maxframe 2.2.0__cp310-cp310-macosx_10_9_universal2.whl → 2.3.0rc1__cp310-cp310-macosx_10_9_universal2.whl

maxframe/dataframe/misc/infer_dtypes.py

@@ -0,0 +1,251 @@
+# Copyright 1999-2025 Alibaba Group Holding Ltd.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from ... import opcodes
+from ...serialization.serializables import AnyField, StringField
+from ..core import DATAFRAME_TYPE, SERIES_TYPE
+from ..operators import DataFrameOperator, DataFrameOperatorMixin
+
+
+class DataFrameInferDtypes(DataFrameOperator, DataFrameOperatorMixin):
+    _op_type_ = opcodes.DATAFRAME_INFER_DTYPES
+
+    infer_method = StringField("infer_method")
+    infer_kwargs = AnyField("infer_kwargs")
+
+    infer_stage = StringField("infer_stage", default=None)
+
+    def __init__(self, output_types=None, **kw):
+        super().__init__(_output_types=output_types, **kw)
+
+    def __call__(self, df):
+        if isinstance(df, DATAFRAME_TYPE):
+            return self.new_dataframe(
+                [df],
+                shape=df.shape,
+                dtypes=None,
+                index_value=df.index_value,
+                columns_value=df.columns_value,
+            )
+        else:
+            assert isinstance(df, SERIES_TYPE)
+            return self.new_series(
+                [df],
+                shape=df.shape,
+                dtype=None,
+                name=df.name,
+                index_value=df.index_value,
+            )
+
+
+def convert_dtypes(
+    df_or_series,
+    infer_objects=True,
+    convert_string=True,
+    convert_integer=True,
+    convert_boolean=True,
+    convert_floating=True,
+    dtype_backend="numpy",
+):
+    """
+    Convert columns to best possible dtypes using dtypes supporting ``pd.NA``.
+
+    Parameters
+    ----------
+    infer_objects : bool, default True
+        Whether object dtypes should be converted to the best possible types.
+    convert_string : bool, default True
+        Whether object dtypes should be converted to ``StringDtype()``.
+    convert_integer : bool, default True
+        Whether, if possible, conversion can be done to integer extension types.
+    convert_boolean : bool, defaults True
+        Whether object dtypes should be converted to ``BooleanDtypes()``.
+    convert_floating : bool, defaults True
+        Whether, if possible, conversion can be done to floating extension types.
+        If `convert_integer` is also True, preference will be give to integer
+        dtypes if the floats can be faithfully casted to integers.
+
+    Returns
+    -------
+    Series or DataFrame
+        Copy of input object with new dtype.
+
+    See Also
+    --------
+    infer_objects : Infer dtypes of objects.
+    to_datetime : Convert argument to datetime.
+    to_timedelta : Convert argument to timedelta.
+    to_numeric : Convert argument to a numeric type.
+
+    Notes
+    -----
+    By default, ``convert_dtypes`` will attempt to convert a Series (or each
+    Series in a DataFrame) to dtypes that support ``pd.NA``. By using the options
+    ``convert_string``, ``convert_integer``, ``convert_boolean`` and
+    ``convert_boolean``, it is possible to turn off individual conversions
+    to ``StringDtype``, the integer extension types, ``BooleanDtype``
+    or floating extension types, respectively.
+
+    For object-dtyped columns, if ``infer_objects`` is ``True``, use the inference
+    rules as during normal Series/DataFrame construction.  Then, if possible,
+    convert to ``StringDtype``, ``BooleanDtype`` or an appropriate integer
+    or floating extension type, otherwise leave as ``object``.
+
+    If the dtype is integer, convert to an appropriate integer extension type.
+
+    If the dtype is numeric, and consists of all integers, convert to an
+    appropriate integer extension type. Otherwise, convert to an
+    appropriate floating extension type.
+
+    .. versionchanged:: 1.2
+        Starting with pandas 1.2, this method also converts float columns
