remove pandas backports to fix test coverage errors

Author: chelsea-lin

.github/workflows/docs.yml

@@ -12,7 +12,7 @@ jobs:
     - name: Setup Python
       uses: actions/setup-python@v5
       with:
-        python-version: "3.9"
+        python-version: "3.10"
     - name: Install nox
       run: |
         python -m pip install --upgrade setuptools pip wheel

db_dtypes/__init__.py

@@ -21,17 +21,17 @@
 import warnings
 
 import numpy
-import packaging.version
 import pandas
 import pandas.api.extensions
 from pandas.errors import OutOfBoundsDatetime
 import pyarrow
 import pyarrow.compute
 
 from db_dtypes import core
+from db_dtypes.json import JSONArray, JSONDtype
 from db_dtypes.version import __version__
-from . import _versions_helpers
 
+from . import _versions_helpers
 
 date_dtype_name = "dbdate"
 time_dtype_name = "dbtime"
@@ -47,15 +47,6 @@
 _NP_BOX_DTYPE = "datetime64[us]"
 
 
-# To use JSONArray and JSONDtype, you'll need Pandas 1.5.0 or later. With the removal
-# of Python 3.7 compatibility, the minimum Pandas version will be updated to 1.5.0.
-if packaging.version.Version(pandas.__version__) >= packaging.version.Version("1.5.0"):
-    from db_dtypes.json import JSONArray, JSONDtype
-else:
-    JSONArray = None
-    JSONDtype = None
-
-
 @pandas.api.extensions.register_extension_dtype
 class TimeDtype(core.BaseDatetimeDtype):
     """
@@ -359,21 +350,12 @@ def __sub__(self, other):
     )
 
 
-if not JSONArray or not JSONDtype:
-    __all__ = [
-        "__version__",
-        "DateArray",
-        "DateDtype",
-        "TimeArray",
-        "TimeDtype",
-    ]
-else:
-    __all__ = [
-        "__version__",
-        "DateArray",
-        "DateDtype",
-        "JSONDtype",
-        "JSONArray",
-        "TimeArray",
-        "TimeDtype",
-    ]
+__all__ = [
+    "__version__",
+    "DateArray",
+    "DateDtype",
+    "JSONDtype",
+    "JSONArray",
+    "TimeArray",
+    "TimeDtype",
+]

db_dtypes/pandas_backports.py

@@ -19,13 +19,9 @@
 the versions in the later versions of pandas.
 """
 
-import operator
-from typing import Any
 
-import numpy
 import packaging.version
 import pandas
-from pandas.api.types import is_integer
 import pandas.compat.numpy.function
 import pandas.core.nanops
 
@@ -36,17 +32,15 @@
 nanany = pandas.core.nanops.nanany
 nanmax = pandas.core.nanops.nanmax
 nanmin = pandas.core.nanops.nanmin
+nanmedian = pandas.core.nanops.nanmedian
 numpy_validate_all = pandas.compat.numpy.function.validate_all
 numpy_validate_any = pandas.compat.numpy.function.validate_any
 numpy_validate_max = pandas.compat.numpy.function.validate_max
 numpy_validate_min = pandas.compat.numpy.function.validate_min
+numpy_validate_median = pandas.compat.numpy.function.validate_median
 
-if pandas_release >= (1, 3):
-    nanmedian = pandas.core.nanops.nanmedian
-    numpy_validate_median = pandas.compat.numpy.function.validate_median
 
-
-def import_default(module_name, force=False, default=None):
+def import_default(module_name, default=None):
     """
     Provide an implementation for a class or function when it can't be imported
 
@@ -57,16 +51,10 @@ def import_default(module_name, force=False, default=None):
     """
 
     if default is None:
-        return lambda func_or_class: import_default(module_name, force, func_or_class)
-
-    if force:
-        return default
+        return lambda func_or_class: import_default(module_name, func_or_class)
 
     name = default.__name__
-    try:
-        module = __import__(module_name, {}, {}, [name])
-    except ModuleNotFoundError:
-        return default
+    module = __import__(module_name, {}, {}, [name])
 
     return getattr(module, name, default)
 
