CLN: Stopped object inference in constructors for pandas objects (#58758)

* CLN: Stopped object inference in constructors for pandas objects

* Adjust tests

Author: mroeschke

doc/source/whatsnew/v3.0.0.rst

@@ -311,6 +311,7 @@ Removal of prior version deprecations/changes
 - Removed the deprecated ``delim_whitespace`` keyword in :func:`read_csv` and :func:`read_table`, use ``sep=r"\s+"`` instead (:issue:`55569`)
 - Require :meth:`SparseDtype.fill_value` to be a valid value for the :meth:`SparseDtype.subtype` (:issue:`53043`)
 - Stopped automatically casting non-datetimelike values (mainly strings) in :meth:`Series.isin` and :meth:`Index.isin` with ``datetime64``, ``timedelta64``, and :class:`PeriodDtype` dtypes (:issue:`53111`)
+- Stopped performing dtype inference in :class:`Index`, :class:`Series` and :class:`DataFrame` constructors when given a pandas object (:class:`Series`, :class:`Index`, :class:`ExtensionArray`), call ``.infer_objects`` on the input to keep the current behavior (:issue:`56012`)
 - Stopped performing dtype inference when setting a :class:`Index` into a :class:`DataFrame` (:issue:`56102`)
 - Stopped performing dtype inference with in :meth:`Index.insert` with object-dtype index; this often affects the index/columns that result when setting new entries into an empty :class:`Series` or :class:`DataFrame` (:issue:`51363`)
 - Removed the "closed" and "unit" keywords in :meth:`TimedeltaIndex.__new__` (:issue:`52628`, :issue:`55499`)

pandas/_testing/__init__.py

@@ -10,7 +10,6 @@
     ContextManager,
     cast,
 )
-import warnings
 
 import numpy as np
 
@@ -290,17 +289,11 @@ def box_expected(expected, box_cls, transpose: bool = True):
         else:
             expected = pd.array(expected, copy=False)
     elif box_cls is Index:
-        with warnings.catch_warnings():
-            warnings.filterwarnings("ignore", "Dtype inference", category=FutureWarning)
-            expected = Index(expected)
+        expected = Index(expected)
     elif box_cls is Series:
-        with warnings.catch_warnings():
-            warnings.filterwarnings("ignore", "Dtype inference", category=FutureWarning)
-            expected = Series(expected)
+        expected = Series(expected)
     elif box_cls is DataFrame:
-        with warnings.catch_warnings():
-            warnings.filterwarnings("ignore", "Dtype inference", category=FutureWarning)
-            expected = Series(expected).to_frame()
+        expected = Series(expected).to_frame()
         if transpose:
             # for vector operations, we need a DataFrame to be a single-row,
             #  not a single-column, in order to operate against non-DataFrame

pandas/core/construction.py

@@ -554,7 +554,7 @@ def sanitize_array(
         # Avoid ending up with a NumpyExtensionArray
         dtype = dtype.numpy_dtype
 
-    data_was_index = isinstance(data, ABCIndex)
+    infer_object = not isinstance(data, (ABCIndex, ABCSeries))
 
     # extract ndarray or ExtensionArray, ensure we have no NumpyExtensionArray
     data = extract_array(data, extract_numpy=True, extract_range=True)
@@ -607,7 +607,7 @@ def sanitize_array(
 
         if dtype is None:
             subarr = data
-            if data.dtype == object and not data_was_index:
+            if data.dtype == object and infer_object:
                 subarr = maybe_infer_to_datetimelike(data)
             elif data.dtype.kind == "U" and using_pyarrow_string_dtype():
                 from pandas.core.arrays.string_ import StringDtype

pandas/core/frame.py

@@ -728,10 +728,6 @@ def __init__(
                 NDFrame.__init__(self, data)
                 return
 
-        is_pandas_object = isinstance(data, (Series, Index, ExtensionArray))
-        data_dtype = getattr(data, "dtype", None)
-        original_dtype = dtype
-
         # GH47215
         if isinstance(index, set):
             raise ValueError("index cannot be a set")
@@ -896,18 +892,6 @@ def __init__(
 
         NDFrame.__init__(self, mgr)
 
-        if original_dtype is None and is_pandas_object and data_dtype == np.object_:
-            if self.dtypes.iloc[0] != data_dtype:
-                warnings.warn(
-                    "Dtype inference on a pandas object "
-                    "(Series, Index, ExtensionArray) is deprecated. The DataFrame "
-                    "constructor will keep the original dtype in the future. "
-                    "Call `infer_objects` on the result to get the old "
-                    "behavior.",
-                    FutureWarning,
-                    stacklevel=2,
-                )
-
     # ----------------------------------------------------------------------
 
     def __dataframe__(

pandas/core/indexes/base.py

@@ -490,8 +490,6 @@ def __new__(
         if not copy and isinstance(data, (ABCSeries, Index)):
             refs = data._references
 
