from __future__ import annotations

import itertools
import pickle

from typing import Any
from unittest.mock import patch, Mock

from datetime import datetime, date, timedelta

import numpy as np
from numpy.testing import (assert_array_equal, assert_approx_equal,
                           assert_array_almost_equal)
import pytest

from matplotlib import _api, cbook
import matplotlib.colors as mcolors
from matplotlib.cbook import delete_masked_points, strip_math


class Test_delete_masked_points:
    def test_bad_first_arg(self):
        with pytest.raises(ValueError):
            delete_masked_points('a string', np.arange(1.0, 7.0))

    def test_string_seq(self):
        a1 = ['a', 'b', 'c', 'd', 'e', 'f']
        a2 = [1, 2, 3, np.nan, np.nan, 6]
        result1, result2 = delete_masked_points(a1, a2)
        ind = [0, 1, 2, 5]
        assert_array_equal(result1, np.array(a1)[ind])
        assert_array_equal(result2, np.array(a2)[ind])

    def test_datetime(self):
        dates = [datetime(2008, 1, 1), datetime(2008, 1, 2),
                 datetime(2008, 1, 3), datetime(2008, 1, 4),
                 datetime(2008, 1, 5), datetime(2008, 1, 6)]
        a_masked = np.ma.array([1, 2, 3, np.nan, np.nan, 6],
                               mask=[False, False, True, True, False, False])
        actual = delete_masked_points(dates, a_masked)
        ind = [0, 1, 5]
        assert_array_equal(actual[0], np.array(dates)[ind])
        assert_array_equal(actual[1], a_masked[ind].compressed())

    def test_rgba(self):
        a_masked = np.ma.array([1, 2, 3, np.nan, np.nan, 6],
                               mask=[False, False, True, True, False, False])
        a_rgba = mcolors.to_rgba_array(['r', 'g', 'b', 'c', 'm', 'y'])
        actual = delete_masked_points(a_masked, a_rgba)
        ind = [0, 1, 5]
        assert_array_equal(actual[0], a_masked[ind].compressed())
        assert_array_equal(actual[1], a_rgba[ind])


class Test_boxplot_stats:
    def setup_method(self):
        np.random.seed(937)
        self.nrows = 37
        self.ncols = 4
        self.data = np.random.lognormal(size=(self.nrows, self.ncols),
                                        mean=1.5, sigma=1.75)
        self.known_keys = sorted([
            'mean', 'med', 'q1', 'q3', 'iqr',
            'cilo', 'cihi', 'whislo', 'whishi',
            'fliers', 'label'
        ])
        self.std_results = cbook.boxplot_stats(self.data)

        self.known_nonbootstrapped_res = {
            'cihi': 6.8161283264444847,
            'cilo': -0.1489815330368689,
            'iqr': 13.492709959447094,
            'mean': 13.00447442387868,
            'med': 3.3335733967038079,
            'fliers': np.array([
                92.55467075,  87.03819018,  42.23204914,  39.29390996
            ]),
            'q1': 1.3597529879465153,
            'q3': 14.85246294739361,
            'whishi': 27.899688243699629,
            'whislo': 0.042143774965502923
        }

        self.known_bootstrapped_ci = {
            'cihi': 8.939577523357828,
            'cilo': 1.8692703958676578,
        }

        self.known_whis3_res = {
            'whishi': 42.232049135969874,
            'whislo': 0.042143774965502923,
            'fliers': np.array([92.55467075, 87.03819018]),
        }

        self.known_res_percentiles = {
            'whislo':   0.1933685896907924,
            'whishi':  42.232049135969874
        }

        self.known_res_range = {
            'whislo': 0.042143774965502923,
            'whishi': 92.554670752188699

