wip: mapping,setting import

wip: mapping type
feat: dump from es with query, limit
2024-03-25 20:22:18 +08:00 · 2024-03-24 20:47:54 +08:00 · 2024-03-23 21:25:19 +08:00 · 2024-03-22 18:05:47 +08:00
15 changed files with 151 additions and 646 deletions
@@ -1,58 +0,0 @@
-name: Auto Build
-on:
-  push:
-    tags:
-      - 'v*'
-
-jobs:
-  build-job:
-    runs-on: ubuntu-latest
-    permissions:
-      id-token: write
-      contents: write
-      pull-requests: write
-      repository-projects: write
-    steps:
-      - name: checkout repository
-        uses: actions/checkout@v4
-
-      - name: fill version
-        run: sed -i -E "s/v[0-9]+.[0-9]+.[0-9]+/${{ github.ref_name }}/g" internal/opt/version.go
-
-      - name: install golang
-        uses: actions/setup-go@v4
-        with:
-          go-version: '1.18'
-
-      - name: build linux amd64
-        run: CGO_ENABLE=0 GOOS=linux GOARCH=amd64 go build -ldflags='-s -w' -o dist/esgo2dump_${{ github.ref_name }}_linux_amd64 .
-
-      - name: build linux arm64
-        run: CGO_ENABLE=0 GOOS=linux GOARCH=arm64 go build -ldflags='-s -w' -o dist/esgo2dump_${{ github.ref_name }}_linux_arm64 .
-
-      - name: build windows amd64
-        run: CGO_ENABLE=0 GOOS=windows GOARCH=amd64 go build -ldflags='-s -w' -o dist/esgo2dump_${{ github.ref_name }}_windows_amd64.exe .
-
-      - name: build windows arm64
-        run: CGO_ENABLE=0 GOOS=windows GOARCH=arm64 go build -ldflags='-s -w' -o dist/esgo2dump_${{ github.ref_name }}_windows_arm64.exe .
-
-      - name: build darwin amd64
-        run: CGO_ENABLE=0 GOOS=darwin GOARCH=amd64 go build -ldflags='-s -w' -o dist/esgo2dump_${{ github.ref_name }}_darwin_amd64 .
-
-      - name: build darwin arm64
-        run: CGO_ENABLE=0 GOOS=darwin GOARCH=arm64 go build -ldflags='-s -w' -o dist/esgo2dump_${{ github.ref_name }}_darwin_arm64 .
-
-      - name: create releases
-        id: create_releases
-        uses: "marvinpinto/action-automatic-releases@latest"
-        with:
-          repo_token: "${{ secrets.GITHUB_TOKEN }}"
-          title: "Release_${{ github.ref_name }}"
-          prerelease: false
-          files: |
-            dist/esgo2dump_${{ github.ref_name }}_linux_amd64
-            dist/esgo2dump_${{ github.ref_name }}_linux_arm64
-            dist/esgo2dump_${{ github.ref_name }}_windows_amd64.exe
-            dist/esgo2dump_${{ github.ref_name }}_windows_arm64.exe
-            dist/esgo2dump_${{ github.ref_name }}_darwin_amd64
-            dist/esgo2dump_${{ github.ref_name }}_darwin_arm64
@@ -5,5 +5,4 @@ data.json
 mapping.json
 setting.json
 output.json
-*.txt
-dist
+*.txt
@@ -1,6 +1,6 @@
-module github.com/loveuer/esgo2dump
+module esgo2dump

-go 1.18
+go 1.20

 require (
 	github.com/elastic/go-elasticsearch/v7 v7.17.10
@@ -2,8 +2,7 @@ package cmd

 import (
 	"context"
-
-	"github.com/loveuer/esgo2dump/internal/opt"
+	"esgo2dump/internal/opt"
 	"github.com/spf13/cobra"
 )

@@ -17,13 +16,9 @@ var (
 		Example: `
 esgo2dump --input=http://127.0.0.1:9200/some_index --output=./data.json

