FIX: Handle sites with more than 1 JSON-LD element (#17095)

A followup to #17007

lib/onebox/json_ld.rb

@@ -12,32 +12,25 @@ module Onebox
     private
 
     def extract(doc)
-      extracted_json = extract_json_from(doc)
+      return {} if Onebox::Helpers::blank?(doc)
-      parsed_json = parse_json(extracted_json)
+
+      doc.css('script[type="application/ld+json"]').each do |element|
+        parsed_json = parse_json(element.text)
 
-      extracted =
         case parsed_json["@type"]
         when MOVIE_JSON_LD_TYPE
-          Onebox::Movie.new(parsed_json)
+          return Onebox::Movie.new(parsed_json).to_h
-        else
-          {}
         end
+      end
 
-      extracted.to_h
+      {}
-    end
-
-    def extract_json_from(doc)
-      return {} if Onebox::Helpers::blank?(doc)
-      json_ld = doc.search('script[type="application/ld+json"]').text
-      return {} if Onebox::Helpers::blank?(json_ld)
-      json_ld
     end
 
     def parse_json(json)
       begin
         JSON[json]
       rescue JSON::ParserError => e
-        Discourse.warn_exception(e, message: "Error parsing JSON-LD json: #{json}")
+        Discourse.warn_exception(e, message: "Error parsing JSON-LD: #{json}")
         {}
       end
     end

spec/lib/onebox/json_ld_spec.rb

@@ -7,27 +7,26 @@ describe Onebox::JsonLd do
     invalid_json = "{\"@type\":invalid-json}"
     doc = Nokogiri::HTML("<script type=\"application/ld+json\">#{invalid_json}</script>")
     Discourse.expects(:warn_exception).with(
-      instance_of(JSON::ParserError), { message: "Error parsing JSON-LD json: #{invalid_json}" }
+      instance_of(JSON::ParserError), { message: "Error parsing JSON-LD: #{invalid_json}" }
     )
 
     json_ld = described_class.new(doc)
-
     expect(json_ld.data).to eq({})
   end
 
-  it 'returns an empty hash if there is no json_ld script tag' do
+  it 'returns an empty hash if there is no JSON-LD script tag' do
     doc = Nokogiri::HTML("<script type=\"something else\"></script>")
     json_ld = described_class.new(doc)
     expect(json_ld.data).to eq({})
   end
 
-  it 'returns an empty hash if there is no json_ld data' do
+  it 'returns an empty hash if there is no JSON-LD data' do
     doc = Nokogiri::HTML("<script type=\"application/ld+json\"></script>")
     json_ld = described_class.new(doc)
     expect(json_ld.data).to eq({})
   end
 
-  it 'returns an empty hash if the type of JSONLD data is not Movie' do
+  it 'returns an empty hash if the type of JSON-LD data is not Movie' do
     doc = Nokogiri::HTML("<script type=\"application/ld+json\">{\"@type\":\"Something Else\",\"aggregateRating\":{\"@type\":\"AggregateRating\",\"ratingCount\":806928,\"bestRating\":10,\"worstRating\":1}}</script>")
     json_ld = described_class.new(doc)
     expect(json_ld.data).to eq({})
@@ -39,6 +38,15 @@ describe Onebox::JsonLd do
     expect(json_ld.data).to eq(expected_movie_hash)
   end
 
+  it 'does not fail when there is more than one JSON-LD element' do
+    doc = Nokogiri::HTML(onebox_response('imdb'))
+    doc.css("body")[0] << "<script type=\"application/ld+json\">{\"@context\":\"http://schema.org\",\"@type\":\"WebPage\",\"url\":\"https:\/\/imdb.com\",\"description\":\"Movies\"}</script>"
+    Discourse.expects(:warn_exception).never
+
+    json_ld = described_class.new(doc)
+    expect(json_ld.data).to eq(expected_movie_hash)
+  end
+
   private
 
   def expected_movie_hash