The shortest path (we all love short paths) to getting a good answer to this question would be to follow-up in this thread with a brief example of the input data that includes examples of any possible quirks the data might exhibit. Also post 20 lines or less of what code you've already written to get started in the project. Then ask a specific question about the portion of the project where you're stuck.

Having read perlintro first will enable you to better understand the advice that follows. perlintro should take about 30 minutes to get through, and it's a reasonable expectation on our part that you become familiar with it first.