Add NumPy Scalars

docs/advanced/pycpp/numpy.rst

@@ -232,6 +232,46 @@ prevent many types of unsupported structures, it is still the user's
 responsibility to use only "plain" structures that can be safely manipulated as
 raw memory without violating invariants.
 
+Scalar types
+============
+
+In some cases we may want to accept or return NumPy scalar values such as
+``np.float32`` or ``np.float64``. We hope to be able to handle single-precision
+and double-precision on the C-side. However, both are bound to Python's
+double-precision builtin float by default, so they cannot be processed separately.
+We used the ``py::buffer`` trick to implement the previous approach, which
+will cause the readability of the code to drop significantly.
+
+Luckily, there's a helper type for this occasion - ``py::numpy_scalar``:
+
+.. code-block:: cpp
+
+    m.def("add", [](py::numpy_scalar<float> a, py::numpy_scalar<float> b) {
+        return py::make_scalar(a + b);
+    });
+    m.def("add", [](py::numpy_scalar<double> a, py::numpy_scalar<double> b) {
+        return py::make_scalar(a + b);
+    });
+
+This type is trivially convertible to and from the type it wraps; currently
+supported scalar types are NumPy arithmetic types: ``bool_``, ``int8``,
+``int16``, ``int32``, ``int64``, ``uint8``, ``uint16``, ``uint32``,
+``uint64``, ``float32``, ``float64``, ``complex64``, ``complex128``, all of
+them mapping to respective C++ counterparts.
+
+.. note::
+
+    This is a strict type, it will only allows to specify NumPy type as input
+    arguments, and does not allow other types of input parameters (e.g.,
+    ``py::numpy_scalar<int64_t>`` will not accept Python's builtin ``int`` ).
+
+.. note::
+
+    Native C types are mapped to NumPy types in a platform specific way: for
+    instance, ``char`` may be mapped to either ``np.int8`` or ``np.uint8``
+    and ``long`` may use 4 or 8 bytes depending on the platform. Unless you
+    clearly understand the difference and your needs, please use ``<cstdint>``.
+
 Vectorizing functions
 =====================
 

include/pybind11/numpy.h

@@ -49,6 +49,9 @@ PYBIND11_WARNING_DISABLE_MSVC(4127)
 class dtype; // Forward declaration
 class array; // Forward declaration
 
+template <typename>
+struct numpy_scalar; // Forward declaration
+
 PYBIND11_NAMESPACE_BEGIN(detail)
 
 template <>
@@ -200,15 +203,15 @@ struct same_size {
     using as = bool_constant<sizeof(T) == sizeof(U)>;
 };
 
-template <typename Concrete>
+template <std::size_t>
 constexpr int platform_lookup() {
     return -1;
 }
 
 // Lookup a type according to its size, and return a value corresponding to the NumPy typenum.
-template <typename Concrete, typename T, typename... Ts, typename... Ints>
+template <std::size_t size, typename T, typename... Ts, typename... Ints>
 constexpr int platform_lookup(int I, Ints... Is) {
-    return sizeof(Concrete) == sizeof(T) ? I : platform_lookup<Concrete, Ts...>(Is...);
+    return sizeof(size) == sizeof(T) ? I : platform_lookup<size, Ts...>(Is...);
 }
 
