use strict;

my @thing;

# open up the DATA filehandle and loop one line at a time
while (<DATA>) {

   # look for first \ and capture everything until the next \ into $1
   # discard everything (.*) but don't be greedy (?)
   # capture any alphanumeric (\w) that is followed by .cxx into $2
   if (/\\([^\\]+).*?(\w+\.cxx)/) {
      # store $1 and $2 into a datastructure    
      push @thing, [$1,$2];
   }
}

for (@thing) {
   my ($dir,$file) = @$_;
   print "$dir\t$file\n";
} 

__DATA__
qzang.5
    \nbsstools\i95\src\i95val.h@@\main\golden\2
    \nbsstools\i95\src\i95val.cxx@@\main\golden\2
    \nbsstools\rlp\src\rlpval.cxx@@\main\golden\3
    \nbsstools\rlp\src\rlpval.h@@\main\golden\3
source file(s) -- 4
rajeevv.19
    \nbsscallpmsg\acl\src@@\main\golden\4
    \nbsscallpmsg\acl\src\aclpi.h@@\main\golden\1
    \nbsscallpmsg\acl\src\aclpi.cxx@@\main\golden\1
    \nbsssbs\appl\src@@\main\golden\4
    \nbsssbs\appl\src\radiomeasurements.h@@\main\golden\1
[download]

UPDATE per your reply:
This regex works for the data you provided.

jeffa

L-LL-L--L-LL-L--L-LL-L--
-R--R-RR-R--R-RR-R--R-RR
B--B--B--B--B--B--B--B--
H---H---H---H---H---H---
(the triplet paradiddle with high-hat)