From an algorithm analysis perspective, avoiding a second pass is a relatively minor optimization. Your algorithm is O(n) with the second pass, and it's still O(n) with the optimization. Unless the performance is so bad that users are timing it with a clock, they won't notice the difference.

If the performance is so bad that users are timing it with a clock, you should probably think about O(log n) possibilities (such as running a binary search against a sorted index) or, depending on the nature of your data, maybe replacing the flat-file implementation with a DBMS.