Add GovspeakPreviewHelper

Author: edavey

app/helpers/govspeak_preview_helper.rb

@@ -0,0 +1,92 @@
+require "delegate"
+
+module GovspeakPreviewHelper
+  include Rails.application.routes.url_helpers
+
+  def govspeak_to_html(govspeak, options = {})
+    processed_govspeak = preprocess_govspeak(govspeak, options)
+    html = markup_to_nokogiri_doc(
+      processed_govspeak,
+      locale: options[:locale],
+    ).to_html
+
+    "<div class=\"govspeak\">#{html}</div>".html_safe
+  end
+
+  def govspeak_headers(govspeak, level = (2..2))
+    build_govspeak_document(govspeak).headers.select do |header|
+      level.cover?(header.level)
+    end
+  end
+
+  def govspeak_header_hierarchy(govspeak)
+    headers = []
+    govspeak_headers(govspeak, 2..3).each do |header|
+      case header.level
+      when 2
+        headers << { header:, children: [] }
+      when 3
+        raise Govspeak::OrphanedHeadingError, header.text if headers.none?
+
+        headers.last[:children] << header
+      end
+    end
+    headers
+  end
+
+  def preprocess_govspeak(govspeak, options)
+    govspeak ||= ""
+    ContentBlockManager::FindAndReplaceEmbedCodesService.call(govspeak) if options[:preview]
+    govspeak = add_heading_numbers(govspeak) if options[:heading_numbering] == :auto
+    govspeak
+  end
+
+private
+
+  def add_heading_numbers(govspeak)
+    h2 = 0
+    h3 = 0
+
+    govspeak.gsub(/^(###|##)\s*(.+)$/) do
+      hashes = Regexp.last_match(1)
+      heading_text = Regexp.last_match(2).strip
+
+      if hashes == "##"
+        h2 += 1
+        h3 = 0
+        num = "#{h2}."
+      else # "###"
+        h2 = 1 if h2.zero?
+        h3 += 1
+        num = "#{h2}.#{h3}"
+      end
+
+      # We have to manually derive and append a slug otherwise when Govspeak
+      # generates the HTML, it includes the <span> and number in the ID. Hence
+      # the `heading_text.parameterize`
+      "#{hashes} <span class=\"number\">#{num} </span>#{heading_text} {##{heading_text.parameterize}}"
+    end
+  end
+
+  def markup_to_nokogiri_doc(govspeak, options = {})
+    govspeak = build_govspeak_document(govspeak, options)
+    doc = Nokogiri::HTML::Document.new
+    doc.encoding = "UTF-8"
+    doc.fragment(govspeak.to_html)
+  end
+
+  def build_govspeak_document(govspeak, options = {})
+    locale = options[:locale]
+
+    Govspeak::Document.new(
+      govspeak,
+      images: [],
+      attachments: [],
+      document_domains: [
+        ContentBlockManager.admin_host,
+        ContentBlockManager.public_host,
+      ],
+      locale:,
+    )
+  end
+end

lib/content_block_manager.rb

@@ -13,6 +13,23 @@ def self.support_email_address
     "feedback-content-modelling@digital.cabinet-office.gov.uk"
   end
 
+  def self.admin_host
+    @admin_host ||= URI(admin_root).host
+  end
+
+  def self.internal_admin_host
+    @internal_admin_host ||=
+      URI(Plek.find("content-block-manager")).host
+  end
+
+  def self.public_host
+    @public_host ||= URI(public_root).host
+  end
+
+  def self.admin_root
+    @admin_root ||= Plek.external_url_for("content-block-manager")
+  end
+
   def self.public_root
     @public_root ||= Plek.website_root
   end

lib/govspeak/orphaned_heading_error.rb

@@ -0,0 +1,10 @@
+module Govspeak
+  class OrphanedHeadingError < StandardError
+    attr_reader :heading
+
+    def initialize(heading)
+      @heading = heading
+      super("Parent heading missing for: #{heading}")
+    end
+  end
+end

