[antlr-interest] Bug in ANTLRv3.g

[Víctor Manuel Sánchez Cartagena]

Thanks Terr. I modified the grammar as you suggested and it doesn't fail
now.
But I discovered a new bug:

If I add a "header" action with a comment inside, I get the following error:

BR.recoverFromMismatchedToken
line 6:0 mismatched input '@' expecting ACTION

But if I remove the comment, I get no errors.

This is the grammar:
grammar plp2;
@header
{    //comment
    import java.lang.String;
}
@members {
  public String refactor_rename_this;
  public String refactor_rename_to;
}
s returns [String value]
    :
                c[""]
    ;
c[String ambito] returns [String value]
    :    'class' ID
        {
                      $value= "class " + $ID.text + " n;" + "} n";
                }
    ;
ID    :    ('a'..'z'|'A'..'Z')('a'..'z'|'A'..'Z'|'0'..'9')*
    ;
ENTERO    :    ('0'..'9')+
    ;

2008/7/30 Terence Parr <parrt at cs.usfca.edu>

> ooops. try this:
>
> ragment
> NESTED_ACTION :
>        '{'
>        (       options {greedy=false; k=2;}
>        :       NESTED_ACTION
>        |       SL_COMMENT
>        |       ML_COMMENT
>        |       ACTION_STRING_LITERAL
>        |       ACTION_CHAR_LITERAL
>        |       .
>        )*
>        '}'
>   ;
> Ter
>
> On Jul 29, 2008, at 2:33 PM, Víctor Manuel Sánchez Cartagena wrote:
>
>  Thank you very much. I changed the rule and now it doesn't fail in: c[""]
>> But it still fails in: $value= "class " + $ID.text + " n;" + "}n";
>>
>> In fact, ANTLR gives an error in every string that starts with }.
>>
>> In this case, I get the following output:
>>
>> line 26:0 mismatched character '<EOF>' expecting '"'
>> line 19:62 no viable alternative at input 'n'
>> BR.recoverFromMismatchedToken
>>
>>
>> 2008/7/29 Terence Parr <parrt at cs.usfca.edu>
>> Ooops. change rule
>>
>> fragment
>> ACTION_STRING_LITERAL
>>       :       '"' (ACTION_ESC|~('\\'|'"'))* '"'
>>       ;
>>
>> + to *
>>
>> Ter
>>
>>
>> On Jul 29, 2008, at 10:53 AM, Víctor Manuel Sánchez Cartagena wrote:
>>
>> I think I have found a bug in the ANTLRv3.g grammar.
>> I build the lexer and the parser with ANTLR, and when I try to parse the
>>  following grammar, it fails, although the grammar is correct.
>>
>> grammar plp2;
>> @members {
>>  public String refactor_rename_this;
>>  public String refactor_rename_to;
>> }
>> s returns [String value]
>>   :
>>               // It doesn't fail:
>>                c["aa"]
>>               // It fails:
>>               //c[""]
>>   ;
>> c[String ambito] returns [String value]
>>   :    'class' ID
>>       {
>>                    // It doesn't fail:
>>                     $value= "class " + $ID.text + " n;" + "{}n";
>>                    // It fails:
>>                    // $value= "class " + $ID.text + " n;" + "}n";
>>               }
>>   ;
>> ID    :    ('a'..'z'|'A'..'Z')('a'..'z'|'A'..'Z'|'0'..'9')*
>>   ;
>> ENTERO    :    ('0'..'9')+
>>   ;
>>
>> Could anyone help me to find out what's wrong in ANTLRv3.g?
>>
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.antlr.org/pipermail/antlr-interest/attachments/20080807/0fa20bd8/attachment-0001.html