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