[antlr-interest] antlr beginner confused over ambiguities

[Sam Barnett-Cormack]

Rick Schumeyer wrote:
> This works, but it gives me one token containing "12:34:56".  I still 
> need to parse the hh,mm,ss values.  What is the best way to do that?
> 
> It seems that I want the parser to be able to tell the lexer "normally 
> you can consider INT:INT:INT as a STRING.  But right now, return tokens 
> for each piece."  Isn't that what a semantic predicate does?  Or am I 
> really missing something? 

Well, you *could* do something funky with multiple-token emitting and so 
on, I suppose, but my immediate reaction would be "hh:mm:ss is a 
fixed-length, fixed-delimeter format. I can do that with a split 
function in the target language". It all depends on what you want to 
consider a "token". From a parse point of view, is a time like that an 
indivisible entity, or three INTs with a delimeter?

Sam

> On Tue, Feb 24, 2009 at 3:32 PM, Gavin Lambert <antlr at mirality.co.nz 
> <mailto:antlr at mirality.co.nz>> wrote:
> 
>     At 08:52 25/02/2009, Rick Schumeyer wrote:
> 
>         If the input string is "12:34:45", then of course antlr returns
>         a STRING token.  I understand why it does that (I think) but I'm
>         not really sure how to fix that.  I'm fuzzy on the difference
>         between a semantic and syntactic predicate, and whether one of
>         those is what I need.
> 
>         grammar STest;
> 
>         line    :    timestamp;
> 
>         timestamp
>            :    INT ':' INT ':' INT
>            ;
> 
>         INT    :    ('0'..'9')+;
> 
>         STRING    :    ('!'..'+' | '-'..'~')+; /* anything but ws or
>         comma */
> 
> 
>     fragment TIMESTAMP: INT ':' INT ':' INT ;
> 
>     INT : ('0'..'9')+ ;
> 
>     STRING
>      : (TIMESTAMP) => TIMESTAMP { $type = TIMESTAMP; }
> 
>      | ('!'..'+' | '-'..'~')+; /* anything but ws or comma */
>      ;
> 
> 
> 
> ------------------------------------------------------------------------
> 
> 
> List: http://www.antlr.org/mailman/listinfo/antlr-interest
> Unsubscribe: http://www.antlr.org/mailman/options/antlr-interest/your-email-address