fix: report allocated disk usage for sparse files (#182)

internal/stat.go

@@ -286,7 +286,7 @@ func find(dir string, filter func(info fs.FileInfo) bool) ([]*file, error) {
 		if !entry.IsDir() {
 			fileChan <- &file{
 				name: entry.Name(),
-				size: fileInfo.Size(),
+				size: diskSize(fileInfo, filepath.Join(dir, entry.Name())),
 			}
 			continue
 		}

internal/sys_darwin.go

@@ -16,6 +16,21 @@
 
 package internal
 
+import (
+	"os"
+	"syscall"
+)
+
 func sysFilter(dir string) bool {
 	return "/dev" != dir
 }
+
+// diskSize returns the actual number of bytes allocated on disk for the file,
+// i.e. allocated blocks (st_blocks * 512), matching `du`'s default behavior.
+// For sparse files this is smaller than the apparent logical size.
+func diskSize(info os.FileInfo, _ string) int64 {
+	if stat, ok := info.Sys().(*syscall.Stat_t); ok {
+		return stat.Blocks * 512 // st_blocks is always in 512-byte units (POSIX)
+	}
+	return info.Size()
+}

internal/sys_linux.go

@@ -16,6 +16,21 @@
 
 package internal
 
+import (
+	"os"
+	"syscall"
+)
+
 func sysFilter(dir string) bool {
 	return dir != "/proc"
 }
+
+// diskSize returns the actual number of bytes allocated on disk for the file,
+// i.e. allocated blocks (st_blocks * 512), matching `du`'s default behavior.
+// For sparse files this is smaller than the apparent logical size.
+func diskSize(info os.FileInfo, _ string) int64 {
+	if stat, ok := info.Sys().(*syscall.Stat_t); ok {
+		return stat.Blocks * 512 // st_blocks is always in 512-byte units (POSIX)
+	}
+	return info.Size()
+}

internal/sys_windows.go

@@ -16,6 +16,39 @@
 
 package internal
 
+import (
+	"os"
+	"syscall"
+	"unsafe"
+)
+
+var (
+	modKernel32                = syscall.NewLazyDLL("kernel32.dll")
+	procGetCompressedFileSizeW = modKernel32.NewProc("GetCompressedFileSizeW")
+)
+
 func sysFilter(_ string) bool {
 	return true
 }
+
+// diskSize returns the actual number of bytes allocated on disk for the file
+// via GetCompressedFileSizeW, matching `du`'s default behavior. For sparse or
+// compressed files this excludes unallocated holes, so it is smaller than the
+// apparent logical size returned by os.FileInfo.Size().
+func diskSize(info os.FileInfo, name string) int64 {
+	p, err := syscall.UTF16PtrFromString(name)
+	if err != nil {
+		return info.Size()
+	}
+
+	var high uint32
+	low, _, _ := procGetCompressedFileSizeW.Call(
+		uintptr(unsafe.Pointer(p)),
+		uintptr(unsafe.Pointer(&high)),
+	)
+	if uint32(low) == 0xFFFFFFFF { // INVALID_FILE_SIZE → call failed
+		return info.Size()
+	}
+
+	return int64(uint64(high)<<32 | uint64(uint32(low)))
+}