Wander Wonder Script Dump v1.1

#1.1    Changed dump format- added delay into the flag box
#1      Initial release
import os
import struct
import sys
asciichar = 'ABCDEFGHIJKLMNOPQRSTUVWXYZ()0123456789 :!'

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

#Tells if the string s is a SHIFT-JIS character or not
def is_jis_char(s):
    if len(s) != 2: #If it's not a string of length 2, then...
        return False
    try:            #Python error handling
        if len(s.decode('cp932')) != 1: #For example, 2 ascii characters will be length 2
            return False                #Two bytes that decode to a SHIFT-JIS character will be length 1
    except UnicodeDecodeError:  #The string couldn't be decoded
        return False            #Not a SHIFT-JIS character
    return True                 #It is a SHIFT-JIS character

def script_dump(filename):
    filedata = get_data(filename + '.orig')  #Get data
    ptr = 0                         #Initialize variables
    end_of_ptr_table = struct.unpack('<I',filedata[0:4])[0]
    ptr_table = []
    while ptr < end_of_ptr_table:   #Load pointers
        ptr_table.append(struct.unpack('<I',filedata[ptr:ptr+4])[0])
        ptr += 4
    flag = True
    chunks = []                     #Load chunks
    for ptr in ptr_table:
        if flag == True:
            flag = False
        else:
            chunks.append([prev_ptr,filedata[prev_ptr:ptr]])
        prev_ptr = ptr
    chunks.append([ptr,filedata[ptr:]])

    #Find open dialog chunk and delay chunk by searching
    for i, (ptr, chunk) in enumerate(chunks):
        if chunk == '400B0000400600004001000040090100400C0000000000004002010040160000'.decode('hex'):
            print 'Open dialog box chunk is 0x{:2x} at address {}.'.format(i,hex(ptr))
            opendialog_chunk = i
        elif chunk == '4002000040033C804002010040160000'.decode('hex'):
            print 'Delay chunk is 0x{:2x} at address {}.'.format(i,hex(ptr))
            delay_chunk = i

    script_chunks = []
    name_chunks = []
    for i, (ptr, chunk) in enumerate(chunks):   #Loop through chunks
        #Finds open dialog and delay chunks
        if chunk[:4] == '\x40\x14' + struct.pack('<H',opendialog_chunk):
            script_chunks.append(i)
        elif chunk[:4] == '\x40\x02\x00\x00' and chunk.find('400500004002010040160000'.decode('hex')) > 0:
            name_chunks.append(i)
    print 'Script Chunks: {}'.format(', '.join([hex(x) for x in script_chunks]))
    print 'Name Chunks: {}'.format(', '.join([hex(x) for x in name_chunks]))

    output = []                 #Initialize variables
    namelist = []
    for j in name_chunks:       #Loop through name chunks
        i, chunk = chunks[j]    #Assign chunk to local variable
        pos = 4                 #Name starts at position 4 within chunk
        while is_jis_char(chunk[pos:pos+2]):
            pos += 2            #Find end of name
        name = chunk[4:pos]
        namelist.append([j, name])  #Add chunk number and name to name list
        output.append(hex(j) + ';')         #Add name to output
        output.append(name + ';;' + 'name' + '\r\n')

