xboxscrape.py

#!/usr/bin/env python

'''
    Tool to read game media type header information from the original
    XBOX's xbe executables.

    In retrospect, it would have been much less fiddly to write this in C/C++

    Info taken from here:
    http://www.caustik.com/cxbx/download/xbe.htm
'''

import argparse
import struct
import codecs
import sys
import os

CERT_LOCATION = 0x0118
BASE_LOCATION = 0x0104

CERT_TITLE_LOCATION = 0x000C
CERT_TITLE_SIZE = 0x0050
CERT_MEDIA_LOCATION = 0x009c
CERT_MEDIA_FLAGS = {
    'hd':          0x00000001,
    'dvd_x2':      0x00000002,
    'dvd_cd':      0x00000004,
    'cd':          0x00000008,
    'dvd_5_ro':    0x00000010,
    'dvd_9_ro':    0x00000020,
    'dvd_5_rw':    0x00000040,
    'dvd_9_rw':    0x00000080,
    'dongle':      0x00000100,
    'media_board': 0x00000200,
    'hd_insecure': 0x40000000,
    'insecure':    0x80000000,
}


def main():
    sys.stdout = codecs.getwriter('utf8')(sys.stdout)

    parser = argparse.ArgumentParser("XBOX file info dumper")
    parser.add_argument('filename', metavar='file',
                        help='Executable to read.', default='default.xbe')
    args = parser.parse_args()

    if not os.path.isfile(args.filename):
        print("Couldn't find '{0}'.".format(args.filename))
        exit(1)

    with file(args.filename, 'rb') as f:
        header_read(f)


def header_read(f):
    f.seek(CERT_LOCATION)
    media_offset, = struct.unpack('i', f.read(4))

    f.seek(BASE_LOCATION)
    base_offset, = struct.unpack('i', f.read(4))

    cert_offset = media_offset - base_offset

    title = title_read(f, cert_offset)

    o = u"Reading header for '{0}':".format(title)

    print(o)

    media_types = media_read(f, cert_offset)

    print("   Media Flags: {0}".format(', '.join(media_types)))


def title_read(f, cert_offset):
    f.seek(cert_offset + CERT_TITLE_LOCATION)

    data = f.read(CERT_TITLE_SIZE)
    text = "unknown"

    try:
        text = data.decode("UTF-16LE").rstrip('\0')
    except:
        pass

    return text


def media_read(f, cert_offset):
    f.seek(cert_offset + CERT_MEDIA_LOCATION)
    data = bytearray(f.read(4))
    field = struct.unpack('i', f.read(4))[0]

    items = []
    for k, v in CERT_MEDIA_FLAGS.items():
        if (v & field != 0):
            items.append(k)

    return items


if __name__ == "__main__":
    main()