BFSAR/BCSAR parser/extractor WIP

import sys
import string
from util import *


class strg(object):
    name = ""
    size = None
    offset = None
    def __init__(self, id):
        self.id = id

class audio(object):
    name = ""
    fileID = None
    strgID = None
    bankID = None
    def __init__(self, id):
        self.id = id

class soundSet(object):
    offset = None
    fileID = None
    strgID = None
    def __init__(self, id):
        self.id = id

class sarFile(object):
    fileName = ""
    path = ""
    offset = None
    ext = ".b"
    internal = False
    fileOffset = None
    fileInfoLength = None
    size = None
    bank = None
    def __init__(self, id):
        self.id = id

class bwav(object):

    def __init__(self, id):
        self.id = id

class warc(object):
    offset = None
    fileID = None
    sectionCount = None
    wavCount = None
    infoLength = None
    wavs = []
    name = ""
    folder = ""
    def __init__(self, id):
        self.id = id

class bank(object):
    offset = None
    fileID = None
    wavs = []
    seqs = []
    name = ""
    folder = ""
    def __init__(self, id):
        self.id = id


readu16 = readu16be
readu32 = readu32be
sar = open(sys.argv[1], 'rb')
sar.seek(1)
assert sar.read(3) == "SAR"
byteOrder = readu16be(sar)
print hex(byteOrder)
if byteOrder == 0xfeff:
    readu16 = readu16be
    readu32 = readu32be
    endian = ">"
#   print "big endian"
elif byteOrder == 0xfffe:
    readu16 = readu16le
    readu32 = readu32le
    endian = "<"
#   print "little endian"
else:
    print "not found"
#print endian
sar.seek(0x10)
sectionCount = readu16(sar)
sar.seek(2, 1)
#print sectionCount
#print sys.argv[1] + " opened"
for s in range(0, sectionCount):
    ID = readu16(sar)
#   print hex(ID)
    sar.seek(2, 1)
    if ID == 0x2000:
        strgOffset = readu32(sar)
        strgSize = readu32(sar)
    elif ID == 0x2001:
        infoOffset = readu32(sar)
        infoSize = readu32(sar)
    elif ID == 0x2002:
        fileOffset = readu32(sar)
        fileSize = readu32(sar)

# Reading STRG section
sar.seek(strgOffset)
assert sar.read(4) == "STRG"
assert readu32(sar) == strgSize

for s in range(0, 2):
    ID = readu16(sar)
#   print hex(ID)
    sar.seek(2, 1)
    if ID == 0x2400:
        strgTableOffset = readu32(sar) + strgOffset + 8
    elif ID == 0x2401:
        strgOtherOffset = readu32(sar) + strgOffset + 8

sar.seek(strgTableOffset)
strgCount = readu32(sar)
names = []
for i in range(strgCount):
    names.append(strg(i))
    ID = readu16(sar)
    sar.seek(2, 1)
    if ID ==  0x1f01:
        names[i].offset = readu32(sar) + strgTableOffset
        names[i].size = readu32(sar)
        tempPos = sar.tell()
        sar.seek(names[i].offset)
        names[i].name = sar.read(names[i].size-1)
#       print names[i].name
        sar.seek(tempPos)
    else:   # string may be null
        pass


# Reading INFO section
sar.seek(infoOffset)
assert sar.read(4) == "INFO"
assert readu32(sar) == infoSize
for i in range(8):
    ID = readu16(sar)
#   print hex(ID)
    sar.seek(2, 1)
    if ID == 0x2100:
        audioTableOffset = readu32(sar) + infoOffset + 8
    elif ID == 0x2101:
        bankTableOffset = readu32(sar) + infoOffset + 8
    elif ID == 0x2102:
        playerTableOffset = readu32(sar) + infoOffset + 8
    elif ID == 0x2103:
        warcTableOffset = readu32(sar) + infoOffset + 8
    elif ID == 0x2104:
        setTableOffset = readu32(sar) + infoOffset + 8
    elif ID == 0x2105:
        groupTableOffset = readu32(sar) + infoOffset + 8
    elif ID == 0x2106:
        fileTableOffset = readu32(sar) + infoOffset + 8
    elif ID == 0x220B:
        fileTableEnd = readu32(sar) + infoOffset + 8
    else:
        print "Unknown table pointer"
        sar.close()
        sys.exit()

# process file table
sar.seek(fileTableOffset)
fileCount = readu32(sar)
files = []
for i in range(fileCount):
    sar.seek(fileTableOffset + 4 + i * 8)
    files.append(sarFile(i))
    ID = readu16(sar)
#   print hex(ID)
    sar.seek(2, 1)
    if ID == 0x220a:
#       print "exists"
        files[i].offset = readu32(sar)
    else:
        print "File doesn't exist"
        sar.seek(4, 1)
    sar.seek(fileTableOffset + files[i].offset)
    ID = readu16(sar)
#   print hex(ID)
    sar.seek(10, 1)
    if ID == 0x220c:
        files[i].internal = True
        ID = readu16(sar)
        sar.seek(2, 1)
        if ID != 0xffff:
            offset = readu32(sar)
            if offset != 0xffffffff:
                files[i].fileOffset = offset + fileOffset + 8
                files[i].size = readu32(sar)
                sar.seek(files[i].fileOffset)
                files[i].ext += string.lower(sar.read(4))
#               print files[i].ext
    elif ID == 0x220d:
        files[i].path = getString(sar)
#       print files[i].path

