|
@@ -27,12 +27,12 @@ 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
|
|
|
+ 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}`
|
|
|
Fields2 = `{"budget":1,"bidamount":1,"title":1,"projectname":1,"winner":1}`
|
|
|
)
|
|
@@ -319,8 +319,8 @@ func (e *ExtractTask) PreInfo(doc map[string]interface{}) (j, jf *ju.Job, isSite
|
|
|
BuyerAddr: qu.ObjToString(doc["buyeraddr"]),
|
|
|
RuleBlock: e.RuleBlock,
|
|
|
}
|
|
|
- if (j.Jsondata != nil||(*j.Jsondata) != nil) && (*j.Jsondata)["jsoncontent"]!= nil{
|
|
|
- delete((*j.Jsondata),"jsoncontent")
|
|
|
+ if (j.Jsondata != nil || (*j.Jsondata) != nil) && (*j.Jsondata)["jsoncontent"] != nil {
|
|
|
+ delete((*j.Jsondata), "jsoncontent")
|
|
|
}
|
|
|
if isextFile {
|
|
|
jf = &ju.Job{
|
|
@@ -339,8 +339,8 @@ func (e *ExtractTask) PreInfo(doc map[string]interface{}) (j, jf *ju.Job, isSite
|
|
|
RuleBlock: e.RuleBlock,
|
|
|
IsFile: isextFile,
|
|
|
}
|
|
|
- if (jf.Jsondata != nil||(*jf.Jsondata) != nil) && (*jf.Jsondata)["jsoncontent"]!= nil{
|
|
|
- delete((*jf.Jsondata),"jsoncontent")
|
|
|
+ if (jf.Jsondata != nil || (*jf.Jsondata) != nil) && (*jf.Jsondata)["jsoncontent"] != nil {
|
|
|
+ delete((*jf.Jsondata), "jsoncontent")
|
|
|
}
|
|
|
}
|
|
|
codeSite := j.SpiderCode
|
|
@@ -974,7 +974,7 @@ func ExtRuleCoreByPkgReg(j *ju.Job, in *RegLuaInfo, e *ExtractTask) {
|
|
|
cfn := e.ClearFn[in.Field]
|
|
|
lock.Unlock()
|
|
|
data := clear.DoClearFn(cfn, []interface{}{strings.TrimSpace(rep[in.Field+"_"+fmt.Sprint(i)]), j.Content})
|
|
|
- if data[len(data)-1].(bool){
|
|
|
+ if data[len(data)-1].(bool) {
|
|
|
j.BlockPackage[k].Budget = qu.Float64All(data[0])
|
|
|
j.BlockPackage[k].IsTrueBudget = true
|
|
|
}
|
|
@@ -984,7 +984,7 @@ func ExtRuleCoreByPkgReg(j *ju.Job, in *RegLuaInfo, e *ExtractTask) {
|
|
|
cfn := e.ClearFn[in.Field]
|
|
|
lock.Unlock()
|
|
|
data := clear.DoClearFn(cfn, []interface{}{strings.TrimSpace(rep[in.Field+"_"+fmt.Sprint(i)]), j.Content})
|
|
|
- if data[len(data)-1].(bool){
|
|
|
+ if data[len(data)-1].(bool) {
|
|
|
j.BlockPackage[k].Bidamount = qu.Float64All(data[0])
|
|
|
j.BlockPackage[k].IsTrueBidamount = true
|
|
|
}
|
|
@@ -1026,7 +1026,7 @@ func ExtRuleCoreByPkgReg(j *ju.Job, in *RegLuaInfo, e *ExtractTask) {
|
|
|
cfn := e.ClearFn[in.Field]
|
|
|
lock.Unlock()
|
|
|
data := clear.DoClearFn(cfn, []interface{}{val, j.Content})
|
|
|
- if data[len(data)-1].(bool){
|
|
|
+ if data[len(data)-1].(bool) {
|
|
|
j.BlockPackage[k].Budget = qu.Float64All(data[0])
|
|
|
j.BlockPackage[k].IsTrueBudget = true
|
|
|
}
|
|
@@ -1037,7 +1037,7 @@ func ExtRuleCoreByPkgReg(j *ju.Job, in *RegLuaInfo, e *ExtractTask) {
|
|
|
cfn := e.ClearFn[in.Field]
|
|
|
lock.Unlock()
|
|
|
data := clear.DoClearFn(cfn, []interface{}{val, j.Content})
|
|
|
- if data[len(data)-1].(bool){
|
|
|
+ if data[len(data)-1].(bool) {
|
|
