fastdd.py

#!/usr/bin/python3

# Copyright (c) 2024 Nicolás Alvarez
# SPDX-License-Identifier: MIT

# Usage: fastdd.py <source> <dest>
#
# Will copy blocks from source to destination,
# unless the destination block already has the correct content.

import sys
import os
from tqdm import tqdm

BLOCK_SIZE=1024*4096
data_written = 0
data_saved = 0

size = os.path.getsize(sys.argv[1])
#size = 7969177600
pbar = tqdm(total=size, unit="B", unit_scale=True)

pos = 0
with open(sys.argv[1], "rb") as inf, open(sys.argv[2], "r+b") as outf, pbar:
    while True:
        outf.seek(pos)
        block = inf.read(BLOCK_SIZE)
        if len(block) == 0: break

        if outf.read(BLOCK_SIZE) != block:
            outf.seek(pos)
            outf.write(block)
            data_written += len(block)
        else:
            data_saved += len(block)
        pbar.update(len(block))
        pos += len(block)

print(f"Wrote {data_written/1024/1024:.1f}MB")
print(f"Avoided writing {data_saved/1024/1024:.1f}MB")