from statsmodels.compat.pandas import MONTH_END

import pandas as pd
import pytest

from statsmodels.datasets import co2, macrodata
from statsmodels.tsa.x13 import (
    _find_x12,
    x13_arima_analysis,
    x13_arima_select_order,
)

x13path = _find_x12()

pytestmark = pytest.mark.skipif(x13path is False,
                                reason='X13/X12 not available')

dta = macrodata.load_pandas().data
index = pd.period_range(start='1959Q1', end='2009Q3', freq='Q')
dta.index = index
quarterly_data = dta.dropna()

dta = co2.load_pandas().data
dta['co2'] = dta.co2.interpolate()
monthly_data = dta.resample(MONTH_END)
# change in pandas 0.18 resample is deferred object
if not isinstance(monthly_data, (pd.DataFrame, pd.Series)):
    monthly_data = monthly_data.mean()

monthly_start_data = dta.resample('MS')
if not isinstance(monthly_start_data, (pd.DataFrame, pd.Series)):
    monthly_start_data = monthly_start_data.mean()

data = (monthly_data, monthly_start_data, monthly_data.co2,
        monthly_start_data.co2, quarterly_data.realgdp,
        quarterly_data[['realgdp']])
ids = ('monthly', 'monthly_start', 'monthly_co2', 'monthly_start_co2',
       'series', 'dataframe')


@pytest.fixture(params=data, ids=ids)
def dataset(request):
    return request.param


def test_x13_arima_select_order(dataset):
    res = x13_arima_select_order(dataset)
    assert isinstance(res.order, tuple)
    assert isinstance(res.sorder, tuple)


@pytest.mark.matplotlib
def test_x13_arima_plot(dataset):
    res = x13_arima_analysis(dataset)
    res.plot()