tests: move xfail out of functions (#265)

* tests: move xfail out of functions

* remove stray print

* Update tests/test_core.py

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* add engine in test_dtype

---------

Co-authored-by: Deepak Cherian <[email protected]>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

Author: 3 people

tests/__init__.py

@@ -1,5 +1,5 @@
 import importlib
-from contextlib import contextmanager
+from contextlib import nullcontext
 
 import numpy as np
 import packaging.version
@@ -46,6 +46,7 @@ def LooseVersion(vstring):
 
 
 has_dask, requires_dask = _importorskip("dask")
+has_numba, requires_numba = _importorskip("numba")
 has_xarray, requires_xarray = _importorskip("xarray")
 
 
@@ -67,15 +68,10 @@ def __call__(self, dsk, keys, **kwargs):
         return dask.get(dsk, keys, **kwargs)
 
 
-@contextmanager
-def dummy_context():
-    yield None
-
-
 def raise_if_dask_computes(max_computes=0):
     # return a dummy context manager so that this can be used for non-dask objects
     if not has_dask:
-        return dummy_context()
+        return nullcontext()
     scheduler = CountingScheduler(max_computes)
     return dask.config.set(scheduler=scheduler)
 

tests/conftest.py

@@ -1,11 +1,10 @@
 import pytest
 
+from . import requires_numba
 
-@pytest.fixture(scope="module", params=["flox", "numpy", "numba"])
+
+@pytest.fixture(
+    scope="module", params=["flox", "numpy", pytest.param("numba", marks=requires_numba)]
+)
 def engine(request):
-    if request.param == "numba":
-        try:
-            import numba  # noqa
-        except ImportError:
-            pytest.xfail()
     return request.param

tests/test_core.py

@@ -105,7 +105,7 @@ def test_alignment_error():
 
 
 @pytest.mark.parametrize("dtype", (float, int))
-@pytest.mark.parametrize("chunk", [False, True])
+@pytest.mark.parametrize("chunk", [False, pytest.param(True, marks=requires_dask)])
 # TODO: make this intp when python 3.8 is dropped
 @pytest.mark.parametrize("expected_groups", [None, [0, 1, 2], np.array([0, 1, 2], dtype=np.int64)])
 @pytest.mark.parametrize(
@@ -145,7 +145,7 @@ def test_groupby_reduce(
 ) -> None:
     array = array.astype(dtype)
     if chunk:
-        if not has_dask or expected_groups is None:
+        if expected_groups is None:
             pytest.skip()
         array = da.from_array(array, chunks=(3,) if array.ndim == 1 else (1, 3))
         by = da.from_array(by, chunks=(3,) if by.ndim == 1 else (1, 3))
@@ -166,7 +166,7 @@ def test_groupby_reduce(
         engine=engine,
     )
     # we use pd.Index(expected_groups).to_numpy() which is always int64
-    # for the values in this tests
+    # for the values in this test
     if expected_groups is None:
         g_dtype = by.dtype
     elif isinstance(expected_groups, np.ndarray):
@@ -191,14 +191,20 @@ def gen_array_by(size, func):
     return array, by
 
 
-@pytest.mark.parametrize("chunks", [None, -1, 3, 4])
+@pytest.mark.parametrize(
+    "chunks",
+    [
+        None,
+        pytest.param(-1, marks=requires_dask),
+        pytest.param(3, marks=requires_dask),
+        pytest.param(4, marks=requires_dask),
+    ],
+)
 @pytest.mark.parametrize("nby", [1, 2, 3])
 @pytest.mark.parametrize("size", ((12,), (12, 9)))
 @pytest.mark.parametrize("add_nan_by", [True, False])
 @pytest.mark.parametrize("func", ALL_FUNCS)
 def test_groupby_reduce_all(nby, size, chunks, func, add_nan_by, engine):
-    if chunks is not None and not has_dask:
-        pytest.skip()
     if "arg" in func and engine == "flox":
         pytest.skip()
 
@@ -390,16 +396,16 @@ def test_numpy_reduce_nd_md():
 def test_groupby_agg_dask(func, shape, array_chunks, group_chunks, add_nan, dtype, engine, reindex):
     """Tests groupby_reduce with dask arrays against groupby_reduce with numpy arrays"""
 
-    rng = np.random.default_rng(12345)
-    array = dask.array.from_array(rng.random(shape), chunks=array_chunks).astype(dtype)
-    array = dask.array.ones(shape, chunks=array_chunks)
-
     if func in ["first", "last"]:
         pytest.skip()
 
     if "arg" in func and (engine == "flox" or reindex):
         pytest.skip()
 
