DEPR: enforce Series/DataFrame awareness-mismatch deprecations (pandas-dev#48739)

* DEPR: enforce deprecation on Series(ts_aware, dtype=naive)

* DEPR: enforce deprecation on Series(tzaware_seq, dtype=naive)

* rename

* move whatsnew

Author: jbrockmendel

doc/source/whatsnew/v2.0.0.rst

@@ -146,6 +146,8 @@ Deprecations
 
 Removal of prior version deprecations/changes
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+- Enforced deprecation disallowing passing a timezone-aware :class:`Timestamp` and ``dtype="datetime64[ns]"`` to :class:`Series` or :class:`DataFrame` constructors (:issue:`41555`)
+- Enforced deprecation disallowing passing a sequence of timezone-aware values and ``dtype="datetime64[ns]"`` to to :class:`Series` or :class:`DataFrame` constructors (:issue:`41555`)
 - Removed Date parser functions :func:`~pandas.io.date_converters.parse_date_time`,
   :func:`~pandas.io.date_converters.parse_date_fields`, :func:`~pandas.io.date_converters.parse_all_fields`
   and :func:`~pandas.io.date_converters.generic_parser` (:issue:`24518`)

pandas/core/dtypes/cast.py

@@ -1358,20 +1358,11 @@ def maybe_cast_to_datetime(
                         # didn't specify one
 
                         if dta.tz is not None:
-                            warnings.warn(
-                                "Data is timezone-aware. Converting "
-                                "timezone-aware data to timezone-naive by "
-                                "passing dtype='datetime64[ns]' to "
-                                "DataFrame or Series is deprecated and will "
-                                "raise in a future version. Use "
-                                "`pd.Series(values).dt.tz_localize(None)` "
-                                "instead.",
-                                FutureWarning,
-                                stacklevel=find_stack_level(),
+                            raise ValueError(
+                                "Cannot convert timezone-aware data to "
+                                "timezone-naive dtype. Use "
+                                "pd.Series(values).dt.tz_localize(None) instead."
                             )
-                            # equiv: dta.view(dtype)
-                            # Note: NOT equivalent to dta.astype(dtype)
-                            dta = dta.tz_localize(None)
 
                         # TODO(2.0): Do this astype in sequence_to_datetimes to
                         #  avoid potential extra copy?
@@ -1678,7 +1669,7 @@ def construct_2d_arraylike_from_scalar(
     shape = (length, width)
 
     if dtype.kind in ["m", "M"]:
-        value = _maybe_unbox_datetimelike_tz_deprecation(value, dtype)
+        value = _maybe_box_and_unbox_datetimelike(value, dtype)
     elif dtype == _dtype_obj:
         if isinstance(value, (np.timedelta64, np.datetime64)):
             # calling np.array below would cast to pytimedelta/pydatetime
@@ -1742,7 +1733,7 @@ def construct_1d_arraylike_from_scalar(
             if not isna(value):
                 value = ensure_str(value)
         elif dtype.kind in ["M", "m"]:
-            value = _maybe_unbox_datetimelike_tz_deprecation(value, dtype)
+            value = _maybe_box_and_unbox_datetimelike(value, dtype)
 
         subarr = np.empty(length, dtype=dtype)
         if length:
@@ -1752,42 +1743,14 @@ def construct_1d_arraylike_from_scalar(
     return subarr
 
 
-def _maybe_unbox_datetimelike_tz_deprecation(value: Scalar, dtype: DtypeObj):
-    """
-    Wrap _maybe_unbox_datetimelike with a check for a timezone-aware Timestamp
-    along with a timezone-naive datetime64 dtype, which is deprecated.
-    """
+def _maybe_box_and_unbox_datetimelike(value: Scalar, dtype: DtypeObj):
     # Caller is responsible for checking dtype.kind in ["m", "M"]
 
     if isinstance(value, datetime):
         # we dont want to box dt64, in particular datetime64("NaT")
         value = maybe_box_datetimelike(value, dtype)
 