+        to the nullable floating extension type.
+
+    In the future, as new dtypes are added that support ``pd.NA``, the results
+    of this method will change to support those new dtypes.
+
+    Examples
+    --------
+    >>> import maxframe.tensor as mt
+    >>> import maxframe.dataframe as md
+    >>> df = md.DataFrame(
+    ...     {
+    ...         "a": md.Series([1, 2, 3], dtype=mt.dtype("int32")),
+    ...         "b": md.Series(["x", "y", "z"], dtype=mt.dtype("O")),
+    ...         "c": md.Series([True, False, mt.nan], dtype=mt.dtype("O")),
+    ...         "d": md.Series(["h", "i", mt.nan], dtype=mt.dtype("O")),
+    ...         "e": md.Series([10, mt.nan, 20], dtype=mt.dtype("float")),
+    ...         "f": md.Series([mt.nan, 100.5, 200], dtype=mt.dtype("float")),
+    ...     }
+    ... )
+
+    Start with a DataFrame with default dtypes.
+
+    >>> df.execute()
+       a  b      c    d     e      f
+    0  1  x   True    h  10.0    NaN
+    1  2  y  False    i   NaN  100.5
+    2  3  z    NaN  NaN  20.0  200.0
+
+    >>> df.dtypes.execute()
+    a      int32
+    b     object
+    c     object
+    d     object
+    e    float64
+    f    float64
+    dtype: object
+
+    Convert the DataFrame to use best possible dtypes.
+
+    >>> dfn = df.convert_dtypes()
+    >>> dfn.execute()
+       a  b      c     d     e      f
+    0  1  x   True     h    10   <NA>
+    1  2  y  False     i  <NA>  100.5
+    2  3  z   <NA>  <NA>    20  200.0
+
+    >>> dfn.dtypes.execute()
+    a      Int32
+    b     string
+    c    boolean
+    d     string
+    e      Int64
+    f    Float64
+    dtype: object
+
+    Start with a Series of strings and missing data represented by ``np.nan``.
+
+    >>> s = md.Series(["a", "b", mt.nan])
+    >>> s.execute()
+    0      a
+    1      b
+    2    NaN
+    dtype: object
+
+    Obtain a Series with dtype ``StringDtype``.
+
+    >>> s.convert_dtypes().execute()
+    0       a
+    1       b
+    2    <NA>
+    dtype: string
+    """
+    dtype_backend = "numpy" if dtype_backend == "numpy_nullable" else dtype_backend
+    op = DataFrameInferDtypes(
+        infer_method="convert_dtypes",
+        infer_kwargs=dict(
+            infer_objects=infer_objects,
+            convert_string=convert_string,
+            convert_integer=convert_integer,
+            convert_boolean=convert_boolean,
+            convert_floating=convert_floating,
+            dtype_backend=dtype_backend,
+        ),
+    )
+    return op(df_or_series)
+
+
+def infer_objects(df_or_series, copy=True):
+    """
+    Attempt to infer better dtypes for object columns.
+
+    Attempts soft conversion of object-dtyped
+    columns, leaving non-object and unconvertible
+    columns unchanged. The inference rules are the
+    same as during normal Series/DataFrame construction.
+
+    Returns
+    -------
+    converted : same type as input object
+
+    See Also
+    --------
+    to_datetime : Convert argument to datetime.
+    to_timedelta : Convert argument to timedelta.
+    to_numeric : Convert argument to numeric type.
+    convert_dtypes : Convert argument to best possible dtype.
+
+    Examples
+    --------
+    >>> import maxframe.dataframe as md
+    >>> df = md.DataFrame({"A": ["a", 1, 2, 3]})
+    >>> df = df.iloc[1:]
+    >>> df.execute()
+       A
+    1  1
+    2  2
+    3  3
+
+    >>> df.dtypes.execute()
+    A    object
+    dtype: object
+
+    >>> df.infer_objects().dtypes.execute()
+    A    int64
+    dtype: object
+    """
+    if (isinstance(df_or_series, SERIES_TYPE) and df_or_series.dtype != "O") or (
+        isinstance(df_or_series, DATAFRAME_TYPE)
+        and all(dt != "O" for dt in df_or_series.dtypes)
+    ):
+        # no objects to cast
+        return df_or_series
+
+    _ = copy  # in MaxFrame data are immutable, thus ignore the parameter
+    op = DataFrameInferDtypes(
+        infer_method="infer_objects",
+        infer_kwargs={},
+    )
+    return op(df_or_series)

