From b18caff5b6309fe9fad42dac6b7342b246223688 Mon Sep 17 00:00:00 2001
From: Jonathan de Jong <jonathandejong02@gmail.com>
Date: Wed, 22 Jan 2025 15:02:05 +0100
Subject: [PATCH] Change `first_of_value` to `first_of_hash`, and add more test
cases (#33647)
---
app/lib/link_details_extractor.rb | 10 +++----
spec/lib/link_details_extractor_spec.rb | 38 +++++++++++++++++++++++++
2 files changed, 43 insertions(+), 5 deletions(-)
diff --git a/app/lib/link_details_extractor.rb b/app/lib/link_details_extractor.rb
index fe7f23f481..a8004f2925 100644
--- a/app/lib/link_details_extractor.rb
+++ b/app/lib/link_details_extractor.rb
@@ -46,7 +46,7 @@ class LinkDetailsExtractor
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 @@ class LinkDetailsExtractor
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
index 36d6f22b00..cb072c4870 100644
--- a/spec/lib/link_details_extractor_spec.rb
+++ b/spec/lib/link_details_extractor_spec.rb
@@ -249,6 +249,44 @@ RSpec.describe LinkDetailsExtractor do
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