How to compose BNF rules using Grammar-Kit

Question

I am trying to convert an EBNF file into working BNF for IntelliJ Grammar-kit.

In the EBNF there are rules as such:

BinOpChar ::= "~" | "!" | "@" | "#" | "$" | "%" | "^" | "&" | "*" | "-"
BinOp ::= BinOpChar, {BinOpChar}

How can I create such rules without resorting to regex? Reason being is that this kind of construct happens very often and it becomes repetitive to do in regex.

To be clear, I would like to be able to create a rule to match @@ from BinOpChar but don't match @ @. Is that possible?

Answer 1

The easiest way is to list each operator independently:

{
  tokens=[
    //...
    op_1='~'
    op_2='!'
    op_3='@'
    op_4='@@'
    op_5='#'
    //...
  ]
}

If you really want to accept all n + n^2 tokens, you will need to use a regular expression:

{
  tokens=[
    //...
    bin_op:'regexp:[~!@#]{1,2}'
    //...
  ]
}

But the idea is, you want to use the lexer to define tokens, and the grammar to define expressions and so forth. So in the grammar if you write:

{
  tokens=[
    space='regexp:\s+'
  ]
}

BinOp ::= BinOpChar [BinOpChar]
BinOpChar ::= "~" | "!" | "@" | "#"

Then it's going to accept @@ and @ @. Does that make sense?