Each node is identified by part of it's hash and the number of bytes within it. Sadly this WARC doesn't have any deduplication in it, but larger datasets might!
Honestly I'm really happy with the simplicity of this tool. With it I can debug large datasets of #ipfs data without needing to touch the network or maintain a web app.
Here's what a CAR with a WebRecorder WACZ file chunked to split the contents looks like: https://hackmd.io/aQnRiqVDSj2e2TZzsV8cKg?view
Each node is identified by part of it's hash and the number of bytes within it. Sadly this WARC doesn't have any deduplication in it, but larger datasets might!