[porn91] Simplify
[youtube-dl] / youtube_dl / extractor / porn91.py
1 # encoding: utf-8
2 from __future__ import unicode_literals
3
4 from ..compat import compat_urllib_parse
5 from .common import InfoExtractor
6
7
8 class Porn91IE(InfoExtractor):
9     IE_NAME = '91porn'
10     _VALID_URL = r'(?:https?://)(?:www\.|)91porn\.com/.+?\?viewkey=(?P<id>[\w\d]+)'
11
12     _TEST = {
13         'url': 'http://91porn.com/view_video.php?viewkey=7e42283b4f5ab36da134',
14         'md5': '6df8f6d028bc8b14f5dbd73af742fb20',
15         'info_dict': {
16             'id': '7e42283b4f5ab36da134',
17             'title': '18岁大一漂亮学妹,水嫩性感,再爽一次!',
18             'ext': 'mp4'
19         }
20     }
21
22     def _real_extract(self, url):
23         video_id = self._match_id(url)
24         url = 'http://91porn.com/view_video.php?viewkey=%s' % video_id
25         self._set_cookie('91porn.com', 'language', 'cn_CN')
26         webpage = self._download_webpage(url, video_id, "get HTML content")
27         title = self._search_regex(
28             r'<div id="viewvideo-title">([^<]+)</div>', webpage, 'title')
29         title = title.replace('\n', '')
30
31         # get real url
32         file_id = self._search_regex(
33             r'so.addVariable\(\'file\',\'(\d+)\'', webpage, 'file id')
34         sec_code = self._search_regex(
35             r'so.addVariable\(\'seccode\',\'([^\']+)\'', webpage, 'sec code')
36         max_vid = self._search_regex(
37             r'so.addVariable\(\'max_vid\',\'(\d+)\'', webpage, 'max vid')
38         url_params = compat_urllib_parse.urlencode({
39             'VID': file_id,
40             'mp4': '1',
41             'seccode': sec_code,
42             'max_vid': max_vid,
43         })
44         info_cn = self._download_webpage(
45             'http://91porn.com/getfile.php?' + url_params, video_id,
46             "get real video url")
47         video_url = self._search_regex(r'file=([^&]+)&', info_cn, 'url')
48
49         return {
50             'id': video_id,
51             'title': title,
52             'url': video_url,
53         }