I would write it something like this:

use strict;
use warnings;

use Data::Dumper;
$Data::Dumper::Indent = 0;

my $file = "test";

# Use lexical file handles and 3-argument form of open
# See http://modernperlbooks.com/mt/2010/04/three-arg-open-migrating-t
+o-modern-perl.html
open my $fh, '<', $file or die "error opening '$file': $!";

# Slurp file contents into string $contents
# See http://modernperlbooks.com/mt/2009/08/a-one-line-slurp-in-perl-5
+.html
my $contents = do { local $/; <$fh> };
close $fh;

# Split contents into fields by XXX on a line by itself (note the /m m
+odifier)
my @fields = split /^XXX$/m, $contents;

# Tidy up newlines/spaces in multiline fields
for (@fields) { tr/\n/ /s; s/^ +//; s/ +$//; }

print Dumper \@fields;
[download]

Update: See Also

• Lexical file handles - Perl Maven
• 3-argument form of open - Effective Perl (item 52)
• File slurping - Re: Re: Slurp a file by merlyn and Re^3: Extracting selected fields form file record (don't use File::Slurp References)