fixup tests!

Author: MarcoGorelli

pandas/core/interchange/buffer.py

@@ -108,6 +108,7 @@ def __init__(
             self._buffer = arr.buffers()[1]
         self._length = len(arr)
         self._dlpack = arr.__dlpack__
+        self._is_validity = is_validity
 
     @property
     def bufsize(self) -> int:

pandas/core/interchange/column.py

@@ -198,6 +198,10 @@ def describe_null(self):
             null_value = 1
             return column_null_dtype, null_value
         if isinstance(self._col.dtype, ArrowDtype):
+            if all(
+                chunk.buffers()[0] is None for chunk in self._col.array._pa_array.chunks
+            ):
+                return ColumnNullType.NON_NULLABLE, None
             return ColumnNullType.USE_BITMASK, 0
         kind = self.dtype[0]
         try:
@@ -364,7 +368,7 @@ def _get_validity_buffer(self) -> tuple[PandasBuffer, Any]:
             arr = self._col.array
             dtype = (DtypeKind.BOOL, 1, ArrowCTypes.BOOL, Endianness.NATIVE)
             if all(chunk.buffers()[0] is None for chunk in arr._pa_array.chunks):
-                return None, dtype
+                return None
             buffer = PandasBufferPyarrow(
                 arr._pa_array, is_validity=True, allow_copy=self._allow_copy
             )

pandas/core/interchange/from_dataframe.py

@@ -298,9 +298,9 @@ def string_column_to_ndarray(col: Column) -> tuple[np.ndarray, Any]:
 
     null_pos = None
     if null_kind in (ColumnNullType.USE_BITMASK, ColumnNullType.USE_BYTEMASK):
-        assert buffers["validity"], "Validity buffers cannot be empty for masks"
-        valid_buff, valid_dtype = buffers["validity"]
-        if valid_buff is not None:
+        validity = buffers["validity"]
+        if validity is not None:
+            valid_buff, valid_dtype = validity
             null_pos = buffer_to_ndarray(
                 valid_buff, valid_dtype, offset=col.offset, length=col.size()
             )
@@ -520,13 +520,14 @@ def set_nulls(
     np.ndarray or pd.Series
         Data with the nulls being set.
     """
+    if validity is None:
+        return data
     null_kind, sentinel_val = col.describe_null
     null_pos = None
 
     if null_kind == ColumnNullType.USE_SENTINEL:
         null_pos = pd.Series(data) == sentinel_val
     elif null_kind in (ColumnNullType.USE_BITMASK, ColumnNullType.USE_BYTEMASK):
-        assert validity, "Expected to have a validity buffer for the mask"
         valid_buff, valid_dtype = validity
         if valid_buff is not None:
             null_pos = buffer_to_ndarray(

pandas/tests/interchange/test_impl.py

@@ -463,7 +463,7 @@ def test_non_str_names_w_duplicates():
         ),
     ],
 )
-def test_pandas_nullable_w_missing_values(
+def test_pandas_nullable_with_missing_values(
     data: list, dtype: str, expected_dtype: str
 ) -> None:
     # https://github.com/pandas-dev/pandas/issues/57643
@@ -481,6 +481,68 @@ def test_pandas_nullable_w_missing_values(
     assert result[2].as_py() is None
 
 
+@pytest.mark.parametrize(
+    ("data", "dtype", "expected_dtype"),
+    [
+        ([1, 2, 3], "Int64", "int64"),
+        ([1, 2, 3], "Int64[pyarrow]", "int64"),
+        ([1, 2, 3], "Int8", "int8"),
+        ([1, 2, 3], "Int8[pyarrow]", "int8"),
+        (
+            [1, 2, 3],
+            "UInt64",
+            "uint64",
+        ),
+        (
+            [1, 2, 3],
+            "UInt64[pyarrow]",
+            "uint64",
+        ),
+        ([1.0, 2.25, 5.0], "Float32", "float32"),
+        ([1.0, 2.25, 5.0], "Float32[pyarrow]", "float32"),
+        ([True, False, False], "boolean", "bool"),
+        ([True, False, False], "boolean[pyarrow]", "bool"),
+        (["much ado", "about", "nothing"], "string[pyarrow_numpy]", "large_string"),
+        (["much ado", "about", "nothing"], "string[pyarrow]", "large_string"),
+        (
+            [datetime(2020, 1, 1), datetime(2020, 1, 2), datetime(2020, 1, 3)],
+            "timestamp[ns][pyarrow]",
+            "timestamp[ns]",
+        ),
+        (
+            [datetime(2020, 1, 1), datetime(2020, 1, 2), datetime(2020, 1, 3)],
+            "timestamp[us][pyarrow]",
+            "timestamp[us]",
+        ),
+        (
+            [
+                datetime(2020, 1, 1, tzinfo=timezone.utc),
+                datetime(2020, 1, 2, tzinfo=timezone.utc),
+                datetime(2020, 1, 3, tzinfo=timezone.utc),
+            ],
+            "timestamp[us, Asia/Kathmandu][pyarrow]",
+            "timestamp[us, tz=Asia/Kathmandu]",
+        ),
+    ],
+)
+def test_pandas_nullable_without_missing_values(
+    data: list, dtype: str, expected_dtype: str
+) -> None:
+    # https://github.com/pandas-dev/pandas/issues/57643
+    pa = pytest.importorskip("pyarrow", "11.0.0")
+    import pyarrow.interchange as pai
+
+    if expected_dtype == "timestamp[us, tz=Asia/Kathmandu]":
+        expected_dtype = pa.timestamp("us", "Asia/Kathmandu")
+
+    df = pd.DataFrame({"a": data}, dtype=dtype)
+    result = pai.from_dataframe(df.__dataframe__())["a"]
+    assert result.type == expected_dtype
+    assert result[0].as_py() == data[0]
+    assert result[1].as_py() == data[1]
+    assert result[2].as_py() == data[2]
+
+
 def test_empty_dataframe():
     # https://github.com/pandas-dev/pandas/issues/56700
     df = pd.DataFrame({"a": []}, dtype="int8")