[Regexp] Possible non-conformance with Perl 5 in java distribution of gnu.regexp

I have a query on the interpretation of a regular expression within the 
java version of GNU regexp. I think it's a bug, but I may be wrong.

The expression I am trying to reproduce is equivalent to the foillowing 
which I verified in a Perl5 script interpreter...

#!/usr/bin/perl
while(<>){
   $contents = $_;
   $contents =~ s/ *(,*) *gzip *(\1*),* */\2/;
   print "$contents";
}

...and should remove the entry gzip from a comma separated list, whilst 
tidying up the commas (in other words it substitutes a comma if gzip is 
bounded on both sides by commas, otherwise it substitutes the empty 
string if gzip is at the start or end of the list), so...


gzip, deplace:dfokj
slkkf, gzip, las;kj
slkjf, lkhjassd , gzip

...gets processed by the perl script to...

deplace:dfokj
slkkf,las;kj
slkjf, lkhjassd

The perl VM which processes this accurately shows the following version 
information...

perl, v5.8.1-RC3

My code do define the RE and its substitute looks like this currently...

             ACCEPT_ZIP_PATTERN = new RE(" *(,*) *gzip *(\1*),* *");
             ACCEPT_ZIP_SUBSTITUTE = "\2";
             processed_content_types =
                 
ACCEPT_ZIP_PATTERN.substituteAll(processed_content_types, 
ACCEPT_ZIP_SUBSTITUTE);

The equivalent values which I have attempted within the java PERL5 
interpreter include...

Multiple different combinations of the backreferences 1 and 2 using 
either \ or $ to represent them.
Switching to the RE_PERL_5_S syntax instead of the default RE_PERL_5 to 
see if that changed anything.

I spent ages trying to debug why it didn't work in java before I 
verified it in the real perl engine. Now I am concerned that the java 
engine is actually not conformant with PERL5

Alternatively what am I doing wrong to invoke backreferences in the 
java distribution of gnu.regexp?

Cefn
http://cefn.com

_______________________________________________
Gnu-regexp-users mailing list
[email protected]
http://lists.gnu.org/mailman/listinfo/gnu-regexp-users