X-Git-Url: http://git.cielonegro.org/gitweb.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fporn91.py;h=3e15533e9d445e96884974af398e0461438bcc8b;hb=a13d06de420f6968425d48030c37e1150ff9ed6a;hp=cdf308f3d7776a5a1366a765cdd170a210d2cb97;hpb=9ff811c5cddbf3481fdcd44e97cf3683a925b33f;p=youtube-dl.git
diff --git a/youtube_dl/extractor/porn91.py b/youtube_dl/extractor/porn91.py
index cdf308f3d..3e15533e9 100644
--- a/youtube_dl/extractor/porn91.py
+++ b/youtube_dl/extractor/porn91.py
@@ -1,10 +1,13 @@
# encoding: utf-8
from __future__ import unicode_literals
-import re
-
from ..compat import compat_urllib_parse
from .common import InfoExtractor
+from ..utils import (
+ parse_duration,
+ int_or_none,
+ ExtractorError,
+)
class Porn91IE(InfoExtractor):
@@ -17,43 +20,54 @@ class Porn91IE(InfoExtractor):
'info_dict': {
'id': '7e42283b4f5ab36da134',
'title': '18å²å¤§ä¸æ¼äº®å¦å¦¹ï¼æ°´å«©æ§æï¼åç½ä¸æ¬¡ï¼',
- 'ext': 'mp4'
+ 'ext': 'mp4',
+ 'duration': 431,
+ 'age_limit': 18,
}
}
def _real_extract(self, url):
- mobj = re.match(self._VALID_URL, url)
- video_id = mobj.group('id')
+ video_id = self._match_id(url)
url = 'http://91porn.com/view_video.php?viewkey=%s' % video_id
self._set_cookie('91porn.com', 'language', 'cn_CN')
- webpage = self._download_webpage(url, video_id, "get HTML content")
+ webpage = self._download_webpage(url, video_id, 'get HTML content')
+
+ if 'ä½ä¸ºæ¸¸å®¢ï¼ä½ æ¯å¤©åªå¯è§ç10个è§é¢' in webpage:
+ raise ExtractorError('91 Porn says: Daily limit 10 videos exceeded', expected=True)
+
title = self._search_regex(
- r'
(?P
.+?)',
- webpage, 'title', flags=re.DOTALL)
+ r'([^<]+)
', webpage, 'title')
title = title.replace('\n', '')
# get real url
- n1 = self._search_regex(
- r'so.addVariable\(\'file\',\'(?P\d+)\'', webpage, 'n1')
- n2 = self._search_regex(
- r'so.addVariable\(\'seccode\',\'(?P.+?)\'', webpage, 'n2')
- n3 = self._search_regex(
- r'so.addVariable\(\'max_vid\',\'(?P\d+)\'', webpage, 'n3')
+ file_id = self._search_regex(
+ r'so.addVariable\(\'file\',\'(\d+)\'', webpage, 'file id')
+ sec_code = self._search_regex(
+ r'so.addVariable\(\'seccode\',\'([^\']+)\'', webpage, 'sec code')
+ max_vid = self._search_regex(
+ r'so.addVariable\(\'max_vid\',\'(\d+)\'', webpage, 'max vid')
url_params = compat_urllib_parse.urlencode({
- 'VID': n1,
+ 'VID': file_id,
'mp4': '1',
- 'seccode': n2,
- 'max_vid': n3,
+ 'seccode': sec_code,
+ 'max_vid': max_vid,
})
- t_url = 'http://91porn.com/getfile.php?' + url_params
- info_cn = self._download_webpage(t_url, video_id, "get real video_url")
- video_url = self._search_regex(r'file=(?Phttp.+?)&', info_cn, 'url')
+ info_cn = self._download_webpage(
+ 'http://91porn.com/getfile.php?' + url_params, video_id,
+ 'get real video url')
+ video_url = self._search_regex(r'file=([^&]+)&', info_cn, 'url')
- # construct info
- info = {
+ duration = parse_duration(self._search_regex(
+ r'æ¶é¿:\s*\s*(\d+:\d+)', webpage, 'duration', fatal=False))
+
+ comment_count = int_or_none(self._search_regex(
+ r'çè¨:\s*\s*(\d+)', webpage, 'comment count', fatal=False))
+
+ return {
'id': video_id,
'title': title,
'url': video_url,
+ 'duration': duration,
+ 'comment_count': comment_count,
+ 'age_limit': self._rta_search(webpage),
}
-
- return info