-    try:
-        value = _maybe_unbox_datetimelike(value, dtype)
-    except TypeError:
-        if (
-            isinstance(value, Timestamp)
-            and value.tzinfo is not None
-            and isinstance(dtype, np.dtype)
-            and dtype.kind == "M"
-        ):
-            warnings.warn(
-                "Data is timezone-aware. Converting "
-                "timezone-aware data to timezone-naive by "
-                "passing dtype='datetime64[ns]' to "
-                "DataFrame or Series is deprecated and will "
-                "raise in a future version. Use "
-                "`pd.Series(values).dt.tz_localize(None)` "
-                "instead.",
-                FutureWarning,
-                stacklevel=find_stack_level(),
-            )
-            new_value = value.tz_localize(None)
-            return _maybe_unbox_datetimelike(new_value, dtype)
-        else:
-            raise
-    return value
+    return _maybe_unbox_datetimelike(value, dtype)
 
 
 def construct_1d_object_array_from_listlike(values: Sized) -> np.ndarray:

pandas/tests/frame/test_constructors.py

@@ -2848,37 +2848,32 @@ def test_constructor_data_aware_dtype_naive(self, tz_aware_fixture, pydt):
         ts = Timestamp("2019", tz=tz)
         if pydt:
             ts = ts.to_pydatetime()
-        ts_naive = Timestamp("2019")
 
-        with tm.assert_produces_warning(FutureWarning):
-            result = DataFrame({0: [ts]}, dtype="datetime64[ns]")
+        msg = (
+            "Cannot convert timezone-aware data to timezone-naive dtype. "
+            r"Use pd.Series\(values\).dt.tz_localize\(None\) instead."
+        )
+        with pytest.raises(ValueError, match=msg):
+            DataFrame({0: [ts]}, dtype="datetime64[ns]")
 
-        expected = DataFrame({0: [ts_naive]})
-        tm.assert_frame_equal(result, expected)
+        msg2 = "Cannot unbox tzaware Timestamp to tznaive dtype"
+        with pytest.raises(TypeError, match=msg2):
+            DataFrame({0: ts}, index=[0], dtype="datetime64[ns]")
 
-        with tm.assert_produces_warning(FutureWarning):
-            result = DataFrame({0: ts}, index=[0], dtype="datetime64[ns]")
-        tm.assert_frame_equal(result, expected)
+        with pytest.raises(ValueError, match=msg):
+            DataFrame([ts], dtype="datetime64[ns]")
 
-        with tm.assert_produces_warning(FutureWarning):
-            result = DataFrame([ts], dtype="datetime64[ns]")
-        tm.assert_frame_equal(result, expected)
+        with pytest.raises(ValueError, match=msg):
+            DataFrame(np.array([ts], dtype=object), dtype="datetime64[ns]")
 
-        with tm.assert_produces_warning(FutureWarning):
-            result = DataFrame(np.array([ts], dtype=object), dtype="datetime64[ns]")
-        tm.assert_frame_equal(result, expected)
+        with pytest.raises(TypeError, match=msg2):
+            DataFrame(ts, index=[0], columns=[0], dtype="datetime64[ns]")
 
-        with tm.assert_produces_warning(FutureWarning):
-            result = DataFrame(ts, index=[0], columns=[0], dtype="datetime64[ns]")
-        tm.assert_frame_equal(result, expected)
-
-        with tm.assert_produces_warning(FutureWarning):
-            df = DataFrame([Series([ts])], dtype="datetime64[ns]")
-        tm.assert_frame_equal(result, expected)
+        with pytest.raises(ValueError, match=msg):
+            DataFrame([Series([ts])], dtype="datetime64[ns]")
 
-        with tm.assert_produces_warning(FutureWarning):
-            df = DataFrame([[ts]], columns=[0], dtype="datetime64[ns]")
-        tm.assert_equal(df, expected)
+        with pytest.raises(ValueError, match=msg):
+            DataFrame([[ts]], columns=[0], dtype="datetime64[ns]")
 
     def test_from_dict(self):
 
