]>
Commit | Line | Data |
---|---|---|
d90b3854 P |
1 | # encoding: utf-8 |
2 | from __future__ import unicode_literals | |
3 | ||
d90b3854 P |
4 | from ..compat import compat_urllib_parse |
5 | from .common import InfoExtractor | |
a80601f8 YCH |
6 | from ..utils import ( |
7 | parse_duration, | |
8 | int_or_none, | |
d05a1dbe | 9 | ExtractorError, |
a80601f8 | 10 | ) |
d90b3854 P |
11 | |
12 | ||
13 | class Porn91IE(InfoExtractor): | |
14 | IE_NAME = '91porn' | |
15 | _VALID_URL = r'(?:https?://)(?:www\.|)91porn\.com/.+?\?viewkey=(?P<id>[\w\d]+)' | |
16 | ||
17 | _TEST = { | |
9ff811c5 YCH |
18 | 'url': 'http://91porn.com/view_video.php?viewkey=7e42283b4f5ab36da134', |
19 | 'md5': '6df8f6d028bc8b14f5dbd73af742fb20', | |
20 | 'info_dict': { | |
21 | 'id': '7e42283b4f5ab36da134', | |
22 | 'title': '18岁大一漂亮学妹,水嫩性感,再爽一次!', | |
a80601f8 YCH |
23 | 'ext': 'mp4', |
24 | 'duration': 431, | |
9ff811c5 | 25 | } |
d90b3854 P |
26 | } |
27 | ||
28 | def _real_extract(self, url): | |
1c222387 | 29 | video_id = self._match_id(url) |
d90b3854 P |
30 | url = 'http://91porn.com/view_video.php?viewkey=%s' % video_id |
31 | self._set_cookie('91porn.com', 'language', 'cn_CN') | |
a2d97130 | 32 | webpage = self._download_webpage(url, video_id, 'get HTML content') |
d05a1dbe YCH |
33 | |
34 | if '作为游客,你每天只可观看10个视频' in webpage: | |
35 | raise ExtractorError('91 Porn says: Daily limit 10 videos exceeded', expected=True) | |
36 | ||
703d78bb | 37 | title = self._search_regex( |
1c222387 | 38 | r'<div id="viewvideo-title">([^<]+)</div>', webpage, 'title') |
703d78bb | 39 | title = title.replace('\n', '') |
d90b3854 P |
40 | |
41 | # get real url | |
1c222387 YCH |
42 | file_id = self._search_regex( |
43 | r'so.addVariable\(\'file\',\'(\d+)\'', webpage, 'file id') | |
44 | sec_code = self._search_regex( | |
45 | r'so.addVariable\(\'seccode\',\'([^\']+)\'', webpage, 'sec code') | |
46 | max_vid = self._search_regex( | |
47 | r'so.addVariable\(\'max_vid\',\'(\d+)\'', webpage, 'max vid') | |
d90b3854 | 48 | url_params = compat_urllib_parse.urlencode({ |
1c222387 | 49 | 'VID': file_id, |
d90b3854 | 50 | 'mp4': '1', |
1c222387 YCH |
51 | 'seccode': sec_code, |
52 | 'max_vid': max_vid, | |
d90b3854 | 53 | }) |
1c222387 YCH |
54 | info_cn = self._download_webpage( |
55 | 'http://91porn.com/getfile.php?' + url_params, video_id, | |
a2d97130 | 56 | 'get real video url') |
1c222387 | 57 | video_url = self._search_regex(r'file=([^&]+)&', info_cn, 'url') |
d90b3854 | 58 | |
a80601f8 YCH |
59 | duration = parse_duration(self._search_regex( |
60 | r'时长:\s*</span>\s*(\d+:\d+)', webpage, 'duration', fatal=False)) | |
61 | ||
62 | comment_count = int_or_none(self._search_regex( | |
63 | r'留言:\s*</span>\s*(\d+)', webpage, 'comment count', fatal=False)) | |
64 | ||
1c222387 | 65 | return { |
d90b3854 P |
66 | 'id': video_id, |
67 | 'title': title, | |
68 | 'url': video_url, | |
a80601f8 YCH |
69 | 'duration': duration, |
70 | 'comment_count': comment_count, | |
d90b3854 | 71 | } |