Nayuta Script Insert v1.3

# v1.3 Inserts 0xC1 opcodes
# v1.2 Specify insert files. Fix more jumps. Does all files in list.
# Insert Nayuta no Kiseki script file / update data.lst
# Usage is python script.py mp_0000
# Where mp_0000.bin is the file you want to insert
# Input file is from the dumper (*.data) with whichever strings you
#   want to insert, in tab-separated format.
# Also need *.orig (original data file) in same directory.
# Also need data.lst from the ROM in same directory.
import os
import struct
import sys
import pdb

filelist = ['mp_0000', 'mp_0000c', 'mp_0000d',
            'mp_0001', 'mp_0001c', 'mp_0001d',
            'mp_0002', 'mp_0002c', 'mp_0002d',
            'mp_0003', 'mp_0003c', 'mp_0003d',
            'mp_0004', 'mp_0004c', 'mp_0004d', 'mp_0004n',
            'mp_0005', 'mp_0005c',
            'mp_0100', 'mp_0100a', 'mp_0100b',
            'mp_0101', 'mp_0101a', 'mp_0101b', 'mp_0101c',
            'mp_0102', 'mp_0104', 'mp_0105', 'mp_0106',
            'mp_0110', 'mp_0111', 'mp_0112', 'mp_0113', 'mp_0114',
            'mp_0115', 'mp_0115a', 'mp_0115b', 'mp_0115c', 'mp_0115d',
            'mp_0116', 'mp_0117', 'mp_0118', 'mp_0119',
            'mp_1000a', 'mp_1000b', 'mp_1000c', 'mp_1000d',
            'mp_1001a', 'mp_1001b', 'mp_1001c', 'mp_1001d',
            'mp_1002a', 'mp_1002b',
            'mp_1010', 'mp_1011',
            'mp_1099', 'mp_1100',
            'mp_2000b', 'mp_2000c',
            'mp_2001a', 'mp_2001b', 'mp_2001c', 'mp_2001d',
            'mp_2002a', 'mp_2002b', 'mp_2002d',
            'mp_2010', 'mp_2099', 'mp_2100',
            'mp_3000b', 'mp_3000c', 'mp_3000d',
            'mp_3001a', 'mp_3001b', 'mp_3001c', 'mp_3001d',
            'mp_3002a', 'mp_3002c', 'mp_3002d',
            'mp_3003c', 'mp_3003d',
            'mp_3099', 'mp_3100',
            'mp_4000a', 'mp_4000c', 'mp_4000d',
            'mp_4001a', 'mp_4001b', 'mp_4001c', 'mp_4001d',
            'mp_4002a', 'mp_4002b', 'mp_4002c', 'mp_4002d',
            'mp_4011', 'mp_4099', 'mp_4100',
            'mp_5000', 'mp_5098', 'mp_5099', 'mp_5100', 'mp_5101',
            'mp_5200', 'mp_5300',
            'mp_6000', 'mp_6001', 'mp_6002', 'mp_6003',
            'mp_6100', 'mp_6100d', 'mp_6200',
            'mp_6300', 'mp_6301', 'mp_6302', 'mp_6303',
            'mp_7020', 'mp_7098',
            'mp_7099', 'mp_7099a', 'mp_7099b', 'mp_7099c',
            'mp_7100', 'mp_7200',
            'mp_8098', 'mp_8099', 'mp_8100', 'mp_8100b',
            'noi', 'system']

def get_data(filename):
    totalbytes = os.path.getsize(filename)
    infile = open(filename, 'rb')
    totalfiledata = infile.read(totalbytes)
    infile.close()
    return totalfiledata

def replacestr(origstr,replacestr,startpos,replacelen):
#Returns a string with a replaced sub-string
#origstr - the original string, replacestr = the string to replace
#startpos - where the replacement string should go
#replacelen - how many characters of the original string to replace
    return origstr[:startpos] + replacestr + origstr[startpos+replacelen:]

def twos_comp(val, bits):
    """compute the 2's compliment of int value val"""
    if( (val&(1<<(bits-1))) != 0 ):
        val = val - (1<<bits)
    return val

def myprogram(filename):

    opcodes = ((0x07,2),(0x40,2),(0x3f,2),(0x41,0x14),
               (0x98,2),(0xc9,8),(0x36,8),(0xce,8),
               (0xC1,'special'))

    # Load input
    inputdata = []
    with open(filename + '.data','rb') as f:
        for line in f:
            line = line.translate(None,'\r\n')
            line = line.split('\t')
            if len(line) < 3:
                continue
            if line[1] != '':
                line[1] = int(line[1],16)
            if line[0] != '' and line[1] == '':
                pass
            elif line[0] != '': #Line has a new opcode
                if line[1] in [op[0] for op in opcodes]:  #It's one of our opcodes
                    if line[4] == "" and line[3] != "":
                        print '%s: Blank opcode %s at %s.' % (filename, hex(line[1]), line[0])
                    line[0] = int(line[0],16)
                    inputdata.append(line[:3])
                    inputdata[-1].append([line[4]]) #append column 5 as set
            else:
                inputdata[-1][-1].append(line[4])

    filedata = get_data(filename + '.orig')

