|
| 1 | +import numpy as np |
| 2 | +import pytest |
| 3 | + |
| 4 | + |
| 5 | +from pandas import Index, IntervalIndex, Timestamp, interval_range |
| 6 | +import pandas.util.testing as tm |
| 7 | + |
| 8 | + |
| 9 | +@pytest.fixture(scope='class', params=[None, 'foo']) |
| 10 | +def name(request): |
| 11 | + return request.param |
| 12 | + |
| 13 | + |
| 14 | +class TestIntervalIndex: |
| 15 | + |
| 16 | + def create_index(self, closed='right'): |
| 17 | + return IntervalIndex.from_breaks(range(11), closed=closed) |
| 18 | + |
| 19 | + @pytest.mark.parametrize("sort", [None, False]) |
| 20 | + def test_union(self, closed, sort): |
| 21 | + index = self.create_index(closed=closed) |
| 22 | + other = IntervalIndex.from_breaks(range(5, 13), closed=closed) |
| 23 | + |
| 24 | + expected = IntervalIndex.from_breaks(range(13), closed=closed) |
| 25 | + result = index[::-1].union(other, sort=sort) |
| 26 | + if sort is None: |
| 27 | + tm.assert_index_equal(result, expected) |
| 28 | + assert tm.equalContents(result, expected) |
| 29 | + |
| 30 | + result = other[::-1].union(index, sort=sort) |
| 31 | + if sort is None: |
| 32 | + tm.assert_index_equal(result, expected) |
| 33 | + assert tm.equalContents(result, expected) |
| 34 | + |
| 35 | + tm.assert_index_equal(index.union(index, sort=sort), index) |
| 36 | + tm.assert_index_equal(index.union(index[:1], sort=sort), index) |
| 37 | + |
| 38 | + # GH 19101: empty result, same dtype |
| 39 | + index = IntervalIndex(np.array([], dtype='int64'), closed=closed) |
| 40 | + result = index.union(index, sort=sort) |
| 41 | + tm.assert_index_equal(result, index) |
| 42 | + |
| 43 | + # GH 19101: empty result, different dtypes |
| 44 | + other = IntervalIndex(np.array([], dtype='float64'), closed=closed) |
| 45 | + result = index.union(other, sort=sort) |
| 46 | + tm.assert_index_equal(result, index) |
| 47 | + |
| 48 | + @pytest.mark.parametrize("sort", [None, False]) |
| 49 | + def test_intersection(self, closed, sort): |
| 50 | + index = self.create_index(closed=closed) |
| 51 | + other = IntervalIndex.from_breaks(range(5, 13), closed=closed) |
| 52 | + |
| 53 | + expected = IntervalIndex.from_breaks(range(5, 11), closed=closed) |
| 54 | + result = index[::-1].intersection(other, sort=sort) |
| 55 | + if sort is None: |
| 56 | + tm.assert_index_equal(result, expected) |
| 57 | + assert tm.equalContents(result, expected) |
| 58 | + |
| 59 | + result = other[::-1].intersection(index, sort=sort) |
| 60 | + if sort is None: |
| 61 | + tm.assert_index_equal(result, expected) |
| 62 | + assert tm.equalContents(result, expected) |
| 63 | + |
| 64 | + tm.assert_index_equal(index.intersection(index, sort=sort), index) |
| 65 | + |
| 66 | + # GH 19101: empty result, same dtype |
| 67 | + other = IntervalIndex.from_breaks(range(300, 314), closed=closed) |
| 68 | + expected = IntervalIndex(np.array([], dtype='int64'), closed=closed) |
| 69 | + result = index.intersection(other, sort=sort) |
| 70 | + tm.assert_index_equal(result, expected) |
| 71 | + |
| 72 | + # GH 19101: empty result, different dtypes |
| 73 | + breaks = np.arange(300, 314, dtype='float64') |
| 74 | + other = IntervalIndex.from_breaks(breaks, closed=closed) |
| 75 | + result = index.intersection(other, sort=sort) |
| 76 | + tm.assert_index_equal(result, expected) |
| 77 | + |
| 78 | + # GH 26225: nested intervals |
| 79 | + index = IntervalIndex.from_tuples([(1, 2), (1, 3), (1, 4), (0, 2)]) |
| 80 | + other = IntervalIndex.from_tuples([(1, 2), (1, 3)]) |
| 81 | + expected = IntervalIndex.from_tuples([(1, 2), (1, 3)]) |
| 82 | + result = index.intersection(other) |
| 83 | + tm.assert_index_equal(result, expected) |
| 84 | + |
| 85 | + # GH 26225: duplicate element |
| 86 | + index = IntervalIndex.from_tuples([(1, 2), (1, 2), (2, 3), (3, 4)]) |