-        is_pandas_object = isinstance(data, (ABCSeries, Index, ExtensionArray))
-
         # range
         if isinstance(data, (range, RangeIndex)):
             result = RangeIndex(start=data, copy=copy, name=name)
@@ -508,7 +506,7 @@ def __new__(
         elif is_ea_or_datetimelike_dtype(data_dtype):
             pass
 
-        elif isinstance(data, (np.ndarray, Index, ABCSeries)):
+        elif isinstance(data, (np.ndarray, ABCMultiIndex)):
             if isinstance(data, ABCMultiIndex):
                 data = data._values
 
@@ -518,7 +516,9 @@ def __new__(
                 # they are actually ints, e.g. '0' and 0.0
                 # should not be coerced
                 data = com.asarray_tuplesafe(data, dtype=_dtype_obj)
-
+        elif isinstance(data, (ABCSeries, Index)):
+            # GH 56244: Avoid potential inference on object types
+            pass
         elif is_scalar(data):
             raise cls._raise_scalar_data_error(data)
         elif hasattr(data, "__array__"):
@@ -571,19 +571,7 @@ def __new__(
         klass = cls._dtype_to_subclass(arr.dtype)
 
         arr = klass._ensure_array(arr, arr.dtype, copy=False)
-        result = klass._simple_new(arr, name, refs=refs)
-        if dtype is None and is_pandas_object and data_dtype == np.object_:
-            if result.dtype != data_dtype:
-                warnings.warn(
-                    "Dtype inference on a pandas object "
-                    "(Series, Index, ExtensionArray) is deprecated. The Index "
-                    "constructor will keep the original dtype in the future. "
-                    "Call `infer_objects` on the result to get the old "
-                    "behavior.",
-                    FutureWarning,
-                    stacklevel=2,
-                )
-        return result  # type: ignore[return-value]
+        return klass._simple_new(arr, name, refs=refs)
 
     @classmethod
     def _ensure_array(cls, data, dtype, copy: bool):

pandas/core/internals/construction.py

@@ -192,6 +192,7 @@ def ndarray_to_mgr(
 ) -> Manager:
     # used in DataFrame.__init__
     # input must be a ndarray, list, Series, Index, ExtensionArray
+    infer_object = not isinstance(values, (ABCSeries, Index, ExtensionArray))
 
     if isinstance(values, ABCSeries):
         if columns is None:
@@ -287,15 +288,14 @@ def ndarray_to_mgr(
     # if we don't have a dtype specified, then try to convert objects
     # on the entire block; this is to convert if we have datetimelike's
     # embedded in an object type
-    if dtype is None and is_object_dtype(values.dtype):
+    if dtype is None and infer_object and is_object_dtype(values.dtype):
         obj_columns = list(values)
         maybe_datetime = [maybe_infer_to_datetimelike(x) for x in obj_columns]
         # don't convert (and copy) the objects if no type inference occurs
         if any(x is not y for x, y in zip(obj_columns, maybe_datetime)):
-            dvals_list = [ensure_block_shape(dval, 2) for dval in maybe_datetime]
             block_values = [
-                new_block_2d(dvals_list[n], placement=BlockPlacement(n))
-                for n in range(len(dvals_list))
+                new_block_2d(ensure_block_shape(dval, 2), placement=BlockPlacement(n))
+                for n, dval in enumerate(maybe_datetime)
             ]
         else:
             bp = BlockPlacement(slice(len(columns)))

pandas/core/series.py

@@ -389,10 +389,6 @@ def __init__(
             self.name = name
             return
 
-        is_pandas_object = isinstance(data, (Series, Index, ExtensionArray))
-        data_dtype = getattr(data, "dtype", None)
-        original_dtype = dtype
-
         if isinstance(data, (ExtensionArray, np.ndarray)):
             if copy is not False:
                 if dtype is None or astype_is_view(data.dtype, pandas_dtype(dtype)):
@@ -438,7 +434,6 @@ def __init__(
                 data = data.astype(dtype)
 
             refs = data._references
-            data = data._values
             copy = False
 
         elif isinstance(data, np.ndarray):
@@ -512,17 +507,6 @@ def __init__(
         self.name = name
         self._set_axis(0, index)
 
