I don't feel I clearly understand the problem ccelt09 is trying to address, but here's an approach (one of many) that might be helpful as a first approximation to a solution. Note: No validation of ranges is done to insure against negative start/end offsets, start/end inversion, overlap, etc.

>perl -wMstrict -le
"my $test = 'ABCDEFGHIJKLMNOPQRSTUVWXYZ';
 ;;
 my %inclusive_ranges = map { $_ => 1 } 7 .. 11, 18 .. 22;
 ;;
 my @excluded = grep ! $inclusive_ranges{$_}, 0 .. length($test) - 1;
 ;;
 my $str = $test;
 print qq{'$str'};
 ;;
 substr $str, $_, 1, '-' for @excluded;
 print qq{'$str'};
"
'ABCDEFGHIJKLMNOPQRSTUVWXYZ'
'-------HIJKL------STUVW---'
[download]

Update: Another way, same caveats. Note: This approach assumes the  \x00 (null) character is never part of input data.

>perl -wMstrict -le
"my $test = 'ABCDEFGHIJKLMNOPQRSTUVWXYZ';
 ;;
 my %inclusive_ranges = map { $_ => 1 } 7 .. 11, 18 .. 22;
 ;;
 my $exclude_mask =
   join '',
   map $inclusive_ranges{$_} ? qq{\xFF} : qq{\x00},
   0 .. length($test) - 1
   ;
 ;;
 my $str = $test;
 print qq{'$str'};
 ;;
 $str &= $exclude_mask;
 $str =~ tr/\x00/-/;
 print qq{'$str'};
"
'ABCDEFGHIJKLMNOPQRSTUVWXYZ'
'-------HIJKL------STUVW---'
[download]