GH1383 Add support for creating Series/Index with dtype='category'

Author: loicdiridollou

pandas-stubs/core/indexes/base.pyi

@@ -42,6 +42,7 @@ from pandas.core.base import (
     NumListLike,
     _ListLike,
 )
+from pandas.core.indexes.category import CategoricalIndex
 from pandas.core.strings.accessor import StringMethods
 from typing_extensions import (
     Never,
@@ -58,6 +59,7 @@ from pandas._typing import (
     AnyAll,
     ArrayLike,
     AxesData,
+    CategoryDtypeArg,
     DropKeep,
     Dtype,
     DtypeArg,
@@ -229,6 +231,16 @@ class Index(IndexOpsMixin[S1]):
         tupleize_cols: bool = ...,
     ) -> TimedeltaIndex: ...
     @overload
+    def __new__(
+        cls,
+        data: AxesData,
+        *,
+        dtype: CategoryDtypeArg,
+        copy: bool = ...,
+        name: Hashable = ...,
+        tupleize_cols: bool = ...,
+    ) -> CategoricalIndex: ...
+    @overload
     def __new__(
         cls,
         data: Sequence[Interval[_OrderableT]] | IndexOpsMixin[Interval[_OrderableT]],

pandas-stubs/core/series.pyi

@@ -353,6 +353,16 @@ class Series(IndexOpsMixin[S1], NDFrame):
         copy: bool = ...,
     ) -> Series[Timestamp]: ...
     @overload
+    def __new__(
+        cls,
+        data: _ListLike,
+        index: AxesData | None = ...,
+        *,
+        dtype: CategoryDtypeArg,
+        name: Hashable = ...,
+        copy: bool = ...,
+    ) -> Series[CategoricalDtype]: ...
+    @overload
     def __new__(
         cls,
         data: PeriodIndex | Sequence[Period],

pyproject.toml

@@ -65,7 +65,7 @@ scipy = { version = ">=1.9.1", python = "<3.14" }
 scipy-stubs = ">=1.15.3.0"
 SQLAlchemy = ">=2.0.39"
 types-python-dateutil = ">=2.8.19"
-beautifulsoup4 = "<=4.13.5"
+beautifulsoup4 = ">=4.13.5"
 html5lib = ">=1.1"
 python-calamine = ">=0.2.0"
 

tests/indexes/test_indexes.py

@@ -15,6 +15,7 @@
 from pandas.core.arrays import DatetimeArray
 from pandas.core.arrays.categorical import Categorical
 from pandas.core.indexes.base import Index
+from pandas.core.indexes.category import CategoricalIndex
 from typing_extensions import (
     Never,
     assert_type,
@@ -1366,6 +1367,12 @@ def test_index_factorize() -> None:
     check(assert_type(idx_uniques, np_1darray | Index | Categorical), pd.Index)
 
 
+def test_index_categorical() -> None:
+    """Test creating an index with Categorical type GH1383."""
+    sr = pd.Index([1], dtype="category")
+    check(assert_type(sr, CategoricalIndex), CategoricalIndex)
+
+
 def test_disallow_empty_index() -> None:
     # From GH 826
     if TYPE_CHECKING_INVALID_USAGE:

tests/series/test_series.py

@@ -50,6 +50,8 @@
     Scalar,
 )
 
+from pandas.core.dtypes.dtypes import CategoricalDtype
+
 from tests import (
     PD_LTE_23,
     TYPE_CHECKING_INVALID_USAGE,
@@ -1824,6 +1826,10 @@ def test_categorical_codes():
     cat = pd.Categorical(["a", "b", "a"])
     check(assert_type(cat.codes, np_1darray[np.signedinteger]), np_1darray[np.int8])
 
+    # GH1383
+    sr = pd.Series([1], dtype="category")
+    check(assert_type(sr, "pd.Series[CategoricalDtype]"), pd.Series, np.integer)
+
 
 def test_relops() -> None:
     # GH 175