Adding Schema.org tags for better discovery (#688)

pulibrary · Oct 23, 2024 · b7bf538 · b7bf538
1 parent 7a6e7dd
commit b7bf538
Show file tree

Hide file tree

Showing 6 changed files with 149 additions and 0 deletions.
diff --git a/app/helpers/schema_org_helper.rb b/app/helpers/schema_org_helper.rb
@@ -0,0 +1,43 @@
+# frozen_string_literal: true
+
+# rubocop:disable Rails/OutputSafety
+module SchemaOrgHelper
+  def keywords_helper(subjects)
+    keywords_json = subjects.map do |subject|
+      '"' + html_escape(subject) + '"'
+    end
+
+    html = "[" + keywords_json.join(",") + "]"
+    html.html_safe
+  end
+
+  def authors_helper(authors)
+    authors_json = authors.each.map do |author|
+      json_str = "\n\t\t\t{\n\t\t\t" + '"name": ' + '"' + author.value + '"'
+      if author.affiliation_name.present?
+        json_str += ",\n\t\t\t" + '"affiliation": ' + '"' + author.affiliation_name + '"'
+      end
+      if author.orcid.present?
+        json_str += ",\n\t\t\t" + '"identifier": ' + '"' + author.orcid + '"'
+      end
+      json_str += "\n\t\t\t}"
+      json_str
+    end
+    html = "[" + authors_json.join(",") + "]"
+    html.html_safe
+  end
+
+  def license_helper(licenses)
+    if licenses.count == 0
+      ""
+    else
+      html = '"license": {'
+      html += "\n\t\t\t" + '"@type": ' + '"Dataset"' + ",\n" \
+              "\t\t\t" + '"text": ' + '"' + licenses[0]['identifier'] + '"' + ",\n" \
+              "\t\t\t" + '"url": ' + '"' + licenses[0]['uri'] + '"'
+      html += "\n\t\t\t},"
+      html.html_safe
+    end
+  end
+end
+# rubocop:enable Rails/OutputSafety
diff --git a/app/views/layouts/blacklight/base.html.erb b/app/views/layouts/blacklight/base.html.erb
@@ -33,6 +33,7 @@
       More info: https://getbootstrap.com/docs/4.6/components/popovers/
     -->
     <script type="text/javascript" src="https://cdn.jsdelivr.net/npm/[email protected]/dist/umd/popper.min.js"></script>
+    <%= render partial: 'shared/schema_org'%>
   </head>
   <body class="<%= render_body_class %>">
     <nav id="skip-link" role="navigation" aria-label="<%= t('blacklight.skip_links.label') %>">

diff --git a/app/views/shared/_schema_org.html.erb b/app/views/shared/_schema_org.html.erb
@@ -0,0 +1,28 @@
+<% if [email protected]?%>
+<!-- Renders Schema.org tags -->
+    <script type ="application/ld+json">
+    { "@context": "http://schema.org",
+        "@type": "Dataset",
+        "identifier": "<%= @document.doi_url %>",
+        "name": "<%= @document.title %>",
+        "author": <%= authors_helper(@document.authors_ordered) %>,
+        "version": "1",
+        "description": "<%= @document.description %>",
+        "keywords": <%= keywords_helper(@document.subject) %>,
+        "citation": 
+        [
+            "<a href="<%= @document.doi_url%>" target="_blank"> "<%= @document.citation.to_s('apa')%>" </a>"
+        ],
+        "schemaVersion": "https://schema.org/version/28.0",
+        <%= license_helper(@document.rights_enhanced) %>
+        "provider":
+        {
+            "@type": "Organization",
+            "name": "Princeton University"
+        }
+    }
+    </script>
+<% end %>
+
+
+
diff --git a/spec/helpers/schema_org_helper_spec.rb b/spec/helpers/schema_org_helper_spec.rb
@@ -0,0 +1,67 @@
+# frozen_string_literal: true
+
+require 'rails_helper'
+RSpec.describe SchemaOrgHelper, type: :helper do
+  let(:author_with_orcid) do
+    {
+      "value" => "some_name",
+      "identifier" => {
+        "scheme" => "ORCID",
+        "value" => "0000-0000-1111-1111"
+      },
+      "affiliations" => [
+        { "value" => "some_affiliation" }
+      ]
+    }
+  end
+
+  let(:author_without_orcid) do
+    {
+      "value" => "some_name"
+    }
+  end
+
+  let(:license) do
+    {
+      "identifier" => "some_identifier",
+      "uri" => "some_uri"
+    }
+  end
+
+  describe "#render_sidebar_related_identifiers" do
+    it "renders keywords" do
+      expect(helper.keywords_helper(['a', 'b'])).to eq '["a","b"]'
+    end
+
+    it "does not render keywords" do
+      expect(helper.keywords_helper([])).to eq '[]'
+    end
+
+    it "renders authors with orcid and affiliation" do
+      author = [Author.new(author_with_orcid)]
+      # Expected output with orcid and affiliation
+      expected_output = "[\n\t\t\t{\n\t\t\t\"name\": \"some_name\",\n\t\t\t\"affiliation\": \"some_affiliation\",\n\t\t\t\"identifier\": \"0000-0000-1111-1111\"\n\t\t\t}]"
+
+      # Call the helper function with the authors array and verify the result
+      expect(helper.authors_helper(author)).to eq expected_output
+    end
+
+    it "renders authors without orcid nor affiliation" do
+      author = [Author.new(author_without_orcid)]
+      # Expected output without orcid and affiliation
+      expected_output = "[\n\t\t\t{\n\t\t\t\"name\": \"some_name\"\n\t\t\t}]"
+
+      # Call the helper function with the authors array and verify the result
+      expect(helper.authors_helper(author)).to eq expected_output
+    end
+
+    it "render one license" do
+      expected_output = "\"license\": {\n\t\t\t\"@type\": \"Dataset\",\n\t\t\t\"text\": \"some_identifier\",\n\t\t\t\"url\": \"some_uri\"\n\t\t\t},"
+      expect(helper.license_helper([license])).to eq expected_output
+    end
+
+    it "renders no license" do
+      expect(helper.license_helper([])).to eq ""
+    end
+  end
+end
diff --git a/spec/system/search_pdc_results_page_spec.rb b/spec/system/search_pdc_results_page_spec.rb
@@ -41,4 +41,9 @@
     click_on "View larger"
     expect(page).to have_content "2022\n2022"
   end
+
+  it 'does not render Schema.org tags' do
+    visit '/?search_field=all_fields&q='
+    expect(page.html.include?('"@context": "http://schema.org",')).to be false
+  end
 end
diff --git a/spec/system/show_pdc_describe_record_spec.rb b/spec/system/show_pdc_describe_record_spec.rb
@@ -59,6 +59,11 @@
     expect(first_filename_spot).to eq("readme.txt")
   end
 
+  it 'renders Schema.org tags' do
+    visit '/catalog/doi-10-34770-bm4s-t361'
+    expect(page.html.include?('"@context": "http://schema.org",')).to be true
+  end
+
   context "when crawler visits the site" do
     before do
       allow_any_instance_of(CatalogController).to receive(:agent_is_crawler?).and_return(true)