TYP: Series/DataFrame/Index are not Hashable (#113)

* TYP: Series/DataFrame are not Hashable

* explicitly set __hash__ also in the sub-classes (they inherit from classes that have proper __hash__)

* cover a few more __hash__; and try to check types at runtime

* just in case it would affect static type checking

* cache only poetry.lock

* remove runtime check

* assert assert_type

* undo change for the runtype hack

* poetry.lock should not be dependent on the OS/python version

* Empty commit

* move ClassVar to the top

* negative type tests

* move to top

Author: twoertwein

.github/workflows/test.yml

@@ -26,26 +26,12 @@ jobs:
     - name: Install Poetry
       run: pip install poetry
 
-    - name: Determine poetry cache-dir
-      run: |
-        echo "::set-output name=PATH::$(poetry config cache-dir)"
-      id: cache_path
-
-    - name: Cache project dependencies
+    - name: Cache poetry.lock
       id : cache
       uses: actions/cache@v3
-      with:
-        path: |
-          ${{ steps.cache_path.outputs.PATH }}
-          poetry.lock
-        key: ${{ matrix.os }}-${{ matrix.python-version }}-poetry-${{ hashFiles('pyproject.toml') }}
-        restore-keys: ${{ matrix.os }}-${{ matrix.python-version }}-poetry-
-
-    - name: Delete poetry.lock on cache miss
-      if: steps.cache.outputs.cache-hit != 'true'
-      uses: JesseTG/[email protected]
       with:
         path: poetry.lock
+        key: poetry-${{ hashFiles('pyproject.toml') }}
 
     - name: Install project dependencies
       run: poetry install -vvv --no-root

pandas-stubs/_typing.pyi

@@ -200,5 +200,5 @@ XMLParsers = Literal["lxml", "etree"]
 # Any plain Python or numpy function
 Function = Union[np.ufunc, Callable[..., Any]]
 GroupByObject = Union[
-    Label, List[Label], Function, Series, np.ndarray, Mapping[Label, Any]
+    Label, List[Label], Function, Series, np.ndarray, Mapping[Label, Any], Index
 ]

pandas-stubs/core/frame.pyi

@@ -5,6 +5,7 @@ import datetime as _dt
 from typing import (
     Any,
     Callable,
+    ClassVar,
     Dict,
     Hashable,
     Iterable,
@@ -183,6 +184,8 @@ class DataFrame(NDFrame, OpsMixin):
         Index,
         Series,
     ]
+    __hash__: ClassVar[None]  # type: ignore[assignment]
+
     def __new__(
         cls,
         data: Optional[Union[_ListLike, DataFrame, Dict[Any, Any]]] = ...,
@@ -428,7 +431,7 @@ class DataFrame(NDFrame, OpsMixin):
         *,
         axis: Axis = ...,
         index: Hashable | Sequence[Hashable] = ...,
-        columns: Hashable | Sequence[Hashable] = ...,
+        columns: Hashable | Sequence[Hashable] | Index = ...,
         level: Optional[Level] = ...,
         inplace: Literal[True],
         errors: IgnoreRaise = ...,
@@ -440,7 +443,7 @@ class DataFrame(NDFrame, OpsMixin):
         *,
         axis: Axis = ...,
         index: Hashable | Sequence[Hashable] = ...,
-        columns: Hashable | Sequence[Hashable] = ...,
+        columns: Hashable | Sequence[Hashable] | Index = ...,
         level: Optional[Level] = ...,
         inplace: Literal[False] = ...,
         errors: IgnoreRaise = ...,
@@ -452,7 +455,7 @@ class DataFrame(NDFrame, OpsMixin):
         *,
         axis: Axis = ...,
         index: Hashable | Sequence[Hashable] = ...,
-        columns: Hashable | Sequence[Hashable] = ...,
+        columns: Hashable | Sequence[Hashable] | Index = ...,
         level: Optional[Level] = ...,
         inplace: bool = ...,
         errors: IgnoreRaise = ...,

