Antlr + StringTemplate

• Antlrworks is not good tool. It does not support semantic predicates.
• The documentation for using Antlr together with StringTemplate is not good enough.
• There are many ways to improve error reporting. A quick fix would be to override emitErrorMessage(String message) in the parser class and simply throw an exception with the provided message:

	grammar T;
	@members {
	@Override
	public void emitErrorMessage(String message) {
	throw new RuntimeException(message);
	}
	}
	definition
	: '(' 'define' ( '(' variable def_formals ')' body ')'
	| variable expression ')'
	)
	;
	def_formals
	: variable* ('.' variable)?
	;
	body
	: ((definition)=> definition)* expression+
	;
	expression
	: INT
	;
	variable
	: ID
	;
	ID : 'a'..'z'+;
	INT : '0'..'9';
	SPACE : ' ' {skip();};

view raw gistfile1.g4 hosted with ❤ by GitHub

which you can test with the class:

	import org.antlr.runtime.*;
	public class Main {
	public static void main(String[] args) {
	String[] tests = {
	"(define x 5)",
	"(define x 5))",
	"(define x)",
	"(define)"
	};
	for(String input : tests) {
	TLexer lexer = new TLexer(new ANTLRStringStream(input));
	TParser parser = new TParser(new CommonTokenStream(lexer));
	System.out.println("\nParsing : " + input);
	try {
	parser.definition();
	} catch(Exception e) {
	System.out.println(" exception -> " + e.getMessage());
	}
	}
	}
	}

view raw gistfile1.java hosted with ❤ by GitHub

After running the class above, you will see the following:

bart@hades:~/Programming/ANTLR/Demos/T$ java -cp antlr-3.3.jar org.antlr.Tool T.g

bart@hades:~/Programming/ANTLR/Demos/T$ javac -cp antlr-3.3.jar *.java

bart@hades:~/Programming/ANTLR/Demos/T$ java -cp .:antlr-3.3.jar Main

Parsing : (define x 5)
Parsing : (define x 5))
Parsing : (define x)
exception -> line 1:9 missing INT at ')'
Parsing : (define)
exception -> line 1:7 no viable alternative at input ')'

As you can see, the input (define x 5)) produces no exception! That is because the lexer has no problems with it (they're all valid tokens) and the parser is simply instructed to consume the definition rule:

definition

: '(' 'define' ( '(' variable def_formals ')' body ')'

| variable expression ')'

) ;
which it does. If you wanted an error because of the dangling ')', then you'd have the add the EOF token at the end of the rule:

definition

: '(' 'define' ( '(' variable def_formals ')' body ')'

| variable expression ')'

)

EOF;