I read data in a anonymous array reference to save memory.

In normal ways, people read file in:

sub read_file {
    my $url = shift;
    my $file_data;
    open my $fh, '<', $url;
    local $/;
    $file_data = <$fh>;
    show_size();
}

show_size();
read_file('name1.txt');
read_file('name2.txt');
read_file('name3.txt');
read_file('name4.txt');
read_file('name5.txt');
show_size();

sub show_size {
    print `top -l 1 | grep perl | awk '{print "MEM="\$8 "\tRPRVT="\$30
+}'`; # Work on Mac OS X 10.10
}
[download]

Output:
MEM=984K+ RPRVT=N/A
MEM=2052K+ RPRVT=N/A
MEM=2056K+ RPRVT=N/A
MEM=2060K+ RPRVT=N/A
MEM=4900K+ RPRVT=N/A
MEM=34M+ RPRVT=N/A
MEM=34M+ RPRVT=N/A

scalar $file_data's data will not release until program come to end. That makes memory increase and increase.

So, I read file by this way:

sub read_file {
    my $url = shift;
    my $file_data = [];
    open my $fh, '<', $url;
    local $/;
    $file_data->[0] = <$fh>;
    show_size();
}

show_size();
read_file('name1.txt');
read_file('name2.txt');
read_file('name3.txt');
read_file('name4.txt');
read_file('name5.txt');
show_size();

sub show_size {
    print `top -l 1 | grep perl | awk '{print "MEM="\$8 "\tRPRVT="\$30
+}'`; # Work on Mac OS X 10.10
}
[download]

Output:
MEM=1048K+ RPRVT=N/A
MEM=2084K+ RPRVT=N/A
MEM=2084K+ RPRVT=N/A
MEM=2084K+ RPRVT=N/A
MEM=4928K+ RPRVT=N/A
MEM=37M+ RPRVT=N/A
MEM=4932K+ RPRVT=N/A

Perl uses reference counting way to manage memory. When ending to the block, $file_data->[0] will automatically release.

But... Is it memory safe? Or Is there something wrong? I don't know all but it just looks great.

The script is written on Mac OS X 10.10.2 and be tested in Perl 5.18.2. Maybe you should modify this adapting for your system.