DEPS: drop np19 support (#47796)

Author: fangchenli

.github/workflows/sdist.yml

@@ -79,9 +79,9 @@ jobs:
       run: |
         case "${{matrix.python-version}}" in
         3.8)
-          pip install numpy==1.19.5 ;;
+          pip install numpy==1.20.3 ;;
         3.9)
-          pip install numpy==1.19.5 ;;
+          pip install numpy==1.20.3 ;;
         3.10)
           pip install numpy==1.21.2 ;;
         esac

ci/deps/actions-38-minimum_versions.yaml

@@ -17,7 +17,7 @@ dependencies:
 
   # required dependencies
   - python-dateutil=2.8.1
-  - numpy=1.19.5
+  - numpy=1.20.3
   - pytz=2020.1
 
   # optional dependencies

doc/source/getting_started/install.rst

@@ -235,7 +235,7 @@ Dependencies
 ================================================================ ==========================
 Package                                                          Minimum supported version
 ================================================================ ==========================
-`NumPy <https://numpy.org>`__                                    1.19.5
+`NumPy <https://numpy.org>`__                                    1.20.3
 `python-dateutil <https://dateutil.readthedocs.io/en/stable/>`__ 2.8.1
 `pytz <https://pypi.org/project/pytz/>`__                        2020.1
 ================================================================ ==========================

doc/source/whatsnew/v1.5.0.rst

@@ -416,7 +416,7 @@ If installed, we now require:
 +-----------------+-----------------+----------+---------+
 | Package         | Minimum Version | Required | Changed |
 +=================+=================+==========+=========+
-| numpy           | 1.19.5          |    X     |    X    |
+| numpy           | 1.20.3          |    X     |    X    |
 +-----------------+-----------------+----------+---------+
 | mypy (dev)      | 0.971           |          |    X    |
 +-----------------+-----------------+----------+---------+

pandas/compat/__init__.py

@@ -17,7 +17,7 @@
 from pandas._typing import F
 from pandas.compat.numpy import (
     is_numpy_dev,
-    np_version_under1p20,
+    np_version_under1p21,
 )
 from pandas.compat.pyarrow import (
     pa_version_under1p01,
@@ -152,7 +152,7 @@ def get_lzma_file() -> type[lzma.LZMAFile]:
 
 __all__ = [
     "is_numpy_dev",
-    "np_version_under1p20",
+    "np_version_under1p21",
     "pa_version_under1p01",
     "pa_version_under2p0",
     "pa_version_under3p0",

pandas/compat/numpy/__init__.py

@@ -6,12 +6,11 @@
 # numpy versioning
 _np_version = np.__version__
 _nlv = Version(_np_version)
-np_version_under1p20 = _nlv < Version("1.20")
+np_version_under1p21 = _nlv < Version("1.21")
 np_version_under1p22 = _nlv < Version("1.22")
 np_version_gte1p22 = _nlv >= Version("1.22")
 is_numpy_dev = _nlv.dev is not None
-_min_numpy_ver = "1.19.5"
-is_numpy_min = _nlv == Version(_min_numpy_ver)
+_min_numpy_ver = "1.20.3"
 
 if is_numpy_dev or not np_version_under1p22:
     np_percentile_argname = "method"

pandas/core/array_algos/putmask.py

@@ -12,7 +12,7 @@
     ArrayLike,
     npt,
 )
-from pandas.compat import np_version_under1p20
+from pandas.compat import np_version_under1p21
 
 from pandas.core.dtypes.cast import infer_dtype_from
 from pandas.core.dtypes.common import is_list_like
@@ -66,7 +66,7 @@ def putmask_without_repeat(
     mask : np.ndarray[bool]
     new : Any
     """
-    if np_version_under1p20:
+    if np_version_under1p21:
         new = setitem_datetimelike_compat(values, mask.sum(), new)
 
     if getattr(new, "ndim", 0) >= 1:
@@ -78,7 +78,6 @@ def putmask_without_repeat(
         shape = np.shape(new)
         # np.shape compat for if setitem_datetimelike_compat
         #  changed arraylike to list e.g. test_where_dt64_2d
-
         if nlocs == shape[-1]:
             # GH#30567
             # If length of ``new`` is less than the length of ``values``,

