G'day cbtshare,

Here's the technique I might have used for this task:

#!/usr/bin/env perl

use strict;
use warnings;
use autodie;

use constant {
    IN_FILE => 'pm_1200636_text.txt',
    HEADER => 0,
    KEY    => 1,
    VALUE  => 2,
};

my %parsed;

{
    open my $fh, '<', IN_FILE;

    my $name;

    while (<$fh>) {
        my @fields = split;

        if ($fields[HEADER] eq 'name') {
            $name = $fields[KEY];
            next;
        }

        if ($fields[HEADER] eq 'device') {
            push @{$parsed{$name}{$fields[KEY]}}, $fields[VALUE];
            next;
        }
    }
}

# For testing only
use Data::Dump;
dd \%parsed;
[download]

This only reads a record at a time, so there should be no memory issues that might occur when slurping entire files. The only data that persists after the anonymous block is %parsed: process that as necessary. Also note that as $fh goes out of scope at the end of the anonymous block, Perl automatically closes this for you (there's no need for a close statement in this instance).

I used the same data as you posted (see the spoiler).

Output from a sample run:

{
  Andrew => { ipad => [2009] },
  Brian  => { ipad => [2001, 2001, 2001] },
  ryan   => { cell => [2009], ipad => [2005] },
}
[download]

See also: "perldsc - Perl Data Structures Cookbook"; autodie; open; and, Data::Dump. Everything else is very straightforward and basic Perl, but feel free to ask if anything is unclear.