# ----------------------------------------------------------------------------
# Copyright (c) 2013--, scikit-bio development team.
#
# Distributed under the terms of the Modified BSD License.
#
# The full license is in the file LICENSE.txt, distributed with this software.
# ----------------------------------------------------------------------------

import io
import unittest
from unittest import TestCase, main

import numpy as np
import numpy.testing as npt
import pandas as pd
import pandas.testing as pdt
import scipy.spatial.distance

try:
    import matplotlib as mpl
except ImportError:
    has_matplotlib = False
else:
    has_matplotlib = True

import skbio.sequence.distance
from skbio import DistanceMatrix, Sequence
from skbio.stats.distance import (
    DissimilarityMatrixError, DistanceMatrixError, MissingIDError,
    DissimilarityMatrix, randdm)
from skbio.stats.distance._base import (_preprocess_input,
                                        _run_monte_carlo_stats)
from skbio.stats.distance._utils import is_symmetric_and_hollow
from skbio.util import assert_data_frame_almost_equal
from skbio.util._testing import assert_series_almost_equal


class DissimilarityMatrixTestData:
    def setUp(self):
        self.dm_1x1_data = [[0.0]]
        self.dm_2x2_data = [[0.0, 0.123], [0.123, 0.0]]
        self.dm_2x2_asym_data = [[0.0, 1.0], [-2.0, 0.0]]
        self.dm_3x3_data = [[0.0, 0.01, 4.2], [0.01, 0.0, 12.0],
                            [4.2, 12.0, 0.0]]
        self.dm_5x5_data = [[0, 1, 2, 3, 4],
                            [5, 0, 6, 7, 8],
                            [9, 1, 0, 2, 3],
                            [4, 5, 6, 0, 7],
                            [8, 9, 1, 2, 0]]


class DissimilarityMatrixTestBase(DissimilarityMatrixTestData):
    matobj = None

    def setUp(self):
        super(DissimilarityMatrixTestBase, self).setUp()
        self.dm_1x1 = self.matobj(self.dm_1x1_data, ['a'])
        self.dm_2x2 = self.matobj(self.dm_2x2_data, ['a', 'b'])
        self.dm_2x2_asym = self.matobj(self.dm_2x2_asym_data,
                                       ['a', 'b'])
        self.dm_3x3 = self.matobj(self.dm_3x3_data, ['a', 'b', 'c'])
        self.dm_5x5 = self.matobj(self.dm_5x5_data, list('abcde'))

        self.dms = [self.dm_1x1, self.dm_2x2, self.dm_2x2_asym, self.dm_3x3]
        self.dm_shapes = [(1, 1), (2, 2), (2, 2), (3, 3)]
        self.dm_sizes = [1, 4, 4, 9]
        self.dm_transposes = [
            self.dm_1x1, self.dm_2x2,
            self.matobj([[0, -2], [1, 0]], ['a', 'b']), self.dm_3x3]
        self.dm_redundant_forms = [np.array(self.dm_1x1_data),
                                   np.array(self.dm_2x2_data),
                                   np.array(self.dm_2x2_asym_data),
                                   np.array(self.dm_3x3_data)]

    def test_avoid_copy_on_construction(self):
        # ((data, expect_copy))
        tests = (([[0, 1], [1, 0]], True),
                 ([(0, 1), (1, 0)], True),
                 (((0, 1), (1, 0)), True),
                 (np.array([[0, 1], [1, 0]], dtype='int'), True),
                 (np.array([[0, 1], [1, 0]], dtype='float'), False),
                 (np.array([[0, 1], [1, 0]], dtype=np.float32), False),
                 (np.array([[0, 1], [1, 0]], dtype=np.float64), False),
                 (np.array([[0, 1], [1, 0]], dtype='double'), False))

        for data, expect in tests:
            obj = DissimilarityMatrix(data)
            self.assertEqual(id(obj.data) != id(data), expect)

    def test_within(self):
        exp = pd.DataFrame([['a', 'a', 0.0],
                            ['a', 'c', 4.2],
                            ['c', 'a', 4.2],
                            ['c', 'c', 0.0]],
                           columns=['i', 'j', 'value'])
        obs = self.dm_3x3.within(['a', 'c'])
        pdt.assert_frame_equal(obs, exp)

    def test_within_order_stability(self):
        exp = pd.DataFrame([['a', 'a', 0.0],
                            ['a', 'c', 4.2],
                            ['c', 'a', 4.2],
                            ['c', 'c', 0.0]],
                           columns=['i', 'j', 'value'])

        # NOTE: order was changed from ['a', 'c'] to ['c', 'a']
        # but the output order in exp is consistent with
        # test_within
        obs = self.dm_3x3.within(['c', 'a'])
        pdt.assert_frame_equal(obs, exp)
        obs = self.dm_3x3.within(['a', 'c'])
        pdt.assert_frame_equal(obs, exp)

    def test_within_missing_id(self):
        with self.assertRaisesRegex(MissingIDError, "not found."):
            self.dm_3x3.within(['x', 'a'])

    def test_between(self):
        exp = pd.DataFrame([['b', 'a', 5.],
                            ['b', 'c', 6.],
                            ['b', 'e', 8.],
                            ['d', 'a', 4.],
                            ['d', 'c', 6.],
                            ['d', 'e', 7.]],
                           columns=['i', 'j', 'value'])

        obs = self.dm_5x5.between(['b', 'd'], ['a', 'c', 'e'])
        pdt.assert_frame_equal(obs, exp)

    def test_between_order_stability(self):
        exp = pd.DataFrame([['b', 'a', 5.],
                            ['b', 'c', 6.],
                            ['b', 'e', 8.],
                            ['d', 'a', 4.],
                            ['d', 'c', 6.],
                            ['d', 'e', 7.]],
                           columns=['i', 'j', 'value'])

        # varying the order of the "i" values, result remains consistent
        # with the test_between result
        obs = self.dm_5x5.between(['d', 'b'], ['a', 'c', 'e'])
        pdt.assert_frame_equal(obs, exp)

        # varying the order of the "j" values, result remains consistent
        # with the test_between result
        obs = self.dm_5x5.between(['b', 'd'], ['a', 'e', 'c'])
        pdt.assert_frame_equal(obs, exp)

