"""Compute per-feature population statistics from training CSV.

Produces feature_stats_v1.json with mean, std, min, max, median, q25, q75
for each feature column. Used by XAI service for z-score computation.

Usage:
    python -m app.training.compute_feature_stats DataSet/features.csv
"""

from __future__ import annotations

import csv
import json
import sys
from pathlib import Path

import numpy as np


def compute_stats(csv_path: str | Path, output_path: str | Path) -> None:
    csv_path = Path(csv_path)
    output_path = Path(output_path)

    with open(csv_path, "r", encoding="utf-8") as f:
        reader = csv.DictReader(f)
        rows = list(reader)
        fieldnames = reader.fieldnames or []

    excluded = {"file_path", "label_int", "duration_sec", "sample_rate"}
    feature_cols = [c for c in fieldnames if c not in excluded]

    stats: dict[str, dict[str, float]] = {}

    for col in feature_cols:
        values = []
        for row in rows:
            try:
                v = float(row[col])
                if not (np.isnan(v) or np.isinf(v)):
                    values.append(v)
            except (ValueError, TypeError):
                continue
        if not values:
            stats[col] = {
                "mean": 0.0, "std": 1.0,
                "min": 0.0, "max": 1.0,
                "median": 0.0, "q25": 0.0, "q75": 0.0,
                "count": 0,
            }
            continue
        arr = np.array(values)
        stats[col] = {
            "mean": float(np.mean(arr)),
            "std": float(np.std(arr)) or 1.0,
            "min": float(np.min(arr)),
            "max": float(np.max(arr)),
            "median": float(np.median(arr)),
            "q25": float(np.percentile(arr, 25)),
            "q75": float(np.percentile(arr, 75)),
            "count": len(values),
        }

    # Also split by class for AI vs Human distributions
    by_class: dict[str, dict[str, dict[str, float]]] = {"ai": {}, "human": {}}
    for cls_label, cls_int in (("ai", "1"), ("human", "0")):
        for col in feature_cols:
            values = []
            for row in rows:
                if row.get("label_int") != cls_int:
                    continue
                try:
                    v = float(row[col])
                    if not (np.isnan(v) or np.isinf(v)):
                        values.append(v)
                except (ValueError, TypeError):
                    continue
            if values:
                arr = np.array(values)
                by_class[cls_label][col] = {
                    "mean": float(np.mean(arr)),
                    "std": float(np.std(arr)) or 1.0,
                    "median": float(np.median(arr)),
                }

    output = {**stats, "_by_class": by_class}
    output_path.parent.mkdir(parents=True, exist_ok=True)
    with open(output_path, "w", encoding="utf-8") as f:
        json.dump(output, f, indent=2)
    print(f"Wrote {output_path} ({len(stats)} features, {len(rows)} samples)")


if __name__ == "__main__":
    csv_in = sys.argv[1] if len(sys.argv) > 1 else "../DataSet/features.csv"
    out = sys.argv[2] if len(sys.argv) > 2 else "models/feature_stats_v1.json"
    compute_stats(csv_in, out)