BUG: .astype() of a Float64Index to a Int64Index

closes pandas-dev#12881
closes pandas-dev#12866
xref pandas-dev#12309

Author: jreback

doc/source/whatsnew/v0.18.1.txt

@@ -267,7 +267,8 @@ Bug Fixes
 - Bug in ``float_format`` option with option not being validated as a callable. (:issue:`12706`)
 - Bug in ``GroupBy.filter`` when ``dropna=False`` and no groups fulfilled the criteria (:issue:`12768`)
 - Bug in ``__name__`` of ``.cum*`` functions (:issue:`12021`)
-
+- Bug in ``.astype()`` of a ``Float64Inde/Int64Index`` to an ``Int64Index`` (:issue:`12881`)
+- Bug in roundtripping an integer based index in ``.to_json()/.read_json()`` when ``orient='index'`` (the default) (:issue:`12866`)
 
 - Bug in ``.drop()`` with a non-unique ``MultiIndex``. (:issue:`12701`)
 - Bug in ``.concat`` of datetime tz-aware and naive DataFrames (:issue:`12467`)

pandas/indexes/numeric.py

@@ -7,7 +7,9 @@
 from pandas.indexes.base import Index, InvalidIndexError
 from pandas.util.decorators import Appender, cache_readonly
 import pandas.core.common as com
-from pandas.core.common import is_dtype_equal, isnull
+from pandas.core.common import (is_dtype_equal, isnull, pandas_dtype,
+                                is_float_dtype, is_object_dtype,
+                                is_integer_dtype)
 import pandas.indexes.base as ibase
 
 
@@ -101,12 +103,7 @@ def __new__(cls, data=None, dtype=None, copy=False, name=None,
             cls._string_data_error(data)
 
         elif issubclass(data.dtype.type, np.integer):
-            # don't force the upcast as we may be dealing
-            # with a platform int
-            if (dtype is None or
-                    not issubclass(np.dtype(dtype).type, np.integer)):
-                dtype = np.int64
-
+            dtype = np.int64
             subarr = np.array(data, dtype=dtype, copy=copy)
         else:
             subarr = np.array(data, dtype=np.int64, copy=copy)
@@ -219,12 +216,9 @@ def __new__(cls, data=None, dtype=None, copy=False, name=None,
         dtype = np.dtype(dtype)
 
         # allow integer / object dtypes to be passed, but coerce to float64
-        if dtype.kind in ['i', 'O']:
+        if dtype.kind in ['i', 'O', 'f']:
             dtype = np.float64
 
-        elif dtype.kind in ['f']:
-            pass
-
         else:
             raise TypeError("cannot support {0} dtype in "
                             "Float64Index".format(dtype))
@@ -245,11 +239,16 @@ def inferred_type(self):
         return 'floating'
 
     def astype(self, dtype):
-        if np.dtype(dtype) not in (np.object, np.float64):
+        dtype = pandas_dtype(dtype)
+        if is_float_dtype(dtype) or is_integer_dtype(dtype):
+            values = self._values.astype(dtype)
+        elif is_object_dtype(dtype):
+            values = self._values
+        else:
             raise TypeError('Setting %s dtype to anything other than '
                             'float64 or object is not supported' %
                             self.__class__)
-        return Index(self._values, name=self.name, dtype=dtype)
+        return Index(values, name=self.name, dtype=dtype)
 
     def _convert_scalar_indexer(self, key, kind=None):
         """

pandas/io/tests/test_json/test_pandas.py

@@ -576,10 +576,7 @@ def test_reconstruction_index(self):
         df = DataFrame([[1, 2, 3], [4, 5, 6]])
         result = read_json(df.to_json())
 
