Merge branch 'master' of github.com:pandas-dev/pandas into any_return_type

Author: yadav-sachin

.github/ISSUE_TEMPLATE/bug_report.yaml

@@ -7,6 +7,7 @@ body:
   - type: checkboxes
     id: checks
     attributes:
+      label: Pandas version checks
       options:
         - label: >
             I have checked that this issue has not already been reported.

.github/ISSUE_TEMPLATE/documentation_improvement.yaml

@@ -6,6 +6,7 @@ labels: [Docs, Needs Triage]
 body:
   - type: checkboxes
     attributes:
+      label: Pandas version checks
       options:
         - label: >
             I have checked that the issue still exists on the latest versions of the docs

.github/ISSUE_TEMPLATE/installation_issue.yaml

@@ -7,6 +7,7 @@ body:
   - type: checkboxes
     id: checks
     attributes:
+      label: Installation check
       options:
         - label: >
             I have read the [installation guide](https://pandas.pydata.org/pandas-docs/stable/getting_started/install.html#installing-pandas).

.github/ISSUE_TEMPLATE/performance_issue.yaml

@@ -7,6 +7,7 @@ body:
   - type: checkboxes
     id: checks
     attributes:
+      label: Pandas version checks
       options:
         - label: >
             I have checked that this issue has not already been reported.

.github/ISSUE_TEMPLATE/submit_question.yml

@@ -11,6 +11,7 @@ body:
         usage questions, we ask that all usage questions are first asked on StackOverflow.
   - type: checkboxes
     attributes:
+      label: Research
       options:
         - label: >
             I have searched the [[pandas] tag](https://stackoverflow.com/questions/tagged/pandas)

ci/run_tests.sh

@@ -5,6 +5,9 @@
 # https://github.com/pytest-dev/pytest/issues/1075
 export PYTHONHASHSEED=$(python -c 'import random; print(random.randint(1, 4294967295))')
 
+# May help reproduce flaky CI builds if set in subsequent runs
+echo PYTHONHASHSEED=$PYTHONHASHSEED
+
 if [[ "not network" == *"$PATTERN"* ]]; then
     export http_proxy=http://1.2.3.4 https_proxy=http://1.2.3.4;
 fi

doc/source/whatsnew/v1.4.0.rst

@@ -660,7 +660,7 @@ Conversion
 
 Strings
 ^^^^^^^
-- Fixed bug in checking for ``string[pyarrow]`` dtype incorrectly raising an ImportError when pyarrow is not installed (:issue:`44327`)
+- Fixed bug in checking for ``string[pyarrow]`` dtype incorrectly raising an ImportError when pyarrow is not installed (:issue:`44276`)
 -
 
 Interval

pandas/_testing/__init__.py

@@ -28,17 +28,12 @@
 from pandas._typing import Dtype
 
 from pandas.core.dtypes.common import (
-    is_datetime64_dtype,
-    is_datetime64tz_dtype,
     is_float_dtype,
     is_integer_dtype,
-    is_period_dtype,
     is_sequence,
-    is_timedelta64_dtype,
     is_unsigned_integer_dtype,
     pandas_dtype,
 )
-from pandas.core.dtypes.dtypes import IntervalDtype
 
 import pandas as pd
 from pandas import (
@@ -112,14 +107,11 @@
 )
 from pandas.core.arrays import (
     BaseMaskedArray,
-    DatetimeArray,
     ExtensionArray,
     PandasArray,
-    PeriodArray,
-    TimedeltaArray,
-    period_array,
 )
 from pandas.core.arrays._mixins import NDArrayBackedExtensionArray
+from pandas.core.construction import extract_array
 
 if TYPE_CHECKING:
     from pandas import (
@@ -161,6 +153,17 @@
     + BYTES_DTYPES
 )
 
+NARROW_NP_DTYPES = [
+    np.float16,
+    np.float32,
+    np.int8,
+    np.int16,
+    np.int32,
+    np.uint8,
+    np.uint16,
+    np.uint32,
+]
+
 NULL_OBJECTS = [None, np.nan, pd.NaT, float("nan"), pd.NA, Decimal("NaN")]
 NP_NAT_OBJECTS = [
     cls("NaT", unit)
@@ -257,13 +260,6 @@ def box_expected(expected, box_cls, transpose=True):
             #  single-row special cases in datetime arithmetic
             expected = expected.T
             expected = pd.concat([expected] * 2, ignore_index=True)
-    elif box_cls is PeriodArray:
-        # the PeriodArray constructor is not as flexible as period_array
-        expected = period_array(expected)
-    elif box_cls is DatetimeArray:
-        expected = DatetimeArray(expected)
-    elif box_cls is TimedeltaArray:
-        expected = TimedeltaArray(expected)
     elif box_cls is np.ndarray or box_cls is np.array:
         expected = np.array(expected)
     elif box_cls is to_array:
@@ -274,21 +270,16 @@ def box_expected(expected, box_cls, transpose=True):
 
 
 def to_array(obj):
+    """
+    Similar to pd.array, but does not cast numpy dtypes to nullable dtypes.
+    """
     # temporary implementation until we get pd.array in place
     dtype = getattr(obj, "dtype", None)
 
