detect hard links

README.md

@@ -93,6 +93,8 @@ Gdu has two modes: interactive (default) and non-interactive.
 
 Non-interactive mode is started automtically when TTY is not detected (using [go-isatty](https://github.com/mattn/go-isatty)), for example if the output is being piped to a file, or it can be started explicitly by using a flag.
 
+Hard links are counted only once.
+
 ## Running tests
 
     make test

analyze/dir.go

@@ -31,7 +31,9 @@ func ProcessDir(path string, progress *CurrentProgress, ignore ShouldDirBeIgnore
 	dir := processDir(path, progress, concurrencyLimitChannel, &wait, ignore)
 	dir.BasePath = filepath.Dir(path)
 	wait.Wait()
-	dir.UpdateStats()
+
+	links := make(AlreadyCountedHardlinks, 10)
+	dir.UpdateStats(links)
 
 	progress.Mutex.Lock()
 	progress.Done = true
@@ -114,11 +116,12 @@ func processDir(path string, progress *CurrentProgress, concurrencyLimitChannel
 				Name:      f.Name(),
 				Flag:      flag,
 				Size:      f.Size(),
-				Usage:     getUsage(f),
 				ItemCount: 1,
 				Parent:    &dir,
 			}
 
+			setPlatformSpecificAttrs(file, f)
+
 			totalSize += f.Size()
 
 			mutex.Lock()

analyze/dir_other.go

@@ -6,6 +6,4 @@ import (
 	"os"
 )
 
-func getUsage(f os.FileInfo) int64 {
-	return int64(0)
-}
+func setPlatformSpecificAttrs(file *File, f os.FileInfo) {}

analyze/dir_test.go

@@ -64,6 +64,25 @@ func TestFlags(t *testing.T) {
 	assert.Equal(t, '@', dir.Files[0].Files[1].Flag)
 }
 
+func TestHardlink(t *testing.T) {
+	fin := testdir.CreateTestDir()
+	defer fin()
+
+	os.Link("test_dir/nested/file2", "test_dir/nested/file3")
+
+	dir := ProcessDir("test_dir", &CurrentProgress{Mutex: &sync.Mutex{}}, func(_ string) bool { return false })
+
+	assert.Equal(t, int64(7+4096*3), dir.Size) // file2 and file3 are counted just once for size
+	assert.Equal(t, int64(4096*5), dir.Usage)  // and usage
+	assert.Equal(t, 6, dir.ItemCount)          // but twice for item count
+
+	// test file3
+	assert.Equal(t, "file3", dir.Files[0].Files[1].Name)
+	assert.Equal(t, int64(2), dir.Files[0].Files[1].Size)
+	assert.Equal(t, int64(4096), dir.Files[0].Files[1].Usage)
+	assert.Equal(t, 'H', dir.Files[0].Files[1].Flag)
+}
+
 func BenchmarkProcessDir(b *testing.B) {
 	fin := testdir.CreateTestDir()
 	defer fin()

analyze/dir_unix.go

@@ -10,12 +10,13 @@ import (
 
 const devBSize = 512
 
-func getUsage(f os.FileInfo) int64 {
-	var usage int64 = 0
-
+func setPlatformSpecificAttrs(file *File, f os.FileInfo) {
 	switch stat := f.Sys().(type) {
 	case *syscall.Stat_t:
-		usage = stat.Blocks * devBSize
+		file.Usage = stat.Blocks * devBSize
+
+		if stat.Nlink > 1 {
+			file.MutliLinkInode = stat.Ino
+		}
 	}
-	return usage
 }

analyze/file.go

@@ -16,8 +16,12 @@ type File struct {
 	IsDir          bool
 	Files          Files
 	Parent         *File
+	MutliLinkInode uint64 // Inode number of file with multiple links (hard link)
 }
 
+// AlreadyCountedHardlinks holds all files with hardlinks that have already been counted
+type AlreadyCountedHardlinks map[uint64]bool
+
 // Path retruns absolute path of the file
 func (f *File) Path() string {
 	if f.BasePath != "" {
@@ -50,18 +54,30 @@ func (f *File) RemoveFile(file *File) error {
 }
 
 // UpdateStats recursively updates size and item count
-func (f *File) UpdateStats() {
+func (f *File) UpdateStats(links AlreadyCountedHardlinks) {
 	if !f.IsDir {
 		return
 	}
 	totalSize := int64(4096)
 	totalUsage := int64(4096)
 	var itemCount int
 	for _, entry := range f.Files {
-		entry.UpdateStats()
+		if entry.IsDir {
+			entry.UpdateStats(links)
+		}
+
+		itemCount += entry.ItemCount
+
+		if entry.MutliLinkInode > 0 {
+			if !links[entry.MutliLinkInode] {
+				links[entry.MutliLinkInode] = true
+			} else {
+				entry.Flag = 'H'
+				continue
+			}
+		}
 		totalSize += entry.Size
 		totalUsage += entry.Usage
-		itemCount += entry.ItemCount
 	}
 	f.ItemCount = itemCount + 1
 	f.Size = totalSize

tui/tui.go

@@ -194,7 +194,9 @@ func (ui *UI) AnalyzePath(path string, analyzer analyze.Analyzer, parentDir *ana
 			ui.currentDir.Parent = parentDir
 			parentDir.Files = parentDir.Files.RemoveByName(ui.currentDir.Name)
 			parentDir.Files = append(parentDir.Files, ui.currentDir)
-			ui.topDir.UpdateStats()
+
+			links := make(analyze.AlreadyCountedHardlinks, 10)
+			ui.topDir.UpdateStats(links)
 		} else {
 			ui.topDirPath = abspath
 			ui.topDir = ui.currentDir