extract_unallocated.py

#!/usr/bin/env python3
# (c) Maxim Suhanov
# 2020

import sys

if len(sys.argv) != 4:
	print('Usage: extract_unallocated.py <bitmap file> <flat image> <cluster size (in bytes)')
	sys.exit(0)

with open(sys.argv[1], 'rb') as bitmap:
	bitmap_buffer = bitmap.read()

volume = open(sys.argv[2], 'rb')
cluster_size = int(sys.argv[3])

if cluster_size <= 0 or cluster_size % 512 != 0:
	raise ValueError('Invalid cluster size')

sector_size = 512
sectors_per_cluster = cluster_size // sector_size

empty_sector = b'EMPTY           ' * 32
original_sector = b'ORIGINAL        ' * 32
null_sector = b'\x00' * 512

cnt_empty = 0
cnt_original = 0
cnt_valid = 0
cnt_null = 0

bit_index = 0
while True:
	try:
		current_bitmap_byte = bitmap_buffer[bit_index // 8]
	except IndexError:
		break

	current_bitmap_bit = (current_bitmap_byte >> (bit_index % 8)) & 1
	if current_bitmap_bit == 0:
		volume.seek(bit_index * cluster_size)

		for __ in range(sectors_per_cluster):
			unallocated_sector = volume.read(sector_size)
			truncated_read = len(unallocated_sector) != sector_size

			if unallocated_sector == empty_sector:
				cnt_empty += 1
			elif unallocated_sector == original_sector:
				cnt_original += 1
			else:
				cnt_valid += 1
				sys.stdout.buffer.write(unallocated_sector)

				if unallocated_sector == null_sector:
					cnt_null += 1

			if truncated_read:
				break

		if truncated_read:
			break

	bit_index += 1

volume.close()

print('Empty (null) sectors: {}, original sectors: {}, valid unallocated sectors: {}, valid sectors with nulls: {}\n'.format(cnt_empty, cnt_original, cnt_valid, cnt_null), file = sys.stderr)