Truncate long richtext descriptions

[rails.git] / lib / rich_text.rb

diff --git a/lib/rich_text.rb b/lib/rich_text.rb
index ec5e9e473789a9d70c7e8fb770cac51e14be8137..03800725ad98cc8af64c561288c4e0ff6269f39a 100644 (file)
--- a/lib/rich_text.rb
+++ b/lib/rich_text.rb
@@ -1,57 +1,179 @@
 module RichText
 module RichText

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

   def self.new(format, text)
     case format
   def self.new(format, text)
     case format

-    when "html"; HTML.new(text || "")
-    when "markdown"; Markdown.new(text || "")
-    else; nil
+    when "html" then HTML.new(text || "")
+    when "markdown" then Markdown.new(text || "")
+    when "text" then Text.new(text || "")

     end
   end
 
     end
   end
 

-  class HTML < String
+  class SimpleFormat

     include ActionView::Helpers::TextHelper
     include ActionView::Helpers::TextHelper

+    include ActionView::Helpers::OutputSafetyHelper
+
+    def sanitize(text, _options = {})
+      Sanitize.clean(text, Sanitize::Config::OSM).html_safe
+    end
+  end
+
+  class Base < String

     include ActionView::Helpers::TagHelper
 
     include ActionView::Helpers::TagHelper
 

-    def to_html
-      linkify(sanitize(simple_format(self)))
+    def spam_score
+      link_count = 0
+      link_size = 0
+
+      doc = Nokogiri::HTML(to_html)
+
+      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
+      end
+
+      spammy_phrases = SPAMMY_PHRASES.count do |phrase|
+        doc.content.include?(phrase)
+      end
+
+      ([link_proportion - 0.2, 0.0].max * 200) +
+        (link_count * 40) +
+        (spammy_phrases * 40)

     end
 
     end
 

-    def to_text
-      self
+    def image
+      nil
+    end
+
+    def image_alt
+      nil

     end
 
     end
 

-  private
+    def description
+      nil
+    end
+
+    protected
+
+    def simple_format(text)
+      SimpleFormat.new.simple_format(text)
+    end

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

-    def linkify(text)
+    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
 

-  class Markdown < String
+  class HTML < Base
+    def to_html
+      linkify(sanitize(simple_format(self)))
+    end
+
+    def to_text
+      to_s
+    end
+  end
+
+  class Markdown < Base

     def to_html
     def to_html

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

     end
 
     def to_text
     end
 
     def to_text

-      self
+      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
+
+    def document
+      @document ||= Kramdown::Document.new(self)
+    end
+
+    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
 

-  private
+    def first_paragraph_element(element)
+      return element if paragraph?(element)

 
 

-    def html_parser
-      @@html_renderer ||= Redcarpet::Render::XHTML.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
-      })
+      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
+          child.children.each do |c|
+            append_text.call(c)
+            break if text.length > MAX_DESCRIPTION_LENGTH
+          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
+
+  class Text < Base
+    def to_html
+      linkify(simple_format(ERB::Util.html_escape(self)))
+    end
+
+    def to_text
+      to_s

     end
   end
 end
     end
   end
 end