    firstpass = True
    for chunk in script_chunks: #Loop through script chunks
        if firstpass == True:
            firstpass = False
        else:
            del output[-1]      #Delete ';' entry if 1st line in opcode
        output.append(hex(chunk) + ';')     #Add chunk # to output
        chunkaddr = chunks[chunk][0]        #Get chunk addr and text
        chunk = chunks[chunk][1]
        pos = 0                             #Init variable
        while pos < len(chunk):
            if ord(chunk[pos]) == 0:        #For byte 0x00, advance to next byte
                pos += 1
                continue
            elif ord(chunk[pos]) == 0x40:   #For byte 0x40, it's an opcode
                if ord(chunk[pos+1]) == 0x17:
                    output.append('0x' + chunk[pos:pos+8].encode('hex') + ';' +
                                  '0x' + chunk[pos:pos+8].encode('hex') + ';' + 'unknown' + '\r\n')
                    output.append(';')
                    pos += 8
                    continue
                output.append('0x' + chunk[pos:pos+4].encode('hex') + ';' + '0x' +
                              chunk[pos:pos+4].encode('hex') + ';')
                if ord(chunk[pos+1]) == 0x14:
                    f_number = struct.unpack('<H',chunk[pos+2:pos+4])[0]
                    if f_number == opendialog_chunk:
                        output.append('open dialog')
                    elif f_number == delay_chunk:
                        output.append('delay')
                    elif f_number in [x[0] for x in namelist]:
                        for i, name in namelist:
                            if f_number == i:
                                break
                        output.append('name: {}'.format(name))
                    else:
                        output.append('function {}'.format(hex(f_number)))
                elif ord(chunk[pos+1]) == 0x00:
                    output.append('end chunk')
                elif ord(chunk[pos+1]) == 0x02:
                    output.append('unknown')
                elif ord(chunk[pos+1]) == 0x05:
                    output.append('linebreak')
                elif ord(chunk[pos+1]) == 0x06:
                    output.append('closebox')
                else:
                    output.append('unknown opcode')
                pos += 4
    ##            elif ord(chunk[pos]) == 0x80:
    ##                output.append(chunk[pos:pos+4].encode('hex') + ';' + chunk[pos:pos+4].encode('hex') + ';')
    ##                output.append('unknown opcode')
    ##                pos += 4
            elif is_jis_char(chunk[pos:pos+2]) or chunk[pos] in asciichar:  #It's text
                start = pos
                while is_jis_char(chunk[pos:pos+2]) or chunk[pos] in asciichar:
                    if chunk[pos] in asciichar:         #If ascii, advance by 1
                        pos += 1
                    elif is_jis_char(chunk[pos:pos+2]): #If S-JIS, advance by 2
                        pos += 2
                output.append(chunk[start:pos] + ';')   #Output text
                if chunk[pos:pos+2] == '\x40\x05':
                    output.append(';linebreak')
                    pos += 4
                elif chunk[pos:pos+4] == '\x00\x00\x40\x05':
                    output.append(';linebreak')
                    pos += 6
                if chunk[pos:pos+4] == '\x40\x14' + struct.pack('<H',delay_chunk):
                    if 'linebreak' in output[-1]:
                        output[-1] += '+delay'
                    else:
                        output.append(';delay')
                    pos += 4
            else:   #Not byte 0x00, byte 0x40, or text
                print 'Unknown character'
                print hex(chunkaddr)        #Start of chunk
                print hex(chunkaddr+pos)    #Postion of miss
                quit()
            output.append('\r\n')           #Output linebreak and indent
            output.append(';')              #for next opcode within chunk
    with open(filename + '.csv','wb') as f: #Write output to file
        f.write(''.join(output))

scripts = ['apend00','bird01','bird02','bird03','bird04','bird05','bird06','birddemo','boss01','boss02','boss2',
           'boss03','boss3','catedral','church01','church02','church03','churdemo','cnt01','coloseum',
           'eddemo01','eddemo02','eddemo03','exdemo01','extra01','extra02','extra03','fir01','fir02','fir03',
           'grave01','grave02','grave03','house','ice01','ice02','ice03','inn01','inn02','inn03','ise01','ise02',
           'ise03','knondemo','last01','last02_0','last02_1','last02_2','last02_3','last03','last04','lastdemo',
           'mnt00','mnt01','mnt02','mnt03','mnt04','mnt05','museum','nxdemo01','nxdemo02','nxdemo03','opdemo00',
           'opdemo01','opdemo02','opdemo03','plane01','plane02','plane03','random','sai01','sai02','sai03',
           'sai04','sand01','sand02','shops','soldier','subgame','thu01','thu02','thu03','tiademo','tower00',
           'tower01','tower02','tower03','tower04','tower05','tower06','tower07','tower08','tower09','town01',
           'town02','town2','town03','tree00','tree01','tree02','tree03','tree04','tutori01','tutori02','tutori03',
           'twdemo01','twdemo02','wind00','wind01','wind02','wood01','wood02','wpshop01','wpshop02','wpshop03']
for script in scripts:                      #Dump each script
    print 'Dumping: {}'.format(script)
    script_dump(script)
    print ''