使用上新字段, 添加另一算分字段

main/suggest-task.go

@@ -3,7 +3,6 @@ package main
 import (
 	"bufio"
 	"context"
-	"crypto/md5"
 		"database/sql"
 	"fmt"
 	_ "github.com/go-sql-driver/mysql"
@@ -19,7 +18,8 @@ import (
 	"time"
 	"unicode"
 	"unicode/utf8"
-		)
+	"crypto/md5"
+)
 
 type Word struct {
 	Keyword        string  `json:"keyword"`
@@ -42,6 +42,16 @@ type Word struct {
 	WordRank       float64 `json:"wordRank"`
 	WordABRank     float64 `json:"wordABRank"`
 	KeywordVersion string  `json:"keywordVersion"`
+
+	WeekUv  int32 `json:"-"`
+	WeekClickUv int32 `json:"-"`
+	WeekAddCartUv int32 `json:"-"`
+	MonthPv int32 `json:"-"`
+	MonthClickCount int32 `json:"-"`
+	MonthAddCartCount int32 `json:"-"`
+	MonthUv int32 `json:"-"`
+	MonthProductClickUv int32 `json:"-"`
+	MonthAddCartUv int32 `json:"-"`
 }
 
 var wordMap sync.Map
@@ -95,11 +105,14 @@ func main() {
 	log.Printf("maxId/10000=%d\n", count)
 
 	for i := 0; i <= count; i++ {
+		wg.Add(1)
 		go queryIndex(i * TABLE_SPLIT_STEP_SIZE, datawareDB, bulkProcessor, &wg)
 	}
 
+
 	wg.Wait()
 	fmt.Println("all thread has read maps")
+
 	checkUnusedData(bulkProcessor)
 
 	err = bulkProcessor.Flush()
@@ -236,10 +249,13 @@ func queryInfo(db *sql.DB) []int  {
 }
 
 func queryIndex(idFlag int, db *sql.DB, bulkProcessor *elastic.BulkProcessor, wg *sync.WaitGroup)  {
-	wg.Add(1)
+
 	// 循环时可能查询到重复数据，应该以id 的上下界来查询
 	var sqlStr = fmt.Sprintf("select id, keyword, year_pv, year_product_click_count, year_add_cart_count, " +
-		"week_pv, week_product_click_count, week_add_cart_count, p_day from app_search_keyword_year_week_p_day where id >= %d and id < %d", idFlag, idFlag + TABLE_SPLIT_STEP_SIZE)
+		"week_pv, week_product_click_count, week_add_cart_count, p_day, " +
+		"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 " +
+		"from app_search_keyword_year_week_p_day where id >= %d and id < %d", idFlag, idFlag + TABLE_SPLIT_STEP_SIZE)
 	results, err := db.Query(sqlStr)
 	if err != nil { log.Print(err.Error()) }
 
@@ -253,8 +269,35 @@ func queryIndex(idFlag int, db *sql.DB, bulkProcessor *elastic.BulkProcessor, wg
 		var weekProductClickCount sql.NullInt64
 		var weekAddCartCount sql.NullInt64
 		var pDay string
+		var weekUv sql.NullInt64
+		var weekClickUv sql.NullInt64
+		var weekAddCartUv sql.NullInt64
+		var monthPv  sql.NullInt64
+		var monthClickCount sql.NullInt64
+		var monthAddCartCount sql.NullInt64
+		var monthUv  sql.NullInt64
+		var monthProductClickUv sql.NullInt64
+		var monthAddCartUv sql.NullInt64
+
+		err = results.Scan(&id,
+			               &keyword,
+			               &yearPv,
+			               &yearProductClickCount,
+			               &yearAddCartCount,
+			               &weekPv,
+			               &weekProductClickCount,
+			               &weekAddCartCount,
+			               &pDay,
+			               &weekUv,
+			               &weekClickUv,
+			               &weekAddCartUv,
+			               &monthPv,
+			               &monthClickCount,
+			               &monthAddCartCount,
+			               &monthUv,
+			               &monthProductClickUv,
+			               &monthAddCartUv)
 
-		err = results.Scan(&id, &keyword, &yearPv, &yearProductClickCount, &yearAddCartCount, &weekPv, &weekProductClickCount, &weekAddCartCount, &pDay)
 		if err != nil { log.Print(err.Error()) }
 
 		if keyword.Valid && len(keyword.String) > 0 && keyword.String != "" {
@@ -268,7 +311,16 @@ func queryIndex(idFlag int, db *sql.DB, bulkProcessor *elastic.BulkProcessor, wg
 					WeekCount: int32(weekPv.Int64),
 					WeekClickCount: int32(weekProductClickCount.Int64),
 					WeekCartCount: int32(weekAddCartCount.Int64),
-					KeywordVersion:pDay}
+					KeywordVersion:pDay,
+			        WeekUv: int32(weekUv.Int64),
+				    WeekClickUv: int32(weekClickUv.Int64),
+				    WeekAddCartUv: int32(weekAddCartUv.Int64),
+				    MonthPv: int32(monthPv.Int64),
+					MonthClickCount: int32(monthClickCount.Int64),
+					MonthAddCartCount: int32(monthAddCartCount.Int64),
+					MonthUv: int32(monthUv.Int64),
+					MonthProductClickUv: int32(monthProductClickUv.Int64),
+					MonthAddCartUv: int32(monthAddCartUv.Int64) }
 