maxframe/dataframe/misc/map.py

@@ -21,8 +21,8 @@ import pandas as pd
 from ... import opcodes
 from ...core import EntityData, OutputType
 from ...serialization.serializables import AnyField, KeyField, StringField
-from ...udf import BuiltinFunction, MarkedFunction
-from ...utils import quiet_stdio
+from ...udf import BuiltinFunction, MarkedFunction, ODPSFunction
+from ...utils import make_dtype, quiet_stdio
 from ..core import SERIES_TYPE
 from ..operators import DataFrameOperator, DataFrameOperatorMixin
 from ..utils import build_series, copy_func_scheduling_hints
@@ -40,6 +40,7 @@ class DataFrameMap(DataFrameOperator, DataFrameOperatorMixin):
         if not self.output_types:
             self.output_types = [OutputType.series]
         if hasattr(self, "arg"):
+            self.arg = ODPSFunction.wrap(self.arg)
             copy_func_scheduling_hints(self.arg, self)
 
     @classmethod
@@ -55,25 +56,34 @@ class DataFrameMap(DataFrameOperator, DataFrameOperatorMixin):
         ) and not isinstance(self.arg, BuiltinFunction)
 
     def __call__(self, series, dtype, skip_infer=False):
-        if dtype is None and not skip_infer:
-            inferred_dtype = None
-            if callable(self.arg):
+        if dtype is not None:
+            dtype = make_dtype(dtype)
+        else:
+            # obtain dtype from existing hints
+            if isinstance(self.arg, ODPSFunction):
+                if self.arg.result_dtype is not None:
+                    dtype = self.arg.result_dtype
+            elif callable(self.arg):
                 # arg is a function, try to inspect the signature
                 sig = inspect.signature(self.arg)
                 return_type = sig.return_annotation
                 if return_type is not inspect._empty:
-                    inferred_dtype = np.dtype(return_type)
-                else:
-                    try:
-                        with quiet_stdio():
-                            # try to infer dtype by calling the function
-                            inferred_dtype = (
-                                build_series(series)
-                                .map(self.arg, na_action=self.na_action)
-                                .dtype
-                            )
-                    except:  # noqa: E722  # nosec
-                        pass
+                    dtype = np.dtype(return_type)
+
+        err_prefix = None
+        if dtype is None and not skip_infer:
+            inferred_dtype = None
+            if callable(self.arg):
+                try:
+                    with quiet_stdio():
+                        # try to infer dtype by calling the function
+                        inferred_dtype = (
+                            build_series(series)
+                            .map(self.arg, na_action=self.na_action)
+                            .dtype
+                        )
+                except:  # noqa: E722  # nosec
+                    pass
             else:
                 if isinstance(self.arg, MutableMapping):
                     inferred_dtype = pd.Series(self.arg).dtype
@@ -86,13 +96,16 @@ class DataFrameMap(DataFrameOperator, DataFrameOperatorMixin):
                     # but for int, due to the nan which may occur,
                     # we cannot infer the dtype
                     dtype = inferred_dtype
+                else:
+                    err_prefix = "int type may not be exact"
             else:
                 dtype = inferred_dtype
 
         if dtype is None:
             if not skip_infer:
+                err_prefix = err_prefix or "cannot infer dtype"
                 raise ValueError(
-                    "cannot infer dtype, it needs to be specified manually for `map`"
+                    f"{err_prefix}, it needs to be specified manually for `map`"
                 )
         else:
             dtype = np.int64 if dtype is int else dtype

