summaryrefslogtreecommitdiff
path: root/lib/gitlab/email/html_to_markdown_parser.rb
diff options
context:
space:
mode:
Diffstat (limited to 'lib/gitlab/email/html_to_markdown_parser.rb')
-rw-r--r--lib/gitlab/email/html_to_markdown_parser.rb29
1 files changed, 29 insertions, 0 deletions
diff --git a/lib/gitlab/email/html_to_markdown_parser.rb b/lib/gitlab/email/html_to_markdown_parser.rb
new file mode 100644
index 00000000000..42dd012308b
--- /dev/null
+++ b/lib/gitlab/email/html_to_markdown_parser.rb
@@ -0,0 +1,29 @@
+# frozen_string_literal: true
+
+require 'nokogiri'
+
+module Gitlab
+ module Email
+ class HtmlToMarkdownParser < Html2Text
+ ADDITIONAL_TAGS = %w[em strong img details].freeze
+ IMG_ATTRS = %w[alt src].freeze
+
+ def self.convert(html)
+ html = fix_newlines(replace_entities(html))
+ doc = Nokogiri::HTML(html)
+
+ HtmlToMarkdownParser.new(doc).convert
+ end
+
+ def iterate_over(node)
+ return super unless ADDITIONAL_TAGS.include?(node.name)
+
+ if node.name == 'img'
+ node.keys.each { |key| node.remove_attribute(key) unless IMG_ATTRS.include?(key) } # rubocop:disable Style/HashEachMethods
+ end
+
+ Kramdown::Document.new(node.to_html, input: 'html').to_commonmark
+ end
+ end
+ end
+end