ENH: Improve performance of numpy scalar __copy__ and __deepcopy__ (#29656)

eendebakpt · web-flow · commit 538654066cb0 · 2025-10-03T13:03:27.000+02:00
The numpy scalars are immutable (expect for subclasses of np.void), so copy or deepcopy can return the same object. This improves performance, since the current implementation converts to array, copies and converts back to scalar.
diff --git a/numpy/_core/src/multiarray/scalartypes.c.src b/numpy/_core/src/multiarray/scalartypes.c.src
@@ -2093,12 +2093,11 @@ gentype_wraparray(PyObject *NPY_UNUSED(scalar), PyObject *args)
 
 /*
  * These gentype_* functions do not take keyword arguments.
- * The proper flag is METH_VARARGS.
+ * The proper flag is METH_VARARGS or METH_NOARGS.
  */
 /**begin repeat
  *
- * #name = tolist, item, __deepcopy__, __copy__,
- *         swapaxes, conj, conjugate, nonzero,
+ * #name = tolist, item, swapaxes, conj, conjugate, nonzero,
  *         fill, transpose#
  */
 static PyObject *
@@ -2108,6 +2107,34 @@ gentype_@name@(PyObject *self, PyObject *args)
 }
 /**end repeat**/
 
+static PyObject *
+gentype___copy__(PyObject *self)
+{
+    // scalars are immutable, so we can return a new reference
+    // the only expections are scalars with void dtype
+    if (PyObject_IsInstance(self, (PyObject *)&PyVoidArrType_Type)) {
+        // path via array
+        return gentype_generic_method(self, NULL, NULL, "__copy__");
+    }
+    return Py_NewRef(self);
+}
+
+static PyObject *
+gentype___deepcopy__(PyObject *self, PyObject *args)
+{
+    // note: maybe the signature needs to be updated as __deepcopy__ can accept the keyword memo
+
+    // scalars are immutable, so we can return a new reference
+    // the only expections are scalars with void dtype
+    // if the number of arguments is not 1, we let gentype_generic_method do the
+    // error handling
+    if (PyObject_IsInstance(self, (PyObject *)&PyVoidArrType_Type) || (PyTuple_Size(args)!=1)) {
+        // path via array
+        return gentype_generic_method(self, args, NULL, "__deepcopy__");
+    }
+    return Py_NewRef(self);
+}
+
 static PyObject *
 gentype_byteswap(PyObject *self, PyObject *args, PyObject *kwds)
 {
@@ -2652,7 +2679,7 @@ static PyMethodDef gentype_methods[] = {
     /* for the copy module */
     {"__copy__",
         (PyCFunction)gentype___copy__,
-        METH_VARARGS, NULL},
+        METH_NOARGS, NULL},
     {"__deepcopy__",
         (PyCFunction)gentype___deepcopy__,
         METH_VARARGS, NULL},
diff --git a/numpy/_core/tests/test_multiarray.py b/numpy/_core/tests/test_multiarray.py
@@ -2476,6 +2476,20 @@ def test__deepcopy__(self, dtype):
         with pytest.raises(AssertionError):
             assert_array_equal(a, b)
 
+    def test__deepcopy___void_scalar(self):
+        # see comments in gh-29643
+        value = np.void('Rex', dtype=[('name', 'U10')])
+        value_deepcopy = value.__deepcopy__(None)
+        value[0] = None
+        assert value_deepcopy[0] == 'Rex'
+
+    @pytest.mark.parametrize("sctype", [np.int64, np.float32, np.float64])
+    def test__deepcopy__scalar(self, sctype):
+        # test optimization from gh-29656
+        value = sctype(1.1)
+        value_deepcopy = value.__deepcopy__(None)
+        assert value is value_deepcopy
+
     def test__deepcopy__catches_failure(self):
         class MyObj:
             def __deepcopy__(self, *args, **kwargs):