pandas/tests/arrays/floating/test_construction.py

@@ -1,8 +1,6 @@
 import numpy as np
 import pytest
 
-from pandas.compat import np_version_under1p20
-
 import pandas as pd
 import pandas._testing as tm
 from pandas.core.arrays import FloatingArray
@@ -54,11 +52,6 @@ def test_floating_array_disallows_float16():
 
 def test_floating_array_disallows_Float16_dtype(request):
     # GH#44715
-    if np_version_under1p20:
-        # https://github.com/numpy/numpy/issues/20512
-        mark = pytest.mark.xfail(reason="numpy does not raise on np.dtype('Float16')")
-        request.node.add_marker(mark)
-
     with pytest.raises(TypeError, match="data type 'Float16' not understood"):
         pd.array([1.0, 2.0], dtype="Float16")
 

pandas/tests/arrays/sparse/test_arithmetics.py

@@ -3,11 +3,8 @@
 import numpy as np
 import pytest
 
-from pandas.compat import np_version_under1p20
-
 import pandas as pd
 import pandas._testing as tm
-from pandas.core import ops
 from pandas.core.arrays.sparse import (
     SparseArray,
     SparseDtype,
@@ -121,19 +118,7 @@ def test_float_scalar(
         self, kind, mix, all_arithmetic_functions, fill_value, scalar, request
     ):
         op = all_arithmetic_functions
-
-        if np_version_under1p20:
-            if op in [operator.floordiv, ops.rfloordiv]:
-                if op is operator.floordiv and scalar != 0:
-                    pass
-                elif op is ops.rfloordiv and scalar == 0:
-                    pass
-                else:
-                    mark = pytest.mark.xfail(raises=AssertionError, reason="GH#38172")
-                    request.node.add_marker(mark)
-
         values = np.array([np.nan, 1, 2, 0, np.nan, 0, 1, 2, 1, np.nan])
-
         a = SparseArray(values, kind=kind, fill_value=fill_value)
         self._check_numeric_ops(a, scalar, values, scalar, mix, op)
 
@@ -171,14 +156,6 @@ def test_float_same_index_with_nans(
     ):
         # when sp_index are the same
         op = all_arithmetic_functions
-
-        if (
-            np_version_under1p20
-            and op is ops.rfloordiv
-            and not (mix and kind == "block")
-        ):
-            mark = pytest.mark.xfail(raises=AssertionError, reason="GH#38172")
-            request.node.add_marker(mark)
         values = np.array([np.nan, 1, 2, 0, np.nan, 0, 1, 2, 1, np.nan])
         rvalues = np.array([np.nan, 2, 3, 4, np.nan, 0, 1, 3, 2, np.nan])
 
@@ -353,13 +330,7 @@ def test_bool_array_logical(self, kind, fill_value):
 
     def test_mixed_array_float_int(self, kind, mix, all_arithmetic_functions, request):
         op = all_arithmetic_functions
-
-        if np_version_under1p20 and op in [operator.floordiv, ops.rfloordiv] and mix:
-            mark = pytest.mark.xfail(raises=AssertionError, reason="GH#38172")
-            request.node.add_marker(mark)
-
         rdtype = "int64"
-
         values = np.array([np.nan, 1, 2, 0, np.nan, 0, 1, 2, 1, np.nan])
         rvalues = np.array([2, 0, 2, 3, 0, 0, 1, 5, 2, 0], dtype=rdtype)
 

pandas/tests/extension/base/casting.py

@@ -1,7 +1,7 @@
 import numpy as np
 import pytest
 
-from pandas.compat import np_version_under1p20
+from pandas.compat import np_version_under1p21
 import pandas.util._test_decorators as td
 
 import pandas as pd
@@ -32,8 +32,7 @@ def test_astype_object_frame(self, all_data):
         assert result._mgr.arrays[0].dtype == np.dtype(object)
 
