TST/CLN: tighten some xfails (pandas-dev#41469)

Author: simonjayhawkins

doc/source/development/code_style.rst

@@ -57,7 +57,8 @@ xfail during the testing phase. To do so, use the ``request`` fixture:
     import pytest
 
     def test_xfail(request):
-        request.node.add_marker(pytest.mark.xfail(reason="Indicate why here"))
+        mark = pytest.mark.xfail(raises=TypeError, reason="Indicate why here")
+        request.node.add_marker(mark)
 
 xfail is not to be used for tests involving failure due to invalid user arguments.
 For these tests, we need to verify the correct exception type and error message

pandas/tests/apply/test_frame_apply.py

@@ -164,8 +164,9 @@ def test_apply_with_string_funcs(request, float_frame, func, args, kwds, how):
     if len(args) > 1 and how == "agg":
         request.node.add_marker(
             pytest.mark.xfail(
+                raises=TypeError,
                 reason="agg/apply signature mismatch - agg passes 2nd "
-                "argument to func"
+                "argument to func",
             )
         )
     result = getattr(float_frame, how)(func, *args, **kwds)

pandas/tests/apply/test_frame_transform.py

@@ -173,7 +173,9 @@ def test_transform_bad_dtype(op, frame_or_series, request):
     # GH 35964
     if op == "rank":
         request.node.add_marker(
-            pytest.mark.xfail(reason="GH 40418: rank does not raise a TypeError")
+            pytest.mark.xfail(
+                raises=ValueError, reason="GH 40418: rank does not raise a TypeError"
+            )
         )
 
     obj = DataFrame({"A": 3 * [object]})  # DataFrame that will fail on most transforms

pandas/tests/apply/test_series_apply.py

@@ -262,7 +262,9 @@ def test_transform_partial_failure(op, request):
     # GH 35964
     if op in ("ffill", "bfill", "pad", "backfill", "shift"):
         request.node.add_marker(
-            pytest.mark.xfail(reason=f"{op} is successful on any dtype")
+            pytest.mark.xfail(
+                raises=AssertionError, reason=f"{op} is successful on any dtype"
+            )
         )
     if op in ("rank", "fillna"):
         pytest.skip(f"{op} doesn't raise TypeError on object")

pandas/tests/arithmetic/test_interval.py

@@ -135,7 +135,8 @@ def test_compare_scalar_na(self, op, interval_array, nulls_fixture, request):
 
         if nulls_fixture is pd.NA and interval_array.dtype.subtype != "int64":
             mark = pytest.mark.xfail(
-                reason="broken for non-integer IntervalArray; see GH 31882"
+                raises=AssertionError,
+                reason="broken for non-integer IntervalArray; see GH 31882",
             )
             request.node.add_marker(mark)
 
@@ -220,7 +221,7 @@ def test_compare_list_like_nan(self, op, interval_array, nulls_fixture, request)
 
         if nulls_fixture is pd.NA and interval_array.dtype.subtype != "i8":
             reason = "broken for non-integer IntervalArray; see GH 31882"
-            mark = pytest.mark.xfail(reason=reason)
+            mark = pytest.mark.xfail(raises=AssertionError, reason=reason)
             request.node.add_marker(mark)
 
         tm.assert_numpy_array_equal(result, expected)

pandas/tests/arrays/masked/test_arithmetic.py

@@ -170,7 +170,9 @@ def test_unary_op_does_not_propagate_mask(data, op, request):
     data, _ = data
     if data.dtype in ["Float32", "Float64"] and op == "__invert__":
         request.node.add_marker(
-            pytest.mark.xfail(reason="invert is not implemented for float ea dtypes")
+            pytest.mark.xfail(
+                raises=TypeError, reason="invert is not implemented for float ea dtypes"
+            )
         )
     s = pd.Series(data)
     result = getattr(s, op)()

pandas/tests/arrays/masked/test_arrow_compat.py

@@ -173,7 +173,7 @@ def test_from_arrow_type_error(request, data):
         # TODO numeric dtypes cast any incoming array to the correct dtype
         # instead of erroring
         request.node.add_marker(
-            pytest.mark.xfail(reason="numeric dtypes don't error but cast")
+            pytest.mark.xfail(raises=None, reason="numeric dtypes don't error but cast")
         )
 
     arr = pa.array(data).cast("string")

pandas/tests/arrays/sparse/test_arithmetics.py

@@ -132,7 +132,7 @@ def test_float_scalar(
                 elif op is ops.rfloordiv and scalar == 0:
                     pass
                 else:
-                    mark = pytest.mark.xfail(reason="GH#38172")
+                    mark = pytest.mark.xfail(raises=AssertionError, reason="GH#38172")
                     request.node.add_marker(mark)
 
         values = self._base([np.nan, 1, 2, 0, np.nan, 0, 1, 2, 1, np.nan])
@@ -177,11 +177,13 @@ def test_float_same_index_with_nans(
         # when sp_index are the same
         op = all_arithmetic_functions
 