@@ -77,105 +65,11 @@ def import_default(module_name, force=False, default=None):
 # 'datetime.date'
 @import_default("pandas.core.arraylike")
 class OpsMixin:
-    def _cmp_method(self, other, op):  # pragma: NO COVER
-        return NotImplemented
-
-    def __eq__(self, other):
-        return self._cmp_method(other, operator.eq)
-
-    def __ne__(self, other):
-        return self._cmp_method(other, operator.ne)
-
-    def __lt__(self, other):
-        return self._cmp_method(other, operator.lt)
-
-    def __le__(self, other):
-        return self._cmp_method(other, operator.le)
-
-    def __gt__(self, other):
-        return self._cmp_method(other, operator.gt)
-
-    def __ge__(self, other):
-        return self._cmp_method(other, operator.ge)
-
-    __add__ = __radd__ = __sub__ = lambda self, other: NotImplemented
+    pass
 
 
 # TODO: use public API once pandas 1.5 / 2.x is released.
 # See: https://github.com/pandas-dev/pandas/pull/45544
-@import_default("pandas.core.arrays._mixins", pandas_release < (1, 3))
+@import_default("pandas.core.arrays._mixins")
 class NDArrayBackedExtensionArray(pandas.core.arrays.base.ExtensionArray):
-    def __init__(self, values, dtype):
-        assert isinstance(values, numpy.ndarray)
-        self._ndarray = values
-        self._dtype = dtype
-
-    @classmethod
-    def _from_backing_data(cls, data):
-        return cls(data, data.dtype)
-
-    def __getitem__(self, index):
-        value = self._ndarray[index]
-        if is_integer(index):
-            return self._box_func(value)
-        return self.__class__(value, self._dtype)
-
-    def __setitem__(self, index, value):
-        self._ndarray[index] = self._validate_setitem_value(value)
-
-    def __len__(self):
-        return len(self._ndarray)
-
-    @property
-    def shape(self):
-        return self._ndarray.shape
-
-    @property
-    def ndim(self) -> int:
-        return self._ndarray.ndim
-
-    @property
-    def size(self) -> int:
-        return self._ndarray.size
-
-    @property
-    def nbytes(self) -> int:
-        return self._ndarray.nbytes
-
-    def copy(self):
-        return self[:]
-
-    def repeat(self, n):
-        return self.__class__(self._ndarray.repeat(n), self._dtype)
-
-    def take(
-        self,
-        indices,
-        *,
-        allow_fill: bool = False,
-        fill_value: Any = None,
-        axis: int = 0,
-    ):
-        from pandas.core.algorithms import take
-
-        if allow_fill:
-            fill_value = self._validate_scalar(fill_value)
-
-        new_data = take(
-            self._ndarray,
-            indices,
-            allow_fill=allow_fill,
-            fill_value=fill_value,
-            axis=axis,
-        )
-        return self._from_backing_data(new_data)
-
-    @classmethod
-    def _concat_same_type(cls, to_concat, axis=0):
-        dtypes = {str(x.dtype) for x in to_concat}
-        if len(dtypes) != 1:
-            raise ValueError("to_concat must have the same dtype (tz)", dtypes)
-
-        new_values = [x._ndarray for x in to_concat]
-        new_values = numpy.concatenate(new_values, axis=axis)
-        return to_concat[0]._from_backing_data(new_values)  # type: ignore[arg-type]
+    pass

tests/unit/test_date.py

@@ -23,7 +23,6 @@
 import pytest
 
 import db_dtypes
-from db_dtypes import pandas_backports
 
 VALUE_PARSING_TEST_CASES = [
     # Min/Max values for pandas.Timestamp.
@@ -48,9 +47,6 @@
     float("nan"),
 ]
 
-if hasattr(pandas, "NA"):
-    NULL_VALUE_TEST_CASES.append(pandas.NA)
-
 
 def test_box_func():
     input_array = db_dtypes.DateArray([])
@@ -261,10 +257,6 @@ def test_date_min_2d():
     )
 