        }

    def test_form_main_list(self):
        assert isinstance(self.std_results, list)

    def test_form_each_dict(self):
        for res in self.std_results:
            assert isinstance(res, dict)

    def test_form_dict_keys(self):
        for res in self.std_results:
            assert set(res) <= set(self.known_keys)

    def test_results_baseline(self):
        res = self.std_results[0]
        for key, value in self.known_nonbootstrapped_res.items():
            assert_array_almost_equal(res[key], value)

    def test_results_bootstrapped(self):
        results = cbook.boxplot_stats(self.data, bootstrap=10000)
        res = results[0]
        for key, value in self.known_bootstrapped_ci.items():
            assert_approx_equal(res[key], value)

    def test_results_whiskers_float(self):
        results = cbook.boxplot_stats(self.data, whis=3)
        res = results[0]
        for key, value in self.known_whis3_res.items():
            assert_array_almost_equal(res[key], value)

    def test_results_whiskers_range(self):
        results = cbook.boxplot_stats(self.data, whis=[0, 100])
        res = results[0]
        for key, value in self.known_res_range.items():
            assert_array_almost_equal(res[key], value)

    def test_results_whiskers_percentiles(self):
        results = cbook.boxplot_stats(self.data, whis=[5, 95])
        res = results[0]
        for key, value in self.known_res_percentiles.items():
            assert_array_almost_equal(res[key], value)

    def test_results_withlabels(self):
        labels = ['Test1', 2, 'Aardvark', 4]
        results = cbook.boxplot_stats(self.data, labels=labels)
        for lab, res in zip(labels, results):
            assert res['label'] == lab

        results = cbook.boxplot_stats(self.data)
        for res in results:
            assert 'label' not in res

    def test_label_error(self):
        labels = [1, 2]
        with pytest.raises(ValueError):
            cbook.boxplot_stats(self.data, labels=labels)

    def test_bad_dims(self):
        data = np.random.normal(size=(34, 34, 34))
        with pytest.raises(ValueError):
            cbook.boxplot_stats(data)

    def test_boxplot_stats_autorange_false(self):
        x = np.zeros(shape=140)
        x = np.hstack([-25, x, 25])
        bstats_false = cbook.boxplot_stats(x, autorange=False)
        bstats_true = cbook.boxplot_stats(x, autorange=True)

        assert bstats_false[0]['whislo'] == 0
        assert bstats_false[0]['whishi'] == 0
        assert_array_almost_equal(bstats_false[0]['fliers'], [-25, 25])

        assert bstats_true[0]['whislo'] == -25
        assert bstats_true[0]['whishi'] == 25
        assert_array_almost_equal(bstats_true[0]['fliers'], [])


class Test_callback_registry:
    def setup_method(self):
        self.signal = 'test'
        self.callbacks = cbook.CallbackRegistry()

    def connect(self, s, func, pickle):
        if pickle:
            return self.callbacks.connect(s, func)
        else:
            return self.callbacks._connect_picklable(s, func)

    def disconnect(self, cid):
        return self.callbacks.disconnect(cid)

    def count(self):
        count1 = len(self.callbacks._func_cid_map.get(self.signal, []))
        count2 = len(self.callbacks.callbacks.get(self.signal))
        assert count1 == count2
        return count1

    def is_empty(self):
        np.testing.break_cycles()
        assert self.callbacks._func_cid_map == {}
        assert self.callbacks.callbacks == {}
        assert self.callbacks._pickled_cids == set()

    def is_not_empty(self):
        np.testing.break_cycles()
        assert self.callbacks._func_cid_map != {}
        assert self.callbacks.callbacks != {}

    def test_cid_restore(self):
        cb = cbook.CallbackRegistry()
        cb.connect('a', lambda: None)
        cb2 = pickle.loads(pickle.dumps(cb))
        cid = cb2.connect('c', lambda: None)
        assert cid == 1

    @pytest.mark.parametrize('pickle', [True, False])
    def test_callback_complete(self, pickle):
        # ensure we start with an empty registry
        self.is_empty()

        # create a class for testing
        mini_me = Test_callback_registry()

        # test that we can add a callback
        cid1 = self.connect(self.signal, mini_me.dummy, pickle)
        assert type(cid1) is int
        self.is_not_empty()