-esgo2dump --input=http://127.0.0.1:9200/some_index --output=http://192.168.1.1:9200/some_index --limit=5000
-
 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_file=my_queries.json`,
+esgo2dump --input=http://127.0.0.1:9200/some_index --output=./data.json --query='{"match": {"name": "some_name"}}'`,
 	}

 	f_input  string
@@ -31,22 +26,16 @@ 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
-
-	f_version bool
 )

 func init() {
 	rootCommand.Flags().BoolVar(&opt.Debug, "debug", false, "")
-	rootCommand.Flags().BoolVarP(&f_version, "version", "v", false, "print esgo2dump version")
 	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, "")
 }

@@ -1,44 +1,21 @@
 package cmd

 import (
-	"bufio"
 	"context"
 	"encoding/json"
+	"errors"
+	"esgo2dump/internal/interfaces"
+	"esgo2dump/internal/opt"
+	"esgo2dump/internal/xes"
+	"esgo2dump/internal/xfile"
 	"fmt"
-	"net/url"
-	"os"
-
-	"github.com/loveuer/esgo2dump/internal/interfaces"
-	"github.com/loveuer/esgo2dump/internal/opt"
-	"github.com/loveuer/esgo2dump/internal/xes"
-	"github.com/loveuer/esgo2dump/internal/xfile"
 	"github.com/sirupsen/logrus"
 	"github.com/spf13/cobra"
+	"io"
+	"net/url"
+	"os"
 )

-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
@@ -50,13 +27,10 @@ func run(cmd *cobra.Command, args []string) error {
 		logrus.SetLevel(logrus.DebugLevel)
 	}

-	if f_version {
-		logrus.Infof("esgo2dump (Version: %s)", opt.Version)
-		return nil
-	}
-
-	if err = check(cmd); err != nil {
-		return err
+	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 {
@@ -72,45 +46,23 @@ 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 err
-		}
-
-		logrus.Info("Dump: write data succeed!!!")
-
-		return nil
+		return executeData(cmd.Context(), ioi, ioo)
 	case "mapping":
 		var mapping map[string]any
-		if mapping, err = ioi.ReadMapping(cmd.Context()); err != nil {
+		if mapping, err = ioi.ReadMapping(); err != nil {
 			return err
 		}

-		if err = ioo.WriteMapping(cmd.Context(), mapping); err != nil {
-			return err
-		}
-
-		logrus.Info("Dump: write mapping succeed!!!")
-
-		return nil
+		return ioo.WriteMapping(mapping)
 	case "setting":
 		var setting map[string]any
-		if setting, err = ioi.ReadSetting(cmd.Context()); err != nil {
+		if setting, err = ioi.ReadSetting(); err != nil {
 			return err
 		}

-		if err = ioo.WriteSetting(cmd.Context(), setting); err != nil {
-			return err
-		}
-
-		logrus.Info("Dump: write setting succeed!!!")
-
-		return nil
+		return ioo.WriteSetting(setting)
 	default:
 		return fmt.Errorf("unknown type=%s", f_type)
 	}
@@ -119,131 +71,38 @@ 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)
-		scanner.Buffer(make([]byte, 1*1024*1024), 5*1024*1024)
-		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()
-						if query != nil {
-							bs, _ := json.Marshal(query)
-							logrus.Infof("Dump: query_file query=%s read done!!!", string(bs))
-						}
-						continue Loop
-					}
-
-					ch <- lines
-				}
-			}
-		}
-	}(ctx)
-
-	var (
+		lines   []*interfaces.ESSource
 		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 {
+		default:
+
+			if lines, err = input.ReadData(f_limit); err != nil {
+				if errors.Is(err, io.EOF) {
+					return nil
+				}
+
 				return err
 			}

-			if len(docs) == 0 {
+			if len(lines) == 0 {
 				return nil
 			}

-			if succeed, err = output.WriteData(ctx, docs); err != nil {
+			if succeed, err = output.WriteData(lines); err != nil {
 				return err
 			}