        # varying the order of the "i" and "j" values, result remains
        # consistent with the test_between result
        obs = self.dm_5x5.between(['d', 'b'], ['a', 'e', 'c'])
        pdt.assert_frame_equal(obs, exp)

    def test_between_overlap(self):
        exp = pd.DataFrame([['b', 'a', 5.],
                            ['b', 'd', 7.],
                            ['b', 'e', 8.],
                            ['d', 'a', 4.],
                            ['d', 'd', 0.],
                            ['d', 'e', 7.]],
                           columns=['i', 'j', 'value'])

        # 'd' in i and j overlap
        with self.assertRaisesRegex(KeyError, ("This constraint can "
                                               "removed with "
                                               "allow_overlap=True.")):
            self.dm_5x5.between(['b', 'd'], ['a', 'd', 'e'])

        obs = self.dm_5x5.between(['b', 'd'], ['a', 'd', 'e'],
                                  allow_overlap=True)
        pdt.assert_frame_equal(obs, exp)

    def test_between_missing_id(self):
        with self.assertRaisesRegex(MissingIDError, "not found."):
            self.dm_3x3.between(['x', 'a'], ['a', 'b', 'c'])

        with self.assertRaisesRegex(MissingIDError, "not found."):
            self.dm_3x3.between(['a', 'b'], ['a', 'x', 'c'])

        with self.assertRaisesRegex(MissingIDError, "not found."):
            self.dm_3x3.between(['a', 'y'], ['a', 'x', 'c'])

    def test_stable_order(self):
        exp = np.array([1, 3, 4], dtype=int)
        obs = self.dm_5x5._stable_order(['d', 'e', 'b'])

        npt.assert_equal(obs, exp)

    def test_subset_to_dataframe(self):
        exp = pd.DataFrame([['b', 'a', 5.],
                            ['b', 'd', 7.],
                            ['b', 'e', 8.],
                            ['d', 'a', 4.],
                            ['d', 'd', 0.],
                            ['d', 'e', 7.]],
                           columns=['i', 'j', 'value'])

        obs = self.dm_5x5._subset_to_dataframe(['b', 'd'], ['a', 'd', 'e'])
        pdt.assert_frame_equal(obs, exp)

        # and the empty edge cases
        exp = pd.DataFrame([],
                           columns=['i', 'j', 'value'],
                           index=pd.RangeIndex(start=0, stop=0))

        obs = self.dm_5x5._subset_to_dataframe([], ['a', 'd', 'e'])
        pdt.assert_frame_equal(obs, exp, check_dtype=False)
        obs = self.dm_5x5._subset_to_dataframe(['b', 'd'], [])
        pdt.assert_frame_equal(obs, exp, check_dtype=False)
        obs = self.dm_5x5._subset_to_dataframe([], [])
        pdt.assert_frame_equal(obs, exp, check_dtype=False)

    def test_init_from_dm(self):
        ids = ['foo', 'bar', 'baz']

        # DissimilarityMatrix -> DissimilarityMatrix
        exp = self.matobj(self.dm_3x3_data, ids)
        obs = self.matobj(self.dm_3x3, ids)
        self.assertEqual(obs, exp)
        # Test that copy of data is not made.
        self.assertTrue(obs.data is self.dm_3x3.data)
        obs.data[0, 1] = 424242
        self.assertTrue(np.array_equal(obs.data, self.dm_3x3.data))

        # DistanceMatrix -> DissimilarityMatrix
        exp = self.matobj(self.dm_3x3_data, ids)
        obs = self.matobj(
            self.matobj(self.dm_3x3_data, ('a', 'b', 'c')), ids)
        self.assertEqual(obs, exp)

        # DissimilarityMatrix -> DistanceMatrix
        with self.assertRaises(DistanceMatrixError):
            DistanceMatrix(self.dm_2x2_asym, ['foo', 'bar'])

    def test_init_non_hollow_dm(self):
        data = [[1, 1], [1, 1]]
        obs = self.matobj(data, ['a', 'b'])
        self.assertTrue(np.array_equal(obs.data, data))
        data_hollow = skbio.stats.distance._utils.is_hollow(obs.data)
        self.assertEqual(data_hollow, False)

    def test_init_no_ids(self):
        exp = self.matobj(self.dm_3x3_data, ('0', '1', '2'))
        obs = self.matobj(self.dm_3x3_data)
        self.assertEqual(obs, exp)
        self.assertEqual(obs['1', '2'], 12.0)

    def test_init_invalid_input(self):
        # Empty data.
        with self.assertRaises(DissimilarityMatrixError):
            self.matobj([], [])

        # Another type of empty data.
        with self.assertRaises(DissimilarityMatrixError):
            self.matobj(np.empty((0, 0)), [])

        # Invalid number of dimensions.
        with self.assertRaises(DissimilarityMatrixError):
            self.matobj([1, 2, 3], ['a'])

        # Dimensions don't match.
        with self.assertRaises(DissimilarityMatrixError):
            self.matobj([[1, 2, 3]], ['a'])

        data = [[0, 1], [1, 0]]

        # Duplicate IDs.
        with self.assertRaises(DissimilarityMatrixError):
            self.matobj(data, ['a', 'a'])

        # Number of IDs don't match dimensions.
        with self.assertRaises(DissimilarityMatrixError):
            self.matobj(data, ['a', 'b', 'c'])
        with self.assertRaises(DissimilarityMatrixError):
            self.matobj(data, [])

    def test_from_iterable_non_hollow_data(self):
        iterable = (x for x in range(4))

        exp = self.matobj([[1, 1, 1, 1],
                           [1, 1, 1, 1],
                           [1, 1, 1, 1],
                           [1, 1, 1, 1]])
        res = self.matobj.from_iterable(iterable, lambda a, b: 1)
        self.assertEqual(res, exp)

    def test_from_iterable_asymmetric_data(self):
        iterable = (x for x in range(4))

        exp = self.matobj([[0, 1, 2, 3],
                           [-1, 0, 1, 2],
                           [-2, -1, 0, 1],
                           [-3, -2, -1, 0]])
        res = self.matobj.from_iterable(iterable, lambda a, b: b - a)
        self.assertEqual(res, exp)

    def test_from_iterable_no_key(self):
        iterable = (x for x in range(4))

