I'm not sure what you mean by "how will the program know which string to modify?". I assume the answer to that would be "by following the algorithm, and modifying the string the algorithm says to".

One of the standard tricks for detecting duplicates is by using a hash:

# Read passwd files from @ARGV, outputting one record per
# username
my %users;
while (<>) {
  $user = split ":",$_,1;
  next if exists $users{$user};
  print;
  $users{$user} = 1;
}
[download]

The first time username foo is found, $users{'foo'} does not exist, so the record is printed out, and $users{'foo'} is set to 1. The second time username foo is found, $users{'foo'} is now 1, so the record is skipped.

You didn't want to skip records, but rather rekey the records, so that the duplicates have different keys -- you wanted your outputted flat-file to have unique keys. Doing that with the passwd example:

# Read passwd files from @ARGV, outputting one record per
# username.  Duplicate usernames are renamed to be unique
my %users;
while (<>) {
  $user = split ":",$_,1;
#  next if exists $users{$user};
  $_ = "a$_" , redo if exists $users{$user};
  print;
  $users{$user} = 1;
}
[download]

The $_ .= "a" , redo statement that replaces the next appends "a" to the incoming line and then reprocesses it. I decided to reprocess it rather than just add "a" because if I had three user "foo" in the input files, I wouldn't want to get users "foo", "afoo", and "afoo", I'd want "foo", "afoo", "aafoo".

I hope this helps.