QB Bulk Script Dumper v1

#How to use
#Need shared.bin in the same folder as this tool and run. That's it.

import struct
import zlib

ARCHIVE_TOC_POS = 0x17da00

def I(f, offset = -1):                  #Function is supposed to unpack a 32-bit integer from a file-like object or from a bytes-like object
    if hasattr(f, 'read'):              #File-like object
        if offset == -1:                #If no 2nd parameter
            return struct.unpack('<I', f.read(4))[0]    #Read data at current position
        else:                           #2nd parameter was specified
            origpos = f.tell()          #Save original position
            f.seek(offset)              #Go to specified position
            val = struct.unpack('<I', f.read(4))[0]     #Read data
            f.seek(origpos)             #Go back to original position
            return val
    elif isinstance(f, bytes):          #Bytes-like object
        if len(f[offset:offset+4]) != 4:                #Checking for EOF error
            raise Exception('Not enough data in file.')
        return struct.unpack('<I', f[offset:offset+4])[0]   #Unpack data
    else:                               #Not a file or bytes object
        raise Exception('Unknown object type.')

ptrs1 = []                              #Archive TOC data container
with open('shared.orig','rb') as f:
    #Stores the archive TOC data
    f.seek(ARCHIVE_TOC_POS + 4)         #Go to entry for # of files
    num_of_files = I(f)
    for x in range(num_of_files):       #Read TOC entries
        offset = I(f)                   #Offset 1st
        f.seek(4, 1)                    #We don't care about unzipped size (so seek past it)
        size = I(f)                     #Zipped size last
        ptrs1.append((offset, size))

    #This is the part that unpacks each file
    for i, (offset, size) in enumerate(ptrs1):      #Loop over the list of TOC data
                                                    #The offset and the zipped size are stored in the TOC data
        f.seek(ARCHIVE_TOC_POS + offset)            #Go to data position
        filedata = zlib.decompress(f.read(size))    #Decompress data
        #The rest of these steps work with the decompressed data
        num_of_pointers = I(filedata, 0x38)         #How many strings are in the file
        text_addr = I(filedata, 0x4c)               #Where is the text located
        pos = I(filedata, 0x44)                     #Where is the text TOC located
        ptrs2 = []                          #Container for text TOC data
        for x in range(num_of_pointers):    #Repeat for as many times as there are strings in the text data
            ptrs2.append(I(filedata, pos))  #Save the pointer
            pos += 4
        filedata = filedata[text_addr:]     #Cut off everything that isn't the text
                                            #Pointers are relative to start of text

        output = []                         #Container for outputted data
        for ptr in ptrs2:                   #For each string pointer
            output.append(hex(ptr) + '\t')  #Store position to output
            s = filedata[ptr:filedata.find(b'\x00',ptr)].decode('cp932')    #Get the string
            lines = s.split('@')            #Split by @ which is the game substring marker
            output.append(lines.pop(0) + '\t')  #Add the "name" info (really 1st substring)
            for line in lines:              #For each line, add it to the output
                output.append(line + '\n\t\t')
            if len(lines) == 0:             #Special case where there is only one line
                output.append('\n')
            else:                           #Delete extra tabs from "output.append(line + '\n\t\t')"
                output[-1] = output[-1][:-2]

        #This outputs the text
        filename = '{:0>3d}'.format(i)      #Formats the "file number" like 000, 001, etc...
        with open(filename + '.tsv','wb') as g:
            for x in output:
                g.write(x.encode('cp932'))