        exp = self.matobj([[0, 1, 2, 3],
                           [1, 0, 1, 2],
                           [2, 1, 0, 1],
                           [3, 2, 1, 0]])
        res = self.matobj.from_iterable(iterable,
                                        lambda a, b: abs(b - a))
        self.assertEqual(res, exp)

    def test_from_iterable_with_key(self):
        iterable = (x for x in range(4))

        exp = self.matobj([[0, 1, 2, 3],
                           [1, 0, 1, 2],
                           [2, 1, 0, 1],
                           [3, 2, 1, 0]], ['0', '1', '4', '9'])
        res = self.matobj.from_iterable(iterable,
                                        lambda a, b: abs(b - a),
                                        key=lambda x: str(x ** 2))
        self.assertEqual(res, exp)

    def test_from_iterable_empty(self):
        with self.assertRaises(DissimilarityMatrixError):
            self.matobj.from_iterable([], lambda x: x)

    def test_from_iterable_single(self):
        exp = self.matobj([[100]])
        res = self.matobj.from_iterable(["boo"], lambda a, b: 100)
        self.assertEqual(res, exp)

    def test_from_iterable_with_keys(self):
        iterable = (x for x in range(4))

        exp = self.matobj([[0, 1, 2, 3],
                           [1, 0, 1, 2],
                           [2, 1, 0, 1],
                           [3, 2, 1, 0]], ['0', '1', '4', '9'])
        res = self.matobj.from_iterable(iterable,
                                        lambda a, b: abs(b - a),
                                        keys=iter(['0', '1', '4', '9'])
                                        )
        self.assertEqual(res, exp)

    def test_from_iterable_with_key_and_keys(self):
        iterable = (x for x in range(4))
        with self.assertRaises(ValueError):
            self.matobj.from_iterable(iterable,
                                      lambda a, b: abs(b - a),
                                      key=str,
                                      keys=['1', '2', '3', '4'])

    def test_from_iterable_scipy_hamming_metric_with_metadata(self):
        # test for #1254
        seqs = [
            Sequence('ACGT'),
            Sequence('ACGA', metadata={'id': 'seq1'}),
            Sequence('AAAA', metadata={'id': 'seq2'}),
            Sequence('AAAA', positional_metadata={'qual': range(4)})
        ]

        exp = self.matobj([
            [0, 0.25, 0.75, 0.75],
            [0.25, 0.0, 0.5, 0.5],
            [0.75, 0.5, 0.0, 0.0],
            [0.75, 0.5, 0.0, 0.0]], ['a', 'b', 'c', 'd'])

        dm = self.matobj.from_iterable(
            seqs,
            metric=scipy.spatial.distance.hamming,
            keys=['a', 'b', 'c', 'd'])

        self.assertEqual(dm, exp)

    def test_from_iterable_skbio_hamming_metric_with_metadata(self):
        # test for #1254
        seqs = [
            Sequence('ACGT'),
            Sequence('ACGA', metadata={'id': 'seq1'}),
            Sequence('AAAA', metadata={'id': 'seq2'}),
            Sequence('AAAA', positional_metadata={'qual': range(4)})
        ]

        exp = self.matobj([
            [0, 0.25, 0.75, 0.75],
            [0.25, 0.0, 0.5, 0.5],
            [0.75, 0.5, 0.0, 0.0],
            [0.75, 0.5, 0.0, 0.0]], ['a', 'b', 'c', 'd'])

        dm = self.matobj.from_iterable(
            seqs,
            metric=skbio.sequence.distance.hamming,
            keys=['a', 'b', 'c', 'd'])

        self.assertEqual(dm, exp)

    def test_data(self):
        for dm, exp in zip(self.dms, self.dm_redundant_forms):
            obs = dm.data
            self.assertTrue(np.array_equal(obs, exp))

        with self.assertRaises(AttributeError):
            self.dm_3x3.data = 'foo'

    def test_ids(self):
        obs = self.dm_3x3.ids
        self.assertEqual(obs, ('a', 'b', 'c'))

        # Test that we overwrite the existing IDs and that the ID index is
        # correctly rebuilt.
        new_ids = ['foo', 'bar', 'baz']
        self.dm_3x3.ids = new_ids
        obs = self.dm_3x3.ids
        self.assertEqual(obs, tuple(new_ids))
        self.assertTrue(np.array_equal(self.dm_3x3['bar'],
                                       np.array([0.01, 0.0, 12.0])))
        with self.assertRaises(MissingIDError):
            self.dm_3x3['b']

    def test_ids_invalid_input(self):
        with self.assertRaises(DissimilarityMatrixError):
            self.dm_3x3.ids = ['foo', 'bar']
        # Make sure that we can still use the dissimilarity matrix after trying
        # to be evil.
        obs = self.dm_3x3.ids
        self.assertEqual(obs, ('a', 'b', 'c'))

    def test_dtype(self):
        for dm in self.dms:
            self.assertEqual(dm.dtype, np.float64)

    def test_shape(self):
        for dm, shape in zip(self.dms, self.dm_shapes):
            self.assertEqual(dm.shape, shape)

    def test_size(self):
        for dm, size in zip(self.dms, self.dm_sizes):
            self.assertEqual(dm.size, size)

    def test_transpose(self):
        for dm, transpose in zip(self.dms, self.dm_transposes):
            self.assertEqual(dm.T, transpose)
            self.assertEqual(dm.transpose(), transpose)
            # We should get a reference to a different object back, even if the
            # transpose is the same as the original.
            self.assertTrue(dm.transpose() is not dm)

    def test_index(self):
        self.assertEqual(self.dm_3x3.index('a'), 0)
        self.assertEqual(self.dm_3x3.index('b'), 1)
        self.assertEqual(self.dm_3x3.index('c'), 2)

        with self.assertRaises(MissingIDError):
            self.dm_3x3.index('d')

        with self.assertRaises(MissingIDError):
            self.dm_3x3.index(1)

    def test_redundant_form(self):
        for dm, redundant in zip(self.dms, self.dm_redundant_forms):
            obs = dm.redundant_form()
            self.assertTrue(np.array_equal(obs, redundant))

    def test_copy(self):
        copy = self.dm_2x2.copy()
        self.assertEqual(copy, self.dm_2x2)
        self.assertFalse(copy.data is self.dm_2x2.data)
        # deepcopy doesn't actually create a copy of the IDs because it is a
        # tuple of strings, which is fully immutable.
        self.assertTrue(copy.ids is self.dm_2x2.ids)

