Python script to extract an AVPK archive

avpk.py

import sys
import io
import struct

def int32(f):
  return struct.unpack('<i', f.read(4))[0]

def string(f):
  bs = []
  while True:
    b = f.read(1)
    if len(b) == 0:
      raise ValueError('Unterminated string')
    if b[0] == 0:
      break
    bs.append(b[0])
  return str(bytes(bs), 'utf-8')

def copy_file(f, out, file_length):
  while file_length > 0:
    buf = f.read(min(file_length, 4096))
    if len(buf) == 0:
      raise ValueError('Truncated file')
    rc = out.write(buf)
    if rc != len(buf):
      raise IOError('Unable to write to file')
    file_length -= rc

def extract(f):
  if f.read(4) != b'AVPK':
    raise ValueError('Bad file signature')
  length = int32(f)
  strings_offset = int32(f)
  strings_length = int32(f)
  files_offset = int32(f)
  files_length = int32(f)
  data_offset = int32(f)
  data_length = int32(f)
  # Parse strings
  f.seek(strings_offset, io.SEEK_SET)
  num_strings = int32(f)
  strings = [(int32(f), int32(f)) for i in range(num_strings)]
  for (name_offset, val_offset) in strings:
    f.seek(strings_offset + name_offset, io.SEEK_SET)
    name = string(f)
    f.seek(strings_offset + val_offset, io.SEEK_SET)
    val = string(f)
    print(name + ": " + val)
  # Parse files
  f.seek(files_offset, io.SEEK_SET)
  num_files = int32(f)
  files = [(int32(f), int32(f), int32(f)) for i in range(num_files)]
  for (file_length, file_name_offset, file_offset) in files:
    f.seek(files_offset + file_name_offset, io.SEEK_SET)
    file_name = string(f)
    print(file_name + " " + str(file_length) + " bytes")
    with open(file_name, 'wb') as out:
      f.seek(data_offset + file_offset, io.SEEK_SET)
      copy_file(f, out, file_length)

with open(sys.argv[1], 'rb') as f:
  extract(f)