/***************************************************************************** * function name : Buiding * params : * return : * * description : 构建索引文件,并将detail存入redis中,这个函数太啰嗦了。 * ******************************************************************************/ func (this *DBBuilder) Buiding() error { var fields string for index, v := range this.Fields { //构造sql语句 fields = fields + "," + v.Name //构造索引数据指针 if v.IsIvt { if v.FType == "T" { this.Fields[index].IvtIdx = utils.NewInvertIdx(utils.TYPE_TEXT, v.Name) this.Fields[index].IvtStrDic = utils.NewStringIdxDic(5001) } if v.FType == "N" { this.Fields[index].IvtIdx = utils.NewInvertIdx(utils.TYPE_NUM, v.Name) if v.IsPK { this.Fields[index].IvtNumDic = utils.NewNumberIdxDic(1) } else { this.Fields[index].IvtNumDic = utils.NewNumberIdxDic(4) } } } if v.IsPlf { if v.FType == "T" { this.Fields[index].PlfText = indexer.NewTextProfile(v.Name) } if v.FType == "N" { this.Fields[index].PlfNumber = indexer.NewNumberProfile(v.Name) } } } fmt.Printf("%v\n", fields) this.DetailIdx = indexer.NewDetail() sql := fmt.Sprintf(this.sql, fields[1:len(fields)]) fmt.Printf("SQL :: %v \n", sql) rows, err := this.Dbadaptor.QueryFormat(sql) if err != nil { this.Logger.Error(" %v", err) return err } defer rows.Close() var doc_id int64 doc_id = 1 segment := utils.NewSegmenter("./data/dictionary.txt") builder := &utils.IndexBuilder{Segmenter: segment, TempIndex: make(map[string][]utils.TmpIdx), TempIndexNum: make(map[string]int64)} for rows.Next() { //values := make([]interface{},len(this.Fields)) values := make([]interface{}, len(this.Fields)) writeCols := make([]string, len(this.Fields)) for i, _ := range writeCols { values[i] = &writeCols[i] } err := rows.Scan(values...) if err != nil { this.Logger.Error("SQL ERROR : %v", err) return err } redis_map := make(map[string]string) for index, v := range writeCols { redis_map[this.Fields[index].Name] = v //v,_ := value.(string) if this.Fields[index].IsIvt { if this.Fields[index].FType == "T" { err := builder.BuildTextIndex(doc_id, v, this.Fields[index].IvtIdx, this.Fields[index].IvtStrDic, this.Fields[index].SType, false) //err := builder.BuildTextIndexTemp(doc_id, v, this.Fields[index].IvtIdx, this.Fields[index].IvtStrDic,this.Fields[index].SType,this.Fields[index].Name) if err != nil { this.Logger.Error("ERROR : %v", err) } } if this.Fields[index].FType == "N" { v_num, err := strconv.ParseInt(v, 0, 0) if err != nil { v_num = 0 this.Logger.Warn("Warning : name : [%v] , value: [%v] , error : [%v]", this.Fields[index].Name, v, err) } err = builder.BuildNumberIndex(doc_id, v_num, this.Fields[index].IvtIdx, this.Fields[index].IvtNumDic, false) //err = builder.BuildNumberIndexTemp(doc_id, v_num, this.Fields[index].IvtIdx, this.Fields[index].IvtNumDic,this.Fields[index].Name) if err != nil { this.Logger.Error("ERROR : %v", err) } } } if this.Fields[index].IsPlf { if this.Fields[index].FType == "T" { //添加日期类型的更新,仅精确到天 add by wuyinghao 2015-08-21 if this.Fields[index].SType == 5 { vl := strings.Split(v, " ") v = vl[0] } err := this.Fields[index].PlfText.PutProfile(doc_id, v) if err != nil { this.Logger.Error("ERROR : %v", err) } } if this.Fields[index].FType == "N" { v_num, err := strconv.ParseInt(v, 0, 0) if err != nil { v_num = 0 this.Logger.Warn("Warning : name : %v , value: %v , error : %v", this.Fields[index].Name, v, err) } err = this.Fields[index].PlfNumber.PutProfile(doc_id, v_num) if err != nil { this.Logger.Error("ERROR : %v", err) } } } } ///// //this.RedisCli.SetFields(doc_id, redis_map) //// if this.DetailIdx.PutDocInfo(doc_id, redis_map) != nil { this.Logger.Error("PutDocInfo doc_id Error : %v \n", err) } fieldlist := make([]string, 0) for k, _ := range redis_map { fieldlist = append(fieldlist, k) } doc_id++ if doc_id%5000 == 0 { this.Logger.Info("processing doc_id : %v \n", doc_id) } //this.Logger.Info("DOC_ID : %v VALUE : %v", doc_id, writeCols) } this.DetailIdx.WriteDetailToFile() //写入全部数据 //builder.WriteAllTempIndexToFile() //builder.WriteIndexToFile() //writeCount:=0 //var writeChan chan int for index, fields := range this.Fields { if this.Fields[index].IsIvt { //utils.WriteToJsonWithChan(fields.IvtIdx, fmt.Sprintf("./index/%v_idx.json", fields.Name),writeChan) //utils.WriteToJson(fields.IvtIdx, fmt.Sprintf("./index/%v_idx.json", fields.Name)) utils.WriteToIndexFile(fields.IvtIdx, fmt.Sprintf("./index/%v_idx.idx", fields.Name)) utils.WriteToJson(fields.IvtIdx, fmt.Sprintf("./index/%v_idx.json", fields.Name)) //writeCount++ if this.Fields[index].FType == "T" { //utils.WriteToJsonWithChan(fields.IvtStrDic, fmt.Sprintf("./index/%v_dic.json", fields.Name),writeChan) utils.WriteToJson(fields.IvtStrDic, fmt.Sprintf("./index/%v_dic.json", fields.Name)) //writeCount++ } if this.Fields[index].FType == "N" { //utils.WriteToJsonWithChan(fields.IvtNumDic, fmt.Sprintf("./index/%v_dic.json", fields.Name),writeChan) utils.WriteToJson(fields.IvtNumDic, fmt.Sprintf("./index/%v_dic.json", fields.Name)) //writeCount++ } } if this.Fields[index].IsPlf { if this.Fields[index].FType == "T" { //utils.WriteToJsonWithChan(fields.PlfText, fmt.Sprintf("./index/%v_pfl.json", fields.Name),writeChan) utils.WriteToJson(fields.PlfText, fmt.Sprintf("./index/%v_pfl.json", fields.Name)) //writeCount++ } if this.Fields[index].FType == "N" { //utils.WriteToJsonWithChan(fields.PlfNumber, fmt.Sprintf("./index/%v_pfl.json", fields.Name),writeChan) utils.WriteToJson(fields.PlfNumber, fmt.Sprintf("./index/%v_pfl.json", fields.Name)) //writeCount++ } } } /* fmt.Printf("writeCount : %v \n ",writeCount) for { select{ case <-writeChan: writeCount-- fmt.Printf("recive writeCount : %v \n ",writeCount) if writeCount == 0 { fmt.Printf("Finish building index...\n") return nil } } } */ return nil }
/***************************************************************************** * function name : Buiding * params : * return : * * description : 构建索引文件,并将detail存入redis中,这个函数太啰嗦了。 * ******************************************************************************/ func (this *DBBuilder) Buiding() error { var fields string for index, v := range this.Fields { //构造sql语句 fields = fields + "," + v.Name //构造索引数据指针 if v.IsIvt { if v.FType == "T" { this.Fields[index].IvtIdx = utils.NewInvertIdx(utils.TYPE_TEXT, v.Name) this.Fields[index].IvtStrDic = utils.NewStringIdxDic(10000) } if v.FType == "N" { this.Fields[index].IvtIdx = utils.NewInvertIdx(utils.TYPE_NUM, v.Name) this.Fields[index].IvtNumDic = utils.NewNumberIdxDic(10000) } } if v.IsPlf { if v.FType == "T" { this.Fields[index].PlfText = indexer.NewTextProfile(v.Name) } if v.FType == "N" { this.Fields[index].PlfNumber = indexer.NewNumberProfile(v.Name) } } } sql := fmt.Sprintf(this.sql, fields[1:len(fields)]) fmt.Printf("SQL :: %v \n", sql) rows, err := this.Dbadaptor.QueryFormat(sql) if err != nil { this.Logger.Error(" %v", err) return err } defer rows.Close() var doc_id int64 doc_id = 1 segment := utils.NewSegmenter("./data/dictionary.txt") builder := &utils.IndexBuilder{segment} for rows.Next() { //values := make([]interface{},len(this.Fields)) values := make([]interface{}, len(this.Fields)) writeCols := make([]string, len(this.Fields)) for i, _ := range writeCols { values[i] = &writeCols[i] } err := rows.Scan(values...) if err != nil { this.Logger.Error("SQL ERROR : %v", err) return err } redis_map := make(map[string]string) for index, v := range writeCols { redis_map[this.Fields[index].Name] = v //v,_ := value.(string) if this.Fields[index].IsIvt { if this.Fields[index].FType == "T" { err := builder.BuildTextIndex(doc_id, v, this.Fields[index].IvtIdx, this.Fields[index].IvtStrDic) if err != nil { this.Logger.Error("ERROR : %v", err) } } if this.Fields[index].FType == "N" { v_num, err := strconv.ParseInt(v, 0, 0) if err != nil { this.Logger.Error("ERROR : %v", err) } err = builder.BuildNumberIndex(doc_id, v_num, this.Fields[index].IvtIdx, this.Fields[index].IvtNumDic) if err != nil { this.Logger.Error("ERROR : %v", err) } } } if this.Fields[index].IsPlf { if this.Fields[index].FType == "T" { err := this.Fields[index].PlfText.PutProfile(doc_id, v) if err != nil { this.Logger.Error("ERROR : %v", err) } } if this.Fields[index].FType == "N" { v_num, err := strconv.ParseInt(v, 0, 0) if err != nil { this.Logger.Error("ERROR : %v", err) } err = this.Fields[index].PlfNumber.PutProfile(doc_id, v_num) if err != nil { this.Logger.Error("ERROR : %v", err) } } } } this.RedisCli.SetFields(doc_id, redis_map) fieldlist := make([]string, 0) for k, _ := range redis_map { fieldlist = append(fieldlist, k) } doc_id++ this.Logger.Info("DOC_ID : %v VALUE : %v", doc_id, writeCols) } for index, fields := range this.Fields { if this.Fields[index].IsIvt { utils.WriteToJson(fields.IvtIdx, fmt.Sprintf("./index/%v_idx.json", fields.Name)) if this.Fields[index].FType == "T" { utils.WriteToJson(fields.IvtStrDic, fmt.Sprintf("./index/%v_dic.json", fields.Name)) } if this.Fields[index].FType == "N" { utils.WriteToJson(fields.IvtNumDic, fmt.Sprintf("./index/%v_dic.json", fields.Name)) } } if this.Fields[index].IsPlf { if this.Fields[index].FType == "T" { utils.WriteToJson(fields.PlfText, fmt.Sprintf("./index/%v_pfl.json", fields.Name)) } if this.Fields[index].FType == "N" { utils.WriteToJson(fields.PlfNumber, fmt.Sprintf("./index/%v_pfl.json", fields.Name)) } } } return nil }