        new_ids = ['hello', 'world']
        copy.ids = new_ids
        self.assertNotEqual(copy.ids, self.dm_2x2.ids)

        copy = self.dm_2x2.copy()
        copy.data[0, 1] = 0.0001
        self.assertFalse(np.array_equal(copy.data, self.dm_2x2.data))

    def test_filter_no_filtering(self):
        # Don't actually filter anything -- ensure we get back a different
        # object.
        obs = self.dm_3x3.filter(['a', 'b', 'c'])
        self.assertEqual(obs, self.dm_3x3)
        self.assertFalse(obs is self.dm_3x3)

    def test_filter_reorder(self):
        # Don't filter anything, but reorder the distance matrix.
        order = ['c', 'a', 'b']
        exp = self.matobj(
            [[0, 4.2, 12], [4.2, 0, 0.01], [12, 0.01, 0]], order)
        obs = self.dm_3x3.filter(order)
        self.assertEqual(obs, exp)

    def test_filter_single_id(self):
        ids = ['b']
        exp = self.matobj([[0]], ids)
        obs = self.dm_2x2_asym.filter(ids)
        self.assertEqual(obs, exp)

    def test_filter_asymmetric(self):
        # 2x2
        ids = ['b', 'a']
        exp = self.matobj([[0, -2], [1, 0]], ids)
        obs = self.dm_2x2_asym.filter(ids)
        self.assertEqual(obs, exp)

        # 3x3
        dm = self.matobj([[0, 10, 53], [42, 0, 22.5], [53, 1, 0]],
                         ('bro', 'brah', 'breh'))
        ids = ['breh', 'brah']
        exp = self.matobj([[0, 1], [22.5, 0]], ids)
        obs = dm.filter(ids)
        self.assertEqual(obs, exp)

    def test_filter_subset(self):
        ids = ('c', 'a')
        exp = self.matobj([[0, 4.2], [4.2, 0]], ids)
        obs = self.dm_3x3.filter(ids)
        self.assertEqual(obs, exp)

        ids = ('b', 'a')
        exp = self.matobj([[0, 0.01], [0.01, 0]], ids)
        obs = self.dm_3x3.filter(ids)
        self.assertEqual(obs, exp)

        # 4x4
        dm = self.matobj([[0, 1, 55, 7], [1, 0, 16, 1],
                         [55, 16, 0, 23], [7, 1, 23, 0]])
        ids = np.asarray(['3', '0', '1'])
        exp = self.matobj([[0, 7, 1], [7, 0, 1], [1, 1, 0]], ids)
        obs = dm.filter(ids)
        self.assertEqual(obs, exp)

    def test_filter_duplicate_ids(self):
        with self.assertRaises(DissimilarityMatrixError):
            self.dm_3x3.filter(['c', 'a', 'c'])

    def test_filter_missing_ids(self):
        with self.assertRaises(MissingIDError):
            self.dm_3x3.filter(['c', 'bro'])

    def test_filter_missing_ids_strict_false(self):
        # no exception should be raised
        ids = ('c', 'a')
        exp = self.matobj([[0, 4.2], [4.2, 0]], ids)
        obs = self.dm_3x3.filter(['c', 'a', 'not found'], strict=False)
        self.assertEqual(obs, exp)

    def test_filter_empty_ids(self):
        with self.assertRaises(DissimilarityMatrixError):
            self.dm_3x3.filter([])

    @unittest.skipUnless(has_matplotlib, "Matplotlib not available.")
    def test_plot_default(self):
        fig = self.dm_1x1.plot()
        self.assertIsInstance(fig, mpl.figure.Figure)
        axes = fig.get_axes()
        self.assertEqual(len(axes), 2)
        ax = axes[0]
        self.assertEqual(ax.get_title(), '')
        xticks = []
        for tick in ax.get_xticklabels():
            xticks.append(tick.get_text())
        self.assertEqual(xticks, ['a'])
        yticks = []
        for tick in ax.get_yticklabels():
            yticks.append(tick.get_text())
        self.assertEqual(yticks, ['a'])

    @unittest.skipUnless(has_matplotlib, "Matplotlib not available.")
    def test_plot_no_default(self):
        ids = ['0', 'one', '2', 'three', '4.000']
        data = ([0, 1, 2, 3, 4], [1, 0, 1, 2, 3], [2, 1, 0, 1, 2],
                [3, 2, 1, 0, 1], [4, 3, 2, 1, 0])
        dm = self.matobj(data, ids)
        fig = dm.plot(cmap='Reds', title='Testplot')
        self.assertIsInstance(fig, mpl.figure.Figure)
        axes = fig.get_axes()
        self.assertEqual(len(axes), 2)
        ax = axes[0]
        self.assertEqual(ax.get_title(), 'Testplot')
        xticks = []
        for tick in ax.get_xticklabels():
            xticks.append(tick.get_text())
        self.assertEqual(xticks, ['0', 'one', '2', 'three', '4.000'])
        yticks = []
        for tick in ax.get_yticklabels():
            yticks.append(tick.get_text())
        self.assertEqual(yticks, ['0', 'one', '2', 'three', '4.000'])

    def test_to_data_frame_1x1(self):
        df = self.dm_1x1.to_data_frame()
        exp = pd.DataFrame([[0.0]], index=['a'], columns=['a'])
        assert_data_frame_almost_equal(df, exp)

    def test_to_data_frame_3x3(self):
        df = self.dm_3x3.to_data_frame()
        exp = pd.DataFrame([[0.0, 0.01, 4.2],
                            [0.01, 0.0, 12.0],
                            [4.2, 12.0, 0.0]],
                           index=['a', 'b', 'c'], columns=['a', 'b', 'c'])
        assert_data_frame_almost_equal(df, exp)

    def test_to_data_frame_default_ids(self):
        df = self.matobj(self.dm_2x2_data).to_data_frame()
        exp = pd.DataFrame([[0.0, 0.123],
                            [0.123, 0.0]],
                           index=['0', '1'], columns=['0', '1'])
        assert_data_frame_almost_equal(df, exp)

    def test_str(self):
        for dm in self.dms:
            obs = str(dm)
            # Do some very light testing here to make sure we're getting a
            # non-empty string back. We don't want to test the exact
            # formatting.
            self.assertTrue(obs)

    def test_eq(self):
        for dm in self.dms:
            copy = dm.copy()
            self.assertTrue(dm == dm)
            self.assertTrue(copy == copy)
            self.assertTrue(dm == copy)
            self.assertTrue(copy == dm)

        self.assertFalse(self.dm_1x1 == self.dm_3x3)

    def test_ne(self):
        # Wrong class.
        self.assertTrue(self.dm_3x3 != 'foo')