-			logrus.Debugf("executeData: output write_data succeed lines=%d", succeed)
-
-			if succeed != len(docs) {
-				return fmt.Errorf("cmd.run: got lines=%d, only succeed=%d", len(docs), succeed)
+			if succeed != len(lines) {
+				return fmt.Errorf("cmd.run: got lines=%d, only succeed=%d", len(lines), succeed)
 			}

-			total += succeed
-
-			logrus.Infof("Dump: succeed=%d total=%d docs succeed!!!", succeed, total)
+			logrus.Infof("Dump: %d docs succeed!!!", succeed)
 		}
+
 	}
 }

@@ -281,7 +140,7 @@ 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, qm)

 ClientByFile:
 	if ioType == interfaces.IOOutput {
@@ -1,21 +1,14 @@
 package interfaces

-import "context"
-
 type DumpIO interface {
-	ReadData(context.Context, int, map[string]any) ([]*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
-
-	ReadSetting(ctx context.Context) (map[string]any, error)
-	WriteSetting(context.Context, map[string]any) error
-
+	ReadData(int) ([]*ESSource, error)
+	ReadMapping() (map[string]any, error)
+	ReadSetting() (map[string]any, error)
+	WriteData(docs []*ESSource) (int, error)
+	WriteMapping(map[string]any) error
+	WriteSetting(map[string]any) error
 	Close() error

-	IOType() IO
+	IsInput() bool
 	IsFile() bool
 }
@@ -25,9 +25,3 @@ type ESResponse struct {
 		Hits     []*ESSource `json:"hits"`
 	} `json:"hits"`
 }
-
-type ESMapping map[string]struct {
-	Mappings struct {
-		Properties map[string]any `json:"properties"`
-	} `json:"mappings"`
-}
@@ -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
 )
@@ -1,3 +0,0 @@
-package opt
-
-const Version = "v0.1.2"
@@ -15,14 +15,3 @@ func Timeout(seconds ...int) context.Context {

 	return ctx
 }
-
-func TimeoutCtx(ctx context.Context, seconds ...int) context.Context {
-	second := 30
-	if len(seconds) > 0 && seconds[0] > 0 {
-		second = seconds[0]
-	}
-
-	timeout, _ := context.WithTimeout(ctx, time.Duration(second)*time.Second)
-
-	return timeout
-}
@@ -5,118 +5,112 @@ import (
 	"context"
 	"crypto/tls"
 	"encoding/json"
+	"esgo2dump/internal/interfaces"
+	"esgo2dump/internal/opt"
+	"esgo2dump/internal/util"
 	"fmt"
-	"net"
+	elastic "github.com/elastic/go-elasticsearch/v7"
+	"github.com/elastic/go-elasticsearch/v7/esapi"
+	"github.com/elastic/go-elasticsearch/v7/esutil"
+	"github.com/sirupsen/logrus"
 	"net/http"
 	"net/url"
 	"strings"
 	"time"
-
-	elastic "github.com/elastic/go-elasticsearch/v7"
-	"github.com/elastic/go-elasticsearch/v7/esapi"
-	"github.com/elastic/go-elasticsearch/v7/esutil"
-	"github.com/loveuer/esgo2dump/internal/interfaces"
-	"github.com/loveuer/esgo2dump/internal/opt"
-	"github.com/loveuer/esgo2dump/internal/util"
-	"github.com/sirupsen/logrus"
 )

-func NewClient(url *url.URL, iot interfaces.IO) (interfaces.DumpIO, error) {
+func NewClient(url *url.URL, qm map[string]any) (interfaces.DumpIO, error) {

