Pure python reimplementation of .cpio.xz content extraction from pbzx file payload for OS X packages.without any third party tools

parse_pbzx.py

#!/usr/bin/python
# tar -Oxf /Volumes/install_app/Packages/Essentials.pkg Payload|python pbzx.py|tar t

import subprocess
import struct
import sys

def pbzx():
    if sys.stdin.read(4) != "pbzx":
        print "Not a pbzx stream"
        sys.exit(1)

    flags = struct.unpack(">Q", sys.stdin.read(8))[0]
    print >> sys.stderr, hex(flags)

    while (flags & 1 << 24):
        flags, length = struct.unpack(">QQ", sys.stdin.read(16))
        xz = None
        while length:
            xbuf = sys.stdin.read(min(4*1024, length))
            if xz == None:
                xz = xbuf[:8] == "\xfd7zXZ\x00\x00\x00"
                if xz:
                    p = subprocess.Popen("zcat", stdin=subprocess.PIPE)
            length -= len(xbuf)
            if xz:
                p.stdin.write(xbuf)
            else:
                sys.stdout.write(xbuf)
        if xz:
            p.stdin.close()
            p.wait()
        # This is required because Popen seems to flush stdout
        sys.stdout.flush()

if __name__ == "__main__":
    pbzx()