API: Disallow sets as index and columns argument in DataFrame constructor

doc/source/whatsnew/v1.5.0.rst

@@ -152,6 +152,7 @@ Other enhancements
 - Added ``numeric_only`` argument to :meth:`Resampler.sum`, :meth:`Resampler.prod`, :meth:`Resampler.min`, :meth:`Resampler.max`, :meth:`Resampler.first`, and :meth:`Resampler.last` (:issue:`46442`)
 - ``times`` argument in :class:`.ExponentialMovingWindow` now accepts ``np.timedelta64`` (:issue:`47003`)
 - :class:`DataError`, :class:`SpecificationError`, :class:`SettingWithCopyError`, and :class:`SettingWithCopyWarning` are now exposed in ``pandas.errors`` (:issue:`27656`)
+- :class:`DataFrame` constructor raises if ``index`` or ``columns`` arguments are sets (:issue:`47215`)
 
 .. ---------------------------------------------------------------------------
 .. _whatsnew_150.notable_bug_fixes:

pandas/_typing.py

@@ -10,7 +10,6 @@
     TYPE_CHECKING,
     Any,
     Callable,
-    Collection,
     Dict,
     Hashable,
     Iterator,
@@ -115,7 +114,7 @@
 Ordered = Optional[bool]
 JSONSerializable = Optional[Union[PythonScalar, List, Dict]]
 Frequency = Union[str, "DateOffset"]
-Axes = Collection[Any]
+Axes = Union[AnyArrayLike, List, Dict, range, Sequence[str]]
 
 RandomState = Union[
     int,

pandas/core/frame.py

@@ -627,6 +627,12 @@ def __init__(
 
         manager = get_option("mode.data_manager")
 
+        # GH47215
+        if index is not None and isinstance(index, set):
+            raise ValueError("index cannot be a set")
+        if columns is not None and isinstance(columns, set):
+            raise ValueError("columns cannot be a set")
+
         if copy is None:
             if isinstance(data, dict):
                 # retain pre-GH#38939 default behavior
@@ -730,10 +736,7 @@ def __init__(
                 if not isinstance(data, np.ndarray) and treat_as_nested(data):
                     # exclude ndarray as we may have cast it a few lines above
                     if columns is not None:
-                        # error: Argument 1 to "ensure_index" has incompatible type
-                        # "Collection[Any]"; expected "Union[Union[Union[ExtensionArray,
-                        # ndarray], Index, Series], Sequence[Any]]"
-                        columns = ensure_index(columns)  # type: ignore[arg-type]
+                        columns = ensure_index(columns)
                     arrays, columns, index = nested_data_to_arrays(
                         # error: Argument 3 to "nested_data_to_arrays" has incompatible
                         # type "Optional[Collection[Any]]"; expected "Optional[Index]"
@@ -771,14 +774,8 @@ def __init__(
             if index is None or columns is None:
                 raise ValueError("DataFrame constructor not properly called!")
 
-            # Argument 1 to "ensure_index" has incompatible type "Collection[Any]";
-            # expected "Union[Union[Union[ExtensionArray, ndarray],
-            # Index, Series], Sequence[Any]]"
-            index = ensure_index(index)  # type: ignore[arg-type]
-            # Argument 1 to "ensure_index" has incompatible type "Collection[Any]";
-            # expected "Union[Union[Union[ExtensionArray, ndarray],
-            # Index, Series], Sequence[Any]]"
-            columns = ensure_index(columns)  # type: ignore[arg-type]
+            index = ensure_index(index)
+            columns = ensure_index(columns)
 
             if not dtype:
                 dtype, _ = infer_dtype_from_scalar(data, pandas_dtype=True)

pandas/core/indexes/base.py

@@ -42,8 +42,8 @@
     tz_compare,
 )
 from pandas._typing import (
-    AnyArrayLike,
     ArrayLike,
+    Axes,
     Dtype,
     DtypeObj,
     F,
@@ -7273,7 +7273,7 @@ def ensure_index_from_sequences(sequences, names=None) -> Index:
         return MultiIndex.from_arrays(sequences, names=names)
 
 
-def ensure_index(index_like: AnyArrayLike | Sequence, copy: bool = False) -> Index:
+def ensure_index(index_like: Axes, copy: bool = False) -> Index:
     """
     Ensure that we have an index from some index-like object.
 

pandas/tests/frame/test_constructors.py

@@ -2976,6 +2976,14 @@ def test_construction_from_ndarray_with_eadtype_mismatched_columns(self):
         with pytest.raises(ValueError, match=msg):
             DataFrame(arr2, columns=["foo", "bar"])
 
+    def test_columns_indexes_raise_on_sets(self):
+        # GH 47215
+        data = [[1, 2, 3], [4, 5, 6]]
+        with pytest.raises(ValueError, match="index cannot be a set"):
+            DataFrame(data, index={"a", "b"})
+        with pytest.raises(ValueError, match="columns cannot be a set"):
+            DataFrame(data, columns={"a", "b", "c"})
+
 
 def get1(obj):  # TODO: make a helper in tm?
     if isinstance(obj, Series):