Change some instances of Nokogiri HTML4 parsing to HTML5 (#31812)

app/helpers/admin/trends/statuses_helper.rb

@@ -5,7 +5,7 @@ module Admin::Trends::StatusesHelper
     text = if status.local?
              status.text.split("\n").first
            else
-             Nokogiri::HTML(status.text).css('html > body > *').first&.text
+             Nokogiri::HTML5(status.text).css('html > body > *').first&.text
            end
 
     return '' if text.blank?

app/lib/emoji_formatter.rb

@@ -24,7 +24,7 @@ class EmojiFormatter
   def to_s
     return html if custom_emojis.empty? || html.blank?
 
-    tree = Nokogiri::HTML.fragment(html)
+    tree = Nokogiri::HTML5.fragment(html)
     tree.xpath('./text()|.//text()[not(ancestor[@class="invisible"])]').to_a.each do |node|
       i                     = -1
       inside_shortname      = false
@@ -43,8 +43,8 @@ class EmojiFormatter
 
           next unless (char_after.nil? || !DISALLOWED_BOUNDING_REGEX.match?(char_after)) && (emoji = emoji_map[shortcode])
 
-          result << Nokogiri::XML::Text.new(text[last_index..shortname_start_index - 1], tree.document) if shortname_start_index.positive?
+          result << tree.document.create_text_node(text[last_index..shortname_start_index - 1]) if shortname_start_index.positive?
-          result << Nokogiri::HTML.fragment(tag_for_emoji(shortcode, emoji))
+          result << tree.document.fragment(tag_for_emoji(shortcode, emoji))
 
           last_index = i + 1
         elsif text[i] == ':' && (i.zero? || !DISALLOWED_BOUNDING_REGEX.match?(text[i - 1]))
@@ -53,7 +53,7 @@ class EmojiFormatter
         end
       end
 
-      result << Nokogiri::XML::Text.new(text[last_index..], tree.document)
+      result << tree.document.create_text_node(text[last_index..])
       node.replace(result)
     end
 

app/lib/plain_text_formatter.rb

@@ -16,7 +16,7 @@ class PlainTextFormatter
     if local?
       text
     else
-      node = Nokogiri::HTML.fragment(insert_newlines)
+      node = Nokogiri::HTML5.fragment(insert_newlines)
       # Elements that are entirely removed with our Sanitize config
       node.xpath('.//iframe|.//math|.//noembed|.//noframes|.//noscript|.//plaintext|.//script|.//style|.//svg|.//xmp').remove
       node.text.chomp

app/services/fetch_oembed_service.rb

@@ -25,7 +25,7 @@ class FetchOEmbedService
     return if html.nil?
 
     @format = @options[:format]
-    page    = Nokogiri::HTML(html)
+    page    = Nokogiri::HTML5(html)
 
     if @format.nil? || @format == :json
       @endpoint_url ||= page.at_xpath('//link[@type="application/json+oembed"]|//link[@type="text/json+oembed"]')&.attribute('href')&.value

app/services/fetch_resource_service.rb

@@ -73,7 +73,7 @@ class FetchResourceService < BaseService
   end
 
   def process_html(response)
-    page      = Nokogiri::HTML(response.body_with_limit)
+    page      = Nokogiri::HTML5(response.body_with_limit)
     json_link = page.xpath('//link[@rel="alternate"]').find { |link| ACTIVITY_STREAM_LINK_TYPES.include?(link['type']) }
 
     process(json_link['href'], terminal: true) unless json_link.nil?

app/services/translate_status_service.rb

@@ -100,7 +100,7 @@ class TranslateStatusService < BaseService
   end
 
   def unwrap_emoji_shortcodes(html)
-    fragment = Nokogiri::HTML.fragment(html)
+    fragment = Nokogiri::HTML5.fragment(html)
     fragment.css('span[translate="no"]').each do |element|
       element.remove_attribute('translate')
       element.replace(element.children) if element.attributes.empty?

lib/sanitize_ext/sanitize_config.rb

@@ -52,7 +52,7 @@ class Sanitize
                  :relative
                end
 
-      current_node.replace(Nokogiri::XML::Text.new(current_node.text, current_node.document)) unless LINK_PROTOCOLS.include?(scheme)
+      current_node.replace(current_node.document.create_text_node(current_node.text)) unless LINK_PROTOCOLS.include?(scheme)
     end
 
     UNSUPPORTED_ELEMENTS_TRANSFORMER = lambda do |env|

lib/tasks/emojis.rake

@@ -13,7 +13,7 @@ def gen_border(codepoint, color)
     view_box[3] += 4
     svg['viewBox'] = view_box.join(' ')
   end
-  g = Nokogiri::XML::Node.new 'g', doc
+  g = doc.create_element('g')
   doc.css('svg > *').each do |elem|
     border_elem = elem.dup