Python malware extraction for 2018-07-05 HWP ROKRAT dropper

#extract malware from: 9e6ff58202f6c1bd2381e8209231efd0ef6855db59db975fb5b75041706ed104
import re
import sys
import zlib
import struct
import hashlib
import oledump
import olefile
import binascii
import cStringIO

from hexdump import hexdump
def md5(content):
    return hashlib.md5(content).hexdigest()

def sha256(content):
    return hashlib.sha256(content).hexdigest()

def writefile(filename,data):
    with open(filename, 'wb') as f:
        f.write(data)

def extract_payload(filename,postscript):

    if re.search(' corpqe\. ',postscript):
        print '[+] Found CVE-2017-8291 (reversed), attempting extraction...'
        if re.search('adobeprint token 1 pop pop exch pop exec \x0a',postscript):
            print '[+] Found PS prior to encoded shellcode and payload, extracting...'
            enc_shellcode_payload = re.search('adobeprint token 1 pop pop exch pop exec \x0a(.+)$',postscript, re.DOTALL).group(1)
            #simple size check
            if len(enc_shellcode_payload) > 100:
                key = ord(enc_shellcode_payload[0])
                print '[+] XOR key for encoded shellcode and payload: 0x%02X' % key
                enc_shellcode_payload = enc_shellcode_payload[1:]
                dec_shellcode_payload = ''
                for c in enc_shellcode_payload:
                    dec_shellcode_payload += chr(ord(c)^key)
                dec_shellcode_payload = bytes(dec_shellcode_payload)
                if re.search('\\x8b\\x04\\x24\\xc3\\xe8',dec_shellcode_payload, re.S):
                    print '[+] Found expected opcode bytes before encoded payload, extracting and decoding payload...'
                    payload_re = re.compile('\\x8b\\x04\\x24\\xc3\\xe8.{4}\\xc3(.{4})(.{4})')
                    payload_size = struct.unpack('<I',payload_re.search(dec_shellcode_payload).group(1))[0]
                    print '[+] Payload size: %i' % payload_size
                    key_payload = payload_re.search(dec_shellcode_payload).group(2)
                    print '[+] Payload 4-byte XOR key: 0x%02X%02X%02X%02X' % (ord(key_payload[0]),ord(key_payload[1]),ord(key_payload[2]),ord(key_payload[3]))
                    decoded = ''
                    payload_pos = payload_re.search(dec_shellcode_payload).span(2)[1]
                    encoded_payload = dec_shellcode_payload[payload_pos:payload_pos+payload_size]
                    for i in range(len(encoded_payload)):
                        decoded += chr(ord(encoded_payload[i])^ord(key_payload[i%4]))
                else:
                    print '[!] Did not find the expected opcode bytes before encoded payload after decoding, exiting...'
                    sys.exit()
        else:
            print '[!] Did not find PS before encoded payload, exiting...'
            sys.exit()


    else:
        print '[!] Did not find expected reversed ".eqproc", exiting...'
        sys.exit()

    md5_decoded = md5(decoded)
    sha256_decoded = sha256(decoded)
    decoded_filename = sha256_decoded

    print '[+] Decoded payload hashes:'
    print '%s' % md5_decoded
    print '%s' % sha256_decoded
    print '[+] Writing decoded file to: %s\n' % decoded_filename
    with open(decoded_filename, 'wb') as f:
        f.write(decoded)

def main():
    #code taken from Didier Stevens oledump script
    #https://raw.githubusercontent.com/DidierStevens/DidierStevensSuite/master/oledump.py
    filename = sys.argv[1]
    oStringIO = cStringIO.StringIO(open(filename, 'rb').read())
    magic = oStringIO.read(6)
    oStringIO.seek(0)
    if magic[0:4] != oledump.OLEFILE_MAGIC:
        print '[!] This only supports OLE, exiting...'
        sys.exit()
    ole = olefile.OleFileIO(oStringIO)
    for orphan, fname, entry_type, stream in oledump.OLEGetStreams(ole):
        if fname[0] == 'BinData':
            if re.search('BIN00\d{2}\.[Ee]?[Pp][Ss]',fname[1]):
                scriptname = fname[1]
                if '.eps' in scriptname:
                    print '[+] Found EPS file: %s' % scriptname
                elif '.ps' in scriptname:
                    print '[+] Found PS file: %s' % scriptname
                try:
                    decompressed = zlib.decompress(stream, -15)
                except Exception, e:
                    print '[!] Problem decompressing zlib stream, error:'
                    print e
                extract_payload(filename,decompressed)

if __name__ == '__main__':
    main()