[antlr-interest] (no subject)

[Imre András]

Gerald Rosenberg <gerald at certiv.net> írta: 


> At 04:36 PM 10/31/2008, =?ISO-8859-2?Q?Imre_Andr=E1s?= wrote:
> 
> >--- unreachable.g --------------------
> >grammar unreachable;
> >
> >ID  :   ('a'..'z'|'A'..'Z')+ ;
> 
> ....
> 
> >sequence returns [String value]:    'SEQUENCE' WS* '{' WS* 
> >e=rightValue {$value=$e.value;} (WS* ',' WS* e=rightValue {$value += 
> >$e.value;})* WS* '}' {return $value;};
> 
> The unreacheable compile analysis-time error is likely due to the 
> fact that ID will consume all alpha characters, leaving no way to 
> match the literal 'SEQUENCE'.
> 
> BTW, since you are skip()ing whitespace, you can remove the WS* from 
> the parser rules with no affect. 
> 
> ady got rid of the unreachable error based on Terence's comments. However, you have a point that the ID and SEQUENCE are ambigous. Interpreting A ::= SEQUENCE{ SEQUENCE } gives a MismatchedTokenException, though parsing continues after a resync.

I put SEQUENCE : 'SEQUENCE' in the grammar, and modified the sequence rule to use it. Now it does not compile because of the ambiguity.

Now how can I explicitly tell that SEQUENCE is a keyword, and not a valid ID?


Thanks,
  András
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.antlr.org/pipermail/antlr-interest/attachments/20081101/f9e38a2e/attachment.html