maxframe/dataframe/misc/repeat.py

@@ -0,0 +1,159 @@
+# Copyright 1999-2025 Alibaba Group Holding Ltd.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from typing import List
+
+import numpy as np
+from pandas.api.types import is_list_like
+
+from ... import opcodes
+from ...core import ENTITY_TYPE, EntityData, get_output_types
+from ...serialization.serializables import AnyField
+from ..operators import DataFrameOperator, DataFrameOperatorMixin
+from ..utils import parse_index, validate_axis
+
+
+class DataFrameRepeat(DataFrameOperator, DataFrameOperatorMixin):
+    _op_type_ = opcodes.REPEAT
+
+    repeats = AnyField("repeats", default=None)
+
+    def __init__(self, output_types=None, **kw):
+        super().__init__(_output_types=output_types, **kw)
+
+    @classmethod
+    def _set_inputs(cls, op: "DataFrameRepeat", inputs: List[EntityData]):
+        super()._set_inputs(op, inputs)
+        if isinstance(op.repeats, ENTITY_TYPE):
+            op.repeats = inputs[1]
+
+    def __call__(self, obj, repeats):
+        self._output_types = get_output_types(obj)
+        test_index = obj.index_value.to_pandas()[:0]
+
+        params = obj.params
+        params["index_value"] = parse_index(test_index, obj, type(self), self.repeats)
+        params["shape"] = (np.nan,)
+
+        inputs = [obj]
+        if isinstance(repeats, ENTITY_TYPE):
+            inputs.append(repeats)
+        return self.new_tileable(inputs, **params)
+
+
+def _repeat(obj, repeats, axis=None):
+    from ...tensor.datasource import tensor
+
+    axis = validate_axis(axis or 0, obj)
+    if is_list_like(repeats):
+        repeats = tensor(repeats)
+    op = DataFrameRepeat(repeats=repeats, axis=axis)
+    return op(obj, repeats)
+
+
+def series_repeat(obj, repeats, axis=None):
+    """
+    Repeat elements of a Series.
+
+    Returns a new Series where each element of the current Series
+    is repeated consecutively a given number of times.
+
+    Parameters
+    ----------
+    repeats : int or array of ints
+        The number of repetitions for each element. This should be a
+        non-negative integer. Repeating 0 times will return an empty
+        Series.
+    axis : None
+        Must be ``None``. Has no effect but is accepted for compatibility
+        with numpy.
+
+    Returns
+    -------
+    Series
+        Newly created Series with repeated elements.
+
+    See Also
+    --------
+    Index.repeat : Equivalent function for Index.
+    numpy.repeat : Similar method for :class:`numpy.ndarray`.
+
+    Examples
+    --------
+    >>> import maxframe.dataframe as md
+    >>> s = md.Series(['a', 'b', 'c'])
+    >>> s.execute()
+    0    a
+    1    b
+    2    c
+    dtype: object
+    >>> s.repeat(2).execute()
+    0    a
+    0    a
+    1    b
+    1    b
+    2    c
+    2    c
+    dtype: object
+    >>> s.repeat([1, 2, 3]).execute()
+    0    a
+    1    b
+    1    b
+    2    c
+    2    c
+    2    c
+    dtype: object
+    """
+    return _repeat(obj, repeats, axis=axis)
+
+
+def index_repeat(obj, repeats, axis=None):
+    """
+    Repeat elements of an Index.
+
+    Returns a new Index where each element of the current Index
+    is repeated consecutively a given number of times.
+
+    Parameters
+    ----------
+    repeats : int or array of ints
+        The number of repetitions for each element. This should be a
+        non-negative integer. Repeating 0 times will return an empty
+        Index.
+    axis : None
+        Must be ``None``. Has no effect but is accepted for compatibility
+        with numpy.
+
+    Returns
+    -------
+    repeated_index : Index
+        Newly created Index with repeated elements.
+
+    See Also
+    --------
+    Series.repeat : Equivalent function for Series.
+    numpy.repeat : Similar method for :class:`numpy.ndarray`.
+
+    Examples
+    --------
+    >>> import maxframe.dataframe as md
+    >>> idx = md.Index(['a', 'b', 'c'])
+    >>> idx.execute()
+    Index(['a', 'b', 'c'], dtype='object')
+    >>> idx.repeat(2).execute()
+    Index(['a', 'a', 'b', 'b', 'c', 'c'], dtype='object')
+    >>> idx.repeat([1, 2, 3]).execute()
+    Index(['a', 'b', 'b', 'c', 'c', 'c'], dtype='object')
+    """
+    return _repeat(obj, repeats, axis=axis)

