Преглед на файлове

新增数据拦截通用方法downloadByDataIntercept

maxiaoshan преди 1 година
родител
ревизия
941304109e
променени са 2 файла, в които са добавени 12 реда и са изтрити 0 реда
  1. 1 0
      src/config.json
  2. 11 0
      src/spider/script.go

+ 1 - 0
src/config.json

@@ -46,6 +46,7 @@
     "renderaddr": "http://splash.spdata.jianyu360.com/render.json",
     "proxyaddr": "http://cc.spdata.jianyu360.com/crawl/proxy/socks5/fetch",
     "proxyauthor": "Basic amlhbnl1MDAxOjEyM3F3ZSFB",
+    "pwaddr": "http://172.17.145.179:8997/web/intercept/request?url=%s",
     "es": {
         "addr": "http://192.168.3.206:9800",
         "pool": 15,

+ 11 - 0
src/spider/script.go

@@ -1260,6 +1260,17 @@ func (s *Script) LoadScript(site, channel, user *string, code, script_file strin
 		}
 		return 1
 	}))
+	//针对中国招标投标公共服务平台三级页瑞数加密下载方法
+	s.L.SetGlobal("downloadByDataIntercept", s.L.NewFunction(func(S *lua.LState) int {
+		url := S.ToString(-4)
+		url_regex := S.ToString(-3)
+		timeout := S.ToInt(-2)
+		proxy := S.ToBool(-1)
+		headers := util.DownloadByDataIntercept(url, url_regex, timeout, proxy)
+		table := util.MapToLuaTable(S, headers)
+		S.Push(table)
+		return 1
+	}))
 	return ""
 }
 func dealHref(pageListUrl, href string) string {