I'd replace $file =~ s/\s+/_/g; with $file =~ y/_//sd; and here is why (this is supposed to be a well known fact)

use Benchmark qw( cmpthese );
cmpthese(
    100_000,
    { 's' => sub {
                 $a='a _ b __ c';
                 $a=~s/_+//g;
             },
      'y' => sub {
                 $a='a _ b __ c';
                 $a=~ y/_//sd;
             }
    }
);
__END__
Benchmark: timing 100000 iterations of s, y...
         s:  9 wallclock secs ( 7.18 usr +  0.00 sys =  7.18 CPU) @ 13
+927.58/s (n=100000)
         y:  2 wallclock secs ( 2.19 usr +  0.00 sys =  2.19 CPU) @ 45
+578.85/s (n=100000)
     Rate    s    y
s 13928/s   -- -69%
y 45579/s 227%   --
[download]

 

______crazyinsomniac_____________________________ Of all the things I've lost, I miss my mind the most. perl -e "$q=$_;map({chr unpack qq;H*;,$_}split(q;;,q*H*));print;$q/$q;"