csvdiff

What it does

• Two million rows, 131 MB, in 2.8 seconds, holding about 60 bytes per row of the old file.
• Treats 1000, 1000.00 and 1.000,00 as one value, and empty, NULL, NaN and \N as one nothing - while keeping the cents in money.
• Calls out broken rows and duplicate keys instead of quietly diffing them.
• Exit code 1 when anything differs, so it works as a gate.

How it is proved

The benchmark is a script in the repository: it builds both exports from a fixed seed, so the timing repeats on any machine.

Rust13 testsprebuilt binaries for Linux, macOS and Windows

People find this looking for

compare two CSV files by key, data migration verification, CSV diff tool large files.

If that is your week and you would rather someone else did it, write to hello@dkautomation.dev or open an issue.