BUG: cant modify df with duplicate index (#17105) (#20939)

Author: fersarr

doc/source/whatsnew/v0.23.0.txt

@@ -1253,6 +1253,7 @@ Indexing
 - Bug in ``Series.is_unique`` where extraneous output in stderr is shown if Series contains objects with ``__ne__`` defined (:issue:`20661`)
 - Bug in ``.loc`` assignment with a single-element list-like incorrectly assigns as a list (:issue:`19474`)
 - Bug in partial string indexing on a ``Series/DataFrame`` with a monotonic decreasing ``DatetimeIndex`` (:issue:`19362`)
+- Bug in performing in-place operations on a ``DataFrame`` with a duplicate ``Index`` (:issue:`17105`)
 - Bug in :meth:`IntervalIndex.get_loc` and :meth:`IntervalIndex.get_indexer` when used with an :class:`IntervalIndex` containing a single interval (:issue:`17284`, :issue:`20921`)
 
 MultiIndex

pandas/core/indexing.py

@@ -1318,7 +1318,7 @@ def _convert_to_indexer(self, obj, axis=None, is_setter=False):
                     (indexer,
                      missing) = labels.get_indexer_non_unique(objarr)
                     # 'indexer' has dupes, create 'check' using 'missing'
-                    check = np.zeros_like(objarr)
+                    check = np.zeros(len(objarr))
                     check[missing] = -1
 
                 mask = check == -1

pandas/tests/frame/test_indexing.py

@@ -1929,6 +1929,32 @@ def test_iloc_duplicates(self):
         expected = df.take([0], axis=1)
         assert_frame_equal(result, expected)
 
+    def test_loc_duplicates(self):
+        # gh-17105
+
+        # insert a duplicate element to the index
+        trange = pd.date_range(start=pd.Timestamp(year=2017, month=1, day=1),
+                               end=pd.Timestamp(year=2017, month=1, day=5))
+
+        trange = trange.insert(loc=5,
+                               item=pd.Timestamp(year=2017, month=1, day=5))
+
+        df = pd.DataFrame(0, index=trange, columns=["A", "B"])
+        bool_idx = np.array([False, False, False, False, False, True])
+
+        # assignment
+        df.loc[trange[bool_idx], "A"] = 6
+
+        expected = pd.DataFrame({'A': [0, 0, 0, 0, 6, 6],
+                                 'B': [0, 0, 0, 0, 0, 0]},
+                                index=trange)
+        tm.assert_frame_equal(df, expected)
+
+        # in-place
+        df = pd.DataFrame(0, index=trange, columns=["A", "B"])
+        df.loc[trange[bool_idx], "A"] += 6
+        tm.assert_frame_equal(df, expected)
+
     def test_iloc_sparse_propegate_fill_value(self):
         from pandas.core.sparse.api import SparseDataFrame
         df = SparseDataFrame({'A': [999, 1]}, default_fill_value=999)