fix: 0 size

Bugfix: the response  elasticsearch6 and 7 are a little different

internal/cmd/cmd.go

@@ -2,6 +2,7 @@ package cmd
 
 import (
 	"context"
+	"github.com/loveuer/esgo2dump/log"
 
 	"github.com/loveuer/esgo2dump/internal/opt"
 	"github.com/spf13/cobra"
@@ -34,7 +35,7 @@ esgo2dump --input=http://127.0.0.1:9200/some_index --output=./data.json --query_
 
 	f_input  string
 	f_output string
-	f_limit  int
+	f_limit  uint64
 	f_type   string
 	f_source string
 	f_sort   string
@@ -48,7 +49,7 @@ esgo2dump --input=http://127.0.0.1:9200/some_index --output=./data.json --query_
 )
 
 func init() {
-	rootCommand.Flags().BoolVar(&opt.Debug, "debug", false, "")
+	rootCommand.PersistentFlags().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")
 
@@ -61,7 +62,13 @@ func init() {
 	rootCommand.Flags().StringVar(&f_sort, "sort", "", "sort, <field>:<direction> format, for example: time:desc or name:asc")
 	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, "")
+	rootCommand.Flags().Uint64VarP(&f_limit, "limit", "l", 100, "")
+
+	rootCommand.PersistentPreRun = func(cmd *cobra.Command, args []string) {
+		if opt.Debug {
+			log.SetLogLevel(log.LogLevelDebug)
+		}
+	}
 }
 
 func Start(ctx context.Context) error {

internal/cmd/run.go

@@ -51,10 +51,6 @@ func run(cmd *cobra.Command, args []string) error {
 		ioo interfaces.DumpIO
 	)
 
-	if opt.Debug {
-		log.SetLogLevel(log.LogLevelDebug)
-	}
-
 	if f_version {
 		fmt.Printf("esgo2dump (Version: %s)\n", opt.Version)
 		os.Exit(0)

internal/interfaces/dumpio.go

@@ -6,7 +6,7 @@ import (
 )
 
 type DumpIO interface {
-	ReadData(ctx context.Context, size int, query map[string]any, includeFields []string, sort []string) (<-chan []*model.ESSource, <-chan error)
+	ReadData(ctx context.Context, size uint64, query map[string]any, includeFields []string, sort []string) (<-chan []*model.ESSource, <-chan error)
 	WriteData(ctx context.Context, docsCh <-chan []*model.ESSource) error
 
 	ReadMapping(context.Context) (map[string]any, error)

internal/util/min.go

@@ -7,3 +7,15 @@ func Min[T ~string | ~int | ~int64 | ~uint64 | ~float64 | ~float32 | ~int32 | ~u
 
 	return b
 }
+
+func AbsMin(a, b uint64) uint64 {
+	if a == 0 {
+		return b
+	}
+
+	if b == 0 {
+		return a
+	}
+
+	return Min(a, b)
+}

internal/xes/xes6.go

@@ -135,7 +135,7 @@ func (c *clientv6) Close() error {
 	return nil
 }
 
-func (c *clientv6) ReadData(ctx context.Context, size int, query map[string]any, source []string, sort []string) (<-chan []*model.ESSource, <-chan error) {
+func (c *clientv6) ReadData(ctx context.Context, size uint64, query map[string]any, source []string, sort []string) (<-chan []*model.ESSource, <-chan error) {
 	dch, ech := es6.ReadData(ctx, c.client, c.index, size, 0, query, source, sort)
 
 	return dch, ech

internal/xes/xes7.go

@@ -70,7 +70,7 @@ func (c *client) Close() error {
 	return nil
 }
 
-func (c *client) ReadData(ctx context.Context, size int, query map[string]any, source []string, sort []string) (<-chan []*model.ESSource, <-chan error) {
+func (c *client) ReadData(ctx context.Context, size uint64, query map[string]any, source []string, sort []string) (<-chan []*model.ESSource, <-chan error) {
 	dch, ech := es7.ReadDataV2(ctx, c.client, c.index, size, 0, query, source, sort)
 