|
j.BlockPackage[k].Bidamount = qu.Float64All(data[0])
|
|
|
j.BlockPackage[k].IsTrueBidamount = true
|
|
|
}
|
|
@@ -1071,7 +1071,7 @@ func getKvByLuaFields(vc *RuleCore, j *ju.Job, et *ExtractTask) (map[string][]ma
|
|
|
kvmap[vc.Field] = append(kvmap[vc.Field], map[string]interface{}{
|
|
|
"code": "winnerorder",
|
|
|
"field": vc.Field,
|
|
|
- "ruletext": "中标候选人_"+ v["sortstr"].(string),
|
|
|
+ "ruletext": "中标候选人_" + v["sortstr"].(string),
|
|
|
"extfrom": v["sortstr"],
|
|
|
"sourcevalue": v["price"],
|
|
|
"value": v["price"],
|
|
@@ -1596,7 +1596,8 @@ type FieldValue struct {
|
|
|
Value interface{}
|
|
|
Count int
|
|
|
}
|
|
|
-var clearWinnerReg =regexp.MustCompile("名称|施工|拟定供应商名称|:|:")
|
|
|
+
|
|
|
+var clearWinnerReg = regexp.MustCompile("名称|施工|拟定供应商名称|:|:")
|
|
|
|
|
|
//分析抽取结果并保存
|
|
|
func AnalysisSaveResult(j, jf *ju.Job, e *ExtractTask) {
|
|
@@ -1630,20 +1631,27 @@ func AnalysisSaveResult(j, jf *ju.Job, e *ExtractTask) {
|
|
|
}
|
|
|
}
|
|
|
}
|
|
|
+ if len(j.PackageInfo) > 10 {
|
|
|
+ for k, v := range j.PackageInfo {
|
|
|
+ j.PackageInfo = map[string]map[string]interface{}{}
|
|
|
+ j.PackageInfo[k] = v
|
|
|
+ break
|
|
|
+ }
|
|
|
+ }
|
|
|
if len(j.PackageInfo) > 0 { //分包信息
|
|
|
tmp["package"] = j.PackageInfo
|
|
|
//包预算,中标金额合并大于抽取就覆盖
|
|
|
var tmpBidamount, tmpBudget float64
|
|
|
//s_winner逗号分隔拼接,分包中标人
|
|
|
- var tmpstr,savewinner []string
|
|
|
+ var tmpstr, savewinner []string
|
|
|
//按包排序
|
|
|
for b, v := range j.PackageInfo {
|
|
|
- if v["winner"]!= nil && v["winner"]!=""{
|
|
|
- tmpstr = append(tmpstr,b)
|
|
|
+ if v["winner"] != nil && v["winner"] != "" {
|
|
|
+ tmpstr = append(tmpstr, b)
|
|
|
}
|
|
|
}
|
|
|
//包预算,中标金额合并大于抽取就覆盖
|
|
|
- if len(j.PackageInfo) >1{
|
|
|
+ if len(j.PackageInfo) > 1 {
|
|
|
//包数大于1累加
|
|
|
for _, v := range j.PackageInfo {
|
|
|
if v["budget"] != nil {
|
|
@@ -1659,10 +1667,10 @@ func AnalysisSaveResult(j, jf *ju.Job, e *ExtractTask) {
|
|
|
if qu.Float64All(tmp["bidamount"]) < tmpBidamount {
|
|
|
tmp["bidamount"] = tmpBidamount
|
|
|
}
|
|
|
- }else {
|
|
|
+ } else {
|
|
|
//包数等于1,tmp没有值取包里的值
|
|
|
if tmp["budget"] == nil || tmp["budget"] == 0 {
|
|
|
- for _,v := range j.PackageInfo {
|
|
|
+ for _, v := range j.PackageInfo {
|
|
|
if v["budget"] != nil {
|
|
|
tmp["budget"] = v["budget"]
|
|
|
}
|
|
@@ -1670,7 +1678,7 @@ func AnalysisSaveResult(j, jf *ju.Job, e *ExtractTask) {
|
|
|
|
|
|
}
|
|
|
if tmp["bidamount"] == nil || tmp["bidamount"] == 0 {
|
|
|
- for _,v := range j.PackageInfo {
|
|
|
+ for _, v := range j.PackageInfo {
|
|
|
if v["bidamount"] != nil {
|
|
|
tmp["bidamount"] = v["bidamount"]
|
|
|
}
|
|
@@ -1679,21 +1687,21 @@ func AnalysisSaveResult(j, jf *ju.Job, e *ExtractTask) {
|
|
|
}
|
|
|
//s_winner逗号分隔拼接,分包中标人
|
|
|
sort.Strings(tmpstr)
|
|
|
- for _,v := range tmpstr{
|
|
|
