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, _options = {})
  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     def image
  53       nil
  54     end
  55
  56     def image_alt
  57       nil
  58     end
  59
  60     protected
  61
  62     def simple_format(text)
  63       SimpleFormat.new.simple_format(text)
  64     end
  65
  66     def sanitize(text)
  67       Sanitize.clean(text, Sanitize::Config::OSM).html_safe
  68     end
  69
  70     def linkify(text, mode = :urls)
  71       if text.html_safe?
  72         Rinku.auto_link(text, mode, tag_builder.tag_options(:rel => "nofollow noopener noreferrer")).html_safe
  73       else
  74         Rinku.auto_link(text, mode, tag_builder.tag_options(:rel => "nofollow noopener noreferrer"))
  75       end
  76     end
  77   end
  78
  79   class HTML < Base
  80     def to_html
  81       linkify(sanitize(simple_format(self)))
  82     end
  83
  84     def to_text
  85       to_s
  86     end
  87   end
  88
  89   class Markdown < Base
  90     def to_html
  91       linkify(sanitize(document.to_html), :all)
  92     end
  93
  94     def to_text
  95       to_s
  96     end
  97
  98     def image
  99       @image_element = first_image_element(document.root) unless defined? @image_element
 100       @image_element.attr["src"] if @image_element
 101     end
 102
 103     def image_alt
 104       @image_element = first_image_element(document.root) unless defined? @image_element
 105       @image_element.attr["alt"] if @image_element
 106     end
 107
 108     private
 109
 110     def document
 111       @document ||= Kramdown::Document.new(self)
 112     end
 113
 114     def first_image_element(element)
 115       return element if image?(element) && element.attr["src"].present?
 116
 117       element.children.find do |child|
 118         nested_image = first_image_element(child)
 119         break nested_image if nested_image
 120       end
 121     end
 122
 123     def image?(element)
 124       element.type == :img || (element.type == :html_element && element.value == "img")
 125     end
 126   end
 127
 128   class Text < Base
 129     def to_html
 130       linkify(simple_format(ERB::Util.html_escape(self)))
 131     end
 132
 133     def to_text
 134       to_s
 135     end
 136   end
 137 end