diff --git a/biz/config/config.go b/biz/config/config.go
index ea6635b..3c320a5 100644
--- a/biz/config/config.go
+++ b/biz/config/config.go
@@ -120,3 +120,16 @@ func LoadConfig(path string) *Config {
 	DefaultConfig = &c
 	return &c
 }
+
+func GetJob(conf *Config, sys, job string) (Job, string, error) {
+	s, err := conf.GetSystem(sys)
+	if err != nil {
+		return Job{}, "", err
+	}
+
+	j, err := s.GetJob(job)
+	if err != nil {
+		return Job{}, "", err
+	}
+	return j, s.Db, nil
+}
diff --git a/biz/db/db.go b/biz/db/db.go
index fcaa2f5..1e4ff6a 100644
--- a/biz/db/db.go
+++ b/biz/db/db.go
@@ -2,10 +2,14 @@ package db
 
 import (
 	"context"
+	"database/sql"
 	"excel_export/biz/export"
+	"fmt"
 	"gorm.io/driver/mysql"
 	"gorm.io/gorm"
 	"runtime/trace"
+	"strconv"
+	"time"
 )
 
 var _ export.DataFetcher = new(Db)
@@ -35,25 +39,76 @@ func (d *Db) Fetch(s string) (*export.Data, error) {
 		return nil, err
 	}
 	defer rows.Close()
-
+	//titles := make([]string, 0, 10)
 	titles, err := rows.Columns()
 	if err != nil {
 		return nil, err
 	}
 
-	var dd []map[string]interface{}
-	d.db.ScanRows(rows, &dd)
-	data := make([]interface{}, len(dd))
-	for i, m := range dd {
-		row := make([]interface{}, 0, len(m))
-		for _, title := range titles {
-			row = append(row, m[title])
-		}
-		data[i] = interface{}(row)
-	}
+	data := getData(rows, d.db, titles)
+
+	//vv := transform(titles, dd)
+	//fmt.Println(vv)
+	//f, err := os.Create("./ff.csv")
+	//
+	//w := csv.NewWriter(f)
+	//w.Write(titles)
+	//w.WriteAll(vv)
+	////w.Flush()
+	//f.Close()
 
