Your problem description wasn't very clear, and it kind of looks like an XY Problem actually, why are you trying to do that? Are you sure there's not a better way to achieve your goal?.

Anyway, it looks like you're asking for tr///, tr/a-zA-Z//cd; will remove any non-alphabetic char from your string, tr/a-zA-Z//cdr; will return a new string equal to the first without the non-alphabetic chars.

#!/usr/bin/perl
use warnings;
use strict;
use feature qw{ say };

my @array1 = qw(fil01_sw100056_02.04.js mn_sw100868_15.17.js
                roof_app01_sw12345_26.02.hex roof_dir01_sw101481_01.32
+.dir);

my @array2 = qw(fil01_sw100056_02.03.js mn_sw100868_15.16.js
                roof_app01_sw101483_25.05.hex roof_dir01_sw101481_01.2
+5.dir);

my $string = 'filswjs';

(my $regex = $string) =~ s/(.)(?=.)/$1.*/g;
say for grep /$regex/, @array1, @array2;
[download]

If I understood correctly, OP wanted to ignore everything but alphabetic ("string") characters, as opposed to numbers ("no") and special characters. So the regex would have to be $regex = join "[^a-zA-Z]*", split //, $string;

Though the more I look at the question, the more it looks like an XY problem

#!/usr/bin/env perl
use strict;
use warnings;
use Data::Dump;

my @array1 = (1,2,3,4);
my @array2 = (4,5,6,7);

my %hash;
@hash{@array1} = undef; #Hash Slice populates keys of hash using value
+s from array

my @common = grep { exists $hash{$_} } @array2;

dd @common; #4
[download]