Support multi-letter sigils

lib/makeup/lexers/elixir_lexer.ex

@@ -257,12 +257,15 @@ defmodule Makeup.Lexers.ElixirLexer do
 
   sigils_interpol =
     for {ldelim, rdelim} <- sigil_delimiters do
-      sigil(ldelim, rdelim, [?a..?z], combinators_inside_string)
+      sigil(ldelim, rdelim, utf8_string([?a..?z], 1), combinators_inside_string)
     end
 
   sigils_no_interpol =
     for {ldelim, rdelim} <- sigil_delimiters do
-      sigil(ldelim, rdelim, [?A..?Z], [escape_delim(rdelim), iex_prompt_inside_string])
+      sigil(ldelim, rdelim, ascii_string([?A..?Z], min: 1), [
+        escape_delim(rdelim),
+        iex_prompt_inside_string
+      ])
     end
 
   all_sigils = sigils_interpol ++ sigils_no_interpol

lib/makeup/lexers/elixir_lexer/helper.ex

@@ -15,8 +15,8 @@ defmodule Makeup.Lexers.ElixirLexer.Helper do
     string("\\" <> rdelim_first_char)
   end
 
-  def sigil(ldelim, rdelim, ranges, middle) do
-    left = string("~") |> utf8_char(ranges) |> string(ldelim)
+  def sigil(ldelim, rdelim, name, middle) do
+    left = string("~") |> concat(name) |> string(ldelim)
     right = string(rdelim)
 
     choices = middle ++ [utf8_char([])]
@@ -31,9 +31,9 @@ defmodule Makeup.Lexers.ElixirLexer.Helper do
   def build_sigil(rest, acc, context, line, offset) do
     type =
       case Enum.at(acc, -2) do
-        sigil when sigil in ~c"sScC" -> :string
-        sigil when sigil in ~c"rR" -> :string_regex
-        sigil when sigil in ~c"TDNU" -> :literal_date
+        sigil when sigil in ~w"s S c C" -> :string
+        sigil when sigil in ~w"r" -> :string_regex
+        sigil when sigil in ~w"T D N U" -> :literal_date
         _ -> :string_sigil
       end
 

test/makeup/lexers/elixir_lexer/elixir_lexer_tokenizer_test.exs

@@ -510,6 +510,10 @@ defmodule ElixirLexerTokenizerTestSnippet do
     assert lex("~U[2020-01-01 01:23:45Z]") == [{:literal_date, %{}, "~U[2020-01-01 01:23:45Z]"}]
   end
 
+  test "custom sigils" do
+    assert lex("~VEC[1 2.0 3]f64") == [{:string_sigil, %{}, "~VEC[1 2.0 3]f64"}]
+  end
+
   describe "strings and sigils" do
     test "unicode codepoints" do
       assert lex(~S["\u0000"]) == [