Make gumbo use configurable

lib/capybara.rb

@@ -38,6 +38,8 @@ module Capybara
     #   See {Capybara.configure}
     # @!method javascript_driver
     #   See {Capybara.configure}
+    # @!method allow_gumbo
+    #   See {Capybara.configure}
     Config::OPTIONS.each do |method|
       def_delegators :config, method, "#{method}="
     end
@@ -86,6 +88,7 @@ module Capybara
     # [test_id = Symbol/String/nil] Optional attribute to match locator aginst with builtin selectors along with id (Default: nil)
     # [predicates_wait = Boolean]  Whether Capybaras predicate matchers use waiting behavior by default (Default: true)
     # [default_normalize_ws = Boolean] Whether text predicates and matchers use normalize whitespace behaviour (Default: false)
+    # [allow_gumbo = Boolean] When `nokogumbo` is available, whether it will be used to parse HTML strings (Default: true)
     #
     # === DSL Options
     #
@@ -366,7 +369,7 @@ module Capybara
     # @return [Nokogiri::HTML::Document]      HTML document
     #
     def HTML(html) # rubocop:disable Naming/MethodName
-      if Nokogiri.respond_to?(:HTML5) # Nokogumbo installed
+      if Nokogiri.respond_to?(:HTML5) && Capybara.allow_gumbo # Nokogumbo installed and allowed for use
         Nokogiri::HTML5(html).tap do |document|
           document.xpath('//textarea').each do |textarea|
             # The Nokogumbo HTML5 parser already returns spec compliant contents
@@ -522,6 +525,7 @@ Capybara.configure do |config|
   config.test_id = nil
   config.predicates_wait = true
   config.default_normalize_ws = false
+  config.allow_gumbo = true
 end
 
 Capybara.register_driver :rack_test do |app|

lib/capybara/config.rb

@@ -7,12 +7,13 @@ module Capybara
   class Config
     extend Forwardable
 
-    OPTIONS = %i[app reuse_server threadsafe default_wait_time server default_driver javascript_driver].freeze
+    OPTIONS = %i[app reuse_server threadsafe default_wait_time server default_driver javascript_driver allow_gumbo].freeze
 
     attr_accessor :app
     attr_reader :reuse_server, :threadsafe
     attr_reader :session_options
     attr_writer :default_driver, :javascript_driver
+    attr_accessor :allow_gumbo
 
     SessionConfig::OPTIONS.each do |method|
       def_delegators :session_options, method, "#{method}="

lib/capybara/spec/spec_helper.rb

@@ -35,6 +35,7 @@ module Capybara
         Capybara.test_id = nil
         Capybara.predicates_wait = true
         Capybara.default_normalize_ws = false
+        Capybara.allow_gumbo = true
         reset_threadsafe
       end
 

spec/basic_node_spec.rb

@@ -110,6 +110,17 @@ RSpec.describe Capybara do
       expect(string.find('//form/input[@name="meh"]')).not_to be_disabled
     end
 
+    it 'drops illegal fragments when using gumbo' do
+      skip "libxml is less strict thatn Gumbo" unless Nokogiri.respond_to?(:HTML5)
+      expect(Capybara.string("<td>1</td>")).not_to have_css('td')
+    end
+
+    it 'can disable use of gumbo' do
+      skip "Test doesn't make sense unlesss nokogumbo is loaded" unless Nokogiri.respond_to?(:HTML5)
+      Capybara.allow_gumbo = false
+      expect(Capybara.string("<td>1</td>")).to have_css('td')
+    end
+
     describe '#title' do
       it 'returns the page title' do
         expect(string.title).to eq('simple_node')