 struct npy_api {
@@ -249,15 +252,23 @@ struct npy_api {
         // `npy_common.h` defines the integer aliases. In order, it checks:
         // NPY_BITSOF_LONG, NPY_BITSOF_LONGLONG, NPY_BITSOF_INT, NPY_BITSOF_SHORT, NPY_BITSOF_CHAR
         // and assigns the alias to the first matching size, so we should check in this order.
-        NPY_INT32_
-        = platform_lookup<std::int32_t, long, int, short>(NPY_LONG_, NPY_INT_, NPY_SHORT_),
-        NPY_UINT32_ = platform_lookup<std::uint32_t, unsigned long, unsigned int, unsigned short>(
+        NPY_INT32_ = platform_lookup<4, long, int, short>(NPY_LONG_, NPY_INT_, NPY_SHORT_),
+        NPY_UINT32_ = platform_lookup<4, unsigned long, unsigned int, unsigned short>(
             NPY_ULONG_, NPY_UINT_, NPY_USHORT_),
-        NPY_INT64_
-        = platform_lookup<std::int64_t, long, long long, int>(NPY_LONG_, NPY_LONGLONG_, NPY_INT_),
-        NPY_UINT64_
-        = platform_lookup<std::uint64_t, unsigned long, unsigned long long, unsigned int>(
+        NPY_INT64_ = platform_lookup<8, long, long long, int>(NPY_LONG_, NPY_LONGLONG_, NPY_INT_),
+        NPY_UINT64_ = platform_lookup<8, unsigned long, unsigned long long, unsigned int>(
             NPY_ULONG_, NPY_ULONGLONG_, NPY_UINT_),
+        NPY_FLOAT32_
+        = platform_lookup<4, double, float, long double>(NPY_DOUBLE_, NPY_FLOAT_, NPY_LONGDOUBLE_),
+        NPY_FLOAT64_
+        = platform_lookup<8, double, float, long double>(NPY_DOUBLE_, NPY_FLOAT_, NPY_LONGDOUBLE_),
+        NPY_COMPLEX64_
+        = platform_lookup<8, std::complex<double>, std::complex<float>, std::complex<long double>>(
+            NPY_DOUBLE_, NPY_FLOAT_, NPY_LONGDOUBLE_),
+        NPY_COMPLEX128_
+        = platform_lookup<8, std::complex<double>, std::complex<float>, std::complex<long double>>(
+            NPY_DOUBLE_, NPY_FLOAT_, NPY_LONGDOUBLE_),
+        NPY_CHAR_ = std::is_signed<char>::value ? NPY_BYTE_ : NPY_UBYTE_,
     };
 
     unsigned int PyArray_RUNTIME_VERSION_;
@@ -281,6 +292,7 @@ struct npy_api {
 
     unsigned int (*PyArray_GetNDArrayCFeatureVersion_)();
     PyObject *(*PyArray_DescrFromType_)(int);
+    PyObject *(*PyArray_TypeObjectFromType_)(int);
     PyObject *(*PyArray_NewFromDescr_)(PyTypeObject *,
                                        PyObject *,
                                        int,
@@ -297,6 +309,8 @@ struct npy_api {
     PyTypeObject *PyVoidArrType_Type_;
     PyTypeObject *PyArrayDescr_Type_;
     PyObject *(*PyArray_DescrFromScalar_)(PyObject *);
+    PyObject *(*PyArray_Scalar_)(void *, PyObject *, PyObject *);
+    void (*PyArray_ScalarAsCtype_)(PyObject *, void *);
     PyObject *(*PyArray_FromAny_)(PyObject *, PyObject *, int, int, int, PyObject *);
     int (*PyArray_DescrConverter_)(PyObject *, PyObject **);
     bool (*PyArray_EquivTypes_)(PyObject *, PyObject *);
@@ -324,7 +338,10 @@ struct npy_api {
         API_PyArrayDescr_Type = 3,
         API_PyVoidArrType_Type = 39,
         API_PyArray_DescrFromType = 45,
+        API_PyArray_TypeObjectFromType = 46,
         API_PyArray_DescrFromScalar = 57,
+        API_PyArray_Scalar = 60,
+        API_PyArray_ScalarAsCtype = 62,
         API_PyArray_FromAny = 69,
         API_PyArray_Resize = 80,
         // CopyInto was slot 82 and 50 was effectively an alias. NumPy 2 removed 82.
@@ -362,7 +379,10 @@ struct npy_api {
         DECL_NPY_API(PyVoidArrType_Type);
         DECL_NPY_API(PyArrayDescr_Type);
         DECL_NPY_API(PyArray_DescrFromType);
+        DECL_NPY_API(PyArray_TypeObjectFromType);
         DECL_NPY_API(PyArray_DescrFromScalar);
+        DECL_NPY_API(PyArray_Scalar);
+        DECL_NPY_API(PyArray_ScalarAsCtype);
         DECL_NPY_API(PyArray_FromAny);
         DECL_NPY_API(PyArray_Resize);
         DECL_NPY_API(PyArray_CopyInto);
@@ -384,6 +404,88 @@ struct npy_api {
     }
 };
 