 	return dch, ech

internal/xfile/xfile.go

@@ -110,14 +110,14 @@ func (c *client) IsFile() bool {
 	return true
 }
 
-func (c *client) ReadData(ctx context.Context, size int, _ map[string]any, _ []string, _ []string) (<-chan []*model.ESSource, <-chan error) {
+func (c *client) ReadData(ctx context.Context, size uint64, _ map[string]any, _ []string, _ []string) (<-chan []*model.ESSource, <-chan error) {
 	var (
 		err   error
-		count = 0
-		list  = make([]*model.ESSource, 0, size)
-		dch   = make(chan []*model.ESSource)
-		ech   = make(chan error)
-		ready = make(chan bool)
+		count uint64 = 0
+		list         = make([]*model.ESSource, 0, size)
+		dch          = make(chan []*model.ESSource)
+		ech          = make(chan error)
+		ready        = make(chan bool)
 	)
 
 	go func(ctx context.Context) {

model/es.go

@@ -7,7 +7,24 @@ type ESSource struct {
 	Sort    []any          `json:"sort"`
 }
 
-type ESResponse struct {
+type ESResponseV6 struct {
+	ScrollId string `json:"_scroll_id"`
+	Took     int    `json:"took"`
+	TimedOut bool   `json:"timed_out"`
+	Shards   struct {
+		Total      int `json:"total"`
+		Successful int `json:"successful"`
+		Skipped    int `json:"skipped"`
+		Failed     int `json:"failed"`
+	} `json:"_shards"`
+	Hits struct {
+		Total    int         `json:"total"`
+		MaxScore float64     `json:"max_score"`
+		Hits     []*ESSource `json:"hits"`
+	} `json:"hits"`
+}
+
+type ESResponseV7 struct {
 	ScrollId string `json:"_scroll_id"`
 	Took     int    `json:"took"`
 	TimedOut bool   `json:"timed_out"`

readme.md

@@ -53,7 +53,8 @@ esgo2dump --input=http://127.0.0.1:9200/some_index --output=./data.json --query_
 - [x] es to file
 - [x] es to es
 - [x] auto create index with mapping
+- [x] support es6
+- [ ] [Feature Request #1](https://github.com/loveuer/esgo2dump/issues/1): Supports more than 10,000 lines of query_file
 - [ ] args: split_size (auto split json output file)
 - [ ] auto create index with mapping,setting
-- [x] support es6
 - [ ] support es8

xes/es6/read.go

@@ -14,7 +14,7 @@ import (
 	"time"
 )
 
-func ReadData(ctx context.Context, client *elastic.Client, index string, size, max int, query map[string]any, source []string, sort []string) (<-chan []*model.ESSource, <-chan error) {
+func ReadData(ctx context.Context, client *elastic.Client, index string, size, max uint64, query map[string]any, source []string, sort []string) (<-chan []*model.ESSource, <-chan error) {
 	var (
 		dataCh = make(chan []*model.ESSource)
 		errCh  = make(chan error)
@@ -24,9 +24,9 @@ func ReadData(ctx context.Context, client *elastic.Client, index string, size, m
 		var (
 			err      error
 			resp     *esapi.Response
-			result   = new(model.ESResponse)
+			result   = new(model.ESResponseV6)
 			scrollId string
-			total    int
+			total    uint64
 		)
 
 		defer func() {
@@ -63,7 +63,7 @@ func ReadData(ctx context.Context, client *elastic.Client, index string, size, m
 		qs := []func(*esapi.SearchRequest){
 			client.Search.WithContext(util.TimeoutCtx(ctx, 20)),
 			client.Search.WithIndex(index),
-			client.Search.WithSize(size),
+			client.Search.WithSize(int(size)),
 			client.Search.WithFrom(0),
 			client.Search.WithScroll(time.Duration(120) * time.Second),
 		}
@@ -106,9 +106,9 @@ func ReadData(ctx context.Context, client *elastic.Client, index string, size, m
 		scrollId = result.ScrollId
 
 		dataCh <- result.Hits.Hits
-		total += len(result.Hits.Hits)
+		total += uint64(len(result.Hits.Hits))
 
