SIM/SID Extractor

import os
import sys
import struct
import io
import zipfile

import vdf
from Crypto.Cipher import AES

def readNullTerminatedString(f,offset):
    prevOffset=f.tell()
    f.seek(offset)

    result=b""
    while 1:
        byte=f.read(1)
        if byte==b"\x00": break
        result+=byte

    f.seek(prevOffset)
    return result.decode()

volHandles=dict()

class Volume:
    def __init__(self,volName):
        self.f=open(volName,"rb")
        self.size=os.fstat(self.f.fileno()).st_size

def openVolume(baseName,disc,vol):
    volName="%s_disk%d_%d.sid" % (baseName,disc,vol)

    if vol not in volHandles:
        while not os.path.isfile(volName):
            # Changing disc - close all volumes.
            closeVolumes()
            input("Need %s, insert disc %d and press Enter..." % (os.path.basename(volName),disc))

        volHandles[vol]=Volume(volName)

    return volHandles[vol].f

def openNextVolume(baseName,disc,vol):
    vol+=1
    volName="%s_disk%d_%d.sid" % (baseName,disc,vol)
    if not os.path.isfile(volName):
        # Change disc.
        disc+=1
        vol=0

    f=openVolume(baseName,disc,vol)
    f.seek(0x00)
    return f,disc,vol

def getVolumeSize(vol):
    return volHandles[vol].size

def closeVolumes():
    for vol in volHandles.values():
        vol.f.close()
    volHandles.clear()

def extractDepot(baseName,curDisc,targetDepot,keystring,outDir):
    if keystring:
        key=bytes.fromhex(keystring+"A8194D02193CD03792937D27590AECBD")
    else:
        key=None

    f=open("%s_disk%d.sim" % (baseName,curDisc),"rb")
    magic,version,discs,stringsSize=struct.unpack("<IIII",f.read(0x10))
    if magic!=0x3FD04C1F: raise Exception("Bad header magic in %s" % (f.name))
    if version!=1: raise Exception("Unknown version %d in %s" % (version,f.name))

    s=io.BytesIO(f.read(stringsSize))
    tableSize,numFiles=struct.unpack("<II",f.read(0x08))
    t=io.BytesIO(f.read(tableSize))
    f.close()

    for i in range(numFiles):
        nameOffset,pathOffset,depot,offset,size,disc,vol,isEnc,pad=struct.unpack("<IIIQQBBBB",t.read(0x20))
        if depot!=targetDepot:
            continue

        name=readNullTerminatedString(s,nameOffset)
        path=readNullTerminatedString(s,pathOffset)
        print(os.path.join(path,name))
        #print(t.tell()-0x20)

        if isEnc and not key:
            print("File is encrypted and no depot key was provided! Skipping...")
            continue

        f2=openVolume(baseName,disc,vol)
        f2.seek(offset)
        #print("%d %d 0x%08x" % (disc,vol,offset))

        os.makedirs(os.path.join(outDir,path),exist_ok=True)
        fname=os.path.join(outDir,path,name)
        out=open(fname,"wb")

        bytesWritten=0
        while bytesWritten!=size:
            if f2.tell()==getVolumeSize(vol):
                # Reached EOF, open next volume.
                f2,disc,vol=openNextVolume(baseName,disc,vol)

            chunkZSize,chunkSize=struct.unpack("<II",f2.read(0x08))
            padSize=(chunkZSize>>24)&0xFF
            chunkZSize&=0x00FFFFFF
            flags=(chunkSize>>24)&0xFF
            chunkSize&=0x00FFFFFF

            data=f2.read(chunkZSize)
            bytesWritten+=chunkSize

            if flags&0x01:
                if not key:
                    print("Encountered encrypted chunk and no depot key was provided! Skipping the file...")
                    out.close()
                    os.remove(fname)
                    break

                decSize=chunkZSize-padSize

                # Decrypt IV.
                cipher=AES.new(key,AES.MODE_ECB)
                iv=cipher.decrypt(data[:0x10])

                # Decrypt data.
                cipher=AES.new(key,AES.MODE_CBC,iv=iv)
                data=cipher.decrypt(data[0x10:])[:decSize]

            if flags&0x02:
                zipf=zipfile.ZipFile(io.BytesIO(data))
                out.write(zipf.read("zip"))
                zipf.close()
            else:
                out.write(data)

        out.close()

    closeVolumes()

if __name__=="__main__":
    if len(sys.argv)<4:
        print("Usage: sim_extract.py <simpath> <depotid> <outpath>")
        sys.exit(0)

    fname=sys.argv[1]
    depotStr=sys.argv[2]
    outDir=sys.argv[3]

    if fname[-4:]!=".sim":
        raise Exception("Not a SIM file")

    kv=vdf.load(open("legacydepotdata.vdf","r"))
    if depotStr in kv["depots"]:
        keystring=kv["depots"][depotStr]
        print("Depot key: %s" % keystring)
    else:
        print("No depot key for depot %s in VDF, assuming the depot is unencrypted." % depotStr)
        keystring=""

    pos=fname.find("_disk")
    if pos==-1:
        raise Exception("Bad SIM name")

    baseName=fname[:pos]
    disc=int(fname[pos+5:-4])
    depot=int(depotStr)
    extractDepot(baseName,disc,depot,keystring,outDir)