Browse Source

更新蜜源数据采集

master
huangjiajun 9 months ago
parent
commit
036a73a042
4 changed files with 79 additions and 1 deletions
  1. +1
    -0
      app/task/init.go
  2. +1
    -0
      app/task/md/cron_key.go
  3. +56
    -1
      app/task/svc/svc_data_collect_cron_miyuan_real_news.go
  4. +21
    -0
      app/task/task_data_collect_cron_miyuan_real_news1.go

+ 1
- 0
app/task/init.go View File

@@ -88,6 +88,7 @@ func initTasks() {
jobs[taskMd.DataCollectCronHaodankuRealNews10] = taskDataCollectCronHaodankuRealNews10 //

jobs[taskMd.DataCollectCronMiyuanRealNews] = taskDataCollectCronMiyuanRealNews //
jobs[taskMd.DataCollectCronMiyuanRealNews1] = taskDataCollectCronMiyuanRealNews1 //
jobs[taskMd.DataCollectCronMiyuanRealNews10] = taskDataCollectCronMiyuanRealNews10 //

}

+ 1
- 0
app/task/md/cron_key.go View File

@@ -5,5 +5,6 @@ const (
DataCollectCronHaodankuRealNews10 = "data_collect_cron_haodanku_real_news10"

DataCollectCronMiyuanRealNews = "data_collect_cron_miyuan_real_news"
DataCollectCronMiyuanRealNews1 = "data_collect_cron_miyuan_real_news1"
DataCollectCronMiyuanRealNews10 = "data_collect_cron_miyuan_real_news10"
)

+ 56
- 1
app/task/svc/svc_data_collect_cron_miyuan_real_news.go View File

@@ -15,7 +15,62 @@ import (
func DataCollectCronMiyuanRealNews(types int) {
key := md.ZhiosRealNewsData
//es.CreateIndexIfNotExists(key, md.ZhiosRealNewsDataField)
ids := []string{"MC", "TM", "JD", "PDD", "CD", "ZD", "MS", "QYH", "TM", "BJ", "HJ"}
ids := []string{"CD", "MS", "TM", "JD", "PDD"}
//ids = []string{"CD"}
num := 10
if types == 0 {
ids = []string{""}
num = 3
}
size := 10
now := time.Now()
fmt.Println("============", time.Since(now))
for _, v := range ids {
for i := 1; i <= num; i++ {
time.Sleep(time.Microsecond * 10) // 等待10毫秒
newsData, err := gzmiyuan.GetGzmiRealData(utils.IntToStr(size), utils.IntToStr(i), v)
utils.FilePutContents("my", utils.SerializeStr(newsData))
if err != nil {
utils.FilePutContents("my", err.Error())
}
if newsData == nil || err != nil {
continue
}
platformMap := map[string]string{"TM": "10010", "BJ": "10009", "HJ": "10008", "MC": "10001", "TB": "10003", "JD": "10004", "PDD": "10005", "CD": "10006", "ZD": "10002", "MS": "10007", "QYH": "10000"}
for _, item := range newsData {
var tmp = md2.RealNewsData{
Id: item.Id,
CateId: utils.SerializeStr([]string{platformMap[item.OpenType]}),
Images: utils.SerializeStr(strings.Split(item.XianbaoImageUrl, ",")),
Platform: "my",
StartTime: utils.TimeStdParseUnix(item.CreateTime),
OldContent: item.XianbaoContent,
}
tmp.Key = php2go.Md5("hdk_real_news_my_" + utils.IntToStr(item.Id))
var uniqueId = tmp.Key
doc, _ := es.FirstDoc(key, uniqueId)
if doc == nil {
goodsList := ToContent(item.XianbaoContent)

tmp.Content = utils.SerializeStr(goodsList)
fmt.Println(tmp)
createDocRet, err := es.CreateDoc(key, uniqueId, tmp)
if err != nil {
return
}
fmt.Printf("CreateDoc ==> %+v \n\n", createDocRet)
}

}
}

}
fmt.Println("==========================end", time.Since(now))
}
func DataCollectCronMiyuanRealNews1(types int) {
key := md.ZhiosRealNewsData
//es.CreateIndexIfNotExists(key, md.ZhiosRealNewsDataField)
ids := []string{"ZD", "MC", "QYH", "TM", "BJ", "HJ"}
//ids = []string{"CD"}
num := 10
if types == 0 {


+ 21
- 0
app/task/task_data_collect_cron_miyuan_real_news1.go View File

@@ -0,0 +1,21 @@
package task

import (
"applet/app/task/svc"
"math/rand"
"time"
)

func taskDataCollectCronMiyuanRealNews1() {
for {
if len(ch) > workerNum {
time.Sleep(time.Millisecond * time.Duration(rand.Intn(1000)))
} else {
goto START
}
}
START:
ch <- 1
svc.DataCollectCronMiyuanRealNews1(1)
<-ch
}

Loading…
Cancel
Save