app/lib/formatter.rb

   1 # frozen_string_literal: true
   2
   3 require 'singleton'
   4 require_relative './sanitize_config'
   5
   6 class Formatter
   7   include Singleton
   8   include RoutingHelper
   9
  10   include ActionView::Helpers::TextHelper
  11
  12   def format(status, **options)
  13     if status.reblog?
  14       prepend_reblog = status.reblog.account.acct
  15       status         = status.proper
  16     else
  17       prepend_reblog = false
  18     end
  19
  20     raw_content = status.text
  21
  22     return '' if raw_content.blank?
  23
  24     unless status.local?
  25       html = reformat(raw_content)
  26       html = encode_custom_emojis(html, status.emojis, options[:autoplay]) if options[:custom_emojify]
  27       return html.html_safe # rubocop:disable Rails/OutputSafety
  28     end
  29
  30     linkable_accounts = status.active_mentions.map(&:account)
  31     linkable_accounts << status.account
  32
  33     html = raw_content
  34     html = "RT @#{prepend_reblog} #{html}" if prepend_reblog
  35     html = encode_and_link_urls(html, linkable_accounts)
  36     html = encode_custom_emojis(html, status.emojis, options[:autoplay]) if options[:custom_emojify]
  37     html = simple_format(html, {}, sanitize: false)
  38     html = html.delete("\n")
  39
  40     html.html_safe # rubocop:disable Rails/OutputSafety
  41   end
  42
  43   def reformat(html)
  44     sanitize(html, Sanitize::Config::MASTODON_STRICT)
  45   end
  46
  47   def plaintext(status)
  48     return status.text if status.local?
  49
  50     text = status.text.gsub(/(<br \/>|<br>|<\/p>)+/) { |match| "#{match}\n" }
  51     strip_tags(text)
  52   end
  53
  54   def simplified_format(account, **options)
  55     html = account.local? ? linkify(account.note) : reformat(account.note)
  56     html = encode_custom_emojis(html, account.emojis, options[:autoplay]) if options[:custom_emojify]
  57     html.html_safe # rubocop:disable Rails/OutputSafety
  58   end
  59
  60   def sanitize(html, config)
  61     Sanitize.fragment(html, config)
  62   end
  63
  64   def format_spoiler(status, **options)
  65     html = encode(status.spoiler_text)
  66     html = encode_custom_emojis(html, status.emojis, options[:autoplay])
  67     html.html_safe # rubocop:disable Rails/OutputSafety
  68   end
  69
  70   def format_display_name(account, **options)
  71     html = encode(account.display_name.presence || account.username)
  72     html = encode_custom_emojis(html, account.emojis, options[:autoplay]) if options[:custom_emojify]
  73     html.html_safe # rubocop:disable Rails/OutputSafety
  74   end
  75
  76   def format_field(account, str, **options)
  77     return reformat(str).html_safe unless account.local? # rubocop:disable Rails/OutputSafety
  78     html = encode_and_link_urls(str, me: true)
  79     html = encode_custom_emojis(html, account.emojis, options[:autoplay]) if options[:custom_emojify]
  80     html.html_safe # rubocop:disable Rails/OutputSafety
  81   end
  82
  83   def linkify(text)
  84     html = encode_and_link_urls(text)
  85     html = simple_format(html, {}, sanitize: false)
  86     html = html.delete("\n")
  87
  88     html.html_safe # rubocop:disable Rails/OutputSafety
  89   end
  90
  91   private
  92
  93   def html_entities
  94     @html_entities ||= HTMLEntities.new
  95   end
  96
  97   def encode(html)
  98     html_entities.encode(html)
  99   end
 100
 101   def encode_and_link_urls(html, accounts = nil, options = {})
 102     entities = utf8_friendly_extractor(html, extract_url_without_protocol: false)
 103
 104     if accounts.is_a?(Hash)
 105       options  = accounts
 106       accounts = nil
 107     end
 108
 109     rewrite(html.dup, entities) do |entity|
 110       if entity[:url]
 111         link_to_url(entity, options)
 112       elsif entity[:hashtag]
 113         link_to_hashtag(entity)
 114       elsif entity[:screen_name]
 115         link_to_mention(entity, accounts)
 116       end
 117     end
 118   end
 119
 120   def count_tag_nesting(tag)
 121     if tag[1] == '/' then -1
 122     elsif tag[-2] == '/' then 0
 123     else 1
 124     end
 125   end
 126
 127   def encode_custom_emojis(html, emojis, animate = false)
 128     return html if emojis.empty?
 129
 130     emoji_map = if animate
 131                   emojis.each_with_object({}) { |e, h| h[e.shortcode] = full_asset_url(e.image.url) }
 132                 else
 133                   emojis.each_with_object({}) { |e, h| h[e.shortcode] = full_asset_url(e.image.url(:static)) }
 134                 end
 135
 136     i                     = -1
 137     tag_open_index        = nil
 138     inside_shortname      = false
 139     shortname_start_index = -1
 140     invisible_depth       = 0
 141
 142     while i + 1 < html.size
 143       i += 1
 144
 145       if invisible_depth.zero? && inside_shortname && html[i] == ':'
 146         shortcode = html[shortname_start_index + 1..i - 1]
 147         emoji     = emoji_map[shortcode]
 148
 149         if emoji
 150           replacement = "<img draggable=\"false\" class=\"emojione\" alt=\":#{encode(shortcode)}:\" title=\":#{encode(shortcode)}:\" src=\"#{encode(emoji)}\" />"
 151           before_html = shortname_start_index.positive? ? html[0..shortname_start_index - 1] : ''
 152           html        = before_html + replacement + html[i + 1..-1]
 153           i          += replacement.size - (shortcode.size + 2) - 1
 154         else
 155           i -= 1
 156         end
 157
 158         inside_shortname = false
 159       elsif tag_open_index && html[i] == '>'
 160         tag = html[tag_open_index..i]
 161         tag_open_index = nil
 162         if invisible_depth.positive?
