feat: put out es bulk err

v0.0.3 beta

.github/workflows/build.yml

@@ -45,7 +45,6 @@ jobs:
         with:
           repo_token: "${{ secrets.GITHUB_TOKEN }}"
           title: "Release_${{ github.ref_name }}"
-          prerelease: false
           files: |
             dist/esgo2dump_${{ github.ref_name }}_linux_amd64
             dist/esgo2dump_${{ github.ref_name }}_linux_arm64

internal/cmd/cmd.go

@@ -21,9 +21,7 @@ esgo2dump --input=http://127.0.0.1:9200/some_index --output=http://192.168.1.1:9
 
 esgo2dump --input=https://username:password@127.0.0.1:9200/some_index --output=./data.json
 
-esgo2dump --input=http://127.0.0.1:9200/some_index --output=./data.json --query='{"match": {"name": "some_name"}}'
+esgo2dump --input=http://127.0.0.1:9200/some_index --output=./data.json --query='{"match": {"name": "some_name"}}'`,
-
-esgo2dump --input=http://127.0.0.1:9200/some_index --output=./data.json --query_file=my_queries.json`,
 	}
 
 	f_input  string
@@ -31,19 +29,16 @@ esgo2dump --input=http://127.0.0.1:9200/some_index --output=./data.json --query_
 	f_limit  int
 	f_type   string
 	f_query  string
-
-	f_query_file string
 )
 
 func init() {
 	rootCommand.Flags().BoolVar(&opt.Debug, "debug", false, "")
 	rootCommand.Flags().IntVar(&opt.Timeout, "timeout", 30, "max timeout seconds per operation with limit")
 
-	rootCommand.Flags().StringVarP(&f_input, "input", "i", "", "*required: input file or es url (example :data.json / http://127.0.0.1:9200/my_index)")
+	rootCommand.Flags().StringVarP(&f_input, "input", "i", "http://127.0.0.1:9200/my_index", "")
 	rootCommand.Flags().StringVarP(&f_output, "output", "o", "output.json", "")
 	rootCommand.Flags().StringVarP(&f_type, "type", "t", "data", "data/mapping/setting")
 	rootCommand.Flags().StringVarP(&f_query, "query", "q", "", `query dsl, example: {"bool":{"must":[{"term":{"name":{"value":"some_name"}}}],"must_not":[{"range":{"age":{"gte":18,"lt":60}}}]}}`)
-	rootCommand.Flags().StringVar(&f_query_file, "query_file", "", `query json file (will execute line by line)`)
 	rootCommand.Flags().IntVarP(&f_limit, "limit", "l", 100, "")
 }
 

internal/cmd/run.go

@@ -1,10 +1,11 @@
 package cmd
 
 import (
-	"bufio"
 	"context"
 	"encoding/json"
+	"errors"
 	"fmt"
+	"io"
 	"net/url"
 	"os"
 
@@ -16,29 +17,6 @@ import (
 	"github.com/spf13/cobra"
 )
 
-func check(cmd *cobra.Command) error {
-	if f_input == "" {
-		return cmd.Help()
-		//return fmt.Errorf("must specify input(example: data.json/http://127.0.0.1:9200/my_index)")
-	}
-
-	if f_limit == 0 || f_limit > 10000 {
-		return fmt.Errorf("invalid limit(1 - 10000)")
-	}
-
-	if f_query != "" && f_query_file != "" {
-		return fmt.Errorf("cannot specify both query and query_file at the same time")
-	}
-
-	switch f_type {
-	case "data", "mapping", "setting":
-	default:
-		return fmt.Errorf("unknown type=%s", f_type)
-	}
-
-	return nil
-}
-
 func run(cmd *cobra.Command, args []string) error {
 	var (
 		err error
@@ -50,8 +28,14 @@ func run(cmd *cobra.Command, args []string) error {
 		logrus.SetLevel(logrus.DebugLevel)
 	}
 
