@array = ('x', 'x', 'y', 'z', 'z', 'z');
my %items;
for my $i (0 .. $#array) {
    $items{$array[$i]} ||= [ $i, $i ];  # only assigns for the first o
+ccurrence
    $items{$array[$i]}[1] = $i;         # replaces for every new occur
+rence
}

use Data::Dumper;
print Dumper \%items;
[download]

$VAR1 = {
          'x' => [
                   0,
                   1
                 ],
          'y' => [
                   2,
                   2
                 ],
          'z' => [
                   3,
                   5
                 ]
        };
[download]

The tricks it uses to achieve this, is using ||=, which only assigns at the first occurrence, provided that what it assigns is always true; and = which always replaces what was there already, in the end leaving it in the state of the last assignment.

In case you have multiple ranges for the same item, it may or may not do what you want.

#! perl -slw
use strict;

my  @array = qw[ x x y  z z z ];

my @idxs = 0;
$array[$_-1] ne $array[$_] and push @idxs, $_-1, $_ for 1 .. $#array;
push @idxs, $#array;

print "@idxs";
[download]

Let me see if I can expand this correctly...

my @idxs = (0);
my $i;

foreach $i (1 .. $#array) {
    if ($array[$i - 1] ne $array[$i]) {
        push @idxs, $i - 1, $i;
    }
}
push @idxs, $#array;

print "@idxs";
[download]

Well, if there is a perl-ish way to do it, I just don't know how to do it. Here is the simple loop over array while storing the start and end of each char way:

@array=qw(x x y z z z);

$c=$array[0];
$start=0;
$end=0;

for $index (1..$#array){
if (!($c eq $array[$index])){
push(@output, $start, $end);
$c=$array[$index];
$start=$index;
$end=$index;
}else{
$end=$index;
}
}
push(@output, $start, $end);

print join(" ", @output), "\n";

Outputs:
0 1 2 2 3 5
[download]

I'm sure there someone else can probably do better.

Cheers

The following seems to do what you want. Not sure if this is the most efficient approach though.

use strict;
use warnings;
use Data::Dumper;

my (@array, $iIndex, $thisValue, $lastValue, @out);

@array         = qw(x x y z z z);
@out        = (0);
$lastValue     = $array[0];

for ($iIndex = 0; $iIndex <= $#array; $iIndex++) {

    $thisValue = $array[$iIndex];
    
    if ($thisValue ne $lastValue) {
        push @out, ($iIndex - 1, $iIndex);
        $lastValue = $thisValue;
    }

}

push @out, $#array;

print Dumper(\@out);
[download]

Regards,
Dom.

Actually, now I come to look at the output from dumper:

$VAR1 = [
          0,
          '1',
          2,
          '2',
          3,
          5
        ];
[download]

I'm wondering about the significance of some of the numbers coming out in quotes. At first I wondered if it had something to do with the brackets on the push line.

Can anyone enlighten me?

Cheers,
Dom.

It is a bug in the XS version of Data::Dumper before v2.11. To see your version id:

perl -MData::Dumper -wle 'print $Data::Dumper::VERSION'
[download]

2.102 is in the base code for Perl 5.6.1, and has the bug.
2.12 is in the base code for Perl 5.8.0, and does not have the bug.


My testing script:

perl -MData::Dumper -wle '$Data::Dumper::Indent=0;print Dumper [ map {
+ ($_, $_+0) } 0..3 ]'
[download]

Good output:

$VAR1 = [0,0,1,1,2,2,3,3];
[download]

Buggy output:

$VAR1 = [0,'0',1,'1',2,'2',3,'3'];
[download]

If you set Useqq, as a side-effect, it will force Data::Dumper to use the pure-Perl forms of its code.
This is slower, but will work around the bug in older versions.

$Data::Dumper::Useqq = 1;
[download]