[antlr-interest] Quick intro to Python backend

[Ben]

Thanks Martijn,

I think I tried to do what you did by overriding lexer::members, and members (and rulecatch) but I think I may be doing it wrong.  Once difference though is that I'm trying to encapsulate the ANTLR code as much as possible---it's not very convenient to write Python inside AntlrWorks, and I've read that the Python runtime is subject to change so I want to use the recommended methods.  So I think this may be more of a Python-specific issue.

-- 
Ben Escoto

----------------- Original message -----------------
From: Martijn Reuvers <martijn.reuvers at gmail.com>
To: Ben <misc7 at emerose.org>
Date: Mon, 7 Sep 2009 20:46:22 +0200
Hi Ben,

I can't help you with the Python part, but below is a quick & dirty
grammar for Java which more or less does do what you want.

Most likely your errors will come from the lexer, hence I throw the
exception there to show you. So in your python program catching that
should tell you it was wrong. The rest of the grammar fills a list
with text or numbers that are parsed, and prints it in the @after of
begin rule (which is only there so a quick printout was possible).

Martijn


grammar test001;

@lexer::members {

  @Override
	public void reportError(RecognitionException e) {
		throw new RuntimeException(e);
	}

}

@members {

	java.util.List result = new java.util.ArrayList();

	public List getResult() {
		return result;
	}
}

begin
@after {
	System.out.println("RESULT=" + result);
}
	:	expr
	;


expr
	:	what (DOT expr)?
	;
	
what
	:	TEXT
		{
			if($TEXT.text != null) {
				result.add($TEXT.text);
			}		
		}
	| NUMBER
		{
			if($NUMBER.text != null) {
				result.add($NUMBER.text);
			}
		}
	;	

DOT
	:	'.'
	;

TEXT
	: 'A'..'Z'+
  ;

NUMBER
	:	'0'..'9'+
	;

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