|
@@ -26,12 +26,12 @@ 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}`
|
|
|
)
|
|
@@ -568,9 +568,8 @@ func (e *ExtractTask) ExtractDetail(j *ju.Job, isSite bool, codeSite string) {
|
|
|
}
|
|
|
}
|
|
|
//函数清理
|
|
|
-
|
|
|
for key, val := range j.Result {
|
|
|
- for _, v := range val {
|
|
|
+ for i, v := range val {
|
|
|
//qu.Debug(key, v.Value)
|
|
|
lockclear.Lock()
|
|
|
var cfn = []string{}
|
|
@@ -584,6 +583,13 @@ func (e *ExtractTask) ExtractDetail(j *ju.Job, isSite bool, codeSite string) {
|
|
|
continue
|
|
|
}
|
|
|
data := clear.DoClearFn(cfn, []interface{}{v.Value, j.Content})
|
|
|
+ if key == "budget" || key == "bidamount" {
|
|
|
+ if istrue, ok := data[len(data)-1].(bool); istrue && ok {
|
|
|
+ j.Result[key][i].IsTrue = true
|
|
|
+ } else {
|
|
|
+ continue
|
|
|
+ }
|
|
|
+ }
|
|
|
before, _ := v.Value.(string)
|
|
|
v.Value = data[0]
|
|
|
BeforeAddClearFnLog(v.Type, "函数清理", j.SourceMid, before, v.MatchType, v, e)
|
|
@@ -1576,6 +1582,11 @@ func AnalysisSaveResult(j, jf *ju.Job, e *ExtractTask) {
|
|
|
} else if v.Field == "projectname" {
|
|
|
tmp[v.Field] = v.Value
|
|
|
break
|
|
|
+ } else if v.Field == "bidamount"||v.Field =="budget"{
|
|
|
+ if v.IsTrue{
|
|
|
+ tmp[v.Field] =v.Value
|
|
|
+ break
|
|
|
+ }
|
|
|
}
|
|
|
}
|
|
|
}
|
|
@@ -1691,19 +1702,19 @@ func AnalysisSaveResult(j, jf *ju.Job, e *ExtractTask) {
|
|
|
//tmp["extract_content"] = j.Content
|
|
|
if e.TaskInfo.TestColl == "" {
|
|
|
if len(tmp) > 0 { //保存抽取结果
|
|
|
- /* if len(e.SiteFields) <= 0 {
|
|
|
- //for field, _ := range e.Fields {
|
|
|
- // if tmp[field] == nil && {
|
|
|
- // tmp[field] = "" //覆盖之前版本数据
|
|
|
- // }
|
|
|
- //}
|
|
|
- } else {
|
|
|
- //for field, _ := range e.SiteFields {
|
|
|
- // if tmp[field] == nil &&{
|
|
|
- // tmp[field] = "" //覆盖之前版本数据
|
|
|
- // }
|
|
|
- //}
|
|
|
- }*/
|
|
|
+ /* if len(e.SiteFields) <= 0 {
|
|
|
+ //for field, _ := range e.Fields {
|
|
|
+ // if tmp[field] == nil && {
|
|
|
+ // tmp[field] = "" //覆盖之前版本数据
|
|
|
+ // }
|
|
|
+ //}
|
|
|
+ } else {
|
|
|
+ //for field, _ := range e.SiteFields {
|
|
|
+ // if tmp[field] == nil &&{
|
|
|
+ // tmp[field] = "" //覆盖之前版本数据
|
|
|
+ // }
|
|
|
+ //}
|
|
|
+ }*/
|
|
|
tmp["repeat"] = 0
|
|
|
tmparr := []map[string]interface{}{
|
|
|
map[string]interface{}{
|
|
@@ -1877,6 +1888,11 @@ func auxInfo(j *ju.Job) map[string][]map[string]interface{} {
|
|
|
standardized = true
|
|
|
}
|
|
|
}
|
|
|
+ if field == "budget"||field == "bidamount"{
|
|
|
+ if !v.IsTrue{
|
|
|
+ continue
|
|
|
+ }
|
|
|
+ }
|
|
|
sfield := map[string]interface{}{
|
|
|
"val": v.Value,
|
|
|
"type": v.Type,
|
|
@@ -1929,7 +1945,7 @@ 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
|