Merge remote-tracking branch 'upstream/pull/5951'

[rails.git] / lib / rich_text.rb

diff --git a/lib/rich_text.rb b/lib/rich_text.rb
index d9c799611a7752f24a54dd582123449d0d8d4216..f65f5137a724b6f4b1cccf4ef4ddfbae88c0f240 100644 (file)
--- a/lib/rich_text.rb
+++ b/lib/rich_text.rb
@@ -25,8 +25,6 @@ module RichText
   end
 
   class Base < String
-    include ActionView::Helpers::TagHelper
-
     def spam_score
       link_count = 0
       link_size = 0
@@ -65,6 +63,48 @@ module RichText
       nil
     end
 
+    def truncate_html(max_length = nil, img_length = 1000)
+      html_doc = to_html
+      return html_doc if max_length.nil?
+
+      doc = Nokogiri::HTML::DocumentFragment.parse(html_doc)
+      keep_or_discards = %w[p h1 h2 h3 h4 h5 h6 pre a table ul ol dl]
+      accumulated_length = 0
+      exceeded_node_parent = nil
+      truncated = false
+
+      doc.traverse do |node|
+        if accumulated_length >= max_length
+          if node == exceeded_node_parent
+            exceeded_node_parent = node.parent
+            node.remove if keep_or_discards.include?(node.name)
+          else
+            node.remove
+          end
+          next
+        end
+
+        next unless node.children.empty?
+
+        if node.text?
+          accumulated_length += node.text.length
+        elsif node.name == "img"
+          accumulated_length += img_length
+        end
+
+        if accumulated_length >= max_length
+          truncated = true
+          exceeded_node_parent = node.parent
+          node.remove
+        end
+      end
+
+      {
+        :truncated => truncated,
+        :html => doc.to_html.html_safe
+      }
+    end
+
     protected
 
     def simple_format(text)
@@ -76,19 +116,35 @@ module RichText
     end
 
     def linkify(text, mode = :urls)
-      link_attr = tag_builder.tag_options(:rel => "nofollow noopener noreferrer")
+      link_attr = 'rel="nofollow noopener noreferrer" dir="auto"'
       Rinku.auto_link(ERB::Util.html_escape(text), mode, link_attr) do |url|
-        %r{^https?://([^/]*)(.*)$}.match(url) do |m|
-          "#{Settings.linkify_hosts_replacement}#{m[2]}" if Settings.linkify_hosts_replacement &&
-                                                            Settings.linkify_hosts&.include?(m[1])
-        end || url
+        url = shorten_host(url, Settings.linkify_hosts, Settings.linkify_hosts_replacement)
+        shorten_host(url, Settings.linkify_wiki_hosts, Settings.linkify_wiki_hosts_replacement) do |path|
+          path.sub(Regexp.new(Settings.linkify_wiki_optional_path_prefix || ""), "")
+        end
       end.html_safe
     end
+
+    private
+
+    def shorten_host(url, hosts, hosts_replacement)
+      %r{^(https?://([^/]*))(.*)$}.match(url) do |m|
+        scheme_host, host, path = m.captures
+        if hosts&.include?(host)
+          path = yield(path) if block_given?
+          if hosts_replacement
+            "#{hosts_replacement}#{path}"
+          else
+            "#{scheme_host}#{path}"
+          end
+        end || url
+      end || url
+    end
   end
 
   class HTML < Base
     def to_html
-      linkify(sanitize(simple_format(self)))
+      linkify(simple_format(self))
     end
 
     def to_text
@@ -124,7 +180,25 @@ module RichText
     private
 
     def document
-      @document ||= Kramdown::Document.new(self)
+      return @document if @document
+
+      @document = Kramdown::Document.new(self)
+
+      should_get_dir_auto = lambda do |el|
+        dir_auto_types = [:p, :header, :codespan, :codeblock, :pre, :ul, :ol, :table, :dl, :math]
+        return true if dir_auto_types.include?(el.type)
+        return true if el.type == :a && el.children.length == 1 && el.children[0].type == :text && el.children[0].value == el.attr["href"]
+
+        false
+      end
+
+      add_dir = lambda do |element|
+        element.attr["dir"] ||= "auto" if should_get_dir_auto.call(element)
+        element.children.each(&add_dir)
+      end
+      add_dir.call(@document.root)
+
+      @document
     end
 
     def first_image_element(element)