        # test that we don't add a second callback
        cid2 = self.connect(self.signal, mini_me.dummy, pickle)
        assert cid1 == cid2
        self.is_not_empty()
        assert len(self.callbacks._func_cid_map) == 1
        assert len(self.callbacks.callbacks) == 1

        del mini_me

        # check we now have no callbacks registered
        self.is_empty()

    @pytest.mark.parametrize('pickle', [True, False])
    def test_callback_disconnect(self, pickle):
        # ensure we start with an empty registry
        self.is_empty()

        # create a class for testing
        mini_me = Test_callback_registry()

        # test that we can add a callback
        cid1 = self.connect(self.signal, mini_me.dummy, pickle)
        assert type(cid1) is int
        self.is_not_empty()

        self.disconnect(cid1)

        # check we now have no callbacks registered
        self.is_empty()

    @pytest.mark.parametrize('pickle', [True, False])
    def test_callback_wrong_disconnect(self, pickle):
        # ensure we start with an empty registry
        self.is_empty()

        # create a class for testing
        mini_me = Test_callback_registry()

        # test that we can add a callback
        cid1 = self.connect(self.signal, mini_me.dummy, pickle)
        assert type(cid1) is int
        self.is_not_empty()

        self.disconnect("foo")

        # check we still have callbacks registered
        self.is_not_empty()

    @pytest.mark.parametrize('pickle', [True, False])
    def test_registration_on_non_empty_registry(self, pickle):
        # ensure we start with an empty registry
        self.is_empty()

        # setup the registry with a callback
        mini_me = Test_callback_registry()
        self.connect(self.signal, mini_me.dummy, pickle)

        # Add another callback
        mini_me2 = Test_callback_registry()
        self.connect(self.signal, mini_me2.dummy, pickle)

        # Remove and add the second callback
        mini_me2 = Test_callback_registry()
        self.connect(self.signal, mini_me2.dummy, pickle)

        # We still have 2 references
        self.is_not_empty()
        assert self.count() == 2

        # Removing the last 2 references
        mini_me = None
        mini_me2 = None
        self.is_empty()

    def dummy(self):
        pass

    def test_pickling(self):
        assert hasattr(pickle.loads(pickle.dumps(cbook.CallbackRegistry())),
                       "callbacks")


def test_callbackregistry_default_exception_handler(capsys, monkeypatch):
    cb = cbook.CallbackRegistry()
    cb.connect("foo", lambda: None)

    monkeypatch.setattr(
        cbook, "_get_running_interactive_framework", lambda: None)
    with pytest.raises(TypeError):
        cb.process("foo", "argument mismatch")
    outerr = capsys.readouterr()
    assert outerr.out == outerr.err == ""

    monkeypatch.setattr(
        cbook, "_get_running_interactive_framework", lambda: "not-none")
    cb.process("foo", "argument mismatch")  # No error in that case.
    outerr = capsys.readouterr()
    assert outerr.out == ""
    assert "takes 0 positional arguments but 1 was given" in outerr.err


def raising_cb_reg(func):
    class TestException(Exception):
        pass

    def raise_runtime_error():
        raise RuntimeError

    def raise_value_error():
        raise ValueError

    def transformer(excp):
        if isinstance(excp, RuntimeError):
            raise TestException
        raise excp

    # old default
    cb_old = cbook.CallbackRegistry(exception_handler=None)
    cb_old.connect('foo', raise_runtime_error)

    # filter
    cb_filt = cbook.CallbackRegistry(exception_handler=transformer)
    cb_filt.connect('foo', raise_runtime_error)

    # filter
    cb_filt_pass = cbook.CallbackRegistry(exception_handler=transformer)
    cb_filt_pass.connect('foo', raise_value_error)

    return pytest.mark.parametrize('cb, excp',
                                   [[cb_old, RuntimeError],
                                    [cb_filt, TestException],
                                    [cb_filt_pass, ValueError]])(func)


