Re: Matching or masking specific characters in an array of strings

[almut]

Here's a bit-logic based approach:

#!/usr/bin/perl -w
use strict;

my %groups = ( 
    group1 => ['cooling', 'rooting', 'hooting', 'looking', 'doormat', 
+'cooking', 'cookies', 'noodles'], # desired result "-oo----"
    group2 => ['yourself', 'southern'], # desired result "-ou--e--"
    group3 => ['arden', 'arlen', 'asked'] # desired result "a--e-"
);

sub show_bits {
    #print unpack("B*", shift), "\n";
}

for my $g (values %groups) {

    my $w1 = $g->[0];
    my $and = "\xff" x length($w1);
    my $or  = "\0"   x length($w1);
    
    for my $w (@$g) {
        print "$w\n";
        #print join(" " x 7, split //, $w), "\n";
        show_bits($w);
        
        $and &= $w;         show_bits($and);
        $or  |= $w;         show_bits($or);
    }
    
    my $xor = $and ^ $or;   show_bits($xor);
    $xor =~ tr/\0/\xff/c;   show_bits($xor);
    my $mask = ~$xor;       show_bits($mask);
    
    my $common = $w1 & $mask;
    $common =~ tr/\0/-/;
    print "$common\n";

    my $i = -1;
    print map {$i++; $_ ne "-" ? "$_ : $i\n" : ()} split //, $common;
    print "\n";
}

__END__
yourself
southern
-ou--e--
o : 1
u : 2
e : 5

cooling
rooting
hooting
looking
doormat
cooking
cookies
noodles
-oo----
o : 1
o : 2

arden
arlen
asked
a--e-
a : 0
e : 3

Uncomment the commented out prints to see how the bit strings are being transformed/combined...

(Update: fixed cut-n-paste error $and ^= $or — thanks johngg)

[duggles]

Thank you very much almut! We can consider this question closed!

I knew I could count on the monks! In addition to my regex skills being way below what they should be, my bit manipulation and use of packing is much worse...

The code is elegant, effective, and I'm certain much faster than what I had. I will be studying this code to see if I can use this sort of logic to help me in future coding efforts.

Thanks again!

If the knowledge and innovative thinking of the perlmonks could only be harnessed for a single purpose - I'm sure world peace would be a piece of cake!!!!

--- duggles ;-)

[nbezzala]

Can someone please explain what is happening here? I'm wondering why a solution like this isn't mentioned in this thread - $w =~ s/^aeiou/-/g;

[jdporter]

They're not talking about "masking away" characters from a given character set in a single string. They're talking about masking multiple strings together in order find out which characters those strings have in common. Hence:

  southern
- yourself
----------
  -ou--e--

(where - (minus) is the masking operation.)