        # Wrong shape.
        self.assertTrue(self.dm_3x3 != self.dm_1x1)

        # Wrong IDs.
        other = self.dm_3x3.copy()
        other.ids = ['foo', 'bar', 'baz']
        self.assertTrue(self.dm_3x3 != other)

        # Wrong data.
        other = self.dm_3x3.copy()
        other.data[1, 0] = 42.42
        self.assertTrue(self.dm_3x3 != other)

        self.assertFalse(self.dm_2x2 != self.dm_2x2)

    def test_contains(self):
        self.assertTrue('a' in self.dm_3x3)
        self.assertTrue('b' in self.dm_3x3)
        self.assertTrue('c' in self.dm_3x3)
        self.assertFalse('d' in self.dm_3x3)

    def test_getslice(self):
        # Slice of first dimension only. Test that __getslice__ defers to
        # __getitem__.
        obs = self.dm_2x2[1:]
        self.assertTrue(np.array_equal(obs, np.array([[0.123, 0.0]])))
        self.assertEqual(type(obs), np.ndarray)

    def test_getitem_by_id(self):
        obs = self.dm_1x1['a']
        self.assertTrue(np.array_equal(obs, np.array([0.0])))

        obs = self.dm_2x2_asym['b']
        self.assertTrue(np.array_equal(obs, np.array([-2.0, 0.0])))

        obs = self.dm_3x3['c']
        self.assertTrue(np.array_equal(obs, np.array([4.2, 12.0, 0.0])))

        with self.assertRaises(MissingIDError):
            self.dm_2x2['c']

    def test_getitem_by_id_pair(self):
        # Same object.
        self.assertEqual(self.dm_1x1['a', 'a'], 0.0)

        # Different objects (symmetric).
        self.assertEqual(self.dm_3x3['b', 'c'], 12.0)
        self.assertEqual(self.dm_3x3['c', 'b'], 12.0)

        # Different objects (asymmetric).
        self.assertEqual(self.dm_2x2_asym['a', 'b'], 1.0)
        self.assertEqual(self.dm_2x2_asym['b', 'a'], -2.0)

        with self.assertRaises(MissingIDError):
            self.dm_2x2['a', 'c']

    def test_getitem_ndarray_indexing(self):
        # Single element access.
        obs = self.dm_3x3[0, 1]
        self.assertEqual(obs, 0.01)

        # Single element access (via two __getitem__ calls).
        obs = self.dm_3x3[0][1]
        self.assertEqual(obs, 0.01)

        # Row access.
        obs = self.dm_3x3[1]
        self.assertTrue(np.array_equal(obs, np.array([0.01, 0.0, 12.0])))
        self.assertEqual(type(obs), np.ndarray)

        # Grab all data.
        obs = self.dm_3x3[:, :]
        self.assertTrue(np.array_equal(obs, self.dm_3x3.data))
        self.assertEqual(type(obs), np.ndarray)

        with self.assertRaises(IndexError):
            self.dm_3x3[:, 3]

    def test_validate_invalid_dtype(self):
        with self.assertRaises(DissimilarityMatrixError):
            self.dm_3x3._validate(np.array([[0, 42], [42, 0]]), ['a', 'b'])

    def test_validate_invalid_shape(self):
        # first check it actually likes good matrices
        self.dm_3x3._validate_shape(np.array([[0., 42.], [42., 0.]]))
        # it checks just the shape, not the content
        self.dm_3x3._validate_shape(np.array([[1., 2.], [3., 4.]]))
        # empty array
        with self.assertRaises(DissimilarityMatrixError):
            self.dm_3x3._validate_shape(np.array([]))
        # invalid shape
        with self.assertRaises(DissimilarityMatrixError):
            self.dm_3x3._validate_shape(np.array([[0., 42.],
                                                  [42., 0.],
                                                  [22., 22.]]))
        with self.assertRaises(DissimilarityMatrixError):
            self.dm_3x3._validate_shape(np.array([[[0., 42.], [42., 0.]],
                                                  [[0., 24.], [24., 0.]]]))

    def test_validate_invalid_ids(self):
        # repeated ids
        with self.assertRaises(DissimilarityMatrixError):
            self.dm_3x3._validate_ids(self.dm_3x3.data, ['a', 'a'])
        # empty ids
        with self.assertRaises(DissimilarityMatrixError):
            self.dm_3x3._validate_ids(self.dm_3x3.data, [])
        # invalid shape
        with self.assertRaises(DissimilarityMatrixError):
            self.dm_3x3._validate_ids(self.dm_3x3.data, ['a', 'b', 'c', 'd'])


class DistanceMatrixTestBase(DissimilarityMatrixTestData):
    matobj = None

    def setUp(self):
        super(DistanceMatrixTestBase, self).setUp()

        self.dm_1x1 = self.matobj(self.dm_1x1_data, ['a'])
        self.dm_2x2 = self.matobj(self.dm_2x2_data, ['a', 'b'])
        self.dm_3x3 = self.matobj(self.dm_3x3_data, ['a', 'b', 'c'])

        self.dms = [self.dm_1x1, self.dm_2x2, self.dm_3x3]
        self.dm_condensed_forms = [np.array([]), np.array([0.123]),
                                   np.array([0.01, 4.2, 12.0])]

    def test_init_from_condensed_form(self):
        data = [1, 2, 3]
        exp = self.matobj([[0, 1, 2],
                           [1, 0, 3],
                           [2, 3, 0]], ['0', '1', '2'])
        res = self.matobj(data)
        self.assertEqual(exp, res)

    def test_init_invalid_input(self):
        # Asymmetric.
        data = [[0.0, 2.0], [1.0, 0.0]]
        with self.assertRaises(DistanceMatrixError):
            self.matobj(data, ['a', 'b'])

