Astral Chain pkz

extract_pkz.py

#"pkzl", 0x10000, full_size, ???
#4 vars in each entry, count, decsize, offset, encsize?
#!! pip install zstd !!

from struct import unpack
from zstd import decompress
from os.path import exists, splitext
from os import makedirs
import sys

def read16(f):
    return unpack("<H", f.read(2))[0]

def read32(f):
    return unpack("<I", f.read(4))[0]

def getstr(f):
    ret = b"";char = f.read(1)
    while char != b"\x00":
        ret += char
        char = f.read(1)
    if f.tell() % 8: #normally don't align but only strings here are
        f.seek(8 - (f.tell() % 8), 1)
    return ret

with open(sys.argv[1], "rb") as f:
    outname = sys.argv[1][:-4]
    assert f.read(4) == b"pkzl" #magic
    assert read32(f) == 0x10000 #??? version?
    f.seek(0, 2)
    full_size = f.tell()
    f.seek(8)
    assert read32(f) == full_size #assert file+8 = filesize
    header = unpack("<5I", f.read(0x14)) #0x20 header
    entries = [unpack("<4Q", f.read(32)) for i in range(header[1])] #4 lwords
    assert f.read(16) == b"ZStandard".ljust(16, b"\x00") #always 0x10 afaict
    filenames = [getstr(f).decode("UTF-8") for i in range(header[1])]
    #variable length, aligned to 8-byte boundary, "UTF-8" so py3's happy >.>

    if not exists(outname): #make output folder if it doesn't exist
        makedirs(outname)
    for i in range(header[1]):
        #print("%016X %016X %016X %016X" % entries[i])
        if entries[i][1] != 0: #entry_size can be 0, skip those? null zstd header
            print(filenames[i]) #pretty print
            with open(outname + "/" + filenames[i], "wb") as o:
                f.seek(entries[i][2]) #seek to entry_offset
                o.write(decompress(f.read(entries[i][3]))) #decompress dec_size