Use the page title as the default FAQPage question

Previously we used "Summary" to name a section of html that didn't have
an h2 tag preceding it.  Unfortunately, Google appears to not include
sections in rich results unless all the text actually appears on the page.

We still get rich snippets, we're just missing the first section.  Hopefully
this will be remedied by using the page title text instead.

This wasn't flagged by the rich text testing tool.

CHANGELOG.md

@@ -9,6 +9,7 @@
 
 ## Unreleased
 
+* Fix some minor issues with the FAQPage schema ([PR #1142](https://github.com/alphagov/govuk_publishing_components/pull/1142))
 * Add error state to select component ([PR #1141](https://github.com/alphagov/govuk_publishing_components/pull/1141))
 * Add size option to label ([PR #1140](https://github.com/alphagov/govuk_publishing_components/pull/1140))
 

lib/govuk_publishing_components/presenters/machine_readable/faq_page_schema.rb

@@ -44,7 +44,7 @@ def questions_and_answers_markup
 
       # Generates a hash of questions and associated information:
       # - question: the text in the h2 tag preceding other markup. Questions are
-      #             used to key the hash. "Summary" is set as the default, as
+      #             used to key the hash. The page title is set as the default, as
       #             there is often a preamble in guides before any h2 is set.
       #
       # - :answer: the markup that is not an h2 tag. It is associated with the
@@ -55,7 +55,7 @@ def questions_and_answers_markup
       def question_and_answers(html)
         doc = Nokogiri::HTML(html)
 
-        question = "Summary"
+        question = page.title
 
         # rubocop:disable Style/IfInsideElse
         doc.xpath("html/body").children.each_with_object({}) do |element, q_and_as|

spec/lib/govuk_publishing_components/presenters/schema_org_spec.rb

@@ -94,7 +94,7 @@
         expect(q_and_a.second["acceptedAnswer"]["text"].strip).to eq("<p>Give it a treat</p>")
       end
 
-      it "handles missing h2s at the start of the body" do
+      it "handles missing h2s at the start of the body by using the page title" do
         part_body = "<p>First catch your dragon</p>
                      <h2 id='step-two'>Step two</h2>
                      <p>Give it a treat</p>
@@ -109,7 +109,7 @@
           body: part_body
         ).structured_data['mainEntity']
 
-        expect(q_and_a.first["name"]).to eq("Summary")
+        expect(q_and_a.first["name"]).to eq("How to train your dragon")
         expect(q_and_a.first["url"]).to eq("http://www.dev.gov.uk/how-to-train-your-dragon")
         expect(q_and_a.first["acceptedAnswer"]["text"].strip).to eq("<p>First catch your dragon</p>")
       end
@@ -151,6 +151,7 @@ def dragon_guide
           canonical_url: "http://www.dev.gov.uk/how-to-train-your-dragon/insurance"
         ) do |random_item|
           random_item.merge(
+            "title" => "How to train your dragon",
             "base_path" => "/how-to-train-your-dragon"
           )
         end