-        if not np_version_under1p20:
-            if op is ops.rfloordiv:
-                if not (mix and kind == "block"):
-                    mark = pytest.mark.xfail(reason="GH#38172")
-                    request.node.add_marker(mark)
+        if (
+            not 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 = self._base([np.nan, 1, 2, 0, np.nan, 0, 1, 2, 1, np.nan])
         rvalues = self._base([np.nan, 2, 3, 4, np.nan, 0, 1, 3, 2, np.nan])
@@ -358,10 +360,13 @@ 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 not np_version_under1p20:
-            if op in [operator.floordiv, ops.rfloordiv] and mix:
-                mark = pytest.mark.xfail(reason="GH#38172")
-                request.node.add_marker(mark)
+        if (
+            not 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"
 

pandas/tests/arrays/string_/test_string.py

@@ -115,10 +115,10 @@ def test_astype_roundtrip(dtype, request):
 def test_add(dtype, request):
     if dtype == "arrow_string":
         reason = (
-            "TypeError: unsupported operand type(s) for +: 'ArrowStringArray' and "
+            "unsupported operand type(s) for +: 'ArrowStringArray' and "
             "'ArrowStringArray'"
         )
-        mark = pytest.mark.xfail(reason=reason)
+        mark = pytest.mark.xfail(raises=TypeError, reason=reason)
         request.node.add_marker(mark)
 
     a = pd.Series(["a", "b", "c", None, None], dtype=dtype)
@@ -143,7 +143,7 @@ def test_add(dtype, request):
 def test_add_2d(dtype, request):
     if dtype == "arrow_string":
         reason = "Failed: DID NOT RAISE <class 'ValueError'>"
-        mark = pytest.mark.xfail(reason=reason)
+        mark = pytest.mark.xfail(raises=None, reason=reason)
         request.node.add_marker(mark)
 
     a = pd.array(["a", "b", "c"], dtype=dtype)
@@ -158,11 +158,8 @@ def test_add_2d(dtype, request):
 
 def test_add_sequence(dtype, request):
     if dtype == "arrow_string":
-        reason = (
-            "TypeError: unsupported operand type(s) for +: 'ArrowStringArray' "
-            "and 'list'"
-        )
-        mark = pytest.mark.xfail(reason=reason)
+        reason = "unsupported operand type(s) for +: 'ArrowStringArray' and 'list'"
+        mark = pytest.mark.xfail(raises=TypeError, reason=reason)
         request.node.add_marker(mark)
 
     a = pd.array(["a", "b", None, None], dtype=dtype)
@@ -179,10 +176,8 @@ def test_add_sequence(dtype, request):
 
 def test_mul(dtype, request):
     if dtype == "arrow_string":
-        reason = (
-            "TypeError: unsupported operand type(s) for *: 'ArrowStringArray' and 'int'"
-        )
-        mark = pytest.mark.xfail(reason=reason)
+        reason = "unsupported operand type(s) for *: 'ArrowStringArray' and 'int'"
+        mark = pytest.mark.xfail(raises=TypeError, reason=reason)
         request.node.add_marker(mark)
 
     a = pd.array(["a", "b", None], dtype=dtype)
@@ -246,7 +241,7 @@ def test_comparison_methods_scalar_pd_na(all_compare_operators, dtype):
 def test_comparison_methods_scalar_not_string(all_compare_operators, dtype, request):
     if all_compare_operators not in ["__eq__", "__ne__"]:
         reason = "comparison op not supported between instances of 'str' and 'int'"
-        mark = pytest.mark.xfail(reason=reason)
+        mark = pytest.mark.xfail(raises=TypeError, reason=reason)
         request.node.add_marker(mark)
 
     op_name = all_compare_operators
@@ -262,11 +257,9 @@ def test_comparison_methods_scalar_not_string(all_compare_operators, dtype, requ
 
 def test_comparison_methods_array(all_compare_operators, dtype, request):
     if dtype == "arrow_string":
-        if all_compare_operators in ["__eq__", "__ne__"]:
-            reason = "NotImplementedError: Neither scalar nor ArrowStringArray"
-        else:
-            reason = "AssertionError: left is not an ExtensionArray"
-        mark = pytest.mark.xfail(reason=reason)
+        mark = pytest.mark.xfail(
+            raises=AssertionError, reason="left is not an ExtensionArray"
+        )
         request.node.add_marker(mark)
 
     op_name = all_compare_operators
@@ -309,8 +302,9 @@ def test_constructor_raises(cls):
 @pytest.mark.parametrize("copy", [True, False])
 def test_from_sequence_no_mutate(copy, cls, request):
     if cls is ArrowStringArray and copy is False:
-        reason = "AssertionError: numpy array are different"
-        mark = pytest.mark.xfail(reason=reason)
+        mark = pytest.mark.xfail(
+            raises=AssertionError, reason="numpy array are different"
+        )
         request.node.add_marker(mark)
 