-		if len(result.Hits.Hits) < size || (max > 0 && total >= max) {
+		if uint64(len(result.Hits.Hits)) < size || (max > 0 && total >= max) {
 			return
 		}
 
@@ -121,7 +121,7 @@ func ReadData(ctx context.Context, client *elastic.Client, index string, size, m
 				return
 			}
 
-			result = new(model.ESResponse)
+			result = new(model.ESResponseV6)
 
 			decoder = json.NewDecoder(resp.Body)
 			if err = decoder.Decode(result); err != nil {
@@ -135,9 +135,9 @@ func ReadData(ctx context.Context, client *elastic.Client, index string, size, m
 			}
 
 			dataCh <- result.Hits.Hits
-			total += len(result.Hits.Hits)
+			total += uint64(len(result.Hits.Hits))
 
-			if len(result.Hits.Hits) < size || (max > 0 && total >= max) {
+			if uint64(len(result.Hits.Hits)) < size || (max > 0 && total >= max) {
 				break
 			}
 		}

xes/es7/read.go

@@ -28,7 +28,7 @@ func ReadData(ctx context.Context, client *elastic.Client, index string, size, m
 		var (
 			err      error
 			resp     *esapi.Response
-			result   = new(model.ESResponse)
+			result   = new(model.ESResponseV7)
 			scrollId string
 			total    int
 		)
@@ -125,7 +125,7 @@ func ReadData(ctx context.Context, client *elastic.Client, index string, size, m
 				return
 			}
 
-			result = new(model.ESResponse)
+			result = new(model.ESResponseV7)
 
 			decoder = json.NewDecoder(resp.Body)
 			if err = decoder.Decode(result); err != nil {
@@ -159,7 +159,7 @@ func ReadDataV2(
 	ctx context.Context,
 	client *elastic.Client,
 	index string,
-	size, max int,
+	size, max uint64,
 	query map[string]any,
 	source []string,
 	sort []string,
@@ -174,9 +174,9 @@ func ReadDataV2(
 			err         error
 			bs          []byte
 			resp        *esapi.Response
-			searchAfter = make([]any, 0)
-			total       = 0
-			body        = make(map[string]any)
+			searchAfter        = make([]any, 0)
+			total       uint64 = 0
+			body               = make(map[string]any)
 			qs          []func(request *esapi.SearchRequest)
 		)
 
@@ -199,11 +199,14 @@ func ReadDataV2(
 			close(errCh)
 		}()
 
+		fina_size := util.AbsMin(size, max-total)
+		log.Debug("es7.read: size = %d, max = %d, total = %d, fina size = %d", size, max, total, fina_size)
+
 		for {
 			qs = []func(*esapi.SearchRequest){
 				client.Search.WithContext(util.TimeoutCtx(ctx, 30)),
 				client.Search.WithIndex(index),
-				client.Search.WithSize(util.Min(size, max-total)),
+				client.Search.WithSize(int(fina_size)),
 				client.Search.WithSort(sorts...),
 			}
 
@@ -221,6 +224,8 @@ func ReadDataV2(
 				return
 			}
 
+			log.Debug("body raw: %s", string(bs))
+
 			qs = append(qs, client.Search.WithBody(bytes.NewReader(bs)))
 			if resp, err = client.Search(qs...); err != nil {
 				errCh <- err
@@ -232,7 +237,7 @@ func ReadDataV2(
 				return
 			}
 
-			var result = new(model.ESResponse)
+			var result = new(model.ESResponseV7)
 			decoder := json.NewDecoder(resp.Body)
 			if err = decoder.Decode(result); err != nil {
 				errCh <- err
@@ -245,9 +250,11 @@ func ReadDataV2(
 			}
 
 			dataCh <- result.Hits.Hits
-			total += len(result.Hits.Hits)
+			total += uint64(len(result.Hits.Hits))
 
-			if len(result.Hits.Hits) < size || (max > 0 && total >= max) {
+			log.Debug("es7.read: total: %d", total)
+
+			if uint64(len(result.Hits.Hits)) < size || (max > 0 && total >= max) {
 				break
 			}