Skip erf tests when no scipy installed

Author: antonwolfy

dpnp/tests/helper.py

@@ -3,7 +3,6 @@
 import dpctl
 import numpy
 import pytest
-import scipy
 from numpy.testing import assert_allclose, assert_array_equal
 
 import dpnp
@@ -499,7 +498,3 @@ def requires_intel_mkl_version(version):  # pragma: no cover
 
     build_deps = numpy.show_config(mode="dicts")["Build Dependencies"]
     return build_deps["blas"]["version"] >= version
-
-
-def scipy_version():
-    return scipy.__version__

dpnp/tests/test_special.py

@@ -1,6 +1,5 @@
 import numpy
 import pytest
-import scipy
 from numpy.testing import assert_allclose
 
 import dpnp
@@ -10,16 +9,18 @@
     generate_random_numpy_array,
     get_all_dtypes,
     get_complex_dtypes,
-    scipy_version,
 )
+from .third_party.cupy.testing import installed, with_requires
 
 
+@with_requires("scipy")
 class TestErf:
-
     @pytest.mark.parametrize(
         "dt", get_all_dtypes(no_none=True, no_float16=False, no_complex=True)
     )
     def test_basic(self, dt):
+        import scipy.special
+
         a = generate_random_numpy_array((2, 5), dtype=dt)
         ia = dpnp.array(a)
 
@@ -28,12 +29,14 @@ def test_basic(self, dt):
 
         # scipy >= 0.16.0 returns float64, but dpnp returns float32
         to_float32 = dt in (dpnp.bool, dpnp.float16)
-        only_type_kind = scipy_version() >= "0.16.0" and to_float32
+        only_type_kind = installed("scipy>=0.16.0") and to_float32
         assert_dtype_allclose(
             result, expected, check_only_type_kind=only_type_kind
         )
 
     def test_nan_inf(self):
+        import scipy.special
+
         a = numpy.array([numpy.nan, -numpy.inf, numpy.inf])
         ia = dpnp.array(a)
 

dpnp/tests/test_strides.py

@@ -2,7 +2,6 @@
 
 import numpy
 import pytest
-import scipy
 from numpy.testing import assert_array_equal
 
 import dpnp
@@ -17,8 +16,8 @@
     get_integer_dtypes,
     get_integer_float_dtypes,
     numpy_version,
-    scipy_version,
 )
+from .third_party.cupy.testing import installed, with_requires
 
 
 @pytest.mark.usefixtures("suppress_divide_invalid_numpy_warnings")
@@ -167,17 +166,20 @@ def test_reduce_hypot(dtype, stride):
     assert_dtype_allclose(result, expected, check_only_type_kind=flag)
 
 
+@with_requires("scipy")
 @pytest.mark.parametrize("dtype", get_float_dtypes(no_float16=False))
 @pytest.mark.parametrize("stride", [2, -1, -3])
 def test_erf(dtype, stride):
+    import scipy.special
+
     x = generate_random_numpy_array(10, dtype=dtype)
     a, ia = x[::stride], dpnp.array(x)[::stride]
 
     result = dpnp.special.erf(ia)
     expected = scipy.special.erf(a)
 
     # scipy >= 0.16.0 returns float64, but dpnp returns float32
-    only_type_kind = scipy_version() >= "0.16.0" and (dtype == dpnp.float16)
+    only_type_kind = installed("scipy>=0.16.0") and (dtype == dpnp.float16)
     assert_dtype_allclose(result, expected, check_only_type_kind=only_type_kind)