#!/usr/bin/env python3
# Author: Simon-Pierre Boucher — simon-pierre.boucher@uqo.ca
"""Baseline feature matrix with as-of comparable summaries.

    python3 scripts/03_features.py [--synthetic] [--no-geo]
→ data/{processed|synthetic}/feature_matrix.parquet (+ results/data/feature_matrix_summary*.json).
Geo features are joined when data/<base>/geo_features.parquet exists (run scripts/04_geo.py first).
"""
from __future__ import annotations

import argparse
import json
import time

import polars as pl

from agentic_valuation.config import PROCESSED, RESULTS, SYNTHETIC
from agentic_valuation.features.matrix import build_feature_matrix, feature_columns


def main() -> None:
    ap = argparse.ArgumentParser(description=__doc__)
    ap.add_argument("--synthetic", action="store_true")
    ap.add_argument("--no-geo", action="store_true", help="ignore geo_features.parquet even if present")
    args = ap.parse_args()
    base = SYNTHETIC if args.synthetic else PROCESSED
    t0 = time.perf_counter()
    analytic = pl.read_parquet(base / "analytic.parquet")
    splits = pl.read_parquet(base / "splits.parquet")
    geo = None
    geo_path = base / "geo_features.parquet"
    if geo_path.exists() and not args.no_geo:
        geo = pl.read_parquet(geo_path)
    fm = build_feature_matrix(analytic, splits, geo=geo, comps_summary=True)
    fm.write_parquet(base / "feature_matrix.parquet", compression="zstd")
    num, cat = feature_columns(fm)
    summary = {
        "synthetic": args.synthetic,
        "rows": fm.height,
        "by_split": {k: int(v) for k, v in fm.group_by("split").len().iter_rows()},
        "numeric_features": num,
        "categorical_features": cat,
        "geo_joined": geo is not None,
        "geo_rows": None if geo is None else geo.height,
        "missingness_pct": {c: round(100.0 * fm[c].null_count() / fm.height, 3) for c in num + cat},
        "nb_n_12m_quantiles": {q: float(fm["nb_n_12m"].quantile(q)) for q in (0.05, 0.25, 0.5, 0.75, 0.95)},
        "nb_mean_dist_km_median": float(fm["nb_mean_dist_km"].median()),
        "seconds": round(time.perf_counter() - t0, 2),
    }
    out = RESULTS / "data" / ("feature_matrix_summary_synthetic.json" if args.synthetic else "feature_matrix_summary.json")
    out.parent.mkdir(parents=True, exist_ok=True)
    out.write_text(json.dumps(summary, indent=1, ensure_ascii=False))
    print(json.dumps({k: summary[k] for k in ("rows", "by_split", "geo_joined", "nb_n_12m_quantiles", "nb_mean_dist_km_median", "seconds")}, ensure_ascii=False, indent=1))


if __name__ == "__main__":
    main()