Get a consensus agreed rebuild of a firmware dump from multiple noisy dumps

clean_firmware.py

import argparse
import numpy as np
from collections import Counter
# atucom
# example: $ python3  clean_firmware.py --threshold=0.8 firmware-dump/full_backup*.bin

def load_dumps(paths):
    dumps = [np.fromfile(path, dtype=np.uint8) for path in paths]
    lengths = [len(d) for d in dumps]
    if len(set(lengths)) != 1:
        raise ValueError("All dumps must be the same length")
    return np.vstack(dumps)

def majority_vote(data, threshold=0.5):
    clean = []
    no_consensus = []

    num_dumps, num_bytes = data.shape
    required_votes = int(threshold * num_dumps)

    for i in range(num_bytes):
        column = data[:, i]
        counts = Counter(column)
        byte, freq = counts.most_common(1)[0]
        if freq > required_votes:
            clean.append(byte)
        else:
            clean.append(0xFF)  # placeholder for unresolved
            no_consensus.append((i, dict(counts)))

    return np.array(clean, dtype=np.uint8), no_consensus

def write_output(clean, no_consensus, out_file, log_file):
    clean.tofile(out_file)
    with open(log_file, 'w') as f:
        for offset, variants in no_consensus:
            variants_str = ', '.join(f"{k:#04x}:{v}" for k, v in variants.items())
            f.write(f"Offset {offset:#08x}: No consensus [{variants_str}]\n")

def main():
    parser = argparse.ArgumentParser(description="Clean firmware from multiple noisy dumps.")
    parser.add_argument('dumps', nargs='+', help="Paths to firmware dump files")
    parser.add_argument('--out', default='reconstructed.bin', help="Output cleaned firmware")
    parser.add_argument('--log', default='no_consensus.log', help="Log file for disagreements")
    parser.add_argument('--threshold', type=float, default=0.5, help="Consensus threshold (0-1)")
    args = parser.parse_args()

    data = load_dumps(args.dumps)
    clean, no_consensus = majority_vote(data, threshold=args.threshold)
    write_output(clean, no_consensus, args.out, args.log)

    print(f"✅ Cleaned firmware saved to {args.out}")
    if no_consensus:
        print(f"⚠️  {len(no_consensus)} bytes had no consensus. See {args.log} for details.")
    else:
        print("🎉 Full consensus across all bytes.")

if __name__ == '__main__':
    main()