Parse any character until

[vincent-lg]

Parse any character until

[josevalim]

You have to build the whole list. You can see how we deal with this in Elixir's syntax highlighter: https://github.com/elixir-makeup/makeup_elixir/blob/master/lib/makeup/lexers/elixir_lexer/atoms.ex.exs :)

[josevalim]

Wouldn't it be utf8_string(not: ?")?

[vincent-lg]

Unfortunately, it doesn't seem to do quite what I want.
Identifiers should be utf-8 letters. I was at a loss about doing it without a regex or at least a true function I could call and which would make sure the string would not be parsed more than necessary. aérien + 5 should accept everything before the space (which is not a valid letter).
I thought I had found the solution with repeat_while:

  vexpr_var =
    repeat_while(
      utf8_char([]),
      {:check_varname, []}
    )
    |> unwrap_and_tag(:var)
    |> label("variable")

  defp check_varname(name, context, _, _) do
    case String.match?(name, ~r/^[[:alpha:]]+$/u) do
      true -> {:cont, context}
      false -> {:halt, context}
    end
  end

But the check_varname function seems to be called only once, after all the characters in the strings have been read. Not exactly what I'd like.
I also tried to play with post_traverse, but I guess this won't do since I need to validate the string.
Perhaps there's no solution. Probably something I just miss.