Merge branch '61841-fix-encoding-error-in-mr-diffs' into 'master'

Fix encoding error in MR diffs

Closes #61841

See merge request gitlab-org/gitlab-ce!32862

app/models/merge_request_diff_file.rb

@@ -17,7 +17,7 @@ class MergeRequestDiffFile < ApplicationRecord
       if merge_request_diff&.stored_externally?
         merge_request_diff.opening_external_diff do |file|
           file.seek(external_diff_offset)
-          file.read(external_diff_size)
+          force_encode_utf8(file.read(external_diff_size))
         end
       else
         super

changelogs/unreleased/61841-fix-encoding-error-in-mr-diffs.yml

+---
+title: Fix encoding error in MR diffs when using external diffs
+merge_request: 32862
+author: Hiroyuki Sato
+type: fixed

spec/models/merge_request_diff_file_spec.rb

@@ -4,6 +4,7 @@ require 'spec_helper'
 
 describe MergeRequestDiffFile do
   describe '#diff' do
+    context 'when diff is not stored' do
       let(:unpacked) { 'unpacked' }
       let(:packed) { [unpacked].pack('m0') }
 
@@ -28,6 +29,38 @@ describe MergeRequestDiffFile do
       end
     end
 
+    context 'when diff is stored in DB' do
+      let(:file) { create(:merge_request).merge_request_diff.merge_request_diff_files.first }
+
+      it 'returns UTF-8 string' do
+        expect(file.diff.encoding).to eq Encoding::UTF_8
+      end
+    end
+
+    context 'when diff is stored in external storage' do
+      let(:file) { create(:merge_request).merge_request_diff.merge_request_diff_files.first }
+      let(:test_dir) { 'tmp/tests/external-diffs' }
+
+      around do |example|
+        FileUtils.mkdir_p(test_dir)
+
+        begin
+          example.run
+        ensure
+          FileUtils.rm_rf(test_dir)
+        end
+      end
+
+      before do
+        stub_external_diffs_setting(enabled: true, storage_path: test_dir)
+      end
+
+      it 'returns UTF-8 string' do
+        expect(file.diff.encoding).to eq Encoding::UTF_8
+      end
+    end
+  end
+
   describe '#utf8_diff' do
     it 'does not raise error when the diff is binary' do
       subject.diff = "\x05\x00\x68\x65\x6c\x6c\x6f"