+    rng = np.random.default_rng(12345)
+    array = dask.array.from_array(rng.random(shape), chunks=array_chunks).astype(dtype)
+    array = dask.array.ones(shape, chunks=array_chunks)
+
     labels = np.array([0, 0, 2, 2, 2, 1, 1, 2, 2, 1, 1, 0])
     if add_nan:
         labels = labels.astype(float)
@@ -612,7 +618,14 @@ def test_groupby_reduce_axis_subset_against_numpy(func, axis, engine):
     assert_equal(actual, expected, tolerance)
 
 
-@pytest.mark.parametrize("reindex,chunks", [(None, None), (False, (2, 2, 3)), (True, (2, 2, 3))])
+@pytest.mark.parametrize(
+    "reindex, chunks",
+    [
+        (None, None),
+        pytest.param(False, (2, 2, 3), marks=requires_dask),
+        pytest.param(True, (2, 2, 3), marks=requires_dask),
+    ],
+)
 @pytest.mark.parametrize(
     "axis, groups, expected_shape",
     [
@@ -624,8 +637,6 @@ def test_groupby_reduce_axis_subset_against_numpy(func, axis, engine):
 def test_groupby_reduce_nans(reindex, chunks, axis, groups, expected_shape, engine):
     def _maybe_chunk(arr):
         if chunks:
-            if not has_dask:
-                pytest.skip()
             return da.from_array(arr, chunks=chunks)
         else:
             return arr
@@ -739,7 +750,14 @@ def test_npg_nanarg_bug(func):
 )
 @pytest.mark.parametrize("method", ["cohorts", "map-reduce"])
 @pytest.mark.parametrize("chunk_labels", [False, True])
-@pytest.mark.parametrize("chunks", ((), (1,), (2,)))
+@pytest.mark.parametrize(
+    "chunks",
+    (
+        (),
+        pytest.param((1,), marks=requires_dask),
+        pytest.param((2,), marks=requires_dask),
+    ),
+)
 def test_groupby_bins(chunk_labels, kwargs, chunks, engine, method) -> None:
     array = [1, 1, 1, 1, 1, 1]
     labels = [0.2, 1.5, 1.9, 2, 3, 20]
@@ -748,8 +766,6 @@ def test_groupby_bins(chunk_labels, kwargs, chunks, engine, method) -> None:
         pytest.xfail()
 
     if chunks:
-        if not has_dask:
-            pytest.skip()
         array = dask.array.from_array(array, chunks=chunks)
         if chunk_labels:
             labels = dask.array.from_array(labels, chunks=chunks)
@@ -825,16 +841,14 @@ def test_rechunk_for_cohorts(chunk_at, expected):
     assert rechunked.chunks == expected
 
 
-@pytest.mark.parametrize("chunks", [None, 3])
+@pytest.mark.parametrize("chunks", [None, pytest.param(3, marks=requires_dask)])
 @pytest.mark.parametrize("fill_value", [123, np.nan])
 @pytest.mark.parametrize("func", ALL_FUNCS)
 def test_fill_value_behaviour(func, chunks, fill_value, engine):
     # fill_value = np.nan tests promotion of int counts to float
     # This is used by xarray
     if func in ["all", "any"] or "arg" in func:
         pytest.skip()
-    if chunks is not None and not has_dask:
-        pytest.skip()
 
     npfunc = _get_array_func(func)
     by = np.array([1, 2, 3, 1, 2, 3])
@@ -1050,11 +1064,8 @@ def test_factorize_values_outside_bins():
     assert_equal(expected, actual)
 
 
-@pytest.mark.parametrize("chunk", [True, False])
+@pytest.mark.parametrize("chunk", [pytest.param(True, marks=requires_dask), False])
 def test_multiple_groupers_bins(chunk) -> None:
-    if chunk and not has_dask:
-        pytest.skip()
-
     xp = dask.array if chunk else np
     array_kwargs = {"chunks": 2} if chunk else {}
     array = xp.ones((5, 2), **array_kwargs, dtype=np.int64)
@@ -1087,9 +1098,9 @@ def test_multiple_groupers_bins(chunk) -> None:
         np.arange(2, 4).reshape(1, 2),
     ],
 )
-@pytest.mark.parametrize("chunk", [True, False])
+@pytest.mark.parametrize("chunk", [pytest.param(True, marks=requires_dask), False])
 def test_multiple_groupers(chunk, by1, by2, expected_groups) -> None:
-    if chunk and (not has_dask or expected_groups is None):
+    if chunk and expected_groups is None:
         pytest.skip()
 
     xp = dask.array if chunk else np
@@ -1233,7 +1244,7 @@ def test_dtype(func, dtype, engine):
         pytest.skip()
     arr = np.ones((4, 12), dtype=dtype)
     labels = np.array(["a", "a", "c", "c", "c", "b", "b", "c", "c", "b", "b", "f"])
-    actual, _ = groupby_reduce(arr, labels, func=func, dtype=np.float64)
+    actual, _ = groupby_reduce(arr, labels, func=func, dtype=np.float64, engine=engine)
     assert actual.dtype == np.dtype("float64")
 
