"""E2c (extension): does the real fly wiring classify better than random wiring?

    python experiments/e2c_connectome.py data/demo_calibration --group date
    python experiments/e2c_connectome.py data/demo_banana --group batch

Three mushroom bodies, everything else identical:
    隨機接線      textbook model, 5 different random seeds
    連結體接線    data/connectome/kc_wiring.npy (run connectome_wiring.py first)
    打亂連結體    same synapse counts per KC, sent to random glomeruli (5 seeds)
If 連結體 beats 打亂連結體, the specific fly wiring matters, not just its statistics.
"""
from __future__ import annotations

import argparse
import sys
from pathlib import Path

import numpy as np
import pandas as pd
from sklearn.model_selection import LeaveOneGroupOut, cross_val_score

sys.path.insert(0, str(Path(__file__).resolve().parents[1]))

from enose.common import load_table, output_dir  # noqa: E402
from enose.connectome import shuffled_wiring  # noqa: E402
from enose.mushroom import MushroomBodyClassifier  # noqa: E402

WIRING = Path(__file__).resolve().parents[1] / "data" / "connectome" / "kc_wiring.npy"
SEEDS: tuple[int, ...] = (0, 1, 2, 3, 4)


def main() -> None:
    parser = argparse.ArgumentParser(description=__doc__, formatter_class=argparse.RawDescriptionHelpFormatter)
    parser.add_argument("folder", type=Path)
    parser.add_argument("--group", default="date")
    arguments = parser.parse_args()
    table, features, labels, _ = load_table(arguments.folder / "features.csv")
    if "rh_level" in table:
        keep = (table["rh_level"] == "mid").to_numpy()
        table, features, labels = table[keep], features[keep], labels[keep]
    groups = table[arguments.group].to_numpy()
    wiring = np.load(WIRING)

    def score(model: MushroomBodyClassifier) -> float:
        return cross_val_score(model, features, labels, groups=groups, cv=LeaveOneGroupOut()).mean()

    rows = [{"接線": "連結體接線", "seed": "-", "準確率": score(MushroomBodyClassifier(connectivity=wiring))}]
    for seed in SEEDS:
        rows.append({"接線": "隨機接線", "seed": seed,
                     "準確率": score(MushroomBodyClassifier(kenyon_cells=wiring.shape[0], random_state=seed))})
        rows.append({"接線": "打亂連結體", "seed": seed,
                     "準確率": score(MushroomBodyClassifier(connectivity=shuffled_wiring(wiring, seed)))})
    result = pd.DataFrame(rows)
    result.to_csv(output_dir(arguments.folder / "results") / "e2c_connectome.csv", index=False)
    print(result.groupby("接線")["準確率"].agg(["mean", "std", "count"]).round(3).to_string())
    print("\n注意：模擬資料的差異不代表真實結果；隨機與打亂組的標準差代表「換一種接線」本身造成的差異。")


if __name__ == "__main__":
    main()