@@ -3051,8 +3046,11 @@ def get1(obj):  # TODO: make a helper in tm?
 
 class TestFromScalar:
     @pytest.fixture(params=[list, dict, None])
-    def constructor(self, request, frame_or_series):
-        box = request.param
+    def box(self, request):
+        return request.param
+
+    @pytest.fixture
+    def constructor(self, frame_or_series, box):
 
         extra = {"index": range(2)}
         if frame_or_series is DataFrame:
@@ -3181,16 +3179,22 @@ def test_out_of_s_bounds_timedelta64(self, constructor):
         dtype = result.dtype if isinstance(result, Series) else result.dtypes.iloc[0]
         assert dtype == object
 
-    def test_tzaware_data_tznaive_dtype(self, constructor):
+    def test_tzaware_data_tznaive_dtype(self, constructor, box, frame_or_series):
         tz = "US/Eastern"
         ts = Timestamp("2019", tz=tz)
-        ts_naive = Timestamp("2019")
 
-        with tm.assert_produces_warning(FutureWarning, match="Data is timezone-aware"):
-            result = constructor(ts, dtype="M8[ns]")
+        if box is None or (frame_or_series is DataFrame and box is dict):
+            msg = "Cannot unbox tzaware Timestamp to tznaive dtype"
+            err = TypeError
+        else:
+            msg = (
+                "Cannot convert timezone-aware data to timezone-naive dtype. "
+                r"Use pd.Series\(values\).dt.tz_localize\(None\) instead."
+            )
+            err = ValueError
 
-        assert np.all(result.dtypes == "M8[ns]")
-        assert np.all(result == ts_naive)
+        with pytest.raises(err, match=msg):
+            constructor(ts, dtype="M8[ns]")
 
 
 # TODO: better location for this test?

pandas/tests/series/test_constructors.py

@@ -1736,24 +1736,23 @@ def test_constructor_data_aware_dtype_naive(self, tz_aware_fixture, pydt):
         ts = Timestamp("2019", tz=tz)
         if pydt:
             ts = ts.to_pydatetime()
-        ts_naive = Timestamp("2019")
 
-        with tm.assert_produces_warning(FutureWarning):
-            result = Series([ts], dtype="datetime64[ns]")
-        expected = Series([ts_naive])
-        tm.assert_series_equal(result, expected)
+        msg = (
+            "Cannot convert timezone-aware data to timezone-naive dtype. "
+            r"Use pd.Series\(values\).dt.tz_localize\(None\) instead."
+        )
+        with pytest.raises(ValueError, match=msg):
+            Series([ts], dtype="datetime64[ns]")
 
-        with tm.assert_produces_warning(FutureWarning):
-            result = Series(np.array([ts], dtype=object), dtype="datetime64[ns]")
-        tm.assert_series_equal(result, expected)
+        with pytest.raises(ValueError, match=msg):
+            Series(np.array([ts], dtype=object), dtype="datetime64[ns]")
 
-        with tm.assert_produces_warning(FutureWarning):
-            result = Series({0: ts}, dtype="datetime64[ns]")
-        tm.assert_series_equal(result, expected)
+        with pytest.raises(ValueError, match=msg):
+            Series({0: ts}, dtype="datetime64[ns]")
 
-        with tm.assert_produces_warning(FutureWarning):
-            result = Series(ts, index=[0], dtype="datetime64[ns]")
-        tm.assert_series_equal(result, expected)
+        msg = "Cannot unbox tzaware Timestamp to tznaive dtype"
+        with pytest.raises(TypeError, match=msg):
+            Series(ts, index=[0], dtype="datetime64[ns]")
 
     def test_constructor_datetime64(self):
         rng = date_range("1/1/2000 00:00:00", "1/1/2000 1:59:50", freq="10s")