diff --git a/doc/source/whatsnew/v1.0.0.rst b/doc/source/whatsnew/v1.0.0.rst index b075a9d8b5e8b..3b55901ec4740 100644 --- a/doc/source/whatsnew/v1.0.0.rst +++ b/doc/source/whatsnew/v1.0.0.rst @@ -247,6 +247,7 @@ Missing MultiIndex ^^^^^^^^^^ +- Constructior for :class:`MultiIndex` verifies that the given ``sortorder`` is compatible with the actual ``lexsort_depth`` if ``verify_integrity`` parameter is ``True`` (the default) (:issue:`28735`) - - @@ -288,6 +289,7 @@ Reshaping - Bug in :meth:`DataFrame.apply` that caused incorrect output with empty :class:`DataFrame` (:issue:`28202`, :issue:`21959`) - Bug in :meth:`DataFrame.stack` not handling non-unique indexes correctly when creating MultiIndex (:issue: `28301`) - Bug :func:`merge_asof` could not use :class:`datetime.timedelta` for ``tolerance`` kwarg (:issue:`28098`) +- Bug in :func:`merge`, did not append suffixes correctly with MultiIndex (:issue:`28518`) Sparse ^^^^^^ diff --git a/pandas/core/indexes/multi.py b/pandas/core/indexes/multi.py index 3273c4f8cd13b..c02561087f83f 100644 --- a/pandas/core/indexes/multi.py +++ b/pandas/core/indexes/multi.py @@ -365,6 +365,15 @@ def _verify_integrity(self, codes=None, levels=None): "Level values must be unique: {values} on " "level {level}".format(values=[value for value in level], level=i) ) + if self.sortorder is not None: + if self.sortorder > self._lexsort_depth(): + raise ValueError( + "Value for sortorder must be inferior or equal " + "to actual lexsort_depth: " + "sortorder {sortorder} with lexsort_depth {lexsort_depth}".format( + sortorder=self.sortorder, lexsort_depth=self._lexsort_depth() + ) + ) codes = [ self._validate_codes(level, code) for level, code in zip(levels, codes) @@ -1783,16 +1792,23 @@ def is_lexsorted(self): @cache_readonly def lexsort_depth(self): if self.sortorder is not None: - if self.sortorder == 0: - return self.nlevels - else: - return 0 + return self.sortorder + + return self._lexsort_depth() + def _lexsort_depth(self) -> int: + """ + Compute and return the lexsort_depth, the number of levels of the + MultiIndex that are sorted lexically + + Returns + ------ + int + """ int64_codes = [ensure_int64(level_codes) for level_codes in self.codes] for k in range(self.nlevels, 0, -1): if libalgos.is_lexsorted(int64_codes[:k]): return k - return 0 def _sort_levels_monotonic(self): diff --git a/pandas/tests/indexes/multi/test_indexing.py b/pandas/tests/indexes/multi/test_indexing.py index d366dbd8bc0a8..ec2e8aa6564a8 100644 --- a/pandas/tests/indexes/multi/test_indexing.py +++ b/pandas/tests/indexes/multi/test_indexing.py @@ -96,7 +96,6 @@ def test_slice_locs_not_contained(): index = MultiIndex( levels=[[0, 2, 4, 6], [0, 2, 4]], codes=[[0, 0, 0, 1, 1, 2, 3, 3, 3], [0, 1, 2, 1, 2, 2, 0, 1, 2]], - sortorder=0, ) result = index.slice_locs((1, 0), (5, 2)) diff --git a/pandas/tests/reshape/merge/test_merge.py b/pandas/tests/reshape/merge/test_merge.py index a04f093ee7818..63de9777756cc 100644 --- a/pandas/tests/reshape/merge/test_merge.py +++ b/pandas/tests/reshape/merge/test_merge.py @@ -2094,3 +2094,33 @@ def test_merge_equal_cat_dtypes2(): # Categorical is unordered, so don't check ordering. tm.assert_frame_equal(result, expected, check_categorical=False) + + +def test_merge_multiindex_columns(): + # Issue #28518 + # Verify that merging two dataframes give the expected labels + # The original cause of this issue come from a bug lexsort_depth and is tested in + # test_lexsort_depth + + letters = ["a", "b", "c", "d"] + numbers = ["1", "2", "3"] + index = pd.MultiIndex.from_product((letters, numbers), names=["outer", "inner"]) + + frame_x = pd.DataFrame(columns=index) + frame_x["id"] = "" + frame_y = pd.DataFrame(columns=index) + frame_y["id"] = "" + + l_suf = "_x" + r_suf = "_y" + result = frame_x.merge(frame_y, on="id", suffixes=((l_suf, r_suf))) + + # Constructing the expected results + expected_labels = [l + l_suf for l in letters] + [l + r_suf for l in letters] + expected_index = pd.MultiIndex.from_product( + [expected_labels, numbers], names=["outer", "inner"] + ) + expected = pd.DataFrame(columns=expected_index) + expected["id"] = "" + + tm.assert_frame_equal(result, expected) diff --git a/pandas/tests/test_multilevel.py b/pandas/tests/test_multilevel.py index dc4db6e7902a8..4a60d3966a9bb 100644 --- a/pandas/tests/test_multilevel.py +++ b/pandas/tests/test_multilevel.py @@ -2056,6 +2056,53 @@ def test_is_lexsorted(self): assert not index.is_lexsorted() assert index.lexsort_depth == 0 + def test_raise_invalid_sortorder(self): + # Test that the MultiIndex constructor raise when a incorrect sortorder is given + # Issue #28518 + + levels = [[0, 1], [0, 1, 2]] + + # Correct sortorder + MultiIndex( + levels=levels, codes=[[0, 0, 0, 1, 1, 1], [0, 1, 2, 0, 1, 2]], sortorder=2 + ) + + with pytest.raises(ValueError, match=r".* sortorder 2 with lexsort_depth 1.*"): + MultiIndex( + levels=levels, + codes=[[0, 0, 0, 1, 1, 1], [0, 1, 2, 0, 2, 1]], + sortorder=2, + ) + + with pytest.raises(ValueError, match=r".* sortorder 1 with lexsort_depth 0.*"): + MultiIndex( + levels=levels, + codes=[[0, 0, 1, 0, 1, 1], [0, 1, 0, 2, 2, 1]], + sortorder=1, + ) + + def test_lexsort_depth(self): + # Test that lexsort_depth return the correct sortorder + # when it was given to the MultiIndex const. + # Issue #28518 + + levels = [[0, 1], [0, 1, 2]] + + index = MultiIndex( + levels=levels, codes=[[0, 0, 0, 1, 1, 1], [0, 1, 2, 0, 1, 2]], sortorder=2 + ) + assert index.lexsort_depth == 2 + + index = MultiIndex( + levels=levels, codes=[[0, 0, 0, 1, 1, 1], [0, 1, 2, 0, 2, 1]], sortorder=1 + ) + assert index.lexsort_depth == 1 + + index = MultiIndex( + levels=levels, codes=[[0, 0, 1, 0, 1, 1], [0, 1, 0, 2, 2, 1]], sortorder=0 + ) + assert index.lexsort_depth == 0 + def test_sort_index_and_reconstruction(self): # 15622