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       Array
 102         .wrap(Settings.linkify&.detection_rules)
 103         .select { |rule| rule.path_template && rule.patterns.is_a?(Array) }
 104         .flat_map do |rule|
 105           expanded_path = "#{rule.host || "#{Settings.server_protocol}://#{Settings.server_url}"}/#{rule.path_template}"
 106           rule.patterns
 107               .select { |pattern| pattern.is_a?(String) }
 108               .map { |pattern| [Regexp.new("(?<=^|#{URL_UNSAFE_CHARS})#{pattern}", Regexp::IGNORECASE), expanded_path] }
 109         end
 110     end
 111
 112     def expand_link_shorthands(text)
 113       gsub_pairs_for_linkify_detection
 114         .reduce(text) { |text, (pattern, replacement)| text.gsub(pattern, replacement) }
 115     end
 116
 117     def expand_host_shorthands(text)
 118       [
 119         [Settings.linkify_hosts, Settings.linkify_hosts_replacement],
 120         [Settings.linkify_wiki_hosts, Settings.linkify_wiki_hosts_replacement]
 121       ]
 122         .select { |hosts, replacement| replacement && hosts&.any? }
 123         .reduce(text) do |text, (hosts, replacement)|
 124           text.gsub(/(?<=^|#{URL_UNSAFE_CHARS})\b#{Regexp.escape(replacement)}/) do
 125             "#{Settings.server_protocol}://#{hosts[0]}"
 126           end
 127         end
 128     end
 129
 130     def auto_link(text, mode)
 131       link_attr = 'rel="nofollow noopener noreferrer" dir="auto"'
 132       Rinku.auto_link(text, mode, link_attr) { |url| format_link_text(url) }
 133     end
 134
 135     def format_link_text(url)
 136       url = shorten_host(url, Settings.linkify_hosts, Settings.linkify_hosts_replacement)
 137       shorten_host(url, Settings.linkify_wiki_hosts, Settings.linkify_wiki_hosts_replacement) do |path|
 138         path.sub(Regexp.new(Settings.linkify_wiki_optional_path_prefix || ""), "")
 139       end
 140     end
 141
 142     def shorten_host(url, hosts, hosts_replacement)
 143       %r{^(https?://([^/]*))(.*)$}.match(url) do |m|
 144         scheme_host, host, path = m.captures
 145         if hosts&.include?(host)
 146           path = yield(path) if block_given?
 147           if hosts_replacement
 148             "#{hosts_replacement}#{path}"
 149           else
 150             "#{scheme_host}#{path}"
 151           end
 152         end || url
 153       end || url
 154     end
 155   end
 156
 157   class HTML < Base
 158     def to_html
 159       linkify(simple_format(self))
 160     end
 161
 162     def to_text
 163       to_s
 164     end
 165   end
 166
 167   class Markdown < Base
 168     def to_html
 169       linkify(sanitize(document.to_html), :all)
 170     end
 171
 172     def to_text
 173       to_s
 174     end
 175
 176     def image
 177       @image_element = first_image_element(document.root) unless defined? @image_element
 178       @image_element.attr["src"] if @image_element
 179     end
 180
 181     def image_alt
 182       @image_element = first_image_element(document.root) unless defined? @image_element
 183       @image_element.attr["alt"] if @image_element
 184     end
 185
 186     def description
 187       return @description if defined? @description
 188
 189       @description = first_truncated_text_content(document.root)
 190     end
 191
 192     private
 193
 194     def document
 195       return @document if @document
 196
 197       @document = Kramdown::Document.new(self)
 198
 199       should_get_dir_auto = lambda do |el|
 200         dir_auto_types = [:p, :header, :codespan, :codeblock, :pre, :ul, :ol, :table, :dl, :math]
 201         return true if dir_auto_types.include?(el.type)
 202         return true if el.type == :a && el.children.length == 1 && el.children[0].type == :text && el.children[0].value == el.attr["href"]
 203
 204         false
 205       end
 206
 207       add_dir = lambda do |element|
 208         element.attr["dir"] ||= "auto" if should_get_dir_auto.call(element)
 209         element.children.each(&add_dir)
 210       end
 211       add_dir.call(@document.root)
 212
 213       @document
 214     end
 215
 216     def first_image_element(element)
 217       return element if image?(element) && element.attr["src"].present?
 218
 219       element.children.find do |child|
 220         nested_image = first_image_element(child)
 221         break nested_image if nested_image
 222       end
 223     end
 224
 225     def first_truncated_text_content(element)
 226       if paragraph?(element)
 227         truncated_text_content(element)
 228       else
 229         element.children.find do |child|
 230           text = first_truncated_text_content(child)
 231           break text unless text.nil?
 232         end
 233       end
 234     end
 235
 236     def truncated_text_content(element)
 237       text = +""
 238
 239       append_text = lambda do |child|
 240         if child.type == :text
 241           text << child.value
 242         else
 243           child.children.each do |c|
 244             append_text.call(c)
 245             break if text.length > DESCRIPTION_MAX_LENGTH
 246           end
 247         end
 248       end
 249       append_text.call(element)
 250
 251       return nil if text.blank?
 252
 253       text_truncated_to_word_break = text.truncate(DESCRIPTION_MAX_LENGTH, :separator => /(?<!\s)\s+/)
 254
 255       if text_truncated_to_word_break.length >= DESCRIPTION_WORD_BREAK_THRESHOLD_LENGTH
 256         text_truncated_to_word_break
 257       else
 258         text.truncate(DESCRIPTION_MAX_LENGTH)
 259       end
 260     end
 261
 262     def image?(element)
 263       element.type == :img || (element.type == :html_element && element.value == "img")
 264     end
 265
 266     def paragraph?(element)
 267       element.type == :p || (element.type == :html_element && element.value == "p")
 268     end
 269   end
 270
 271   class Text < Base
 272     def to_html
 273       linkify(simple_format(ERB::Util.html_escape(self)))
 274     end
 275
 276     def to_text
 277       to_s
 278     end
 279   end
 280 end