Fix code search results with non-ASCII data43531-500-error-searching-wiki-incompatible-character-encodings-utf-8-and-ascii-8bit

Gitlab::Git::Popen#popen now returns ASCII-8BIT (binary) data always, so we need
to handle that explicitly.

3 files changed, 35 insertions, 1 deletions

diff --git a/changelogs/unreleased/43531-500-error-searching-wiki-incompatible-character-encodings-utf-8-and-ascii-8bit.yml b/changelogs/unreleased/43531-500-error-searching-wiki-incompatible-character-encodings-utf-8-and-ascii-8bit.yml
new file mode 100644
index 00000000000..173710412a5
--- /dev/null
+++ b/changelogs/unreleased/43531-500-error-searching-wiki-incompatible-character-encodings-utf-8-and-ascii-8bit.yml
@@ -0,0 +1,5 @@
+---
+title: Fix code and wiki search results pages when non-ASCII text is displayed
+merge_request: 17413
+author:
+type: fixed
diff --git a/lib/gitlab/search_results.rb b/lib/gitlab/search_results.rb
index 5a5ae7f19d4..781783f4d97 100644
--- a/lib/gitlab/search_results.rb
+++ b/lib/gitlab/search_results.rb
@@ -1,6 +1,8 @@
 module Gitlab
   class SearchResults
     class FoundBlob
+      include EncodingHelper
+
       attr_reader :id, :filename, :basename, :ref, :startline, :data, :project_id
 
       def initialize(opts = {})
@@ -9,7 +11,7 @@ module Gitlab
         @basename = opts.fetch(:basename, nil)
         @ref = opts.fetch(:ref, nil)
         @startline = opts.fetch(:startline, nil)
-        @data = opts.fetch(:data, nil)
+        @data = encode_utf8(opts.fetch(:data, nil))
         @per_page = opts.fetch(:per_page, 20)
         @project_id = opts.fetch(:project_id, nil)
       end
diff --git a/spec/lib/gitlab/project_search_results_spec.rb b/spec/lib/gitlab/project_search_results_spec.rb
index 1ebb0105cf5..d8250e4b4c6 100644
--- a/spec/lib/gitlab/project_search_results_spec.rb
+++ b/spec/lib/gitlab/project_search_results_spec.rb
@@ -1,3 +1,4 @@
+# coding: utf-8
 require 'spec_helper'
 
 describe Gitlab::ProjectSearchResults do
@@ -105,6 +106,32 @@ describe Gitlab::ProjectSearchResults do
         end
       end
 
+      context 'when the search returns non-ASCII data' do
+        context 'with UTF-8' do
+          let(:results) { project.repository.search_files_by_content("файл", 'master') }
+
+          it 'returns results as UTF-8' do
+            expect(subject.filename).to eq('encoding/russian.rb')
+            expect(subject.basename).to eq('encoding/russian')
+            expect(subject.ref).to eq('master')
+            expect(subject.startline).to eq(1)
+            expect(subject.data).to eq("Хороший файл")
+          end
+        end
+
+        context 'with ISO-8859-1' do
+          let(:search_result) { "master:encoding/iso8859.txt\x001\x00\xC4\xFC\nmaster:encoding/iso8859.txt\x002\x00\nmaster:encoding/iso8859.txt\x003\x00foo\n".force_encoding(Encoding::ASCII_8BIT) }
+
+          it 'returns results as UTF-8' do
+            expect(subject.filename).to eq('encoding/iso8859.txt')
+            expect(subject.basename).to eq('encoding/iso8859')
+            expect(subject.ref).to eq('master')
+            expect(subject.startline).to eq(1)
+            expect(subject.data).to eq("Äü\n\nfoo")
+          end
+        end
+      end
+
       context "when filename has extension" do
         let(:search_result) { "master:CONTRIBUTE.md\x005\x00- [Contribute to GitLab](#contribute-to-gitlab)\n" }