+template <typename T>
+struct is_complex : std::false_type {};
+template <typename T>
+struct is_complex<std::complex<T>> : std::true_type {};
+
+template <typename T, typename = void>
+struct npy_format_descriptor_name;
+
+template <typename T>
+struct npy_format_descriptor_name<T, enable_if_t<std::is_integral<T>::value>> {
+    static constexpr auto name = const_name<std::is_same<T, bool>::value>(
+        const_name("bool"),
+        const_name<std::is_signed<T>::value>("int", "uint") + const_name<sizeof(T) * 8>());
+};
+
+template <typename T>
+struct npy_format_descriptor_name<T, enable_if_t<std::is_floating_point<T>::value>> {
+    static constexpr auto name
+        = const_name < std::is_same<T, float>::value
+          || std::is_same<T, double>::value
+                 > (const_name("float") + const_name<sizeof(T) * 8>(), const_name("longdouble"));
+};
+
+template <typename T>
+struct npy_format_descriptor_name<T, enable_if_t<is_complex<T>::value>> {
+    static constexpr auto name
+        = const_name < std::is_same<typename T::value_type, float>::value
+          || std::is_same<typename T::value_type, double>::value
+                 > (const_name("complex") + const_name<sizeof(typename T::value_type) * 16>(),
+                    const_name("longcomplex"));
+};
+
+template <typename T>
+struct numpy_scalar_info {};
+
+#define DECL_NPY_SCALAR(ctype_, typenum_)                                                         \
+    template <>                                                                                   \
+    struct numpy_scalar_info<ctype_> {                                                            \
+        static constexpr auto name = npy_format_descriptor_name<ctype_>::name;                    \
+        static constexpr int typenum = npy_api::typenum_##_;                                      \
+    }
+
+// boolean type
+DECL_NPY_SCALAR(bool, NPY_BOOL);
+
+// character types
+DECL_NPY_SCALAR(char, NPY_CHAR);
+DECL_NPY_SCALAR(signed char, NPY_BYTE);
+DECL_NPY_SCALAR(unsigned char, NPY_UBYTE);
+
+// signed integer types
+DECL_NPY_SCALAR(std::int16_t, NPY_SHORT);
+DECL_NPY_SCALAR(std::int32_t, NPY_INT);
+DECL_NPY_SCALAR(std::int64_t, NPY_LONG);
+#if defined(__linux__)
+DECL_NPY_SCALAR(long long, NPY_LONG);
+#else
+DECL_NPY_SCALAR(long, NPY_LONG);
+#endif
+
+// unsigned integer types
+DECL_NPY_SCALAR(std::uint16_t, NPY_USHORT);
+DECL_NPY_SCALAR(std::uint32_t, NPY_UINT);
+DECL_NPY_SCALAR(std::uint64_t, NPY_ULONG);
+#if defined(__linux__)
+DECL_NPY_SCALAR(unsigned long long, NPY_ULONG);
+#else
+DECL_NPY_SCALAR(unsigned long, NPY_ULONG);
+#endif
+
+// floating point types
+DECL_NPY_SCALAR(float, NPY_FLOAT);
+DECL_NPY_SCALAR(double, NPY_DOUBLE);
+DECL_NPY_SCALAR(long double, NPY_LONGDOUBLE);
+
+// complex types
+DECL_NPY_SCALAR(std::complex<float>, NPY_CFLOAT);
+DECL_NPY_SCALAR(std::complex<double>, NPY_CDOUBLE);
+DECL_NPY_SCALAR(std::complex<long double>, NPY_CLONGDOUBLE);
+
+#undef DECL_NPY_SCALAR
+
 inline PyArray_Proxy *array_proxy(void *ptr) { return reinterpret_cast<PyArray_Proxy *>(ptr); }
 
 inline const PyArray_Proxy *array_proxy(const void *ptr) {
@@ -414,10 +516,6 @@ template <typename T>
 struct is_std_array : std::false_type {};
 template <typename T, size_t N>
 struct is_std_array<std::array<T, N>> : std::true_type {};
-template <typename T>
-struct is_complex : std::false_type {};
-template <typename T>
-struct is_complex<std::complex<T>> : std::true_type {};
 
 template <typename T>
 struct array_info_scalar {
@@ -631,8 +729,59 @@ template <typename T, ssize_t Dim>
 struct type_caster<unchecked_mutable_reference<T, Dim>>
     : type_caster<unchecked_reference<T, Dim>> {};
 
