Deobfuscating a VBS Script With Custom Encoding

Published: 2023-06-14
Last Updated: 2023-06-14 07:11:20 UTC
by Didier Stevens (Version: 1)

A reader asked us for help with the deobfuscation of a VBS script (4cd33a3c0d5f655b1bec2be6cbde096ddef696fdcd1685a512703e08514346c0).

It contains line comments with gibberish (REM and ') to obfuscate the real script:

These can be removed with a grep -v and regexes ^REM and ^', like this:

This looks like an encoded payload. After these concatenations, we have more concatenations, but of single characters, and these do not seem to be encoded:

They can be grepped for:

And with re-search.py, the value of each string can be extracted (str-u: double-quoted string, unquoted):

All these lines can be joined together with sets.py:

This looks like a reversed URL. python-per-line.py can be used to reverse each line (line[::-1] is the Python expression to reverse a line):

A similar command can be used for the other variable (nP):

The obfuscated payload can be extracted like this:

I will dedicate a diary entry on analysis methods of this type of encoding, but let me already explain here very briefly how it works.

Every characted of the script is represented by a token consisting of two upper-case letters. For example, character e is represented by token IZ.

The first 256 tokens in the above string define the translation table, and the rest of the tokens are the encoded payload. Here is a Python script to do the decoding:

def SplitLength(str, length):
    return [str[i:i + length] for i in range(0, len(str), length)]

def Decode(str, sizeToken, sizeTokenTable):
    translation = str[:sizeToken * sizeTokenTable]
    code = str[sizeToken * sizeTokenTable:]
    
    dTranslate = {token:counter for counter, token in enumerate(SplitLength(translation, sizeToken))}
    
    return ''.join([chr(dTranslate[token]) for token in SplitLength(code, sizeToken)])

The "dTranslate =" line builds a translation dictionary out of the 256 first tokens.

And the next line does the translation of the remaining tokens.