        # Non-hollow
        data = [[1.0, 2.0], [2.0, 1.0]]
        with self.assertRaises(DistanceMatrixError):
            self.matobj(data, ['a', 'b'])

        # Ensure that the superclass validation is still being performed.
        with self.assertRaises(DissimilarityMatrixError):
            self.matobj([[1, 2, 3]], ['a'])

    def test_init_nans(self):
        with self.assertRaisesRegex(DistanceMatrixError, r'NaNs'):
            self.matobj([[0.0, np.nan], [np.nan, 0.0]], ['a', 'b'])

    def test_from_iterable_no_key(self):
        iterable = (x for x in range(4))

        exp = self.matobj([[0, 1, 2, 3],
                           [1, 0, 1, 2],
                           [2, 1, 0, 1],
                           [3, 2, 1, 0]])
        res = self.matobj.from_iterable(iterable, lambda a, b: abs(b - a))
        self.assertEqual(res, exp)

    def test_from_iterable_validate_equal_valid_data(self):
        validate_true = self.matobj.from_iterable((x for x in range(4)),
                                                  lambda a, b: abs(b - a),
                                                  validate=True)
        validate_false = self.matobj.from_iterable((x for x in range(4)),
                                                   lambda a, b: abs(b - a),
                                                   validate=False)
        self.assertEqual(validate_true, validate_false)

    def test_from_iterable_validate_false(self):
        iterable = (x for x in range(4))

        exp = self.matobj([[0, 1, 2, 3],
                           [1, 0, 1, 2],
                           [2, 1, 0, 1],
                           [3, 2, 1, 0]])
        res = self.matobj.from_iterable(iterable, lambda a, b: abs(b - a),
                                        validate=False)
        self.assertEqual(res, exp)

    def test_from_iterable_validate_non_hollow(self):
        iterable = (x for x in range(4))
        with self.assertRaises(DistanceMatrixError):
            self.matobj.from_iterable(iterable, lambda a, b: 1)

    def test_from_iterable_validate_false_non_symmetric(self):
        exp = self.matobj([[0, 1, 2, 3],
                           [1, 0, 1, 2],
                           [2, 1, 0, 1],
                           [3, 2, 1, 0]])
        res = self.matobj.from_iterable((x for x in range(4)),
                                        lambda a, b: a - b,
                                        validate=False)
        self.assertEqual(res, exp)

    def test_from_iterable_validate_asym(self):
        iterable = (x for x in range(4))
        with self.assertRaises(DistanceMatrixError):
            self.matobj.from_iterable(iterable, lambda a, b: b - a)

    def test_from_iterable_with_key(self):
        iterable = (x for x in range(4))

        exp = self.matobj([[0, 1, 2, 3],
                           [1, 0, 1, 2],
                           [2, 1, 0, 1],
                           [3, 2, 1, 0]], ['0', '1', '4', '9'])
        res = self.matobj.from_iterable(iterable, lambda a, b: abs(b - a),
                                        key=lambda x: str(x**2))
        self.assertEqual(res, exp)

    def test_from_iterable_empty(self):
        with self.assertRaises(DissimilarityMatrixError):
            self.matobj.from_iterable([], lambda x: x)

    def test_from_iterable_single(self):
        exp = self.matobj([[0]])
        res = self.matobj.from_iterable(["boo"], lambda a, b: 0)
        self.assertEqual(res, exp)

    def test_from_iterable_with_keys(self):
        iterable = (x for x in range(4))

        exp = self.matobj([[0, 1, 2, 3],
                           [1, 0, 1, 2],
                           [2, 1, 0, 1],
                           [3, 2, 1, 0]], ['0', '1', '4', '9'])
        res = self.matobj.from_iterable(iterable, lambda a, b: abs(b - a),
                                        keys=iter(['0', '1', '4', '9']))
        self.assertEqual(res, exp)

    def test_from_iterable_with_key_and_keys(self):
        iterable = (x for x in range(4))
        with self.assertRaises(ValueError):
            self.matobj.from_iterable(iterable, lambda a, b: abs(b - a),
                                      key=str, keys=['1', '2', '3', '4'])

    def test_from_iterable_scipy_hamming_metric_with_metadata(self):
        # test for #1254
        seqs = [
            Sequence('ACGT'),
            Sequence('ACGA', metadata={'id': 'seq1'}),
            Sequence('AAAA', metadata={'id': 'seq2'}),
            Sequence('AAAA', positional_metadata={'qual': range(4)})
        ]

        exp = self.matobj([
            [0, 0.25, 0.75, 0.75],
            [0.25, 0.0, 0.5, 0.5],
            [0.75, 0.5, 0.0, 0.0],
            [0.75, 0.5, 0.0, 0.0]], ['a', 'b', 'c', 'd'])

        dm = self.matobj.from_iterable(
            seqs,
            metric=scipy.spatial.distance.hamming,
            keys=['a', 'b', 'c', 'd'])

        self.assertEqual(dm, exp)

    def test_from_iterable_skbio_hamming_metric_with_metadata(self):
        # test for #1254
        seqs = [
            Sequence('ACGT'),
            Sequence('ACGA', metadata={'id': 'seq1'}),
            Sequence('AAAA', metadata={'id': 'seq2'}),
            Sequence('AAAA', positional_metadata={'qual': range(4)})
        ]

        exp = self.matobj([
            [0, 0.25, 0.75, 0.75],
            [0.25, 0.0, 0.5, 0.5],
            [0.75, 0.5, 0.0, 0.0],
            [0.75, 0.5, 0.0, 0.0]], ['a', 'b', 'c', 'd'])

        dm = self.matobj.from_iterable(
            seqs,
            metric=skbio.sequence.distance.hamming,
            keys=['a', 'b', 'c', 'd'])

        self.assertEqual(dm, exp)

    def test_condensed_form(self):
        for dm, condensed in zip(self.dms, self.dm_condensed_forms):
            obs = dm.condensed_form()
            self.assertTrue(np.array_equal(obs, condensed))

    def test_permute_condensed(self):
        # Can't really permute a 1x1 or 2x2...
        for _ in range(2):
            obs = self.dm_1x1.permute(condensed=True)
            npt.assert_equal(obs, np.array([]))

        for _ in range(2):
            obs = self.dm_2x2.permute(condensed=True)
            npt.assert_equal(obs, np.array([0.123]))

        dm_copy = self.dm_3x3.copy()

        obs = self.dm_3x3.permute(condensed=True, seed=3)
        npt.assert_equal(obs, np.array([12.0, 4.2, 0.01]))

        obs = self.dm_3x3.permute(condensed=True, seed=2)
        npt.assert_equal(obs, np.array([4.2, 12.0, 0.01]))

