404-intelligent-service/config.py

#!/usr/bin/env python3
# -*- encoding: utf-8 -*-

# URL与URL描述的对应关系. 这些关系从网站的HTML 的 <a>标签中提取得到

mirror_site_folder = "/data/pub/websites"

ignored_title = {"(更多…)", "trackback", "取消回复", "RSS 2.0", "Prev", "Next", }

domain_mapping = {
"www.taobaodba.com" : "http://taobaodba.ruoguschool.com",
"www.searchtb.com" : "http://searchtb.ruoguschool.com",
"www.tbdata.org" : "http://tbdata.ruoguschool.com",
"www.taobaotest.com" : "http://taobaotest.ruoguschool.com",
"csrd.aliapp.com" : "http://csrd.ruoguschool.com",
"www.alidata.org" : "http://alidata.ruoguschool.com/archives",
"blog.linezing.com" : "http://linezing.ruoguschool.com",
"techshow.ctrip.com" : "http://techshow.ruoguschool.com",
"rdc.taobao.com": "http://rdc.ruoguschool.com",
"lamp.baidu.com": "http://lamp.ruoguschool.com",
}