From 44906f1f3b91b9def7e1424fec18a3237b1a02f0 Mon Sep 17 00:00:00 2001
From: Konstantin Lebedev <9497591+kmlebedev@users.noreply.github.com>
Date: Wed, 27 Sep 2023 17:40:51 +0500
Subject: [PATCH] fix: avoid error file name too long when writing a file
 (#4876)

---
 weed/command/filer_replication.go             |  2 +-
 weed/replication/sink/localsink/local_sink.go |  2 +-
 weed/util/file_util.go                        | 21 ++++++++++++++++--
 weed/util/file_util_test.go                   | 22 +++++++++++++++++++
 4 files changed, 43 insertions(+), 4 deletions(-)
 create mode 100644 weed/util/file_util_test.go

diff --git a/weed/command/filer_replication.go b/weed/command/filer_replication.go
index ec965a5e3..4fca8158a 100644
--- a/weed/command/filer_replication.go
+++ b/weed/command/filer_replication.go
@@ -99,7 +99,7 @@ func runFilerReplicate(cmd *Command, args []string) bool {
 		if m.OldEntry != nil && m.NewEntry == nil {
 			glog.V(1).Infof("delete: %s", key)
 		} else if m.OldEntry == nil && m.NewEntry != nil {
-			glog.V(1).Infof("   add: %s", key)
+			glog.V(1).Infof("add: %s", key)
 		} else {
 			glog.V(1).Infof("modify: %s", key)
 		}
diff --git a/weed/replication/sink/localsink/local_sink.go b/weed/replication/sink/localsink/local_sink.go
index 70f5cfc9d..134d3985a 100644
--- a/weed/replication/sink/localsink/local_sink.go
+++ b/weed/replication/sink/localsink/local_sink.go
@@ -90,7 +90,7 @@ func (localsink *LocalSink) CreateEntry(key string, entry *filer_pb.Entry, signa
 		return os.Mkdir(key, os.FileMode(entry.Attributes.FileMode))
 	}
 
-	dstFile, err := os.OpenFile(key, os.O_RDWR|os.O_CREATE|os.O_TRUNC, os.FileMode(entry.Attributes.FileMode))
+	dstFile, err := os.OpenFile(util.ToShortFileName(key), os.O_RDWR|os.O_CREATE|os.O_TRUNC, os.FileMode(entry.Attributes.FileMode))
 	if err != nil {
 		return err
 	}
diff --git a/weed/util/file_util.go b/weed/util/file_util.go
index 346de76db..430b6bc86 100644
--- a/weed/util/file_util.go
+++ b/weed/util/file_util.go
@@ -1,16 +1,20 @@
 package util
 
 import (
+	"bytes"
+	"crypto/sha256"
 	"errors"
+	"fmt"
+	"github.com/seaweedfs/seaweedfs/weed/glog"
 	"os"
 	"os/user"
 	"path/filepath"
 	"strings"
 	"time"
-
-	"github.com/seaweedfs/seaweedfs/weed/glog"
 )
 
+const maxFilenameLength = 255
+
 func TestFolderWritable(folder string) (err error) {
 	fileInfo, err := os.Stat(folder)
 	if err != nil {
@@ -106,6 +110,19 @@ func FileNameBase(filename string) string {
 	return filename[:lastDotIndex]
 }
 
+func ToShortFileName(path string) string {
+	fileName := filepath.Base(path)
+	if fileNameBytes := []byte(fileName); len(fileNameBytes) > maxFilenameLength {
+		shaStr := fmt.Sprintf("%x", sha256.Sum256(fileNameBytes))
+		fileNameBase := FileNameBase(fileName)
+		fileExt := fileName[len(fileNameBase):]
+		fileNameBaseBates := bytes.ToValidUTF8([]byte(fileNameBase)[:maxFilenameLength-len([]byte(fileExt))-8], []byte{})
+		shortFileName := string(fileNameBaseBates) + shaStr[len(shaStr)-8:]
+		return filepath.Join(filepath.Dir(path), shortFileName) + fileExt
+	}
+	return path
+}
+
 // Copied from os.WriteFile(), adding file sync.
 // see https://github.com/golang/go/issues/20599
 func WriteFile(name string, data []byte, perm os.FileMode) error {
diff --git a/weed/util/file_util_test.go b/weed/util/file_util_test.go
new file mode 100644
index 000000000..a1f924fed
--- /dev/null
+++ b/weed/util/file_util_test.go
@@ -0,0 +1,22 @@
+package util
+
+import (
+	"testing"
+)
+
+func TestToShortFileName(t *testing.T) {
+	tests := []struct {
+		in    string
+		value string
+	}{
+		{"/data/a/b/c/d.txt", "/data/a/b/c/d.txt"},
+		{"/data/a/b/c/очень_длинное_имя_файла_c_подробным_указанием_наименования_и_содержания_стандартизованных_форм_за_анварь_-_июнь_2023_года(РОГА_И_КОПЫТА_ООО).txt", "/data/a/b/c/очень_длинное_имя_файла_c_подробным_указанием_наименования_и_содержания_стандартизованных_форм_за_анварь_-_июнь_2023_года(РОГА_И_КОПЫТ354fcaf4.txt"},
+		{"/data/a/b/c/очень_длинное_имя_файла_c_подробным_указанием_наименования_и_содержания_стандартизованных_форм_за_анварь_-_июнь_2023_года(РОГА_И_КОПЫТА_ООО)_without_extension", "/data/a/b/c/очень_длинное_имя_файла_c_подробным_указанием_наименования_и_содержания_стандартизованных_форм_за_анварь_-_июнь_2023_года(РОГА_И_КОПЫТА_О21a6e47a"},
+	}
+	for _, p := range tests {
+		got := ToShortFileName(p.in)
+		if got != p.value {
+			t.Errorf("failed to test: got %v, want %v", got, p.value)
+		}
+	}
+}