Re: Regex (lookahead) Confusion

[Roger]

while (<DATA>) {
    chomp;
    if (! /([smtwhfa])(?=.*?\1)/) {
       print "$_ : OK\n";
    } else {
       print "$_ : Not OK\n";
    }
}


__DATA__
smsa
smta
stmwhas

And the output -

smsa : Not OK
smta : OK
stmwhas : Not OK

Update: Ah! Thanks people for pointing out that my regex didn't check for non-allowed characters. bart's solution is perfect.

[allolex]

This is a really good effort, but it fails if characters not matching the given character set are included in the string.

Unfortunately, I don't have a solution, either because in my mind, there would have to be a lookbehind on a string of variable length, which (of course) will not work.

#!/usr/bin/perl
use strict;
use warnings;

while(<DATA>) {

        chomp;

        if (! /([smtwhfa])(?=.*?\1)/) {

                print "$_ : OK\n";
        }

        else {
                print "$_ : Not OK\n";
        }
}

__DATA__
swma
smqa
smsa
fhtm
ttma
t2ms

__END__

swma : OK
smqa : OK
smsa : Not OK
fhtm : OK
ttma : Not OK
t2ms : OK

[Anonymous Monk]

Not quite what the OP wanted. It was required that the word was completely made of allowed characters only. Try this:

while (<DATA>) {
    chomp;
    if (! /([smtwhfa])(?=.*?\1)/) {
       print "$_ : OK\n";
    } else {
       print "$_ : Not OK\n";
    }
}


__DATA__
smsa
smta
stmwhas

BADsmtaEXAMPLE

And the output

smsa : Not OK
smta : OK
stmwhas : Not OK
 : OK
BADsmtaEXAMPLE : OK

[allolex]

Except that this appears to be exactly the same code as Roger used.

[Anonymous Monk]

True. The point is that he didn't test it with the right data.

I was saying that his code will give wrong results for the two lines I added, just to prove my assertion.

[flyingmoose]

Roger, can you please explain this a bit to us mere regex mortals? Thanks!

[Roger]

if (! /([smtwhfa])(?=.*?\1)/) {
    |    |         |
    |    |         +----- followed by this character
    |    +--------------- find any of these characters
    +-------------------- but take the negative of the match


bart's regex below is much better, see his reply and follow-up's.