wip: wcag structure

lib/tasks/import.rake

@@ -1,20 +1,66 @@
 # frozen_string_literal: true
+
 URL = "https://outline-rocks.github.io/wcag/translations/WCAG21-de/"
 WCAG_22_EN_URL ="https://www.w3.org/TR/WCAG22/"
+WCAG_22_EN_QREF_URL = "https://www.w3.org/WAI/WCAG22/quickref/"
+
+def translate(input)
+  DeepL.translate(input, "EN", "DE")
+end
 
 def import_wcag22en
   doc = Nokogiri::HTML5(URI.open(WCAG_22_EN_URL))
-  standards = [Standard.find_by(name_de: "WCAG 2.2"), Standard.find_by(name_de: "EN 301 549")]
+  qrefdoc = Nokogiri::HTML5(URI.open(WCAG_22_EN_QREF_URL))
+  quick_criteria = qrefdoc.css(".guidelines section:has(h4)").each_with_object({}) do |node, h|
+    h[node.css("h4 strong").first.content] = {
+      quick_criterion_en: node.css(".sc-content .sc-text p").first.content,
+      link_url: node.css(".understanding a").first.attribute("href").value,
+      perm_url: "#{WCAG_22_EN_QREF_URL}##{node.css("article").first.attribute("id").value}"
+    }
+  rescue
+    debugger
+  end
+  link_category = LinkCategory.find_by(name: "Verstehen")
+  qr_category = LinkCategory.find_by(name: "WCAG Quick Reference")
+  # debugger
+  # raise ActiveRecord::Rollback
+
+  standards = [ Standard.find_by(name_de: "WCAG 2.2"), Standard.find_by(name_de: "EN 301 549") ]
   doc.css("section.principle").each do |principle_node|
     _principle_id = principle_node.attributes["id"].value
     principle_title = principle_node.css("h2").first.content.scan(/([a-zA-Z]+)/)
-    principle = Principle.find_by!(name_en: principle_title)
+    principle = Principle.find_or_create_by!(name_en: principle_title)
+
     principle_node.css("section.guideline").each do |guideline_node|
+      next unless guideline_node.css("h3").first
+
+      puts guideline_node.css("h3").first&.content
+      puts guideline_node.css_path
+
+      g_title = guideline_node.css("h3")
+                              .first
+                              .content
+                              .scan(/Guideline \d+\.\d+ (.*)/)
+                              .first
+                              .first
+      g_number = guideline_node.css("h3 bdi")
+                               .first
+                               .content
+                               .scan(/\d+\.(\d+)/)
+                               .first
+                               .first
+      g_text = guideline_node.css("> p").first.content
+
+      guideline = Guideline.find_or_create_by(principle: principle, number: g_number, name_en: g_title)
+      guideline.update(description_en: g_text, description_de: translate(g_text), name_de: translate(g_title)) if guideline.description_de.blank?
+
       guideline_node.css("section.guideline").each do |sc|
+        puts sc.css_path
+        puts sc.css("h4 bdi")
         sc_number, sc_title = sc.css("h4")
                                 .first
                                 .content
-                                .scan(/Success Criterion (\d+\.\d+\.\d+) (.*)/)
+                                .scan(/Success Criterion \d+\.\d+\.(\d+) (.*)/)
                                 .first
         sc_level = sc.css("p.conformance-level").first&.content&.scan(/\(Level (A+)\)/)&.first&.last
         sc_url = sc.css("a.self-link").first.attr("href")
@@ -23,20 +69,35 @@ def import_wcag22en
                       .select { _1.class == Nokogiri::XML::Element }
                       .select { _1.attr("class").nil? || _1.attr("class") == "note" }
                       .reduce("") { |str, node| str + node.to_s }
-                      
-        check = Check.find_or_initialize_by(external_number: sc_number)
+
+        check = Check.find_or_initialize_by(guideline_id: guideline.id, number: sc_number)
         new_standards = Set.new(check.standards)
         new_standards += standards
-        check.name_de              = sc_title
+        check.name_de              = translate sc_title if check.name_de.blank?
         check.name_en              = sc_title
-        check.principle            = principle
         check.standards            = new_standards
         check.applicable_to_app    = check.applicable_to_web = true
-        check.external_number      = sc_number
         check.external_url         = "#{WCAG_22_EN_URL}#{sc_url}"
         check.conformity_level     = sc_level&.to_sym
-        check.conformity_notice_de = sc_conformity_notice
-        check.criterion_de         = "<div>#{full_text}</div>".gsub('href="#', %(href="#{WCAG_22_EN_URL}#))
+        check.conformity_notice_de = translate sc_conformity_notice
+        check.criterion_de         = "<div>#{translate full_text}</div>".gsub('href="#', %(href="#{WCAG_22_EN_URL}#)) if check.criterion_de.blank?
+        qref_attrs = quick_criteria["#{principle.id}.#{guideline.number}.#{sc_number}"]
+        check.quick_criterion_en = qref_attrs&.dig :quick_criterion_en
+        check.quick_criterion_de = qref_attrs&.dig :quick_criterion_en
+
+        if qref_attrs&.dig(:link_url)
+          link = Link.find_or_create_by(link_category:, url: qref_attrs[:link_url])
+          link.update(text: "#{check.full_number} Verstehen")
+          check.links << link
+        end
+
+        if qref_attrs&.dig(:perm_url)
+          link = Link.find_or_create_by(link_category: qr_category, url: qref_attrs[:perm_url])
+          link.update(text: "#{check.full_number} Quick Reference")
+          check.links << link
+        end
+        # check.criterion_details_en = qref_attrs&.dig :criterion_details_en
+        # check.criterion_details_de = qref_attrs&.dig :criterion_details_en
         check.save!
       end
     end