Remove Metadata from ZIP Files

stripzip.py

# Based on https://github.com/thypon/python-stripzip
import mmap
import os
import sys
from struct import Struct

# Remove zip metadata.
def zip_strip(zip):
    # Remove extra data.
    def remove_extra(mm, offset, length, compressed_size = 0):
        extra_header_struct = Struct('<HH')
        zip64_extra_struct = Struct('<HHQQ')
        mlen = offset + length

        while offset < mlen:
            values = list(extra_header_struct.unpack_from(mm, offset))
            _, header_length = values
            extra_struct = Struct('<HH' + 'B'*header_length)
            values = list(extra_struct.unpack_from(mm, offset))
            header_id, header_length, rest = values[0], values[1], values[2:]

            # Clear metadata.
            if header_id in (0x5455, 0x7875, 0x0a, 0x0A):
                values[0] = 0x0000
                for i in range(2, len(values)):
                    values[i] = 0x00
                extra_struct.pack_into(mm, offset, *values)
            if header_id == 0x0001:
                values = list(zip64_extra_struct.unpack_from(mm, offset))
                header_id, header_length, uncompressed_size, compressed_size = values

            offset += extra_header_struct.size + header_length

        return compressed_size

    FILE_HEADER_SIGNATURE = 0x04034b50
    CENDIR_HEADER_SIGNATURE = 0x02014b50

    signature_struct = Struct('<L')
    local_file_header_struct = Struct('<LHHHHHLLLHH')
    central_directory_header_struct = Struct('<LHHHHHHLLLHHHHHLL')
    archive_size = os.fstat(zip.fileno()).st_size
    mm = mmap.mmap(zip.fileno(), 0)
    offset = 0

    while offset < archive_size:
        if signature_struct.unpack_from(mm, offset) != (FILE_HEADER_SIGNATURE,):
            break
        values = list(local_file_header_struct.unpack_from(mm, offset))
        _, _, _, _, _, _, _, compressed_size, _, name_length, extra_field_length = values
        # Reset date and time values.
        values[4] = 0
        values[5] = 0
        local_file_header_struct.pack_into(mm, offset, *values)
        offset += local_file_header_struct.size + name_length
        if extra_field_length != 0:
            compressed_size = remove_extra(
                mm, offset, extra_field_length, compressed_size)
        offset += compressed_size + extra_field_length

    while offset < archive_size:
        if signature_struct.unpack_from(mm, offset) != (CENDIR_HEADER_SIGNATURE,):
            break
        values = list(central_directory_header_struct.unpack_from(mm, offset))
        _, _, _, _, _, _, _, _, _, _, file_name_length, extra_field_length, file_comment_length, _, _, _, _ = values
        # Reset date and time values.
        values[5] = 0
        values[6] = 0
        central_directory_header_struct.pack_into(mm, offset, *values)
        offset += central_directory_header_struct.size + \
            file_name_length + extra_field_length + file_comment_length
        if extra_field_length != 0:
            remove_extra(mm, offset-extra_field_length, extra_field_length)

    # Abort if file is invalid.
    if offset == 0:
        print('Invalid zip file.')
        sys.exit(1)

def main():
    if len(sys.argv) < 2:
        print('Select a zip file.')
        return

    # Process zip file.
    zip = open(sys.argv[1], 'r+b')
    zip_strip(zip)
    zip.close()

# Run script.
main()