fix: properly parse UTF-8(multibyte) file paths in git output

kaorukobo · jcouball · commit 8e6a11e5f374 · 2025-08-20T14:33:43.000-07:00
diff --git a/lib/git/lib.rb b/lib/git/lib.rb
@@ -643,7 +643,7 @@ def ls_tree(sha, opts = {})
       args << opts[:path] if opts[:path]
 
       command_lines('ls-tree', *args).each do |line|
-        (info, filenm) = line.split("\t")
+        (info, filenm) = split_status_line(line)
         (mode, type, sha) = info.split
         data[type][filenm] = { mode: mode, sha: sha }
       end
@@ -905,9 +905,9 @@ def ls_files(location = nil)
       location ||= '.'
       {}.tap do |files|
         command_lines('ls-files', '--stage', location).each do |line|
-          (info, file) = line.split("\t")
+          (info, file) = split_status_line(line)
           (mode, sha, stage) = info.split
-          files[unescape_quoted_path(file)] = {
+          files[file] = {
             path: file, mode_index: mode, sha_index: sha, stage: stage
           }
         end
@@ -956,7 +956,9 @@ def ignored_files
     end
 
     def untracked_files
-      command_lines('ls-files', '--others', '--exclude-standard', chdir: @git_work_dir)
+      command_lines('ls-files', '--others', '--exclude-standard', chdir: @git_work_dir).map do |f|
+        unescape_quoted_path(f)
+      end
     end
 
     def config_remote(name)
@@ -1602,7 +1604,7 @@ def self.warn_if_old_command(lib) # rubocop:disable Naming/PredicateMethod
 
     def parse_diff_path_status(args)
       command_lines('diff', *args).each_with_object({}) do |line, memo|
-        status, path = line.split("\t")
+        status, path = split_status_line(line)
         memo[path] = status
       end
     end
@@ -1727,7 +1729,7 @@ def parse_diff_stats_output(lines)
 
     def parse_stat_lines(lines)
       lines.map do |line|
-        insertions_s, deletions_s, filename = line.split("\t")
+        insertions_s, deletions_s, filename = split_status_line(line)
         {
           filename: filename,
           insertions: insertions_s.to_i,
@@ -1736,6 +1738,12 @@ def parse_stat_lines(lines)
       end
     end
 
+    def split_status_line(line)
+      parts = line.split("\t")
+      parts[-1] = unescape_quoted_path(parts[-1]) if parts.any?
+      parts
+    end
+
     def build_final_stats_hash(file_stats)
       {
         total: build_total_stats(file_stats),
@@ -1965,7 +1973,7 @@ def diff_as_hash(diff_command, opts = [])
       # update index before diffing to avoid spurious diffs
       command('status')
       command_lines(diff_command, *opts).each_with_object({}) do |line, memo|
-        info, file = line.split("\t")
+        info, file = split_status_line(line)
         mode_src, mode_dest, sha_src, sha_dest, type = info.split
 
         memo[file] = {
diff --git a/tests/units/test_status.rb b/tests/units/test_status.rb
@@ -235,4 +235,54 @@ def test_changed_cache
       assert(!git.status.changed?('test_file_1'))
     end
   end
+
+  def test_multibyte_path
+    # a name consisting of UTF-8 characters
+    multibyte_name = "\u30DE\u30EB\u30C1\u30D0\u30A4\u30C8\u6587\u5B57\u30D5\u30A1\u30A4\u30EB\u263A"
+
+    in_temp_dir do |_path|
+      `git init`
+
+      File.write('file1', 'contents1')
+      `git add file1`
+      `git commit -m "my message"`
+
+      git = Git.open('.')
+
+      # Test added
+      File.write("#{multibyte_name}_added.txt", 'contents_mb_added')
+      `git add #{multibyte_name}_added.txt`
+
+      status = git.status
+      assert_equal(1, status.added.size)
+      assert_equal(["#{multibyte_name}_added.txt"], status.added.keys)
+
+      # Test untracked
+      File.write("#{multibyte_name}_untracked.txt", 'contents_mb_untracked')
+
+      status = git.status
+      assert_equal(1, status.untracked.size)
+      assert_equal(["#{multibyte_name}_untracked.txt"], status.untracked.keys)
+
+      # Test changed
+      File.write("#{multibyte_name}_changed.txt", 'original_content')
+      `git add #{multibyte_name}_changed.txt`
+      `git commit -m "add multibyte file"`
+      File.write("#{multibyte_name}_changed.txt", 'modified_content')
+
+      status = git.status
+      assert_equal(1, status.changed.size)
+      assert_equal(["#{multibyte_name}_changed.txt"], status.changed.keys)
+
+      # Test deleted
+      File.write("#{multibyte_name}_deleted.txt", 'to_be_deleted')
+      `git add #{multibyte_name}_deleted.txt`
+      `git commit -m "add file to be deleted"`
+      File.delete("#{multibyte_name}_deleted.txt")
+
+      status = git.status
+      assert_equal(1, status.deleted.size)
+      assert_equal(["#{multibyte_name}_deleted.txt"], status.deleted.keys)
+    end
+  end
 end
