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