Filter out trademark and copyright from emojis

and display them as unicode characters rather than
images.

changelogs/unreleased/17652-trademark-and-copyright-symbols-converted-to-smiley.yml

+---
+title: Do not convert unicode versions of trademark, copyright,
+  and registered trademark to emoji
+merge_request: 45457
+author:
+type: fixed

lib/banzai/filter/emoji_filter.rb

@@ -8,6 +8,7 @@ module Banzai
     # Based on HTML::Pipeline::EmojiFilter
     class EmojiFilter < HTML::Pipeline::Filter
       IGNORED_ANCESTOR_TAGS = %w(pre code tt).to_set
+      IGNORE_UNICODE_EMOJIS = %w(™ © ®).freeze
 
       def call
         doc.search(".//text()").each do |node|
@@ -60,7 +61,11 @@ module Banzai
 
       # Build a regexp that matches all valid unicode emojis names.
       def self.emoji_unicode_pattern
-        @emoji_unicode_pattern ||= /(#{Gitlab::Emoji.emojis_unicodes.map { |moji| Regexp.escape(moji) }.join('|')})/
+        @emoji_unicode_pattern ||=
+          begin
+            filtered_emojis = Gitlab::Emoji.emojis_unicodes - IGNORE_UNICODE_EMOJIS
+            /(#{filtered_emojis.map { |moji| Regexp.escape(moji) }.join('|')})/
+          end
       end
 
       private

spec/lib/banzai/filter/emoji_filter_spec.rb

@@ -21,6 +21,20 @@ RSpec.describe Banzai::Filter::EmojiFilter do
     expect(doc.to_html).to match Regexp.escape(exp)
   end
 
+  it 'ignores unicode versions of trademark, copyright, and registered trademark' do
+    exp = act = '<p>™ © ®</p>'
+    doc = filter(act)
+    expect(doc.to_html).to match Regexp.escape(exp)
+  end
+
+  it 'replaces name versions of trademark, copyright, and registered trademark' do
+    doc = filter('<p>:tm: :copyright: :registered:</p>')
+
+    expect(doc.css('gl-emoji')[0].text).to eq '™'
+    expect(doc.css('gl-emoji')[1].text).to eq '©'
+    expect(doc.css('gl-emoji')[2].text).to eq '®'
+  end
+
   it 'correctly encodes the URL' do
     doc = filter('<p>:+1:</p>')
     expect(doc.css('gl-emoji').first.text).to eq '👍'