A related problem I once had fun with. Process a file and only print lines once (unlike uniq no assumption of sorted input). That isn't quite what you wanted, but here was the fastest version I could come up with in Perl:

perl -ne 'print if 1 == ++$s{$_}' < in > out
[download]

Here's the fastest and shortest way I could find to process an input file, and print the unique lines inside it:

perl -ne "$s{$_} ||= print" < in > out

Explanation: The ||= operator has a behaviour called "short-circuit". If the left-hand side of the operation evaluates to true, the operation will short-circuit and stop - essentially the right-hand side will be ignored and never evaluated. But if the left side is false, then whatever the right-hand side evaluates to will be assigned to the left.

In this case, we are processing each line from a file called "in". The first time we see $s{$_} it will evaluate to false, setting off the right hand side to be evaluted. This will print $_, return a 1, then set $s{$_} to 1.

The next time we see $s{$_} inside the loop, it is true, so the ||= never again runs the right-hand side operation. This results in the above code only printing a line the first time it's seen, and no more.

can't resist...

perl -ne '!$s{$_}++ && print' < in > out
[download]

   MeowChow                                   
               s aamecha.s a..a\u$&owag.print

That executes more slowly. You have to waste time creating new variables because of using postincrement rather than preincrement.

update: Well most of those nodes discuss how to create a completely unique list, which is not exactly what you want, so here:

update2: Fixed some yucky code... =)

my @in = (1,2,3,4,4,4,5,5,6,7,8,9,9,1,2,3,3);

## one way

my @out1 = $in[0];
do { push @out1, $_ if $out1[$#out1] ne $_ } for @in;

## or another way

my $t = '';
my @out2 = grep { $t eq $_ ? 0 : ($t = $_, 1) } @in;
[download]

   MeowChow                                   
               s aamecha.s a..a\u$&owag.print