Truncate long richtext descriptions

[rails.git] / lib / rich_text.rb

diff --git a/lib/rich_text.rb b/lib/rich_text.rb
index bb2baddc81991832ef24e115b93183e3a1c4e865..03800725ad98cc8af64c561288c4e0ff6269f39a 100644 (file)
--- a/lib/rich_text.rb
+++ b/lib/rich_text.rb
@@ -1,4 +1,10 @@
 module RichText
 module RichText

+  SPAMMY_PHRASES = [
+    "Business Description:", "Additional Keywords:"
+  ].freeze
+
+  MAX_DESCRIPTION_LENGTH = 500
+

   def self.new(format, text)
     case format
     when "html" then HTML.new(text || "")
   def self.new(format, text)
     case format
     when "html" then HTML.new(text || "")


@@ -11,7 +17,7 @@ module RichText
     include ActionView::Helpers::TextHelper
     include ActionView::Helpers::OutputSafetyHelper
 
     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
       Sanitize.clean(text, Sanitize::Config::OSM).html_safe
     end
   end


@@ -25,18 +31,36 @@ module RichText
 
       doc = Nokogiri::HTML(to_html)
 
 
       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
 
         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
 
       end
 

-      [link_proportion - 0.2, 0.0].max * 200 + link_count * 40
+      ([link_proportion - 0.2, 0.0].max * 200) +
+        (link_count * 40) +
+        (spammy_phrases * 40)
+    end
+
+    def image
+      nil
+    end
+
+    def image_alt
+      nil
+    end
+
+    def description
+      nil

     end
 
     protected
     end
 
     protected


@@ -45,11 +69,15 @@ module RichText
       SimpleFormat.new.simple_format(text)
     end
 
       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?
       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
       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
       end
     end
   end


@@ -62,42 +90,80 @@ module RichText
     def to_text
       to_s
     end
     def to_text
       to_s
     end

-
-    private
-
-    def sanitize(text)
-      Sanitize.clean(text, Sanitize::Config::OSM).html_safe
-    end

   end
 
   class Markdown < Base
     def to_html
   end
 
   class Markdown < Base
     def to_html

-      html_parser.render(self).html_safe
+      linkify(sanitize(document.to_html), :all)

     end
 
     def to_text
       to_s
     end
 
     end
 
     def to_text
       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
+      @paragraph_element = first_paragraph_element(document.root) unless defined? @paragraph_element
+      truncated_text_content(@paragraph_element) if @paragraph_element
+    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
 
     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

+    end

 
 

-      def autolink(link, link_type)
-        if link_type == :email
-          "<a rel=\"nofollow\" href=\"mailto:#{link}\">#{link}</a>"
+    def first_paragraph_element(element)
+      return element if paragraph?(element)
+
+      element.children.find do |child|
+        nested_paragraph = first_paragraph_element(child)
+        break nested_paragraph if nested_paragraph
+      end
+    end
+
+    def truncated_text_content(element)
+      text = ""
+
+      append_text = lambda do |child|
+        if child.type == :text
+          text << child.value

         else
         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
       end

+      append_text.call(element)
+
+      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
 
     end
   end