有内存不足问题，暂时提交，试用旧版本

main/suggest-task-dependency.go

 package main
 
 import (
+						"net/http"
+	"time"
+	"encoding/json"
+	"bytes"
+	"io/ioutil"
+	"container/list"
 	"strings"
 	"github.com/mozillazg/go-pinyin"
-	"fmt"
 	"strconv"
-	)
+	"fmt"
+)
 
 type ENV struct  {
 	DataWareDB string
@@ -17,6 +23,12 @@ type ENV struct  {
 	SensitiveFolder string
 }
 
+type Message struct {
+   Title string
+   Phones *list.List
+   Body *list.List
+}
+
 var test_env  = &ENV{
 	DataWareDB: "DataWarehouse_test:FihdZW7o1XKtDETZexOG@tcp(test01-secooDataWarehouse.master.com:3306)/secooDataWarehouse",
 	ErpDB: "3306_test:iS6CXpYqgZ8Mhjui@tcp(10.4.3.223:3306)/secooErpDB",
@@ -79,7 +91,7 @@ func convertToPinyin(str string) string {
 func cleanKeyword(keyword string) string {
 	out, err := t2s.Convert(keyword)
 	if err != nil { fmt.Println(err)  }
-	keyword = strings.ToLower(strings.Trim(DBC2SBC(strings.TrimSpace(out)),"\ufffc|,"))
+	keyword = strings.ToLower(strings.Trim(DBC2SBC(strings.TrimSpace(out)),"\ufffc|,|."))
 	return strings.Join(strings.Fields(keyword)," ")
 }
 
@@ -112,3 +124,41 @@ func DBC2SBC(s string) string {
 	}
 	return strings.Join(strLst, "")
 }
+
+// 发送POST请求
+// url：         请求地址
+// data：        POST请求提交的数据
+// contentType： 请求体格式，如：application/json
+func Post(url string, data interface{}, contentType string) string {
+
+	// 超时时间：5秒
+	client := &http.Client{Timeout: 5 * time.Second}
+	jsonStr, _ := json.Marshal(data)
+	resp, err := client.Post(url, contentType, bytes.NewBuffer(jsonStr))
+	if err != nil {
+		panic(err)
+	}
+	defer resp.Body.Close()
+
+	result, _ := ioutil.ReadAll(resp.Body)
+	return string(result)
+}
+
+
+func sendSuggestNotify() {
+
+	msg := Message{
+		Title:"提示词数据异常",
+	}
+
+	body := list.New()
+	body.PushBack("提示词数据太少")
+	msg.Body = body
+
+	phones := list.New()
+	phones.PushBack("17621863255,13894895183")
+	msg.Phones = phones
+
+	Post("http://matrix-inform.secoolocal.com/user/sendToUser", msg,"application/json")
+}
+

main/suggest-task.go

@@ -19,7 +19,8 @@ import (
 	"unicode"
 	"unicode/utf8"
 	"crypto/md5"
-)
+	)
+
 
 type Word struct {
 	Keyword        string  `json:"keyword"`
@@ -53,6 +54,7 @@ type Word struct {
 	MonthProductClickUv int32 `json:"-"`
 	MonthAddCartUv int32 `json:"-"`
 	SuggestTags string `json:"suggestTags"`
+	UpdateTime int64 `json:"updateTime"`
 }
 
 var wordMap sync.Map
@@ -70,16 +72,21 @@ var prefixFilterArr = []string{"https://", "http://", "dg", "d & g", "dolce&gabb
 
 const TABLE_SPLIT_STEP_SIZE = 10000
 
+const LEVEL_SIZE = 1
+
 const MAX_TAG_SIZE = 5
 
+var UPDATE_TIME = time.Now().UnixNano() / 1e6
+
 func main() {
 	startTime := time.Now()
 
+
 	datawareDB, err := sql.Open("mysql", RUN_ENV.DataWareDB)
 	if err != nil { log.Print(err.Error()) }
 	datawareDB.SetConnMaxLifetime(10*time.Minute)
-	datawareDB.SetMaxOpenConns(50)
-	datawareDB.SetMaxIdleConns(50)
+	datawareDB.SetMaxOpenConns(350)
+	datawareDB.SetMaxIdleConns(100)
 
 	var client *elastic.Client
 	if RUN_ENV.EsUser != "" {
@@ -91,7 +98,7 @@ func main() {
 
 	bulkProcessor, err := elastic.NewBulkProcessorService(client).
 		Workers(50).
-		BulkActions(5000).
+		BulkActions(2000).
 		FlushInterval(500*time.Millisecond).
 		After(after).
 		Do(context.Background())
@@ -107,13 +114,25 @@ func main() {
 	count := arr[1] / TABLE_SPLIT_STEP_SIZE
 	log.Printf("maxId/10000=%d\n", count)
 