    # Load pointers (file header)
    pointers = []
    pos = 0x3C
    EOF = struct.unpack('<I',filedata[pos:pos+4])[0]
    while pos < EOF:
        pointers.append([pos,struct.unpack('<I',filedata[pos:pos+4])[0]])
        pos += 0x28

    # Load more pointers (relative pointers in opcodes)
    relpointers = []
    pos = 0
    for opcode in ('\x7b\x80\xe0\x82','\x7c\x80\xe0\x82','\x7d\x80\xdd\x82',
                   '\x80\x80\xdd\x82','\x81\x80\xdd\x82','\x83\x80\xdd\x82'):
        while True:
            pos = filedata.find(opcode,pos+1)
            if pos == -1:
                break
            if opcode[0] in ('\x7b','\x7c'):
                pos += 4
                pos += struct.unpack('<I',filedata[pos:pos+4])[0] + 6 #Jump past 82E0 opcode
                relpointers.append([pos,struct.unpack('<I',filedata[pos:pos+4])[0]])
            elif opcode[0] in ('\x7d','\x80','\x81'):
                pos += 4
                val = struct.unpack('<I',filedata[pos:pos+4])[0]
                if opcode[0] == '\x80':
                    val = twos_comp(val,32)
                relpointers.append([pos,val])
            elif opcode[0] == '\x83':
                pos += 10 #Jump past 82DD opcode
                relpointers.append([pos,struct.unpack('<I',filedata[pos:pos+4])[0]])

    # Replace strings
    offset = 0
    for (count,(addr,op,name,lines)) in enumerate(inputdata):
        addr += offset
        text = ""
        # Build the 0x82DF opcodes (that have strings in them)
        if op == 0x07:
            for line in lines:
                text += line + '\\n'
            text = text[:-2]
            text = '\xDF\x82' + struct.pack('<I',len(text)) + text
        else:
            for line in lines:
                text += '\xDF\x82' + struct.pack('<I',len(line)) + line
        name = '\xDF\x82' + struct.pack('<I',len(name)) + name
        if op in (0x40,0x98):
            text = name + text
        # Go to the start of the first DF opcode
        # Then look for the end of the string list
        for opcode, jumplen in opcodes:
            if op == opcode:
                break
        else:
            print "Failed opcode lookup!"
            quit()
        if op == 0xC1:
            addr += 2
            for x in range(3):
                addr += 2
                datalen = struct.unpack('<I',filedata[addr:addr+4])[0]
                addr += datalen + 4
        else:
            addr += jumplen
        startpos = addr
        if filedata[addr] != '\xDF':
            pdb.set_trace()
            print 'Fail'
            quit()
        else:
            while filedata[addr] == '\xDF':
                addr += struct.unpack('<I',filedata[addr+2:addr+6])[0] + 6
        endpos = addr
        origlen = endpos - startpos
        newlen = len(text)
        # Replace the string
        filedata = replacestr(filedata,text,startpos,endpos-startpos)
        # Update pointers
        for i,(pointerpos,pointertgt) in enumerate(pointers):
            if pointertgt > startpos:
                pointers[i][1] = pointertgt + newlen - origlen
        # Update relative pointers
        for i,(pointerpos,pointertgt) in enumerate(relpointers):
            if pointertgt < 0:
                if pointerpos + pointertgt > startpos:
                    relpointers[i][0] += newlen - origlen
                elif pointerpos > startpos:
                    relpointers[i][0] += newlen - origlen
                    relpointers[i][1] += -newlen + origlen
            else:
                if pointerpos > startpos:
                    relpointers[i][0] += newlen - origlen
                elif pointerpos + pointertgt > startpos:
                    relpointers[i][1] += newlen - origlen
        # Update offset
        offset += newlen - origlen

    # Write-back pointers
    for pointerpos, pointertgt in pointers:
        pointertgt = struct.pack('<I',pointertgt)
        filedata = replacestr(filedata,pointertgt,pointerpos,4)

    # Write-back relative pointers
    for pointerpos, pointertgt in relpointers:
        if pointertgt > -1:
            pointertgt = struct.pack('<I',pointertgt)
        else:
            pointertgt = int(hex(pointertgt & 0xFFFFFFFF)[2:-1],16)
            pointertgt = struct.pack('<I',pointertgt)
        filedata = replacestr(filedata,pointertgt,pointerpos,4)

    # Write the output
    outfile = open(filename + '.bin','wb')
    outfile.write(filedata)
    outfile.close()

    # Update data.lst
    datasize = struct.pack('<I',len(filedata))
    filedata = get_data('data.lst')
    pos = filedata.find(filename + (8-len(filename))*'\x00',0xCD00)
    filedata = replacestr(filedata,datasize,pos+8,4)

    outfile = open('data.lst','wb')
    outfile.write(filedata)
    outfile.close()

if __name__ == '__main__':
##    for testing in IDLE
##    sys.argv.append('mp_0000')
    for filename in filelist:
        myprogram(filename)