         # earlier numpy raises TypeError on e.g. np.dtype(np.int64) == "Int64"
-        #  instead of returning False
-        if not np_version_under1p20:
+        if not np_version_under1p21:
             # check that we can compare the dtypes
             comp = result.dtypes == df.dtypes
             assert not comp.any()

pandas/tests/extension/test_sparse.py

@@ -17,7 +17,6 @@
 import numpy as np
 import pytest
 
-from pandas.compat import np_version_under1p20
 from pandas.errors import PerformanceWarning
 
 from pandas.core.dtypes.common import is_object_dtype
@@ -415,12 +414,9 @@ def test_astype_object_frame(self, all_data):
             result = df.astype(object)
         assert is_object_dtype(result._mgr.arrays[0].dtype)
 
-        # earlier numpy raises TypeError on e.g. np.dtype(np.int64) == "Int64"
-        #  instead of returning False
-        if not np_version_under1p20:
-            # check that we can compare the dtypes
-            comp = result.dtypes == df.dtypes
-            assert not comp.any()
+        # check that we can compare the dtypes
+        comp = result.dtypes == df.dtypes
+        assert not comp.any()
 
     def test_astype_str(self, data):
         with tm.assert_produces_warning(FutureWarning, match="astype from Sparse"):

pandas/tests/frame/indexing/test_where.py

@@ -4,8 +4,6 @@
 import numpy as np
 import pytest
 
-from pandas.compat import np_version_under1p20
-
 from pandas.core.dtypes.common import is_scalar
 
 import pandas as pd
@@ -1006,7 +1004,6 @@ def _check_where_equivalences(df, mask, other, expected):
     tm.assert_frame_equal(df, expected)
 
 
-@pytest.mark.xfail(np_version_under1p20, reason="failed on Numpy 1.19.5")
 def test_where_dt64_2d():
     dti = date_range("2016-01-01", periods=6)
     dta = dti._data.reshape(3, 2)

pandas/tests/frame/methods/test_quantile.py

@@ -1,7 +1,10 @@
 import numpy as np
 import pytest
 
-from pandas.compat.numpy import np_percentile_argname
+from pandas.compat.numpy import (
+    np_percentile_argname,
+    np_version_under1p21,
+)
 
 import pandas as pd
 from pandas import (
@@ -655,7 +658,7 @@ def compute_quantile(self, obj, qs):
             result = obj.quantile(qs, numeric_only=False)
         return result
 
-    def test_quantile_ea(self, obj, index):
+    def test_quantile_ea(self, request, obj, index):
 
         # result should be invariant to shuffling
         indexer = np.arange(len(index), dtype=np.intp)
@@ -665,6 +668,11 @@ def test_quantile_ea(self, obj, index):
         qs = [0.5, 0, 1]
         result = self.compute_quantile(obj, qs)
 
+        if np_version_under1p21 and index.dtype == "timedelta64[ns]":
+            msg = "failed on Numpy 1.20.3; TypeError: data type 'Int64' not understood"
+            mark = pytest.mark.xfail(reason=msg, raises=TypeError)
+            request.node.add_marker(mark)
+
         exp_dtype = index.dtype
         if index.dtype == "Int64":
             # match non-nullable casting behavior
@@ -700,7 +708,7 @@ def test_quantile_ea_with_na(self, obj, index):
 
     # TODO(GH#39763): filtering can be removed after GH#39763 is fixed
     @pytest.mark.filterwarnings("ignore:Using .astype to convert:FutureWarning")
-    def test_quantile_ea_all_na(self, obj, index):
+    def test_quantile_ea_all_na(self, request, obj, index):
         obj.iloc[:] = index._na_value
 
         # TODO(ArrayManager): this casting should be unnecessary after GH#39763 is fixed
@@ -715,14 +723,19 @@ def test_quantile_ea_all_na(self, obj, index):
         qs = [0.5, 0, 1]
         result = self.compute_quantile(obj, qs)
 
