lib/rich_text.rb

   1 # frozen_string_literal: true
   2
   3 module RichText
   4   SPAMMY_PHRASES = [
   5     "Business Description:", "Additional Keywords:"
   6   ].freeze
   7
   8   MAX_DESCRIPTION_LENGTH = 500
   9
  10   def self.new(format, text)
  11     case format
  12     when "html" then HTML.new(text || "")
  13     when "markdown" then Markdown.new(text || "")
  14     when "text" then Text.new(text || "")
  15     end
  16   end
  17
  18   class SimpleFormat
  19     include ActionView::Helpers::TextHelper
  20     include ActionView::Helpers::OutputSafetyHelper
  21
  22     def sanitize(text, _options = {})
  23       Sanitize.clean(text, Sanitize::Config::OSM).html_safe
  24     end
  25   end
  26
  27   class Base < String
  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, :dir => "auto")
  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       link_attr = 'rel="nofollow noopener noreferrer"'
  78       Rinku.auto_link(ERB::Util.html_escape(text), mode, link_attr) do |url|
  79         %r{^https?://([^/]*)(.*)$}.match(url) do |m|
  80           "#{Settings.linkify_hosts_replacement}#{m[2]}" if Settings.linkify_hosts_replacement &&
  81                                                             Settings.linkify_hosts&.include?(m[1])
  82         end || url
  83       end.html_safe
  84     end
  85   end
  86
  87   class HTML < Base
  88     def to_html
  89       linkify(sanitize(simple_format(self)))
  90     end
  91
  92     def to_text
  93       to_s
  94     end
  95   end
  96
  97   class Markdown < Base
  98     def to_html
  99       linkify(sanitize(document.to_html), :all)
 100     end
 101
 102     def to_text
 103       to_s
 104     end
 105
 106     def image
 107       @image_element = first_image_element(document.root) unless defined? @image_element
 108       @image_element.attr["src"] if @image_element
 109     end
 110
 111     def image_alt
 112       @image_element = first_image_element(document.root) unless defined? @image_element
 113       @image_element.attr["alt"] if @image_element
 114     end
 115
 116     def description
 117       return @description if defined? @description
 118
 119       @description = first_truncated_text_content(document.root)
 120     end
 121
 122     private
 123
 124     def document
 125       @document ||= Kramdown::Document.new(self)
 126     end
 127
 128     def first_image_element(element)
 129       return element if image?(element) && element.attr["src"].present?
 130
 131       element.children.find do |child|
 132         nested_image = first_image_element(child)
 133         break nested_image if nested_image
 134       end
 135     end
 136
 137     def first_truncated_text_content(element)
 138       if paragraph?(element)
 139         truncated_text_content(element)
 140       else
 141         element.children.find do |child|
 142           text = first_truncated_text_content(child)
 143           break text unless text.nil?
 144         end
 145       end
 146     end
 147
 148     def truncated_text_content(element)
 149       text = +""
 150
 151       append_text = lambda do |child|
 152         if child.type == :text
 153           text << child.value
 154         else
 155           child.children.each do |c|
 156             append_text.call(c)
 157             break if text.length > MAX_DESCRIPTION_LENGTH
 158           end
 159         end
 160       end
 161       append_text.call(element)
 162
 163       return nil if text.blank?
 164
 165       text.truncate(MAX_DESCRIPTION_LENGTH)
 166     end
 167
 168     def image?(element)
 169       element.type == :img || (element.type == :html_element && element.value == "img")
 170     end
 171
 172     def paragraph?(element)
 173       element.type == :p || (element.type == :html_element && element.value == "p")
 174     end
 175   end
 176
 177   class Text < Base
 178     def to_html
 179       linkify(simple_format(ERB::Util.html_escape(self)))
 180     end
 181
 182     def to_text
 183       to_s
 184     end
 185   end
 186 end