-	for i := 0; i <= count; i++ {
+	if arr[1] < 2800000 {
+		log.Printf("data is too little ,return")
+		sendSuggestNotify()
+		return
+	}
+
+	for i:= 0; i < count; i = i + LEVEL_SIZE  {
+		for j := 0; j < LEVEL_SIZE; j++ {
+			log.Printf("add to wait %d", i+j)
 			wg.Add(1)
-		go queryIndex(i * TABLE_SPLIT_STEP_SIZE, datawareDB, bulkProcessor, &wg)
+			go queryIndex( (i + j ) *TABLE_SPLIT_STEP_SIZE, datawareDB, bulkProcessor, &wg)
+		}
+		wg.Wait()
+		log.Println("wait finish ")
 	}
 
 
-	wg.Wait()
+
+
 	fmt.Println("all thread has read maps")
 
 	checkUnusedData(bulkProcessor)
@@ -286,9 +305,14 @@ func queryIndex(idFlag int, db *sql.DB, bulkProcessor *elastic.BulkProcessor, wg
 		"week_uv, week_product_click_uv, week_add_cart_uv, 	" +
 		"month_pv, 	month_product_click_count, 	month_add_cart_count, 	month_uv, 	month_product_click_uv, month_add_cart_uv, prepare_tags " +
 		"from app_search_keyword_year_week_p_day where id >= %d and id < %d", idFlag, idFlag + TABLE_SPLIT_STEP_SIZE)
+
+	log.Print(sqlStr)
 	results, err := db.Query(sqlStr)
 	if err != nil { log.Print(err.Error()) }
 
+
+	log.Print("read database success ")
+
 	for results.Next() {
 		var id int
 		var keyword sql.NullString
@@ -335,8 +359,6 @@ func queryIndex(idFlag int, db *sql.DB, bulkProcessor *elastic.BulkProcessor, wg
 		if keyword.Valid && len(keyword.String) > 0 && keyword.String != "" {
 			key := cleanKeyword(keyword.String)
 
-
-
 			var w = &Word{
 					Keyword:key,
 					YearCount: int32(yearPv.Int64),
@@ -355,7 +377,8 @@ func queryIndex(idFlag int, db *sql.DB, bulkProcessor *elastic.BulkProcessor, wg
 					MonthUv: int32(monthUv.Int64),
 					MonthProductClickUv: int32(monthProductClickUv.Int64),
 					MonthAddCartUv: int32(monthAddCartUv.Int64) ,
-			        SuggestTags: prepareTags.String}
+			        SuggestTags: prepareTags.String,
+			        UpdateTime: UPDATE_TIME}
 
 			if v, isExist := wordMap.Load(key); isExist {
 				merge(w,v)
@@ -404,22 +427,26 @@ func after(executionId int64, requests []elastic.BulkableRequest, response *elas
 
 func processWord(w *Word) {
 	w.KeywordPinYin = convertToPinyin(w.Keyword)
+
+	// 年点击加购率
 	w.YearClickRatio = calculateRatio(w.YearClickCount, w.YearCount)
 	w.YearCartRatio = calculateRatio(w.YearCartCount, w.YearCount)
+
+	// 周点击加购率
 	w.WeekClickRatio = calculateRatio(w.WeekClickCount, w.WeekCount)
 	w.WeekCartRatio = calculateRatio(w.WeekCartCount, w.WeekCount)
 
-	// 非默认值，加权
+	// 年加购率 再加权
 	if w.YearCount != 0 && w.YearCartCount != 0 {
 		w.YearCartRatio *= 3
 	}
 
-	// 非默认值，加权
+	// 周加购率 再加权
 	if w.WeekCount != 0 && w.WeekCartCount != 0 {
 		w.WeekCartRatio *= 3
 	}
 
-	// 非默认值，加权
+	// 周点击率 再加权
 	if w.WeekCount != 0 && w.WeekClickCount != 0 {
 		w.WeekClickRatio *= 2
 	}
@@ -452,22 +479,22 @@ func isFilterWord(w *Word) bool {
 	if w.IsSensitive { return true }
 
 	// 过滤掉太长的词 每个中文字占3个byte
-	if utf8.RuneCountInString(w.Keyword) <= 1 || len(w.Keyword) > 60 { return true }
+	if utf8.RuneCountInString(w.Keyword) <= 1 || len(w.Keyword) > 50 { return true }
 
 	// 过滤掉商品id,商品id是有7位数字组成
 	if len(w.Keyword) > 6 && isAllDigit(w.Keyword) { return true }
 