@raising_cb_reg
def test_callbackregistry_custom_exception_handler(monkeypatch, cb, excp):
    monkeypatch.setattr(
        cbook, "_get_running_interactive_framework", lambda: None)
    with pytest.raises(excp):
        cb.process('foo')


def test_callbackregistry_signals():
    cr = cbook.CallbackRegistry(signals=["foo"])
    results = []
    def cb(x): results.append(x)
    cr.connect("foo", cb)
    with pytest.raises(ValueError):
        cr.connect("bar", cb)
    cr.process("foo", 1)
    with pytest.raises(ValueError):
        cr.process("bar", 1)
    assert results == [1]


def test_callbackregistry_blocking():
    # Needs an exception handler for interactive testing environments
    # that would only print this out instead of raising the exception
    def raise_handler(excp):
        raise excp
    cb = cbook.CallbackRegistry(exception_handler=raise_handler)
    def test_func1():
        raise ValueError("1 should be blocked")
    def test_func2():
        raise ValueError("2 should be blocked")
    cb.connect("test1", test_func1)
    cb.connect("test2", test_func2)

    # block all of the callbacks to make sure they aren't processed
    with cb.blocked():
        cb.process("test1")
        cb.process("test2")

    # block individual callbacks to make sure the other is still processed
    with cb.blocked(signal="test1"):
        # Blocked
        cb.process("test1")
        # Should raise
        with pytest.raises(ValueError, match="2 should be blocked"):
            cb.process("test2")

    # Make sure the original callback functions are there after blocking
    with pytest.raises(ValueError, match="1 should be blocked"):
        cb.process("test1")
    with pytest.raises(ValueError, match="2 should be blocked"):
        cb.process("test2")


@pytest.mark.parametrize('line, result', [
    ('a : no_comment', 'a : no_comment'),
    ('a : "quoted str"', 'a : "quoted str"'),
    ('a : "quoted str" # comment', 'a : "quoted str"'),
    ('a : "#000000"', 'a : "#000000"'),
    ('a : "#000000" # comment', 'a : "#000000"'),
    ('a : ["#000000", "#FFFFFF"]', 'a : ["#000000", "#FFFFFF"]'),
    ('a : ["#000000", "#FFFFFF"] # comment', 'a : ["#000000", "#FFFFFF"]'),
    ('a : val  # a comment "with quotes"', 'a : val'),
    ('# only comment "with quotes" xx', ''),
])
def test_strip_comment(line, result):
    """Strip everything from the first unquoted #."""
    assert cbook._strip_comment(line) == result


def test_strip_comment_invalid():
    with pytest.raises(ValueError, match="Missing closing quote"):
        cbook._strip_comment('grid.color: "aa')


def test_sanitize_sequence():
    d = {'a': 1, 'b': 2, 'c': 3}
    k = ['a', 'b', 'c']
    v = [1, 2, 3]
    i = [('a', 1), ('b', 2), ('c', 3)]
    assert k == sorted(cbook.sanitize_sequence(d.keys()))
    assert v == sorted(cbook.sanitize_sequence(d.values()))
    assert i == sorted(cbook.sanitize_sequence(d.items()))
    assert i == cbook.sanitize_sequence(i)
    assert k == cbook.sanitize_sequence(k)


fail_mapping: tuple[tuple[dict, dict], ...] = (
    ({'a': 1, 'b': 2}, {'alias_mapping': {'a': ['b']}}),
    ({'a': 1, 'b': 2}, {'alias_mapping': {'a': ['a', 'b']}}),
)

pass_mapping: tuple[tuple[Any, dict, dict], ...] = (
    (None, {}, {}),
    ({'a': 1, 'b': 2}, {'a': 1, 'b': 2}, {}),
    ({'b': 2}, {'a': 2}, {'alias_mapping': {'a': ['a', 'b']}}),
)


@pytest.mark.parametrize('inp, kwargs_to_norm', fail_mapping)
def test_normalize_kwargs_fail(inp, kwargs_to_norm):
    with pytest.raises(TypeError), \
         _api.suppress_matplotlib_deprecation_warning():
        cbook.normalize_kwargs(inp, **kwargs_to_norm)