 	var (
-		address     = fmt.Sprintf("%s://%s", url.Scheme, url.Host)
-		urlIndex    = strings.TrimPrefix(url.Path, "/")
-		urlUsername string
-		urlPassword string
-		errCh       = make(chan error)
-		cliCh       = make(chan *elastic.Client)
+		err      error
+		endpoint = fmt.Sprintf("%s://%s", url.Scheme, url.Host)
+		c        *elastic.Client
+		infoResp *esapi.Response
+		index    = strings.TrimPrefix(url.Path, "/")
+		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) {
-		var (
-			err      error
-			cli      *elastic.Client
-			infoResp *esapi.Response
-		)
-
-		if cli, err = elastic.NewClient(
-			elastic.Config{
-				Addresses:     endpoints,
-				Username:      username,
-				Password:      password,
-				CACert:        nil,
-				RetryOnStatus: []int{429},
-				MaxRetries:    3,
-				RetryBackoff:  nil,
-				Transport: &http.Transport{
-					TLSClientConfig: &tls.Config{InsecureSkipVerify: true},
-					DialContext:     (&net.Dialer{Timeout: 10 * time.Second}).DialContext,
-				},
+	if c, err = elastic.NewClient(
+		elastic.Config{
+			Addresses:     []string{endpoint},
+			Username:      username,
+			Password:      password,
+			CACert:        nil,
+			RetryOnStatus: []int{429},
+			MaxRetries:    3,
+			RetryBackoff:  nil,
+			Transport: &http.Transport{
+				TLSClientConfig: &tls.Config{InsecureSkipVerify: true},
 			},
-		); err != nil {
-			logrus.Debugf("xes.NewClient: elastic new client with endpont=%s err=%v", endpoints, err)
-			errCh <- err
-			return
-		}
-
-		if infoResp, err = cli.Info(); err != nil {
-			logrus.Debugf("xes.NewClient: ping err=%v", err)
-			errCh <- err
-			return
-		}
-
-		if infoResp.StatusCode != 200 {
-			err = fmt.Errorf("info xes status=%d", infoResp.StatusCode)
-			logrus.Debugf("xes.NewClient: status err=%v", err)
-			errCh <- err
-			return
-		}
-
-		cliCh <- cli
+		},
+	); err != nil {
+		logrus.Debugf("xes.NewClient: elastic new client with endpont=%s err=%v", endpoint, err)
+		return nil, err
 	}

-	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
+	if infoResp, err = c.Info(); err != nil {
+		logrus.Debugf("xes.NewClient: ping err=%v", err)
+		return nil, err
 	}
+
+	if infoResp.StatusCode != 200 {
+		return nil, fmt.Errorf("info xes status=%d", infoResp.StatusCode)
+	}
+
+	return &client{c: c, index: index, queryMap: qm}, nil
 }

 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 {
-	if r.StatusCode == 200 {
-		return nil
+func (c *client) ReadSetting() (map[string]any, error) {
+	r, err := c.c.Indices.GetSettings(
+		c.c.Indices.GetSettings.WithIndex(c.index),
+	)
+	if err != nil {
+		return nil, err
 	}

-	return fmt.Errorf("status=%d msg=%s", r.StatusCode, r.String())
+	if r.StatusCode != 200 {
+		return nil, fmt.Errorf("status=%d, msg=%s", r.StatusCode, r.String())
+	}
+
+	m := make(map[string]any)
+	decoder := json.NewDecoder(r.Body)
+	if err = decoder.Decode(&m); err != nil {
+		return nil, err
+	}
+
+	return m, nil
 }

-func (c *client) IOType() interfaces.IO {
-	return c.iot
+func (c *client) WriteSetting(m map[string]any) error {
+	//TODO implement me
+	panic("implement me")
+}
+
+func (c *client) IsInput() bool {
+	//TODO implement me
+	panic("implement me")
 }

 func (c *client) IsFile() bool {
@@ -127,34 +121,11 @@ func (c *client) Close() error {
 	return nil
 }

