瀏覽代碼

新增字段infoformat

zhangjinkun 5 年之前
父節點
當前提交
b490ca6987
共有 1 個文件被更改,包括 10 次插入10 次删除
  1. 10 10
      src/jy/extract/extract.go

+ 10 - 10
src/jy/extract/extract.go

@@ -27,13 +27,13 @@ import (
 var (
 	lock, lockrule, lockclear, locktag, blocktag sync.RWMutex
 
-	cut     = ju.NewCut()                          //获取正文并清理
-	ExtLogs map[*TaskInfo][]map[string]interface{} //抽取日志
-	TaskList      map[string]*ExtractTask          //任务列表
-	ClearTaskList map[string]*ClearTask            //清理任务列表
-	saveLimit     = 100                            //抽取日志批量保存
-	PageSize      = 5000                           //查询分页
-	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}`
+	cut           = ju.NewCut()                          //获取正文并清理
+	ExtLogs       map[*TaskInfo][]map[string]interface{} //抽取日志
+	TaskList      map[string]*ExtractTask                //任务列表
+	ClearTaskList map[string]*ClearTask                  //清理任务列表
+	saveLimit     = 100                                  //抽取日志批量保存
+	PageSize      = 5000                                 //查询分页
+	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}`
 	Fields2       = `{"budget":1,"bidamount":1,"title":1,"projectname":1,"winner":1}`
 )
 
@@ -1861,9 +1861,9 @@ func AnalysisSaveResult(j, jf *ju.Job, e *ExtractTask) {
 		} else { //测试结果
 			delete(tmp, "_id")
 			if len(j.BlockPackage) > 0 { //分包详情
-				if len(j.BlockPackage) >10{
+				if len(j.BlockPackage) > 10 {
 					tmp["epackage"] = "分包异常"
-				}else {
+				} else {
 					bs, _ := json.Marshal(j.BlockPackage)
 					tmp["epackage"] = string(bs)
 				}
@@ -2058,7 +2058,7 @@ func (e *ExtractTask) QualityAudit(resulttmp map[string]interface{}) {
 func (e *ExtractTask) RedisMatch(field, fv string, val map[string]interface{}) {
 	defer qu.Catch()
 	i := redis.GetInt(field, field+"_"+fv) //查找redis
-	if i == 0 { //reids未找到,执行规则匹配
+	if i == 0 {                            //reids未找到,执行规则匹配
 		val[field+"_isredis"] = false
 		e.RuleMatch(field, fv, val) //规则匹配
 	} else { //redis找到,打标识存库