Dump a .pyc file to XML, with decompilation.

pyc2xml.py

import pdb
import cgi
import dis
import time
import struct
import marshal

def dump_code_object(code, ident=0):
    print '%s<code_object>' % (' '*ident,)
    ident += 2
    print '%s<co_argcount type="int">%d</co_argcount>' % (' '*ident, code.co_argcount)

    print '%s<co_cellvars type="tuple">' % (' '*ident)
    for cellvar in code.co_cellvars:
        print '%s<co_cellvar>%s</co_cellvar>' % (' '*(ident+2), cellvar)
    print '%s</co_cellvars>' % (' '*ident)


    print '%s<co_code>' % (' '*ident)
    print '%s<disassemble>' % (' '*(ident+2))
    print '<![CDATA['
    dis.dis(code)
    print ']]>'
    print '%s</disassemble>' % (' '*(ident+2))
    print '%s<binary>' % (' '*(ident+2))
    print '%s%r' % (' '*(ident+4), code.co_code)
    print '%s</binary>' % (' '*(ident+2))
    print '%s</co_code>' % (' '*ident)

    print '%s<co_consts type="tuple">' % (' '*ident)
    for const in code.co_consts:
        if type(const) == type(code):
            print '%s<co_const>' % (' '*(ident+2),)
            dump_code_object(const, ident+4)
            print '%s</co_const>' % (' '*(ident+2),)
            #print '%s<co_const>%s</co_const>' % (' '*(ident+2), cgi.escape(str(const)))
        else:
            print '%s<co_const>%s</co_const>' % (' '*(ident+2), const)
    print '%s</co_consts>' % (' '*ident)

    print '%s<co_filename>%s</co_filename>' % (' '*ident, code.co_filename)

    print '%s<co_firstlineno type="int">%s</co_firstlineno>' % (' '*ident, code.co_firstlineno)

    print '%s<co_flags>%s</co_flags>' % (' '*ident, code.co_flags)

    print '%s<co_freevars type="tuple">' % (' '*ident)
    for freevar in code.co_freevars:
        print '%s<co_freevar>%s</co_freevar>' % (' '*(ident+2), freevar)
    print '%s</co_freevars>' % (' '*ident)

    print '%s<co_lnotab>%s</co_lnotab>' % (' '*ident, `code.co_lnotab`)

    print '%s<co_name type="str">%r</co_name>' % (' '*ident, cgi.escape(code.co_name))

    print '%s<co_names type="tuple">' % (' '*ident)
    for name in code.co_names:
        print '%s<co_name type="str">%r</co_name>' % (' '*(ident+2), cgi.escape(name))
    print '%s</co_names>' % (' '*ident)

    print '%s<co_nlocals type="int">%d</co_nlocals>' % (' '*ident, code.co_nlocals)

    print '%s<co_stacksize type="int">%d</co_stacksize>' % (' '*ident, code.co_stacksize)

    print '%s<co_varnames type="tuple">' % (' '*ident)
    for varname in code.co_varnames:
        print '%s<co_varname>%r</co_varname>' % (' '*(ident+2), varname)
    print '%s</co_varnames>' % (' '*ident)

    ident -= 2
    print '%s</code_object>' % (' '*ident,)

def dump_pyc(data):
    ident = 0

    print '<?xml version="1.0" encoding="UTF-8" ?>'
    print '%s<pyc>' % (' '*ident,)
    ident += 2

    magic_s = pycdata[0:4]
    magic = struct.unpack('=l', magic_s)[0] & 0xffff
    print '%s<magic type="int">%d</magic>' % (' '*ident, magic)

    time_s = pycdata[4:8]
    time_i = struct.unpack('=l', time_s)[0]
    ctime = time.ctime(time_i)
    print '%s<timestamp>' % (' '*ident,)
    print '%s<time>%s</time>' % (' '*(ident+2), time_i)
    print '%s<ctime>%s</ctime>' % (' '*(ident+2), ctime)
    print '%s</timestamp>' % (' '*ident,)

    code = marshal.loads(pycdata[8:])
    dump_code_object(code, ident)

    ident -= 2
    print '%s</pyc>' % (' '*ident,)

if __name__ == '__main__':
    pycdata = open('test.pyc').read()
    dump_pyc(pycdata)

#######################################################
# ['co_argcount', 'co_cellvars', 'co_code', 'co_consts', 'co_filename',
#  'co_firstlineno', 'co_flags', 'co_freevars', 'co_lnotab', 'co_name',
#  'co_names', 'co_nlocals', 'co_stacksize', 'co_varnames']