Implemented dpnp.hstack() and dpnp.atleast_1d() functions.

dpnp/dpnp_iface_manipulation.py

@@ -117,13 +117,52 @@ def atleast_1d(*arys):
 
     For full documentation refer to :obj:`numpy.atleast_1d`.
 
-    Limitations
-    -----------
-    Input arrays is supported as :obj:`dpnp.ndarray`.
+    Parameters
+    ----------
+    arys : {dpnp_array, usm_ndarray}
+        One or more input arrays.
+
+    Returns
+    -------
+    out : dpnp.ndarray
+        An array, or list of arrays, each with ``a.ndim >= 1``.
+        Copies are made only if necessary.
+
+    See Also
+    --------
+    atleast_2d, atleast_3d
+
+    Examples
+    --------
+    >>> import dpnp as np
+    >>> np.atleast_1d(1.0)
+    array([1.])
+
+    >>> x = np.arange(9.0).reshape(3,3)
+    >>> np.atleast_1d(x)
+    array([[0., 1., 2.],
+           [3., 4., 5.],
+           [6., 7., 8.]])
+    >>> np.atleast_1d(x) is x
+    True
+
+    >>> np.atleast_1d(1, [3, 4])
+    [array([1]), array([3, 4])]
 
     """
 
-    return call_origin(numpy.atleast_1d, *arys)
+    res = []
+    for ary in arys:
+        ary = dpnp.asanyarray(ary)
+        if ary.ndim == 0:
+            result = ary.reshape(1)
+        else:
+            result = ary
+        res.append(result)
+    if len(res) == 1:
+        return res[0]
+    else:
+        return res
 
 
 def atleast_2d(*arys):
@@ -242,7 +281,9 @@ def broadcast_to(array, /, shape, subok=False):
     return call_origin(numpy.broadcast_to, array, shape=shape, subok=subok)
 
 
-def concatenate(arrays, /, *, axis=0, out=None, dtype=None, **kwargs):
+def concatenate(
+    arrays, /, *, axis=0, out=None, dtype=None, casting="same_kind", **kwargs
+):
     """
     Join a sequence of arrays along an existing axis.
 
@@ -299,6 +340,8 @@ def concatenate(arrays, /, *, axis=0, out=None, dtype=None, **kwargs):
         pass
     elif dtype is not None:
         pass
+    elif casting != "same_kind":
+        pass
     else:
         usm_arrays = [dpnp.get_usm_ndarray(x) for x in arrays]
         usm_res = dpt.concat(usm_arrays, axis=axis)
@@ -310,6 +353,7 @@ def concatenate(arrays, /, *, axis=0, out=None, dtype=None, **kwargs):
         axis=axis,
         out=out,
         dtype=dtype,
+        casting=casting,
         **kwargs,
     )
 
@@ -659,23 +703,62 @@ def flipud(m):
     return m[::-1, ...]
 
 
-def hstack(tup):
+def hstack(tup, *, dtype=None, casting="same_kind"):
     """
     Stack arrays in sequence horizontally (column wise).
 
     For full documentation refer to :obj:`numpy.hstack`.
 
-    """
+    Returns
+    -------
+    out : dpnp.ndarray
+        The stacked array which has one more dimension than the input arrays.
 
-    # TODO:
-    # `call_origin` cannot convert sequence of array to sequence of
-    # nparrays
-    tup_new = []
-    for tp in tup:
-        tpx = dpnp.asnumpy(tp) if not isinstance(tp, numpy.ndarray) else tp
-        tup_new.append(tpx)
+    Limitations
+    -----------
+    Each array in `tup` is supported as either :class:`dpnp.ndarray`
+    or :class:`dpctl.tensor.usm_ndarray`. Otherwise ``TypeError`` exception
+    will be raised.
+    Parameters `dtype` and `casting` are supported with default value.
+    Otherwise the function will be executed sequentially on CPU.
+
+    See Also
+    --------
+    :obj:`dpnp.concatenate` : Join a sequence of arrays along an existing axis.
+    :obj:`dpnp.stack` : Join a sequence of arrays along a new axis.
+    :obj:`dpnp.vstack` : Stack arrays in sequence vertically (row wise).
+    :obj:`dpnp.block` : Assemble an nd-array from nested lists of blocks.
+    :obj:`dpnp.split` : Split array into a list of multiple sub-arrays of equal size.
 
