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     def description
  61       nil
  62     end
  63
  64     protected
  65
  66     def simple_format(text)
  67       SimpleFormat.new.simple_format(text)
  68     end
  69
  70     def sanitize(text)
  71       Sanitize.clean(text, Sanitize::Config::OSM).html_safe
  72     end
  73
  74     def linkify(text, mode = :urls)
  75       if text.html_safe?
  76         Rinku.auto_link(text, mode, tag_builder.tag_options(:rel => "nofollow noopener noreferrer")).html_safe
  77       else
  78         Rinku.auto_link(text, mode, tag_builder.tag_options(:rel => "nofollow noopener noreferrer"))
  79       end
  80     end
  81   end
  82
  83   class HTML < Base
  84     def to_html
  85       linkify(sanitize(simple_format(self)))
  86     end
  87
  88     def to_text
  89       to_s
  90     end
  91   end
  92
  93   class Markdown < Base
  94     def to_html
  95       linkify(sanitize(document.to_html), :all)
  96     end
  97
  98     def to_text
  99       to_s
 100     end
 101
 102     def image
 103       @image_element = first_image_element(document.root) unless defined? @image_element
 104       @image_element.attr["src"] if @image_element
 105     end
 106
 107     def image_alt
 108       @image_element = first_image_element(document.root) unless defined? @image_element
 109       @image_element.attr["alt"] if @image_element
 110     end
 111
 112     def description
 113       @paragraph_element = first_paragraph_element(document.root) unless defined? @paragraph_element
 114       text_content(@paragraph_element) if @paragraph_element
 115     end
 116
 117     private
 118
 119     def document
 120       @document ||= Kramdown::Document.new(self)
 121     end
 122
 123     def first_image_element(element)
 124       return element if image?(element) && element.attr["src"].present?
 125
 126       element.children.find do |child|
 127         nested_image = first_image_element(child)
 128         break nested_image if nested_image
 129       end
 130     end
 131
 132     def first_paragraph_element(element)
 133       return element if paragraph?(element)
 134
 135       element.children.find do |child|
 136         nested_paragraph = first_paragraph_element(child)
 137         break nested_paragraph if nested_paragraph
 138       end
 139     end
 140
 141     def text_content(element)
 142       text = ""
 143
 144       append_text = lambda do |child|
 145         if child.type == :text
 146           text << child.value
 147         else
 148           child.children.each { |c| append_text.call(c) }
 149         end
 150       end
 151       append_text.call(element)
 152
 153       text
 154     end
 155
 156     def image?(element)
 157       element.type == :img || (element.type == :html_element && element.value == "img")
 158     end
 159
 160     def paragraph?(element)
 161       element.type == :p
 162     end
 163   end
 164
 165   class Text < Base
 166     def to_html
 167       linkify(simple_format(ERB::Util.html_escape(self)))
 168     end
 169
 170     def to_text
 171       to_s
 172     end
 173   end
 174 end