Grammars with complex identifiers

[GordianDziwis]

Currently I am writing a grammar with sparql. Sparql differs from programming languages, that the rules for a single valid identifier token are quite complex.

For example the rules for a variable name are:

    var: $ => seq(
      choice(
        '?',
        '$'
      ),
      $._var_name
    ),

 _pn_chars_base: $ => token(choice(
      /[A-Z]/,
      /[a-z]/,
      // /[\u00C0-\u00D6]/,
      // /[\u00D8-\u00F6]/,
      // /[\u00F8-\u02FF]/,
      // /[\u0370-\u037D]/,
      // /[\u037F-\u1FFF]/,
      // /[\u200C-\u200D]/,
      // /[\u2070-\u218F]/,
      // /[\u2C00-\u2FEF]/,
      // /[\u3001-\uD7FF]/,
      // /[\uF900-\uFDCF]/,
      // /[\uFDF0-\uFFFD]/,
      // /[\u{10000}-\u{EFFFF}]/u
    )),

    // [165]
    _pn_chars_u: $ => choice(
      $._pn_chars_base,
      '_'
    ),

    // [166]
    _var_name: $ => prec.right(seq(
      choice(
        $._pn_chars_u,
        /[0-9]/
      ),
      repeat(choice(
        $._pn_chars_u,
        /[0-9]/,
        /[\u00B7]/,
        /[\u0300-\u036F]/,
        /[\u203F-\u2040]/
      ))
    )),

This is a simple case, this it would be a unmaintainable mess to express the rule for identifiers as a combination of regexes and the token function.
And when I write the rules for identifier with non terminal rules, I can not use the token function and treesitter assumes that whitespaces are allowed between the pseudo tokens inside an identifier.

How can I handle this?

[GordianDziwis]

Ok, defining all those ranges as arrays seems to work:

const PN_CHARS_BASE = [
      /[A-Z]/,
      /[a-z]/,
      // /[\u00C0-\u00D6]/,
      // /[\u00D8-\u00F6]/,
      // /[\u00F8-\u02FF]/,
      // /[\u0370-\u037D]/,
      // /[\u037F-\u1FFF]/,
      // /[\u200C-\u200D]/,
      // /[\u2070-\u218F]/,
      // /[\u2C00-\u2FEF]/,
      // /[\u3001-\uD7FF]/,
      // /[\uF900-\uFDCF]/,
      // /[\uFDF0-\uFFFD]/,
      // /[\u{10000}-\u{EFFFF}]/u
]

    _pn_chars_u: $ => choice(
      ...PN_CHARS_BASE,
      '_'
    ),