@pytest.mark.parametrize('inp, expected, kwargs_to_norm',
                         pass_mapping)
def test_normalize_kwargs_pass(inp, expected, kwargs_to_norm):
    with _api.suppress_matplotlib_deprecation_warning():
        # No other warning should be emitted.
        assert expected == cbook.normalize_kwargs(inp, **kwargs_to_norm)


def test_warn_external_frame_embedded_python():
    with patch.object(cbook, "sys") as mock_sys:
        mock_sys._getframe = Mock(return_value=None)
        with pytest.warns(UserWarning, match=r"\Adummy\Z"):
            _api.warn_external("dummy")


def test_to_prestep():
    x = np.arange(4)
    y1 = np.arange(4)
    y2 = np.arange(4)[::-1]

    xs, y1s, y2s = cbook.pts_to_prestep(x, y1, y2)

    x_target = np.asarray([0, 0, 1, 1, 2, 2, 3], dtype=float)
    y1_target = np.asarray([0, 1, 1, 2, 2, 3, 3], dtype=float)
    y2_target = np.asarray([3, 2, 2, 1, 1, 0, 0], dtype=float)

    assert_array_equal(x_target, xs)
    assert_array_equal(y1_target, y1s)
    assert_array_equal(y2_target, y2s)

    xs, y1s = cbook.pts_to_prestep(x, y1)
    assert_array_equal(x_target, xs)
    assert_array_equal(y1_target, y1s)


def test_to_prestep_empty():
    steps = cbook.pts_to_prestep([], [])
    assert steps.shape == (2, 0)


def test_to_poststep():
    x = np.arange(4)
    y1 = np.arange(4)
    y2 = np.arange(4)[::-1]

    xs, y1s, y2s = cbook.pts_to_poststep(x, y1, y2)

    x_target = np.asarray([0, 1, 1, 2, 2, 3, 3], dtype=float)
    y1_target = np.asarray([0, 0, 1, 1, 2, 2, 3], dtype=float)
    y2_target = np.asarray([3, 3, 2, 2, 1, 1, 0], dtype=float)

    assert_array_equal(x_target, xs)
    assert_array_equal(y1_target, y1s)
    assert_array_equal(y2_target, y2s)

    xs, y1s = cbook.pts_to_poststep(x, y1)
    assert_array_equal(x_target, xs)
    assert_array_equal(y1_target, y1s)


def test_to_poststep_empty():
    steps = cbook.pts_to_poststep([], [])
    assert steps.shape == (2, 0)


def test_to_midstep():
    x = np.arange(4)
    y1 = np.arange(4)
    y2 = np.arange(4)[::-1]

    xs, y1s, y2s = cbook.pts_to_midstep(x, y1, y2)

    x_target = np.asarray([0, .5, .5, 1.5, 1.5, 2.5, 2.5, 3], dtype=float)
    y1_target = np.asarray([0, 0, 1, 1, 2, 2, 3, 3], dtype=float)
    y2_target = np.asarray([3, 3, 2, 2, 1, 1, 0, 0], dtype=float)

    assert_array_equal(x_target, xs)
    assert_array_equal(y1_target, y1s)
    assert_array_equal(y2_target, y2s)

    xs, y1s = cbook.pts_to_midstep(x, y1)
    assert_array_equal(x_target, xs)
    assert_array_equal(y1_target, y1s)


def test_to_midstep_empty():
    steps = cbook.pts_to_midstep([], [])
    assert steps.shape == (2, 0)


@pytest.mark.parametrize(
    "args",
    [(np.arange(12).reshape(3, 4), 'a'),
     (np.arange(12), 'a'),
     (np.arange(12), np.arange(3))])
def test_step_fails(args):
    with pytest.raises(ValueError):
        cbook.pts_to_prestep(*args)


def test_grouper():
    class Dummy:
        pass
    a, b, c, d, e = objs = [Dummy() for _ in range(5)]
    g = cbook.Grouper()
    g.join(*objs)
    assert set(list(g)[0]) == set(objs)
    assert set(g.get_siblings(a)) == set(objs)

    for other in objs[1:]:
        assert g.joined(a, other)

    g.remove(a)
    for other in objs[1:]:
        assert not g.joined(a, other)

    for A, B in itertools.product(objs[1:], objs[1:]):
        assert g.joined(A, B)


