gh-89188: Implement PyUnicode_KIND() as a function (#129412)

Implement PyUnicode_KIND() and PyUnicode_DATA() as function, in
addition to the macros with the same names. The macros rely on C bit
fields which have compiler-specific layout.

diff --git a/Include/cpython/unicodeobject.h b/Include/cpython/unicodeobject.h
index 287de52b96202c3b800d02ec70654d84033c9c57..cea69dd12809990c9c5534a5cac8e35523c19b57 100644 (file)
--- a/Include/cpython/unicodeobject.h
+++ b/Include/cpython/unicodeobject.h
@@ -240,6 +240,8 @@ enum PyUnicode_Kind {
     PyUnicode_4BYTE_KIND = 4
 };
 
+PyAPI_FUNC(int) PyUnicode_KIND(PyObject *op);
+
 // PyUnicode_KIND(): Return one of the PyUnicode_*_KIND values defined above.
 //
 // gh-89653: Converting this macro to a static inline function would introduce
@@ -264,13 +266,15 @@ static inline void* _PyUnicode_NONCOMPACT_DATA(PyObject *op) {
     return data;
 }
 
-static inline void* PyUnicode_DATA(PyObject *op) {
+PyAPI_FUNC(void*) PyUnicode_DATA(PyObject *op);
+
+static inline void* _PyUnicode_DATA(PyObject *op) {
     if (PyUnicode_IS_COMPACT(op)) {
         return _PyUnicode_COMPACT_DATA(op);
     }
     return _PyUnicode_NONCOMPACT_DATA(op);
 }
-#define PyUnicode_DATA(op) PyUnicode_DATA(_PyObject_CAST(op))
+#define PyUnicode_DATA(op) _PyUnicode_DATA(_PyObject_CAST(op))
 
 /* Return pointers to the canonical representation cast to unsigned char,
    Py_UCS2, or Py_UCS4 for direct character access.

diff --git a/Misc/NEWS.d/next/C_API/2025-01-29-11-58-38.gh-issue-89188.BsfLr3.rst b/Misc/NEWS.d/next/C_API/2025-01-29-11-58-38.gh-issue-89188.BsfLr3.rst
new file mode 100644 (file)
index 0000000..7ff225a
--- /dev/null
+++ b/Misc/NEWS.d/next/C_API/2025-01-29-11-58-38.gh-issue-89188.BsfLr3.rst
@@ -0,0 +1,3 @@
+Implement :c:func:`PyUnicode_KIND` and :c:func:`PyUnicode_DATA` as function,
+in addition to the macros with the same names. The macros rely on C bit
+fields which have compiler-specific layout. Patch by Victor Stinner.

diff --git a/Objects/unicodeobject.c b/Objects/unicodeobject.c
index d9952f764bb178a41de7340c45d699a881a4534c..c6f13f60ad741fc4f7e8c494493298bb87c64deb 100644 (file)
--- a/Objects/unicodeobject.c
+++ b/Objects/unicodeobject.c
@@ -16486,3 +16486,24 @@ PyInit__string(void)
 {
     return PyModuleDef_Init(&_string_module);
 }
+
+
+#undef PyUnicode_KIND
+int PyUnicode_KIND(PyObject *op)
+{
+    if (!PyUnicode_Check(op)) {
+        PyErr_Format(PyExc_TypeError, "expect str, got %T", op);
+        return -1;
+    }
+    return _PyASCIIObject_CAST(op)->state.kind;
+}
+
+#undef PyUnicode_DATA
+void* PyUnicode_DATA(PyObject *op)
+{
+    if (!PyUnicode_Check(op)) {
+        PyErr_Format(PyExc_TypeError, "expect str, got %T", op);
+        return NULL;
+    }
+    return _PyUnicode_DATA(op);
+}