        # Ensure dm hasn't changed after calling permute() on it a couple of
        # times.
        self.assertEqual(self.dm_3x3, dm_copy)

    def test_permute_not_condensed(self):
        obs = self.dm_1x1.permute()
        self.assertEqual(obs, self.dm_1x1)
        self.assertFalse(obs is self.dm_1x1)

        obs = self.dm_2x2.permute()
        self.assertEqual(obs, self.dm_2x2)
        self.assertFalse(obs is self.dm_2x2)

        exp = self.matobj([[0, 12, 4.2],
                           [12, 0, 0.01],
                           [4.2, 0.01, 0]], self.dm_3x3.ids)
        obs = self.dm_3x3.permute(seed=3)
        self.assertEqual(obs, exp)

        exp = self.matobj([[0, 4.2, 12],
                           [4.2, 0, 0.01],
                           [12, 0.01, 0]], self.dm_3x3.ids)
        obs = self.dm_3x3.permute(seed=2)
        self.assertEqual(obs, exp)

    def test_eq(self):
        # Compare DistanceMatrix to DissimilarityMatrix, where both have the
        # same data and IDs.
        eq_dm = DissimilarityMatrix(self.dm_3x3_data, ['a', 'b', 'c'])
        self.assertTrue(self.dm_3x3 == eq_dm)
        self.assertTrue(eq_dm == self.dm_3x3)

    def test_to_series_1x1(self):
        series = self.dm_1x1.to_series()

        exp = pd.Series([], index=[], dtype='float64')
        assert_series_almost_equal(series, exp)

    def test_to_series_2x2(self):
        series = self.dm_2x2.to_series()

        exp = pd.Series([0.123], index=pd.Index([('a', 'b')]))
        assert_series_almost_equal(series, exp)

    def test_to_series_4x4(self):
        dm = self.matobj([
            [0.0, 0.2, 0.3, 0.4],
            [0.2, 0.0, 0.5, 0.6],
            [0.3, 0.5, 0.0, 0.7],
            [0.4, 0.6, 0.7, 0.0]], ['a', 'b', 'c', 'd'])

        series = dm.to_series()

        exp = pd.Series([0.2, 0.3, 0.4, 0.5, 0.6, 0.7],
                        index=pd.Index([('a', 'b'), ('a', 'c'), ('a', 'd'),
                                        ('b', 'c'), ('b', 'd'), ('c', 'd')]))
        assert_series_almost_equal(series, exp)

    def test_to_series_default_ids(self):
        series = self.matobj(self.dm_2x2_data).to_series()

        exp = pd.Series([0.123], index=pd.Index([('0', '1')]))
        assert_series_almost_equal(series, exp)

    def test_validate_asym_shape(self):
        # first check it actually likes good matrices
        data_good = np.array([[0., 42.], [42., 0.]])
        data_sym, data_hollow = is_symmetric_and_hollow(data_good)
        self.assertEqual(data_sym, True)
        del data_sym
        self.assertEqual(data_hollow, True)
        del data_hollow
        data_sym = skbio.stats.distance._utils.is_symmetric(data_good)
        self.assertEqual(data_sym, True)
        del data_sym
        data_hollow = skbio.stats.distance._utils.is_hollow(data_good)
        self.assertEqual(data_hollow, True)
        del data_hollow
        self.dm_3x3._validate_shape(data_good)
        del data_good

        # _validate_shap checks just the shape, not the content
        bad_data = np.array([[1., 2.], [3., 4.]])
        data_sym, data_hollow = is_symmetric_and_hollow(bad_data)
        self.assertEqual(data_sym, False)
        del data_sym
        self.assertEqual(data_hollow, False)
        del data_hollow
        data_sym = skbio.stats.distance._utils.is_symmetric(bad_data)
        self.assertEqual(data_sym, False)
        del data_sym
        data_hollow = skbio.stats.distance._utils.is_hollow(bad_data)
        self.assertEqual(data_hollow, False)
        del data_hollow
        self.dm_3x3._validate_shape(bad_data)
        del bad_data

        # re-try with partially bad data
        bad_data = np.array([[0., 2.], [3., 0.]])
        data_sym, data_hollow = is_symmetric_and_hollow(bad_data)
        self.assertEqual(data_sym, False)
        del data_sym
        self.assertEqual(data_hollow, True)
        del data_hollow
        data_sym = skbio.stats.distance._utils.is_symmetric(bad_data)
        self.assertEqual(data_sym, False)
        del data_sym
        data_hollow = skbio.stats.distance._utils.is_hollow(bad_data)
        self.assertEqual(data_hollow, True)
        del data_hollow
        self.dm_3x3._validate_shape(bad_data)
        del bad_data
    
    def test_rename(self):
        # Test successful renaming with a dictionary in strict mode (default)
        dm = DistanceMatrix([[0, 1], [1, 0]], ids=['a', 'b'])
        rename_dict = {'a': 'x', 'b': 'y'}
        dm.rename(rename_dict)
        exp = ('x', 'y')
        self.assertEqual(dm.ids, exp)

        # Test successful renaming with a function in strict mode (default)
        dm = DistanceMatrix([[0, 1], [1, 0]], ids=['a', 'b'])
        rename_func = lambda x: x + '_1'
        dm.rename(rename_func)
        exp = ('a_1', 'b_1')
        self.assertEqual(dm.ids, exp)

        # Test renaming in non-strict mode where one ID is not included in the dictionary
        dm = DistanceMatrix([[0, 1], [1, 0]], ids=['a', 'b'])
        rename_dict = {'a': 'x'}  # 'b' will retain its original ID
        dm.rename(rename_dict, strict=False)
        exp = ('x', 'b')
        self.assertEqual(dm.ids, exp)

