find-duplicates finds duplicate files quickly based on the xxHashes of their contents.

$ go install github.com/twpayne/find-duplicates@latest

$ find-duplicates
{
  "cdb8979062cbdf9c169563ccc54704f0": [
    ".git/refs/remotes/origin/main",
    ".git/refs/heads/main",
    ".git/ORIG_HEAD"
  ]
}

find-duplicates [options] [paths...]

paths are directories to walk recursively. If no paths are given then the current directory is walked.

The output is a JSON object with properties for each observed xxHash and values arrays of filenames with contents with that xxHash.

Options are:

--keep-going or -k keep going after errors.

--output=<file> or -o <file> write output to <file>, default is stdout.

--threshold=<int> or -t <int> sets the minimum number of files with the same content to be considered duplicates. The default is 2.

--statistics or -s prints statistics to stderr.

find-duplicates aims to be as fast as possible by doing as little work as possible, using each CPU core efficiently, and using all the CPU cores on your machine.

It consists of multiple components:

1. Firstly, it walks the the filesystem concurrently, spawning one goroutine per subdirectory.
2. Secondly, with the observation that files can only be duplicates if they are the same size, it only reads file contents once it has found at more than one file with the same size. This significantly reduces both the number of syscalls and the amount of data read. Furthermore, as the shortest possible runtime is the time taken to read the largest file, larger files are read earlier.
3. Thirdly, files contents are hashed with a fast, non-cryptographic hash.

All components run concurrently.

• "Finding duplicate files unbelievably fast: a small CLI project using Go's concurrency" talk from Zürich Gophers.

MIT