/***************************************************************************** * function name : Buiding * params : * return : * * description : 构建索引文件,并将detail存入redis中,这个函数太啰嗦了。 * ******************************************************************************/ func (this *DBBuilder) Buiding() error { var fields string for index, v := range this.Fields { //构造sql语句 fields = fields + "," + v.Name //新建插件 cumstom := plugins.NewPlus(v.Name) cumstom.Init() //构造索引数据指针 if v.IsIvt { if v.FType == "T" { this.Fields[index].IvtIdx = utils.NewInvertIdx(utils.TYPE_TEXT, v.Name) this.Fields[index].IvtStrDic = utils.NewStringIdxDic(v.Name) this.Fields[index].CustomeInter = cumstom } if v.FType == "N" { this.Fields[index].IvtIdx = utils.NewInvertIdx(utils.TYPE_NUM, v.Name) this.Fields[index].IvtNumDic = utils.NewNumberIdxDic(v.Name) this.Fields[index].CustomeInter = cumstom } if v.FType == "I" { this.Fields[index].IvtIdx = utils.NewInvertIdx(utils.TYPE_TEXT, v.Name) this.Fields[index].IvtStrDic = utils.NewStringIdxDic(v.Name) this.Fields[index].CustomeInter = cumstom } } if v.IsPlf { if v.FType == "T" { this.Fields[index].PlfText = indexer.NewTextProfile(v.Name) this.Fields[index].PlfText.SetCustomInterface(cumstom) } if v.FType == "N" { this.Fields[index].PlfNumber = indexer.NewNumberProfile(v.Name) this.Fields[index].PlfNumber.SetCustomInterface(cumstom) } if v.FType == "I" { this.Fields[index].PlfByte = indexer.NewByteProfile(v.Name) this.Fields[index].PlfByte.SetCustomInterface(cumstom) } } } //fmt.Printf("%v\n", fields) this.DetailIdx = indexer.NewDetail() sql := fmt.Sprintf(this.sql, fields[1:len(fields)]) fmt.Printf("SQL :: %v \n", sql) rows, err := this.Dbadaptor.QueryFormat(sql) if err != nil { this.Logger.Error(" %v", err) return err } defer rows.Close() var doc_id int64 doc_id = 1 segment := utils.NewSegmenter("./data/dictionary.txt") builder := &utils.IndexBuilder{Segmenter: segment, TempIndex: make(map[string][]utils.TmpIdx), TempIndexNum: make(map[string]int64)} for rows.Next() { //values := make([]interface{},len(this.Fields)) values := make([]interface{}, len(this.Fields)) writeCols := make([]string, len(this.Fields)) for i, _ := range writeCols { values[i] = &writeCols[i] } err := rows.Scan(values...) if err != nil { this.Logger.Error("SQL ERROR : %v", err) return err } redis_map := make(map[string]string) for index, v := range writeCols { redis_map[this.Fields[index].Name] = v //v,_ := value.(string) if this.Fields[index].IsIvt { if this.Fields[index].FType == "T" || this.Fields[index].FType == "I" { err := builder.BuildTextIndex(doc_id, v, this.Fields[index].IvtIdx, this.Fields[index].IvtStrDic, this.Fields[index].SType, false, this.Fields[index].CustomeInter) //err := builder.BuildTextIndexTemp(doc_id, v, this.Fields[index].IvtIdx, this.Fields[index].IvtStrDic,this.Fields[index].SType,this.Fields[index].Name) if err != nil { this.Logger.Error("ERROR : %v", err) } } if this.Fields[index].FType == "N" { v_num, err := strconv.ParseInt(v, 0, 0) if err != nil { v_num = 0 //this.Logger.Warn("Warning : name : [%v] , value: [%v] , error : [%v]", this.Fields[index].Name, v, err) } err = builder.BuildNumberIndex(doc_id, v_num, this.Fields[index].IvtIdx, this.Fields[index].IvtNumDic, this.Fields[index].SType, false, this.Fields[index].CustomeInter) //err = builder.BuildNumberIndexTemp(doc_id, v_num, this.Fields[index].IvtIdx, this.Fields[index].IvtNumDic,this.Fields[index].Name) if err != nil { this.Logger.Error("ERROR : %v", err) } } } if this.Fields[index].IsPlf { if this.Fields[index].FType == "T" { //添加日期类型的更新,仅精确到天 add by wuyinghao 2015-08-21 if this.Fields[index].SType == 5 { vl := strings.Split(v, " ") v = vl[0] } else if this.Fields[index].SType == 9 { v = this.Fields[index].PlfText.GetCustomInterface().BuildStringProfile(v) } err := this.Fields[index].PlfText.PutProfile(doc_id, v) if err != nil { this.Logger.Error("ERROR : %v", err) } } if this.Fields[index].FType == "N" { v_num, err := strconv.ParseInt(v, 0, 0) if err != nil { v_num = 0 //this.Logger.Warn("Warning : name : %v , value: %v , error : %v", this.Fields[index].Name, v, err) } if this.Fields[index].SType == 9 { v_num = this.Fields[index].PlfNumber.GetCustomInterface().BuildIntProfile(v) } err = this.Fields[index].PlfNumber.PutProfile(doc_id, v_num) if err != nil { this.Logger.Error("ERROR : %v", err) } } if this.Fields[index].FType == "I" { v_byte := []byte(v) if this.Fields[index].SType == 9 { v_byte = this.Fields[index].PlfByte.GetCustomInterface().BuildByteProfile(v_byte) } err = this.Fields[index].PlfByte.PutProfile(doc_id, v_byte) if err != nil { this.Logger.Error("ERROR : %v", err) } } } } ///// //this.RedisCli.SetFields(doc_id, redis_map) //// //for k,v := range redis_map{ // fmt.Printf("\"%v\":\"%v\",",k,v) //} //fmt.Printf("\n") if this.DetailIdx.PutDocInfo(doc_id, redis_map) != nil { this.Logger.Error("PutDocInfo doc_id Error : %v \n", err) } fieldlist := make([]string, 0) for k, _ := range redis_map { fieldlist = append(fieldlist, k) } doc_id++ if doc_id%5000 == 0 { this.Logger.Info("processing doc_id : %v \n", doc_id) } //this.Logger.Info("DOC_ID : %v VALUE : %v", doc_id, writeCols) } writeCount := 0 writeChan := make(chan string, 1000) this.DetailIdx.WriteDetailToFile() this.Logger.Info("Make BitMap File") utils.MakeBitmapFile() for index, _ := range this.Fields { if this.Fields[index].IsIvt { //utils.WriteToIndexFile(this.Fields[index].IvtIdx, fmt.Sprintf("./index/%v_idx.idx", this.Fields[index].Name)) this.Fields[index].IvtIdx.WriteToFile() if this.Fields[index].FType == "T" || this.Fields[index].FType == "I" { this.Fields[index].IvtStrDic.WriteToFile() } if this.Fields[index].FType == "N" { this.Fields[index].IvtNumDic.WriteToFile() } } if this.Fields[index].IsPlf { if this.Fields[index].FType == "T" { this.Fields[index].PlfText.WriteToFile() } if this.Fields[index].FType == "N" { this.Fields[index].PlfNumber.WriteToFile() } if this.Fields[index].FType == "I" { this.Fields[index].PlfByte.WriteToFile() //go this.Fields[index].PlfByte.WriteToFileWithChan(writeChan) //writeCount++ } } } //fmt.Printf("Waiting %v threads\n ", writeCount) if writeCount == 0 { close(writeChan) return nil } for { select { case file_name := <-writeChan: writeCount-- fmt.Printf("Write [%v] finished \n ", file_name) if writeCount == 0 { fmt.Printf("Finish building all index...\n") close(writeChan) return nil } } } return nil }