You could do a single split on either a '|' or a ':' then process the resulting stream two items (key and value) at a time filtering out the 'special' pairs. You could use the core List::Util->pairs() if on 5.20 or later, or List::MoreUtils->natatime() from CPAN if not, to process in pairs. I use groupsOf() which I wrote when when working in a restricted environment with early Perl versions and no chance of installing non-core modules.

use strict;
use warnings;

use Data::Dumper;

sub groupsOf (&$@);

my $dbStr =
   q{special:1001:area_code:617|special:1001:zip_code:02205|special:10
+01:dow:0|special:1001:tod:14};

my $rhRes = {
   groupsOf { $_[ 0 ] eq q{special} ? () : @_ } 2, split m{\||:}, $dbS
+tr
   };

print Data::Dumper->Dumpxs( [ $rhRes ], [ qw{ rhRes } ] );

sub groupsOf (&$@)
{
    my $rcToRun  = shift;
    my $groupsOf = shift;

    my $rcDoIt;
    $rcDoIt = sub
       {
           $rcToRun->(
              map shift, 1 .. ( @_ < $groupsOf ? @_ : $groupsOf )
              ),
           @_ ? &$rcDoIt : ();
       };
    &$rcDoIt;
}
[download]

The output.

$rhRes = {
           'tod' => '14',
           'zip_code' => '02205',
           'dow' => '0',
           'area_code' => '617'
         };
[download]

I hope this is of interest.

Update: Here's a version using List::Util->pairs(), the output is essentially the same.

use strict;
use warnings;

use List::Util qw{ pairs };
use Data::Dumper;

my $dbStr =
   q{special:1001:area_code:617|special:1001:zip_code:02205|special:10
+01:dow:0|special:1001:tod:14};

my $rhRes = {
   map   { @{ $_ } }
   grep  { $_->[ 0 ] ne q{special} }
   pairs split m{\||:}, $dbStr
   };

print Data::Dumper->Dumpxs( [ $rhRes ], [ qw{ rhRes } ] );
[download]