zipstream.py

   1 #! /usr/bin/env python3
   2
   3 import sys
   4 from struct import Struct
   5 from streams import streamcopy, DelegateWriter
   6 from binascii import crc32
   7 from io import BufferedIOBase
   8
   9 FILE_SIG = b"PK\x03\x04"
  10 FILE_STRUCT = Struct("< BB H H L LLL HH")
  11 FLAGS_ENCRYPTED = 0
  12 FLAGS_STREAMED = 3
  13 FLAGS_STRONG_ENC = 6
  14 FLAGS_UTF8 = 11
  15 METHOD_STORE = 0
  16
  17 def iter_files(zip):
  18     while True:
  19         sig = zip.read(4)
  20         if not sig:
  21             break
  22         if len(sig) != 4 or not sig.startswith(b"PK"):
  23             raise ValueError("Invalid Zip record signature")
  24         if sig != FILE_SIG:
  25             break
  26         yield read_file_header(zip)
  27
  28 def read_file_header(zip):
  29     (_, _, flags, method, _, crc, comp, uncomp, namelen, extra) = (
  30         FILE_STRUCT.unpack(zip.read(FILE_STRUCT.size)))
  31
  32     name = zip.read(namelen)
  33     if len(name) != namelen:
  34         raise EOFError("EOF in file name")
  35     encoding = "utf-8" if flags >> FLAGS_UTF8 & 1 else "cp437"
  36     name = name.decode(encoding)
  37
  38     if flags >> FLAGS_STREAMED & 1:
  39         raise NotImplementedError("Data descriptor")
  40     return (flags, method, crc, comp, uncomp, name, extra)
  41
  42 def extract():
  43     zip = sys.stdin.buffer
  44     out = sys.stdout.buffer
  45
  46     for [flags, method, crc, comp, uncomp, name, extra] in iter_files(zip):
  47         if name:
  48             print(name, file=sys.stderr)
  49
  50         streamcopy(zip, DelegateWriter(), extra)
  51
  52         skip = False
  53         if flags & (1 << FLAGS_ENCRYPTED | 1 << FLAGS_STRONG_ENC):
  54             print("Encrypted file", file=sys.stderr)
  55             skip = True
  56         if method != METHOD_STORE:
  57             print("Compression method {!r}".format(method), file=sys.stderr)
  58             skip = True
  59         if name.endswith((".txt", ".URL")):
  60             skip = True
  61
  62         if skip:
  63             streamcopy(zip, DelegateWriter(), comp)
  64         else:
  65             crcwriter = Crc32Writer()
  66             streamcopy(zip, DelegateWriter(out, crcwriter), uncomp)
  67             if crcwriter.crc != crc:
  68                 fmt = "Extracted CRC {:08X} != expected CRC {:08X}"
  69                 print(fmt.format(crcwriter.crc, crc), file=sys.stderr)
  70
  71             streamcopy(zip, DelegateWriter(), comp - uncomp)
  72
  73 class Crc32Writer(BufferedIOBase):
  74     def __init__(self):
  75         self.crc = 0
  76     def write(self, b):
  77         self.crc = crc32(b, self.crc)
  78
  79 if __name__ == "__main__":
  80     extract()