fengweiqiang 4 vuotta sitten
vanhempi
commit
bc2cbd6478
2 muutettua tiedostoa jossa 30 lisäystä ja 3 poistoa
  1. 28 1
      src/jy/extract/extract.go
  2. 2 2
      src/jy/extract/extractudp.go

+ 28 - 1
src/jy/extract/extract.go

@@ -34,6 +34,33 @@ var (
 	Fields        = `{"title":1,"summary":1,"detail":1,"contenthtml":1,"site":1,"spidercode":1,"toptype":1,"subtype":1,"bidstatus":1,"area":1,"city":1,"comeintime":1,"publishtime":1,"sensitive":1,"projectinfo":1,"jsondata":1,"href":1,"infoformat":1,"attach_text":1,"dataging":1,"review_experts":1,"purchasing":1}`
 	//Fields        = `{"title":1,"summary":1,"detail":1,"contenthtml":1,"site":1,"spidercode":1,"toptype":1,"subtype":1,"bidstatus":1,"area":1,"city":1,"comeintime":1,"publishtime":1,"sensitive":1,"projectinfo":1,"jsondata":1,"href":1,"infoformat":1,"attach_text":1,"dataging":1,"new_attach_text":1,"createtime":1,"currency":1,"id":1,"company_email":1,"buyerclass":1,"tagname":1,"company_phone":1,"appid":1,"industry":1,"projectscope":1,"item":1,"s_subscopeclass":1,"matchkey":1,"jybxhref":1,"legal_person":1,"matchtype":1,"review_experts":1,"purchasing":1}`
 	Fields2 = `{"budget":1,"bidamount":1,"title":1,"projectname":1,"winner":1}`
+	spidercode = map[string]bool{
+		"gd_zhsggzyjyzx_jsgc_fjczbgg":true,
+		"js_szgyyqggzyjyzx_jsgc_zjfbgs":true,
+		"zj_tzsyhggzyjyzx_jsgc_kbqk":true,
+		"hb_tmsggzyjyxxw_jsgc_kbqk":true,
+		"zj_nbsyyggzyjyw_jsgc_kbqk":true,
+		"zj_zjsggzyjyzx_jyxx_kbjg":true,
+		"zj_zjzdgcjyw_ztbjglxx_kbjg":true,
+		"zj_lssggzyjyw_jsgc_kbsk":true,
+		"zj_qzslyxggzyjyzx_gggs_xkbjl":true,
+		"sc_mssggzydzjypt_jsgc_kbjl":true,
+		"sc_pzhsggzyjyfwzx_jsgc_kbylb":true,
+		"a_zgzbtbggfwpt_wasjgf_ss_kbjl":true,
+		"a_hbszbtbggfwpt_kbjl":true,
+		"a_szsjsgcjyfwzxbafzx_kbqkgs":true,
+		"a_szldzbyxgs_kbxx":true,
+		"zj_zssssxggzyjyw_gcjs_kbjggs":true,
+		"gd_szszfhjsj_kbqkgs":true,
+		"a_gjggzyjypt_gcjs_kbjl":true,
+		"a_gjggzyjypt_gcjs_kbjl_new":true,
+		"zj_tzsyhggzyjyzx_kbjggg":true,
+		"a_zgzbtbggfwpy_wasjgf_kbjl_lsbl":true,
+		"ah_czsggzyjyw_jsgc_kbjl":true,
+		"ah_czsggzyjyw_zfcg_kbxx":true,
+		"ah_whsggzyjyfww_kbxx_cgxm":true,
+		"ah_whsggzyjyfww_kbxx_gcxm":true,
+	}
 )
 
 //启动测试抽取
@@ -103,7 +130,7 @@ func RunExtractTestTask(ext *ExtractTask, startId, num string) bool {
 			//if qu.ObjToString(v["sensitive"]) != ""||ggtest.MatchString(qu.ObjToString(v[""])) { //去除含敏感词数据
 			//	continue
 			//}
-			if qu.ObjToString(v["spidercode"]) == "a_gjggzyjypt_gcjs_kbjl" || "a_hbszbtbggfwpt_kbjl" == qu.ObjToString(v["spidercode"]) { //临时开标记录
+			if spidercode[qu.ObjToString(v["spidercode"])] { //临时开标记录
 				continue
 			}
 			var j, jf *ju.Job

+ 2 - 2
src/jy/extract/extractudp.go

@@ -162,7 +162,7 @@ func ExtractByUdp(sid, eid string, ra *net.UDPAddr, instanceId ...string) {
 				//	log.Debug(index, qu.BsonIdToSId(v["_id"]), "//去除含敏感词数据")
 				//	continue
 				//}
-				if qu.ObjToString(v["spidercode"]) == "a_gjggzyjypt_gcjs_kbjl" || "a_hbszbtbggfwpt_kbjl" == qu.ObjToString(v["spidercode"]) { //临时开标记录
+				if spidercode[qu.ObjToString(v["spidercode"])] { //临时开标记录
 					log.Debug(index, qu.BsonIdToSId(v["_id"]), "//开标记录")
 					continue
 				}
@@ -231,7 +231,7 @@ func ExtractByUdp(sid, eid string, ra *net.UDPAddr, instanceId ...string) {
 				//	log.Debug(index, qu.BsonIdToSId(v["_id"]), "//去除含敏感词数据")
 				//	continue
 				//}
-				if qu.ObjToString(v["spidercode"]) == "a_gjggzyjypt_gcjs_kbjl"  || "a_hbszbtbggfwpt_kbjl" == qu.ObjToString(v["spidercode"]) { //临时开标记录
+				if spidercode[qu.ObjToString(v["spidercode"])] { //临时开标记录
 					log.Debug(index, qu.BsonIdToSId(v["_id"]), "//开标记录")
 					continue
 				}