Create a class in the normal way and use overload to overload the reference to produce the value in a string context:

c:\test>p1
package Enhanced::String;

use overload '""' => \&asString;

sub new {
    my $class = shift;
    return bless { string => shift }, $class;
}

sub substr{ 
    return substr( $_[0]{ string }, $_[1], $_[2] ) 
};;

sub asString { 
    return $_[ 0 ]{ string } 
};

package main;

my $obj = Enhanced::String->new( 'the string' );;

print $obj->asString;;
the string

print $obj;;
the string

print $obj->substr( 4, 3 );;
str
[download]

thanks a ton! that did it

package Email;
use overload '""'=> \&ADDRESS , fallback=> 1;

my     $RE_EMAIL= qr/^[\w\-\+\._]+\@[a-zA-Z0-9][-a-zA-Z0-9\.]*\.[a-zA-
+Z]+$/ ;

sub new {
    my  $proto= $_[0];
    my  $class= ref($proto) || $proto;
    my  $self= bless( {} ,  $class );
    $self->{ADDRESS}= undef;
    $self->ADDRESS( $_[1] );
    return $self;
}

sub ADDRESS {
    my $self= shift;
    if ( @_ ) { 
        my     $email= shift;
        $email=~ $RE_EMAIL or die "Not a valid email address";
        $self->{ADDRESS}= $email;
    }
    return $self->{ADDRESS};
}

sub _as_str {
    my  ( $self )= @_;
    return $self->ADDRESS;
    
}
[download]

Why not just validate the user input early on, and send them back to the form entry if they do not provide an email address? I am fond of using Data::FormValidator and HTML::FillInForm for this. If you use CGI::Application then CGI::Application::Plugin::ValidateRM encapsulates this behaviour for you quite nicely.

I already do that for user submitted data.

I need this functionality for data that is coming out of my database, that I perform various markup or logic operations on. If I pull an email address out of the DB, all of my various functions still check it to verfiy that it is an email when they're passed it.

I don't want to accidentally send an email to a non-email, or turn a non-http address into an href. Those functions need to catch errors, and every function I write checks the input data.

With some custom classes, I can just check the ISA before running a regex. thats way more efficient than running a regex over and over again. And if I'm not using a class, but a string, then I just run the regex.