lib/rich_text.rb

   1 module RichText
   2   SPAMMY_PHRASES = [
   3     "Business Description:", "Additional Keywords:"
   4   ].freeze
   5
   6   MAX_DESCRIPTION_LENGTH = 500
   7
   8   def self.new(format, text)
   9     case format
  10     when "html" then HTML.new(text || "")
  11     when "markdown" then Markdown.new(text || "")
  12     when "text" then Text.new(text || "")
  13     end
  14   end
  15
  16   class SimpleFormat
  17     include ActionView::Helpers::TextHelper
  18     include ActionView::Helpers::OutputSafetyHelper
  19
  20     def sanitize(text, _options = {})
  21       Sanitize.clean(text, Sanitize::Config::OSM).html_safe
  22     end
  23   end
  24
  25   class Base < String
  26     include ActionView::Helpers::TagHelper
  27
  28     def spam_score
  29       link_count = 0
  30       link_size = 0
  31
  32       doc = Nokogiri::HTML(to_html)
  33
  34       if doc.content.empty?
  35         link_proportion = 0
  36       else
  37         doc.xpath("//a").each do |link|
  38           link_count += 1
  39           link_size += link.content.length
  40         end
  41
  42         link_proportion = link_size.to_f / doc.content.length
  43       end
  44
  45       spammy_phrases = SPAMMY_PHRASES.count do |phrase|
  46         doc.content.include?(phrase)
  47       end
  48
  49       ([link_proportion - 0.2, 0.0].max * 200) +
  50         (link_count * 40) +
  51         (spammy_phrases * 40)
  52     end
  53
  54     def image
  55       nil
  56     end
  57
  58     def image_alt
  59       nil
  60     end
  61
  62     def description
  63       nil
  64     end
  65
  66     protected
  67
  68     def simple_format(text)
  69       SimpleFormat.new.simple_format(text)
  70     end
  71
  72     def sanitize(text)
  73       Sanitize.clean(text, Sanitize::Config::OSM).html_safe
  74     end
  75
  76     def linkify(text, mode = :urls)
  77       if text.html_safe?
  78         Rinku.auto_link(text, mode, tag_builder.tag_options(:rel => "nofollow noopener noreferrer")).html_safe
  79       else
  80         Rinku.auto_link(text, mode, tag_builder.tag_options(:rel => "nofollow noopener noreferrer"))
  81       end
  82     end
  83   end
  84
  85   class HTML < Base
  86     def to_html
  87       linkify(sanitize(simple_format(self)))
  88     end
  89
  90     def to_text
  91       to_s
  92     end
  93   end
  94
  95   class Markdown < Base
  96     def to_html
  97       linkify(sanitize(document.to_html), :all)
  98     end
  99
 100     def to_text
 101       to_s
 102     end
 103
 104     def image
 105       @image_element = first_image_element(document.root) unless defined? @image_element
 106       @image_element.attr["src"] if @image_element
 107     end
 108
 109     def image_alt
 110       @image_element = first_image_element(document.root) unless defined? @image_element
 111       @image_element.attr["alt"] if @image_element
 112     end
 113
 114     def description
 115       @paragraph_element = first_paragraph_element(document.root) unless defined? @paragraph_element
 116       truncated_text_content(@paragraph_element) if @paragraph_element
 117     end
 118
 119     private
 120
 121     def document
 122       @document ||= Kramdown::Document.new(self)
 123     end
 124
 125     def first_image_element(element)
 126       return element if image?(element) && element.attr["src"].present?
 127
 128       element.children.find do |child|
 129         nested_image = first_image_element(child)
 130         break nested_image if nested_image
 131       end
 132     end
 133
 134     def first_paragraph_element(element)
 135       return element if paragraph?(element)
 136
 137       element.children.find do |child|
 138         nested_paragraph = first_paragraph_element(child)
 139         break nested_paragraph if nested_paragraph
 140       end
 141     end
 142
 143     def truncated_text_content(element)
 144       text = ""
 145
 146       append_text = lambda do |child|
 147         if child.type == :text
 148           text << child.value
 149         else
 150           child.children.each do |c|
 151             append_text.call(c)
 152             break if text.length > MAX_DESCRIPTION_LENGTH
 153           end
 154         end
 155       end
 156       append_text.call(element)
 157
 158       text.truncate(MAX_DESCRIPTION_LENGTH)
 159     end
 160
 161     def image?(element)
 162       element.type == :img || (element.type == :html_element && element.value == "img")
 163     end
 164
 165     def paragraph?(element)
 166       element.type == :p || (element.type == :html_element && element.value == "p")
 167     end
 168   end
 169
 170   class Text < Base
 171     def to_html
 172       linkify(simple_format(ERB::Util.html_escape(self)))
 173     end
 174
 175     def to_text
 176       to_s
 177     end
 178   end
 179 end