Import wcag 2.2 import

lib/tasks/import.rake

@@ -3,56 +3,40 @@ URL = "https://outline-rocks.github.io/wcag/translations/WCAG21-de/"
 WCAG_22_EN_URL ="https://www.w3.org/TR/WCAG22/"
 
 def import_wcag22en
-  puts "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")]
-  standard = Standard.find_by(name_de: "WCAG 2.2")
-
   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]+)/)
-    puts principle_title
     principle = Principle.find_by!(name_en: principle_title)
-
     principle_node.css("section.guideline").each do |guideline_node|
-      # puts guideline_node.attr("id")
-      # puts guideline_node.css("h3").size
-      # next unless guideline_node.css("h3").first
-
-      puts "> guideline"
-      # debugger
-      # guideline_title = guideline_node.css("h3 bdi").first&.content
-      # guideline_text = guideline_node.css("p").first
-      # guideline_number = guideline_title.scan(/Guideline \d+\.(\d+).*/).first.last
-      # guideline = principle.guidelines.find_or_create_by(number: guideline_number)
-      # guideline.title = guideline_title.scan(/Richtlinie \d+\.\d+(.*)/).first.last
-      # guideline.text = guideline_text
-
-      # guideline.save!
-
       guideline_node.css("section.guideline").each do |sc|
-        sc_title = sc.css("h4").first.content.scan(/Success Criterion \d+\.\d+\.\d+ (.*)/).first.last
+        sc_number, sc_title = sc.css("h4")
-        sc_number = sc.css("h4").first.content.scan(/Success Criterion (\d+\.\d+\.\d+).*/).first.last
+                                .first
-        puts sc_number
+                                .content
-        sc_level = sc.css("p").first&.content&.scan(/\(Level (A+)\)/)&.first&.last
+                                .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")
-        sc_conformity_notice = sc.css("div.note p").to_a.map{"<p>#{_1.content}</p>"}.join("\n")
+        sc_conformity_notice = sc.css("div.note p").to_a.map(&:to_s).join("\n")
-        text = "<div>#{sc.css('p').to_a[1..].map{"<p>#{_1.content}</p>"}.join}"
+        full_text = sc.children
-        text += "<p>#{sc.css('dl').first&.to_s}</p>" if sc.css("dl").first
+                      .select { _1.class == Nokogiri::XML::Element }
-
+                      .select { _1.attr("class").nil? || _1.attr("class") == "note" }
-
+                      .reduce("") { |str, node| str + node.to_s }
-        # _sc_number = sc_title.scan(/Erfolgskriterium \d+\.\d+\.(\d+).*/).first.last
                       
         check = Check.find_or_initialize_by(external_number: sc_number)
-        check.name_de = sc_title
+        new_standards = Set.new(check.standards)
-        check.principle = principle
+        new_standards += standards
-        check.standards << standard unless check.standards.include?(standard)
+        check.name_de              = sc_title
-        check.applicable_to_app = check.applicable_to_web = true
+        check.name_en              = sc_title
-        check.external_number = sc_number
+        check.principle            = principle
-        check.external_url = "#{WCAG_22_EN_URL}#{sc_url}"
+        check.standards            = new_standards
-        check.conformity_level = sc_level&.to_sym
+        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 = "#{text}</div>".gsub('href="#', %(href="#{WCAG_22_EN_URL}#))
+        check.criterion_de         = "<div>#{full_text}</div>".gsub('href="#', %(href="#{WCAG_22_EN_URL}#))
         check.save!
       end
     end
@@ -130,4 +114,11 @@ namespace :import do
       # raise ActiveRecord::Rollback
     end
   end
+
+  desc "Import EN 301 549"
+  task en301549de: :environment do
+    Check.transaction do
+      import_en301549de
+    end
+  end
 end