test_features.py

import numpy as np

from keywraith.synthetic import KeystrokeStreamGenerator, MachineProfile
from keywraith.features import (
    extract_features,
    features_to_vector,
    sessions_to_matrix,
    FEATURE_NAMES,
)


def test_feature_vector_shape_and_order():
    gen = KeystrokeStreamGenerator(seed=1)
    feats = extract_features(gen.human_session(100))
    assert set(feats) == set(FEATURE_NAMES)
    vec = features_to_vector(feats)
    assert vec.shape == (len(FEATURE_NAMES),)


def test_burst_has_low_cv_and_high_superhuman_fraction():
    gen = KeystrokeStreamGenerator(seed=2)
    burst = extract_features(
        gen.machine_session(300, MachineProfile(mode="burst", device_delay_ms=5.0))
    )
    human = extract_features(gen.human_session(300))
    assert burst["iki_cv"] < human["iki_cv"]
    assert burst["frac_iki_sub20ms"] > 0.9
    assert human["frac_iki_sub20ms"] < 0.1


def test_human_has_corrections_machine_does_not():
    gen = KeystrokeStreamGenerator(seed=3)
    human = extract_features(gen.human_session(600))
    machine = extract_features(gen.machine_session(600))
    assert machine["correction_rate"] == 0.0
    assert human["correction_rate"] >= 0.0  # may be 0 in a short draw, but non-negative


def test_matrix_labels():
    gen = KeystrokeStreamGenerator(seed=4)
    ds = gen.dataset(n_human=15, n_machine=15)
    X, y, ids = sessions_to_matrix(ds)
    assert X.shape == (30, len(FEATURE_NAMES))
    assert set(y.tolist()) == {0, 1}
    assert len(ids) == 30


def test_empty_session_is_safe():
    from keywraith.synthetic import Session

    feats = extract_features(Session("empty", "human", "human", []))
    assert all(np.isfinite(v) for v in feats.values())