Merge pull request #34 from alphagov/link-token-mapper-inline-links

Extend LinkTokenMapper to handle anchor tags

lib/answer_composition/link_token_mapper.rb

@@ -6,11 +6,15 @@ def initialize
       @mapping = {}
     end
 
-    def map_links_to_tokens(html_content)
+    def map_links_to_tokens(html_content, exact_path)
       doc = Nokogiri::HTML::DocumentFragment.parse(html_content)
 
       doc.css("a").each do |link|
-        href = link["href"]
+        href = begin
+          URI.join(ensure_absolute_govuk_url(exact_path), link["href"]).to_s
+        rescue URI::InvalidURIError
+          link["href"]
+        end
 
         if mapping.key?(href)
           link["href"] = mapping[href]
@@ -61,9 +65,7 @@ def rewrite_link(link_element)
 
       if (url = link_for_token(token))
         link_element.tap do |el|
-          # We frequently host GOV.UK chat in environments off www.gov.uk
-          # and need links not to be relative so that they will work.
-          el.attr["href"] = ensure_absolute_govuk_url(url)
+          el.attr["href"] = url
 
           # If we have a link where the text is e.g. "link_1" then we should replace
           # it with "source". Showing "link_1" to the user makes it seem like something
@@ -82,11 +84,9 @@ def rewrite_link(link_element)
     end
 
     def ensure_absolute_govuk_url(url)
-      begin
-        relative_uri = URI(url)
-      rescue URI::InvalidURIError
-        return url
-      end
+      # We frequently host GOV.UK chat in environments off www.gov.uk
+      # and need links not to be relative so that they will work.
+      relative_uri = URI(url)
 
       return url if relative_uri.absolute?
 

lib/answer_composition/pipeline/openai_structured_answer_composer.rb

@@ -89,7 +89,7 @@ def system_prompt_context
           page_title: result.title,
           page_description: result.description,
           context_headings: result.heading_hierarchy,
-          context_content: link_token_mapper.map_links_to_tokens(result.html_content),
+          context_content: link_token_mapper.map_links_to_tokens(result.html_content, result.exact_path),
         }
       end
     end

spec/lib/answer_composition/link_token_mapper_spec.rb

@@ -12,17 +12,18 @@
         <li>prove you're self-employed, for example to claim Tax-Free Childcare</li>
         <li>make voluntary <a id="foo" href="https://www.gov.uk/national-insurance/what-national-insurance-is">National Insurance</a> payments</li>
         <li>fill in a <a href="/tax-returns">Tax return</a> each tax year.</li>
+        <li>this is <a href="#some-heading">an anchor tag</a>.</li>
       </ul>
     HTML
   end
 
   describe "#map_links_to_tokens" do
     it "replaces href attributes with tokens" do
-      amended_html = described_class.new.map_links_to_tokens(html)
+      amended_html = described_class.new.map_links_to_tokens(html, "/exact-path")
       parsed_html = Nokogiri::HTML::DocumentFragment.parse(amended_html)
       links = parsed_html.css("a")
 
-      expect(links.length).to eq(4)
+      expect(links.length).to eq(5)
 
       expect(links[0]["href"]).to eq("link_1")
       expect(links[0].text).to eq("Tax return")
@@ -33,9 +34,12 @@
       expect(links[2]["href"]).to eq("link_3")
       expect(links[2].text).to eq("National Insurance")
 
-      #  Duplicate link, so gets the same token as the first link
+      # Duplicate link, so gets the same token as the first link
       expect(links[3]["href"]).to eq("link_1")
       expect(links[3].text).to eq("Tax return")
+
+      expect(links[4]["href"]).to eq("link_4")
+      expect(links[4].text).to eq("an anchor tag")
     end
   end
 
@@ -60,7 +64,7 @@
   describe "#replace_tokens_with_links" do
     it "replaces token-based links with stored links that are absolute URIs" do
       mapper = described_class.new
-      mapper.map_links_to_tokens(html)
+      mapper.map_links_to_tokens(html, "/exact-path")
 
       source = <<~MARKDOWN
         # Tax
@@ -73,6 +77,7 @@
 
         * prove you're self-employed, for example to claim Tax-Free Childcare
         * make voluntary [National Insurance](link_3) payments
+        * do something with an [anchor tag](link_4)
 
         [1]: link_2
       MARKDOWN
@@ -93,11 +98,15 @@
       expect(output)
         .to include("[National Insurance][3]")
         .and include("[3]: https://www.gov.uk/national-insurance/what-national-insurance-is")
+
+      expect(output)
+        .to include("[anchor tag][4]")
+        .and include("[4]: https://www.test.gov.uk/exact-path#some-heading")
     end
 
     it "replaces link text that has not been substituted" do
       mapper = described_class.new
-      mapper.map_links_to_tokens(html)
+      mapper.map_links_to_tokens(html, "/exact-path")
 
       markdown = <<~MARKDOWN
         Send a tax return ([link_1][1])
@@ -112,7 +121,7 @@
     it "handles invalid URIs" do
       html = '<p>Send a tax return to <a href="mailto:<[email protected]>">us</a></p>'
       mapper = described_class.new
-      mapper.map_links_to_tokens(html)
+      mapper.map_links_to_tokens(html, "/exact-path")
 
       markdown = <<~MARKDOWN
         You should send a tax return to [us](link_1)
@@ -140,9 +149,9 @@
   describe "#link_for_token" do
     it "returns the link for a given token" do
       mapper = described_class.new
-      mapper.map_links_to_tokens(html)
+      mapper.map_links_to_tokens(html, "/exact-path")
 
-      expect(mapper.link_for_token("link_1")).to eq("/tax-returns")
+      expect(mapper.link_for_token("link_1")).to eq("https://www.test.gov.uk/tax-returns")
     end
 
     it "returns nil if the token is not in the mapping" do