test/support/govspeak_preview_helper_test.rb

@@ -0,0 +1,127 @@
+require "test_helper"
+
+class GovspeakPreviewHelperTest < ActionView::TestCase
+  extend Minitest::Spec::DSL
+
+  it "should not alter urls to other sites" do
+    html = govspeak_to_html("no [change](http://external.example.com/page.html)")
+    assert_select_within_html html, "a[href=?]", "http://external.example.com/page.html", text: "change"
+  end
+
+  it "should not alter mailto urls" do
+    html = govspeak_to_html("no [change](mailto:dave@example.com)")
+    assert_select_within_html html, "a[href=?]", "mailto:dave@example.com", text: "change"
+  end
+
+  it "should not alter invalid urls" do
+    html = govspeak_to_html("no [change](not a valid url)")
+    assert_select_within_html html, "a[href=?]", "not%20a%20valid%20url", text: "change"
+  end
+
+  it "should not alter partial urls" do
+    html = govspeak_to_html("no [change](http://)")
+    assert_select_within_html html, "a[href=?]", "http://", text: "change"
+  end
+
+  it "should wrap output with a govspeak class" do
+    html = govspeak_to_html("govspeak-text")
+    assert_select_within_html html, ".govspeak", text: "govspeak-text"
+  end
+
+  it "should mark the govspeak output as html safe" do
+    html = govspeak_to_html("govspeak-text")
+    assert html.html_safe?
+  end
+
+  it "should produce UTF-8 for HTML entities" do
+    html = govspeak_to_html("a ['funny'](/url) thing")
+    assert_select_within_html html, "a", text: "‘funny’"
+  end
+
+  it "does not change css class on buttons" do
+    html = govspeak_to_html("{button}[Link text](https://www.gov.uk){/button}")
+    assert_select_within_html html, "a.govuk-button", "Link text"
+  end
+
+  it "should only extract level two headers by default" do
+    text = "# Heading 1\n\n## Heading 2\n\n### Heading 3"
+    headers = govspeak_headers(text)
+    assert_equal [Govspeak::Header.new("Heading 2", 2, "heading-2")], headers
+  end
+
+  it "should extract header hierarchy from level 2+3 headings" do
+    text = "# Heading 1\n\n## Heading 2a\n\n### Heading 3a\n\n### Heading 3b\n\n#### Ignored heading\n\n## Heading 2b"
+    headers = govspeak_header_hierarchy(text)
+    assert_equal [
+      {
+        header: Govspeak::Header.new("Heading 2a", 2, "heading-2a"),
+        children: [
+          Govspeak::Header.new("Heading 3a", 3, "heading-3a"),
+          Govspeak::Header.new("Heading 3b", 3, "heading-3b"),
+        ],
+      },
+      {
+        header: Govspeak::Header.new("Heading 2b", 2, "heading-2b"),
+        children: [],
+      },
+    ],
+                 headers
+  end
+
+  it "should raise exception when extracting header hierarchy with orphaned level 3 headings" do
+    e = assert_raise(Govspeak::OrphanedHeadingError) { govspeak_header_hierarchy("### Heading 3") }
+    assert_equal "Heading 3", e.heading
+  end
+
+  it "adds numbers to h2 headings" do
+    input = "# main\n\n## first\n\n## second"
+    output = '<div class="govspeak"><h1 id="main">main</h1> <h2 id="first"> <span class="number">1. </span>first</h2> <h2 id="second"> <span class="number">2. </span>second</h2></div>'
+    assert_equivalent_html output, govspeak_to_html(input, heading_numbering: :auto).gsub(/\s+/, " ")
+  end
+
+  it "adds sub-numbers to h3 tags" do
+    input = "## first\n\n### first point one\n\n### first point two\n\n## second\n\n### second point one"
+    expected_output1 = '<h2 id="first"> <span class="number">1. </span>first</h2>'
+    expected_output_1a = '<h3 id="first-point-one"> <span class="number">1.1 </span>first point one</h3>'
+    expected_output_1b = '<h3 id="first-point-two"> <span class="number">1.2 </span>first point two</h3>'
+    expected_output2 = '<h2 id="second"> <span class="number">2. </span>second</h2>'
+    expected_output_2a = '<h3 id="second-point-one"> <span class="number">2.1 </span>second point one</h3>'
+    actual_output = govspeak_to_html(input, heading_numbering: :auto).gsub(/\s+/, " ")
+    assert_match %r{#{expected_output1}}, actual_output
+    assert_match %r{#{expected_output_1a}}, actual_output
+    assert_match %r{#{expected_output_1b}}, actual_output
+    assert_match %r{#{expected_output2}}, actual_output
+    assert_match %r{#{expected_output_2a}}, actual_output
+  end
+
+  it "should not corrupt character encoding of numbered headings" do
+    input = "# café"
+    actual_output = govspeak_to_html(input, heading_numbering: :auto)
+    assert actual_output.include?("café</h1>")
+  end
+
+  describe "admin flavour of govspeak" do
+    it "should wrap admin output with a govspeak class" do
+      html = govspeak_to_html("govspeak-text", { preview: true })
+      assert_select_within_html html, ".govspeak", text: "govspeak-text"
+    end
+
+    it "should mark the admin govspeak output as html safe" do
+      html = govspeak_to_html("govspeak-text", { preview: true })
+      assert html.html_safe?
+    end
+
+    it "should call the embed codes helper" do
+      input = "Here is some Govspeak"
+      expected = "Expected output"
+      ContentBlockManager::FindAndReplaceEmbedCodesService.expects(:call).with(input).returns(expected)
+      govspeak_to_html(input, { preview: true })
+    end
+  end
+
+private
+
+  def collapse_whitespace(string)
+    string.gsub(/\s+/, " ").strip
+  end
+end