Add latest changes from gitlab-org/gitlab@master

lib/gitlab/email/html_to_markdown_parser.rb

@@ -5,25 +5,46 @@ require 'nokogiri'
 module Gitlab
   module Email
     class HtmlToMarkdownParser < Html2Text
-      ADDITIONAL_TAGS = %w[em strong img details].freeze
-      IMG_ATTRS = %w[alt src].freeze
+      extend Gitlab::Utils::Override
 
+      # List of tags to be converted by Markdown.
+      #
+      # All attributes are removed except for the defined ones.
+      #
+      #   <tag> => [<attribute to keep>, ...]
+      ALLOWED_TAG_ATTRIBUTES = {
+        'em' => [],
+        'strong' => [],
+        'details' => [],
+        'img' => %w[alt src]
+      }.freeze
+      private_constant :ALLOWED_TAG_ATTRIBUTES
+
+      # This redefinition can be removed once https://github.com/soundasleep/html2text_ruby/pull/30
+      # is merged and released.
       def self.convert(html)
         html = fix_newlines(replace_entities(html))
         doc = Nokogiri::HTML(html)
 
-        HtmlToMarkdownParser.new(doc).convert
+        new(doc).convert
       end
 
-      def iterate_over(node)
-        return super unless ADDITIONAL_TAGS.include?(node.name)
+      private
 
-        if node.name == 'img'
-          node.keys.each { |key| node.remove_attribute(key) unless IMG_ATTRS.include?(key) } # rubocop:disable Style/HashEachMethods
-        end
+      override :iterate_over
+      def iterate_over(node)
+        allowed_attributes = ALLOWED_TAG_ATTRIBUTES[node.name]
+        return super unless allowed_attributes
+
+        remove_attributes(node, allowed_attributes)
 
         Kramdown::Document.new(node.to_html, input: 'html').to_commonmark
       end
+
+      def remove_attributes(node, allowed_attributes)
+        to_remove = (node.keys - allowed_attributes)
+        to_remove.each { |key| node.remove_attribute(key) }
+      end
     end
   end
 end