-func (c *client) ResetOffset() {
-	defer func() {
-		c.scrollId = ""
-	}()
-
-	bs, _ := json.Marshal(map[string]string{
-		"scroll_id": c.scrollId,
-	})
-
-	rr, err := c.c.ClearScroll(
-		c.c.ClearScroll.WithContext(util.Timeout(3)),
-		c.c.ClearScroll.WithBody(bytes.NewReader(bs)),
-	)
-	if err != nil {
-		logrus.Warnf("ResetOffset: clear scroll id=%s err=%v", c.scrollId, err)
-		return
-	}
-
-	if rr.StatusCode != 200 {
-		logrus.Warnf("ResetOffset: clear scroll id=%s msg=%s", c.scrollId, rr.String())
-	}
-}
-func (c *client) WriteData(ctx context.Context, docs []*interfaces.ESSource) (int, error) {
+func (c *client) WriteData(docs []*interfaces.ESSource) (int, error) {
 	var (
 		err     error
 		indexer esutil.BulkIndexer
 		count   int
-		be      error
 	)
 	if indexer, err = esutil.NewBulkIndexer(esutil.BulkIndexerConfig{
 		Client:  c.c,
@@ -171,37 +142,32 @@ 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,
 			DocumentID: doc.DocId,
 			Body:       bytes.NewReader(bs),
-			OnFailure: func(ctx context.Context, item esutil.BulkIndexerItem, item2 esutil.BulkIndexerResponseItem, bulkErr error) {
-				be = bulkErr
-			},
 		}); err != nil {
 			return 0, err
 		}
 		count++
 	}

-	if err = indexer.Close(util.TimeoutCtx(ctx, opt.Timeout)); err != nil {
+	if err = indexer.Close(util.Timeout(opt.Timeout)); err != nil {
 		return 0, err
 	}

-	if be != nil {
-		return 0, be
-	}
-
 	stats := indexer.Stats()
 	if stats.NumFailed > 0 {
-		return count, fmt.Errorf("write to xes failed_count=%d bulk_count=%d", stats.NumFailed, count)
+		return count, fmt.Errorf("write to xes failed=%d", stats.NumFailed)
 	}

 	return count, nil
 }

-func (c *client) ReadData(ctx context.Context, i int, query map[string]any) ([]*interfaces.ESSource, error) {
+func (c *client) ReadData(i int) ([]*interfaces.ESSource, error) {
 	var (
 		err    error
 		resp   *esapi.Response
@@ -210,15 +176,15 @@ func (c *client) ReadData(ctx context.Context, i int, query map[string]any) ([]*

 	if c.scrollId == "" {
 		qs := []func(*esapi.SearchRequest){
-			c.c.Search.WithContext(util.TimeoutCtx(ctx, opt.Timeout)),
+			c.c.Search.WithContext(util.Timeout(opt.Timeout)),
 			c.c.Search.WithIndex(c.index),
 			c.c.Search.WithSize(i),
 			c.c.Search.WithFrom(0),
-			c.c.Search.WithScroll(time.Duration(opt.Timeout*2) * time.Second),
+			c.c.Search.WithScroll(time.Duration(opt.ScrollDurationSeconds) * time.Second),
 		}

-		if query != nil && len(query) > 0 {
-			queryBs, _ := json.Marshal(map[string]any{"query": query})
+		if len(c.queryMap) > 0 {
+			queryBs, _ := json.Marshal(map[string]any{"query": c.queryMap})
 			qs = append(qs, c.c.Search.WithBody(bytes.NewReader(queryBs)))
 		}

@@ -242,7 +208,7 @@ func (c *client) ReadData(ctx context.Context, i int, query map[string]any) ([]*

 	if resp, err = c.c.Scroll(
 		c.c.Scroll.WithScrollID(c.scrollId),
-		c.c.Scroll.WithScroll(time.Duration(opt.Timeout*2)*time.Second),
+		c.c.Scroll.WithScroll(time.Duration(opt.ScrollDurationSeconds)*time.Second),
 	); err != nil {
 		return result.Hits.Hits, nil
 	}
@@ -255,7 +221,7 @@ func (c *client) ReadData(ctx context.Context, i int, query map[string]any) ([]*
 	return result.Hits.Hits, nil
 }

