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"
@@ -34,9 +35,10 @@ 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
 	f_query  string
 
 	f_query_file string
@@ -47,7 +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")
 
@@ -57,9 +59,16 @@ func init() {
 	rootCommand.Flags().StringVar(&es_oversion, "o-version", "7", "output(es) version")
 	rootCommand.Flags().StringVarP(&f_type, "type", "t", "data", "data/mapping/setting")
 	rootCommand.Flags().StringVarP(&f_source, "source", "s", "", "query source, use ';' to separate")
+	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

@@ -11,6 +11,7 @@ import (
 	"net/url"
 	"os"
 	"strings"
+	"sync"
 
 	"github.com/loveuer/esgo2dump/internal/interfaces"
 	"github.com/loveuer/esgo2dump/internal/opt"
@@ -50,10 +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)
@@ -192,24 +189,27 @@ func executeData(ctx context.Context, input, output interfaces.DumpIO) error {
 
 		e2ch = make(chan error)
 		wch  = make(chan []*model.ESSource)
+		wg   = sync.WaitGroup{}
 	)
 
 	go func() {
-		defer func() {
-			close(wch)
-			close(e2ch)
-		}()
-
+		wg.Add(1)
 		if err = output.WriteData(ctx, wch); err != nil {
 			e2ch <- err
 		}
+
+		wg.Done()
 	}()
 
 	log.Info("Query: got queries=%d", len(queries))
 
 Loop:
-	for _, query := range queries {
-		dch, ech = input.ReadData(ctx, f_limit, query, sources)
+	for qi, query := range queries {
+		bs, _ := json.Marshal(query)
+
+		log.Debug("Query[%d]: %s", qi, string(bs))
+
+		dch, ech = input.ReadData(ctx, f_limit, query, sources, []string{f_sort})
 
 		for {
 			select {
@@ -233,6 +233,10 @@ Loop:
 		}
 	}
 
+	close(wch)
+
+	wg.Wait()
+
 	return nil
 }
 

internal/interfaces/dumpio.go

@@ -6,7 +6,7 @@ import (
 )
 
 type DumpIO interface {
-	ReadData(ctx context.Context, size int, query map[string]any, includeFields []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

@@ -0,0 +1,21 @@
+package util
+
+func Min[T ~string | ~int | ~int64 | ~uint64 | ~float64 | ~float32 | ~int32 | ~uint32 | ~int16 | ~uint16 | ~int8 | ~uint8](a, b T) T {
+	if a <= b {
+		return a
+	}
+
+	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

@@ -135,8 +135,8 @@ func (c *clientv6) Close() error {
 	return nil
 }
 
-func (c *clientv6) ReadData(ctx context.Context, size int, query map[string]any, source []string) (<-chan []*model.ESSource, <-chan error) {
-	dch, ech := es6.ReadData(ctx, c.client, c.index, size, 0, query, source)
+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

@@ -70,63 +70,8 @@ func (c *client) Close() error {
 	return nil
 }
 
-//func (c *client) WriteData(ctx context.Context, docs []*model.ESSource) (int, error) {
-//	var (
-//		err     error
-//		indexer esutil.BulkIndexer
-//		count   int
-//		be      error
-//	)
-//	if indexer, err = esutil.NewBulkIndexer(esutil.BulkIndexerConfig{
-//		Client:     c.client,
-//		Index:      c.index,
-//		ErrorTrace: true,
-//		OnError: func(ctx context.Context, err error) {
-//
-//		},
-//	}); err != nil {
-//		return 0, err
-//	}
-//
-//	for _, doc := range docs {
-//		var bs []byte
-//
-//		if bs, err = json.Marshal(doc.Content); err != nil {
-//			return 0, err
-//		}
-//
-//		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 {
-//		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, nil
-//}
-
-func (c *client) ReadData(ctx context.Context, size int, query map[string]any, source []string) (<-chan []*model.ESSource, <-chan error) {
-	dch, ech := es7.ReadData(ctx, c.client, c.index, size, 0, query, source)
+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) (<-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

@@ -4,9 +4,27 @@ type ESSource struct {
 	DocId   string         `json:"_id"`
 	Index   string         `json:"_index"`
 	Content map[string]any `json:"_source"`
+	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

@@ -10,10 +10,11 @@ import (
 	"github.com/loveuer/esgo2dump/internal/util"
 	"github.com/loveuer/esgo2dump/log"
 	"github.com/loveuer/esgo2dump/model"
+	"github.com/samber/lo"
 	"time"
 )
 
-func ReadData(ctx context.Context, client *elastic.Client, index string, size, max int, query map[string]any, source []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)
@@ -23,9 +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() {
@@ -62,7 +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),
 		}
@@ -71,6 +72,16 @@ func ReadData(ctx context.Context, client *elastic.Client, index string, size, m
 			qs = append(qs, client.Search.WithSourceIncludes(source...))
 		}
 
