Fix author names as arrays in linked data. (#30957)

2025-02-11 09:25:08 +00:00 · 2024-07-08 18:04:36 +02:00 · 2024-07-08 18:04:36 +02:00 · fa8e972722
commit fa8e972722
parent f1300ad284
2 changed files with 32 additions and 2 deletions
--- a/app/lib/link_details_extractor.rb
+++ b/app/lib/link_details_extractor.rb
@ -62,7 +62,8 @@ class LinkDetailsExtractor
    end

    def author_name
-      author['name']
+      name = author['name']
+      name.is_a?(Array) ? name.join(', ') : name
    end

    def author_url
@ -294,7 +295,7 @@ class LinkDetailsExtractor
  def html_entities_decode(string)
    return if string.nil?

-    unicode_string = string.encode('UTF-8')
+    unicode_string = string.to_s.encode('UTF-8')
    raise EncodingError, 'cannot convert string to valid UTF-8' unless unicode_string.valid_encoding?

    html_entities.decode(unicode_string)
--- a/spec/lib/link_details_extractor_spec.rb
+++ b/spec/lib/link_details_extractor_spec.rb
@ -192,6 +192,35 @@ RSpec.describe LinkDetailsExtractor do

      include_examples 'structured data'
    end
+
+    context 'with author names as array' do
+      let(:ld_json) do
+        {
+          '@context' => 'https://schema.org',
+          '@type' => 'NewsArticle',
+          'headline' => 'A lot of authors',
+          'description' => 'But we decided to cram them into one',
+          'author' => {
+            '@type' => 'Person',
+            'name' => ['Author 1', 'Author 2'],
+          },
+        }.to_json
+      end
+      let(:html) { <<~HTML }
+        <!doctype html>
+        <html>
+        <body>
+          <script type="application/ld+json">
+            #{ld_json}
+          </script>
+        </body>
+        </html>
+      HTML
+
+      it 'joins author names' do
+        expect(subject.author_name).to eq 'Author 1, Author 2'
+      end
+    end
  end

  context 'when Open Graph protocol data is present' do