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) do |path|
  81           path.sub(Regexp.new(Settings.linkify_wiki_optional_path_prefix || ""), "")
  82         end
  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         scheme_host, host, path = m.captures
  91         if hosts&.include?(host)
  92           path = yield(path) if block_given?
  93           if hosts_replacement
  94             "#{hosts_replacement}#{path}"
  95           else
  96             "#{scheme_host}#{path}"
  97           end
  98         end || url
  99       end || url
 100     end
 101   end
 102
 103   class HTML < Base
 104     def to_html
 105       linkify(simple_format(self))
 106     end
 107
 108     def to_text
 109       to_s
 110     end
 111   end
 112
 113   class Markdown < Base
 114     def to_html
 115       linkify(sanitize(document.to_html), :all)
 116     end
 117
 118     def to_text
 119       to_s
 120     end
 121
 122     def image
 123       @image_element = first_image_element(document.root) unless defined? @image_element
 124       @image_element.attr["src"] if @image_element
 125     end
 126
 127     def image_alt
 128       @image_element = first_image_element(document.root) unless defined? @image_element
 129       @image_element.attr["alt"] if @image_element
 130     end
 131
 132     def description
 133       return @description if defined? @description
 134
 135       @description = first_truncated_text_content(document.root)
 136     end
 137
 138     private
 139
 140     def document
 141       return @document if @document
 142
 143       @document = Kramdown::Document.new(self)
 144
 145       should_get_dir_auto = lambda do |el|
 146         dir_auto_types = [:p, :header, :codespan, :codeblock, :pre, :ul, :ol, :table, :dl, :math]
 147         return true if dir_auto_types.include?(el.type)
 148         return true if el.type == :a && el.children.length == 1 && el.children[0].type == :text && el.children[0].value == el.attr["href"]
 149
 150         false
 151       end
 152
 153       add_dir = lambda do |element|
 154         element.attr["dir"] ||= "auto" if should_get_dir_auto.call(element)
 155         element.children.each(&add_dir)
 156       end
 157       add_dir.call(@document.root)
 158
 159       @document
 160     end
 161
 162     def first_image_element(element)
 163       return element if image?(element) && element.attr["src"].present?
 164
 165       element.children.find do |child|
 166         nested_image = first_image_element(child)
 167         break nested_image if nested_image
 168       end
 169     end
 170
 171     def first_truncated_text_content(element)
 172       if paragraph?(element)
 173         truncated_text_content(element)
 174       else
 175         element.children.find do |child|
 176           text = first_truncated_text_content(child)
 177           break text unless text.nil?
 178         end
 179       end
 180     end
 181
 182     def truncated_text_content(element)
 183       text = +""
 184
 185       append_text = lambda do |child|
 186         if child.type == :text
 187           text << child.value
 188         else
 189           child.children.each do |c|
 190             append_text.call(c)
 191             break if text.length > MAX_DESCRIPTION_LENGTH
 192           end
 193         end
 194       end
 195       append_text.call(element)
 196
 197       return nil if text.blank?
 198
 199       text.truncate(MAX_DESCRIPTION_LENGTH)
 200     end
 201
 202     def image?(element)
 203       element.type == :img || (element.type == :html_element && element.value == "img")
 204     end
 205
 206     def paragraph?(element)
 207       element.type == :p || (element.type == :html_element && element.value == "p")
 208     end
 209   end
 210
 211   class Text < Base
 212     def to_html
 213       linkify(simple_format(ERB::Util.html_escape(self)))
 214     end
 215
 216     def to_text
 217       to_s
 218     end
 219   end
 220 end