maxframe/dataframe/misc/tests/test_misc.py

@@ -22,7 +22,7 @@ from .... import opcodes
 from ....core import OutputType
 from ....dataframe import DataFrame
 from ....tensor.core import TENSOR_TYPE
-from ....udf import with_running_options
+from ....udf import ODPSFunction, with_running_options
 from ... import eval as maxframe_eval
 from ... import get_dummies, to_numeric
 from ...arithmetic import DataFrameGreater, DataFrameLess
@@ -613,3 +613,37 @@ def test_pivot_table():
     t = df.pivot_table(index=["A", "B"], columns="C", aggfunc="sum")
     assert isinstance(t.op, DataFramePivotTable)
     assert t.shape == (np.nan, np.nan)
+
+
+def test_map_with_functions():
+    raw = pd.Series([1, 2, 3], name="s_name")
+    series = from_pandas_series(raw, chunk_size=2)
+
+    # inferred type may not be exact
+    def fn1(val):
+        return val
+
+    with pytest.raises(ValueError, match="int type"):
+        series.map(fn1)
+    mapped = series.map(fn1, dtype="float64", skip_infer=True)
+    assert mapped.dtype == np.dtype("float64")
+
+    # test when type infer is valid
+    def fn2(val):
+        return val * 1.0
+
+    mapped = series.map(fn2)
+    assert mapped.dtype == np.dtype("float64")
+
+    # test function with type annotations
+    def fn3(val) -> int:
+        return val
+
+    mapped = series.map(fn3)
+    assert mapped.dtype == np.dtype("int64")
+
+    # test odps function
+    odps_func = ODPSFunction("test_odps_udf", dtype=np.float64)
+    mapped = series.map(odps_func)
+    assert isinstance(mapped.op.arg, ODPSFunction)
+    assert mapped.dtype == np.dtype("float64")

maxframe/dataframe/missing/checkna.py

@@ -22,6 +22,7 @@ from ... import tensor as mt
 from ...core import ENTITY_TYPE, OutputType
 from ...serialization.serializables import BoolField
 from ...tensor.core import TENSOR_TYPE
+from ...utils import get_pd_option
 from ..core import DATAFRAME_TYPE, INDEX_TYPE, SERIES_TYPE, MultiIndex
 from ..operators import DataFrameOperator, DataFrameOperatorMixin
 
@@ -138,7 +139,7 @@ def isna(obj):
     2     True
     dtype: bool
     """
-    use_inf_as_na = pd.get_option("mode.use_inf_as_na")
+    use_inf_as_na = get_pd_option("mode.use_inf_as_na", False)
     if isinstance(obj, MultiIndex):
         raise NotImplementedError("isna is not defined for MultiIndex")
     elif isinstance(obj, ENTITY_TYPE):
@@ -213,7 +214,7 @@ def notna(obj):
     2    False
     dtype: bool
     """
-    use_inf_as_na = pd.get_option("mode.use_inf_as_na")
+    use_inf_as_na = get_pd_option("mode.use_inf_as_na", False)
     if isinstance(obj, MultiIndex):
         raise NotImplementedError("isna is not defined for MultiIndex")
     elif isinstance(obj, ENTITY_TYPE):