-func (c *client) ReadMapping(ctx context.Context) (map[string]any, error) {
+func (c *client) ReadMapping() (map[string]any, error) {
 	r, err := c.c.Indices.GetMapping(
 		c.c.Indices.GetMapping.WithIndex(c.index),
 	)
@@ -275,73 +241,7 @@ func (c *client) ReadMapping(ctx context.Context) (map[string]any, error) {

 	return m, nil
 }
-func (c *client) WriteMapping(ctx context.Context, m map[string]any) error {
-	var (
-		err    error
-		bs     []byte
-		result *esapi.Response
-	)
-
-	for idxKey := range m {
-		if bs, err = json.Marshal(m[idxKey]); err != nil {
-			return err
-		}
-
-		if result, err = c.c.Indices.Create(
-			c.index,
-			c.c.Indices.Create.WithContext(util.TimeoutCtx(ctx, opt.Timeout)),
-			c.c.Indices.Create.WithBody(bytes.NewReader(bs)),
-		); err != nil {
-			return err
-		}
-
-		if err = c.checkResponse(result); err != nil {
-			return err
-		}
-	}
-
-	return nil
-}
-
-func (c *client) ReadSetting(ctx context.Context) (map[string]any, error) {
-	r, err := c.c.Indices.GetSettings(
-		c.c.Indices.GetSettings.WithContext(util.TimeoutCtx(ctx, opt.Timeout)),
-		c.c.Indices.GetSettings.WithIndex(c.index),
-	)
-	if err != nil {
-		return nil, err
-	}
-
-	if r.StatusCode != 200 {
-		return nil, fmt.Errorf("status=%d, msg=%s", r.StatusCode, r.String())
-	}
-
-	m := make(map[string]any)
-	decoder := json.NewDecoder(r.Body)
-	if err = decoder.Decode(&m); err != nil {
-		return nil, err
-	}
-
-	return m, nil
-}
-
-func (c *client) WriteSetting(ctx context.Context, m map[string]any) error {
-	var (
-		err    error
-		bs     []byte
-		result *esapi.Response
-	)
-
-	if bs, err = json.Marshal(m); err != nil {
-		return err
-	}
-
-	if result, err = c.c.Indices.PutSettings(
-		bytes.NewReader(bs),
-		c.c.Indices.PutSettings.WithContext(util.TimeoutCtx(ctx, opt.Timeout)),
-	); err != nil {
-		return err
-	}
-
-	return c.checkResponse(result)
+func (c *client) WriteMapping(m map[string]any) error {
+	//TODO implement me
+	panic("implement me")
 }
@@ -1,13 +1,9 @@
 package xes

 import (
-	"bufio"
-	"fmt"
-	"os"
-	"testing"
-
+	"esgo2dump/internal/util"
 	elastic "github.com/elastic/go-elasticsearch/v7"
-	"github.com/loveuer/esgo2dump/internal/util"
+	"testing"
 )

 func TestGetESMapping(t *testing.T) {
@@ -40,68 +36,3 @@ 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++
-	}
-}
@@ -2,60 +2,28 @@ package xfile

 import (
 	"bufio"
-	"context"
 	"encoding/json"
-	"github.com/loveuer/esgo2dump/internal/opt"
-	"io"
+	"esgo2dump/internal/interfaces"
+	"github.com/sirupsen/logrus"
 	"os"
-
-	"github.com/loveuer/esgo2dump/internal/interfaces"
 )

 type client struct {
 	f       *os.File
-	iot     interfaces.IO
 	scanner *bufio.Scanner
 }

-func (c *client) ReadMapping(ctx context.Context) (map[string]any, error) {
-	var (
-		err error
-		bs  []byte
-	)
-
-	if bs, err = io.ReadAll(c.f); err != nil {
-		return nil, err
-	}
-
-	m := make(map[string]any)
-
-	if err = json.Unmarshal(bs, &m); err != nil {
-		return nil, err
-	}
-
-	return m, nil
+func (c *client) ReadMapping() (map[string]any, error) {
+	//TODO implement me
+	panic("implement me")
 }

