git.squeep.com Git - akkoma/blob - lib/pleroma/formatter.ex

   1 defmodule Pleroma.Formatter do
   2   alias Pleroma.User
   3   alias Pleroma.Web.MediaProxy
   4
   5   @tag_regex ~r/\#\w+/u
   6   def parse_tags(text, data \\ %{}) do
   7     Regex.scan(@tag_regex, text)
   8     |> Enum.map(fn ["#" <> tag = full_tag] -> {full_tag, String.downcase(tag)} end)
   9     |> (fn map ->
  10           if data["sensitive"] in [true, "True", "true", "1"],
  11             do: [{"#nsfw", "nsfw"}] ++ map,
  12             else: map
  13         end).()
  14   end
  15
  16   def parse_mentions(text) do
  17     # Modified from https://www.w3.org/TR/html5/forms.html#valid-e-mail-address
  18     regex =
  19       ~r/@[a-zA-Z0-9.!#$%&'*+\/=?^_`{|}~-]*@?[a-zA-Z0-9](?:[a-zA-Z0-9-]{0,61}[a-zA-Z0-9])?(?:\.[a-zA-Z0-9](?:[a-zA-Z0-9-]{0,61}[a-zA-Z0-9])?)*/u
  20
  21     Regex.scan(regex, text)
  22     |> List.flatten()
  23     |> Enum.uniq()
  24     |> Enum.map(fn "@" <> match = full_match ->
  25       {full_match, User.get_cached_by_nickname(match)}
  26     end)
  27     |> Enum.filter(fn {_match, user} -> user end)
  28   end
  29
  30   @finmoji [
  31     "a_trusted_friend",
  32     "alandislands",
  33     "association",
  34     "auroraborealis",
  35     "baby_in_a_box",
  36     "bear",
  37     "black_gold",
  38     "christmasparty",
  39     "crosscountryskiing",
  40     "cupofcoffee",
  41     "education",
  42     "fashionista_finns",
  43     "finnishlove",
  44     "flag",
  45     "forest",
  46     "four_seasons_of_bbq",
  47     "girlpower",
  48     "handshake",
  49     "happiness",
  50     "headbanger",
  51     "icebreaker",
  52     "iceman",
  53     "joulutorttu",
  54     "kaamos",
  55     "kalsarikannit_f",
  56     "kalsarikannit_m",
  57     "karjalanpiirakka",
  58     "kicksled",
  59     "kokko",
  60     "lavatanssit",
  61     "losthopes_f",
  62     "losthopes_m",
  63     "mattinykanen",
  64     "meanwhileinfinland",
  65     "moominmamma",
  66     "nordicfamily",
  67     "out_of_office",
  68     "peacemaker",
  69     "perkele",
  70     "pesapallo",
  71     "polarbear",
  72     "pusa_hispida_saimensis",
  73     "reindeer",
  74     "sami",
  75     "sauna_f",
  76     "sauna_m",
  77     "sauna_whisk",
  78     "sisu",
  79     "stuck",
  80     "suomimainittu",
  81     "superfood",
  82     "swan",
  83     "the_cap",
  84     "the_conductor",
  85     "the_king",
  86     "the_voice",
  87     "theoriginalsanta",
  88     "tomoffinland",
  89     "torillatavataan",
  90     "unbreakable",
  91     "waiting",
  92     "white_nights",
  93     "woollysocks"
  94   ]
  95
  96   @finmoji_with_filenames Enum.map(@finmoji, fn finmoji ->
  97                             {finmoji, "/finmoji/128px/#{finmoji}-128.png"}
  98                           end)
  99
 100   @emoji_from_file (with {:ok, default} <- File.read("config/emoji.txt") do
 101                       custom =
 102                         with {:ok, custom} <- File.read("config/custom_emoji.txt") do
 103                           custom
 104                         else
 105                           _e -> ""
 106                         end
 107
 108                       (default <> "\n" <> custom)
 109                       |> String.trim()
 110                       |> String.split(~r/\n+/)
 111                       |> Enum.map(fn line ->
 112                         [name, file] = String.split(line, ~r/,\s*/)
 113                         {name, file}
 114                       end)
 115                     else
 116                       _ -> []
 117                     end)
 118
 119   @emoji_from_globs (
 120                       static_path = Path.join(:code.priv_dir(:pleroma), "static")
 121
 122                       globs =
 123                         Application.get_env(:pleroma, :emoji, [])
 124                         |> Keyword.get(:shortcode_globs, [])
 125
 126                       paths =
 127                         Enum.map(globs, fn glob ->
 128                           Path.join(static_path, glob)
 129                           |> Path.wildcard()
 130                         end)
 131                         |> Enum.concat()
 132
 133                       Enum.map(paths, fn path ->
 134                         shortcode = Path.basename(path, Path.extname(path))
 135                         external_path = Path.join("/", Path.relative_to(path, static_path))
 136                         {shortcode, external_path}
 137                       end)
 138                     )
 139
 140   @emoji @finmoji_with_filenames ++ @emoji_from_globs ++ @emoji_from_file