 	return &export.Data{
 		Title: titles,
 		Data:  data,
 	}, nil
 }
+
+func getData(rows *sql.Rows, db *gorm.DB, titles []string) [][]string {
+	result := make([][]string, 0, 10)
+	for rows.Next() {
+		var row map[string]interface{}
+		db.ScanRows(rows, &row)
+		result = append(result, transformRow(titles, row))
+	}
+	return result
+}
+
+func transform(titles []string, data []map[string]interface{}) [][]string {
+	result := make([][]string, len(data))
+	for i, m := range data {
+		result[i] = transformRow(titles, m)
+	}
+	return result
+}
+
+func transformRow(titles []string, data map[string]interface{}) []string {
+	row := make([]string, 0, len(data))
+	for _, title := range titles {
+		col := data[title]
+		switch v := col.(type) {
+		case string:
+			row = append(row, v)
+		case time.Time:
+			row = append(row, v.Format("2006-01-02 15:04:05"))
+		case int, int8, int16, int32, int64:
+			row = append(row, fmt.Sprintf("%d", v))
+		case float64:
+			// When formatting floats, do not use fmt.Sprintf("%v", n), this will cause numbers below 1e-4 to be printed in
+			// scientific notation. Scientific notation is not a valid way to store numbers in XML.
+			// Also not not use fmt.Sprintf("%f", n), this will cause numbers to be stored as X.XXXXXX. Which means that
+			// numbers will lose precision and numbers with fewer significant digits such as 0 will be stored as 0.000000
+			// which causes tests to fail.
+			row = append(row, strconv.FormatFloat(v, 'f', -1, 64))
+		case float32:
+			row = append(row, strconv.FormatFloat(float64(v), 'f', -1, 32))
+		case []byte:
+			row = append(row, string(v))
+		case nil:
+			row = append(row, "")
+		default:
+			row = append(row, fmt.Sprintf("%v", v))
+		}
+	}
+	return row
+
+}
diff --git a/biz/db/db_test.go b/biz/db/db_test.go
index b029ff6..cd07e5e 100644
--- a/biz/db/db_test.go
+++ b/biz/db/db_test.go
@@ -20,7 +20,7 @@ func TestDb_Fetch(t *testing.T) {
 	ret, err := db.Fetch(sql)
 	assert.Nil(t, err)
 	fmt.Printf("%v \n", ret)
-	item := ret.Data[0].([]interface{})
+	item := ret.Data[0]
 	for _, i := range item {
 		fmt.Printf("%v", i)
 	}
diff --git a/biz/export/csv_exporter.go b/biz/export/csv_exporter.go
new file mode 100644
index 0000000..2104515
--- /dev/null
+++ b/biz/export/csv_exporter.go
@@ -0,0 +1,61 @@
+package export
+
+import "fmt"
+
+type CSVExporter struct {
+	mFetcher DataFetcher
+	file     FileAdapter
+	count    int
+	last     interface{}
+}
+
+func NewCVSExporter(fetcher DataFetcher, file FileAdapter) DataExporter {
+	return &CSVExporter{
+		mFetcher: fetcher,
+		file:     file,
+	}
+}
+
+func (ee *CSVExporter) Fetcher(fetcher DataFetcher) {
+	ee.mFetcher = fetcher
+}
+
+func (ee *CSVExporter) File(file FileAdapter) {
+	ee.file = file
+}
+
+func (ee *CSVExporter) Export(sql, pk string) error {
+	data, err := ee.mFetcher.Fetch(sql)
+	if err != nil {
+		return fmt.Errorf("数据获取错误:%w", err)
+	}
+	ee.count = len(data.Data)
+	//fmt.Printf("Excel Exporter.Excel, got %v data\n", len(data))
+	//ee.file.Open()
+	ee.file.WriteTitle(data.Title)
+	var last interface{}
+	for _, val := range data.Data {
+		last = val
+		ee.file.Write(last)
+	}
+
+	if row, ok := last.([]interface{}); ok {
+		ee.last = row[ee.getPkIndex(data.Title, pk)]
+	}
+
+	//ee.file.Close()
+	return nil
+}
+
+func (ee *CSVExporter) Last() (int, interface{}) {
+	return ee.count, ee.last
+}
+
+func (ee *CSVExporter) getPkIndex(titles []string, pk string) int {
+	for i, title := range titles {
+		if title == pk {
+			return i
+		}
+	}
+	return -1
+}
diff --git a/biz/export/csv_exporter_test.go b/biz/export/csv_exporter_test.go
new file mode 100644
index 0000000..da38678
--- /dev/null
+++ b/biz/export/csv_exporter_test.go
@@ -0,0 +1,26 @@
+package export
+
+import (
+	"github.com/stretchr/testify/assert"
