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     include ActionView::Helpers::TagHelper
  29
  30     def spam_score
  31       link_count = 0
  32       link_size = 0
  33
  34       doc = Nokogiri::HTML(to_html)
  35
  36       if doc.content.empty?
  37         link_proportion = 0
  38       else
  39         doc.xpath("//a").each do |link|
  40           link_count += 1
  41           link_size += link.content.length
  42         end
  43
  44         link_proportion = link_size.to_f / doc.content.length
  45       end
  46
  47       spammy_phrases = SPAMMY_PHRASES.count do |phrase|
  48         doc.content.include?(phrase)
  49       end
  50
  51       ([link_proportion - 0.2, 0.0].max * 200) +
  52         (link_count * 40) +
  53         (spammy_phrases * 40)
  54     end
  55
  56     def image
  57       nil
  58     end
  59
  60     def image_alt
  61       nil
  62     end
  63
  64     def description
  65       nil
  66     end
  67
  68     protected
  69
  70     def simple_format(text)
  71       SimpleFormat.new.simple_format(text, :dir => "auto")
  72     end
  73
  74     def sanitize(text)
  75       Sanitize.clean(text, Sanitize::Config::OSM).html_safe
  76     end
  77
  78     def linkify(text, mode = :urls)
  79       link_attr = tag_builder.tag_options(:rel => "nofollow noopener noreferrer")
  80       Rinku.auto_link(ERB::Util.html_escape(text), mode, link_attr) do |url|
  81         url = shorten_host(url, Settings.linkify_hosts, Settings.linkify_hosts_replacement)
  82         shorten_host(url, Settings.linkify_wiki_hosts, Settings.linkify_wiki_hosts_replacement)
  83       end.html_safe
  84     end
  85
  86     private
  87
  88     def shorten_host(url, hosts, hosts_replacement)
  89       %r{^https?://([^/]*)(.*)$}.match(url) do |m|
  90         "#{hosts_replacement}#{m[2]}" if hosts_replacement && hosts&.include?(m[1])
  91       end || url
  92     end
  93   end
  94
  95   class HTML < Base
  96     def to_html
  97       linkify(sanitize(simple_format(self)))
  98     end
  99
 100     def to_text
 101       to_s
 102     end
 103   end
 104
 105   class Markdown < Base
 106     def to_html
 107       linkify(sanitize(document.to_html), :all)
 108     end
 109
 110     def to_text
 111       to_s
 112     end
 113
 114     def image
 115       @image_element = first_image_element(document.root) unless defined? @image_element
 116       @image_element.attr["src"] if @image_element
 117     end
 118
 119     def image_alt
 120       @image_element = first_image_element(document.root) unless defined? @image_element
 121       @image_element.attr["alt"] if @image_element
 122     end
 123
 124     def description
 125       return @description if defined? @description
 126
 127       @description = first_truncated_text_content(document.root)
 128     end
 129
 130     private
 131
 132     def document
 133       @document ||= Kramdown::Document.new(self)
 134     end
 135
 136     def first_image_element(element)
 137       return element if image?(element) && element.attr["src"].present?
 138
 139       element.children.find do |child|
 140         nested_image = first_image_element(child)
 141         break nested_image if nested_image
 142       end
 143     end
 144
 145     def first_truncated_text_content(element)
 146       if paragraph?(element)
 147         truncated_text_content(element)
 148       else
 149         element.children.find do |child|
 150           text = first_truncated_text_content(child)
 151           break text unless text.nil?
 152         end
 153       end
 154     end
 155
 156     def truncated_text_content(element)
 157       text = +""
 158
 159       append_text = lambda do |child|
 160         if child.type == :text
 161           text << child.value
 162         else
 163           child.children.each do |c|
 164             append_text.call(c)
 165             break if text.length > MAX_DESCRIPTION_LENGTH
 166           end
 167         end
 168       end
 169       append_text.call(element)
 170
 171       return nil if text.blank?
 172
 173       text.truncate(MAX_DESCRIPTION_LENGTH)
 174     end
 175
 176     def image?(element)
 177       element.type == :img || (element.type == :html_element && element.value == "img")
 178     end
 179
 180     def paragraph?(element)
 181       element.type == :p || (element.type == :html_element && element.value == "p")
 182     end
 183   end
 184
 185   class Text < Base
 186     def to_html
 187       linkify(simple_format(ERB::Util.html_escape(self)))
 188     end
 189
 190     def to_text
 191       to_s
 192     end
 193   end
 194 end