convert_strings.py

#!/usr/bin/python
from __future__ import print_function, division

import struct


def read_int32(file):
    return struct.unpack("i", file.read(4))[0]


def read_string(file, length):
    return file.read(length)


def read_lengthstring(file):
    return read_string(file, read_int32(file))


def convert_valuestring(file, offset, expectedname):
    # We'll want to return back to the point in the file we started from.
    restore = file.tell()
    rval = False    # We use this to let the caller know if we did anything.
    file.seek(offset)   # This should take us right to the start of the object.
    file.read(4)    # Skip the total object length
    byte = file.read(1)
    if byte[0] != 1:
        print("Expected a byte of 0x01; got %r instead" % byte)
    length = file.read(1)[0]
    if length == 255:   # Length is >= 255; need to read a 4-byte integer.
        length = read_int32(file)
    expectedlength = length - 4
    secondlength = read_int32(file)
    if not secondlength == expectedlength:
        print("[name] secondlength/expectedlength mismatch: got %d, expected "
              "%d" % (secondlength, expectedlength))
    name = read_string(file, secondlength)
    if not name == expectedname:
        print("name/expectedname mismatch: got %r, expected %r" %
              (name, expectedname))
    if file.read(1)[0] == 3:
        # This string needs converting.
        file.seek(-1, 1)
        file.write(b"\x02")
        length = file.read(1)[0]
        if length == 255:
            length = read_int32(file)
        expectedlength = length - 4
        secondlength = read_int32(file)
        if not secondlength == expectedlength:
            print("[string] secondlength/expectedlength mismatch: got %d, "
                  "expected %d" % (secondlength, expectedlength))
        memory = file.tell()    # We'll come back to here to convert.
        length = read_int32(file)
        if length > secondlength:
            print("[string] Unexpectedly long third length: %d (compared to"
                  "%d)" % (length, secondlength))
        # We start getting characters now.
        chars = []
        length = secondlength - 4
        while length > 0:
            chars.append(file.read(1))
            length -= 1
            byte = file.read(1)
            if byte[0]:
                print("Unexpectedly non-null byte %r at %d bytes from the end "
                      "of the string" % (byte, length))
            length -= 1
        if length < 0:
            print("Length ended up at %d" % length)
        # print("DEBUG: detected string: %r" % b"".join(chars))
        byte = file.read(1)
        if not byte[0] == 6:
            print("Expected a 0x06 byte at the end of the string; found %r "
                  "instead." % byte)
        file.seek(memory)
        file.write(b"".join(chars))
        if not (chars and chars[-1] == 0):   # If we didn't just write a null
            file.write(b"\0")   # write one to make sure
        rval = True
    file.seek(restore)
    return rval


def find_all_valuestrings(file):
    total = 0
    filetype = read_lengthstring(file)
    if not filetype == b"Diversions Entertainment Objects File2\0":
        print("Filetype did not match expectations: %r" % filetype)
    byte = file.read(1)
    if not byte[0] == 2:
        print("Expected 0x02 byte; got %r instead" % byte)
    byteseq = file.read(4)
    if not byteseq == b"\0\0\0\0":
        print("Expected 4 null bytes, got %r instead" % byteseq)
    endheader = read_int32(file)    # Get the header length
    endheader += file.tell()    # I want to know when I've exceeded it.
    byte = file.read(1)
    if not byte[0] == 9:
        print("Expected 0x09 byte; got %r instead" % byte)
    length = file.read(1)[0]
    if length == 255:
        length = read_int32(file)
    byte = file.read(1)
    if not byte[0] == 4:
        print("[header] Expected 0x04 byte; got %r instead" % byte)
    length = file.read(1)[0]
    if length == 255:
        length = read_int32(file)
    file.seek(length, 1)    # Skip the rest of the header
    byte = file.read(1)
    if not byte[0] == 4:
        print("[predirectory] Expected 0x04 byte; got %r instead" % byte)
    byteseq = file.read(4)
    # if not byteseq == b"\0\0\0\0":
    #     print("DEBUG: Any idea what %r means?" % byteseq)
    byte = file.read(1)
    if not byte[0] == 1:
        print("[predirectory] Expected 0x01 byte; got %r instead" % byte)
    length = file.read(1)[0]
    if length == 255:
        length = read_int32(file)
    file.seek(length, 1)    # Skip the indecipherable blob
    file.read(4)    # Can't do anything with this number yet.
    while file.tell() < endheader:
        byte = file.read(1)
        if not byte[0] == 4:
            if file.tell() < endheader:     # We haven't left the directory yet
                print("[directory<] Expected 0x04 byte; got %r instead" % byte)
        else:
            if file.read(1)[0] == 255:  # Skip directory entry length
                read_int32(file)
            name = read_lengthstring(file)
            objtype = read_lengthstring(file)
            file.read(16)   # Skip the GUID
            byte = file.read(1)
            if byte[0]:
                print("[directory] Expected null byte; got %r instead" % byte)
            file.read(8)    # Can't do anything with this yet.
            byte = file.read(1)
            if not byte[0] == 4:
                print("[directory>] Expected 0x04 byte; got %r instead" % byte)
            offset = read_int32(file)
            if objtype == b"Value\\String\0":
                # print("DEBUG: Trying to convert string named %r, located at "
                #       "offset 0x%x" % (name, offset))
                if convert_valuestring(file, offset, name):
                    total += 1
            # else:
            #     print("DEBUG: Ignoring object of type %r" % (objtype))
    if not endheader == file.tell():
        print("endheader mismatch: %r, current offset %r" %
              (endheader, file.tell()))
    # print("DEBUG: These numbers should match: %r == %r" %
    #       (endheader, file.tell()), end="...and they ")
    # if endheader == file.tell():
    #     print("do.")
    # else:
    #     print("don't!!!")
    return total

if __name__ == "__main__":
    import sys
    if len(sys.argv) > 1:
        file_to_open = sys.argv[1]
    else:
        file_to_open = input("Filename:")
    with open(file_to_open, "r+b") as f:
        total = find_all_valuestrings(f)
    print("Converted %d string(s) to be v2172-compatible." % total)