-	if err = check(cmd); err != nil {
+	if f_limit == 0 || f_limit > 10000 {
-		return err
+		return fmt.Errorf("invalid limit(1 - 10000)")
+	}
+
+	switch f_type {
+	case "data", "mapping", "setting":
+	default:
+		return fmt.Errorf("unknown type=%s", f_type)
 	}
 
 	if ioi, err = newIO(f_input, interfaces.IOInput); err != nil {
@@ -67,19 +51,9 @@ func run(cmd *cobra.Command, args []string) error {
 		_ = ioo.Close()
 	}()
 
-	if (f_query_file != "" || f_query != "") && ioi.IsFile() {
-		return fmt.Errorf("with file input, query or query_file can't be supported")
-	}
-
 	switch f_type {
 	case "data":
-		if err = executeData(cmd.Context(), ioi, ioo); err != nil {
+		return executeData(cmd.Context(), ioi, ioo)
-			return err
-		}
-
-		logrus.Info("Dump: write data succeed!!!")
-
-		return nil
 	case "mapping":
 		var mapping map[string]any
 		if mapping, err = ioi.ReadMapping(cmd.Context()); err != nil {
@@ -114,126 +88,33 @@ func run(cmd *cobra.Command, args []string) error {
 func executeData(ctx context.Context, input, output interfaces.DumpIO) error {
 	var (
 		err     error
-		ch      = make(chan []*interfaces.ESSource, 1)
-		errCh   = make(chan error)
-		queries = make([]map[string]any, 0)
-	)
-
-	if f_query != "" {
-		query := make(map[string]any)
-		if err = json.Unmarshal([]byte(f_query), &query); err != nil {
-			return fmt.Errorf("invalid query err=%v", err)
-		}
-
-		queries = append(queries, query)
-	}
-
-	if f_query_file != "" {
-		var (
-			qf *os.File
-		)
-
-		if qf, err = os.Open(f_query_file); err != nil {
-			return fmt.Errorf("open query_file err=%v", err)
-		}
-
-		defer func() {
-			_ = qf.Close()
-		}()
-
-		scanner := bufio.NewScanner(qf)
-		lineCount := 1
-		for scanner.Scan() {
-			line := scanner.Text()
-			oq := make(map[string]any)
-			if err = json.Unmarshal([]byte(line), &oq); err != nil {
-				return fmt.Errorf("query file line=%d invalid err=%v", lineCount, err)
-			}
-
-			queries = append(queries, oq)
-
-			if len(queries) > 10000 {
-				return fmt.Errorf("query_file support max lines=%d", 10000)
-			}
-
-			lineCount++
-		}
-
-	}
-
-	if len(queries) == 0 {
-		queries = append(queries, nil)
-	}
-
-	go func(c context.Context) {
-		var (
 		lines   []*interfaces.ESSource
-		)
-
-		defer func() {
-			close(ch)
-		}()
-
-	Loop:
-		for _, query := range queries {
-			for {
-				select {
-				case <-c.Done():
-					return
-				default:
-					if lines, err = input.ReadData(c, f_limit, query); err != nil {
-						errCh <- err
-						return
-					}
-
-					logrus.Debugf("executeData: input read_data got lines=%d", len(lines))
-
-					if len(lines) == 0 {
-						input.ResetOffset()
-						continue Loop
-					}
-
-					ch <- lines
-				}
-			}
-		}
-	}(ctx)
-
-	var (
 		succeed int
-		total   int
-		docs    []*interfaces.ESSource
-		ok      bool
 	)
 
 	for {
-		select {
-		case <-ctx.Done():
-		case err = <-errCh:
-			return err
-		case docs, ok = <-ch:
-			if !ok {
-				return err
-			}
 
-			if len(docs) == 0 {
+		if lines, err = input.ReadData(ctx, f_limit); err != nil {
+			if errors.Is(err, io.EOF) {
 				return nil
 			}
 
-			if succeed, err = output.WriteData(ctx, docs); err != nil {
 			return err
 		}
 
-			logrus.Debugf("executeData: output write_data succeed lines=%d", succeed)
+		if len(lines) == 0 {
-
+			return nil
-			if succeed != len(docs) {
-				return fmt.Errorf("cmd.run: got lines=%d, only succeed=%d", len(docs), succeed)
 		}
 
