#!/usr/bin/env python3
"""Look at the constant pool of a class of the remake and see what is hidden.

The remake obfuscates the string constants: they are not readable in the class
file.  This script pulls the UTF-8 entries out of the constant pool and tries
to work out how they were scrambled, because the same cipher is probably what
protects the .obj and .rrs files.

Usage:  inspect-constants.py <class file> [...]
"""

import struct
import sys
from pathlib import Path

TAGS = {
    1: "Utf8", 3: "Integer", 4: "Float", 5: "Long", 6: "Double",
    7: "Class", 8: "String", 9: "Fieldref", 10: "Methodref", 11: "InterfaceMethodref",
    12: "NameAndType", 15: "MethodHandle", 16: "MethodType", 17: "Dynamic",
    18: "InvokeDynamic", 19: "Module", 20: "Package",
}


def read_pool(data):
    """The constant pool of a class: a list of (tag, value) pairs."""
    if data[:4] != b"\xca\xfe\xba\xbe":
        raise ValueError("not a class file")
    count = struct.unpack_from(">H", data, 8)[0]
    pool = []
    pos = 10
    i = 1
    while i < count:
        tag = data[pos]
        pos += 1
        if tag == 1:
            length = struct.unpack_from(">H", data, pos)[0]
            pos += 2
            pool.append((tag, data[pos:pos + length]))
            pos += length
        elif tag in (3, 4):
            pool.append((tag, data[pos:pos + 4]))
            pos += 4
        elif tag in (5, 6):
            pool.append((tag, data[pos:pos + 8]))
            pos += 8
        elif tag in (7, 8, 16, 19, 20):
            pool.append((tag, struct.unpack_from(">H", data, pos)[0]))
            pos += 2
        elif tag in (9, 10, 11, 12, 17, 18):
            pool.append((tag, struct.unpack_from(">HH", data, pos)))
            pos += 4
        elif tag == 15:
            pool.append((tag, data[pos:pos + 3]))
            pos += 3
        else:
            raise ValueError("unknown tag %d at %d" % (tag, pos))
        if tag in (5, 6):
            i += 2
        else:
            i += 1
    return pool, pos


def printable(data):
    good = 0
    for b in data:
        if 32 <= b < 127:
            good += 1
    return good / max(1, len(data))


def crack(text):
    """The single byte key that makes a string readable, or None."""
    best = None
    for key in range(1, 256):
        out = bytes(b ^ key for b in text)
        score = printable(out)
        if score > 0.98 and all(32 <= b < 127 or b in (9, 10, 13) for b in out):
            best = (key, out)
            break
    return best


def main(argv):
    if len(argv) < 2:
        print(__doc__)
        return 1
    for name in argv[1:]:
        path = Path(name)
        data = path.read_bytes()
        pool, _ = read_pool(data)
        strings = [v for t, v in pool if t == 1]
        readable = [s for s in strings if printable(s) > 0.95]
        hidden = [s for s in strings if printable(s) <= 0.95]
        print("=== %s: %d strings, %d readable, %d hidden"
              % (path.name, len(strings), len(readable), len(hidden)))
        for s in readable[:12]:
            print("   plain  %r" % s[:60])
        for s in hidden[:8]:
            cracked = crack(s)
            if cracked:
                print("   xor    %r  key 0x%02X" % (s[:40], cracked[0]))
            else:
                print("   hidden %r" % s[:40])
    return 0


if __name__ == "__main__":
    sys.exit(main(sys.argv))