-    return call_origin(numpy.hstack, tup_new)
+    Examples
+    --------
+    >>> import dpnp as np
+    >>> a = np.array((1,2,3))
+    >>> b = np.array((4,5,6))
+    >>> np.hstack((a,b))
+    array([1, 2, 3, 4, 5, 6])
+
+    >>> a = np.array([[1],[2],[3]])
+    >>> b = np.array([[4],[5],[6]])
+    >>> np.hstack((a,b))
+    array([[1, 4],
+           [2, 5],
+           [3, 6]])
+
+    """
+
+    if not hasattr(tup, "__getitem__"):
+        raise TypeError(
+            "Arrays to stack must be passed as a sequence type such as list or tuple."
+        )
+    arrs = dpnp.atleast_1d(*tup)
+    if not isinstance(arrs, list):
+        arrs = [arrs]
+    # As a special case, dimension 0 of 1-dimensional arrays is "horizontal"
+    if arrs and arrs[0].ndim == 1:
+        return dpnp.concatenate(arrs, axis=0, dtype=dtype, casting=casting)
+    else:
+        return dpnp.concatenate(arrs, axis=1, dtype=dtype, casting=casting)
 
 
 def moveaxis(a, source, destination):
@@ -1131,7 +1214,7 @@ def stack(arrays, /, *, axis=0, out=None, dtype=None, **kwargs):
     Each array in `arrays` is supported as either :class:`dpnp.ndarray`
     or :class:`dpctl.tensor.usm_ndarray`. Otherwise ``TypeError`` exception
     will be raised.
-    Parameters `out` and `dtype are supported with default value.
+    Parameters `out` and `dtype` are supported with default value.
     Keyword argument ``kwargs`` is currently unsupported.
     Otherwise the function will be executed sequentially on CPU.
 

tests/skipped_tests.tbl

@@ -34,7 +34,6 @@ tests/third_party/cupy/fft_tests/test_fft.py::TestFftn_param_23_{axes=None, norm
 
 tests/third_party/intel/test_zero_copy_test1.py::test_dpnp_interaction_with_dpctl_memory
 
-tests/test_arraymanipulation.py::TestHstack::test_generator
 tests/test_arraymanipulation.py::TestVstack::test_generator
 
 tests/test_linalg.py::test_cond[-1-[[1, 2, 3], [4, 5, 6], [7, 8, 9]]]

tests/skipped_tests_gpu.tbl

@@ -216,7 +216,6 @@ tests/third_party/cupy/random_tests/test_distributions.py::TestDistributionsMult
 tests/third_party/cupy/random_tests/test_distributions.py::TestDistributionsMultivariateNormal_param_3_{d=4, shape=(3, 2)}::test_normal
 
 tests/third_party/intel/test_zero_copy_test1.py::test_dpnp_interaction_with_dpctl_memory
-tests/test_arraymanipulation.py::TestHstack::test_generator
 tests/test_arraymanipulation.py::TestVstack::test_generator
 
 tests/test_linalg.py::test_cond[-1-[[1, 2, 3], [4, 5, 6], [7, 8, 9]]]

tests/test_arraymanipulation.py

@@ -335,27 +335,23 @@ class TestHstack:
     def test_non_iterable(self):
         assert_raises(TypeError, dpnp.hstack, 1)
 
-    @pytest.mark.usefixtures("allow_fall_back_on_numpy")
     def test_empty_input(self):
-        assert_raises(ValueError, dpnp.hstack, ())
+        assert_raises(TypeError, dpnp.hstack, ())
 
-    @pytest.mark.usefixtures("allow_fall_back_on_numpy")
     def test_0D_array(self):
         b = dpnp.array(2)
         a = dpnp.array(1)
         res = dpnp.hstack([a, b])
         desired = dpnp.array([1, 2])
         assert_array_equal(res, desired)
 
-    @pytest.mark.usefixtures("allow_fall_back_on_numpy")
     def test_1D_array(self):
         a = dpnp.array([1])
         b = dpnp.array([2])
         res = dpnp.hstack([a, b])
         desired = dpnp.array([1, 2])
         assert_array_equal(res, desired)
 
-    @pytest.mark.usefixtures("allow_fall_back_on_numpy")
     def test_2D_array(self):
         a = dpnp.array([[1], [2]])
         b = dpnp.array([[1], [2]])
@@ -364,9 +360,9 @@ def test_2D_array(self):
         assert_array_equal(res, desired)
 
     def test_generator(self):
-        with assert_warns(FutureWarning):
+        with pytest.raises(TypeError):
             dpnp.hstack((numpy.arange(3) for _ in range(2)))
-        with assert_warns(FutureWarning):
+        with pytest.raises(TypeError):
             dpnp.hstack(map(lambda x: x, numpy.ones((3, 2))))
 
 
@@ -607,6 +603,44 @@ def test_generator(self):
             dpnp.vstack((numpy.arange(3) for _ in range(2)))
 