-			total += succeed
+		if succeed, err = output.WriteData(ctx, lines); err != nil {
-
+			return err
-			logrus.Infof("Dump: succeed=%d total=%d docs succeed!!!", succeed, total)
 		}
+
+		if succeed != len(lines) {
+			return fmt.Errorf("cmd.run: got lines=%d, only succeed=%d", len(lines), succeed)
+		}
+
+		logrus.Infof("Dump: %d docs succeed!!!", succeed)
 	}
 }
 
@@ -271,7 +152,7 @@ func newIO(source string, ioType interfaces.IO) (interfaces.DumpIO, error) {
 
 	logrus.Debugf("newIO.%s: source as url=%+v", ioType.Code(), *iurl)
 
-	return xes.NewClient(iurl, ioType)
+	return xes.NewClient(iurl, ioType, qm)
 
 ClientByFile:
 	if ioType == interfaces.IOOutput {

internal/interfaces/dumpio.go

@@ -3,11 +3,9 @@ package interfaces
 import "context"
 
 type DumpIO interface {
-	ReadData(context.Context, int, map[string]any) ([]*ESSource, error)
+	ReadData(context.Context, int) ([]*ESSource, error)
 	WriteData(ctx context.Context, docs []*ESSource) (int, error)
 
-	ResetOffset()
-
 	ReadMapping(context.Context) (map[string]any, error)
 	WriteMapping(context.Context, map[string]any) error
 

internal/opt/var.go

@@ -7,7 +7,4 @@ const (
 var (
 	Debug   bool
 	Timeout int
-
-	BuffSize    = 5 * 1024 * 1024   // 5M
-	MaxBuffSize = 100 * 1024 * 1024 // 100M, default elastic_search doc max size
 )

internal/xes/xes.go

@@ -21,40 +21,34 @@ import (
 	"github.com/sirupsen/logrus"
 )
 
-func NewClient(url *url.URL, iot interfaces.IO) (interfaces.DumpIO, error) {
+func NewClient(url *url.URL, iot interfaces.IO, qm map[string]any) (interfaces.DumpIO, error) {
 
 	var (
-		address     = fmt.Sprintf("%s://%s", url.Scheme, url.Host)
+		err      error
-		urlIndex    = strings.TrimPrefix(url.Path, "/")
+		endpoint = fmt.Sprintf("%s://%s", url.Scheme, url.Host)
-		urlUsername string
+		c        *elastic.Client
-		urlPassword string
+		infoResp *esapi.Response
-		errCh       = make(chan error)
+		index    = strings.TrimPrefix(url.Path, "/")
-		cliCh       = make(chan *elastic.Client)
+		username string
+		password string
 	)
 
 	if url.User != nil {
-		urlUsername = url.User.Username()
+		username = url.User.Username()
 		if p, ok := url.User.Password(); ok {
-			urlPassword = p
+			password = p
 		}
 	}
 
-	logrus.Debugf("xes.NewClient: endpoint=%s index=%s (username=%s password=%s)", address, urlIndex, urlUsername, urlPassword)
+	logrus.Debugf("xes.NewClient: endpoint=%s index=%s (username=%s password=%s)", endpoint, index, username, password)
 
-	if urlIndex == "" {
+	if index == "" {
 		return nil, fmt.Errorf("please specify index name: (like => http://127.0.0.1:9200/my_index)")
 	}
 
-	ncFunc := func(endpoints []string, username, password, index string) {
+	if c, err = elastic.NewClient(
-		var (
-			err      error
-			cli      *elastic.Client
-			infoResp *esapi.Response
-		)
-
-		if cli, err = elastic.NewClient(
 		elastic.Config{
-				Addresses:     endpoints,
+			Addresses:     []string{endpoint},
 			Username:      username,
 			Password:      password,
 			CACert:        nil,
@@ -63,48 +57,33 @@ func NewClient(url *url.URL, iot interfaces.IO) (interfaces.DumpIO, error) {
 			RetryBackoff:  nil,
 			Transport: &http.Transport{
 				TLSClientConfig: &tls.Config{InsecureSkipVerify: true},
-					DialContext:     (&net.Dialer{Timeout: 10 * time.Second}).DialContext,
+				DialContext:     (&net.Dialer{Timeout: 5 * time.Second}).DialContext,
 			},
 		},
 	); err != nil {
-			logrus.Debugf("xes.NewClient: elastic new client with endpont=%s err=%v", endpoints, err)
+		logrus.Debugf("xes.NewClient: elastic new client with endpont=%s err=%v", endpoint, err)
-			errCh <- err
+		return nil, err
-			return
 	}
 