+	"os"
+	"testing"
+)
+
+func TestCSVExporter_Export(t *testing.T) {
+	data := NewMysqlDataFetcher("aaa")
+	pwd, _ := os.Getwd()
+	file := NewCSV(pwd+"/aa-{begin}.csv", 5, map[string]string{"begin": "202301"})
+	e := NewCVSExporter(data, file)
+	file.Open()
+	err := e.Export("aa", "字段1")
+	file.Close()
+	assert.Nil(t, err)
+
+	assert.FileExists(t, pwd+"/aa-202301_0.csv")
+	assert.FileExists(t, pwd+"/aa-202301_1.csv")
+	assert.NoFileExists(t, pwd+"/aa-202301_2.csv")
+
+	_ = os.Remove(pwd + "/aa-202301_0.csv")
+	_ = os.Remove(pwd + "/aa-202301_1.csv")
+
+}
diff --git a/biz/export/csv_test.go b/biz/export/csv_test.go
new file mode 100644
index 0000000..9bf882e
--- /dev/null
+++ b/biz/export/csv_test.go
@@ -0,0 +1,31 @@
+package export
+
+import (
+	"github.com/stretchr/testify/assert"
+	"os"
+	"testing"
+)
+
+func TestCsv_Write(t *testing.T) {
+	pwd, _ := os.Getwd()
+	e := NewCSV(pwd+"/aa-{begin}.csv", 5, map[string]string{"begin": "202301"})
+
+	e.Open()
+
+	e.WriteTitle([]string{"姓名", "年龄"})
+	data := make([]interface{}, 2)
+	data[0] = "张三"
+	for i := 0; i < 9; i++ {
+		data[1] = 10 + i
+		e.Write(data)
+	}
+	e.Close()
+
+	assert.FileExists(t, pwd+"/aa-202301_0.csv")
+	assert.FileExists(t, pwd+"/aa-202301_1.csv")
+	assert.NoFileExists(t, pwd+"/aa-202301_2.csv")
+
+	_ = os.Remove(pwd + "/aa-202301_0.csv")
+	_ = os.Remove(pwd + "/aa-202301_1.csv")
+
+}
diff --git a/biz/export/cvs.go b/biz/export/cvs.go
new file mode 100644
index 0000000..5e45509
--- /dev/null
+++ b/biz/export/cvs.go
@@ -0,0 +1,127 @@
+package export
+
+import (
+	"bufio"
+	"encoding/csv"
+	"errors"
+	"fmt"
+	"io"
+	"os"
+	"reflect"
+)
+
+type CSV struct {
+	fc     *os.File
+	csv    *csv.Writer
+	f      *File
+	isNew  bool
+	titles []string
+}
+
+func NewCSV(fileName string, limit int, param map[string]string) *CSV {
+	return &CSV{
+		f: NewFile(fileName, limit, param),
+	}
+}
+
+func (e *CSV) slice() {
+	if e.f.slice() {
+		e.reset()
+	}
+}
+
+func (e *CSV) SetParam(param map[string]string) {
+	e.f.param = param
+}
+
+func (e *CSV) reset() {
+	e.save()
+	e.f.NextFile()
+	e.Open()
+	e.WriteTitle(nil)
+	e.slice()
+}
+
+func (e *CSV) Open() error {
+	var err error
+	if e.f.IsFileExist() {
+		e.fc, err = os.OpenFile(e.f.FileName(), os.O_APPEND|os.O_WRONLY, 0644)
+		if err != nil {
+			return err
+		}
+		e.f.SetRow(e.getLineCount(e.fc))
+	} else {
+		e.isNew = true
+		e.fc, err = os.Create(e.f.FileName())
+	}
+
+	if err == nil {
+		e.csv = csv.NewWriter(e.fc)
+	}
+
+	return nil
+}
+
+func (e *CSV) getLineCount(file io.Reader) (line int) {
+	reader := bufio.NewReader(file)
+	line = 0
+	for {
+		_, isPrefix, err := reader.ReadLine()
+		if err != nil {
+			break
+		}
+		if !isPrefix {
+			line++
+		}
+	}
+	return line
+}
+
+func (e *CSV) save() error {
+	e.csv.Flush()
+	e.fc.Close()
+	return nil
+}
+
+func (e *CSV) WriteTitle(titles []string) error {
+
+	if titles != nil {
+		e.titles = titles
+	}
+
+	if e.titles != nil && e.isNew {
+		e.Write(e.titles)
+		e.isNew = false
+	}
+	return nil
+}
+
+func (e *CSV) Write(data interface{}) error {
+	if e.f.slice() {
+		e.reset()
+	}
+
+	v := reflect.ValueOf(data)
+	if v.Kind() == reflect.Ptr {
+		v = v.Elem()
+	}
+	if v.Kind() != reflect.Slice {
+		return errors.New("数据无效，不是切片类型")
+	}
+
+	switch val := data.(type) {
+	case []string:
+		return e.csv.Write(val)
+	case []interface{}:
+		strs := make([]string, len(val))
+		for i, v := range val {
+			strs[i] = fmt.Sprintf("%v", v)
+		}
+		return e.csv.Write(strs)
+	}
+	return nil
+}
+
+func (e *CSV) Close() error {
+	return e.save()
+}
diff --git a/biz/export/excel_exporter.go b/biz/export/excel_exporter.go
index 4ce9fae..d4924c0 100644
--- a/biz/export/excel_exporter.go
+++ b/biz/export/excel_exporter.go
@@ -1,6 +1,9 @@
 package export
 