 
+class TestAtleast1d:
+    def test_0D_array(self):
+        a = dpnp.array(1)
+        b = dpnp.array(2)
+        res = [dpnp.atleast_1d(a), dpnp.atleast_1d(b)]
+        desired = [dpnp.array([1]), dpnp.array([2])]
+        assert_array_equal(res, desired)
+
+    def test_1D_array(self):
+        a = dpnp.array([1, 2])
+        b = dpnp.array([2, 3])
+        res = [dpnp.atleast_1d(a), dpnp.atleast_1d(b)]
+        desired = [dpnp.array([1, 2]), dpnp.array([2, 3])]
+        assert_array_equal(res, desired)
+
+    def test_2D_array(self):
+        a = dpnp.array([[1, 2], [1, 2]])
+        b = dpnp.array([[2, 3], [2, 3]])
+        res = [dpnp.atleast_1d(a), dpnp.atleast_1d(b)]
+        desired = [a, b]
+        assert_array_equal(res, desired)
+
+    def test_3D_array(self):
+        a = dpnp.array([[1, 2], [1, 2]])
+        b = dpnp.array([[2, 3], [2, 3]])
+        a = dpnp.array([a, a])
+        b = dpnp.array([b, b])
+        res = [dpnp.atleast_1d(a), dpnp.atleast_1d(b)]
+        desired = [a, b]
+        assert_array_equal(res, desired)
+
+    def test_r1array(self):
+        assert dpnp.atleast_1d(3).shape == (1,)
+        assert dpnp.atleast_1d(3j).shape == (1,)
+        assert dpnp.atleast_1d(3.0).shape == (1,)
+        assert dpnp.atleast_1d([[2, 3], [4, 5]]).shape == (2, 2)
+
+
 class TestRollaxis:
     data = [
         (0, 0),

tests/third_party/cupy/manipulation_tests/test_dims.py

@@ -18,12 +18,10 @@ def check_atleast(self, func, xp):
         f = numpy.float32(1)
         return func(a, b, c, d, e, f)
 
-    @pytest.mark.usefixtures("allow_fall_back_on_numpy")
     @testing.numpy_cupy_array_equal()
     def test_atleast_1d1(self, xp):
         return self.check_atleast(xp.atleast_1d, xp)
 
-    @pytest.mark.usefixtures("allow_fall_back_on_numpy")
     @testing.numpy_cupy_array_equal()
     def test_atleast_1d2(self, xp):
         a = testing.shaped_arange((1, 3, 2), xp)

tests/third_party/cupy/manipulation_tests/test_join.py

@@ -270,30 +270,27 @@ def test_dstack_single_element_3(self, xp):
         a = testing.shaped_arange((1,), xp)
         return xp.dstack((a,))
 
-    @pytest.mark.skip("dpnp.hstack() is not implemented yet")
     @testing.numpy_cupy_array_equal()
     def test_hstack_vectors(self, xp):
         a = xp.arange(3)
         b = xp.arange(2, -1, -1)
         return xp.hstack((a, b))
 
-    @pytest.mark.skip("dpnp.hstack() is not implemented yet")
     @testing.numpy_cupy_array_equal()
     def test_hstack_scalars(self, xp):
         a = testing.shaped_arange((), xp)
         b = testing.shaped_arange((), xp)
         c = testing.shaped_arange((), xp)
         return xp.hstack((a, b, c))
 
-    @pytest.mark.skip("dpnp.hstack() is not implemented yet")
     @testing.numpy_cupy_array_equal()
     def test_hstack(self, xp):
         a = testing.shaped_arange((2, 1), xp)
         b = testing.shaped_arange((2, 2), xp)
         c = testing.shaped_arange((2, 3), xp)
         return xp.hstack((a, b, c))
 
-    @pytest.mark.skip("dpnp.hstack() is not implemented yet")
+    @pytest.mark.usefixtures("allow_fall_back_on_numpy")
     @testing.with_requires("numpy>=1.24.0")
     @testing.for_all_dtypes_combination(names=["dtype1", "dtype2"])
     @testing.numpy_cupy_array_equal(accept_error=TypeError)
@@ -302,18 +299,9 @@ def test_hstack_dtype(self, xp, dtype1, dtype2):
         b = testing.shaped_arange((3, 4), xp, dtype1)
         return xp.hstack((a, b), dtype=dtype2)
 
-    @pytest.mark.skip("dpnp.hstack() is not implemented yet")
+    @pytest.mark.usefixtures("allow_fall_back_on_numpy")
     @testing.with_requires("numpy>=1.24.0")
-    @pytest.mark.parametrize(
-        "casting",
-        [
-            "no",
-            "equiv",
-            "safe",
-            "same_kind",
-            "unsafe",
-        ],
-    )
+    @testing.for_castings()
     @testing.for_all_dtypes_combination(names=["dtype1", "dtype2"])
     @testing.numpy_cupy_array_equal(
         accept_error=(TypeError, numpy.ComplexWarning)