pandas-stubs/core/generic.pyi

@@ -2,6 +2,7 @@ import sys
 from typing import (
     Any,
     Callable,
+    ClassVar,
     Dict,
     Hashable,
     Iterator,
@@ -48,6 +49,8 @@ _bool = bool
 _str = str
 
 class NDFrame(PandasObject, indexing.IndexingMixin):
+    __hash__: ClassVar[None]  # type: ignore[assignment]
+
     def __new__(
         cls,
         data: BlockManager,
@@ -89,7 +92,6 @@ class NDFrame(PandasObject, indexing.IndexingMixin):
     def bool(self) -> _bool: ...
     def __abs__(self) -> NDFrame: ...
     def __round__(self, decimals: int = ...) -> NDFrame: ...
-    def __hash__(self): ...
     def __iter__(self) -> Iterator: ...
     def keys(self): ...
     def iteritems(self): ...

pandas-stubs/core/indexes/base.pyi

@@ -1,5 +1,6 @@
 from typing import (
     Callable,
+    ClassVar,
     Dict,
     Hashable,
     Iterable,
@@ -41,6 +42,8 @@ class InvalidIndexError(Exception): ...
 _str = str
 
 class Index(IndexOpsMixin, PandasObject):
+    __hash__: ClassVar[None]  # type: ignore[assignment]
+
     def __new__(
         cls,
         data: Iterable = ...,
@@ -169,7 +172,6 @@ class Index(IndexOpsMixin, PandasObject):
     def where(self, cond, other=...): ...
     def is_type_compatible(self, kind) -> bool: ...
     def __contains__(self, key) -> bool: ...
-    def __hash__(self) -> int: ...
     def __setitem__(self, key, value) -> None: ...
     @overload
     def __getitem__(

pandas-stubs/core/indexes/frozen.pyi

@@ -8,4 +8,4 @@ class FrozenList(PandasObject, list):
     def __eq__(self, other) -> bool: ...
     def __mul__(self, other): ...
     def __reduce__(self): ...
-    def __hash__(self): ...
+    def __hash__(self) -> int: ...  # type: ignore[override]

pandas-stubs/core/series.pyi

@@ -5,6 +5,7 @@ from datetime import (
 from typing import (
     Any,
     Callable,
+    ClassVar,
     Dict,
     Generic,
     Hashable,
@@ -138,6 +139,8 @@ class _LocIndexerSeries(_LocIndexer, Generic[S1]):
 class Series(IndexOpsMixin, NDFrame, Generic[S1]):
 
     _ListLike = Union[ArrayLike, Dict[_str, np.ndarray], List, Tuple, Index]
+    __hash__: ClassVar[None]
+
     @overload
     def __new__(
         cls,

tests/test_frame.py

@@ -7,6 +7,7 @@
     TYPE_CHECKING,
     Any,
     Dict,
+    Hashable,
     Iterable,
     List,
     Tuple,
@@ -1141,3 +1142,20 @@ def test_frame_ndarray_assignmment() -> None:
 
     df_b = pd.DataFrame({"a": [0.0] * 10, "b": [1.0] * 10})
     df_b.iloc[:, :] = np.array([[-1.0, np.inf]] * 10)
+
+
+def test_not_hashable() -> None:
+    # GH 113
+    assert assert_type(pd.DataFrame.__hash__, None) is None
+    assert assert_type(pd.DataFrame().__hash__, None) is None
+    assert assert_type(pd.Series.__hash__, None) is None
+    assert assert_type(pd.Series([], dtype=object).__hash__, None) is None
+    assert assert_type(pd.Index.__hash__, None) is None
+    assert assert_type(pd.Index([]).__hash__, None) is None
+
+    def test_func(h: Hashable):
+        pass
+
+    test_func(pd.DataFrame())  # type: ignore[arg-type]
+    test_func(pd.Series([], dtype=object))  # type: ignore[arg-type]
+    test_func(pd.Index([]))  # type: ignore[arg-type]