| 87 | + other = IntervalIndex.from_tuples([(1, 2), (2, 3)]) |
| 88 | + expected = IntervalIndex.from_tuples([(1, 2), (1, 2), (2, 3)]) |
| 89 | + result = index.intersection(other) |
| 90 | + tm.assert_index_equal(result, expected) |
| 91 | + |
| 92 | + @pytest.mark.parametrize("sort", [None, False]) |
| 93 | + def test_difference(self, closed, sort): |
| 94 | + index = IntervalIndex.from_arrays([1, 0, 3, 2], |
| 95 | + [1, 2, 3, 4], |
| 96 | + closed=closed) |
| 97 | + result = index.difference(index[:1], sort=sort) |
| 98 | + expected = index[1:] |
| 99 | + if sort is None: |
| 100 | + expected = expected.sort_values() |
| 101 | + tm.assert_index_equal(result, expected) |
| 102 | + |
| 103 | + # GH 19101: empty result, same dtype |
| 104 | + result = index.difference(index, sort=sort) |
| 105 | + expected = IntervalIndex(np.array([], dtype='int64'), closed=closed) |
| 106 | + tm.assert_index_equal(result, expected) |
| 107 | + |
| 108 | + # GH 19101: empty result, different dtypes |
| 109 | + other = IntervalIndex.from_arrays(index.left.astype('float64'), |
| 110 | + index.right, closed=closed) |
| 111 | + result = index.difference(other, sort=sort) |
| 112 | + tm.assert_index_equal(result, expected) |
| 113 | + |
| 114 | + @pytest.mark.parametrize("sort", [None, False]) |
| 115 | + def test_symmetric_difference(self, closed, sort): |
| 116 | + index = self.create_index(closed=closed) |
| 117 | + result = index[1:].symmetric_difference(index[:-1], sort=sort) |
| 118 | + expected = IntervalIndex([index[0], index[-1]]) |
| 119 | + if sort is None: |
| 120 | + tm.assert_index_equal(result, expected) |
| 121 | + assert tm.equalContents(result, expected) |
| 122 | + |
| 123 | + # GH 19101: empty result, same dtype |
| 124 | + result = index.symmetric_difference(index, sort=sort) |
| 125 | + expected = IntervalIndex(np.array([], dtype='int64'), closed=closed) |
| 126 | + if sort is None: |
| 127 | + tm.assert_index_equal(result, expected) |
| 128 | + assert tm.equalContents(result, expected) |
| 129 | + |
| 130 | + # GH 19101: empty result, different dtypes |
| 131 | + other = IntervalIndex.from_arrays(index.left.astype('float64'), |
| 132 | + index.right, closed=closed) |
| 133 | + result = index.symmetric_difference(other, sort=sort) |
| 134 | + tm.assert_index_equal(result, expected) |
| 135 | + |
| 136 | + @pytest.mark.parametrize('op_name', [ |
| 137 | + 'union', 'intersection', 'difference', 'symmetric_difference']) |
| 138 | + @pytest.mark.parametrize("sort", [None, False]) |
| 139 | + def test_set_operation_errors(self, closed, op_name, sort): |
| 140 | + index = self.create_index(closed=closed) |
| 141 | + set_op = getattr(index, op_name) |
| 142 | + |
| 143 | + # non-IntervalIndex |
| 144 | + msg = ('the other index needs to be an IntervalIndex too, but ' |
| 145 | + 'was type Int64Index') |
| 146 | + with pytest.raises(TypeError, match=msg): |
| 147 | + set_op(Index([1, 2, 3]), sort=sort) |
| 148 | + |
| 149 | + # mixed closed |
| 150 | + msg = ('can only do set operations between two IntervalIndex objects ' |
| 151 | + 'that are closed on the same side') |
| 152 | + for other_closed in {'right', 'left', 'both', 'neither'} - {closed}: |
| 153 | + other = self.create_index(closed=other_closed) |
| 154 | + with pytest.raises(ValueError, match=msg): |
| 155 | + set_op(other, sort=sort) |
| 156 | + |
| 157 | + # GH 19016: incompatible dtypes |
| 158 | + other = interval_range(Timestamp('20180101'), periods=9, closed=closed) |
| 159 | + msg = ('can only do {op} between two IntervalIndex objects that have ' |
| 160 | + 'compatible dtypes').format(op=op_name) |
| 161 | + with pytest.raises(TypeError, match=msg): |
| 162 | + set_op(other, sort=sort) |
0 commit comments