SARCExtract.py v0.1 by NWPlayer123

import os, sys, struct

def hexstr(data, length): #Convert input into hex string
    return hex(data).lstrip("0x").rstrip("L").zfill(length).upper()
def binr(byte):
    return bin(byte).lstrip("0b").zfill(8)
def uint8(data, pos):
    return struct.unpack(">B", data[pos:pos + 1])[0]
def uint16(data, pos):
    return struct.unpack(">H", data[pos:pos + 2])[0]
def uint24(data, pos):
    return struct.unpack(">I", "\00" + data[pos:pos + 3])[0] #HAX
def uint32(data, pos):
    return struct.unpack(">I", data[pos:pos + 4])[0]
def check(length, size, percent):
    length = float(length);size = float(size)
    test = round(length / size, 2) #Percent complete as decimal
    test = test * 100 #Percent
    if test % 5 == 0: #Count up by 5s so we don't spam
        if percent != test: #New Number
            print(str(test)[:-2] + "%")
            percent = test
    return percent
def calchash(name, multiplier):
    result = 0
    for x in xrange(len(name)):
        result = ord(name[x]) + result * multiplier
    return result
def getstr(data):
    x = data.find("\x00")
    if x != -1:
        return data[:x]
    else:
        return data

class Yaz0(object):
    def decompress(self, data):
        print("Reading Yaz0....")
        pos = 0
        magic = data[pos:pos + 4];pos += 4
        if magic != "Yaz0":
            print("Not a Yaz0 file")
            sys.exit(1)
        size = uint32(data, pos);pos += 4 #Uncompressed filesize
        unk  = uint32(data, pos);pos += 4 #Unknown (0x80 ???)
        pad  = uint32(data, pos);pos += 4 #Padding
        out = [];dstpos = 0
        print("Decompressing Yaz0....")
        percent = 0
        while len(out) < size: #Read Entire File
            code = uint8(data, pos);code = binr(code);pos += 1
            for x in xrange(8):
                percent = check(len(out), size, percent)
                if len(out) >= size: break
                if code[x] == "1":
                    out.append(uint8(data, pos));pos += 1
                if code[x] == "0":
                    rle = hexstr(uint16(data, pos), 4);pos += 2
                    if rle[0] == "0":
                        read = uint8(data, pos);read += 0x12;pos += 1
                    else:
                        read = int(rle[0], 16);read += 2
                    b = int(rle[1], 16)
                    bckpos = ((b << 8) | int(rle[2:], 16)) + 1
                    dstpos = len(out) - bckpos
                    for x in xrange(read):
                        try:
                            out.append(out[dstpos + x])
                        except:
                            print(pos)
        SARChive = SARC()
        SARChive.extract(out, 1)

class SARC(object):
    def extract(self, data, mode):
        print("Reading SARC....")
        pos = 0
        if mode == 1:
            magic = ""
            for x in xrange(4):
                magic += chr(data[x]);pos += 1
        if mode == 0:
            magic = data[pos:pos + 4];pos += 4
        name, ext = os.path.splitext(sys.argv[1])
        if magic != "SARC":
            if mode == 1: #Sent here by Yaz0
                print("Not a SARC Archive!")
                print("Writing Decompressed File....")
                f = open(name + ".bin", "wb")
                for x in xrange(len(data)):
                    f.write(struct.pack(">B", data[x]))
                sys.exit(1)
            if mode == 0:
                print("Not a SARC Archive!")
                sys.exit(1)
        if mode == 1:
            data2 = data;data = ""
            for x in xrange(len(data2)):
                data += chr(data2[x])
        hdr = uint16(data, pos);pos += 2 #Length of header, always 0x14
        order = uint16(data, pos);pos += 2 #Byte Order Mark
        if order == 65279: #0xFEFF - Big Endian
            pass
        else:
            print("Little endian not supported!")
            sys.exit(1)
        size = uint32(data, pos);pos += 4 #Size of the entire file
        doff = uint32(data, pos);pos += 4 #Start of data section
        unk  = uint32(data, pos);pos += 4 #???
        #---------------------------------------------------------------
        magic2 = data[pos:pos + 4];pos += 4
        if magic2 != "SFAT": #File Attribute Table
            print("Unknown Data Section: " + magic2)
            sys.exit(1)
        hdr2 = uint16(data, pos);pos += 2 #Length of header, always 0xC
        nodec = uint16(data, pos);pos += 2 #Node Count
        hashr = uint32(data, pos);pos += 4 #Hash Multiplier ??? - Always 0x65
        nodes = [];percent = 0
        print("Reading File Attribute Table...")
        for x in xrange(nodec):
            percent = check(x, nodec, percent)
            hash = data[pos:pos + 4];pos += 4
            hash = calchash(hash, hashr)
            unk  = uint8(data, pos);pos += 1
            off  = uint24(data, pos);pos += 3 #Name Offset
            srt  = uint32(data, pos);pos += 4 #File Offset Start
            end  = uint32(data, pos);pos += 4 #File Offset End
            nodes.append([hash, unk, off, srt, end])
        #---------------------------------------------------------------
        magic3 = data[pos:pos + 4];pos += 4
        if magic3 != "SFNT": #File Name Table
            print("Unknown Data Section: " + magic3)
            sys.exit(1)
        hdr3 = uint16(data, pos);pos += 2
        unk2 = uint16(data, pos);pos += 2
        strings = [];percent = 0
        print("Reading Filenames....")
        for x in xrange(nodec):
            percent = check(x, nodec, percent)
            string = getstr(data[pos:]);pos += len(string)
            while uint8(data, pos) == 0: pos += 1 #Move to the next string
            strings.append(string)
        #---------------------------------------------------------------
        print("Writing Files....")
        try:
            os.mkdir(name)
        except:
            print("Folder already exists, continuing....")
        for x in xrange(nodec):
            print("File " + str(x) + " of " + str(nodec))
            f = open(name + "/" + strings[x], "wb")
            f.write(data[nodes[x][3] + doff:nodes[x][4] + doff])
            f.close()

def main():
    print("SARCExtract v0.1 by NWPlayer123")
    f = open(sys.argv[1], "rb")
    data = f.read()
    f.close()
    magic = data[0:4]
    if magic == "Yaz0":
        SARChive = Yaz0()
        SARChive.decompress(data)
    elif magic == "SARC":
        SARChive = SARC()
        SARChive.extract(data, 0)
    else:
        print("Unknown File Format!")
        sys.exit(1)

if __name__ == "__main__":
    main()