pak_extract

import os
import struct
from collections import namedtuple
from sys import argv

# this code will extract the contents of Quake 1 format pak files
# usage python pak_extract.py <filename>
# the pak contents will be extracted under a folder named after the pak file

# type definitions
PAKHeader = namedtuple('PAKHeader', "format, dir_ofs, dir_size")
fmt_PAKHeader = '<4s2i'

PAKEntry = namedtuple('PAKDirectory', "path, ofs, size")
fmt_PAKEntry = '<56s2I'

def extract(pak_path):
    with open(pak_path, 'rb') as pak_file:
        header_data = pak_file.read(struct.calcsize(fmt_PAKHeader))
        header = PAKHeader._make(struct.unpack(fmt_PAKHeader, header_data))

        entry_struct = struct.Struct(fmt_PAKEntry)
        entry_size = struct.calcsize(fmt_PAKEntry)
        num_entries = int(header.dir_size / struct.calcsize(fmt_PAKEntry))

        print("File: %s (found directory at %d containing %d entries)" % (pak_path, header.dir_ofs, num_entries))

        pak_file.seek(header.dir_ofs)
        pak_directory_data = pak_file.read(header.dir_size)

        for i in range(0, num_entries):
            entry_ofs = i * entry_size
            entry_data = pak_directory_data[entry_ofs:entry_ofs+entry_size]
            entry = PAKEntry._make(entry_struct.unpack_from(entry_data))
            # some entries (e.g. PAK0.PAK/progs.dat) are not correctly 0 padded,
            # so this is required as .decode doesn't work
            entry_path = ""
            for j, b in enumerate(entry.path):
                if b == '\x00':
                    entry_path = entry.path[0:j].decode('ascii')
                    break
                elif j == 55:
                    entry_path = entry.path.decode('ascii')
            # save the file
            if entry_path != "":
                print("[%d] %s (%d bytes at %d)" % (i, entry.path, entry.size, entry.ofs))
                file_name = os.path.basename(entry_path)
                file_path = os.path.dirname(entry_path)
                # read the file data
                pak_file.seek(entry.ofs)
                file_data = pak_file.read(entry.size)
                # prepare the output path
                output_dir = os.path.splitext(pak_path)[0] + "/" + file_path
                if not os.path.exists(output_path):
                    os.makedirs(output_dir)
                output_path = output_dir + "/" + file_name
                # write the file
                with open(output_path, 'wb') as output_file:
                    output_file.write(file_data)

extract(argv[1])