 163           invisible_depth += count_tag_nesting(tag)
 164         elsif tag == '<span class="invisible">'
 165           invisible_depth = 1
 166         end
 167       elsif html[i] == '<'
 168         tag_open_index   = i
 169         inside_shortname = false
 170       elsif !tag_open_index && html[i] == ':'
 171         inside_shortname      = true
 172         shortname_start_index = i
 173       end
 174     end
 175
 176     html
 177   end
 178
 179   def rewrite(text, entities)
 180     chars = text.to_s.to_char_a
 181
 182     # Sort by start index
 183     entities = entities.sort_by do |entity|
 184       indices = entity.respond_to?(:indices) ? entity.indices : entity[:indices]
 185       indices.first
 186     end
 187
 188     result = []
 189
 190     last_index = entities.reduce(0) do |index, entity|
 191       indices = entity.respond_to?(:indices) ? entity.indices : entity[:indices]
 192       result << encode(chars[index...indices.first].join)
 193       result << yield(entity)
 194       indices.last
 195     end
 196
 197     result << encode(chars[last_index..-1].join)
 198
 199     result.flatten.join
 200   end
 201
 202   UNICODE_ESCAPE_BLACKLIST_RE = /\p{Z}|\p{P}/
 203
 204   def utf8_friendly_extractor(text, options = {})
 205     old_to_new_index = [0]
 206
 207     escaped = text.chars.map do |c|
 208       output = begin
 209         if c.ord.to_s(16).length > 2 && UNICODE_ESCAPE_BLACKLIST_RE.match(c).nil?
 210           CGI.escape(c)
 211         else
 212           c
 213         end
 214       end
 215
 216       old_to_new_index << old_to_new_index.last + output.length
 217
 218       output
 219     end.join
 220
 221     # Note: I couldn't obtain list_slug with @user/list-name format
 222     # for mention so this requires additional check
 223     special = Extractor.extract_urls_with_indices(escaped, options).map do |extract|
 224       # exactly one of :url, :hashtag, :screen_name, :cashtag keys is present
 225       key = (extract.keys & [:url, :hashtag, :screen_name, :cashtag]).first
 226
 227       new_indices = [
 228         old_to_new_index.find_index(extract[:indices].first),
 229         old_to_new_index.find_index(extract[:indices].last),
 230       ]
 231
 232       has_prefix_char = [:hashtag, :screen_name, :cashtag].include?(key)
 233       value_indices = [
 234         new_indices.first + (has_prefix_char ? 1 : 0), # account for #, @ or $
 235         new_indices.last - 1,
 236       ]
 237
 238       next extract.merge(
 239         :indices => new_indices,
 240         key => text[value_indices.first..value_indices.last]
 241       )
 242     end
 243
 244     standard = Extractor.extract_entities_with_indices(text, options)
 245
 246     Extractor.remove_overlapping_entities(special + standard)
 247   end
 248
 249   def link_to_url(entity, options = {})
 250     url        = Addressable::URI.parse(entity[:url])
 251     html_attrs = { target: '_blank', rel: 'nofollow noopener' }
 252
 253     html_attrs[:rel] = "me #{html_attrs[:rel]}" if options[:me]
 254
 255     Twitter::Autolink.send(:link_to_text, entity, link_html(entity[:url]), url, html_attrs)
 256   rescue Addressable::URI::InvalidURIError, IDN::Idna::IdnaError
 257     encode(entity[:url])
 258   end
 259
 260   def link_to_mention(entity, linkable_accounts)
 261     acct = entity[:screen_name]
 262
 263     return link_to_account(acct) unless linkable_accounts
 264
 265     account = linkable_accounts.find { |item| TagManager.instance.same_acct?(item.acct, acct) }
 266     account ? mention_html(account) : "@#{encode(acct)}"
 267   end
 268
 269   def link_to_account(acct)
 270     username, domain = acct.split('@')
 271
 272     domain  = nil if TagManager.instance.local_domain?(domain)
 273     account = EntityCache.instance.mention(username, domain)
 274
 275     account ? mention_html(account) : "@#{encode(acct)}"
 276   end
 277
 278   def link_to_hashtag(entity)
 279     hashtag_html(entity[:hashtag])
 280   end
 281
 282   def link_html(url)
 283     url    = Addressable::URI.parse(url).to_s
 284     prefix = url.match(/\Ahttps?:\/\/(www\.)?/).to_s
 285     text   = url[prefix.length, 30]
 286     suffix = url[prefix.length + 30..-1]
 287     cutoff = url[prefix.length..-1].length > 30
 288
 289     "<span class=\"invisible\">#{encode(prefix)}</span><span class=\"#{cutoff ? 'ellipsis' : ''}\">#{encode(text)}</span><span class=\"invisible\">#{encode(suffix)}</span>"
 290   end
 291
 292   def hashtag_html(tag)
 293     "<a href=\"#{encode(tag_url(tag.downcase))}\" class=\"mention hashtag\" rel=\"tag\">#<span>#{encode(tag)}</span></a>"
 294   end
 295
 296   def mention_html(account)
 297     "<span class=\"h-card\"><a href=\"#{encode(TagManager.instance.url_for(account))}\" class=\"u-url mention\">@<span>#{encode(account.username)}</span></a></span>"
 298   end
 299 end