MAINT: More useful error msg on Index overflow

Display a more friendly error message when there
is an OverflowError during Index construction.

Partially addresses pandas-devgh-15832.

pandas/core/indexes/base.py

@@ -306,7 +306,14 @@ def __new__(cls, data=None, dtype=None, copy=False, name=None,
                     if is_integer_dtype(dtype):
                         inferred = lib.infer_dtype(data)
                         if inferred == 'integer':
-                            data = np.array(data, copy=copy, dtype=dtype)
+                            try:
+                                data = np.array(data, copy=copy, dtype=dtype)
+                            except OverflowError:
+                                # gh-15823: a more user-friendly error message
+                                raise OverflowError(
+                                    "the elements provided in the data cannot "
+                                    "all be casted to the dtype {dtype}"
+                                    .format(dtype=dtype))
                         elif inferred in ['floating', 'mixed-integer-float']:
                             if isna(data).any():
                                 raise ValueError('cannot convert float '

pandas/tests/indexes/test_base.py

@@ -474,6 +474,13 @@ def test_constructor_nonhashable_name(self, indices):
         tm.assert_raises_regex(TypeError, message,
                                indices.set_names, names=renamed)
 
+    def test_constructor_overflow_int64(self):
+        # see gh-15832
+        msg = ("the elements provided in the data cannot "
+               "all be casted to the dtype int64")
+        with tm.assert_raises_regex(OverflowError, msg):
+            Index([np.iinfo(np.uint64).max - 1], dtype="int64")
+
     def test_view_with_args(self):
 
         restricted = ['unicodeIndex', 'strIndex', 'catIndex', 'boolIndex',