Detroit Sound Extractor

import os
import struct
import io

# Parameters
idxFile=r"D:\Games\Steam\steamapps\common\Detroit Become Human\BigFile_PC.idx"
bnkDir=r"E:\GameRips\detroit\pc"
wemDir=r"E:\GameRips\detroit\pc\wem"

extractBnk=True
extractWem=False



# Script starts here
idx=dict()
datHandles=dict()
bigFileName=None

class FileEntry:
    def __init__(self,f):
        data=struct.unpack(">7I",f.read(0x1c))
        self.typ=data[0]
        self.count=data[1]
        self.id=data[2]
        self.offset=data[3]
        self.size=data[4]
        self.extraSize=data[5]
        self.datNum=data[6]

def parseIdx(fname):
    name,ext=os.path.splitext(fname)
    global bigFileName
    bigFileName=name

    f=open(fname,"rb")
    f.seek(0x65)
    numFiles=struct.unpack(">I",f.read(0x04))[0]
    for i in range(numFiles):
        entry=FileEntry(f)
        idx[entry.typ,entry.id]=entry
    f.close()

def loadFile(ref):
    entry=idx[ref]
    if entry.datNum in datHandles:
        f=datHandles[entry.datNum]
    else:
        name,ext=os.path.splitext(bigFileName)
        newExt=".dat" if entry.datNum==0 else ".d%02d" % (entry.datNum)
        datName=name+newExt
        f=open(datName,"rb")
        datHandles[entry.datNum]=f

    f.seek(entry.offset)
    data=f.read(entry.size)
    return data

def readHeader(f,expectedMagic=None):
    magic,ver=struct.unpack("<8sI",f.read(0x0c))
    if expectedMagic and magic!=expectedMagic:
        raise Exception("Bad header magic at 0x%08x! Expected %s, got %s!" %
                        (f.tell()-0x0c,expectedMagic,magic))

def readString(f):
    size=struct.unpack("<I",f.read(0x04))[0]
    return f.read(size).decode()

def readQZip(f):
    magic,flag=struct.unpack("<4sB",f.read(0x05))
    if magic!=b"QZIP":
        raise Exception("Bad QZIP magic at 0x%08x!" % (f.tell()-0x05))

def readComCont(f,dc=False):
    readHeader(f,b"COM_CONT")
    size,numRefs=struct.unpack("<II",f.read(0x08))
    refs=list()
    for i in range(numRefs):
        typ,id,flag=struct.unpack("<IIB",f.read(0x09))
        refs.append((typ,id))

    # Legacy thing from Fahrenheit and early Heavy Rain, always empty.
    readHeader(f,b"LOADCONT")
    if not dc:
        size=struct.unpack("<I",f.read(0x04))[0]
        f.seek(size,1)

    return refs

def closeDats():
    for f in datHandles.values():
        f.close()
    datHandles.clear()

def extractSoundBanks(outpath):
    print("Extracting BNK...")
    os.makedirs(outpath,exist_ok=True)

    for ref in idx:
        if ref[0]!=1022:
            continue

        #print("%s: %s" % (ref[0],ref[1]))
        f=io.BytesIO(loadFile(ref))

        comRefs=readComCont(f)
        readHeader(f,b"CSNDBNK_")
        size=struct.unpack("<I",f.read(0x04))[0]
        bnkId=readString(f)
        bnkName=readString(f)
        print("%s: %s" % (bnkId,bnkName))

        f2=io.BytesIO(loadFile(comRefs[0]))

        # HACK: There's no offsets table for assets inside Data Container
        # so we just look for CSNDBKDT string.
        pos=f2.getvalue().find(b"CSNDBKDT")
        if pos==-1:
            print("No BNK data for %s!" % (bnkName))
            continue

        f2.seek(pos)
        readHeader(f2)
        size=struct.unpack("<I",f2.read(0x04))[0]
        bnkData=f2.read(size)
        out=open(os.path.join(outpath,bnkName+".bnk"),"wb")
        out.write(bnkData)
        out.close()

    closeDats()

def extractSoundData(outpath):
    print("Extracting WEM...")
    os.makedirs(outpath,exist_ok=True)

    for ref in idx:
        if ref[0]!=1033:
            continue

        f=io.BytesIO(loadFile(ref))
        readQZip(f)
        readHeader(f,b"CSNDDATA")
        size=struct.unpack("<I",f.read(0x04))[0]
        unkStr=readString(f)
        unkNum=struct.unpack("<I",f.read(0x04))[0]
        wemName=readString(f)
        wemSize=struct.unpack("<I",f.read(0x04))[0]
        wemData=f.read(wemSize)
        #print(wemName)
        out=open(os.path.join(outpath,wemName+".wem"),"wb")
        out.write(wemData)
        out.close()

    closeDats()

if __name__=="__main__":
    parseIdx(idxFile)

    if extractBnk:
        extractSoundBanks(bnkDir)

    if extractWem:
        extractSoundData(wemDir)