-func (c *client) ReadSetting(ctx context.Context) (map[string]any, error) {
-	var (
-		err error
-		bs  []byte
-	)
-
-	if bs, err = io.ReadAll(c.f); err != nil {
-		return nil, err
-	}
-
-	m := make(map[string]any)
-
-	if err = json.Unmarshal(bs, &m); err != nil {
-		return nil, err
-	}
-
-	return m, nil
+func (c *client) ReadSetting() (map[string]any, error) {
+	//TODO implement me
+	panic("implement me")
 }

-func (c *client) WriteMapping(ctx context.Context, m map[string]any) error {
+func (c *client) WriteMapping(m map[string]any) error {
 	bs, err := json.Marshal(m)
 	if err != nil {
 		return err
@@ -66,7 +34,7 @@ func (c *client) WriteMapping(ctx context.Context, m map[string]any) error {
 	return err
 }

-func (c *client) WriteSetting(ctx context.Context, m map[string]any) error {
+func (c *client) WriteSetting(m map[string]any) error {
 	bs, err := json.Marshal(m)
 	if err != nil {
 		return err
@@ -77,17 +45,16 @@ func (c *client) WriteSetting(ctx context.Context, m map[string]any) error {
 	return err
 }

-func (c *client) IOType() interfaces.IO {
-	return c.iot
+func (c *client) IsInput() bool {
+	//TODO implement me
+	panic("implement me")
 }

 func (c *client) IsFile() bool {
 	return true
 }

-func (c *client) ResetOffset() {}
-
-func (c *client) WriteData(ctx context.Context, docs []*interfaces.ESSource) (int, error) {
+func (c *client) WriteData(docs []*interfaces.ESSource) (int, error) {
 	var (
 		err   error
 		bs    []byte
@@ -111,7 +78,7 @@ 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(i int) ([]*interfaces.ESSource, error) {
 	var (
 		err   error
 		count = 0
@@ -121,6 +88,8 @@ 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
@@ -146,12 +115,10 @@ func (c *client) Close() error {
 }

 func NewClient(file *os.File, ioType interfaces.IO) (interfaces.DumpIO, error) {
-	c := &client{f: file, iot: ioType}
+	c := &client{f: file}

 	if ioType == interfaces.IOInput {
 		c.scanner = bufio.NewScanner(c.f)
-		buf := make([]byte, opt.BuffSize)
-		c.scanner.Buffer(buf, opt.MaxBuffSize)
 	}

 	return c, nil
@@ -2,11 +2,10 @@ package main

 import (
 	"context"
+	"esgo2dump/internal/cmd"
 	"os/signal"
 	"syscall"

-	"github.com/loveuer/esgo2dump/internal/cmd"
-
 	"github.com/sirupsen/logrus"
 )

@@ -1,51 +0,0 @@
-# esgo2dump
-# dump elasticsearch with golang
-
---
-
- 当前仅支持 elasticsearch 7
-
---
-
-### install
-
- with golang >= 1.18
-
-  `go install github.com/loveuer/esgo2dump@latest`
-
- download pre-build release:
-
-  [releases](https://github.com/loveuer/esgo2dump/releases)
-
-### usage
-
-`esgo2dump -h`
-
-```bash
-esgo2dump --input=http://127.0.0.1:9200/some_index --output=./data.json
-
-esgo2dump --input=http://127.0.0.1:9200/some_index --output=http://192.168.1.1:9200/some_index --limit=5000
-
-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_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
-
- [x] data dump
- [x] mapping dump
- [x] es to file
- [x] es to es
- [x] auto create index with mapping
- [ ] auto create index with mapping,setting
- [ ] support es8
Author	SHA1	Message	Date
loveuer	d3524d0f05	wip: mapping,setting import	2024-03-25 20:22:18 +08:00
loveuer	b0b0ec2b58	wip: mapping type	2024-03-24 20:47:54 +08:00
loveuer	f1af294ced	feat: dump from es with query, limit	2024-03-23 21:25:19 +08:00
loveuer	983533237f	wip: basic feat	2024-03-22 18:05:47 +08:00