lib/rich_text.rb

   1 module RichText
   2   SPAMMY_PHRASES = [
   3     "Business Description:", "Additional Keywords:"
   4   ].freeze
   5
   6   def self.new(format, text)
   7     case format
   8     when "html" then HTML.new(text || "")
   9     when "markdown" then Markdown.new(text || "")
  10     when "text" then Text.new(text || "")
  11     end
  12   end
  13
  14   class SimpleFormat
  15     include ActionView::Helpers::TextHelper
  16     include ActionView::Helpers::OutputSafetyHelper
  17
  18     def sanitize(text)
  19       Sanitize.clean(text, Sanitize::Config::OSM).html_safe
  20     end
  21   end
  22
  23   class Base < String
  24     include ActionView::Helpers::TagHelper
  25
  26     def spam_score
  27       link_count = 0
  28       link_size = 0
  29
  30       doc = Nokogiri::HTML(to_html)
  31
  32       if doc.content.empty?
  33         link_proportion = 0
  34       else
  35         doc.xpath("//a").each do |link|
  36           link_count += 1
  37           link_size += link.content.length
  38         end
  39
  40         link_proportion = link_size.to_f / doc.content.length.to_f
  41       end
  42
  43       spammy_phrases = SPAMMY_PHRASES.count do |phrase|
  44         doc.content.include?(phrase)
  45       end
  46
  47       [link_proportion - 0.2, 0.0].max * 200 +
  48         link_count * 40 +
  49         spammy_phrases * 40
  50     end
  51
  52     protected
  53
  54     def simple_format(text)
  55       SimpleFormat.new.simple_format(text)
  56     end
  57
  58     def linkify(text)
  59       if text.html_safe?
  60         Rinku.auto_link(text, :urls, tag_options(:rel => "nofollow")).html_safe
  61       else
  62         Rinku.auto_link(text, :urls, tag_options(:rel => "nofollow"))
  63       end
  64     end
  65   end
  66
  67   class HTML < Base
  68     def to_html
  69       linkify(sanitize(simple_format(self)))
  70     end
  71
  72     def to_text
  73       to_s
  74     end
  75
  76     private
  77
  78     def sanitize(text)
  79       Sanitize.clean(text, Sanitize::Config::OSM).html_safe
  80     end
  81   end
  82
  83   class Markdown < Base
  84     def to_html
  85       Markdown.html_parser.render(self).html_safe
  86     end
  87
  88     def to_text
  89       to_s
  90     end
  91
  92     def self.html_renderer
  93       @html_renderer ||= Redcarpet::Render::XHTML.new(:filter_html => true, :safe_links_only => true, :link_attributes => { :rel => "nofollow" })
  94     end
  95
  96     def self.html_parser
  97       @html_parser ||= Redcarpet::Markdown.new(html_renderer, :no_intra_emphasis => true, :autolink => true, :space_after_headers => true)
  98     end
  99   end
 100
 101   class Text < Base
 102     def to_html
 103       linkify(simple_format(ERB::Util.html_escape(self)))
 104     end
 105
 106     def to_text
 107       to_s
 108     end
 109   end
 110 end