TST: skip->xfail (#44647)

Author: jbrockmendel

pandas/tests/arrays/masked/test_arithmetic.py

@@ -48,9 +48,11 @@ def test_array_scalar_like_equivalence(data, all_arithmetic_operators):
         tm.assert_extension_array_equal(result, expected)
 
 
-def test_array_NA(data, all_arithmetic_operators):
-    if "truediv" in all_arithmetic_operators:
-        pytest.skip("division with pd.NA raises")
+def test_array_NA(data, all_arithmetic_operators, request):
+    if "truediv" in all_arithmetic_operators and data[0].dtype.kind != "f":
+        mark = pytest.mark.xfail(reason="division with pd.NA fails")
+        request.node.add_marker(mark)
+
     data, _ = data
     op = tm.get_op_from_name(all_arithmetic_operators)
     check_skip(data, all_arithmetic_operators)

pandas/tests/base/test_conversion.py

@@ -265,10 +265,12 @@ def test_numpy_array_all_dtypes(any_numpy_dtype):
         ),
     ],
 )
-def test_array(arr, attr, index_or_series):
+def test_array(arr, attr, index_or_series, request):
     box = index_or_series
     if arr.dtype.name in ("Int64", "Sparse[int64, 0]") and box is pd.Index:
-        pytest.skip(f"No index type for {arr.dtype}")
+        mark = pytest.mark.xfail(reason="Needs EA-Backed Index")
+        request.node.add_marker(mark)
+
     result = box(arr, copy=False).array
 
     if attr:
@@ -341,9 +343,6 @@ def test_to_numpy(arr, expected, index_or_series_or_array, request):
     box = index_or_series_or_array
     thing = box(arr)
 
-    if arr.dtype.name in ("Int64", "Sparse[int64, 0]") and box is pd.Index:
-        pytest.skip(f"No index type for {arr.dtype}")
-
     if arr.dtype.name == "int64" and box is pd.array:
         mark = pytest.mark.xfail(reason="thing is Int64 and to_numpy() returns object")
         request.node.add_marker(mark)

pandas/tests/base/test_misc.py

@@ -15,7 +15,6 @@
 
 import pandas as pd
 from pandas import (
-    DataFrame,
     Index,
     Series,
 )
@@ -33,10 +32,11 @@
         ("floordiv", "//"),
     ],
 )
-@pytest.mark.parametrize("klass", [Series, DataFrame])
-def test_binary_ops_docstring(klass, op_name, op):
+def test_binary_ops_docstring(frame_or_series, op_name, op):
     # not using the all_arithmetic_functions fixture with _get_opstr
     # as _get_opstr is used internally in the dynamic implementation of the docstring
+    klass = frame_or_series
+
     operand1 = klass.__name__.lower()
     operand2 = "other"
     expected_str = " ".join([operand1, op, operand2])
@@ -134,12 +134,11 @@ def test_searchsorted(index_or_series_obj):
     assert 0 <= index <= len(obj)
 
 
-def test_access_by_position(index):
+def test_access_by_position(index_flat):
+    index = index_flat
 
     if len(index) == 0:
         pytest.skip("Test doesn't make sense on empty data")
-    elif isinstance(index, pd.MultiIndex):
-        pytest.skip("Can't instantiate Series from MultiIndex")
 
     series = Series(index)
     assert index[0] == series.iloc[0]

pandas/tests/extension/test_datetime.py

@@ -114,12 +114,7 @@ def test_combine_add(self, data_repeated):
 
 
 class TestInterface(BaseDatetimeTests, base.BaseInterfaceTests):
-    def test_array_interface(self, data):
-        if data.tz:
-            # np.asarray(DTA) is currently always tz-naive.
-            pytest.skip("GH-23569")
-        else:
-            super().test_array_interface(data)
+    pass
 
 
 class TestArithmeticOps(BaseDatetimeTests, base.BaseArithmeticOpsTests):

pandas/tests/indexes/test_base.py

@@ -822,14 +822,13 @@ def test_isin_nan_common_float64(self, request, nulls_fixture):
                 pytest.mark.xfail(reason="Float64Index cannot contain pd.NA")
             )
 
-        tm.assert_numpy_array_equal(
-            Float64Index([1.0, nulls_fixture]).isin([np.nan]), np.array([False, True])
-        )
+        idx = Float64Index([1.0, nulls_fixture])
+        res = idx.isin([np.nan])
+        tm.assert_numpy_array_equal(res, np.array([False, True]))
 
         # we cannot compare NaT with NaN
-        tm.assert_numpy_array_equal(
-            Float64Index([1.0, nulls_fixture]).isin([pd.NaT]), np.array([False, False])
-        )
+        res = idx.isin([pd.NaT])
+        tm.assert_numpy_array_equal(res, np.array([False, False]))
 
