lib/rich_text.rb

   1 # frozen_string_literal: true
   2
   3 module RichText
   4   DESCRIPTION_MAX_LENGTH = 500
   5   DESCRIPTION_WORD_BREAK_THRESHOLD_LENGTH = 450
   6
   7   def self.new(format, text)
   8     case format
   9     when "html" then HTML.new(text || "")
  10     when "markdown" then Markdown.new(text || "")
  11     when "text" then Text.new(text || "")
  12     end
  13   end
  14
  15   class SimpleFormat
  16     include ActionView::Helpers::TextHelper
  17     include ActionView::Helpers::OutputSafetyHelper
  18
  19     def sanitize(text, _options = {})
  20       Sanitize.clean(text, Sanitize::Config::OSM).html_safe
  21     end
  22   end
  23
  24   class Base < String
  25     def image
  26       nil
  27     end
  28
  29     def image_alt
  30       nil
  31     end
  32
  33     def description
  34       nil
  35     end
  36
  37     def truncate_html(max_length = nil, img_length = 1000)
  38       html_doc = to_html
  39       return html_doc if max_length.nil?
  40
  41       doc = Nokogiri::HTML::DocumentFragment.parse(html_doc)
  42       keep_or_discards = %w[p h1 h2 h3 h4 h5 h6 pre a table ul ol dl]
  43       accumulated_length = 0
  44       exceeded_node_parent = nil
  45       truncated = false
  46
  47       doc.traverse do |node|
  48         if accumulated_length >= max_length
  49           if node == exceeded_node_parent
  50             exceeded_node_parent = node.parent
  51             node.remove if keep_or_discards.include?(node.name)
  52           else
  53             node.remove
  54           end
  55           next
  56         end
  57
  58         next unless node.children.empty?
  59
  60         if node.text?
  61           accumulated_length += node.text.length
  62         elsif node.name == "img"
  63           accumulated_length += img_length
  64         end
  65
  66         if accumulated_length >= max_length
  67           truncated = true
  68           exceeded_node_parent = node.parent
  69           node.remove
  70         end
  71       end
  72
  73       {
  74         :truncated => truncated,
  75         :html => doc.to_html.html_safe
  76       }
  77     end
  78
  79     protected
  80
  81     def simple_format(text)
  82       SimpleFormat.new.simple_format(text, :dir => "auto")
  83     end
  84
  85     def sanitize(text)
  86       Sanitize.clean(text, Sanitize::Config::OSM).html_safe
  87     end
  88
  89     def linkify(text, mode = :urls)
  90       link_attr = 'rel="nofollow noopener noreferrer" dir="auto"'
  91       Rinku.auto_link(ERB::Util.html_escape(text), mode, link_attr) do |url|
  92         url = shorten_host(url, Settings.linkify_hosts, Settings.linkify_hosts_replacement)
  93         shorten_host(url, Settings.linkify_wiki_hosts, Settings.linkify_wiki_hosts_replacement) do |path|
  94           path.sub(Regexp.new(Settings.linkify_wiki_optional_path_prefix || ""), "")
  95         end
  96       end.html_safe
  97     end
  98
  99     private
 100
 101     def shorten_host(url, hosts, hosts_replacement)
 102       %r{^(https?://([^/]*))(.*)$}.match(url) do |m|
 103         scheme_host, host, path = m.captures
 104         if hosts&.include?(host)
 105           path = yield(path) if block_given?
 106           if hosts_replacement
 107             "#{hosts_replacement}#{path}"
 108           else
 109             "#{scheme_host}#{path}"
 110           end
 111         end || url
 112       end || url
 113     end
 114   end
 115
 116   class HTML < Base
 117     def to_html
 118       linkify(simple_format(self))
 119     end
 120
 121     def to_text
 122       to_s
 123     end
 124   end
 125
 126   class Markdown < Base
 127     def to_html
 128       linkify(sanitize(document.to_html), :all)
 129     end
 130
 131     def to_text
 132       to_s
 133     end
 134
 135     def image
 136       @image_element = first_image_element(document.root) unless defined? @image_element
 137       @image_element.attr["src"] if @image_element
 138     end
 139
 140     def image_alt
 141       @image_element = first_image_element(document.root) unless defined? @image_element
 142       @image_element.attr["alt"] if @image_element
 143     end
 144
 145     def description
 146       return @description if defined? @description
 147
 148       @description = first_truncated_text_content(document.root)
 149     end
 150
 151     private
 152
 153     def document
 154       return @document if @document
 155
 156       @document = Kramdown::Document.new(self)
 157
 158       should_get_dir_auto = lambda do |el|
 159         dir_auto_types = [:p, :header, :codespan, :codeblock, :pre, :ul, :ol, :table, :dl, :math]
 160         return true if dir_auto_types.include?(el.type)
 161         return true if el.type == :a && el.children.length == 1 && el.children[0].type == :text && el.children[0].value == el.attr["href"]
 162
 163         false
 164       end
 165
 166       add_dir = lambda do |element|
 167         element.attr["dir"] ||= "auto" if should_get_dir_auto.call(element)
 168         element.children.each(&add_dir)
 169       end
 170       add_dir.call(@document.root)
 171
 172       @document
 173     end
 174
 175     def first_image_element(element)
 176       return element if image?(element) && element.attr["src"].present?
 177
 178       element.children.find do |child|
 179         nested_image = first_image_element(child)
 180         break nested_image if nested_image
 181       end
 182     end
 183
 184     def first_truncated_text_content(element)
 185       if paragraph?(element)
 186         truncated_text_content(element)
 187       else
 188         element.children.find do |child|
 189           text = first_truncated_text_content(child)
 190           break text unless text.nil?
 191         end
 192       end
 193     end
 194
 195     def truncated_text_content(element)
 196       text = +""
 197
 198       append_text = lambda do |child|
 199         if child.type == :text
 200           text << child.value
 201         else
 202           child.children.each do |c|
 203             append_text.call(c)
 204             break if text.length > DESCRIPTION_MAX_LENGTH
 205           end
 206         end
 207       end
 208       append_text.call(element)
 209
 210       return nil if text.blank?
 211
 212       text_truncated_to_word_break = text.truncate(DESCRIPTION_MAX_LENGTH, :separator => /(?<!\s)\s+/)
 213
 214       if text_truncated_to_word_break.length >= DESCRIPTION_WORD_BREAK_THRESHOLD_LENGTH
 215         text_truncated_to_word_break
 216       else
 217         text.truncate(DESCRIPTION_MAX_LENGTH)
 218       end
 219     end
 220
 221     def image?(element)
 222       element.type == :img || (element.type == :html_element && element.value == "img")
 223     end
 224
 225     def paragraph?(element)
 226       element.type == :p || (element.type == :html_element && element.value == "p")
 227     end
 228   end
 229
 230   class Text < Base
 231     def to_html
 232       linkify(simple_format(ERB::Util.html_escape(self)))
 233     end
 234
 235     def to_text
 236       to_s
 237     end
 238   end
 239 end