Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/twentyfourvideo.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .common import InfoExtractor
   5 from ..utils import (
   6     parse_iso8601,
   7     int_or_none,
   8     xpath_attr,
   9     xpath_element,
  10 )
  11
  12
  13 class TwentyFourVideoIE(InfoExtractor):
  14     IE_NAME = '24video'
  15     _VALID_URL = r'https?://(?:www\.)?24video\.(?:net|me|xxx|sex|tube)/(?:video/(?:view|xml)/|player/new24_play\.swf\?id=)(?P<id>\d+)'
  16
  17     _TESTS = [{
  18         'url': 'http://www.24video.net/video/view/1044982',
  19         'md5': 'e09fc0901d9eaeedac872f154931deeb',
  20         'info_dict': {
  21             'id': '1044982',
  22             'ext': 'mp4',
  23             'title': 'Эротика каменного века',
  24             'description': 'Как смотрели порно в каменном веке.',
  25             'thumbnail': r're:^https?://.*\.jpg$',
  26             'uploader': 'SUPERTELO',
  27             'duration': 31,
  28             'timestamp': 1275937857,
  29             'upload_date': '20100607',
  30             'age_limit': 18,
  31             'like_count': int,
  32             'dislike_count': int,
  33         },
  34     }, {
  35         'url': 'http://www.24video.net/player/new24_play.swf?id=1044982',
  36         'only_matching': True,
  37     }, {
  38         'url': 'http://www.24video.me/video/view/1044982',
  39         'only_matching': True,
  40     }, {
  41         'url': 'http://www.24video.tube/video/view/2363750',
  42         'only_matching': True,
  43     }]
  44
  45     def _real_extract(self, url):
  46         video_id = self._match_id(url)
  47
  48         webpage = self._download_webpage(
  49             'http://www.24video.sex/video/view/%s' % video_id, video_id)
  50
  51         title = self._og_search_title(webpage)
  52         description = self._html_search_regex(
  53             r'<(p|span)[^>]+itemprop="description"[^>]*>(?P<description>[^<]+)</\1>',
  54             webpage, 'description', fatal=False, group='description')
  55         thumbnail = self._og_search_thumbnail(webpage)
  56         duration = int_or_none(self._og_search_property(
  57             'duration', webpage, 'duration', fatal=False))
  58         timestamp = parse_iso8601(self._search_regex(
  59             r'<time id="video-timeago" datetime="([^"]+)" itemprop="uploadDate">',
  60             webpage, 'upload date'))
  61
  62         uploader = self._html_search_regex(
  63             r'class="video-uploaded"[^>]*>\s*<a href="/jsecUser/movies/[^"]+"[^>]*>([^<]+)</a>',
  64             webpage, 'uploader', fatal=False)
  65
  66         view_count = int_or_none(self._html_search_regex(
  67             r'<span class="video-views">(\d+) просмотр',
  68             webpage, 'view count', fatal=False))
  69         comment_count = int_or_none(self._html_search_regex(
  70             r'<a[^>]+href="#tab-comments"[^>]*>(\d+) комментари',
  71             webpage, 'comment count', fatal=False))
  72
  73         # Sets some cookies
  74         self._download_xml(
  75             r'http://www.24video.sex/video/xml/%s?mode=init' % video_id,
  76             video_id, 'Downloading init XML')
  77
  78         video_xml = self._download_xml(
  79             'http://www.24video.sex/video/xml/%s?mode=play' % video_id,
  80             video_id, 'Downloading video XML')
  81
  82         video = xpath_element(video_xml, './/video', 'video', fatal=True)
  83
  84         formats = [{
  85             'url': xpath_attr(video, '', 'url', 'video URL', fatal=True),
  86         }]
  87
  88         like_count = int_or_none(video.get('ratingPlus'))
  89         dislike_count = int_or_none(video.get('ratingMinus'))
  90         age_limit = 18 if video.get('adult') == 'true' else 0
  91
  92         return {
  93             'id': video_id,
  94             'title': title,
  95             'description': description,
  96             'thumbnail': thumbnail,
  97             'uploader': uploader,
  98             'duration': duration,
  99             'timestamp': timestamp,
 100             'view_count': view_count,
 101             'comment_count': comment_count,
 102             'like_count': like_count,
 103             'dislike_count': dislike_count,
 104             'age_limit': age_limit,
 105             'formats': formats,
 106         }