-		if infoResp, err = cli.Info(); err != nil {
+	if infoResp, err = c.Info(); err != nil {
 		logrus.Debugf("xes.NewClient: ping err=%v", err)
-			errCh <- err
+		return nil, err
-			return
 	}
 
 	if infoResp.StatusCode != 200 {
-			err = fmt.Errorf("info xes status=%d", infoResp.StatusCode)
+		return nil, fmt.Errorf("info xes status=%d", infoResp.StatusCode)
-			logrus.Debugf("xes.NewClient: status err=%v", err)
-			errCh <- err
-			return
 	}
 
-		cliCh <- cli
+	return &client{c: c, index: index, queryMap: qm, iot: iot}, nil
-	}
-
-	go ncFunc([]string{address}, urlUsername, urlPassword, urlIndex)
-
-	select {
-	case <-util.Timeout(10).Done():
-		return nil, fmt.Errorf("dial es=%s err=%v", address, context.DeadlineExceeded)
-	case c := <-cliCh:
-		return &client{c: c, index: urlIndex, iot: iot}, nil
-	case e := <-errCh:
-		return nil, e
-	}
 }
 
 type client struct {
 	c        *elastic.Client
 	iot      interfaces.IO
 	index    string
+	from     int
 	scrollId string
+	queryMap map[string]any
 }
 
 func (c *client) checkResponse(r *esapi.Response) error {
@@ -127,9 +106,6 @@ func (c *client) Close() error {
 	return nil
 }
 
-func (c *client) ResetOffset() {
-	c.scrollId = ""
-}
 func (c *client) WriteData(ctx context.Context, docs []*interfaces.ESSource) (int, error) {
 	var (
 		err     error
@@ -152,6 +128,8 @@ func (c *client) WriteData(ctx context.Context, docs []*interfaces.ESSource) (in
 			return 0, err
 		}
 
+		logrus.Debugf("xes.Write: doc content=%s", string(bs))
+
 		if err = indexer.Add(context.Background(), esutil.BulkIndexerItem{
 			Action:     "index",
 			Index:      c.index,
@@ -182,7 +160,7 @@ func (c *client) WriteData(ctx context.Context, docs []*interfaces.ESSource) (in
 	return count, nil
 }
 
-func (c *client) ReadData(ctx context.Context, i int, query map[string]any) ([]*interfaces.ESSource, error) {
+func (c *client) ReadData(ctx context.Context, i int) ([]*interfaces.ESSource, error) {
 	var (
 		err    error
 		resp   *esapi.Response
@@ -198,8 +176,8 @@ func (c *client) ReadData(ctx context.Context, i int, query map[string]any) ([]*
 			c.c.Search.WithScroll(time.Duration(opt.ScrollDurationSeconds) * time.Second),
 		}
 
-		if query != nil && len(query) > 0 {
+		if len(c.queryMap) > 0 {
-			queryBs, _ := json.Marshal(map[string]any{"query": query})
+			queryBs, _ := json.Marshal(map[string]any{"query": c.queryMap})
 			qs = append(qs, c.c.Search.WithBody(bytes.NewReader(queryBs)))
 		}
 

internal/xes/xes_test.go

@@ -1,9 +1,6 @@
 package xes
 
 import (
-	"bufio"
-	"fmt"
-	"os"
 	"testing"
 
 	elastic "github.com/elastic/go-elasticsearch/v7"
@@ -40,68 +37,3 @@ func TestGetESMapping(t *testing.T) {
 
