Fix browse elements timeouts

[rails.git] / lib / rich_text.rb

diff --git a/lib/rich_text.rb b/lib/rich_text.rb
index 2b7e310094058cd7f22670743b2681510057ccd3..bdf9c37ca7359eec1f2cea57c2243178d8ce7791 100644 (file)
--- a/lib/rich_text.rb
+++ b/lib/rich_text.rb
@@ -1,17 +1,23 @@
 module RichText
+  SPAMMY_PHRASES = [
+    "Business Description:", "Additional Keywords:"
+  ].freeze
+
+  MAX_DESCRIPTION_LENGTH = 500
+
   def self.new(format, text)
     case format
-    when "html"; HTML.new(text || "")
-    when "markdown"; Markdown.new(text || "")
-    when "text"; Text.new(text || "")
-    else; nil
+    when "html" then HTML.new(text || "")
+    when "markdown" then Markdown.new(text || "")
+    when "text" then Text.new(text || "")
     end
   end
 
   class SimpleFormat
     include ActionView::Helpers::TextHelper
+    include ActionView::Helpers::OutputSafetyHelper
 
-    def sanitize(text)
+    def sanitize(text, _options = {})
       Sanitize.clean(text, Sanitize::Config::OSM).html_safe
     end
   end
@@ -25,31 +31,53 @@ module RichText
 
       doc = Nokogiri::HTML(to_html)
 
-      if doc.content.length > 0
+      if doc.content.empty?
+        link_proportion = 0
+      else
         doc.xpath("//a").each do |link|
           link_count += 1
           link_size += link.content.length
         end
 
-        link_proportion = link_size.to_f / doc.content.length.to_f
-      else
-        link_proportion = 0
+        link_proportion = link_size.to_f / doc.content.length
+      end
+
+      spammy_phrases = SPAMMY_PHRASES.count do |phrase|
+        doc.content.include?(phrase)
       end
 
-      return [link_proportion - 0.2, 0.0].max * 200 + link_count * 20
+      ([link_proportion - 0.2, 0.0].max * 200) +
+        (link_count * 40) +
+        (spammy_phrases * 40)
     end
 
-  protected
+    def image
+      nil
+    end
+
+    def image_alt
+      nil
+    end
+
+    def description
+      nil
+    end
+
+    protected
 
     def simple_format(text)
       SimpleFormat.new.simple_format(text)
     end
 
-    def linkify(text)
+    def sanitize(text)
+      Sanitize.clean(text, Sanitize::Config::OSM).html_safe
+    end
+
+    def linkify(text, mode = :urls)
       if text.html_safe?
-        Rinku.auto_link(text, :urls, tag_options(:rel => "nofollow")).html_safe
+        Rinku.auto_link(text, mode, tag_builder.tag_options(:rel => "nofollow noopener noreferrer")).html_safe
       else
-        Rinku.auto_link(text, :urls, tag_options(:rel => "nofollow"))
+        Rinku.auto_link(text, mode, tag_builder.tag_options(:rel => "nofollow noopener noreferrer"))
       end
     end
   end
@@ -60,48 +88,87 @@ module RichText
     end
 
     def to_text
-      self.to_s
-    end
-
-  private
-
-    def sanitize(text)
-      Sanitize.clean(text, Sanitize::Config::OSM).html_safe
+      to_s
     end
   end
 
   class Markdown < Base
     def to_html
-      html_parser.render(self).html_safe
+      linkify(sanitize(document.to_html), :all)
     end
 
     def to_text
-      self.to_s
+      to_s
+    end
+
+    def image
+      @image_element = first_image_element(document.root) unless defined? @image_element
+      @image_element.attr["src"] if @image_element
+    end
+
+    def image_alt
+      @image_element = first_image_element(document.root) unless defined? @image_element
+      @image_element.attr["alt"] if @image_element
+    end
+
+    def description
+      return @description if defined? @description
+
+      @description = first_truncated_text_content(document.root)
     end
 
-  private
+    private
 
-    def html_parser
-      @@html_renderer ||= Renderer.new({
-        :filter_html => true, :safe_links_only => true
-      })
-      @@html_parser ||= Redcarpet::Markdown.new(@@html_renderer, {
-        :no_intra_emphasis => true, :autolink => true, :space_after_headers => true
-      })
+    def document
+      @document ||= Kramdown::Document.new(self)
     end
 
-    class Renderer < Redcarpet::Render::XHTML
-      def link(link, title, alt_text)
-        "<a rel=\"nofollow\" href=\"#{link}\">#{alt_text}</a>"
+    def first_image_element(element)
+      return element if image?(element) && element.attr["src"].present?
+
+      element.children.find do |child|
+        nested_image = first_image_element(child)
+        break nested_image if nested_image
       end
+    end
 
-      def autolink(link, link_type)
-        if link_type == :email
-          "<a rel=\"nofollow\" href=\"mailto:#{link}\">#{link}</a>"
+    def first_truncated_text_content(element)
+      if paragraph?(element)
+        truncated_text_content(element)
+      else
+        element.children.find do |child|
+          text = first_truncated_text_content(child)
+          break text unless text.nil?
+        end
+      end
+    end
+
+    def truncated_text_content(element)
+      text = ""
+
+      append_text = lambda do |child|
+        if child.type == :text
+          text << child.value
         else
-          "<a rel=\"nofollow\" href=\"#{link}\">#{link}</a>"
+          child.children.each do |c|
+            append_text.call(c)
+            break if text.length > MAX_DESCRIPTION_LENGTH
+          end
         end
-      end 
+      end
+      append_text.call(element)
+
+      return nil if text.blank?
+
+      text.truncate(MAX_DESCRIPTION_LENGTH)
+    end
+
+    def image?(element)
+      element.type == :img || (element.type == :html_element && element.value == "img")
+    end
+
+    def paragraph?(element)
+      element.type == :p || (element.type == :html_element && element.value == "p")
     end
   end
 
@@ -111,7 +178,7 @@ module RichText
     end
 
     def to_text
-      self.to_s
+      to_s
     end
   end
 end