The usual approach to sorting is to read everything that you want to sort (e.g., into an array), sort it, and then save the result. You're reading the unsorted file a line at a time. You can sort that way--by merging each new line into it's correct position in some data structure, but that's more work, and you won't get the benefit of Perl's built-in sort.

Try something like this:

  my @unsorted = <TOSORT>;
  my @sorted = sort byPhone @unsorted;
[download]

This lets you isolate details for how records will be compared while sorted into a separate subroutine. (See perldoc -f sort for details). Then, you'll do something like

  sub byPhone {
      return (split(/\|/, $a))[1] cmp (split(/\|/, $b))[1];
  }
[download]

to extract and compare phone numbers. Exchange $a and $b if you want to reverse the order.