From 71c1df1af906e0e94cb34880ad03c63b3386124c Mon Sep 17 00:00:00 2001
From: simlecode <69969590+simlecode@users.noreply.github.com>
Date: Tue, 15 Jul 2025 11:23:12 +0800
Subject: [PATCH] feat: add random rename source file flag

---
 chunk.go               | 19 ++++++++++++++++---
 cmd/graphsplit/main.go | 14 ++++++++++----
 extra_file.go          | 11 +++++++----
 3 files changed, 33 insertions(+), 11 deletions(-)

diff --git a/chunk.go b/chunk.go
index be01765..09f7e41 100644
--- a/chunk.go
+++ b/chunk.go
@@ -163,6 +163,7 @@ func Chunk(ctx context.Context,
 	parallel int,
 	cb GraphBuildCallback,
 	ef *ExtraFile,
+	randomRenameSourceFile bool,
 ) error {
 	var cumuSize int64 = 0
 	graphSliceCount := 0
@@ -186,7 +187,10 @@ func Chunk(ctx context.Context,
 	}
 	files := GetFileListAsync(args)
 	for item := range files {
-		item := tryRenameFileName([]Finfo{item})[0]
+		item := item
+		if randomRenameSourceFile {
+			item = tryRenameFileName([]Finfo{item})[0]
+		}
 		// log.Infof("name: %s", item.Name)
 		fileSize := item.Info.Size()
 		switch {
@@ -221,7 +225,11 @@ func Chunk(ctx context.Context,
 				SeekStart: seekStart,
 				SeekEnd:   seekEnd,
 			}
-			graphFiles = append(graphFiles, tryRenameFileName([]Finfo{fi})...)
+			if randomRenameSourceFile {
+				graphFiles = append(graphFiles, tryRenameFileName([]Finfo{fi})...)
+			} else {
+				graphFiles = append(graphFiles, fi)
+			}
 			fileSliceCount++
 			// todo build ipld from graphFiles
 			BuildIpldGraph(ctx, append(ef.getFiles(), graphFiles...), GenGraphName(graphName, graphSliceCount, sliceTotal), parentPath, carDir, parallel, cb, expectSliceSize, ef)
@@ -247,7 +255,12 @@ func Chunk(ctx context.Context,
 					SeekStart: seekStart,
 					SeekEnd:   seekEnd,
 				}
-				graphFiles = append(graphFiles, tryRenameFileName([]Finfo{fi})...)
+				if randomRenameSourceFile {
+					graphFiles = append(graphFiles, tryRenameFileName([]Finfo{fi})...)
+				} else {
+					graphFiles = append(graphFiles, fi)
+				}
+
 				fileSliceCount++
 				if seekEnd-seekStart == partSliceSize-1 {
 					// todo build ipld from graphFiles
diff --git a/cmd/graphsplit/main.go b/cmd/graphsplit/main.go
index d1d2477..c228365 100644
--- a/cmd/graphsplit/main.go
+++ b/cmd/graphsplit/main.go
@@ -76,6 +76,11 @@ var chunkCmd = &cli.Command{
 			Value: false,
 			Usage: "rename carfile to piece",
 		},
+		&cli.BoolFlag{
+			Name:  "random-rename-source-file",
+			Value: false,
+			Usage: "random rename source file name",
+		},
 		&cli.BoolFlag{
 			Name:  "add-padding",
 			Value: false,
@@ -98,6 +103,7 @@ var chunkCmd = &cli.Command{
 		parentPath := c.String("parent-path")
 		carDir := c.String("car-dir")
 		graphName := c.String("graph-name")
+		randomRenameSourceFile := c.Bool("random-rename-source-file")
 		if !graphsplit.ExistDir(carDir) {
 			return fmt.Errorf("the path of car-dir does not exist")
 		}
@@ -138,8 +144,8 @@ var chunkCmd = &cli.Command{
 		if sliceSize+int(extraFileSliceSize) > 32*graphsplit.Gib {
 			return fmt.Errorf("slice size %d + extra file slice size %d exceeds 32 GiB", sliceSize, extraFileSliceSize)
 		}
-		log.Infof("extra file slice size: %d", extraFileSliceSize)
-		rf, err := graphsplit.NewRealFile(strings.TrimSuffix(cfg.ExtraFilePath, "/"), int64(extraFileSliceSize), int64(sliceSize))
+		log.Infof("extra file slice size: %d, random rename source file: %v", extraFileSliceSize, randomRenameSourceFile)
+		rf, err := graphsplit.NewRealFile(strings.TrimSuffix(cfg.ExtraFilePath, "/"), int64(extraFileSliceSize), int64(sliceSize), randomRenameSourceFile)
 		if err != nil {
 			return err
 		}
@@ -158,11 +164,11 @@ var chunkCmd = &cli.Command{
 		fmt.Println("loop: ", loop)
 		if !loop {
 			fmt.Println("chunking once...")
-			return graphsplit.Chunk(ctx, int64(sliceSize), parentPath, targetPath, carDir, graphName, int(parallel), cb, rf)
+			return graphsplit.Chunk(ctx, int64(sliceSize), parentPath, targetPath, carDir, graphName, int(parallel), cb, rf, randomRenameSourceFile)
 		}
 		fmt.Println("loop chunking...")
 		for {
-			err = graphsplit.Chunk(ctx, int64(sliceSize), parentPath, targetPath, carDir, graphName, int(parallel), cb, rf)
+			err = graphsplit.Chunk(ctx, int64(sliceSize), parentPath, targetPath, carDir, graphName, int(parallel), cb, rf, randomRenameSourceFile)
 			if err != nil {
 				return fmt.Errorf("failed to chunk: %v", err)
 			}
diff --git a/extra_file.go b/extra_file.go
index 8471f63..8cd2a5b 100644
--- a/extra_file.go
+++ b/extra_file.go
@@ -15,7 +15,7 @@ type ExtraFile struct {
 	pieceRawSize int64
 }
 
-func NewRealFile(path string, sliceSize int64, pieceRawSize int64) (*ExtraFile, error) {
+func NewRealFile(path string, sliceSize int64, pieceRawSize int64, randomRenameSourceFile bool) (*ExtraFile, error) {
 	rf := &ExtraFile{path: path, sliceSize: sliceSize, pieceRawSize: pieceRawSize}
 	if path != "" {
 		finfo, err := os.Stat(path)
@@ -25,18 +25,21 @@ func NewRealFile(path string, sliceSize int64, pieceRawSize int64) (*ExtraFile,
 		if !finfo.IsDir() {
 			return nil, fmt.Errorf("the path %s is not a directory", path)
 		}
-		rf.walk()
+		rf.walk(randomRenameSourceFile)
 	}
 
 	return rf, nil
 }
 
-func (rf *ExtraFile) walk() {
+func (rf *ExtraFile) walk(randomRenameSourceFile bool) {
 	files := GetFileListAsync([]string{rf.path})
 	for item := range files {
 		rf.files = append(rf.files, item)
 	}
-	rf.files = tryRenameFileName(rf.files)
+	if randomRenameSourceFile {
+		rf.files = tryRenameFileName(rf.files)
+	}
+
 }
 
 func (rf *ExtraFile) getFiles() []Finfo {