]>
Commit | Line | Data |
---|---|---|
c5ba203e AS |
1 | from __future__ import unicode_literals |
2 | ||
dcc2a706 | 3 | import re |
4 | ||
5 | from .common import InfoExtractor | |
1cc79574 | 6 | from ..compat import ( |
dcc2a706 | 7 | compat_urllib_request, |
ee8de13e | 8 | compat_urllib_parse_unquote, |
1cc79574 PH |
9 | ) |
10 | from ..utils import ( | |
607dbbad S |
11 | parse_duration, |
12 | str_to_int, | |
dcc2a706 | 13 | ) |
14 | ||
607dbbad | 15 | |
dcc2a706 | 16 | class XTubeIE(InfoExtractor): |
16ea8179 | 17 | _VALID_URL = r'https?://(?:www\.)?(?P<url>xtube\.com/watch\.php\?v=(?P<id>[^/?&#]+))' |
dcc2a706 | 18 | _TEST = { |
c5ba203e | 19 | 'url': 'http://www.xtube.com/watch.php?v=kVTUy_G222_', |
c5ba203e AS |
20 | 'md5': '092fbdd3cbe292c920ef6fc6a8a9cdab', |
21 | 'info_dict': { | |
607dbbad S |
22 | 'id': 'kVTUy_G222_', |
23 | 'ext': 'mp4', | |
24 | 'title': 'strange erotica', | |
9789d753 | 25 | 'description': 'contains:an ET kind of thing', |
607dbbad S |
26 | 'uploader': 'greenshowers', |
27 | 'duration': 450, | |
28 | 'age_limit': 18, | |
dcc2a706 | 29 | } |
30 | } | |
31 | ||
32 | def _real_extract(self, url): | |
16ea8179 | 33 | video_id = self._match_id(url) |
dcc2a706 | 34 | |
35 | req = compat_urllib_request.Request(url) | |
36 | req.add_header('Cookie', 'age_verified=1') | |
37 | webpage = self._download_webpage(req, video_id) | |
38 | ||
16ea8179 S |
39 | video_title = self._html_search_regex( |
40 | r'<p class="title">([^<]+)', webpage, 'title') | |
607dbbad | 41 | video_uploader = self._html_search_regex( |
16ea8179 | 42 | [r"var\s+contentOwnerId\s*=\s*'([^']+)", |
324b2c78 | 43 | r'By:\s*<a href="/community/profile\.php\?user=([^"]+)'], |
16ea8179 | 44 | webpage, 'uploader', fatal=False) |
607dbbad | 45 | video_description = self._html_search_regex( |
16ea8179 S |
46 | r'<p class="fieldsDesc">([^<]+)', |
47 | webpage, 'description', fatal=False) | |
607dbbad | 48 | duration = parse_duration(self._html_search_regex( |
16ea8179 S |
49 | r'<span class="bold">Runtime:</span> ([^<]+)</p>', |
50 | webpage, 'duration', fatal=False)) | |
51 | view_count = str_to_int(self._html_search_regex( | |
52 | r'<span class="bold">Views:</span> ([\d,\.]+)</p>', | |
53 | webpage, 'view count', fatal=False)) | |
54 | comment_count = str_to_int(self._html_search_regex( | |
55 | r'<div id="commentBar">([\d,\.]+) Comments</div>', | |
56 | webpage, 'comment count', fatal=False)) | |
aa488e13 | 57 | |
16ea8179 S |
58 | formats = [] |
59 | for format_id, video_url in re.findall( | |
60 | r'flashvars\.quality_(.+?)\s*=\s*"([^"]+)"', webpage): | |
61 | fmt = { | |
ee8de13e | 62 | 'url': compat_urllib_parse_unquote(video_url), |
aa488e13 | 63 | 'format_id': format_id, |
16ea8179 S |
64 | } |
65 | m = re.search(r'^(?P<height>\d+)[pP]', format_id) | |
66 | if m: | |
67 | fmt['height'] = int(m.group('height')) | |
68 | formats.append(fmt) | |
69 | ||
70 | if not formats: | |
ee8de13e | 71 | video_url = compat_urllib_parse_unquote(self._search_regex( |
16ea8179 S |
72 | r'flashvars\.video_url\s*=\s*"([^"]+)"', |
73 | webpage, 'video URL')) | |
74 | formats.append({'url': video_url}) | |
75 | ||
aa488e13 | 76 | self._sort_formats(formats) |
dcc2a706 | 77 | |
78 | return { | |
79 | 'id': video_id, | |
80 | 'title': video_title, | |
81 | 'uploader': video_uploader, | |
82 | 'description': video_description, | |
607dbbad S |
83 | 'duration': duration, |
84 | 'view_count': view_count, | |
85 | 'comment_count': comment_count, | |
aa488e13 | 86 | 'formats': formats, |
dcc2a706 | 87 | 'age_limit': 18, |
9f5809b3 | 88 | } |
89 | ||
22a6f150 | 90 | |
9f5809b3 | 91 | class XTubeUserIE(InfoExtractor): |
92 | IE_DESC = 'XTube user profile' | |
93 | _VALID_URL = r'https?://(?:www\.)?xtube\.com/community/profile\.php\?(.*?)user=(?P<username>[^&#]+)(?:$|[&#])' | |
22a6f150 PH |
94 | _TEST = { |
95 | 'url': 'http://www.xtube.com/community/profile.php?user=greenshowers', | |
96 | 'info_dict': { | |
97 | 'id': 'greenshowers', | |
05900629 | 98 | 'age_limit': 18, |
22a6f150 PH |
99 | }, |
100 | 'playlist_mincount': 155, | |
101 | } | |
9f5809b3 | 102 | |
103 | def _real_extract(self, url): | |
104 | mobj = re.match(self._VALID_URL, url) | |
105 | username = mobj.group('username') | |
106 | ||
107 | profile_page = self._download_webpage( | |
108 | url, username, note='Retrieving profile page') | |
109 | ||
110 | video_count = int(self._search_regex( | |
2514d263 | 111 | r'<strong>%s\'s Videos \(([0-9]+)\)</strong>' % username, profile_page, |
9f5809b3 | 112 | 'video count')) |
113 | ||
114 | PAGE_SIZE = 25 | |
115 | urls = [] | |
116 | page_count = (video_count + PAGE_SIZE + 1) // PAGE_SIZE | |
117 | for n in range(1, page_count + 1): | |
118 | lpage_url = 'http://www.xtube.com/user_videos.php?page=%d&u=%s' % (n, username) | |
119 | lpage = self._download_webpage( | |
120 | lpage_url, username, | |
121 | note='Downloading page %d/%d' % (n, page_count)) | |
122 | urls.extend( | |
123 | re.findall(r'addthis:url="([^"]+)"', lpage)) | |
124 | ||
125 | return { | |
126 | '_type': 'playlist', | |
127 | 'id': username, | |
05900629 | 128 | 'age_limit': 18, |
9f5809b3 | 129 | 'entries': [{ |
130 | '_type': 'url', | |
131 | 'url': eurl, | |
132 | 'ie_key': 'XTube', | |
133 | } for eurl in urls] | |
134 | } |