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   URL_UNSAFE_CHARS = "[^\\w!#$%&'*+,./:;=?@_~^\\-]"
   7
   8   def self.new(format, text)
   9     case format
  10     when "html" then HTML.new(text || "")
  11     when "markdown" then Markdown.new(text || "")
  12     when "text" then Text.new(text || "")
  13     end
  14   end
  15
  16   class SimpleFormat
  17     include ActionView::Helpers::TextHelper
  18     include ActionView::Helpers::OutputSafetyHelper
  19
  20     def sanitize(text, _options = {})
  21       Sanitize.clean(text, Sanitize::Config::OSM).html_safe
  22     end
  23   end
  24
  25   class Base < String
  26     def image
  27       nil
  28     end
  29
  30     def image_alt
  31       nil
  32     end
  33
  34     def description
  35       nil
  36     end
  37
  38     def truncate_html(max_length = nil, img_length = 1000)
  39       html_doc = to_html
  40       return html_doc if max_length.nil?
  41
  42       doc = Nokogiri::HTML::DocumentFragment.parse(html_doc)
  43       keep_or_discards = %w[p h1 h2 h3 h4 h5 h6 pre a table ul ol dl]
  44       accumulated_length = 0
  45       exceeded_node_parent = nil
  46       truncated = false
  47
  48       doc.traverse do |node|
  49         if accumulated_length >= max_length
  50           if node == exceeded_node_parent
  51             exceeded_node_parent = node.parent
  52             node.remove if keep_or_discards.include?(node.name)
  53           else
  54             node.remove
  55           end
  56           next
  57         end
  58
  59         next unless node.children.empty?
  60
  61         if node.text?
  62           accumulated_length += node.text.length
  63         elsif node.name == "img"
  64           accumulated_length += img_length
  65         end
  66
  67         if accumulated_length >= max_length
  68           truncated = true
  69           exceeded_node_parent = node.parent
  70           node.remove
  71         end
  72       end
  73
  74       {
  75         :truncated => truncated,
  76         :html => doc.to_html.html_safe
  77       }
  78     end
  79
  80     protected
  81
  82     def simple_format(text)
  83       SimpleFormat.new.simple_format(text, :dir => "auto")
  84     end
  85
  86     def sanitize(text)
  87       Sanitize.clean(text, Sanitize::Config::OSM).html_safe
  88     end
  89
  90     def linkify(text, mode = :urls)
  91       ERB::Util.html_escape(text)
  92                .then { |html| expand_link_shorthands(html) }
  93                .then { |html| expand_host_shorthands(html) }
  94                .then { |html| auto_link(html, mode) }
  95                .html_safe
  96     end
  97
  98     private
  99
 100     def gsub_pairs_for_linkify_detection
 101       []
 102     end
 103
 104     def expand_link_shorthands(text)
 105       gsub_pairs_for_linkify_detection
 106         .reduce(text) { |text, (pattern, replacement)| text.gsub(pattern, replacement) }
 107     end
 108
 109     def expand_host_shorthands(text)
 110       Array
 111         .wrap(Settings.linkify&.normalisation_rules)
 112         .select { |rule| rule.host_replacement && rule.hosts&.any? }
 113         .reduce(text) do |text, rule|
 114           text.gsub(/(?<=^|#{URL_UNSAFE_CHARS})\b#{Regexp.escape(rule.host_replacement)}/) do
 115             "#{Settings.server_protocol}://#{rule.hosts[0]}"
 116           end
 117         end
 118     end
 119
 120     def auto_link(text, mode)
 121       link_attr = 'rel="nofollow noopener noreferrer" dir="auto"'
 122       Rinku.auto_link(text, mode, link_attr) { |url| format_link_text(url) }
 123     end
 124
 125     def format_link_text(url)
 126       url = Array
 127             .wrap(Settings.linkify&.normalisation_rules)
 128             .reduce(url) do |normalised_url, rule|
 129               shorten_host(normalised_url, rule.hosts, rule.host_replacement) do |path|
 130                 path.sub(Regexp.new(rule.optional_path_prefix || ""), "")
 131               end
 132       end
 133       Array.wrap(Settings.linkify&.display_rules)
 134            .select { |rule| rule.pattern && rule.replacement }
 135            .reduce(url) { |url, rule| url.sub(Regexp.new(rule.pattern), rule.replacement) }
 136     end
 137
 138     def shorten_host(url, hosts, hosts_replacement)
 139       %r{^(https?://([^/]*))(.*)$}.match(url) do |m|
 140         scheme_host, host, path = m.captures
 141         if hosts&.include?(host)
 142           path = yield(path) if block_given?