 	t.Log("get source:", r.String())
 }
-
-func TestScanWithInterrupt(t *testing.T) {
-	filename := "test_scan.txt"
-	f, err := os.OpenFile(filename, os.O_RDWR|os.O_CREATE, 0644)
-	if err != nil {
-		t.Error(1, err)
-		return
-	}
-	defer func() {
-		os.Remove(filename)
-	}()
-	f.WriteString(`line 01
-line 02
-line 03
-line 04
-line 05
-line 06
-line 07
-line 08
-line 09
-line 10
-line 11
-line 12
-line 13
-line 14
-line 15`)
-	f.Close()
-
-	of, err := os.Open(filename)
-	if err != nil {
-		t.Error(2, err)
-		return
-	}
-
-	scanner := bufio.NewScanner(of)
-
-	count := 0
-	for scanner.Scan() {
-		text := scanner.Text()
-		fmt.Printf("[line: %2d] = %s\n", count, text)
-		count++
-
-		if count > 5 {
-			break
-		}
-	}
-
-	count = 0
-	for scanner.Scan() {
-		text := scanner.Text()
-		fmt.Printf("[line: %2d] = %s\n", count, text)
-		count++
-
-		if count > 5 {
-			break
-		}
-	}
-
-	count = 0
-	for scanner.Scan() {
-		text := scanner.Text()
-		fmt.Printf("[line: %2d] = %s\n", count, text)
-		count++
-	}
-}

internal/xfile/xfile.go

@@ -4,11 +4,11 @@ import (
 	"bufio"
 	"context"
 	"encoding/json"
-	"github.com/loveuer/esgo2dump/internal/opt"
 	"io"
 	"os"
 
 	"github.com/loveuer/esgo2dump/internal/interfaces"
+	"github.com/sirupsen/logrus"
 )
 
 type client struct {
@@ -85,8 +85,6 @@ func (c *client) IsFile() bool {
 	return true
 }
 
-func (c *client) ResetOffset() {}
-
 func (c *client) WriteData(ctx context.Context, docs []*interfaces.ESSource) (int, error) {
 	var (
 		err   error
@@ -111,7 +109,7 @@ func (c *client) WriteData(ctx context.Context, docs []*interfaces.ESSource) (in
 	return count, nil
 }
 
-func (c *client) ReadData(ctx context.Context, i int, _ map[string]any) ([]*interfaces.ESSource, error) {
+func (c *client) ReadData(ctx context.Context, i int) ([]*interfaces.ESSource, error) {
 	var (
 		err   error
 		count = 0
@@ -121,6 +119,8 @@ func (c *client) ReadData(ctx context.Context, i int, _ map[string]any) ([]*inte
 	for c.scanner.Scan() {
 		line := c.scanner.Text()
 
+		logrus.Debugf("xfile.Read: line=%s", line)
+
 		item := new(interfaces.ESSource)
 		if err = json.Unmarshal([]byte(line), item); err != nil {
 			return list, err
@@ -150,8 +150,6 @@ func NewClient(file *os.File, ioType interfaces.IO) (interfaces.DumpIO, error) {
 
 	if ioType == interfaces.IOInput {
 		c.scanner = bufio.NewScanner(c.f)
-		buf := make([]byte, opt.BuffSize)
-		c.scanner.Buffer(buf, opt.MaxBuffSize)
 	}
 
 	return c, nil

readme.md

@@ -28,16 +28,7 @@ esgo2dump --input=http://127.0.0.1:9200/some_index --output=http://192.168.1.1:9
 
 esgo2dump --input=https://username:password@127.0.0.1:9200/some_index --output=./data.json
 
-esgo2dump --input=http://127.0.0.1:9200/some_index --output=./data.json --query='{"match": {"name": "some_name"}}'
+esgo2dump --input=http://127.0.0.1:9200/some_index --output=./data.json --query='{"match": {"name": "some_name"}}'`,
-
-esgo2dump --input=http://127.0.0.1:9200/some_index --output=./data.json --query_file=my_queries.json
-```
-
-- example_queries.json
-```json
-{"bool":{"should":[{"term":{"user_id":{"value":"123"}}},{"term":{"user_id":{"value":"456"}}}]}}
-{"bool":{"should":[{"term":{"user_id":{"value":"abc"}}},{"term":{"user_id":{"value":"def"}}}]}}
-{"bool":{"should":[{"term":{"user_id":{"value":"ABC"}}},{"term":{"user_id":{"value":"DEF"}}}]}}
 ```
 
 ### roadmap