Change first_of_value to first_of_hash, and add more test cases (#…

…33647)
mastodon · Jan 22, 2025 · b18caff · b18caff
1 parent e155aab
commit b18caff
Show file tree

Hide file tree

Showing 2 changed files with 43 additions and 5 deletions.
diff --git a/app/lib/link_details_extractor.rb b/app/lib/link_details_extractor.rb
@@ -46,7 +46,7 @@ def type
     end
 
     def image
-      obj = first_of_value(json['image'])
+      obj = first_of_hash(json['image'])
 
       return obj['url'] if obj.is_a?(Hash)
 
@@ -85,15 +85,15 @@ def valid?
     private
 
     def author
-      first_of_value(json['author']) || {}
+      first_of_hash(json['author']) || {}
     end
 
     def publisher
-      first_of_value(json['publisher']) || {}
+      first_of_hash(json['publisher']) || {}
     end
 
-    def first_of_value(arr)
-      arr.is_a?(Array) ? arr.first : arr
+    def first_of_hash(arr)
+      arr.is_a?(Array) ? arr.flatten.find { |item| item.is_a?(Hash) } : arr
     end
 
     def root_array(root)

diff --git a/spec/lib/link_details_extractor_spec.rb b/spec/lib/link_details_extractor_spec.rb
@@ -249,6 +249,44 @@
         expect(subject.author_name).to eq 'Author 1, Author 2'
       end
     end
+
+    context 'with embedded arrays' 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'],
+          }]],
+          'publisher' => [[{
+            '@type' => 'NewsMediaOrganization',
+            'name' => 'Pet News',
+            'url' => 'https://example.com',
+          }]],
+        }.to_json
+      end
+      let(:html) { <<~HTML }
+        <!doctype html>
+        <html>
+        <body>
+          <script type="application/ld+json">
+            #{ld_json}
+          </script>
+        </body>
+        </html>
+      HTML
+
+      it 'gives correct author_name' do
+        expect(subject.author_name).to eq 'Author 1'
+      end
+
+      it 'gives provider_name' do
+        expect(subject.provider_name).to eq 'Pet News'
+      end
+    end
   end
 
   context 'when Open Graph protocol data is present' do