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         url = shorten_host(url, Settings.linkify_hosts, Settings.linkify_hosts_replacement)
  80         shorten_host(url, Settings.linkify_wiki_hosts, Settings.linkify_wiki_hosts_replacement)
  81       end.html_safe
  82     end
  83
  84     private
  85
  86     def shorten_host(url, hosts, hosts_replacement)
  87       %r{^https?://([^/]*)(.*)$}.match(url) do |m|
  88         "#{hosts_replacement}#{m[2]}" if hosts_replacement && hosts&.include?(m[1])
  89       end || url
  90     end
  91   end
  92
  93   class HTML < Base
  94     def to_html
  95       linkify(simple_format(self))
  96     end
  97
  98     def to_text
  99       to_s
 100     end
 101   end
 102
 103   class Markdown < Base
 104     def to_html
 105       linkify(sanitize(document.to_html), :all)
 106     end
 107
 108     def to_text
 109       to_s
 110     end
 111
 112     def image
 113       @image_element = first_image_element(document.root) unless defined? @image_element
 114       @image_element.attr["src"] if @image_element
 115     end
 116
 117     def image_alt
 118       @image_element = first_image_element(document.root) unless defined? @image_element
 119       @image_element.attr["alt"] if @image_element
 120     end
 121
 122     def description
 123       return @description if defined? @description
 124
 125       @description = first_truncated_text_content(document.root)
 126     end
 127
 128     private
 129
 130     def document
 131       @document ||= Kramdown::Document.new(self)
 132     end
 133
 134     def first_image_element(element)
 135       return element if image?(element) && element.attr["src"].present?
 136
 137       element.children.find do |child|
 138         nested_image = first_image_element(child)
 139         break nested_image if nested_image
 140       end
 141     end
 142
 143     def first_truncated_text_content(element)
 144       if paragraph?(element)
 145         truncated_text_content(element)
 146       else
 147         element.children.find do |child|
 148           text = first_truncated_text_content(child)
 149           break text unless text.nil?
 150         end
 151       end
 152     end
 153
 154     def truncated_text_content(element)
 155       text = +""
 156
 157       append_text = lambda do |child|
 158         if child.type == :text
 159           text << child.value
 160         else
 161           child.children.each do |c|
 162             append_text.call(c)
 163             break if text.length > MAX_DESCRIPTION_LENGTH
 164           end
 165         end
 166       end
 167       append_text.call(element)
 168
 169       return nil if text.blank?
 170
 171       text.truncate(MAX_DESCRIPTION_LENGTH)
 172     end
 173
 174     def image?(element)
 175       element.type == :img || (element.type == :html_element && element.value == "img")
 176     end
 177
 178     def paragraph?(element)
 179       element.type == :p || (element.type == :html_element && element.value == "p")
 180     end
 181   end
 182
 183   class Text < Base
 184     def to_html
 185       linkify(simple_format(ERB::Util.html_escape(self)))
 186     end
 187
 188     def to_text
 189       to_s
 190     end
 191   end
 192 end