perl: make script fast to use big file - perl

Question

My problem is how to make my script fast (I use big files)
I have the script above it add "bbb" between words if the words exist in an other file that contain sequences of words
for exemple file2.txt : i eat big pizza .my big pizza ...
file1.txt (sequences):
eat big pizza
big pizza
the result Newfile
i eatbbbbigbbbpizza.my bigbbbpizza ...
my script:
use strict;
use warnings;
use autodie;
open Newfile ,">./newfile.txt" or die "Cannot create Newfile.txt";
my %replacement;
my ($f1, $f2) = ('file1.txt', 'file2.txt');
open(my $fh, $f1);
my #seq;
foreach (<$fh> )
{
chomp;
s/^\s+|\s+$//g;
push #seq, $_;
}
close $fh;
#seq = sort bylen #seq;
open($fh, $f2);
foreach (<$fh> ) {
foreach my $r (#seq) {
my $t = $r;
$t =~ s/\h+/bbb/g;
s/$r/$t/g;
}
print Newfile ;
}
close $fh;
close Newfile ;
exit 0;
sub bylen {
length($b) <=> length($a);
}

Answer 1

Instead of an array
my #seq;
define your words as a hash.
my %seq;
Instead of pushing the words
push #seq, $_;
store the words in the hash. Precalculate the replacement and move it out of the loop.
my $t = $_;
$t =~ s/\h+/bbb/g;
$seq{$_} = $t;
Precalculate the words in front of the outer loop:
my #seq = keys %seq;
And use hash look-ups to find the replacement in the inner loop:
my $t = $seq{$r};
This might be a bit faster, but do not expect too much.
In most cases it is better to reduce the problem by preparing the input in a way, which makes the solution easier. For example grep -f is much faster than your Perl loops. Use grep to find the lines, which need a replacement, and do the replacement with Perl or Sed.
Another way is to parallel the job. You can divide your input in n parts and run n processes on n CPUs in parallel. See the GNU parallel tutorial.

Answer 2

What about a regexp like this (beware that this approach can cause security concerns) ?
use strict;
use warnings;
open (my $Newfile, '>', 'newfile.txt') or die "Cannot create Newfile.txt: $!";
my ($f1, $f2) = qw(file1.txt file2.txt);
open (my $fh, $f1) or die "Can't open $f1 for reading: $!";
my #seq = map {split ' ', $_ } <$fh>;
close $fh;
# an improvement would be to use an hash to avoid dupplicates
my $regexp = '(' . join('|', #seq) . ')';
open($fh, $f2) or die "Can't open $f2 for reading: $!";
foreach my $line (<$fh> ) {
$line =~ s/$regexp/$1bbb/g;
print $Newfile $line;
}
close $fh;
close $Newfile ;
exit 0;