LL(1) grammars supporting an efficient error handling