 143           if hosts_replacement
 144             "#{hosts_replacement}#{path}"
 145           else
 146             "#{scheme_host}#{path}"
 147           end
 148         end || url
 149       end || url
 150     end
 151   end
 152
 153   class HTML < Base
 154     def to_html
 155       linkify(simple_format(self))
 156     end
 157
 158     def to_text
 159       to_s
 160     end
 161   end
 162
 163   class Markdown < Base
 164     def to_html
 165       linkify(sanitize(document.to_html), :all)
 166     end
 167
 168     def to_text
 169       to_s
 170     end
 171
 172     def image
 173       @image_element = first_image_element(document.root) unless defined? @image_element
 174       @image_element.attr["src"] if @image_element
 175     end
 176
 177     def image_alt
 178       @image_element = first_image_element(document.root) unless defined? @image_element
 179       @image_element.attr["alt"] if @image_element
 180     end
 181
 182     def description
 183       return @description if defined? @description
 184
 185       @description = first_truncated_text_content(document.root)
 186     end
 187
 188     private
 189
 190     def document
 191       return @document if @document
 192
 193       @document = Kramdown::Document.new(self)
 194
 195       should_get_dir_auto = lambda do |el|
 196         dir_auto_types = [:p, :header, :codespan, :codeblock, :pre, :ul, :ol, :table, :dl, :math]
 197         return true if dir_auto_types.include?(el.type)
 198         return true if el.type == :a && el.children.length == 1 && el.children[0].type == :text && el.children[0].value == el.attr["href"]
 199
 200         false
 201       end
 202
 203       add_dir = lambda do |element|
 204         element.attr["dir"] ||= "auto" if should_get_dir_auto.call(element)
 205         element.children.each(&add_dir)
 206       end
 207       add_dir.call(@document.root)
 208
 209       @document
 210     end
 211
 212     def first_image_element(element)
 213       return element if image?(element) && element.attr["src"].present?
 214
 215       element.children.find do |child|
 216         nested_image = first_image_element(child)
 217         break nested_image if nested_image
 218       end
 219     end
 220
 221     def first_truncated_text_content(element)
 222       if paragraph?(element)
 223         truncated_text_content(element)
 224       else
 225         element.children.find do |child|
 226           text = first_truncated_text_content(child)
 227           break text unless text.nil?
 228         end
 229       end
 230     end
 231
 232     def truncated_text_content(element)
 233       text = +""
 234
 235       append_text = lambda do |child|
 236         if child.type == :text
 237           text << child.value
 238         else
 239           child.children.each do |c|
 240             append_text.call(c)
 241             break if text.length > DESCRIPTION_MAX_LENGTH
 242           end
 243         end
 244       end
 245       append_text.call(element)
 246
 247       return nil if text.blank?
 248
 249       text_truncated_to_word_break = text.truncate(DESCRIPTION_MAX_LENGTH, :separator => /(?<!\s)\s+/)
 250
 251       if text_truncated_to_word_break.length >= DESCRIPTION_WORD_BREAK_THRESHOLD_LENGTH
 252         text_truncated_to_word_break
 253       else
 254         text.truncate(DESCRIPTION_MAX_LENGTH)
 255       end
 256     end
 257
 258     def image?(element)
 259       element.type == :img || (element.type == :html_element && element.value == "img")
 260     end
 261
 262     def paragraph?(element)
 263       element.type == :p || (element.type == :html_element && element.value == "p")
 264     end
 265   end
 266
 267   class Text < Base
 268     def to_html
 269       linkify(simple_format(ERB::Util.html_escape(self)))
 270     end
 271
 272     def to_text
 273       to_s
 274     end
 275
 276     private
 277
 278     def gsub_pairs_for_linkify_detection
 279       Array
 280         .wrap(Settings.linkify&.detection_rules)
 281         .select { |rule| rule.path_template && rule.patterns.is_a?(Array) }
 282         .flat_map do |rule|
 283           expanded_path = "#{rule.host || "#{Settings.server_protocol}://#{Settings.server_url}"}/#{rule.path_template}"
 284           rule.patterns
 285               .select { |pattern| pattern.is_a?(String) }
 286               .map { |pattern| [Regexp.new("(?<=^|#{URL_UNSAFE_CHARS})#{pattern}", Regexp::IGNORECASE), expanded_path] }
 287         end
 288     end
 289   end
 290 end