1 from __future__ import unicode_literals
6 from .common import InfoExtractor
7 from ..compat import compat_urllib_parse_unquote
15 class XTubeIE(InfoExtractor):
16 _VALID_URL = r'(?:xtube:|https?://(?:www\.)?xtube\.com/(?:watch\.php\?.*\bv=|video-watch/(?P<display_id>[^/]+)-))(?P<id>[^/?&#]+)'
20 'url': 'http://www.xtube.com/watch.php?v=kVTUy_G222_',
21 'md5': '092fbdd3cbe292c920ef6fc6a8a9cdab',
25 'title': 'strange erotica',
26 'description': 'contains:an ET kind of thing',
27 'uploader': 'greenshowers',
33 'url': 'http://www.xtube.com/video-watch/strange-erotica-625837',
34 'only_matching': True,
36 'url': 'xtube:625837',
37 'only_matching': True,
40 def _real_extract(self, url):
41 mobj = re.match(self._VALID_URL, url)
42 video_id = mobj.group('id')
43 display_id = mobj.group('display_id')
47 url = 'http://www.xtube.com/watch.php?v=%s' % video_id
49 req = sanitized_Request(url)
50 req.add_header('Cookie', 'age_verified=1; cookiesAccepted=1')
51 webpage = self._download_webpage(req, display_id)
53 flashvars = self._parse_json(
55 r'xt\.playerOps\s*=\s*({.+?});', webpage, 'player ops'),
56 video_id)['flashvars']
58 title = flashvars.get('title') or self._search_regex(
59 r'<h1>([^<]+)</h1>', webpage, 'title')
60 video_url = compat_urllib_parse_unquote(flashvars['video_url'])
61 duration = int_or_none(flashvars.get('video_duration'))
63 uploader = self._search_regex(
64 r'<input[^>]+name="contentOwnerId"[^>]+value="([^"]+)"',
65 webpage, 'uploader', fatal=False)
66 description = self._search_regex(
67 r'</h1>\s*<p>([^<]+)', webpage, 'description', fatal=False)
68 view_count = str_to_int(self._search_regex(
69 r'<dt>Views:</dt>\s*<dd>([\d,\.]+)</dd>',
70 webpage, 'view count', fatal=False))
71 comment_count = str_to_int(self._html_search_regex(
72 r'>Comments? \(([\d,\.]+)\)<',
73 webpage, 'comment count', fatal=False))
77 'display_id': display_id,
80 'description': description,
83 'view_count': view_count,
84 'comment_count': comment_count,
89 class XTubeUserIE(InfoExtractor):
90 IE_DESC = 'XTube user profile'
91 _VALID_URL = r'https?://(?:www\.)?xtube\.com/profile/(?P<id>[^/]+-\d+)'
93 'url': 'http://www.xtube.com/profile/greenshowers-4056496',
95 'id': 'greenshowers-4056496',
98 'playlist_mincount': 155,
101 def _real_extract(self, url):
102 user_id = self._match_id(url)
105 for pagenum in itertools.count(1):
106 request = sanitized_Request(
107 'http://www.xtube.com/profile/%s/videos/%d' % (user_id, pagenum),
109 'Cookie': 'popunder=4',
110 'X-Requested-With': 'XMLHttpRequest',
114 page = self._download_json(
115 request, user_id, 'Downloading videos JSON page %d' % pagenum)
117 html = page.get('html')
121 for _, video_id in re.findall(r'data-plid=(["\'])(.+?)\1', html):
122 entries.append(self.url_result('xtube:%s' % video_id, XTubeIE.ie_key()))
124 page_count = int_or_none(page.get('pageCount'))
125 if not page_count or pagenum == page_count:
128 playlist = self.playlist_result(entries, user_id)
129 playlist['age_limit'] = 18