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