[antlr-interest] Problem with Duplicating nodes/trees

[Terence Parr]

I believe I have fixed these in the 3.1 beta

Ter
On Feb 8, 2008, at 12:10 PM, Grzegorz Cieslewski wrote:

> I did some more testing and found a simpler case that demonstrates  
> the problem
>
> grammar T2;
> options {output=AST;}
> tokens
> {
> 	TEST;
> }
> a : b -> ^(TEST b b)
> ;
> b 	:	INT;
> c	: 	ID;
>
> ID : 'a'..'z'+ ;
> INT : '0'..'9'+;
> WS : (' '|'\n') {$channel=HIDDEN;} ;
>
> For input "123" I was expecting ^(TEST 123 123) but I get ^(TEST 123).
> It seems to me that this is a bug.
>
>
> On Feb 8, 2008 10:58 AM, Grzegorz Cieslewski  
> <cieslewski at hcs.ufl.edu> wrote:
>> I looked through the code that ANTLR generates.  It seems to me that
>> the function
>> org.antlr.runtime.tree.RewriteRuleElementStream.next() is not
>> following the spec.
>> It does not perform duplication correctly when the size()==1.
>>
>>
>>
>> On Feb 8, 2008 9:38 AM, Mark Volkmann <r.mark.volkmann at gmail.com>  
>> wrote:
>>> On Feb 8, 2008 8:00 AM, Grzegorz Cieslewski  
>>> <cieslewski at hcs.ufl.edu> wrote:
>>>> Hello All,
>>>>
>>>> I have encountered a problem with tree duplication during AST
>>>> construction.  I boiled my problem to following grammar.
>>>>
>>>> grammar T2;
>>>> options {output=AST;}
>>>> tokens
>>>> {
>>>>        TEST;
>>>> }
>>>> a : c (',' c)* b -> ^(TEST c b)+
>>>> ;
>>>> b       :       INT;
>>>> c       :       ID;
>>>>
>>>> ID : 'a'..'z'+ ;
>>>> INT : '0'..'9'+;
>>>> WS : (' '|'\n') {$channel=HIDDEN;} ;
>>>>
>>>> For input like "x, y, z 1", I was expecting to get a tree looking  
>>>> like
>>>> ^(TEST x 1) ^(TEST y 1) ^(TEST z 1),
>>>> but the antlr seems not to duplicate the b rule and I get ^(TEST  
>>>> x 1)
>>>> ^(TEST y) ^(TEST z).
>>>>
>>>> Does any one know what am I doing wrong?  Is this a bug or a  
>>>> "feature".
>>>
>>> Based on section 7.5 of the book, in the subsection titled
>>> "Duplicating Nodes and Trees", what you have looks correct. The
>>> relevant statement is this.
>>>
>>> "In a rewrite rule, ANTLR duplicates any element with cardinality  
>>> one
>>> when referenced more than once or encountered more than once because
>>> of an EBNF * or + suffice operator."
>>>
>>> b is your element with a cardinality of one and it is encountered  
>>> more
>>> than once in your rewrite rule with a + operator, but it isn't being
>>> duplicated.
>>>
>>> I tried your grammar in the latest intermediate build and it doesn't
>>> work. I hope someone else has an idea why. I think it's possible
>>> you've found a bug.
>>>
>>> --
>>> R. Mark Volkmann
>>> Object Computing, Inc.
>>>
>>
>>
>>
>>
>> --
>> =====================================================
>> Grzegorz Cieslewski
>> Research Assistant
>> High-performance Computing & Simulation (HCS) Research Laboratory
>> University of Florida, Dept. of Electrical and Computer Engineering
>> 330 Benton Hall, Gainesville, FL, 32611-6200
>> Phone: (352) 392-9041
>> Email: cieslewski at hcs.ufl.edu
>> Web: www.hcs.ufl.edu
>> =====================================================
>>
>
>
>
> -- 
> =====================================================
> Grzegorz Cieslewski
> Research Assistant
> High-performance Computing & Simulation (HCS) Research Laboratory
> University of Florida, Dept. of Electrical and Computer Engineering
> 330 Benton Hall, Gainesville, FL, 32611-6200
> Phone: (352) 392-9041
> Email: cieslewski at hcs.ufl.edu
> Web: www.hcs.ufl.edu
> =====================================================