Extract .SOUND files from the videogame The Witness

witness_lz4d.py

import struct
import sys
import os

MAX_OUTPUT_SIZE = 1 << 24 # 16 megabytes

def load_binary_file(fn, nbytes=100000000):
    data = []
    with open(fn, "rb") as src:
        byte = src.read(1)
        idx = 0
        while (len(byte) > 0) and (idx < nbytes):
            data.append(ord(byte))
            byte = src.read(1)
            idx += 1
    
    #print(data[:200])    
    return data

def getSoundFiles(mypath = '.'):
    return [f for f in os.listdir(mypath) if f.endswith('.sound')]

def main(srcfn):
    indata = load_binary_file(srcfn)
    inbytes = len(indata)
    outlen = struct.unpack('<I',bytearray(indata[8:12]))[0]
    if (outlen < 0) or (outlen > MAX_OUTPUT_SIZE):
        print("Outsize larger than max size {}; aborting. File might not be compressed.".format(MAX_OUTPUT_SIZE))
        return
    outdata = bytearray(outlen)
    inptr=12
    outptr=0
    printout=0
    while(outptr < outlen):
        copybytes    = (indata[inptr] & 0xf0) >> 4
        matchbytes = (indata[inptr] & 0x0f)
        inptr += 1
        if (copybytes == 15):
            while( True ):
                copybytes += indata[inptr]
                inptr += 1
                if (indata[inptr-1] != 0xff):
                    break

        #print("Copying %d bytes" % copybytes)
        if (copybytes > 0):
            outdata[outptr:outptr+copybytes] = indata[inptr:inptr+copybytes]
            inptr += copybytes
            outptr += copybytes

        if (inptr >= inbytes):
            break
        offset = indata[inptr] + (indata[inptr+1] << 8)
        inptr += 2
        if (matchbytes == 15):
            while( True ):
                matchbytes += indata[inptr]
                inptr += 1
                if (indata[inptr-1] != 0xff):
                    break
        matchbytes += 4
        #print("Offset is %d, copying %d match bytes" % (offset, matchbytes))
        
        if (matchbytes >= offset): #overlapping copy
            for i in range(matchbytes):
                outdata[outptr+i] = outdata[outptr-offset+i] # slow
        else:
                outdata[outptr:outptr+matchbytes] = outdata[outptr-offset:outptr-offset+matchbytes]
        outptr += matchbytes    

        if (outptr > printout):
            # print("%d/%d" % (outptr,outlen))
            printout += 256*1024

    # print("%d/%d" % (outptr,outlen))
    
    
    if bytearray(indata).find(b'Ogg') > -1:
        destfn = srcfn+'.ogg'
    else:
        destfn = srcfn+'.wav'
        
    with open(destfn,"wb") as outfile:
        startPos = bytearray(outdata).find(b'RIFF4')
        if startPos > -1:
            outfile.write(outdata[outdata.find(b'RIFF4'):-15])
        else:
            outfile.write(bytearray(indata[16:-15]))

if __name__ == '__main__':
    if (len(sys.argv) < 2):
        #print("need input and output file names")
        count = 0
        filesList = getSoundFiles()
        length = len(filesList)
        for s in filesList:
            count +=1
            main(s)
            print(str(count)+"/"+str(length)+"\t", s)
    else:
        main(sys.argv[1])

it works thanks!

Oh my God, thank you so much!