1 from __future__ import unicode_literals
5 from .common import InfoExtractor
12 class XXXYMoviesIE(InfoExtractor):
13 _VALID_URL = r'https?://(?:www\.)?xxxymovies\.com/videos/(?P<id>\d+)/(?P<display_id>[^/]+)'
15 'url': 'http://xxxymovies.com/videos/138669/ecstatic-orgasm-sofcore/',
16 'md5': '810b1bdbbffff89dd13bdb369fe7be4b',
19 'display_id': 'ecstatic-orgasm-sofcore',
21 'title': 'Ecstatic Orgasm Sofcore',
27 def _real_extract(self, url):
28 mobj = re.match(self._VALID_URL, url)
30 video_id = mobj.group('id')
31 display_id = mobj.group('display_id')
33 webpage = self._download_webpage(url, video_id)
35 video_url = self._html_search_regex(
36 r"video_url\s*:\s*'([^']+)'", webpage, 'video URL')
38 title = self._html_search_regex(
39 r'<title>(.*?)\s*-\s*XXXYMovies.com</title>', webpage, 'title')
41 thumbnail = self._html_search_regex(
42 r'preview_url\s*:\s*\'(.*?)\'', webpage, 'thumbnail', fatal=False)
44 categories = self._html_search_meta(
45 'keywords', webpage, 'categories', default='').split(',')
47 duration = parse_duration(self._search_regex(
48 r'<span>Duration:</span>\s*(\d+:\d+)', webpage, 'duration', fatal=False))
50 view_count = int_or_none(self._html_search_regex(
51 r'<div class="video_views">\s*(\d+)', webpage, 'view count', fatal=False))
55 'display_id': display_id,
58 'thumbnail': thumbnail,
59 'categories': categories,
61 'view_count': view_count,