+		if len(sort) > 0 {
+			sorts := lo.Filter(sort, func(item string, index int) bool {
+				return item != ""
+			})
+
+			if len(sorts) > 0 {
+				qs = append(qs, client.Search.WithSort(sorts...))
+			}
+		}
+
 		if query != nil && len(query) > 0 {
 			queryBs, _ := json.Marshal(map[string]any{"query": query})
 			qs = append(qs, client.Search.WithBody(bytes.NewReader(queryBs)))
@@ -95,9 +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
 		}
 
@@ -110,7 +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 {
@@ -124,9 +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

@@ -10,10 +10,15 @@ import (
 	"github.com/loveuer/esgo2dump/internal/util"
 	"github.com/loveuer/esgo2dump/log"
 	"github.com/loveuer/esgo2dump/model"
+	"github.com/samber/lo"
 	"time"
 )
 
-func ReadData(ctx context.Context, client *elastic.Client, index string, size, max int, query map[string]any, source []string) (<-chan []*model.ESSource, <-chan error) {
+// ReadData
+// Deprecated
+// @param[source]: a list of include fields to extract and return from the _source field.
+// @param[sort]:   a list of <field>:<direction> pairs.
+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) {
 	var (
 		dataCh = make(chan []*model.ESSource)
 		errCh  = make(chan error)
@@ -23,7 +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
 		)
@@ -71,6 +76,16 @@ func ReadData(ctx context.Context, client *elastic.Client, index string, size, m
 			qs = append(qs, client.Search.WithSourceIncludes(source...))
 		}
 
+		if len(sort) > 0 {
+			sorts := lo.Filter(sort, func(item string, index int) bool {
+				return item != ""
+			})
+
+			if len(sorts) > 0 {
+				qs = append(qs, client.Search.WithSort(sorts...))
+			}
+		}
+
 		if query != nil && len(query) > 0 {
 			queryBs, _ := json.Marshal(map[string]any{"query": query})
 			qs = append(qs, client.Search.WithBody(bytes.NewReader(queryBs)))
@@ -110,7 +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 {
@@ -134,3 +149,120 @@ func ReadData(ctx context.Context, client *elastic.Client, index string, size, m
 
 	return dataCh, errCh
 }
+
+// ReadDataV2 es7 read data
+/*
+	- @param[source]: a list of include fields to extract and return from the _source field.
+    - @param[sort]:   a list of <field>:<direction> pairs.
+*/
+func ReadDataV2(
+	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)
+	)
+
+	go func() {
+		var (
+			err         error
+			bs          []byte
+			resp        *esapi.Response
+			searchAfter        = make([]any, 0)
+			total       uint64 = 0
+			body               = make(map[string]any)
+			qs          []func(request *esapi.SearchRequest)
+		)
+
+		if sort == nil {
+			sort = []string{}
+		}
+
+		if query != nil && len(query) > 0 {
+			body["query"] = query
+		}
+
+		sort = append(sort, "_id:ASC")
+
+		sorts := lo.Filter(sort, func(item string, index int) bool {
+			return item != ""
+		})
+
+		defer func() {
+			close(dataCh)
+			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(int(fina_size)),
+				client.Search.WithSort(sorts...),
+			}
+
+			if len(source) > 0 {
+				qs = append(qs, client.Search.WithSourceIncludes(source...))
+			}
+
+			delete(body, "search_after")
+			if len(searchAfter) > 0 {
+				body["search_after"] = searchAfter
+			}
+
+			if bs, err = json.Marshal(body); err != nil {
+				errCh <- err
+				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
+				return
+			}
+
+			if resp.StatusCode != 200 {
+				errCh <- fmt.Errorf("resp status=%d, resp=%s", resp.StatusCode, resp.String())
+				return
+			}
+
+			var result = new(model.ESResponseV7)
+			decoder := json.NewDecoder(resp.Body)
+			if err = decoder.Decode(result); err != nil {
+				errCh <- err
+				return
+			}
+
+			if resp.StatusCode != 200 {
+				errCh <- fmt.Errorf("resp status=%d, resp=%s", resp.StatusCode, resp.String())
+				return
+			}
+
+			dataCh <- result.Hits.Hits
+			total += uint64(len(result.Hits.Hits))
+
+			log.Debug("es7.read: total: %d", total)
+
+			if uint64(len(result.Hits.Hits)) < size || (max > 0 && total >= max) {
+				break
+			}
+
+			searchAfter = result.Hits.Hits[len(result.Hits.Hits)-1].Sort
+		}
+
+	}()
+
+	return dataCh, errCh
+
+}