 
-@pytest.mark.skipif(
-    not hasattr(pandas_backports, "numpy_validate_median"),
-    reason="median not available with this version of pandas",
-)
 @pytest.mark.parametrize(
     "values, expected",
     [
@@ -287,10 +279,6 @@ def test_date_median(values, expected):
     assert series.median() == expected
 
 
-@pytest.mark.skipif(
-    not hasattr(pandas_backports, "numpy_validate_median"),
-    reason="median not available with this version of pandas",
-)
 def test_date_median_2d():
     input_array = db_dtypes.DateArray(
         numpy.array(

tests/unit/test_dtypes.py

@@ -14,14 +14,11 @@
 
 import datetime
 
-import packaging.version
 import pytest
 
 pd = pytest.importorskip("pandas")
 np = pytest.importorskip("numpy")
 
-pandas_release = packaging.version.parse(pd.__version__).release
-
 SAMPLE_RAW_VALUES = dict(
     dbdate=(datetime.date(2021, 2, 2), "2021-2-3", pd.NaT),
     dbtime=(datetime.time(1, 2, 2), "1:2:3.5", pd.NaT),
@@ -538,39 +535,35 @@ def test_min_max_median(dtype):
         a = cls(data)
         assert a.min() == sample_values[0]
         assert a.max() == sample_values[-1]
-        if pandas_release >= (1, 3):
-            assert (
-                a.median() == datetime.time(1, 2, 4)
-                if dtype == "dbtime"
-                else datetime.date(2021, 2, 3)
-            )
+        assert (
+            a.median() == datetime.time(1, 2, 4)
+            if dtype == "dbtime"
+            else datetime.date(2021, 2, 3)
+        )
 
     empty = cls([])
     assert empty.min() is pd.NaT
     assert empty.max() is pd.NaT
-    if pandas_release >= (1, 3):
-        assert empty.median() is pd.NaT
+    assert empty.median() is pd.NaT
     empty = cls([None])
     assert empty.min() is pd.NaT
     assert empty.max() is pd.NaT
     assert empty.min(skipna=False) is pd.NaT
     assert empty.max(skipna=False) is pd.NaT
-    if pandas_release >= (1, 3):
-        with pytest.warns(RuntimeWarning, match="empty slice"):
-            # It's weird that we get the warning here, and not
-            # below. :/
-            assert empty.median() is pd.NaT
-        assert empty.median(skipna=False) is pd.NaT
+    with pytest.warns(RuntimeWarning, match="empty slice"):
+        # It's weird that we get the warning here, and not
+        # below. :/
+        assert empty.median() is pd.NaT
+    assert empty.median(skipna=False) is pd.NaT
 
     a = _make_one(dtype)
     assert a.min() == sample_values[0]
     assert a.max() == sample_values[1]
-    if pandas_release >= (1, 3):
-        assert (
-            a.median() == datetime.time(1, 2, 2, 750000)
-            if dtype == "dbtime"
-            else datetime.date(2021, 2, 2)
-        )
+    assert (
+        a.median() == datetime.time(1, 2, 2, 750000)
+        if dtype == "dbtime"
+        else datetime.date(2021, 2, 2)
+    )
 
 
 def test_date_add():
@@ -584,8 +577,7 @@ def test_date_add():
     do = pd.DateOffset(days=1)
     expect = dates.astype("object") + do
     np.testing.assert_array_equal(dates + do, expect)
-    if pandas_release >= (1, 1):
-        np.testing.assert_array_equal(do + dates, expect)
+    np.testing.assert_array_equal(do + dates, expect)
 
     with pytest.raises(TypeError):
         dates + times.astype("timedelta64")

tests/unit/test_time.py

@@ -20,7 +20,6 @@
 
 # To register the types.
 import db_dtypes  # noqa
-from db_dtypes import pandas_backports
 
 
 def test_box_func():
@@ -117,10 +116,6 @@ def test_time_parsing_errors(value, error):
         pandas.Series([value], dtype="dbtime")
 
 
-@pytest.mark.skipif(
-    not hasattr(pandas_backports, "numpy_validate_median"),
-    reason="median not available with this version of pandas",
-)
 @pytest.mark.parametrize(
     "values, expected",
     [