I would implement the algorithm in nearly the same way, but I think I would implement it as a filter-like command which you can easily call in pipe sequences. So read-in with the magic <> (this allows for STDIN or infiles) and print to STDOUT (which can be easily redirected into outfile from the command line). This has the additional advantage of shortening the program a bit:

#!/usr/bin/perl -w
use strict;
 
my $uclc = shift or Usage();

# fixed small bug
Usage() unless ($uclc eq 'lc' or $uclc eq 'uc');

# to speed up the loop (no string compare on every line)
my $lc = $uclc eq 'lc';
 
while (<>) {
  print $lc ? lc : uc;
}
 
sub Usage {
   die << EO_USE;
Usage: uclc.pl (lc|uc) [infile]
reads from STDIN if no infile given
EO_USE
}
[download]

Update: Oops, thanks tilly, fixed: while instead of foreach.

-- Hofmator