#!/usr/bin/env python3
"""Recover the kit caption/localisation vocabulary from the UNPACKED CardsDLL.

Why CardsDLL and not FIFA17.exe: CardsDLL is not packed, so a MISS here is
meaningful. FIFA17.exe is Denuvo-packed and only partially readable -- a hit
there is useful, a miss proves nothing. Every run therefore prints a positive
control first; if the control fails, the run is void and no negative may be
quoted from it.

Usage: python3 cardsdll_kit_strings.py [path-to-CardsDLL]
"""
from __future__ import annotations

import os
import re
import sys

DEFAULT = os.path.expanduser(
    "~/.cache/openfut-investigation/bin/CardsDLL_Win64_retail.dll"
)

# Strings that MUST be present. If any is missing the search is broken.
CONTROLS = [b"activeHomeKit", b"cardsubtypeid", b"resourceId", b"activeAwayKit"]

# The kit caption vocabulary this project has referred to, plus neighbours worth
# knowing about either way.
PROBES = [
    b"FUT_UC_KITS", b"TeamName_Abbr15_", b"TeamName_Abbr15", b"TeamName_",
    b"FUT_UC_", b"StadiumName_", b"Badge", b"Stadium",
    b"activeBadge", b"activeBall", b"activeStadium",
    b"kit", b"Kit", b"KIT",
    b"home", b"Home", b"HOME", b"away", b"Away", b"AWAY",
    b"locked", b"Locked", b"LOCKED", b"unlock",
    b"category", b"year", b"teamid", b"teamId",
    b"DataProvider", b"itemData", b"itemType", b"itemState",
]


def ascii_strings(data, minlen=4):
    for m in re.finditer(rb"[ -~]{%d,}" % minlen, data):
        yield m.start(), m.group()


def main():
    path = sys.argv[1] if len(sys.argv) > 1 else DEFAULT
    data = open(path, "rb").read()
    print(f"{os.path.basename(path)}  {len(data)} bytes")

    print("\n-- positive control (a miss voids every negative below) --")
    ok = True
    for c in CONTROLS:
        n = data.count(c)
        print(f"   {c.decode():16s} {n}")
        if n == 0:
            ok = False
    if not ok:
        print("   CONTROL FAILED — do not quote negatives from this run.")
        return 1

    print("\n-- probe counts --")
    for p in PROBES:
        print(f"   {p.decode():18s} {data.count(p)}")

    # Whole-string table: every standalone string containing kit-ish substrings.
    print("\n-- standalone strings matching kit/team/caption vocabulary --")
    pat = re.compile(rb"(?i)(kit|teamname|abbr|stadiumname|fut_uc|locked|unlock)")
    seen = set()
    for off, s in ascii_strings(data, 5):
        if pat.search(s) and s not in seen:
            seen.add(s)
            print(f"   @{off:#08x}  {s.decode('latin1')[:110]}")
    print(f"   ({len(seen)} distinct)")
    return 0


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