 			if v, isExist := wordMap.Load(key); isExist {
 				merge(w,v)
@@ -296,6 +348,16 @@ func merge(word *Word, v interface{}) {
 	word.WeekCount += t.WeekCount
 	word.WeekCartCount += t.WeekCartCount
 	word.WeekClickCount += t.WeekClickCount
+
+    word.WeekUv += t.WeekUv
+	word.WeekClickUv += t.WeekClickUv
+	word.WeekAddCartUv += t.WeekAddCartUv
+	word.MonthPv += t.MonthPv
+	word.MonthClickCount += t.MonthClickCount
+	word.MonthAddCartCount += t.MonthAddCartCount
+	word.MonthUv += t.MonthUv
+	word.MonthProductClickUv += t.MonthProductClickUv
+	word.MonthAddCartUv += t.MonthAddCartUv
 }
 
 func after(executionId int64, requests []elastic.BulkableRequest, response *elastic.BulkResponse, err error) {
@@ -421,19 +483,53 @@ func calculateWordRank(w *Word) {
 }
 
 func calculateWordABRank(w *Word) {
+
+	monthClickRatio := calculateRatio(w.MonthProductClickUv, w.MonthUv)
+	monthCartRatio := calculateRatio(w.MonthAddCartUv, w.MonthUv)
+	weekClickRatioNew := calculateRatio(w.WeekClickUv, w.WeekUv)
+	weekCartRatioNew := calculateRatio(w.WeekAddCartUv, w.WeekUv)
+
+	// 月点击
+	if w.MonthProductClickUv != 0 && w.MonthUv != 0 {
+		monthClickRatio *= 1.5
+	}
+
+	// 月加购，加权
+	if w.MonthAddCartUv != 0 && w.MonthUv != 0 {
+		monthCartRatio *= 3
+	}
+
+	// 周点击，加权
+	if w.WeekClickUv != 0 && w.WeekUv != 0 {
+		weekClickRatioNew *= 2
+	}
+
+	// 周加购，加权
+	if w.WeekAddCartUv != 0 && w.WeekUv != 0 {
+		weekCartRatioNew *= 3
+	}
+
+
 	wordABRank := 10000.0
 	wordABRank += 3000 * calculateLengthFactor(len(w.Keyword))
-	wordABRank += 2000 * calculateCountFactor(w.YearCount, 1)
-	wordABRank += 2000 * calculateCountFactor(w.WeekCount, 52)
+	wordABRank += 2000 * calculateCountFactor(w.MonthUv, 1)
+	wordABRank += 2000 * calculateCountFactor(w.WeekUv, 4)
 
 	// 点击
-	// 年点击改为 2000
-	wordABRank += 2000 * calculateRatioFactor(w.YearClickRatio, w.YearClickCount)
-	wordABRank += 3000 * calculateRatioFactor(w.WeekClickRatio, w.WeekClickCount)
+	//  年
+	wordABRank += 2000 * calculateCountFactor(w.YearCount, 1)
+	// 月
+	wordABRank += 3000 * calculateRatioFactor(monthClickRatio, w.MonthProductClickUv)
+	// 周
+	wordABRank += 3000 * calculateRatioFactor(weekClickRatioNew, w.WeekUv)
 
 	// 加购
+	// 年
 	wordABRank += 3000 * calculateRatioFactor(w.YearCartRatio, w.YearCartCount)
-	wordABRank += 3000 * calculateRatioFactor(w.WeekCartRatio, w.WeekCartCount)
+	// 月
+	wordABRank += 3000 * calculateRatioFactor(monthCartRatio, w.MonthUv)
+	// 周
+	wordABRank += 3000 * calculateRatioFactor(weekCartRatioNew, w.WeekUv)
 
 	if w.IsBrand { wordABRank *= 1.8 }
 	if w.IsCategory { wordABRank *= 1.2 }

main/test.go

 package main
 
 import (
-	"encoding/json"
 		"fmt"
+	"math"
 )
 
 type B struct {
@@ -15,18 +15,29 @@ type B struct {
 }
 
 func main() {
-	b := B{
-		Keyword: "赵延超",
-		KeywordPinYin: "zhaoyanchao",
-		YearCount: 1000,
-		YearCartCount: 100,
-		YearClickCount: 10,
-		ZhaoCount: 2}
-
-	if jsonBytes,errs := json.Marshal(b); errs == nil {
-		fmt.Print(string(jsonBytes))
+
+	fmt.Println(calculateRatioFactor2(0,0))
+
+	fmt.Println(calculateRatioFactor2(0.2,1))
+	fmt.Println(calculateRatioFactor2(0.2,10))
+}
+
+
+func calculateRatioFactor2(ratio float64, count int32) float64 {
+	var rank float64
+	switch {
+	case count > 1 && count < 10 : rank = 1.2
+	case count >= 10 && count < 20 : rank = 1.4
+	case count >= 20 && count < 50 : rank = 1.6
+	case count >= 50 && count < 100 : rank = 1.8
+	case count >= 100 && count < 200 : rank = 2.0
+	case count >= 200 && count < 500 : rank = 2.2
+	case count >= 500 : rank = 2.5
+	default:rank = 1.0
 	}
 
+	//根据搜索转化率，转换为热度因子
+	return math.Log10(math.Sqrt(ratio + 10)) * rank
 }