[antlr-interest] Re: Unified grammar and # directives for a C-like language

[uprightness_of_character]

I posted another message (I'm not sure whether it's lost or on its 
way) explaining the problems with the two proposed solutions to the 
'#'-directive problem. In short:

* Jim O'Connor's suggestion engenders an ambiguity that antlr resolves 
in the undesired direction;
* Monty's suggestion recognizes '#'-directives that start in column 1, 
but without leading spaces.

I've had a lot to learn from your answers - thanks a lot - and I 
synthesized something that works. I wonder if this is elegant enough, 
or if there's an easier and clearer solution. 

I defined my whitespaces like this:

// Whitespace (no newlines) -- ignored
WS 
    { const bool firstCol = getColumn() == 1; }
    :
    (
        ' '
        | '\t'
        | '\f'
    )+
    { 
        $setType(antlr::Token::SKIP);
    }
    ( { firstCol }? '#' { $setType(HASH_DIRECTIVE_BEGIN); } )? 
    ;

And I also defined:

HASH_DIRECTIVE_BEGIN : { getColumn()==1}? '#' ;

These two rules match the '#' directives correctly. I am a bit unhappy 
that the concept is spread in two rules, so if there's a better 
solution, please let me know!


Andrei



 

Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/