def test_grouper_private():
    class Dummy:
        pass
    objs = [Dummy() for _ in range(5)]
    g = cbook.Grouper()
    g.join(*objs)
    # reach in and touch the internals !
    mapping = g._mapping

    for o in objs:
        assert o in mapping

    base_set = mapping[objs[0]]
    for o in objs[1:]:
        assert mapping[o] is base_set


def test_flatiter():
    x = np.arange(5)
    it = x.flat
    assert 0 == next(it)
    assert 1 == next(it)
    ret = cbook._safe_first_finite(it)
    assert ret == 0

    assert 0 == next(it)
    assert 1 == next(it)


def test__safe_first_finite_all_nan():
    arr = np.full(2, np.nan)
    ret = cbook._safe_first_finite(arr)
    assert np.isnan(ret)


def test__safe_first_finite_all_inf():
    arr = np.full(2, np.inf)
    ret = cbook._safe_first_finite(arr)
    assert np.isinf(ret)


def test_reshape2d():

    class Dummy:
        pass

    xnew = cbook._reshape_2D([], 'x')
    assert np.shape(xnew) == (1, 0)

    x = [Dummy() for _ in range(5)]

    xnew = cbook._reshape_2D(x, 'x')
    assert np.shape(xnew) == (1, 5)

    x = np.arange(5)
    xnew = cbook._reshape_2D(x, 'x')
    assert np.shape(xnew) == (1, 5)

    x = [[Dummy() for _ in range(5)] for _ in range(3)]
    xnew = cbook._reshape_2D(x, 'x')
    assert np.shape(xnew) == (3, 5)

    # this is strange behaviour, but...
    x = np.random.rand(3, 5)
    xnew = cbook._reshape_2D(x, 'x')
    assert np.shape(xnew) == (5, 3)

    # Test a list of lists which are all of length 1
    x = [[1], [2], [3]]
    xnew = cbook._reshape_2D(x, 'x')
    assert isinstance(xnew, list)
    assert isinstance(xnew[0], np.ndarray) and xnew[0].shape == (1,)
    assert isinstance(xnew[1], np.ndarray) and xnew[1].shape == (1,)
    assert isinstance(xnew[2], np.ndarray) and xnew[2].shape == (1,)

    # Test a list of zero-dimensional arrays
    x = [np.array(0), np.array(1), np.array(2)]
    xnew = cbook._reshape_2D(x, 'x')
    assert isinstance(xnew, list)
    assert len(xnew) == 1
    assert isinstance(xnew[0], np.ndarray) and xnew[0].shape == (3,)

    # Now test with a list of lists with different lengths, which means the
    # array will internally be converted to a 1D object array of lists
    x = [[1, 2, 3], [3, 4], [2]]
    xnew = cbook._reshape_2D(x, 'x')
    assert isinstance(xnew, list)
    assert isinstance(xnew[0], np.ndarray) and xnew[0].shape == (3,)
    assert isinstance(xnew[1], np.ndarray) and xnew[1].shape == (2,)
    assert isinstance(xnew[2], np.ndarray) and xnew[2].shape == (1,)

    # We now need to make sure that this works correctly for Numpy subclasses
    # where iterating over items can return subclasses too, which may be
    # iterable even if they are scalars. To emulate this, we make a Numpy
    # array subclass that returns Numpy 'scalars' when iterating or accessing
    # values, and these are technically iterable if checking for example
    # isinstance(x, collections.abc.Iterable).

    class ArraySubclass(np.ndarray):

        def __iter__(self):
            for value in super().__iter__():
                yield np.array(value)

        def __getitem__(self, item):
            return np.array(super().__getitem__(item))

    v = np.arange(10, dtype=float)
    x = ArraySubclass((10,), dtype=float, buffer=v.data)

    xnew = cbook._reshape_2D(x, 'x')

    # We check here that the array wasn't split up into many individual
    # ArraySubclass, which is what used to happen due to a bug in _reshape_2D
    assert len(xnew) == 1
    assert isinstance(xnew[0], ArraySubclass)

