import pytest
import pandas as pd
from scheduling_analysis import core

@pytest.fixture
def sample_data():
    return [
        {"run_id": "r1", "metric_name": "max_outlier_ms", "metric_value": 10.0, "mechanism": "sched_a"},
        {"run_id": "r2", "metric_name": "max_outlier_ms", "metric_value": 25.0, "mechanism": "sched_b"},
        {"run_id": "r1", "metric_name": "band_width_h", "metric_value": 5.0, "mechanism": "sched_a"},
        {"run_id": "r2", "metric_name": "band_width_h", "metric_value": 7.5, "mechanism": "sched_b"}
    ]


def test_analyze_scheduling_effects_basic(sample_data):
    result = core.analyze_scheduling_effects(sample_data)
    assert isinstance(result, dict)
    assert set(result.keys()) == {"max_outlier_effect", "resonance_band_shift", "metrics"}

    # numerical plausibility
    assert isinstance(result["max_outlier_effect"], float)
    assert isinstance(result["resonance_band_shift"], float)
    assert abs(result["max_outlier_effect"] - 15.0) < 1e-6
    assert abs(result["resonance_band_shift"] - 2.5) < 1e-6


def test_empty_data_raises():
    with pytest.raises(ValueError):
        core.analyze_scheduling_effects([])


def test_invalid_data_type():
    # Not a list should raise a TypeError
    with pytest.raises(TypeError):
        core.analyze_scheduling_effects("invalid")


def test_partial_metrics_present():
    data = [
        {"run_id": "r1", "metric_name": "max_outlier_ms", "metric_value": 10.0, "mechanism": "sched_a"},
        {"run_id": "r1", "metric_name": "band_width_h", "metric_value": 5.0, "mechanism": "sched_a"}
    ]
    result = core.analyze_scheduling_effects(data)
    assert result["max_outlier_effect"] == pytest.approx(0.0)
    assert result["resonance_band_shift"] == pytest.approx(0.0)
    assert isinstance(result["metrics"], dict)