-	// 品牌词 类目词 人工干预词 不做过滤
-	if w.IsBrand || w.IsCategory || w.IsManual { return false }
-
-	// 年数据过滤
-	if w.YearCount == 0 || w.YearClickCount == 0  { return true }
-
 	// 前缀过滤
 	for _, v := range prefixFilterArr {
 		if strings.HasPrefix(w.Keyword, v) { return true }
 	}
 
+	// 品牌词 类目词 人工干预词 不做过滤
+	if w.IsBrand || w.IsCategory || w.IsManual { return false }
+
+	// 年数据过滤
+	if w.YearCount < 2 || w.YearClickCount < 2  { return true }
+
 	// 判断是否是热搜词 一年内搜索次数大于50或者一周内搜索次数大于5
 	if isHotSearchWord(w) {
 		// 搜索次数比较多 转化率或者点击率较高的 不过滤
@@ -505,14 +532,22 @@ func calculateRatio(numerator int32, denominator int32) float64 {
 }
 
 func calculateWordRank(w *Word) {
+
 	wordRank := 10000.0
+	// 长度因子
 	wordRank += 3000 * calculateLengthFactor(len(w.Keyword))
+	// 年数量因子
 	wordRank += 2000 * calculateCountFactor(w.YearCount, 1)
+	// 周数量因子
 	wordRank += 2000 * calculateCountFactor(w.WeekCount, 52)
 
+	// 年点击率因子
 	wordRank += 3000 * calculateRatioFactor(w.YearClickRatio, w.YearClickCount)
+	// 周点击率因子
 	wordRank += 3000 * calculateRatioFactor(w.WeekClickRatio, w.WeekClickCount)
+	// 年加购率因子
 	wordRank += 3000 * calculateRatioFactor(w.YearCartRatio, w.YearCartCount)
+	// 周加购率因子
 	wordRank += 3000 * calculateRatioFactor(w.WeekCartRatio, w.WeekCartCount)
 	if w.IsBrand { wordRank *= 1.8 }
 	if w.IsCategory { wordRank *= 1.2 }
@@ -522,8 +557,11 @@ func calculateWordRank(w *Word) {
 
 func calculateWordABRank(w *Word) {
 
+	// 月点击加购率
 	monthClickRatio := calculateRatio(w.MonthProductClickUv, w.MonthUv)
 	monthCartRatio := calculateRatio(w.MonthAddCartUv, w.MonthUv)
+
+	// 周点击加购率(和A相比, count 换成了uv)
 	weekClickRatioNew := calculateRatio(w.WeekClickUv, w.WeekUv)
 	weekCartRatioNew := calculateRatio(w.WeekAddCartUv, w.WeekUv)
 
@@ -549,24 +587,28 @@ func calculateWordABRank(w *Word) {
 
 
 	wordABRank := 10000.0
+	// 长度因子
 	wordABRank += 3000 * calculateLengthFactor(len(w.Keyword))
-	wordABRank += 2000 * calculateCountFactor(w.MonthUv, 1)
-	wordABRank += 2000 * calculateCountFactor(w.WeekUv, 4)
 
-	// 点击
-	//  年
+	// 月数量因子
+	wordABRank += 2000 * calculateCountFactor(w.MonthUv, 4)
+	// 周数量因子
+	wordABRank += 2000 * calculateCountFactor(w.WeekUv, 52)
+	// 年数量因子
 	wordABRank += 2000 * calculateCountFactor(w.YearCount, 1)
-	// 月
+
+	// 点击
+	// 月点击率因子
 	wordABRank += 3000 * calculateRatioFactor(monthClickRatio, w.MonthProductClickUv)
-	// 周
+	// 周点击率因子
 	wordABRank += 3000 * calculateRatioFactor(weekClickRatioNew, w.WeekUv)
 
 	// 加购
-	// 年
+	// 年加购率因子
 	wordABRank += 3000 * calculateRatioFactor(w.YearCartRatio, w.YearCartCount)
-	// 月
+	// 月加购率因子
 	wordABRank += 3000 * calculateRatioFactor(monthCartRatio, w.MonthUv)
-	// 周
+	// 周加购率因子
 	wordABRank += 3000 * calculateRatioFactor(weekCartRatioNew, w.WeekUv)
 
 	if w.IsBrand { wordABRank *= 1.8 }

main/test.go

@@ -16,16 +16,8 @@ type B struct {
 }
 func main() {
 
-	var arr = strings.Split("",",")
-	var s = ""
-	for i,leng := 0, len(arr); i< MAX_TAG_SIZE && i < leng; i++ {
-		if i == MAX_TAG_SIZE-1 || i == leng -1 {
-			s = s + arr[i]
-		} else {
-			s = s + arr[i] + ","
-		}
-	}
-	fmt.Print(s)
+	prefix := strings.HasPrefix("tod's", "tod's")
+	fmt.Print(prefix)
 
 }