findbytes.py

# If you see very long/bad variable names and questionable design choice, probably because I wrote this in sleep deprivation

import sys

def findBytesInTarget(findingBytes, target):
    foundList = []
    findingOffestInTarget = 0
    while True:
        offset = target.find(findingBytes, findingOffestInTarget)
        if offset is -1:
            break
        foundList.append(offset)
        if len(foundList) > 1:
            break

        findingOffestInTarget = offset + 1
    
    return foundList

#return offset if found once, -1 if not found, (-1 - occurrence count) if found more than once
def findOffsetInTargetWithLen(source, target, offsetInSource, byteLen):
    foundList = []
    for shift in range(0, byteLen, 4):
        findingOffestAfterShift = offsetInSource - shift
        findingBytes = source[findingOffestAfterShift:findingOffestAfterShift+byteLen]

        foundList = findBytesInTarget(findingBytes, target)

        if len(foundList) is 1:
            ret = findBytesInTarget(findingBytes, source)
            if len(ret) is 1:
                return foundList[0] + shift
            else:
                return -2

    return -1 - len(foundList)

STARTING_FIND_LEN = 16
MAX_INCREASE = 0x40
# Return [offset, len searched], -1 if not found
def findOffestInTarget(source, target, offsetInSource):
    MIN_FIND_LEN = 8

    findLenIncrease = 0
    findLenUpperStop = False
    while True:
        findLenUpper = STARTING_FIND_LEN + findLenIncrease + 4

        if not findLenUpperStop:
            ret = findOffsetInTargetWithLen(source, target, offsetInSource, findLenUpper)
        if ret >= 0:
            return [ret, findLenUpper]
        if ret is -1:
            findLenUpperStop = True
        if findLenUpperStop is True:
            break

        findLenIncrease = findLenIncrease + 4
        
    findLenIncrease = 0
    findLenLowerStop = False
    while True:
        findLenLower = STARTING_FIND_LEN - findLenIncrease
        if findLenLower < MIN_FIND_LEN:
            break

        if not findLenLowerStop:
            ret = findOffsetInTargetWithLen(source, target, offsetInSource, findLenLower)
        if ret >= 0:
            return [ret, findLenLower]
        if ret < -2:
            findLenLowerStop = True
        if findLenLowerStop is True:
            break

        findLenIncrease = findLenIncrease + 4
    
    return [-1, -1]

def main(argc, argv):
    if argc < 4:
        print('Too few args')
        print('Usage: %s [old_file] [new_file] [text_file] [(optional) start line #] [(optional) end line #]' % argv[0])
        print('   or: %s [old_file] [new_file] [offset from old_file]' % argv[0])
        return

    fileByteFrom = argv[1]
    fileFindOffstIn = argv[2]

    lineNumberStart = 1
    if argc >= 5:
        lineNumberStart = int(argv[4], 10)
    lineNumberEnd = -1
    if argc >= 6:
        lineNumberEnd = int(argv[5], 10)

    offsetList = []
    try:
        offset = int(argv[3], 16)
        offsetList.append(argv[3])
    except ValueError:
        with open(argv[3], 'r') as f:
            offsetList = f.readlines()

    if lineNumberEnd < 0 or lineNumberEnd > len(offsetList):
        lineNumberEnd = len(offsetList)

    with open(fileByteFrom, 'rb') as f:
        sourceFile = f.read()
    with open(fileFindOffstIn, 'rb') as f:
        targetFile = f.read()

    for lineIndex in range(lineNumberStart-1, lineNumberEnd):
        offset = -1
        try:
            offset = int(offsetList[lineIndex][:8], 16)

            ret = 0
            increasing = 0
            finding_offset = offset + increasing
            while increasing <= MAX_INCREASE and finding_offset + STARTING_FIND_LEN < len(sourceFile):

                ret = findOffestInTarget(sourceFile, targetFile, finding_offset)
                if ret[0] >= increasing:
                    ret[0] = ret[0] - increasing
                    break

                increasing = increasing + 4
                finding_offset = offset + increasing

            print('{:08X}'.format(ret[0]), end='')
            print(offsetList[lineIndex][8:-1], end='')
            if ret[0] is -1:
                print('\nWARNING: ^ failed to find matching bytes for offset, please freak out')
            else:
                if increasing > 0:
                    print(' // Found match at {:08X}+{:X} ({:08X})'.format(ret[0], increasing, ret[0]+increasing), end='')
                print(' // Confidence level: {}'.format(ret[1]))
            
        except ValueError:
            print(offsetList[lineIndex], end='')

if __name__ == '__main__':
    main(len(sys.argv), sys.argv)