https://xvideos.com/video4588838/biker_takes_his_girl',
  62         'only_matching': True
  63     }, {
  64         'url': 'https://xvideos.es/video4588838/biker_takes_his_girl',
  65         'only_matching': True
  66     }, {
  67         'url': 'https://www.xvideos.es/video4588838/biker_takes_his_girl',
  68         'only_matching': True
  69     }, {
  70         'url': 'http://xvideos.es/video4588838/biker_takes_his_girl',
  71         'only_matching': True
  72     }, {
  73         'url': 'http://www.xvideos.es/video4588838/biker_takes_his_girl',
  74         'only_matching': True
  75     }, {
  76         'url': 'http://fr.xvideos.com/video4588838/biker_takes_his_girl',
  77         'only_matching': True
  78     }, {
  79         'url': 'https://fr.xvideos.com/video4588838/biker_takes_his_girl',
  80         'only_matching': True
  81     }, {
  82         'url': 'http://it.xvideos.com/video4588838/biker_takes_his_girl',
  83         'only_matching': True
  84     }, {
  85         'url': 'https://it.xvideos.com/video4588838/biker_takes_his_girl',
  86         'only_matching': True
  87     }, {
  88         'url': 'http://de.xvideos.com/video4588838/biker_takes_his_girl',
  89         'only_matching': True
  90     }, {
  91         'url': 'https://de.xvideos.com/video4588838/biker_takes_his_girl',
  92         'only_matching': True
  93     }, {
  94         'url': 'https://flashservice.xvideos.com/embedframe/ucuvbkfda4e',
  95         'only_matching': True,
  96     }, {
  97         'url': 'https://www.xvideos.com/embedframe/ucuvbkfda4e',
  98         'only_matching': True,
  99     }, {
 100         'url': 'http://static-hw.xvideos.com/swf/xv-player.swf?id_video=ucuvbkfda4e',
 101         'only_matching': True,
 102     }, {
 103         'url': 'https://xvideos.es/video.ucuvbkfda4e/a_beautiful_red-haired_stranger_was_refused_but_still_came_to_my_room_for_sex',
 104         'only_matching': True
 105     }]
 106
 107     def _real_extract(self, url):
 108         video_id = self._match_id(url)
 109         webpage = self._download_webpage(url, video_id)
 110
 111         mobj = re.search(r'<h1 class="inlineError">(.+?)</h1>', webpage)
 112         if mobj:
 113             raise ExtractorError('%s said: %s' % (self.IE_NAME, clean_html(mobj.group(1))), expected=True)
 114
 115         title = self._html_search_regex(
 116             (r'<title>(?P<title>.+?)\s+-\s+XVID',
 117              r'setVideoTitle\s*\(\s*(["\'])(?P<title>(?:(?!\1).)+)\1'),
 118             webpage, 'title', default=None,
 119             group='title') or self._og_search_title(webpage)
 120
 121         thumbnails = []
 122         for preference, thumbnail in enumerate(('', '169')):
 123             thumbnail_url = self._search_regex(
 124                 r'setThumbUrl%s\(\s*(["\'])(?P<thumbnail>(?:(?!\1).)+)\1' % thumbnail,
 125                 webpage, 'thumbnail', default=None, group='thumbnail')
 126             if thumbnail_url:
 127                 thumbnails.append({
 128                     'url': thumbnail_url,
 129                     'preference': preference,
 130                 })
 131
 132         duration = int_or_none(self._og_search_property(
 133             'duration', webpage, default=None)) or parse_duration(
 134             self._search_regex(
 135                 r'<span[^>]+class=["\']duration["\'][^>]*>.*?(\d[^<]+)',
 136                 webpage, 'duration', fatal=False))
 137
 138         formats = []
 139
 140         video_url = compat_urllib_parse_unquote(self._search_regex(
 141             r'flv_url=(.+?)&', webpage, 'video URL', default=''))
 142         if video_url:
 143             formats.append({
 144                 'url': video_url,
 145                 'format_id': 'flv',
 146             })
 147
 148         for kind, _, format_url in re.findall(
 149                 r'setVideo([^(]+)\((["\'])(http.+?)\2\)', webpage):
 150             format_id = kind.lower()
 151             if format_id == 'hls':
 152                 hls_formats = self._extract_m3u8_formats(
 153                     format_url, video_id, 'mp4',
 154                     entry_protocol='m3u8_native', m3u8_id='hls', fatal=False)
 155                 self._check_formats(hls_formats, video_id)
 156                 formats.extend(hls_formats)
 157             elif format_id in ('urllow', 'urlhigh'):
 158                 formats.append({
 159                     'url': format_url,
 160                     'format_id': '%s-%s' % (determine_ext(format_url, 'mp4'), format_id[3:]),
 161                     'quality': -2 if format_id.endswith('low') else None,
 162                 })
 163
 164         return {
 165             'id': video_id,
 166             'formats': formats,
 167             'title': title,
 168             'duration': duration,
 169             'thumbnails': thumbnails,
 170             'age_limit': 18,
 171         }
 172
 173
 174 class XVideosQuickiesIE(InfoExtractor):
 175     IE_NAME = 'xvideos:quickies'
 176     _VALID_URL = r'https?://(?P<domain>(?:[^/]+\.)?xvideos2?\.com)/amateur-channels/[^#]+#quickies/a/(?P<id>\d+)'
 177     _TESTS = [{
 178         'url': 'https://www.xvideos.com/amateur-channels/wifeluna#quickies/a/47258683',
 179         'md5': '16e322a93282667f1963915568f782c1',
 180         'info_dict': {
 181             'id': '47258683',
 182             'ext': 'mp4',
 183             'title': 'Verification video',
 184             'age_limit': 18,
 185             'duration': 16,
 186             'thumbnail': r're:^https://cdn.*-pic.xvideos-cdn.com/.+\.jpg',
 187         }
 188     }]
 189
 190     def _real_extract(self, url):
 191         domain, id_ = self._match_valid_url(url).group('domain', 'id')
 192         return self.url_result(f'https://{domain}/video{id_}/_', XVideosIE, id_)