東方神医伝 extract

import os
import sys

#Constants taken from insani tools.
LITTLE_ENDIAN = 1
BIG_ENDIAN = -1

BYTE_LENGTH = 1
SHORT_LENGTH = 2
INT_LENGTH = 4
LONG_LENGTH = 8

ERROR_NONE = 0
ERROR_WARNING = 1
ERROR_ABORT = 2

def readbytes(infile, bytecount, key=-1):
    ret = infile.read(bytecount)

    return ret

def converttoint(bytes, size=INT_LENGTH, endian=LITTLE_ENDIAN):
    result = 0
    for i in xrange(size):
        temp=long(ord(bytes[i]))
        if endian == LITTLE_ENDIAN :
            result |= (temp << (8*i))
        elif endian == BIG_ENDIAN :
            result = (result << 8) | temp
        else :
                raise Exception('converttoint: Unknown endian specification')

        return result

def convertfromint(value, size=INT_LENGTH, endian=LITTLE_ENDIAN):
    result = ""
    for i in xrange(size):
        if endian == LITTLE_ENDIAN :
            #result |= (temp << (8*i))
            temp = (value >> (8*i)) & 0x00FF
        elif endian == BIG_ENDIAN :
            #result = (result << 8) | temp
            temp = (value >> (8*(size-i-1))) & 0x00FF
        else :
                raise Exception('convertfromint: Unknown endian specification')
            result += chr(temp)

        return result

#program

files = sys.argv[1:]
for filename in files:
    infile = open(filename, 'rb')

    print "Reading file", filename

    #Header:
    #4 bytes for the total size of the index, not counting this header.

    filecount = converttoint(readbytes(infile, 4, 0)) / 0x10C
    print "Expecting", filecount, "files."

    #Each file entry is
    #4 bytes    4 bytes     104 bytes
    #File start Filesize    Filename

    #Total of x10C bytes per entry.
    #The unusual length of the filename makes me think the last 4 bytes might be a useful int
    #I think the filename is just a filename. It isn't all zeroes, but the data does appear to be junk
    #(One filename included "png" several times). On rebuild it is probably safe to just zero fill the filename.

    entries = []

    for i in xrange(filecount):
        entry = {}
        entry['filestart'] = converttoint(readbytes(infile, 4))
        entry['filesize'] = converttoint(readbytes(infile, 4))
        nameraw = readbytes(infile, 0x100)
        #nameraw = readbytes(infile, 0x104)
        entry['name'] = nameraw[0:nameraw.find("\x00")]     #Truncate to the first \0.
        entry['unk'] = converttoint(readbytes(infile, 4))   #Maybe the filename is exactly x100?

        entries.append(entry)


    #print "Current position is now %d / %X" % (infile.tell(), infile.tell())

    #for entry in entries:
    #   for i in entry:
    #       print i + ":", entry[i]
    #   print ""

    #sys.exit(0)

    dirname = filename[0:filename.find(".")]
    if not os.path.isdir(dirname):
        os.mkdir(dirname)

    i = 0
    for entry in entries:
        filename = entry['name'].strip(" \n\t\x00")

        dirsegment, filesegment = os.path.split(filename)
        #filename = os.path.join(dirsegment, ("%03d_" % i) + filesegment)       #Can be done to enforce correct order on repacking
        if (not os.path.isdir(os.path.join(dirname, dirsegment))):
            os.makedirs(os.path.join(dirname, dirsegment))

        infile.seek(entry['filestart'])
        data = readbytes(infile, entry['filesize'])
        outfile = open(os.path.join(dirname, filename), 'wb')
        outfile.write(data)
        outfile.close()

        print "Created file %s of size %08X from position %08X" % (filename, entry['filesize'], entry['filestart'])
        #print "Unk is %08X " % entry['unk']    #It's mostly 0012F79C, also a few 0012FAC4. No idea what it would be used for.

        i += 1

    infile.close()