maxframe/dataframe/reduction/__init__.py

@@ -17,7 +17,7 @@ from .all import DataFrameAll
 from .any import DataFrameAny
 from .argmax import DataFrameArgMax
 from .argmin import DataFrameArgMin
-from .core import CustomReduction
+from .core import CustomReduction, NamedAgg
 from .count import DataFrameCount
 from .cummax import DataFrameCummax
 from .cummin import DataFrameCummin
@@ -31,6 +31,7 @@ from .max import DataFrameMax
 from .mean import DataFrameMean
 from .median import DataFrameMedian
 from .min import DataFrameMin
+from .mode import DataFrameMode
 from .nunique import DataFrameNunique
 from .prod import DataFrameProd
 from .reduction_size import DataFrameSize
@@ -47,8 +48,8 @@ def _install():
     from .aggregation import aggregate
     from .all import all_dataframe, all_index, all_series
     from .any import any_dataframe, any_index, any_series
-    from .argmax import argmax_series
-    from .argmin import argmin_series
+    from .argmax import argmax_series_index
+    from .argmin import argmin_series_index
     from .count import count_dataframe, count_series
     from .cov import cov_dataframe, cov_series
     from .cummax import cummax
@@ -62,6 +63,7 @@ def _install():
     from .mean import mean_dataframe, mean_series
     from .median import median_dataframe, median_series
     from .min import min_dataframe, min_index, min_series
+    from .mode import mode_dataframe, mode_series
     from .nunique import nunique_dataframe, nunique_series
     from .prod import prod_dataframe, prod_series
     from .reduction_size import size_dataframe, size_series
@@ -76,8 +78,8 @@ def _install():
         ("aggregate", aggregate, aggregate),
         ("all", all_series, all_dataframe),
         ("any", any_series, any_dataframe),
-        ("argmax", argmax_series, None),
-        ("argmin", argmin_series, None),
+        ("argmax", argmax_series_index, None),
+        ("argmin", argmin_series_index, None),
         ("count", count_series, count_dataframe),
         ("cov", cov_series, cov_dataframe),
         ("cummax", cummax, cummax),
@@ -92,6 +94,7 @@ def _install():
         ("mean", mean_series, mean_dataframe),
         ("median", median_series, median_dataframe),
         ("min", min_series, min_dataframe),
+        ("mode", mode_series, mode_dataframe),
         ("nunique", nunique_series, nunique_dataframe),
         ("prod", prod_series, prod_dataframe),
         ("product", prod_series, prod_dataframe),
@@ -118,6 +121,8 @@ def _install():
         setattr(t, "any", any_index)
         setattr(t, "min", min_index)
         setattr(t, "max", max_index)
+        setattr(t, "argmin", argmin_series_index)
+        setattr(t, "argmax", argmax_series_index)
 
 
 _install()

maxframe/dataframe/reduction/aggregation.py

@@ -38,7 +38,7 @@ from ...serialization.serializables import (
 )
 from ...typing_ import TileableType
 from ...udf import BuiltinFunction
-from ...utils import lazy_import, pd_release_version
+from ...utils import get_pd_option, lazy_import, pd_release_version
 from ..operators import DataFrameOperator, DataFrameOperatorMixin
 from ..utils import build_df, build_empty_df, build_series, parse_index, validate_axis
 from .core import (
@@ -92,8 +92,8 @@ class DataFrameAggregate(DataFrameOperator, DataFrameOperatorMixin):
     _op_type_ = opcodes.AGGREGATE
 
     raw_func = AnyField("raw_func")
-    raw_func_kw = DictField("raw_func_kw")
-    func = AnyField("func")
+    raw_func_kw = DictField("raw_func_kw", default=None)
+    func = AnyField("func", default=None)
     func_rename = ListField("func_rename", default=None)
     axis = AnyField("axis", default=0)
     numeric_only = BoolField("numeric_only", default=None)
@@ -199,7 +199,7 @@ class DataFrameAggregate(DataFrameOperator, DataFrameOperatorMixin):
         normalize_reduction_funcs(self, ndim=df.ndim)
         compile_reduction_funcs(self, df)
         if output_type is None or dtypes is None:
-            with enter_mode(kernel=False, build=False):
+            with enter_mode(kernel=False, build=False, mock=True):
                 dtypes, index = self._calc_result_shape(df)
         else:
             self.output_types = [output_type]
@@ -231,7 +231,7 @@ class DataFrameAggregate(DataFrameOperator, DataFrameOperatorMixin):
             return self.new_series(
                 [df],
                 shape=new_shape,
-                dtype=dtypes[0],
+                dtype=dtypes.iloc[0],
                 name=dtypes.index[0],
                 index_value=new_index,
             )
@@ -456,7 +456,7 @@ def aggregate(df, func=None, axis=0, **kw):
     min    1
     """
     axis = validate_axis(axis, df)
-    use_inf_as_na = kw.pop("_use_inf_as_na", pd.get_option("mode.use_inf_as_na"))
+    use_inf_as_na = kw.pop("_use_inf_as_na", get_pd_option("mode.use_inf_as_na", False))
     if func == "unique":
         # workaround for direct call of unique function which
         #  returns a tensor directly