Correctly escape UTF-8 path elements for uploads

changelogs/unreleased/42159-utf8-uploads.yml

+---
+title: Correctly escape UTF-8 path elements for uploads
+merge_request: 16560
+author:
+type: fixed

lib/banzai/filter/relative_link_filter.rb

@@ -50,7 +50,7 @@ module Banzai
       end
 
       def process_link_to_upload_attr(html_attr)
-        path_parts = [html_attr.value]
+        path_parts = [Addressable::URI.unescape(html_attr.value)]
 
         if group
           path_parts.unshift(relative_url_root, 'groups', group.full_path, '-')
@@ -58,13 +58,13 @@ module Banzai
           path_parts.unshift(relative_url_root, project.full_path)
         end
 
-        path = File.join(*path_parts)
+        path = Addressable::URI.escape(File.join(*path_parts))
 
         html_attr.value =
           if context[:only_path]
             path
           else
-            URI.join(Gitlab.config.gitlab.base_url, path).to_s
+            Addressable::URI.join(Gitlab.config.gitlab.base_url, path).to_s
           end
       end
 

spec/lib/banzai/filter/relative_link_filter_spec.rb

@@ -278,18 +278,19 @@ describe Banzai::Filter::RelativeLinkFilter do
         expect(doc.at_css('a')['href']).to eq 'http://example.com'
       end
 
-      it 'supports Unicode filenames' do
+      it 'supports unescaped Unicode filenames' do
         path = '/uploads/한글.png'
-        escaped = Addressable::URI.escape(path)
+        doc = filter(link(path))
 
-        # Stub these methods so the file doesn't actually need to be in the repo
-        allow_any_instance_of(described_class)
-          .to receive(:file_exists?).and_return(true)
-        allow_any_instance_of(described_class)
-          .to receive(:image?).with(path).and_return(true)
+        expect(doc.at_css('a')['href']).to eq("/#{project.full_path}/uploads/%ED%95%9C%EA%B8%80.png")
+      end
 
+      it 'supports escaped Unicode filenames' do
+        path = '/uploads/한글.png'
+        escaped = Addressable::URI.escape(path)
         doc = filter(image(escaped))
-        expect(doc.at_css('img')['src']).to match "/#{project.full_path}/uploads/%ED%95%9C%EA%B8%80.png"
+
+        expect(doc.at_css('img')['src']).to eq("/#{project.full_path}/uploads/%ED%95%9C%EA%B8%80.png")
       end
     end