-import "fmt"
+import (
+	"fmt"
+	"log"
+)
 
 type ExcelExporter struct {
 	mFetcher DataFetcher
@@ -33,15 +36,13 @@ func (ee *ExcelExporter) Export(sql, pk string) error {
 	//fmt.Printf("Excel Exporter.Excel, got %v data\n", len(data))
 	//ee.file.Open()
 	ee.file.WriteTitle(data.Title)
-	var last interface{}
+	var last []string
 	for _, val := range data.Data {
 		last = val
 		ee.file.Write(last)
 	}
 
-	if row, ok := last.([]interface{}); ok {
-		ee.last = row[getPkIndex(data.Title, pk)]
-	}
+	ee.last = last[getPkIndex(data.Title, pk)]
 
 	//ee.file.Close()
 	return nil
@@ -52,6 +53,10 @@ func (ee *ExcelExporter) Last() (int, interface{}) {
 }
 
 func getPkIndex(titles []string, pk string) int {
+	if pk == "" {
+		log.Println("pk is not empty")
+		return -1
+	}
 	for i, title := range titles {
 		if title == pk {
 			return i
diff --git a/biz/export/excel_exporter_test.go b/biz/export/excel_exporter_test.go
index 7e57b28..2c464b7 100644
--- a/biz/export/excel_exporter_test.go
+++ b/biz/export/excel_exporter_test.go
@@ -11,7 +11,9 @@ func TestExcelExporter_Export(t *testing.T) {
 	pwd, _ := os.Getwd()
 	file := NewExcel(pwd+"/aa-{begin}.xlsx", 5, map[string]string{"begin": "202301"})
 	e := NewExcelExporter(data, file)
-	err := e.Export("aa")
+	file.Open()
+	err := e.Export("aa", "字段1")
+	file.Close()
 	assert.Nil(t, err)
 
 	assert.FileExists(t, pwd+"/aa-202301_0.xlsx")
diff --git a/biz/export/export.go b/biz/export/export.go
index 1e4896a..ee971d7 100644
--- a/biz/export/export.go
+++ b/biz/export/export.go
@@ -10,7 +10,7 @@ type (
 
 	Data struct {
 		Title []string
-		Data  []interface{}
+		Data  [][]string
 	}
 	DataFetcher interface {
 		Fetch(sql string) (*Data, error)
diff --git a/biz/export/mysql_data_fetcher.go b/biz/export/mysql_data_fetcher.go
index 40ea077..d8e48f3 100644
--- a/biz/export/mysql_data_fetcher.go
+++ b/biz/export/mysql_data_fetcher.go
@@ -2,6 +2,7 @@ package export
 
 import (
 	"math/rand"
+	"strconv"
 )
 
 type MysqlDataFetcher struct {
@@ -9,9 +10,9 @@ type MysqlDataFetcher struct {
 }
 
 func (mf *MysqlDataFetcher) Fetch(sql string) (*Data, error) {
-	rows := make([]interface{}, 0, 6)
+	rows := make([][]string, 0, 6)
 	// 插入6个随机数组成的切片，模拟查询要返回的数据集
-	rows = append(rows, rand.Perm(5), rand.Perm(5), rand.Perm(5), rand.Perm(5), rand.Perm(5), rand.Perm(5))
+	rows = append(rows, row(), row(), row(), row(), row(), row())
 	return &Data{
 		Title: []string{"字段1", "字段2", "字段3", "字段4", "字段5"},
 		Data:  rows,
@@ -23,3 +24,12 @@ func NewMysqlDataFetcher(configStr string) DataFetcher {
 		Config: configStr,
 	}
 }
+
+func row() []string {
+	strs := make([]string, 5)
+	nums := rand.Perm(5)
+	for i, num := range nums {
+		strs[i] = strconv.Itoa(num)
+	}
+	return strs
+}
diff --git a/cmd/cmd/export.go b/cmd/cmd/excel.go
similarity index 56%
rename from cmd/cmd/export.go
rename to cmd/cmd/excel.go
index 4c7f389..e0d2a56 100644
--- a/cmd/cmd/export.go
+++ b/cmd/cmd/excel.go
@@ -9,8 +9,18 @@ import (
 	"time"
 )
 
