twrp_extract_files.py

#!/usr/bin/env python3

BACKUP_FILE = 'backup.ab'

# Constants
# ref: https://github.com/omnirom/android_bootable_recovery/blob/android-7.1/adbbu/twadbstream.h
TWRP = b'TWRP' + b'\x00\x00\x00\x00'

TWSTREAMHDR = b'twstreamheader'
TWFN = b'twfilename'
TWIMG = b'twimage'
TWDATA = b'twdatablock'
MD5TRAILER = b'md5trailer'
TWENDADB = b'twendadb'


class ParseError(Exception):
    pass


with open(BACKUP_FILE, 'rb') as backup_file:
    initial = backup_file.read(10240)  # size picked randomly
    backup_file.seek(initial.find(TWRP))

    while True:
        block_header = backup_file.read(8)
        if block_header != TWRP:
            pos = backup_file.tell() - 8
            raise ParseError(
                f'Expected TWRP header, got: {block_header} ({pos})'
            )
            
        block_type = backup_file.read(16).rstrip(b'\x00')
        
        if block_type in (TWSTREAMHDR, TWENDADB):
            # AdbBackupControlType
            block_crc = backup_file.read(4)
            backup_file.seek(484, 1)

            if block_type == TWENDADB:
                # We should have reached the end of the backup file
                break

        elif block_type in (TWFN, TWIMG):
            # twfilehdr
            block_size = backup_file.read(8)
            block_compressed = backup_file.read(8)
            block_crc = backup_file.read(4)

            name = backup_file.read(468).rstrip(b'\x00')
            file_name = name.split(b'/')[-1].decode()
            output = open(file_name, 'wb')

            print('Extracting', block_type.decode(), name.decode())

        elif block_type in TWDATA:
            # AdbBackupControlType + data
            block_crc = backup_file.read(4)
            backup_file.seek(484, 1)

            # Write out data in block
            output.write(backup_file.read(1048064))  # (1024 * 1024) - 512

        elif block_type == MD5TRAILER:
            # AdbBackupFileTrailer
            block_crc = backup_file.read(4)
            block_ident = backup_file.read(4)
            block_md5 = backup_file.read(40)
            backup_file.seek(440, 1)

            output.close()

        else:
            pos = backup_file.tell() - 24
            raise ParseError(
                f"Unknown block type: '{block_type}' ({pos})"
            )