-        if original_dtype is None and is_pandas_object and data_dtype == np.object_:
-            if self.dtype != data_dtype:
-                warnings.warn(
-                    "Dtype inference on a pandas object "
-                    "(Series, Index, ExtensionArray) is deprecated. The Series "
-                    "constructor will keep the original dtype in the future. "
-                    "Call `infer_objects` on the result to get the old behavior.",
-                    FutureWarning,
-                    stacklevel=find_stack_level(),
-                )
-
     def _init_dict(
         self, data: Mapping, index: Index | None = None, dtype: DtypeObj | None = None
     ):

pandas/tests/copy_view/test_constructors.py

@@ -228,12 +228,12 @@ def test_dataframe_from_series_or_index_different_dtype(index_or_series):
     assert df._mgr._has_no_reference(0)
 
 
-def test_dataframe_from_series_infer_datetime():
+def test_dataframe_from_series_dont_infer_datetime():
     ser = Series([Timestamp("2019-12-31"), Timestamp("2020-12-31")], dtype=object)
-    with tm.assert_produces_warning(FutureWarning, match="Dtype inference"):
-        df = DataFrame(ser)
-    assert not np.shares_memory(get_array(ser), get_array(df, 0))
-    assert df._mgr._has_no_reference(0)
+    df = DataFrame(ser)
+    assert df.dtypes.iloc[0] == np.dtype(object)
+    assert np.shares_memory(get_array(ser), get_array(df, 0))
+    assert not df._mgr._has_no_reference(0)
 
 
 @pytest.mark.parametrize("index", [None, [0, 1, 2]])

pandas/tests/frame/test_constructors.py

@@ -2702,21 +2702,14 @@ def test_frame_string_inference_block_dim(self):
             df = DataFrame(np.array([["hello", "goodbye"], ["hello", "Hello"]]))
         assert df._mgr.blocks[0].ndim == 2
 
-    def test_inference_on_pandas_objects(self):
+    @pytest.mark.parametrize("klass", [Series, Index])
+    def test_inference_on_pandas_objects(self, klass):
         # GH#56012
-        idx = Index([Timestamp("2019-12-31")], dtype=object)
-        with tm.assert_produces_warning(FutureWarning, match="Dtype inference"):
-            result = DataFrame(idx, columns=["a"])
-        assert result.dtypes.iloc[0] != np.object_
-        result = DataFrame({"a": idx})
+        obj = klass([Timestamp("2019-12-31")], dtype=object)
+        result = DataFrame(obj, columns=["a"])
         assert result.dtypes.iloc[0] == np.object_
 
-        ser = Series([Timestamp("2019-12-31")], dtype=object)
-
-        with tm.assert_produces_warning(FutureWarning, match="Dtype inference"):
-            result = DataFrame(ser, columns=["a"])
-        assert result.dtypes.iloc[0] != np.object_
-        result = DataFrame({"a": ser})
+        result = DataFrame({"a": obj})
         assert result.dtypes.iloc[0] == np.object_
 
     def test_dict_keys_returns_rangeindex(self):

pandas/tests/indexes/base_class/test_constructors.py

@@ -59,18 +59,12 @@ def test_index_string_inference(self):
             ser = Index(["a", 1])
         tm.assert_index_equal(ser, expected)
 
-    def test_inference_on_pandas_objects(self):
+    @pytest.mark.parametrize("klass", [Series, Index])
+    def test_inference_on_pandas_objects(self, klass):
         # GH#56012
-        idx = Index([pd.Timestamp("2019-12-31")], dtype=object)
-        with tm.assert_produces_warning(FutureWarning, match="Dtype inference"):
-            result = Index(idx)
-        assert result.dtype != np.object_
-
-        ser = Series([pd.Timestamp("2019-12-31")], dtype=object)
-
-        with tm.assert_produces_warning(FutureWarning, match="Dtype inference"):
-            result = Index(ser)
-        assert result.dtype != np.object_
+        obj = klass([pd.Timestamp("2019-12-31")], dtype=object)
+        result = Index(obj)
+        assert result.dtype == np.object_
 
     def test_constructor_not_read_only(self):
         # GH#57130

pandas/tests/indexes/test_base.py

@@ -104,23 +104,20 @@ def test_constructor_copy(self, using_infer_string):
     )
     def test_constructor_from_index_dtlike(self, cast_as_obj, index):
         if cast_as_obj:
-            with tm.assert_produces_warning(FutureWarning, match="Dtype inference"):
-                result = Index(index.astype(object))
-        else:
-            result = Index(index)
-
-        tm.assert_index_equal(result, index)
-
-        if isinstance(index, DatetimeIndex):
-            assert result.tz == index.tz
-            if cast_as_obj:
+            result = Index(index.astype(object))
+            assert result.dtype == np.dtype(object)
+            if isinstance(index, DatetimeIndex):
                 # GH#23524 check that Index(dti, dtype=object) does not
                 #  incorrectly raise ValueError, and that nanoseconds are not
                 #  dropped
                 index += pd.Timedelta(nanoseconds=50)
                 result = Index(index, dtype=object)
                 assert result.dtype == np.object_
                 assert list(result) == list(index)
+        else:
+            result = Index(index)
+
+            tm.assert_index_equal(result, index)
 
