Convert strings with an unknown number of hex strings embedded in them to strings using regex

So I have a list of strings (content from Snort rules), and I am trying to convert the hex portions of them to UTF-8/ASCII, so I can send the content over netcat.

The method I have now works fine for strings with single hex characters (i.e. 3A), but breaks when there’s a series of hex characters (i.e. 3A 4B 00 FF)

My current solution is:

import re
import codecs

def convert_hex(match):
  string = match.group(1)
  string = string.replace(" ", "")
  decode_hex = codecs.getdecoder("hex_codec")
  try:
    result = decode_hex(string)[0]
  except:
    result = bytes.fromhex((lambda s: ("%s%s00" * (len(s)//2)) % tuple(s))(string)).decode('utf-16-le')
  return result.decode("utf-8")


strings = ['|0A|Referer|3A| res|3A|/C|3A|', 'RemoteNC Control Password|3A|', '/bbs/search.asp', 'User-Agent|3A| Mozilla/4.0 |28|compatible|3B| MSIE 5.0|3B| Windows NT 5.0|29|']

converted_strings = []

for string in strings:
    for i in range(len(string)):
        string = re.sub(r"|(.{2})|", convert_hex, string)
    converted_strings.append(string)

For the strings in strings, this works, but for a string like:

|08 00 00 00 27 C7 CC 6B C2 FD 13 0E|

it breaks.

I tried changing the regex to:

re.sub(r"|.*([A-Fa-f0-9]{2}).*|")

but that only converts the last hex.

I need this solution to work for strings like Hello|3A|World, |3A 00 FF|, and Hello|3A 00|World

I know it’s an issue with the regexp, but I’m not sure what exactly.

Any help would be much appreciated.

Answer

It looks like a substring is either always hex i.e. (?:[A-Fa-f0-9]{2}s)+[A-Fa-f0-9]{2} or not hex at all between | symbols?

This works:

for string in strings:
    for i in range(len(string)):
        string = re.sub(r"(?<=|)((?:[A-Fa-f0-9]{2}s)*[A-Fa-f0-9]{2})(?=|)", convert_hex, string)
    converted_strings.append(string)

(extra parentheses for a capturing group 1 – you could leave out one pair of parentheses and change your function to act on group(0) instead)

But it breaks on your example |08 00 00 00 27 C7 CC 6B C2 FD 13 0E|, as that doesn’t appear to be a valid UTF-8 encoding. The resulting error:

UnicodeDecodeError: 'utf-8' codec can't decode byte 0xc7 in position 5: invalid continuation byte

However, a valid UTF-8 encoded multi-byte string like '|74 65 73 74 20 f0 9f 98 80|' works just fine:

import re
import codecs

def convert_hex(match):
  string = match.group(1)
  string = string.replace(" ", "")
  decode_hex = codecs.getdecoder("hex_codec")
  try:
    result = decode_hex(string)[0]
  except:
    result = bytes.fromhex((lambda s: ("%s%s00" * (len(s)//2)) % tuple(s))(string)).decode('utf-16-le')
  return result.decode("utf-8")


strings = ['|74 65 73 74 20 f0 9f 98 80|']

converted_strings = []

for string in strings:
    for i in range(len(string)):
        string = re.sub(r"(?<=|)((?:[A-Fa-f0-9]{2}s)*[A-Fa-f0-9]{2})(?=|)", convert_hex, string)
    converted_strings.append(string)

print(converted_strings)

Result:

['|test 😀|']

If you don’t really need a printable representation of the data, you could just have your function return the bytes object and only apply the function to matching parts – instead of constructing a new string.

Based on what @Selcuk was saying, perhaps a result with byte-strings makes more sense – this works on all three types of input:

import re
import codecs

def convert_hex(match):
  string = match.group(1)
  string = string.replace(b" ", b"")
  decode_hex = codecs.getdecoder("hex_codec")
  try:
    result = decode_hex(string)[0]
  except:
    result = bytes.fromhex((lambda s: ("%s%s00" * (len(s)//2)) % tuple(s))(string)).decode('utf-16-le')
  return result


strings = ['|0A|Referer|3A| res|3A|/C|3A|', '|74 65 73 74 20 f0 9f 98 80|', '|08 00 00 00 27 C7 CC 6B C2 FD 13 0E|']

converted_strings = []

for string in strings:
    string = re.sub(rb"(?<=|)((?:[A-Fa-f0-9]{2}s)*[A-Fa-f0-9]{2})(?=|)", convert_hex, string.encode())
    converted_strings.append(string)

print(converted_strings)

Result:

[b'|n|Referer|:| res|:|/C|:|', b'|test xf0x9fx98x80|', b"|x08x00x00x00'xc7xcckxc2xfdx13x0e|"]

No encoding issues, because no encoding is chosen. (Note that I didn’t attempt to change convert_hex too much – there’s some encoding juggling in there that you may need to look at, I just got it to work for bytes)