 141
 142   def emojify(text, emoji \\ @emoji)
 143   def emojify(text, nil), do: text
 144
 145   def emojify(text, emoji) do
 146     Enum.reduce(emoji, text, fn {emoji, file}, text ->
 147       emoji = HtmlSanitizeEx.strip_tags(emoji)
 148       file = HtmlSanitizeEx.strip_tags(file)
 149
 150       String.replace(
 151         text,
 152         ":#{emoji}:",
 153         "<img height='32px' width='32px' alt='#{emoji}' title='#{emoji}' src='#{
 154           MediaProxy.url(file)
 155         }' />"
 156       )
 157       |> HtmlSanitizeEx.basic_html()
 158     end)
 159   end
 160
 161   def get_emoji(text) when is_binary(text) do
 162     Enum.filter(@emoji, fn {emoji, _} -> String.contains?(text, ":#{emoji}:") end)
 163   end
 164
 165   def get_emoji(_), do: []
 166
 167   def get_custom_emoji() do
 168     @emoji
 169   end
 170
 171   @link_regex ~r/[0-9a-z+\-\.]+:[0-9a-z$-_.+!*'(),]+/ui
 172
 173   # IANA got a list https://www.iana.org/assignments/uri-schemes/ but
 174   # Stuff like ipfs isn’t in it
 175   # There is very niche stuff
 176   @uri_schemes [
 177     "https://",
 178     "http://",
 179     "dat://",
 180     "dweb://",
 181     "gopher://",
 182     "ipfs://",
 183     "ipns://",
 184     "irc:",
 185     "ircs:",
 186     "magnet:",
 187     "mailto:",
 188     "mumble:",
 189     "ssb://",
 190     "xmpp:"
 191   ]
 192
 193   # TODO: make it use something other than @link_regex
 194   def html_escape(text) do
 195     Regex.split(@link_regex, text, include_captures: true)
 196     |> Enum.map_every(2, fn chunk ->
 197       {:safe, part} = Phoenix.HTML.html_escape(chunk)
 198       part
 199     end)
 200     |> Enum.join("")
 201   end
 202
 203   @doc "changes scheme:... urls to html links"
 204   def add_links({subs, text}) do
 205     additionnal_schemes =
 206       Application.get_env(:pleroma, :uri_schemes, [])
 207       |> Keyword.get(:additionnal_schemes, [])
 208
 209     links =
 210       text
 211       |> String.split([" ", "\t", "<br>"])
 212       |> Enum.filter(fn word -> String.starts_with?(word, @uri_schemes ++ additionnal_schemes) end)
 213       |> Enum.filter(fn word -> Regex.match?(@link_regex, word) end)
 214       |> Enum.map(fn url -> {Ecto.UUID.generate(), url} end)
 215       |> Enum.sort_by(fn {_, url} -> -String.length(url) end)
 216
 217     uuid_text =
 218       links
 219       |> Enum.reduce(text, fn {uuid, url}, acc -> String.replace(acc, url, uuid) end)
 220
 221     subs =
 222       subs ++
 223         Enum.map(links, fn {uuid, url} ->
 224           {:safe, link} = Phoenix.HTML.Link.link(url, to: url)
 225
 226           link =
 227             link
 228             |> IO.iodata_to_binary()
 229
 230           {uuid, link}
 231         end)
 232
 233     {subs, uuid_text}
 234   end
 235
 236   @doc "Adds the links to mentioned users"
 237   def add_user_links({subs, text}, mentions) do
 238     mentions =
 239       mentions
 240       |> Enum.sort_by(fn {name, _} -> -String.length(name) end)
 241       |> Enum.map(fn {name, user} -> {name, user, Ecto.UUID.generate()} end)
 242
 243     uuid_text =
 244       mentions
 245       |> Enum.reduce(text, fn {match, _user, uuid}, text ->
 246         String.replace(text, match, uuid)
 247       end)
 248
 249     subs =
 250       subs ++
 251         Enum.map(mentions, fn {match, %User{ap_id: ap_id, info: info}, uuid} ->
 252           ap_id = info["source_data"]["url"] || ap_id
 253
 254           short_match = String.split(match, "@") |> tl() |> hd()
 255
 256           {uuid,
 257            "<span><a class='mention' href='#{ap_id}'>@<span>#{short_match}</span></a></span>"}
 258         end)
 259
 260     {subs, uuid_text}
 261   end
 262
 263   @doc "Adds the hashtag links"
 264   def add_hashtag_links({subs, text}, tags) do
 265     tags =
 266       tags
 267       |> Enum.sort_by(fn {name, _} -> -String.length(name) end)
 268       |> Enum.map(fn {name, short} -> {name, short, Ecto.UUID.generate()} end)
 269
 270     uuid_text =
 271       tags
 272       |> Enum.reduce(text, fn {match, _short, uuid}, text ->
 273         String.replace(text, match, uuid)
 274       end)
 275
 276     subs =
 277       subs ++
 278         Enum.map(tags, fn {tag_text, tag, uuid} ->
 279           url = "<a href='#{Pleroma.Web.base_url()}/tag/#{tag}' rel='tag'>#{tag_text}</a>"
 280           {uuid, url}
 281         end)
 282
 283     {subs, uuid_text}
 284   end
 285
 286   def finalize({subs, text}) do
 287     Enum.reduce(subs, text, fn {uuid, replacement}, result_text ->
 288       String.replace(result_text, uuid, replacement)
 289     end)
 290   end
 291 end