     @pytest.mark.parametrize(
         "index,has_tz",

pandas/tests/series/accessors/test_dt_accessor.py

@@ -256,9 +256,8 @@ def test_dt_accessor_limited_display_api(self):
         tm.assert_almost_equal(results, sorted(set(ok_for_dt + ok_for_dt_methods)))
 
         # Period
-        idx = period_range("20130101", periods=5, freq="D", name="xxx").astype(object)
-        with tm.assert_produces_warning(FutureWarning, match="Dtype inference"):
-            ser = Series(idx)
+        idx = period_range("20130101", periods=5, freq="D", name="xxx")
+        ser = Series(idx)
         results = get_dir(ser)
         tm.assert_almost_equal(
             results, sorted(set(ok_for_period + ok_for_period_methods))

pandas/tests/series/methods/test_equals.py

@@ -82,15 +82,13 @@ def test_equals_matching_nas():
     left = Series([np.datetime64("NaT")], dtype=object)
     right = Series([np.datetime64("NaT")], dtype=object)
     assert left.equals(right)
-    with tm.assert_produces_warning(FutureWarning, match="Dtype inference"):
-        assert Index(left).equals(Index(right))
+    assert Index(left).equals(Index(right))
     assert left.array.equals(right.array)
 
     left = Series([np.timedelta64("NaT")], dtype=object)
     right = Series([np.timedelta64("NaT")], dtype=object)
     assert left.equals(right)
-    with tm.assert_produces_warning(FutureWarning, match="Dtype inference"):
-        assert Index(left).equals(Index(right))
+    assert Index(left).equals(Index(right))
     assert left.array.equals(right.array)
 
     left = Series([np.float64("NaN")], dtype=object)

pandas/tests/series/test_constructors.py

@@ -1318,9 +1318,8 @@ def test_constructor_periodindex(self):
         pi = period_range("20130101", periods=5, freq="D")
         s = Series(pi)
         assert s.dtype == "Period[D]"
-        with tm.assert_produces_warning(FutureWarning, match="Dtype inference"):
-            expected = Series(pi.astype(object))
-        tm.assert_series_equal(s, expected)
+        expected = Series(pi.astype(object))
+        assert expected.dtype == object
 
     def test_constructor_dict(self):
         d = {"a": 0.0, "b": 1.0, "c": 2.0}
@@ -2137,20 +2136,14 @@ def test_series_string_inference_na_first(self):
             result = Series([pd.NA, "b"])
         tm.assert_series_equal(result, expected)
 
-    def test_inference_on_pandas_objects(self):
+    @pytest.mark.parametrize("klass", [Series, Index])
+    def test_inference_on_pandas_objects(self, klass):
         # GH#56012
-        ser = Series([Timestamp("2019-12-31")], dtype=object)
-        with tm.assert_produces_warning(None):
-            # This doesn't do inference
-            result = Series(ser)
+        obj = klass([Timestamp("2019-12-31")], dtype=object)
+        # This doesn't do inference
+        result = Series(obj)
         assert result.dtype == np.object_
 
-        idx = Index([Timestamp("2019-12-31")], dtype=object)
-
-        with tm.assert_produces_warning(FutureWarning, match="Dtype inference"):
-            result = Series(idx)
-        assert result.dtype != np.object_
-
 
 class TestSeriesConstructorIndexCoercion:
     def test_series_constructor_datetimelike_index_coercion(self):

pandas/tests/tseries/frequencies/test_inference.py

@@ -23,7 +23,6 @@
     date_range,
     period_range,
 )
-import pandas._testing as tm
 from pandas.core.arrays import (
     DatetimeArray,
     TimedeltaArray,
@@ -202,17 +201,6 @@ def test_infer_freq_custom(base_delta_code_pair, constructor):
     assert frequencies.infer_freq(index) is None
 
 
-@pytest.mark.parametrize(
-    "freq,expected", [("Q", "QE-DEC"), ("Q-NOV", "QE-NOV"), ("Q-OCT", "QE-OCT")]
-)
-def test_infer_freq_index(freq, expected):
-    rng = period_range("1959Q2", "2009Q3", freq=freq)
-    with tm.assert_produces_warning(FutureWarning, match="Dtype inference"):
-        rng = Index(rng.to_timestamp("D", how="e").astype(object))
-
-    assert rng.inferred_freq == expected
-
-
 @pytest.mark.parametrize(
     "expected,dates",
     list(