-        self.assertEqual(result.index.dtype, np.float64)
-        self.assertEqual(result.columns.dtype, np.float64)
-        assert_frame_equal(result, df, check_index_type=False,
-                           check_column_type=False)
+        assert_frame_equal(result, df)
 
         df = DataFrame({'a': [1, 2, 3], 'b': [4, 5, 6]}, index=['A', 'B', 'C'])
         result = read_json(df.to_json())
@@ -776,23 +773,20 @@ def test_timedelta(self):
 
         s = Series([timedelta(23), timedelta(seconds=5)])
         self.assertEqual(s.dtype, 'timedelta64[ns]')
-        # index will be float dtype
-        assert_series_equal(s, pd.read_json(s.to_json(), typ='series')
-                            .apply(converter),
-                            check_index_type=False)
+
+        result = pd.read_json(s.to_json(), typ='series').apply(converter)
+        assert_series_equal(result, s)
 
         s = Series([timedelta(23), timedelta(seconds=5)],
-                   index=pd.Index([0, 1], dtype=float))
+                   index=pd.Index([0, 1]))
         self.assertEqual(s.dtype, 'timedelta64[ns]')
-        assert_series_equal(s, pd.read_json(
-            s.to_json(), typ='series').apply(converter))
+        result = pd.read_json(s.to_json(), typ='series').apply(converter)
+        assert_series_equal(result, s)
 
         frame = DataFrame([timedelta(23), timedelta(seconds=5)])
         self.assertEqual(frame[0].dtype, 'timedelta64[ns]')
         assert_frame_equal(frame, pd.read_json(frame.to_json())
-                           .apply(converter),
-                           check_index_type=False,
-                           check_column_type=False)
+                           .apply(converter))
 
         frame = DataFrame({'a': [timedelta(days=23), timedelta(seconds=5)],
                            'b': [1, 2],
@@ -801,7 +795,7 @@ def test_timedelta(self):
         result = pd.read_json(frame.to_json(date_unit='ns'))
         result['a'] = pd.to_timedelta(result.a, unit='ns')
         result['c'] = pd.to_datetime(result.c)
-        assert_frame_equal(frame, result, check_index_type=False)
+        assert_frame_equal(frame, result)
 
     def test_mixed_timedelta_datetime(self):
         frame = DataFrame({'a': [timedelta(23), pd.Timestamp('20130101')]},

pandas/tests/indexes/test_numeric.py

@@ -231,6 +231,34 @@ def test_astype(self):
         self.assertTrue(i.equals(result))
         self.check_is_index(result)
 
+        # GH 12881
+        # a float astype int
+        for dtype in ['int16', 'int32', 'int64']:
+            i = Float64Index([0, 1, 2])
+            result = i.astype(dtype)
+            expected = Int64Index([0, 1, 2])
+            tm.assert_index_equal(result, expected)
+
+            i = Float64Index([0, 1.1, 2])
+            result = i.astype(dtype)
+            expected = Int64Index([0, 1, 2])
+            tm.assert_index_equal(result, expected)
+
+        for dtype in ['float32', 'float64']:
+            i = Float64Index([0, 1, 2])
+            result = i.astype(dtype)
+            expected = i
+            tm.assert_index_equal(result, expected)
+
+            i = Float64Index([0, 1.1, 2])
+            result = i.astype(dtype)
+            expected = Index(i.values.astype(dtype))
+            tm.assert_index_equal(result, expected)
+
+        # invalid
+        for dtype in ['M8[ns]', 'm8[ns]']:
+            self.assertRaises(TypeError, lambda: i.astype(dtype))
+
     def test_equals(self):
 
         i = Float64Index([1.0, 2.0])

pandas/tests/test_common.py

@@ -652,27 +652,6 @@ def test_ensure_int32():
     assert (result.dtype == np.int32)
 
 
-def test_ensure_platform_int():
-
-    # verify that when we create certain types of indices
-    # they remain the correct type under platform conversions
-    from pandas.core.index import Int64Index
-
-    # int64
-    x = Int64Index([1, 2, 3], dtype='int64')
-    assert (x.dtype == np.int64)
-
-    pi = com._ensure_platform_int(x)
-    assert (pi.dtype == np.int_)
-
-    # int32
-    x = Int64Index([1, 2, 3], dtype='int32')
-    assert (x.dtype == np.int32)
-
-    pi = com._ensure_platform_int(x)
-    assert (pi.dtype == np.int_)
-
-
 def test_is_re():
     passes = re.compile('ad'),
     fails = 'x', 2, 3, object()