|
|
|
@ -1,10 +1,6 @@
|
|
|
|
|
# coding: utf-8
|
|
|
|
|
from __future__ import unicode_literals
|
|
|
|
|
|
|
|
|
|
from ..compat import (
|
|
|
|
|
compat_urllib_parse_unquote,
|
|
|
|
|
compat_urllib_parse_urlencode,
|
|
|
|
|
)
|
|
|
|
|
from .common import InfoExtractor
|
|
|
|
|
from ..utils import (
|
|
|
|
|
parse_duration,
|
|
|
|
@ -19,7 +15,7 @@ class Porn91IE(InfoExtractor):
|
|
|
|
|
|
|
|
|
|
_TEST = {
|
|
|
|
|
'url': 'http://91porn.com/view_video.php?viewkey=7e42283b4f5ab36da134',
|
|
|
|
|
'md5': '6df8f6d028bc8b14f5dbd73af742fb20',
|
|
|
|
|
'md5': '7fcdb5349354f40d41689bd0fa8db05a',
|
|
|
|
|
'info_dict': {
|
|
|
|
|
'id': '7e42283b4f5ab36da134',
|
|
|
|
|
'title': '18岁大一漂亮学妹,水嫩性感,再爽一次!',
|
|
|
|
@ -43,24 +39,7 @@ class Porn91IE(InfoExtractor):
|
|
|
|
|
r'<div id="viewvideo-title">([^<]+)</div>', webpage, 'title')
|
|
|
|
|
title = title.replace('\n', '')
|
|
|
|
|
|
|
|
|
|
# get real url
|
|
|
|
|
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': file_id,
|
|
|
|
|
'mp4': '1',
|
|
|
|
|
'seccode': sec_code,
|
|
|
|
|
'max_vid': max_vid,
|
|
|
|
|
})
|
|
|
|
|
info_cn = self._download_webpage(
|
|
|
|
|
'http://91porn.com/getfile.php?' + url_params, video_id,
|
|
|
|
|
'Downloading real video url')
|
|
|
|
|
video_url = compat_urllib_parse_unquote(self._search_regex(
|
|
|
|
|
r'file=([^&]+)&', info_cn, 'url'))
|
|
|
|
|
info_dict = self._parse_html5_media_entries(url, webpage, video_id)[0]
|
|
|
|
|
|
|
|
|
|
duration = parse_duration(self._search_regex(
|
|
|
|
|
r'时长:\s*</span>\s*(\d+:\d+)', webpage, 'duration', fatal=False))
|
|
|
|
@ -68,11 +47,12 @@ class Porn91IE(InfoExtractor):
|
|
|
|
|
comment_count = int_or_none(self._search_regex(
|
|
|
|
|
r'留言:\s*</span>\s*(\d+)', webpage, 'comment count', fatal=False))
|
|
|
|
|
|
|
|
|
|
return {
|
|
|
|
|
info_dict.update({
|
|
|
|
|
'id': video_id,
|
|
|
|
|
'title': title,
|
|
|
|
|
'url': video_url,
|
|
|
|
|
'duration': duration,
|
|
|
|
|
'comment_count': comment_count,
|
|
|
|
|
'age_limit': self._rta_search(webpage),
|
|
|
|
|
}
|
|
|
|
|
})
|
|
|
|
|
|
|
|
|
|
return info_dict
|
|
|
|
|