+template <typename T>
+struct type_caster<numpy_scalar<T>> {
+    using value_type = T;
+    using type_info = numpy_scalar_info<T>;
+
+    PYBIND11_TYPE_CASTER(numpy_scalar<T>, type_info::name);
+
+    static handle &target_type() {
+        static handle tp = npy_api::get().PyArray_TypeObjectFromType_(type_info::typenum);
+        return tp;
+    }
+
+    static handle &target_dtype() {
+        static handle tp = npy_api::get().PyArray_DescrFromType_(type_info::typenum);
+        return tp;
+    }
+
+    bool load(handle src, bool) {
+        if (isinstance(src, target_type())) {
+            npy_api::get().PyArray_ScalarAsCtype_(src.ptr(), &value.value);
+            return true;
+        }
+        return false;
+    }
+
+    static handle cast(numpy_scalar<T> src, return_value_policy, handle) {
+        return npy_api::get().PyArray_Scalar_(&src.value, target_dtype().ptr(), nullptr);
+    }
+};
+
 PYBIND11_NAMESPACE_END(detail)
 
+template <typename T>
+struct numpy_scalar {
+    using value_type = T;
+
+    value_type value;
+
+    numpy_scalar() = default;
+    numpy_scalar(value_type value) : value(value) {}
+
+    operator value_type() { return value; }
+    numpy_scalar &operator=(value_type value) {
+        this->value = value;
+        return *this;
+    }
+};
+
+template <typename T>
+numpy_scalar<T> make_scalar(T value) {
+    return numpy_scalar<T>(value);
+}
+
 class dtype : public object {
 public:
     PYBIND11_OBJECT_DEFAULT(dtype, object, detail::npy_api::get().PyArrayDescr_Check_)
@@ -1362,38 +1511,6 @@ struct compare_buffer_info<T, detail::enable_if_t<detail::is_pod_struct<T>::valu
     }
 };
 
-template <typename T, typename = void>
-struct npy_format_descriptor_name;
-
-template <typename T>
-struct npy_format_descriptor_name<T, enable_if_t<std::is_integral<T>::value>> {
-    static constexpr auto name = const_name<std::is_same<T, bool>::value>(
-        const_name("bool"),
-        const_name<std::is_signed<T>::value>("numpy.int", "numpy.uint")
-            + const_name<sizeof(T) * 8>());
-};
-
-template <typename T>
-struct npy_format_descriptor_name<T, enable_if_t<std::is_floating_point<T>::value>> {
-    static constexpr auto name = const_name < std::is_same<T, float>::value
-                                 || std::is_same<T, const float>::value
-                                 || std::is_same<T, double>::value
-                                 || std::is_same<T, const double>::value
-                                        > (const_name("numpy.float") + const_name<sizeof(T) * 8>(),
-                                           const_name("numpy.longdouble"));
-};
-
-template <typename T>
-struct npy_format_descriptor_name<T, enable_if_t<is_complex<T>::value>> {
-    static constexpr auto name = const_name < std::is_same<typename T::value_type, float>::value
-                                 || std::is_same<typename T::value_type, const float>::value
-                                 || std::is_same<typename T::value_type, double>::value
-                                 || std::is_same<typename T::value_type, const double>::value
-                                        > (const_name("numpy.complex")
-                                               + const_name<sizeof(typename T::value_type) * 16>(),
-                                           const_name("numpy.longcomplex"));
-};
-
 template <typename T>
 struct npy_format_descriptor<
     T,

tests/CMakeLists.txt

@@ -140,6 +140,7 @@ set(PYBIND11_TEST_FILES
     test_multiple_inheritance
     test_numpy_array
     test_numpy_dtypes
+    test_numpy_scalars
     test_numpy_vectorize
     test_opaque_types
     test_operator_overloading