Jar Infection Checker (python)

scanner.py

# Adapted by @ivanpu to Python from Overwolf's scanner, because I couldn't launch it
# Original code: https://github.com/overwolf/jar-infection-scanner/
# Check for updates to this script: https://gist.github.com/ivanpu/c5347bf107fa900ac79f1fcf2f056e7c

from __future__ import annotations

import zipfile
from argparse import ArgumentParser
from pathlib import Path

SIGNATURES: list[bytes] = [
    b"\x38\x54\x59\x04\x10\x35\x54\x59\x05\x10\x2E\x54\x59\x06\x10\x32\x54\x59\x07\x10\x31\x54\x59\x08\x10\x37\x54\x59\x10\x06\x10\x2E\x54\x59\x10\x07\x10\x31\x54\x59\x10\x08\x10\x34\x54\x59\x10\x09\x10\x34\x54\x59\x10\x0A\x10\x2E\x54\x59\x10\x0B\x10\x31\x54\x59\x10\x0C\x10\x33\x54\x59\x10\x0D\x10\x30\x54\xB7",
    b"\x68\x54\x59\x04\x10\x74\x54\x59\x05\x10\x74\x54\x59\x06\x10\x70\x54\x59\x07\x10\x3a\x54\x59\x08\x10\x2f\x54\x59\x10\x06\x10\x2f\x54\x59\x10\x07\x10\x66\x54\x59\x10\x08\x10\x69\x54\x59\x10\x09\x10\x6c\x54\x59\x10\x0a\x10\x65\x54\x59\x10\x0b\x10\x73\x54\x59\x10\x0c\x10\x2e\x54\x59\x10\x0a\x10\x73\x54\x59\x10\x0e\x10\x6b\x54\x59\x10\x0f\x10\x79\x54\x59\x10\x10\x10\x72\x54\x59\x10\x11\x10\x61\x54\x59\x10\x12\x10\x67\x54\x59\x10\x13\x10\x65\x54\x59\x10\x14\x10\x2e\x54\x59\x10\x15\x10\x64",
    b"\x2d\x54\x59\x04\x10\x6a\x54\x59\x05\x10\x61\x54\x59\x06\x10\x72",
]


def check_jar_file(jar_file_path: Path) -> bool:
    def check_zip_dir(path: zipfile.Path) -> bool:
        for entry in path.iterdir():
            if entry.is_file() and entry.name.endswith(".class"):
                buffer = entry.read_bytes()
                if any(sig in buffer for sig in SIGNATURES):
                    return True
            elif entry.is_dir():
                if check_zip_dir(entry):
                    return True
        return False

    try:
        return check_zip_dir(zipfile.Path(jar_file_path))
    except Exception as e:
        print(f"Error while extracting {jar_file_path}: {e}")
    return False


def main():
    parser = ArgumentParser()
    parser.add_argument("directories", nargs="+", type=Path, help="directories to scan", metavar="directory")

    args = parser.parse_args()
    for path in args.directories:  # type: Path
        jars = path.rglob("*.[jJ][aA][rR]")
        for jar in jars:
            if check_jar_file(jar):
                print("!!! INFECTED:", jar)


if __name__ == "__main__":
    main()

While I did try to make the code backwards compatible as much as possible (I was assuming python 3.7+), I missed to check the zipfile documentation thoroughly:
zipfile.Path was added in Python 3.8
zipfile.Path.suffix was added in Python 3.11

@silmaril42
If you want something a bit more elegant than you currently have, you can use this:

if entry.is_file() and entry.name.endswith(".class"):

Which I think I'll actually update into the script, to lower the Python dependency to 3.8.

Yes, this looks much nicer! 😃 👍