This headline gave me flashbacks.
I was tired of my team having to load multiple 100MB+ tree killer reports from the repository just to access ~10 records of control totals. I poked around with grep but wasn't landing on the solution. After much googling, I turned to sed. I have retained next to nothing of what I learned from that little adventure, but long story short those tree-killers started sprouting wee control total reports thanks to some minor additions to the job scheduler. Perfect for importing into Excel to appease the bean-counters.