     @pytest.mark.parametrize("level", [0, -1])
     @pytest.mark.parametrize(

pandas/tests/indexing/test_coercion.py

@@ -610,10 +610,12 @@ def test_where_object(self, index_or_series, fill_val, exp_dtype):
         "fill_val,exp_dtype",
         [(1, np.int64), (1.1, np.float64), (1 + 1j, np.complex128), (True, object)],
     )
-    def test_where_int64(self, index_or_series, fill_val, exp_dtype):
+    def test_where_int64(self, index_or_series, fill_val, exp_dtype, request):
         klass = index_or_series
         if klass is pd.Index and exp_dtype is np.complex128:
-            pytest.skip("Complex Index not supported")
+            mark = pytest.mark.xfail(reason="Complex Index not supported")
+            request.node.add_marker(mark)
+
         obj = klass([1, 2, 3, 4])
         assert obj.dtype == np.int64
         cond = klass([True, False, True, False])
@@ -632,10 +634,12 @@ def test_where_int64(self, index_or_series, fill_val, exp_dtype):
         "fill_val, exp_dtype",
         [(1, np.float64), (1.1, np.float64), (1 + 1j, np.complex128), (True, object)],
     )
-    def test_where_float64(self, index_or_series, fill_val, exp_dtype):
+    def test_where_float64(self, index_or_series, fill_val, exp_dtype, request):
         klass = index_or_series
         if klass is pd.Index and exp_dtype is np.complex128:
-            pytest.skip("Complex Index not supported")
+            mark = pytest.mark.xfail(reason="Complex Index not supported")
+            request.node.add_marker(mark)
+
         obj = klass([1.1, 2.2, 3.3, 4.4])
         assert obj.dtype == np.float64
         cond = klass([True, False, True, False])

pandas/tests/io/parser/test_skiprows.py

@@ -182,7 +182,7 @@ def test_skip_row_with_newline_and_quote(all_parsers, data, exp_data):
 @pytest.mark.parametrize(
     "line_terminator", ["\n", "\r\n", "\r"]  # "LF"  # "CRLF"  # "CR"
 )
-def test_skiprows_lineterminator(all_parsers, line_terminator):
+def test_skiprows_lineterminator(all_parsers, line_terminator, request):
     # see gh-9079
     parser = all_parsers
     data = "\n".join(
@@ -203,7 +203,8 @@ def test_skiprows_lineterminator(all_parsers, line_terminator):
     )
 
     if parser.engine == "python" and line_terminator == "\r":
-        pytest.skip("'CR' not respect with the Python parser yet")
+        mark = pytest.mark.xfail(reason="'CR' not respect with the Python parser yet")
+        request.node.add_marker(mark)
 
     data = data.replace("\n", line_terminator)
     result = parser.read_csv(

pandas/tests/io/pytables/test_round_trip.py

@@ -9,6 +9,7 @@
 import pytest
 
 from pandas._libs.tslibs import Timestamp
+from pandas.compat import is_platform_windows
 
 import pandas as pd
 from pandas import (
@@ -350,7 +351,10 @@ def test_timeseries_preepoch(setup_path):
     try:
         _check_roundtrip(ts, tm.assert_series_equal, path=setup_path)
     except OverflowError:
-        pytest.skip("known failure on some windows platforms")
+        if is_platform_windows():
+            pytest.xfail("known failure on some windows platforms")
+        else:
+            raise
 
 
 @pytest.mark.parametrize(

pandas/tests/reductions/test_reductions.py

@@ -79,17 +79,18 @@ def test_ops(self, opname, obj):
             ("boolean", True),
         ],
     )
-    def test_nanminmax(self, opname, dtype, val, index_or_series):
+    def test_nanminmax(self, opname, dtype, val, index_or_series, request):
         # GH#7261
         klass = index_or_series
 
         if dtype in ["Int64", "boolean"] and klass == Index:
-            pytest.skip("EAs can't yet be stored in an index")
+            mark = pytest.mark.xfail(reason="Need EA-backed Index")
+            request.node.add_marker(mark)
 
         def check_missing(res):
             if dtype == "datetime64[ns]":
                 return res is NaT
-            elif dtype == "Int64":
+            elif dtype in ["Int64", "boolean"]:
                 return res is pd.NA
             else:
                 return isna(res)

pandas/tests/series/indexing/test_setitem.py

@@ -652,6 +652,7 @@ def test_series_where(self, obj, key, expected, val, is_inplace):
 
     def test_index_where(self, obj, key, expected, val, request):
         if Index(obj).dtype != obj.dtype:
+            # TODO(ExtensionIndex): Should become unreachable
             pytest.skip("test not applicable for this dtype")
 
         mask = np.zeros(obj.shape, dtype=bool)
@@ -667,6 +668,7 @@ def test_index_where(self, obj, key, expected, val, request):
 
     def test_index_putmask(self, obj, key, expected, val):
         if Index(obj).dtype != obj.dtype:
+            # TODO(ExtensionIndex): Should become unreachable
             pytest.skip("test not applicable for this dtype")
 
         mask = np.zeros(obj.shape, dtype=bool)

pandas/tests/series/methods/test_diff.py

@@ -11,21 +11,22 @@
 
 class TestSeriesDiff:
     def test_diff_np(self):
-        pytest.skip("skipping due to Series no longer being an ndarray")
+        # TODO(__array_function__): could make np.diff return a Series
+        #  matching ser.diff()
 