    # check list of strings:
    x = ['a', 'b', 'c', 'c', 'dd', 'e', 'f', 'ff', 'f']
    xnew = cbook._reshape_2D(x, 'x')
    assert len(xnew[0]) == len(x)
    assert isinstance(xnew[0], np.ndarray)


def test_reshape2d_pandas(pd):
    # separate to allow the rest of the tests to run if no pandas...
    X = np.arange(30).reshape(10, 3)
    x = pd.DataFrame(X, columns=["a", "b", "c"])
    Xnew = cbook._reshape_2D(x, 'x')
    # Need to check each row because _reshape_2D returns a list of arrays:
    for x, xnew in zip(X.T, Xnew):
        np.testing.assert_array_equal(x, xnew)


def test_reshape2d_xarray(xr):
    # separate to allow the rest of the tests to run if no xarray...
    X = np.arange(30).reshape(10, 3)
    x = xr.DataArray(X, dims=["x", "y"])
    Xnew = cbook._reshape_2D(x, 'x')
    # Need to check each row because _reshape_2D returns a list of arrays:
    for x, xnew in zip(X.T, Xnew):
        np.testing.assert_array_equal(x, xnew)


def test_index_of_pandas(pd):
    # separate to allow the rest of the tests to run if no pandas...
    X = np.arange(30).reshape(10, 3)
    x = pd.DataFrame(X, columns=["a", "b", "c"])
    Idx, Xnew = cbook.index_of(x)
    np.testing.assert_array_equal(X, Xnew)
    IdxRef = np.arange(10)
    np.testing.assert_array_equal(Idx, IdxRef)


def test_index_of_xarray(xr):
    # separate to allow the rest of the tests to run if no xarray...
    X = np.arange(30).reshape(10, 3)
    x = xr.DataArray(X, dims=["x", "y"])
    Idx, Xnew = cbook.index_of(x)
    np.testing.assert_array_equal(X, Xnew)
    IdxRef = np.arange(10)
    np.testing.assert_array_equal(Idx, IdxRef)


def test_contiguous_regions():
    a, b, c = 3, 4, 5
    # Starts and ends with True
    mask = [True]*a + [False]*b + [True]*c
    expected = [(0, a), (a+b, a+b+c)]
    assert cbook.contiguous_regions(mask) == expected
    d, e = 6, 7
    # Starts with True ends with False
    mask = mask + [False]*e
    assert cbook.contiguous_regions(mask) == expected
    # Starts with False ends with True
    mask = [False]*d + mask[:-e]
    expected = [(d, d+a), (d+a+b, d+a+b+c)]
    assert cbook.contiguous_regions(mask) == expected
    # Starts and ends with False
    mask = mask + [False]*e
    assert cbook.contiguous_regions(mask) == expected
    # No True in mask
    assert cbook.contiguous_regions([False]*5) == []
    # Empty mask
    assert cbook.contiguous_regions([]) == []


def test_safe_first_element_pandas_series(pd):
    # deliberately create a pandas series with index not starting from 0
    s = pd.Series(range(5), index=range(10, 15))
    actual = cbook._safe_first_finite(s)
    assert actual == 0


def test_warn_external(recwarn):
    _api.warn_external("oops")
    assert len(recwarn) == 1
    assert recwarn[0].filename == __file__


def test_array_patch_perimeters():
    # This compares the old implementation as a reference for the
    # vectorized one.
    def check(x, rstride, cstride):
        rows, cols = x.shape
        row_inds = [*range(0, rows-1, rstride), rows-1]
        col_inds = [*range(0, cols-1, cstride), cols-1]
        polys = []
        for rs, rs_next in zip(row_inds[:-1], row_inds[1:]):
            for cs, cs_next in zip(col_inds[:-1], col_inds[1:]):
                # +1 ensures we share edges between polygons
                ps = cbook._array_perimeter(x[rs:rs_next+1, cs:cs_next+1]).T
                polys.append(ps)
        polys = np.asarray(polys)
        assert np.array_equal(polys,
                              cbook._array_patch_perimeters(
                                  x, rstride=rstride, cstride=cstride))

    def divisors(n):
        return [i for i in range(1, n + 1) if n % i == 0]

    for rows, cols in [(5, 5), (7, 14), (13, 9)]:
        x = np.arange(rows * cols).reshape(rows, cols)
        for rstride, cstride in itertools.product(divisors(rows - 1),
                                                  divisors(cols - 1)):
            check(x, rstride=rstride, cstride=cstride)


def test_setattr_cm():
    class A:
        cls_level = object()
        override = object()

        def __init__(self):
            self.aardvark = 'aardvark'
            self.override = 'override'
            self._p = 'p'

        def meth(self):
            ...