        # Test that renaming with strict=True raises an error if not all IDs are included
        dm = DistanceMatrix([[0, 1], [1, 0]], ids=['a', 'b'])
        rename_dict = {'a': 'x'}  # Missing 'b'
        with self.assertRaises(ValueError):
            dm.rename(rename_dict, strict=True)


class RandomDistanceMatrixTests(TestCase):
    def test_default_usage(self):
        exp = DistanceMatrix(np.asarray([[0.0]]), ['1'])
        obs = randdm(1)
        self.assertEqual(obs, exp)

        obs = randdm(2)
        self.assertEqual(obs.shape, (2, 2))
        self.assertEqual(obs.ids, ('1', '2'))

        obs1 = randdm(5)
        num_trials = 10
        found_diff = False
        for _ in range(num_trials):
            obs2 = randdm(5)

            if obs1 != obs2:
                found_diff = True
                break

        self.assertTrue(found_diff)

    def test_large_matrix_for_symmetry(self):
        obs3 = randdm(100)
        self.assertEqual(obs3, obs3.T)

    def test_ids(self):
        ids = ['foo', 'bar', 'baz']
        obs = randdm(3, ids=ids)
        self.assertEqual(obs.shape, (3, 3))
        self.assertEqual(obs.ids, tuple(ids))

    def test_constructor(self):
        exp = DissimilarityMatrix(np.asarray([[0.0]]), ['1'])
        obs = randdm(1, constructor=DissimilarityMatrix)
        self.assertEqual(obs, exp)
        self.assertEqual(type(obs), DissimilarityMatrix)

    def test_random_fn(self):
        def myrand(size):
            # One dm to rule them all...
            data = np.empty(size)
            data.fill(42)
            return data

        exp = DistanceMatrix(np.asarray([[0, 42, 42], [42, 0, 42],
                                         [42, 42, 0]]), ['1', '2', '3'])
        obs = randdm(3, random_fn=myrand)
        self.assertEqual(obs, exp)

    def test_random_seed(self):
        obs = randdm(5, random_fn=42).data
        exp = np.array([[0., 0.97562235, 0.37079802, 0.22723872, 0.75808774],
                        [0.97562235, 0., 0.92676499, 0.55458479, 0.35452597],
                        [0.37079802, 0.92676499, 0., 0.06381726, 0.97069802],
                        [0.22723872, 0.55458479, 0.06381726, 0., 0.89312112],
                        [0.75808774, 0.35452597, 0.97069802, 0.89312112, 0.]])
        npt.assert_almost_equal(obs, exp)

    def test_invalid_input(self):
        # Invalid dimensions.
        with self.assertRaises(DissimilarityMatrixError):
            randdm(0)

        # Invalid dimensions.
        with self.assertRaises(ValueError):
            randdm(-1)

        # Invalid number of IDs.
        with self.assertRaises(DissimilarityMatrixError):
            randdm(2, ids=['foo'])


class CategoricalStatsHelperFunctionTests(TestCase):
    def setUp(self):
        self.dm = DistanceMatrix([[0.0, 1.0, 2.0],
                                  [1.0, 0.0, 3.0],
                                  [2.0, 3.0, 0.0]], ['a', 'b', 'c'])
        self.grouping = [1, 2, 1]
        # Ordering of IDs shouldn't matter, nor should extra IDs.
        self.df = pd.read_csv(
            io.StringIO('ID,Group\nb,Group2\na,Group1\nc,Group1\nd,Group3'),
            index_col=0)
        self.df_missing_id = pd.read_csv(
            io.StringIO('ID,Group\nb,Group2\nc,Group1'), index_col=0)

    def test_preprocess_input_with_valid_input(self):
        # Should obtain same result using grouping vector or data frame.
        exp = (3, 2, np.array([0, 1, 0]),
               (np.array([0, 0, 1]), np.array([1, 2, 2])),
               np.array([1., 2., 3.]))

        obs = _preprocess_input(self.dm, self.grouping, None)
        npt.assert_equal(obs, exp)

        obs = _preprocess_input(self.dm, self.df, 'Group')
        npt.assert_equal(obs, exp)

    def test_preprocess_input_raises_error(self):
        # Requires a DistanceMatrix.
        with self.assertRaises(TypeError):
            _preprocess_input(
                DissimilarityMatrix([[0, 2], [3, 0]], ['a', 'b']),
                [1, 2], None)

        # Requires column if DataFrame.
        with self.assertRaises(ValueError):
            _preprocess_input(self.dm, self.df, None)

        # Cannot provide column if not data frame.
        with self.assertRaises(ValueError):
            _preprocess_input(self.dm, self.grouping, 'Group')

        # Column must exist in data frame.
        with self.assertRaises(ValueError):
            _preprocess_input(self.dm, self.df, 'foo')

        # All distance matrix IDs must be in data frame.
        with self.assertRaises(ValueError):
            _preprocess_input(self.dm, self.df_missing_id, 'Group')

        # Grouping vector length must match number of objects in dm.
        with self.assertRaises(ValueError):
            _preprocess_input(self.dm, [1, 2], None)

        # Grouping vector cannot have only unique values.
        with self.assertRaises(ValueError):
            _preprocess_input(self.dm, [1, 2, 3], None)

        # Grouping vector cannot have only a single group.
        with self.assertRaises(ValueError):
            _preprocess_input(self.dm, [1, 1, 1], None)

    def test_run_monte_carlo_stats_with_permutations(self):
        obs = _run_monte_carlo_stats(lambda e: 42, self.grouping, 50)
        npt.assert_equal(obs, (42, 1.0))

    def test_run_monte_carlo_stats_no_permutations(self):
        obs = _run_monte_carlo_stats(lambda e: 42, self.grouping, 0)
        npt.assert_equal(obs, (42, np.nan))

    def test_run_monte_carlo_stats_invalid_permutations(self):
        with self.assertRaises(ValueError):
            _run_monte_carlo_stats(lambda e: 42, self.grouping, -1)


class DissimilarityMatrixTests(DissimilarityMatrixTestBase, TestCase):
    matobj = DissimilarityMatrix

    def setUp(self):
        super(DissimilarityMatrixTests, self).setUp()


class DistanceMatrixTests(DistanceMatrixTestBase, TestCase):
    matobj = DistanceMatrix

    def setUp(self):
        super(DistanceMatrixTests, self).setUp()


if __name__ == '__main__':
    main()