-        # no longer works as the return type of np.diff is now nd.array
-        s = Series(np.arange(5))
+        ser = Series(np.arange(5))
 
-        r = np.diff(s)
-        tm.assert_series_equal(Series([np.nan, 0, 0, 0, np.nan]), r)
+        res = np.diff(ser)
+        expected = np.array([1, 1, 1, 1])
+        tm.assert_numpy_array_equal(res, expected)
 
     def test_diff_int(self):
         # int dtype
         a = 10000000000000000
         b = a + 1
-        s = Series([a, b])
+        ser = Series([a, b])
 
-        result = s.diff()
+        result = ser.diff()
         assert result[1] == 1
 
     def test_diff_tz(self):
@@ -43,22 +44,24 @@ def test_diff_tz(self):
         expected = ts - ts
         tm.assert_series_equal(result, expected)
 
+    def test_diff_dt64(self):
         # datetime diff (GH#3100)
-        s = Series(date_range("20130102", periods=5))
-        result = s.diff()
-        expected = s - s.shift(1)
+        ser = Series(date_range("20130102", periods=5))
+        result = ser.diff()
+        expected = ser - ser.shift(1)
         tm.assert_series_equal(result, expected)
 
         # timedelta diff
         result = result - result.shift(1)  # previous result
         expected = expected.diff()  # previously expected
         tm.assert_series_equal(result, expected)
 
+    def test_diff_dt64tz(self):
         # with tz
-        s = Series(
+        ser = Series(
             date_range("2000-01-01 09:00:00", periods=5, tz="US/Eastern"), name="foo"
         )
-        result = s.diff()
+        result = ser.diff()
         expected = Series(TimedeltaIndex(["NaT"] + ["1 days"] * 4), name="foo")
         tm.assert_series_equal(result, expected)
 
@@ -68,14 +71,14 @@ def test_diff_tz(self):
     )
     def test_diff_bool(self, input, output, diff):
         # boolean series (test for fixing #17294)
-        s = Series(input)
-        result = s.diff()
+        ser = Series(input)
+        result = ser.diff()
         expected = Series(output)
         tm.assert_series_equal(result, expected)
 
     def test_diff_object_dtype(self):
         # object series
-        s = Series([False, True, 5.0, np.nan, True, False])
-        result = s.diff()
-        expected = s - s.shift(1)
+        ser = Series([False, True, 5.0, np.nan, True, False])
+        result = ser.diff()
+        expected = ser - ser.shift(1)
         tm.assert_series_equal(result, expected)

pandas/tests/series/test_ufunc.py

@@ -171,12 +171,15 @@ def test_binary_ufunc_scalar(ufunc, sparse, flip, arrays_for_binary_ufunc):
 @pytest.mark.parametrize("sparse", SPARSE, ids=SPARSE_IDS)
 @pytest.mark.parametrize("shuffle", SHUFFLE)
 @pytest.mark.filterwarnings("ignore:divide by zero:RuntimeWarning")
-def test_multiple_output_binary_ufuncs(ufunc, sparse, shuffle, arrays_for_binary_ufunc):
+def test_multiple_output_binary_ufuncs(
+    ufunc, sparse, shuffle, arrays_for_binary_ufunc, request
+):
     # Test that
     #  the same conditions from binary_ufunc_scalar apply to
     #  ufuncs with multiple outputs.
     if sparse and ufunc is np.divmod:
-        pytest.skip("sparse divmod not implemented.")
+        mark = pytest.mark.xfail(reason="sparse divmod not implemented")
+        request.node.add_marker(mark)
 
     a1, a2 = arrays_for_binary_ufunc
     # work around https://github.com/pandas-dev/pandas/issues/26987

pandas/tests/strings/test_strings.py

@@ -364,9 +364,12 @@ def test_len_mixed():
         ("rindex", "E", 0, 5, [4, 3, 1, 4]),
     ],
 )
-def test_index(method, sub, start, end, index_or_series, any_string_dtype, expected):
+def test_index(
+    method, sub, start, end, index_or_series, any_string_dtype, expected, request
+):
     if index_or_series is Index and not any_string_dtype == "object":
-        pytest.skip("Index cannot yet be backed by a StringArray/ArrowStringArray")
+        mark = pytest.mark.xfail(reason="Need EA-backed Index")
+        request.node.add_marker(mark)
 
     obj = index_or_series(
         ["ABCDEFG", "BCDEFEF", "DEFGHIJEF", "EFGHEF"], dtype=any_string_dtype