"""Feature matrix assembly tests (RESEARCH Pattern 0)."""

from pathlib import Path

import numpy as np
import pytest

from model.features import (
    ANOMALY_FEATURES,
    CATEGORICAL_FEATURES,
    CLASSES,
    CLASSIFIER_FEATURES,
    load_anomaly_features,
    load_split,
)
from model.synth.state_machines import GENERATORS


def test_classes_match_generators_order() -> None:
    """CLASSES MUST mirror GENERATORS insertion order (byte-identicality anchor)."""
    assert CLASSES == list(GENERATORS.keys())
    assert len(CLASSES) == 10


def test_classifier_features_shape() -> None:
    """20 features = 9 numerics + 8 categoricals + 3 misc."""
    assert len(CLASSIFIER_FEATURES) == 20
    assert len(ANOMALY_FEATURES) == 9
    assert len(CATEGORICAL_FEATURES) == 8


def test_classifier_features_no_leakage_columns() -> None:
    """`bssid` and `timestamp` MUST NOT be in CLASSIFIER_FEATURES (Pitfall 2)."""
    assert "bssid" not in CLASSIFIER_FEATURES
    assert "timestamp" not in CLASSIFIER_FEATURES


@pytest.mark.skipif(
    not Path("data/train.parquet").exists(),
    reason="data/train.parquet not generated (run `make synth` first)",
)
def test_load_split_train_shape() -> None:
    X, y, names = load_split(Path("data/train.parquet"))
    assert X.dtype == np.float64
    assert y.dtype == np.int64
    assert X.shape[0] == 3_000_000  # 10k samples × 10 classes × 30 frames
    assert X.shape[1] == len(CLASSIFIER_FEATURES)
    assert names == list(CLASSIFIER_FEATURES)
    assert y.min() >= 0 and y.max() < len(CLASSES)


@pytest.mark.skipif(
    not Path("data/train.parquet").exists(),
    reason="data/train.parquet not generated (run `make synth` first)",
)
def test_load_anomaly_features_shape() -> None:
    X_anom, y, ts = load_anomaly_features(Path("data/train.parquet"))
    assert X_anom.shape[0] == 3_000_000
    assert X_anom.shape[1] == len(ANOMALY_FEATURES)
    assert X_anom.dtype == np.float64
    assert ts.shape[0] == 3_000_000
    # All real classes (not -1 baseline) on train.parquet
    assert (y >= 0).all() and (y < len(CLASSES)).all()