#!/usr/local/bin/perl -wi.bak
use strict;

my @recs;
while (<>) {
    chomp;
    push @{$recs[int(($.-1) / 5)]}, split /,?\s/;
    print map join('|', @$_) . "\n", $recs[int($.-1) / 5]
        if !($. % 5);
}
[download]

% reformat.pl <file>
[download]

my $data =~ tr/ /|/;
$data =~ tr/,//d; # get rid of the comma, if you want
[download]

If there's a possibility that any of the fields will contain spaces or pipes or commas, look at the Text::CSV module from CPAN. Writing a regex to take care of this will not only test your skills at planning for all contingencies, it will cause you to go bald.

Anyway:

use strict;

open FH, "foo" or die "Can't open: $!";
my @recs;
while (<FH>) {
    chomp;
    push @{$recs[int(($.-1) / 5)]}, $_;
}
close FH or die "Can't close: $!";

open FH, ">foo.new" or die "Can't open: $!";
for my $ref (@recs) {
    # break up the city, state, zip into 3 parts
    my($city, $state, $zip);
    if ($ref->[2] =~ /(.*?),\s(.*?)\s(.*)/) {
        ($city, $state, $zip) = ($1, $2, $3);
    }

    # join it all together into a pipe-separated
    # record, then write it out
    my $new = join "|", @{$ref}[0,1], $city, $state,
        $zip, @{$ref}[3,4];
    print FH $new, "\n";
}
close FH;
[download]

Deuglification, then:

use strict;
open FH, "foo" or die "Can't open: $!";
my @recs;
foreach my $line (<FH>) {
    chomp $line;
    push @recs, [ split /,?\s/, $line ];
}
close FH or die "Can't close: $!";

open FH, ">foo.new" or die "Can't open: $!";
foreach my $line_ref (@recs) {
    my $line = join '|', @$line_ref[0 .. 4];
    print FH $line, "\n";
}
close FH;
[download]

Note that this is also untested. I maintain that a proper use of split is better than an apple a day.

Interesting bits for the Original Poster:

• We push an array reference onto @recs
• split can take an arbitrarily complex regex, instead of just a single character. Use it liberally!
• We use an array slice to get at only the first few fields we want.

Yes, but then, mine worked. :) Just kidding.

You're just pushing each line arbitrarily onto @recs--you need to group them in sets of 5, because that's what the original file looked like. Plus why are you using foreach in the original reading-in-the-file loop? That

foreach my $line (<FH>)
[download]

reads the entire file into a temp array--not a horrible thing in many cases, but still, if we can process on a line by line basis, we may as well do so. :)

I like your use of split quite a lot, though.

So, combining your ideas and mine:

use strict;

 open FH, "foo" or die "Can't open: $!";
 my @recs;
 while (<FH>) {
     chomp;
     push @{$recs[int(($.-1) / 5)]}, split /,?\s/;
 }
 close FH or die "Can't close: $!";

 open FH, ">foo.new" or die "Can't open: $!";
 print FH map join('|', @$_) . "\n", @recs;
 close FH;
[download]

Notice I took the array slice out--I think the op wanted everything in the array. If not, though, he/she should stick it back in, just

@$ref[0..4]
[download]

instead of

@$ref
[download]

And I'm now using map, just cause map is great.