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
  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 sanitize(text)
  59       Sanitize.clean(text, Sanitize::Config::OSM).html_safe
  60     end
  61
  62     def linkify(text, mode = :urls)
  63       if text.html_safe?
  64         Rinku.auto_link(text, mode, tag_builder.tag_options(:rel => "nofollow noopener noreferrer")).html_safe
  65       else
  66         Rinku.auto_link(text, mode, tag_builder.tag_options(:rel => "nofollow noopener noreferrer"))
  67       end
  68     end
  69   end
  70
  71   class HTML < Base
  72     def to_html
  73       linkify(sanitize(simple_format(self)))
  74     end
  75
  76     def to_text
  77       to_s
  78     end
  79   end
  80
  81   class Markdown < Base
  82     def to_html
  83       linkify(sanitize(Kramdown::Document.new(self).to_html), :all)
  84     end
  85
  86     def to_text
  87       to_s
  88     end
  89   end
  90
  91   class Text < Base
  92     def to_html
  93       linkify(simple_format(ERB::Util.html_escape(self)))
  94     end
  95
  96     def to_text
  97       to_s
  98     end
  99   end
 100 end