# process wave archive table
warcs = []
sar.seek(warcTableOffset)
warcCount = readu32(sar)
for i in range(warcCount):
    warcs.append(warc(i))
    sar.seek(warcTableOffset + 4 + i * 8)
    ID = readu16(sar)
#   print hex(ID)
    sar.seek(2, 1)
    if ID == 0x2207:
        warcs[i].offset = readu32(sar)
        sar.seek(warcTableOffset + warcs[i].offset)
#       print hex(sar.tell())
        warcs[i].fileID = readu32(sar)

#looping through warcs again to see which ones have a name
for i in range(warcCount):
    fileID = warcs[i].fileID
    if i < warcCount - 1:
        warcs[i].infoLength = warcs[i+1].offset - warcs[i].offset
    else:
        warcs[i].infoLength = groupTableOffset - warcTableOffset - warcs[i].offset
    if warcs[i].infoLength > 0xc:
        sar.seek(warcTableOffset + warcs[i].offset + 0xc)
        strgID = readu32(sar)
        warcs[i].name = names[strgID].name
    else:
        warcs[i].name = "WARC_%s" % hex(i)
    files[fileID].fileName = warcs[i].name + files[fileID].ext
    print files[fileID].fileName
    # now to process wave archive
    if files[fileID].internal and files[fileID].fileOffset != None:
        sar.seek(files[fileID].fileOffset+1)
#       print hex(sar.tell())
#       print sar.read(4)
        assert sar.read(3) == "WAR"
        sar.seek(files[fileID].fileOffset)
        """
        war = open(files[fileID].fileName, "wb")
        war.write(sar.read(files[fileID].size))
        war.close()
        """
        # to be finished

# process bank table
banks = []
sar.seek(bankTableOffset)
bankCount = readu32(sar)
print bankCount
for i in range(bankCount):
    banks.append(bank(i))
    sar.seek(bankTableOffset + 4 + i * 8)
    ID = readu16(sar)
#   print hex(ID)
    sar.seek(2, 1)
    if ID == 0x2206:
        banks[i].offset = readu32(sar)
        sar.seek(bankTableOffset + banks[i].offset)
        fileID = banks[i].fileID = readu32(sar)
        sar.seek(0xc,1)
        strgID = banks[i].strgID = readu32(sar)
        banks[i].name = names[strgID].name
        # now to process bank
        if files[fileID].internal and files[fileID].fileOffset != None:
            sar.seek(files[fileID].fileOffset+1)
#           print hex(sar.tell())
#           print sar.read(4)
            assert sar.read(3) == "BNK"
            # to be finished

# process audio table
sounds = []
sar.seek(audioTableOffset)
audioCount = readu32(sar)
print audioCount
for i in range(audioCount):
    sounds.append(audio(i))
    sar.seek(audioTableOffset + 4 + i * 8)
    ID = readu16(sar)
#   print hex(ID)
    sar.seek(2, 1)
    if ID == 0x2200:
        sounds[i].offset = readu32(sar)
        sar.seek(audioTableOffset + sounds[i].offset)
        fileID = sounds[i].fileID = readu32(sar)
        sar.seek(8,1)
        typeID = readu16(sar)
        sar.seek(10,1)
        strgID = sounds[i].strgID = readu32(sar)
        sounds[i].name = names[strgID].name
        files[fileID].fileName = sounds[i].name + files[fileID].ext
        if files[fileID].internal:
            pass
#           print files[fileID].fileName
        if typeID == 0x2201:    # External stream
            pass
        elif typeID == 0x2202:  # Wave sound
            pass
#           print files[fileID].fileName
            sar.seek(files[fileID].fileOffset+1)
            assert sar.read(3) == "WSD"
            sar.seek(files[fileID].fileOffset)
            """
            wsd = open(files[fileID].fileName, "wb")
            wsd.write(sar.read(files[fileID].size))
            wsd.close()
            """
        elif typeID == 0x2203:  # Sequence
            sar.seek(0x40,1)
            if fileID == 207:
                print hex(sar.tell())
            bankID = sounds[i].bankID = readByte(sar)
#           print hex(bankID)
#           banks[sounds[i].bankID].seqs.append(1)
            sar.seek(files[fileID].fileOffset+1)
            assert sar.read(3) == "SEQ"

print hex(bankCount)

#for i in range(bankCount):
#   print banks[i].name
#   print "\t" + str(len(banks[i].seqs))
#   for j in range(len(banks[i].seqs)):
#       print "\t" + files[banks[i].seqs[j]].fileName


sar.close()