Raise ValueError when using levels with non-unique values in MultiIndex constructor

doc/source/whatsnew/v0.21.0.txt

@@ -432,7 +432,7 @@ Other API Changes
 - :class:`Period` is now immutable, and will now raise an ``AttributeError`` when a user tries to assign a new value to the ``ordinal`` or ``freq`` attributes (:issue:`17116`).
 - :func:`to_datetime` when passed a tz-aware ``origin=`` kwarg will now raise a more informative ``ValueError`` rather than a ``TypeError`` (:issue:`16842`)
 - Renamed non-functional ``index`` to ``index_col`` in :func:`read_stata` to improve API consistency (:issue:`16342`)
-
+- :class:`MultiIndex` constructor now checks if the values of each level are unique when ``verify_integrity=True`` (:issue:`17464`)
 
 .. _whatsnew_0210.deprecations:
 

pandas/core/indexes/multi.py

@@ -147,6 +147,11 @@ def _verify_integrity(self, labels=None, levels=None):
                                  " level  (%d). NOTE: this index is in an"
                                  " inconsistent state" % (i, label.max(),
                                                           len(level)))
+        for i, level in enumerate(levels):
+            if not level.is_unique:
+                raise ValueError("Level values must be unique: {0!r}"
+                                 " on level {1}".format([value for value
+                                                         in level], i))
 
     def _get_levels(self):
         return self._levels

pandas/tests/indexes/test_multi.py

@@ -588,6 +588,14 @@ def test_constructor_mismatched_label_levels(self):
             with tm.assert_raises_regex(ValueError, label_error):
                 self.index.copy().labels = [[0, 0, 0, 0], [0, 0]]
 
+    def test_constructor_non_unique_level_values(self):
+        # GH #17464
+        with tm.assert_raises_regex(ValueError, '^Level values'):
+            MultiIndex(levels=[[0, 1], [0, 0, 1, 1]],
+                       labels=[[0, 0, 0, 0, 1, 1, 1, 1],
+                               [0, 0, 1, 1, 0, 0, 1, 1]],
+                       names=[u'idx0', u'idx1'])
+
     def assert_multiindex_copied(self, copy, original):
         # Levels should be (at least, shallow copied)
         tm.assert_copy(copy.levels, original.levels)