I have island grammar, that becomes known on parsing stage. How can I parse it with separate lexer/parser, adding result to common tree ? In my case, such grammar is inside BLOCK token in rule output, output: OUTPUT ID BLOCK;