Merge pull request #159 from alphagov/whitehall-govspeak-upgrade

Whitehall Govspeak upgrade

Author: kevindew

.ruby-version

@@ -1 +1 @@
-2.6.1
+2.6.3

govspeak.gemspec

@@ -37,6 +37,7 @@ library for use in the UK Government Single Domain project'
   s.add_dependency 'i18n', '~> 0.7'
   s.add_dependency 'kramdown', '~> 1.15.0'
   s.add_dependency 'nokogiri', '~> 1.5'
+  s.add_dependency 'nokogumbo', '~> 2'
   s.add_dependency 'rinku', '~> 2.0'
   s.add_dependency "sanitize", "~> 5"
 

lib/govspeak.rb

@@ -4,6 +4,8 @@
 require 'htmlentities'
 require 'kramdown'
 require 'kramdown/parser/govuk'
+require 'nokogiri'
+require 'nokogumbo'
 require 'rinku'
 require 'govuk_publishing_components'
 require 'govspeak/header_extractor'

lib/govspeak/html_validator.rb

@@ -18,7 +18,7 @@ def valid?
 
   # Make whitespace in html tags consistent
   def normalise_html(html)
-    Nokogiri::HTML.parse(html).to_s
+    Nokogiri::HTML5.fragment(html).to_s
   end
 
   def govspeak_to_html

lib/govspeak/post_processor.rb

@@ -1,5 +1,3 @@
-require 'nokogiri'
-
 module Govspeak
   class PostProcessor
     @extensions = []

test/html_validator_test.rb

@@ -101,4 +101,11 @@ class HtmlValidatorTest < Minitest::Test
     assert Govspeak::HtmlValidator.new("{button start}[Start now](https://gov.uk){/button}").valid?
     assert Govspeak::HtmlValidator.new("{button start cross-domain-tracking:UA-XXXXXX-Y}[Start now](https://gov.uk){/button}").valid?
   end
+
+  test "allow HTML tables with and without tbody elements" do
+    # An upgrade of govspeak broke HTML table entries as tbody elements were inserted.
+    # An example of one of these is: https://www.gov.uk/government/publications/what-works-network-membership-requirements/what-works-network
+    assert Govspeak::HtmlValidator.new("<table><tr><td>Hello</td></tr></table>").valid?, "No <tbody> is valid"
+    assert Govspeak::HtmlValidator.new("<table><tbody><tr><td>Hello</td></tr></tbody></table>").valid?, "<tbody> is valid"
+  end
 end