parser.js

var regexp = LSD.RegExp({
  'fn_tail': '\\.',
  'fn_arguments': 'inside_round',
  'fn_name': 'unicode',
  'fn': '(fn_tail)?(fn_name)\\((fn_arguments)\\)',
  
  'block_arguments': 'inside_curly',
  'block_body': 'inside_curly',
  'block': '\\{(block_arguments)?(block_body)\\}',
  
  'double_string': '"((?:[^"]|\\")*)"/',
  'single_string': "'((?:[^']|\\')*)'",
  'string': '(double_string)|(single_string)'
  
  'whitespace': '\\s+',
  'comma': ',',
  'semicolon': ';',
  'separator': '(whitespace|comma|semicolon)'
  
  'index': '\\[\\s*(inside_square)\\s*\\]',
  
  'token_tail': '\\.',
  'token_name': 'unicode',
  'token': '(token_tail)?(token_name)',
  
  'length_number': '[-+]?(?:\\d+\\.\\d*|\\d*\\.\\d+)'
  'length_unit': 'em|px|pt|%|fr|deg|(?=$|[^a-zA-Z0-9.])'
  'length': '(length_number)(length_unit)',
  
  'operator': '([-+]|[\\/%^~=><*\\^!|&$]+)'
}, {
  fn: function(tail, name, args) {

  },

  block: function(args, body) {

  },

  index: function(content) {

  },

  token: function(name) {

  },

  length: function(number, unit) {

  },

  operator: function(name) {

  },
  
  separator: function(token) {
    switch (token) {
      case ';':
        break;
      case ',':
        break;
    }
  }
})

var results = regexp.exec(string);

Very nice.
I don't like your syntax for nested selectors though.
Check out http://www.geocities.jp/kosako3/oniguruma/doc/RE.txt #9 subexp call.

My idea is that any string can be replaced with another sub-expression, but you might have a point. Adding square brackets is not big of a deal though.

http://xregexp.com/plugins/#unicode is an example of something similar.