 

tests/test_xarray.py

@@ -16,7 +16,7 @@
     dask.config.set(scheduler="sync")
 
 try:
-    # Should test against legacy xarray implementation
+    # test against legacy xarray implementation
     xr.set_options(use_flox=False)
 except ValueError:
     pass
@@ -31,15 +31,15 @@
 @pytest.mark.parametrize("add_nan", [True, False])
 @pytest.mark.parametrize("skipna", [True, False])
 def test_xarray_reduce(skipna, add_nan, min_count, engine, reindex):
+    if skipna is False and min_count is not None:
+        pytest.skip()
+
     arr = np.ones((4, 12))
 
     if add_nan:
         arr[1, ...] = np.nan
         arr[[0, 2], [3, 4]] = np.nan
 
-    if skipna is False and min_count is not None:
-        pytest.skip()
-
     labels = np.array(["a", "a", "c", "c", "c", "b", "b", "c", "c", "b", "b", "f"])
     labels = np.array(labels)
     labels2 = np.array([1, 2, 2, 1])
@@ -77,11 +77,8 @@ def test_xarray_reduce(skipna, add_nan, min_count, engine, reindex):
 
 # TODO: sort
 @pytest.mark.parametrize("pass_expected_groups", [True, False])
-@pytest.mark.parametrize("chunk", (True, False))
+@pytest.mark.parametrize("chunk", (pytest.param(True, marks=requires_dask), False))
 def test_xarray_reduce_multiple_groupers(pass_expected_groups, chunk, engine):
-    if not has_dask and chunk:
-        pytest.skip()
-
     if chunk and pass_expected_groups is False:
         pytest.skip()
 
@@ -126,11 +123,8 @@ def test_xarray_reduce_multiple_groupers(pass_expected_groups, chunk, engine):
 
 
 @pytest.mark.parametrize("pass_expected_groups", [True, False])
-@pytest.mark.parametrize("chunk", (True, False))
+@pytest.mark.parametrize("chunk", (pytest.param(True, marks=requires_dask), False))
 def test_xarray_reduce_multiple_groupers_2(pass_expected_groups, chunk, engine):
-    if not has_dask and chunk:
-        pytest.skip()
-
     if chunk and pass_expected_groups is False:
         pytest.skip()
 
@@ -317,14 +311,12 @@ def test_multi_index_groupby_sum(engine):
     assert_equal(expected, actual)
 
 
-@pytest.mark.parametrize("chunks", (None, 2))
+@pytest.mark.parametrize("chunks", (None, pytest.param(2, marks=requires_dask)))
 def test_xarray_groupby_bins(chunks, engine):
     array = xr.DataArray([1, 1, 1, 1, 1], dims="x")
     labels = xr.DataArray([1, 1.5, 1.9, 2, 3], dims="x", name="labels")
 
     if chunks:
-        if not has_dask:
-            pytest.skip()
         array = array.chunk({"x": chunks})
         labels = labels.chunk({"x": chunks})
 
@@ -472,11 +464,8 @@ def test_alignment_error():
 @pytest.mark.parametrize("add_nan", [True, False])
 @pytest.mark.parametrize("dtype_out", [np.float64, "float64", np.dtype("float64")])
 @pytest.mark.parametrize("dtype", [np.float32, np.float64])
-@pytest.mark.parametrize("chunk", (True, False))
+@pytest.mark.parametrize("chunk", (pytest.param(True, marks=requires_dask), False))
 def test_dtype(add_nan, chunk, dtype, dtype_out, engine):
-    if chunk and not has_dask:
-        pytest.skip()
-
     xp = dask.array if chunk else np
     data = xp.linspace(0, 1, 48, dtype=dtype).reshape((4, 12))
 
@@ -508,12 +497,9 @@ def test_dtype(add_nan, chunk, dtype, dtype_out, engine):
     xr.testing.assert_allclose(expected, actual.transpose("labels", ...), **tolerance64)
 
 
-@pytest.mark.parametrize("chunk", [True, False])
+@pytest.mark.parametrize("chunk", [pytest.param(True, marks=requires_dask), False])
 @pytest.mark.parametrize("use_flox", [True, False])
 def test_dtype_accumulation(use_flox, chunk):
-    if chunk and not has_dask:
-        pytest.skip()
-
     datetimes = pd.date_range("2010-01", "2015-01", freq="6H", inclusive="left")
     samples = 10 + np.cos(2 * np.pi * 0.001 * np.arange(len(datetimes))) * 1
     samples += np.random.randn(len(datetimes))