Python script to convert exported JavaScript back into a PICO-8 cartridge

pico8jstocart.py

import os.path
import re
import sys


# LZ-ish decompression scheme borrowed from picolove:
# https://github.com/gamax92/picolove/blob/master/cart.lua
compression_map = b"\n 0123456789abcdefghijklmnopqrstuvwxyz!#%(){}[]<>+=/*:;.,~_"
def decompress(code):
    lua = bytearray()
    mode = 0
    copy = None
    i = 7
    codelen = (code[4] << 8) | code[5]
    while len(lua) < codelen:
        i = i + 1
        byte = code[i]
        if mode == 1:
            lua.append(byte)
            mode = 0
        elif mode == 2:
            # copy from buffer
            offset = len(lua) - ((copy - 0x3c) * 16 + (byte & 0xf))
            length = (byte >> 4) + 2
            lua.extend(lua[offset:offset + length])
            mode = 0
        elif byte == 0x00:
            # output next byte
            mode = 1
        elif 0x01 <= byte <= 0x3b:
            # output this byte from map
            lua.append(compression_map[byte - 1])
        elif byte >= 0x3c:
            # copy previous bytes
            mode = 2
            copy = byte

    return bytes(lua)


def main():
    if len(sys.argv) != 3:
        print("usage: pico8jstocart.py infile.js outfile.p8")
        sys.exit(1)

    infile, outfile = sys.argv[1:]

    if os.path.exists(outfile):
        print("Cowardly refusing to overwrite", outfile)
        sys.exit(1)

    with open(infile) as f:
        jsblob = f.read()
    m = re.search(r'\bvar _cartdat=[[]([0-9,\n]+)[]]', jsblob)
    if m:
        jsdata = m.group(1)
    else:
        raise ValueError("Can't find _cartdat")

    data = bytes(int(number) for number in jsdata.split(','))

    with open(outfile, 'w', encoding='latin1') as f:
        def write(*args):
            print(*args, file=f)

        write("pico-8 cartridge // http://www.pico-8.com")
        write("version 8")

        # ROM layout largely matches the documented RAM layout:
        # 0x0000 sprites
        # 0x1000 shared sprite/map region
        # 0x2000 map
        # 0x3000 sprite flags
        # 0x3100 music
        # 0x3200 sounds
        # 0x4300 Lua

        # Lua is first in the cartridge, last in ROM
        lua = data[67 * 256:]
        if lua[:4] == b':c:\x00':
            lua = decompress(lua)
        lua = lua.decode('latin1')
        write("__lua__")
        write(lua)

        # Next is the spritesheet.  Somewhat inconveniently, the nybbles need
        # flipping, because little-endian.
        write("__gfx__")
        for i in range(128):
            write(''.join(data[j:j+1].hex()[::-1] for j in range(i * 64, (i + 1) * 64)))

        # Sprite flags can be written out pretty much verbatim
        write("__gff__")
        write(data[0x3000:0x3080].hex())
        write(data[0x3080:0x3100].hex())

        # Map too!
        write("__map__")
        for offset in range(0x2000, 0x3000, 128):
            write(data[offset:offset+128].hex())

        # Sound is more of a pain.  It's stored in the ROM as a compact 4 nybbles
        # per note, but the cartridge uses an expanded format with 5 nybbles per
        # note.
        write("__sfx__")
        for offset in range(0x3200, 0x4300, 68):
            sfxdata = data[offset:offset+64]
            # Flags are at the end of the record in the ROM, but the beginning in
            # the cart.  Luckily they can be dumped as-is.
            flags = data[offset+64:offset+68]
            buf = [flags.hex()]

            for n in range(0, len(sfxdata), 2):
                b0, b1 = sfxdata[n:n+2]
                note = b0 & 0x3f
                props = (b1 << 2) | (b0 >> 6)
                instrument = (props >> 0) & 0x7
                volume = (props >> 3) & 0x7
                effect = (props >> 6) & 0x7
                buf.append(f"{note:02x}{instrument:01x}{volume:01x}{effect:01x}")

            write(''.join(buf))

        # Music also comes in a compressed form, but a rather simpler one
        write("__music__")
        for offset in range(0x3100, 0x3200, 4):
            tracks = bytearray(data[offset:offset+4])
            flags = 0
            for t, track in enumerate(tracks):
                if track & 0x80:
                    flags |= 1 << t
                    tracks[t] ^= 0x80
            write(f"{flags:02x}", tracks.hex())

        # And finally, the cart ends with a blank line.  And done!
        write()


if __name__ == '__main__':
    main()

It doesn't work properly for the lua code. The rest is exported nicely, but the lua code isn't properly decoded. I've tried to solve it, changing the encoding from latin1 to utf-8, ascii, etc, but it doesn't produce any legible output, so pico8 doesn't pick it up either.