Скрипт поиска одинаковых файлов (по sha256)

find_dups.py

#!/usr/bin/env python3

import sys
import os
import stat
import hashlib


def index_dir(d):
    s = os.stat(d)
    if stat.S_ISREG(s.st_mode):
        yield d
    elif stat.S_ISDIR(s.st_mode):
        for i in os.listdir(d):
            yield from index_dir(os.path.join(d, i))


def hash_file(i):
    h = hashlib.sha256()
    with open(i, "rb") as f:
        h.update(f.read(2 * 1024 * 1024))

    return h.digest()


def get_key(h, v):
    for k, i in h.items():
        if i == v:
            return k


def main():
    index = []
    for i in sys.argv[1:]:
        index.extend(index_dir(i))

    index = {i: hash_file(i) for i in index}

    while index:
        k, v = index.popitem()
        if v in index.values():
            print(os.path.relpath(k, '.'), os.path.relpath(get_key(index, v), '.'))


if __name__ == '__main__':
    main()