     nan_arr = np.array(["a", np.nan], dtype=object)
@@ -333,8 +327,8 @@ def test_from_sequence_no_mutate(copy, cls, request):
 
 def test_astype_int(dtype, request):
     if dtype == "arrow_string":
-        reason = "TypeError: Cannot interpret 'Int64Dtype()' as a data type"
-        mark = pytest.mark.xfail(reason=reason)
+        reason = "Cannot interpret 'Int64Dtype()' as a data type"
+        mark = pytest.mark.xfail(raises=TypeError, reason=reason)
         request.node.add_marker(mark)
 
     arr = pd.array(["1", pd.NA, "3"], dtype=dtype)
@@ -349,12 +343,10 @@ def test_astype_float(dtype, any_float_allowed_nullable_dtype, request):
 
     if dtype == "arrow_string":
         if any_float_allowed_nullable_dtype in {"Float32", "Float64"}:
-            reason = "TypeError: Cannot interpret 'Float32Dtype()' as a data type"
+            reason = "Cannot interpret 'Float32Dtype()' as a data type"
         else:
-            reason = (
-                "TypeError: float() argument must be a string or a number, not 'NAType'"
-            )
-        mark = pytest.mark.xfail(reason=reason)
+            reason = "float() argument must be a string or a number, not 'NAType'"
+        mark = pytest.mark.xfail(raises=TypeError, reason=reason)
         request.node.add_marker(mark)
 
     ser = pd.Series(["1.1", pd.NA, "3.3"], dtype=dtype)
@@ -376,8 +368,8 @@ def test_reduce(skipna, dtype):
 @pytest.mark.parametrize("skipna", [True, False])
 def test_min_max(method, skipna, dtype, request):
     if dtype == "arrow_string":
-        reason = "AttributeError: 'ArrowStringArray' object has no attribute 'max'"
-        mark = pytest.mark.xfail(reason=reason)
+        reason = "'ArrowStringArray' object has no attribute 'max'"
+        mark = pytest.mark.xfail(raises=AttributeError, reason=reason)
         request.node.add_marker(mark)
 
     arr = pd.Series(["a", "b", "c", None], dtype=dtype)
@@ -394,13 +386,12 @@ def test_min_max(method, skipna, dtype, request):
 def test_min_max_numpy(method, box, dtype, request):
     if dtype == "arrow_string":
         if box is pd.array:
-            reason = (
-                "TypeError: '<=' not supported between instances of 'str' and "
-                "'NoneType'"
-            )
+            raises = TypeError
+            reason = "'<=' not supported between instances of 'str' and 'NoneType'"
         else:
-            reason = "AttributeError: 'ArrowStringArray' object has no attribute 'max'"
-        mark = pytest.mark.xfail(reason=reason)
+            raises = AttributeError
+            reason = "'ArrowStringArray' object has no attribute 'max'"
+        mark = pytest.mark.xfail(raises=raises, reason=reason)
         request.node.add_marker(mark)
 
     arr = box(["a", "b", "c", None], dtype=dtype)
@@ -425,10 +416,10 @@ def test_fillna_args(dtype, request):
 
     if dtype == "arrow_string":
         reason = (
-            "AssertionError: Regex pattern \"Cannot set non-string value '1' into "
+            "Regex pattern \"Cannot set non-string value '1' into "
             "a StringArray.\" does not match 'Scalar must be NA or str'"
         )
-        mark = pytest.mark.xfail(reason=reason)
+        mark = pytest.mark.xfail(raises=AssertionError, reason=reason)
         request.node.add_marker(mark)
 
     arr = pd.array(["a", pd.NA], dtype=dtype)

pandas/tests/arrays/test_datetimelike.py

@@ -306,7 +306,9 @@ def test_searchsorted_castable_strings(self, arr1d, box, request):
                 # If we have e.g. tzutc(), when we cast to string and parse
                 #  back we get pytz.UTC, and then consider them different timezones
                 #  so incorrectly raise.
-                mark = pytest.mark.xfail(reason="timezone comparisons inconsistent")
+                mark = pytest.mark.xfail(
+                    raises=TypeError, reason="timezone comparisons inconsistent"
+                )
                 request.node.add_marker(mark)
 
         arr = arr1d
@@ -471,7 +473,9 @@ def test_setitem_strs(self, arr1d, request):
                 # If we have e.g. tzutc(), when we cast to string and parse
                 #  back we get pytz.UTC, and then consider them different timezones
                 #  so incorrectly raise.
-                mark = pytest.mark.xfail(reason="timezone comparisons inconsistent")
+                mark = pytest.mark.xfail(
+                    raises=TypeError, reason="timezone comparisons inconsistent"
+                )
                 request.node.add_marker(mark)
 
         # Setting list-like of strs