        @classmethod
        def classy(cls):
            ...

        @staticmethod
        def static():
            ...

        @property
        def prop(self):
            return self._p

        @prop.setter
        def prop(self, val):
            self._p = val

    class B(A):
        ...

    other = A()

    def verify_pre_post_state(obj):
        # When you access a Python method the function is bound
        # to the object at access time so you get a new instance
        # of MethodType every time.
        #
        # https://docs.python.org/3/howto/descriptor.html#functions-and-methods
        assert obj.meth is not obj.meth
        # normal attribute should give you back the same instance every time
        assert obj.aardvark is obj.aardvark
        assert a.aardvark == 'aardvark'
        # and our property happens to give the same instance every time
        assert obj.prop is obj.prop
        assert obj.cls_level is A.cls_level
        assert obj.override == 'override'
        assert not hasattr(obj, 'extra')
        assert obj.prop == 'p'
        assert obj.monkey == other.meth
        assert obj.cls_level is A.cls_level
        assert 'cls_level' not in obj.__dict__
        assert 'classy' not in obj.__dict__
        assert 'static' not in obj.__dict__

    a = B()

    a.monkey = other.meth
    verify_pre_post_state(a)
    with cbook._setattr_cm(
            a, prop='squirrel',
            aardvark='moose', meth=lambda: None,
            override='boo', extra='extra',
            monkey=lambda: None, cls_level='bob',
            classy='classy', static='static'):
        # because we have set a lambda, it is normal attribute access
        # and the same every time
        assert a.meth is a.meth
        assert a.aardvark is a.aardvark
        assert a.aardvark == 'moose'
        assert a.override == 'boo'
        assert a.extra == 'extra'
        assert a.prop == 'squirrel'
        assert a.monkey != other.meth
        assert a.cls_level == 'bob'
        assert a.classy == 'classy'
        assert a.static == 'static'

    verify_pre_post_state(a)


def test_format_approx():
    f = cbook._format_approx
    assert f(0, 1) == '0'
    assert f(0, 2) == '0'
    assert f(0, 3) == '0'
    assert f(-0.0123, 1) == '-0'
    assert f(1e-7, 5) == '0'
    assert f(0.0012345600001, 5) == '0.00123'
    assert f(-0.0012345600001, 5) == '-0.00123'
    assert f(0.0012345600001, 8) == f(0.0012345600001, 10) == '0.00123456'


def test_safe_first_element_with_none():
    datetime_lst = [date.today() + timedelta(days=i) for i in range(10)]
    datetime_lst[0] = None
    actual = cbook._safe_first_finite(datetime_lst)
    assert actual is not None and actual == datetime_lst[1]


def test_strip_math():
    assert strip_math(r'1 \times 2') == r'1 \times 2'
    assert strip_math(r'$1 \times 2$') == '1 x 2'
    assert strip_math(r'$\rm{hi}$') == 'hi'


@pytest.mark.parametrize('fmt, value, result', [
    ('%.2f m', 0.2, '0.20 m'),
    ('{:.2f} m', 0.2, '0.20 m'),
    ('{} m', 0.2, '0.2 m'),
    ('const', 0.2, 'const'),
    ('%d or {}', 0.2, '0 or {}'),
    ('{{{:,.0f}}}', 2e5, '{200,000}'),
    ('{:.2%}', 2/3, '66.67%'),
    ('$%g', 2.54, '$2.54'),
])
def test_auto_format_str(fmt, value, result):
    """Apply *value* to the format string *fmt*."""
    assert cbook._auto_format_str(fmt, value) == result
    assert cbook._auto_format_str(fmt, np.float64(value)) == result