+        if np_version_under1p21 and index.dtype == "timedelta64[ns]":
+            msg = "failed on Numpy 1.20.3; TypeError: data type 'Int64' not understood"
+            mark = pytest.mark.xfail(reason=msg, raises=TypeError)
+            request.node.add_marker(mark)
+
         expected = index.take([-1, -1, -1], allow_fill=True, fill_value=index._na_value)
         expected = Series(expected, index=qs, name="A")
         if expected.dtype == "Int64":
             expected = expected.astype("Float64")
         expected = type(obj)(expected)
         tm.assert_equal(result, expected)
 
-    def test_quantile_ea_scalar(self, obj, index):
+    def test_quantile_ea_scalar(self, request, obj, index):
         # scalar qs
 
         # result should be invariant to shuffling
@@ -733,6 +746,11 @@ def test_quantile_ea_scalar(self, obj, index):
         qs = 0.5
         result = self.compute_quantile(obj, qs)
 
+        if np_version_under1p21 and index.dtype == "timedelta64[ns]":
+            msg = "failed on Numpy 1.20.3; TypeError: data type 'Int64' not understood"
+            mark = pytest.mark.xfail(reason=msg, raises=TypeError)
+            request.node.add_marker(mark)
+
         exp_dtype = index.dtype
         if index.dtype == "Int64":
             exp_dtype = "Float64"

pandas/tests/frame/methods/test_replace.py

@@ -6,8 +6,6 @@
 import numpy as np
 import pytest
 
-from pandas.compat import np_version_under1p20
-
 import pandas as pd
 from pandas import (
     DataFrame,
@@ -1316,12 +1314,6 @@ def test_replace_commutative(self, df, to_replace, exp):
     )
     def test_replace_replacer_dtype(self, request, replacer):
         # GH26632
-        if np.isscalar(replacer) and replacer.dtype.itemsize < 8:
-            request.node.add_marker(
-                pytest.mark.xfail(
-                    np_version_under1p20, reason="np.putmask doesn't coerce dtype"
-                )
-            )
         df = DataFrame(["a"])
         result = df.replace({"a": replacer, "b": replacer})
         expected = DataFrame([replacer])

pandas/tests/indexing/test_iloc.py

@@ -10,7 +10,6 @@
 import numpy as np
 import pytest
 
-from pandas.compat.numpy import is_numpy_min
 from pandas.errors import IndexingError
 import pandas.util._test_decorators as td
 
@@ -1199,7 +1198,6 @@ def test_iloc_getitem_int_single_ea_block_view(self):
         arr[2] = arr[-1]
         assert ser[0] == arr[-1]
 
-    @pytest.mark.xfail(is_numpy_min, reason="Column A gets coerced to integer type")
     def test_iloc_setitem_multicolumn_to_datetime(self, using_array_manager):
 
         # GH#20511

pandas/tests/series/test_constructors.py

@@ -13,7 +13,6 @@
     iNaT,
     lib,
 )
-from pandas.compat.numpy import np_version_under1p20
 import pandas.util._test_decorators as td
 
 from pandas.core.dtypes.common import (
@@ -1904,9 +1903,6 @@ def test_constructor_dtype_timedelta_ns_s_astype_int64(self):
     @pytest.mark.filterwarnings(
         "ignore:elementwise comparison failed:DeprecationWarning"
     )
-    @pytest.mark.xfail(
-        np_version_under1p20, reason="np.array([td64nat, float, float]) raises"
-    )
     @pytest.mark.parametrize("func", [Series, DataFrame, Index, pd.array])
     def test_constructor_mismatched_null_nullable_dtype(
         self, func, any_numeric_ea_dtype

setup.cfg

@@ -31,9 +31,7 @@ project_urls =
 [options]
 packages = find:
 install_requires =
-    numpy>=1.18.5; platform_machine!='aarch64' and platform_machine!='arm64' and python_version<'3.10'
-    numpy>=1.19.2; platform_machine=='aarch64' and python_version<'3.10'
-    numpy>=1.20.0; platform_machine=='arm64' and python_version<'3.10'
+    numpy>=1.20.3; python_version<'3.10'
     numpy>=1.21.0; python_version>='3.10'
     python-dateutil>=2.8.1
     pytz>=2020.1