BUG: fix nout=2 out handling (divmod)

Author: ev-br

torch_np/_normalizations.py

@@ -80,7 +80,7 @@ def normalize_ndarray(arg, name=None):
     from ._ndarray import ndarray
 
     if not isinstance(arg, ndarray):
-        raise TypeError("'out' must be an array")
+        raise TypeError(f"'{name}' must be an array")
     return arg.tensor
 
 
@@ -135,8 +135,9 @@ def maybe_copy_to(out, result, promote_scalar_result=False):
         out.tensor.copy_(result)
         return out
     elif isinstance(result, (tuple, list)):
-        # FIXME: this is broken (there is no copy_to)
-        return type(result)(map(copy_to, zip(result, out)))
+        return type(result)(
+            maybe_copy_to(o, r, promote_scalar_result) for o, r in zip(out, result)
+        )
     else:
         assert False  # We should never hit this path
 

torch_np/_ufuncs.py

@@ -31,7 +31,7 @@ def _ufunc_postprocess(result, out, casting):
 _binary = [
     name
     for name in dir(_binary_ufuncs_impl)
-    if not name.startswith("_") and name not in ["torch", "matmul"]
+    if not name.startswith("_") and name not in ["torch", "matmul", "divmod"]
 ]
 
 
@@ -106,6 +106,7 @@ def matmul(
 #
 # nin=2, nout=2
 #
+@normalizer
 def divmod(
     x1: ArrayLike,
     x2: ArrayLike,
@@ -122,15 +123,19 @@ def divmod(
     signature=None,
     extobj=None,
 ):
-    num_outs = sum(x is None for x in [out1, out2])
-    if sum_outs == 1:
+    # make sure we either have no out arrays at all, or there is either
+    # out1, out2, or out=tuple, but not both
+    out1t, out2t = out
+    num_outs = sum(x is not None for x in [out1, out2])
+    if num_outs == 1:
         raise ValueError("both out1 and out2 need to be provided")
-    if sum_outs != 0 and out != (None, None):
-        raise ValueError("Either provide out1 and out2, or out.")
-    if out is not None:
+    else:
         out1, out2 = out
-    if out1.shape != out2.shape or out1.dtype != out2.dtype:
-        raise ValueError("out1, out2 must be compatible")
+        if num_outs == 2:
+            if out1 is not None or out2 is not None:
+                raise TypeError(
+                    "cannot specify 'out' as both a positional and keyword argument"
+                )
 
     tensors = _ufunc_preprocess(
         (x1, x2), True, casting, order, dtype, subok, signature, extobj

torch_np/tests/test_basic.py

@@ -7,6 +7,7 @@
 
 import torch_np as w
 import torch_np._ufuncs as _ufuncs
+from torch_np.testing import assert_equal
 
 # These function receive one array_like arg and return one array_like result
 one_arg_funcs = [
@@ -445,3 +446,45 @@ def test_typecast(self):
         # force the type cast
         w.copyto(dst, src, casting="unsafe")
         assert (dst == src).all()
+
+
+class TestDivmod:
+    def test_divmod_out(self):
+        x1 = w.arange(8, 15)
+        x2 = w.arange(4, 11)
+
+        out = (w.empty_like(x1), w.empty_like(x1))
+
+        quot, rem = w.divmod(x1, x2, out=out)
+
+        assert_equal(quot, x1 // x2)
+        assert_equal(rem, x1 % x2)
+
+        out1, out2 = out
+        assert quot is out[0]
+        assert rem is out[1]
+
+    def test_divmod_out_list(self):
+        x1 = [4, 5, 6]
+        x2 = [2, 1, 2]
+
+        out = (w.empty_like(x1), w.empty_like(x1))
+
+        quot, rem = w.divmod(x1, x2, out=out)
+
+        assert quot is out[0]
+        assert rem is out[1]
+
+    def test_divmod_no_out(self):
+        # check that the out= machinery handles no out at all
+        x1 = w.array([4, 5, 6])
+        x2 = w.array([2, 1, 2])
+        quot, rem = w.divmod(x1, x2)
+
+        assert_equal(quot, x1 // x2)
+        assert_equal(rem, x1 % x2)
+
+    def test_divmod_out_both_pos_and_kw(self):
+        o = w.empty(1)
+        with assert_raises(TypeError):
+            w.divmod(1, 2, o, o, out=(o, o))