+ for _, v := range tmpstr {
|
|
|
svvvv := qu.ObjToString(j.PackageInfo[v]["winner"])
|
|
|
savevvv := clearWinnerReg.ReplaceAllString(svvvv, "")
|
|
|
- if savevvv == ""{
|
|
|
+ if savevvv == "" {
|
|
|
continue
|
|
|
}
|
|
|
- savewinner = append(savewinner,savevvv)
|
|
|
+ savewinner = append(savewinner, savevvv)
|
|
|
}
|
|
|
- if (savewinner == nil || len(savewinner)==0) && tmp["winner"]!=nil{
|
|
|
+ if (savewinner == nil || len(savewinner) == 0) && tmp["winner"] != nil {
|
|
|
tmp["s_winner"] = tmp["winner"]
|
|
|
- }else if savewinner != nil{
|
|
|
- tmp["s_winner"] = strings.Join(savewinner,",")
|
|
|
+ } else if savewinner != nil {
|
|
|
+ tmp["s_winner"] = strings.Join(savewinner, ",")
|
|
|
}
|
|
|
|
|
|
- }else if tmp["winner"]!= nil && tmp["winner"]!=""{
|
|
|
+ } else if tmp["winner"] != nil && tmp["winner"] != "" {
|
|
|
//没有分包取winner
|
|
|
tmp["s_winner"] = tmp["winner"]
|
|
|
}
|
|
@@ -1853,8 +1861,12 @@ 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
|
|
|
- bs, _ := json.Marshal(j.BlockPackage)
|
|
|
- tmp["epackage"] = string(bs)
|
|
|
+ if len(j.BlockPackage) >10{
|
|
|
+ tmp["epackage"] = "分包异常"
|
|
|
+ }else {
|
|
|
+ bs, _ := json.Marshal(j.BlockPackage)
|
|
|
+ tmp["epackage"] = string(bs)
|
|
|
+ }
|
|
|
}
|
|
|
tmp["result"] = result
|
|
|
tmp["resultf"] = resultf
|
|
@@ -2046,7 +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
|
|
@@ -2138,20 +2150,20 @@ func resetWinnerorder(j *ju.Job) {
|
|
|
if maxlen > 0 {
|
|
|
winners = append(winners, &ju.ExtField{Code: "winnerorder", Field: "winner", ExtFrom: "j.Winnerorder", Value: j.Winnerorder[0]["entname"], Score: 0.5})
|
|
|
if j.Winnerorder[0]["price"] != nil {
|
|
|
- tmpPrice := clear.ObjToMoney([]interface{}{j.Winnerorder[0]["price"],""})
|
|
|
- if tmpPrice[len(tmpPrice)-1].(bool){
|
|
|
- bidamounts = append(bidamounts, &ju.ExtField{Code: "winnerorder", Field: "bidamount", ExtFrom: "j.Winnerorder",SourceValue:j.Winnerorder[0]["price"], Value: tmpPrice[0], Score: 0.5})
|
|
|
+ tmpPrice := clear.ObjToMoney([]interface{}{j.Winnerorder[0]["price"], ""})
|
|
|
+ if tmpPrice[len(tmpPrice)-1].(bool) {
|
|
|
+ bidamounts = append(bidamounts, &ju.ExtField{Code: "winnerorder", Field: "bidamount", ExtFrom: "j.Winnerorder", SourceValue: j.Winnerorder[0]["price"], Value: tmpPrice[0], Score: 0.5})
|
|
|
}
|
|
|
}
|
|
|
}
|
|
|
if j.Result["winner"] == nil && len(winners) > 0 {
|
|
|
j.Result["winner"] = winners
|
|
|
- } else {
|
|
|
+ } else if len(winners) > 0 {
|
|
|
j.Result["winner"] = append(j.Result["winner"], winners...)
|
|
|
}
|
|
|
if j.Result["bidamount"] == nil && len(bidamounts) > 0 {
|
|
|
j.Result["bidamount"] = bidamounts
|
|
|
- } else {
|
|
|
+ } else if len(bidamounts) > 0 {
|
|
|
j.Result["bidamount"] = append(j.Result["bidamount"], bidamounts...)
|
|
|
}
|
|
|
//j.Result["winner"] = winners
|