hashmap.py

#!/usr/bin/python3

import sys
import hashlib

BLOCKSIZE=0x200
f1d = {}
f2h = []

with open(sys.argv[1], 'rb') as f:
    f1 = bytearray(f.read())
for i in range(0, len(f1), BLOCKSIZE):
    s = f1d.setdefault(hashlib.sha1(f1[i:i+BLOCKSIZE]).hexdigest(), [])
    s.append(i)
del f1

with open(sys.argv[2], 'rb') as f:
    f2 = bytearray(f.read())
for i in range(0, len(f2), BLOCKSIZE):
    f2h.append(hashlib.sha1(f2[i:i+BLOCKSIZE]).hexdigest())
del f2

def getoffs(c):
    h = f2h[c]
    if h == "5c3eb80066420002bc3dcc7ca4ab6efad7ed4ae5":
        return "zeroes"
    if h == "9d0ac37bb3ec8c95990fd37a962a17a95ce97aa0":
        return "0xFF"
    if h == "3da96bc604d8819346a5ca15bb73d38f891cd08a":
        return "NOR 0x10000 repeating"
    offs = f1d.get(h, None)
    if offs is None:
        return "not in NOR"
    return offs

def dospecialrun(code, c):
    run = 1
    while True:
        offs = getoffs(c+run)
        if offs == code:
            run += 1
        else:
            break
    return run

def runstreak(c):
    offs = getoffs(c)
    if offs == 0xDEAD0000:
        return 1
    run = 1
    if type(offs) != list:
        run = dospecialrun(offs, c)
        if run > 1:
            print("ROM [0x{:X}, 0x{:X}) => {}".format(c*BLOCKSIZE, (c + run)*BLOCKSIZE, offs))
        else:
            print("ROM 0x{:X} => {}".format(c*BLOCKSIZE, offs))
        return run
    l = len(offs)
    if l > 1:
        print("ROM 0x{:X} => NOR {}".format(c*BLOCKSIZE, ", ".join(map(lambda x: "0x{:X}".format(x), offs))));
        return 1
    if l == 1:
        startnor, lastnor = offs[0], offs[0]
        while True:
            oln = lastnor
            if c+run >= len(f2h):
                break
            offs = f1d.get(f2h[c+run], None)
            if offs is None:
                break
            l = len(offs)
            if l == 0:
                break
            if l > 1:
                if lastnor+BLOCKSIZE in offs:
                    lastnor += BLOCKSIZE
                    run += 1
                    continue
                else:
                    break
            if l == 1:
                lastnor = offs[0]
                if lastnor - oln == BLOCKSIZE:
                    run += 1
                    continue
                else:
                    break
        if run == 1:
            print("ROM 0x{:X} => NOR 0x{:X}".format(c*BLOCKSIZE, startnor));
        else:
            print("ROM [0x{:X}, 0x{:X}) => NOR [0x{:X}, 0x{:X})".format(c*BLOCKSIZE, (c + run)*BLOCKSIZE, startnor, oln+BLOCKSIZE))
        return run

l = len(f2h)
c = 0
while c < l:
    c += runstreak(c)