-func Export(conf *config.Config, sysName, jobName string, begin, end time.Time, batch int) error {
-	job, dbStr, err := GetJob(conf, sysName, jobName)
+type Excel struct {
+	conf *config.Config
+}
+
+func NewExcel(conf *config.Config) *Excel {
+	return &Excel{
+		conf: conf,
+	}
+}
+
+func (e *Excel) Export(sysName, jobName string, begin, end time.Time, batch int) error {
+	job, dbStr, err := config.GetJob(e.conf, sysName, jobName)
 	if err != nil {
 		return err
 	}
@@ -20,27 +30,14 @@ func Export(conf *config.Config, sysName, jobName string, begin, end time.Time,
 		return err
 	}
 
-	return JobHandler(job, d, map[string]interface{}{
+	return e.JobHandler(job, d, map[string]interface{}{
 		"begin": begin,
 		"end":   end,
 		"last":  0,
 	}, batch)
 }
 
-func GetJob(conf *config.Config, sys, job string) (config.Job, string, error) {
-	s, err := conf.GetSystem(sys)
-	if err != nil {
-		return config.Job{}, "", err
-	}
-
-	j, err := s.GetJob(job)
-	if err != nil {
-		return config.Job{}, "", err
-	}
-	return j, s.Db, nil
-}
-
-func JobHandler(job config.Job, d export.DataFetcher, params map[string]interface{}, batch int) error {
+func (e *Excel) JobHandler(job config.Job, d export.DataFetcher, params map[string]interface{}, batch int) error {
 
 	f := export.NewExcel(job.File, job.Size, map[string]string{
 		"begin": params["begin"].(time.Time).Format("20060102"),
@@ -49,16 +46,17 @@ func JobHandler(job config.Job, d export.DataFetcher, params map[string]interfac
 
 	for i, task := range job.Tasks {
 		fmt.Printf("执行导出任务：%d\n", i+1)
-		if err := TaskExport(d, task, params, f, batch); err != nil {
+		if err := e.TaskExport(d, task, params, f, batch); err != nil {
 			return err
 		}
 	}
 	return nil
 }
 
-func TaskExport(d export.DataFetcher, t config.Task, params map[string]interface{}, f export.FileAdapter, batch int) error {
+func (e *Excel) TaskExport(d export.DataFetcher, t config.Task, params map[string]interface{}, f export.FileAdapter, batch int) error {
 	//todo 最多分1000个批次进行处理
 	f.Open()
+	defer f.Close()
 	for i := 0; i < 1000; i++ {
 		sql := t.GetSql(params)
 
@@ -75,6 +73,5 @@ func TaskExport(d export.DataFetcher, t config.Task, params map[string]interface
 		params["last"] = last
 		time.Sleep(time.Microsecond * 30)
 	}
-	f.Close()
 	return nil
 }
diff --git a/cmd/cmd/root.go b/cmd/cmd/root.go
index 9bee009..4698946 100644
--- a/cmd/cmd/root.go
+++ b/cmd/cmd/root.go
@@ -86,7 +86,8 @@ func rootRun(cmd *cobra.Command, args []string) {
 		CmdError(cmd, "无效的参数：%s", err.Error())
 	}
 	b := time.Now()
-	Export(config.DefaultConfig, sName, jName, begin, end, batch)
+	ee := NewExcel(config.DefaultConfig)
+	ee.Export(sName, jName, begin, end, batch)
 	e := time.Now()
 	fmt.Println("耗时：" + e.Sub(b).String())