Question

I don’t see why the following antlr grammar does not generate the tree for the input “b b b”
but it generates the tree for the input “a a a”

statement
  |
a a a 

Do you have some ideas? Thanks.

grammar Test2;

options {
  language = Java;
}

statement: ( a|b )*;

a: 'a';

b: 'b';

WS: ('\n'|' '|'\t'|'r'|'\f')+ {$channel=HIDDEN;};

Can anyone tries this one in ANTLR IDE eclipse plugin
http://antlrv3ide.sourceforge.net/

In my Eclipse, it does behave strangely as I reported above.

Answer 1

It recognizes both both "a a a" and "b b b", as it should. To check for yourself, add a little debug-print statement to your statement rule:

grammar Test2;

options {
  language = Java;
}

statement: ( a|b )* {System.out.println("parsed: " + $text);};

a: 'a';

b: 'b';

WS: ('\n'|' '|'\t'|'r'|'\f')+ {$channel=HIDDEN;};

and then test the parser with the following class:

import org.antlr.runtime.*;

public class Main {
  public static void main(String[] args) throws Exception {
    Test2Lexer lexer = new Test2Lexer(new ANTLRStringStream(args[0]));
    Test2Parser parser = new Test2Parser(new CommonTokenStream(lexer));
    parser.statement();
  }
}

After generating a lexer & parser and compiling all .java source files:

java -cp antlr-3.3.jar org.antlr.Tool Test2.g
javac -cp antlr-3.3.jar *.java

you can test the parser with "a a a" as input:

java -cp .:antlr-3.3.jar Main "a a a"
parsed: a a a

and with "b b b" as input:

java -cp .:antlr-3.3.jar Main "b b b"
parsed: b b b

As you can see, in both cases no error is reported, and the input is printed back to the console.

My guess is that you’re using ANTLRWorks’ interpreter. Don’t. It is notoriously buggy. Either use ANTLRWorks’ debugger (it’s great!), or use a small test class you write yourself (similar to my Main class).

EDIT

Note that the ANTLR IDE Eclipse plugin uses the interpreter from ANTLRWorks, AFAIK.