If you have a recursive data structure -- say an AoA[oA[oA]...]] -- and you want to iterate over the leaves; then the typical Perl solution is a recursive callback iterator something like this:

#! perl -slw
use strict;
use Data::Dump qw[ pp ];

sub genNested {
    my $n = shift;
    map{ rand() < 0.5 ? [ rand(1000), genNested( $n-1 ) ] : rand( 1000
+ ) } 1 .. $n;
}

sub cbIterator (&$) {
    my( $code, $tree ) = @_;
    for my $i ( 0 .. $#$tree ) {
        if( ref( $tree->[ $i ] ) eq 'ARRAY' ) {
            &cbIterator( $code, $tree->[ $i ] );
        }
        else {
            local $_ = $tree->[ $i ];
            $code->();
        }
    }
}

my @nested = genNested( 3 );
pp \@nested;

cbIterator{
    print;
} \@nested;

__END__
C:\test\nhash>nestedIterator.pl
[
  [
    "711.822509765625",
    "334.869384765625",
    ["942.19970703125", ["600.4638671875"]],
  ],
  "978.3935546875",
  "965.240478515625",
]

711.822509765625
334.869384765625
942.19970703125
600.4638671875
978.3935546875
965.240478515625
[download]

Which is fine unless you need to be able to quit the iteration at an arbitrary point. (ie. short circuit rather completing the full iteration).

What I need is an iterative iterator for a recursive data structure. What I've got so far is this:

#! perl -slw
use strict;
use Data::Dump qw[ pp ];

sub genNested {
    my $n = shift;
    map{ rand() < 0.5 ? [ rand(1000), genNested( $n-1 ) ] : rand( 1000
+ ) } 1 .. $n;
}

sub genIterator {
    my $tree = shift;
    my @stack = [ $tree, 0 ];
    return sub {
        ## Fill in the blank (please :).
    }
}

my @nested = genNested( 3 );
pp \@nested;

my $iter = genIterator( \@nested );

while( my $next = $iter->() ) {
    last if $next == $something;
    ## do stuff with $next.
}
[download]

But my mind's gone blank. Any offers for how to fill the blank?