-    if is_period_dtype(dtype):
-        return period_array(obj)
-    elif is_datetime64_dtype(dtype) or is_datetime64tz_dtype(dtype):
-        return DatetimeArray._from_sequence(obj)
-    elif is_timedelta64_dtype(dtype):
-        return TimedeltaArray._from_sequence(obj)
-    elif isinstance(obj, pd.core.arrays.BooleanArray):
-        return obj
-    elif isinstance(dtype, IntervalDtype):
-        return pd.core.arrays.IntervalArray(obj)
-    else:
-        return np.array(obj)
+    if dtype is None:
+        return np.asarray(obj)
+
+    return extract_array(obj, extract_numpy=True)
 
 
 # -----------------------------------------------------------------------------

pandas/conftest.py

@@ -599,11 +599,6 @@ def index_with_missing(request):
 # ----------------------------------------------------------------
 # Series'
 # ----------------------------------------------------------------
-@pytest.fixture
-def empty_series():
-    return Series([], index=[], dtype=np.float64)
-
-
 @pytest.fixture
 def string_series():
     """
@@ -672,31 +667,12 @@ def series_with_multilevel_index():
     return ser
 
 
-_narrow_dtypes = [
-    np.float16,
-    np.float32,
-    np.int8,
-    np.int16,
-    np.int32,
-    np.uint8,
-    np.uint16,
-    np.uint32,
-]
 _narrow_series = {
     f"{dtype.__name__}-series": tm.makeFloatSeries(name="a").astype(dtype)
-    for dtype in _narrow_dtypes
+    for dtype in tm.NARROW_NP_DTYPES
 }
 
 
-@pytest.fixture(params=_narrow_series.keys())
-def narrow_series(request):
-    """
-    Fixture for Series with low precision data types
-    """
-    # copy to avoid mutation, e.g. setting .name
-    return _narrow_series[request.param].copy()
-
-
 _index_or_series_objs = {**indices_dict, **_series, **_narrow_series}
 
 
@@ -712,11 +688,6 @@ def index_or_series_obj(request):
 # ----------------------------------------------------------------
 # DataFrames
 # ----------------------------------------------------------------
-@pytest.fixture
-def empty_frame():
-    return DataFrame()
-
-
 @pytest.fixture
 def int_frame():
     """

pandas/tests/arithmetic/test_numeric.py

@@ -1400,11 +1400,15 @@ def test_integer_array_add_list_like(
     if Series == box_pandas_1d_array:
         expected = Series(expected_data, dtype="Int64")
     elif Series == box_1d_array:
-        expected = Series(expected_data, dtype="object")
+        if box_pandas_1d_array is tm.to_array:
+            expected = Series(expected_data, dtype="Int64")
+        else:
+            expected = Series(expected_data, dtype="object")
     elif Index in (box_pandas_1d_array, box_1d_array):
         expected = Int64Index(expected_data)
     else:
-        expected = np.array(expected_data, dtype="object")
+        # box_pandas_1d_array is tm.to_array; preserves IntegerArray
+        expected = array(expected_data, dtype="Int64")
 
     tm.assert_equal(left, expected)
     tm.assert_equal(right, expected)

pandas/tests/base/test_misc.py

@@ -18,6 +18,7 @@
     Index,
     Series,
 )
+import pandas._testing as tm
 
 
 @pytest.mark.parametrize(
@@ -109,8 +110,9 @@ def test_memory_usage_components_series(series_with_simple_index):
     assert total_usage == non_index_usage + index_usage
 
 
-def test_memory_usage_components_narrow_series(narrow_series):
-    series = narrow_series
+@pytest.mark.parametrize("dtype", tm.NARROW_NP_DTYPES)
+def test_memory_usage_components_narrow_series(dtype):
+    series = tm.makeFloatSeries(name="a").astype(dtype)
     total_usage = series.memory_usage(index=True)
     non_index_usage = series.memory_usage(index=False)
     index_usage = series.index.memory_usage()

pandas/tests/frame/methods/test_replace.py

@@ -1384,8 +1384,9 @@ def test_replace_value_category_type(self):
 
     @pytest.mark.xfail(
         reason="category dtype gets changed to object type after replace, see #35268",
+        raises=AssertionError,
     )
-    def test_replace_dict_category_type(self, input_category_df, expected_category_df):
+    def test_replace_dict_category_type(self):
         """
         Test to ensure category dtypes are maintained
         after replace with dict values

pandas/tests/io/json/test_pandas.py

@@ -244,7 +244,8 @@ def test_roundtrip_categorical(self, request, orient, convert_axes, numpy):
 
     @pytest.mark.parametrize("convert_axes", [True, False])
     @pytest.mark.parametrize("numpy", [True, False])
-    def test_roundtrip_empty(self, orient, convert_axes, numpy, empty_frame):
+    def test_roundtrip_empty(self, orient, convert_axes, numpy):
+        empty_frame = DataFrame()
         data = empty_frame.to_json(orient=orient)
         result = read_json(data, orient=orient, convert_axes=convert_axes, numpy=numpy)
         expected = empty_frame.copy()
@@ -673,7 +674,8 @@ def test_series_roundtrip_object(self, orient, numpy, dtype, object_series):
         tm.assert_series_equal(result, expected)
 
     @pytest.mark.parametrize("numpy", [True, False])
-    def test_series_roundtrip_empty(self, orient, numpy, empty_series):
+    def test_series_roundtrip_empty(self, orient, numpy):
+        empty_series = Series([], index=[], dtype=np.float64)
         data = empty_series.to_json(orient=orient)
         result = read_json(data, typ="series", orient=orient, numpy=numpy)