WIP get string buffer encoding on demand

Include/cpython/unicodeobject.h

@@ -695,6 +695,7 @@ static inline int Py_UNICODE_ISALNUM(Py_UCS4 ch) {
            || Py_UNICODE_ISNUMERIC(ch));
 }
 
+PyAPI_FUNC(int) PyUnicode_GetBuffer(PyObject *unicode, Py_buffer *view, int kind);
 
 /* === Misc functions ===================================================== */
 

Modules/_testcapi/unicode.c

@@ -1999,6 +1999,25 @@ test_string_from_format(PyObject *self, PyObject *Py_UNUSED(ignored))
 #undef CHECK_FORMAT_0
 }
 
+static PyObject *
+unicode_get_buffer(PyObject *module, PyObject *args)
+{
+    PyObject *unicode;
+    int kind;
+
+    if (!PyArg_ParseTuple(args, "Oi:unicode_get_buffer", &unicode, &kind)) {
+        return NULL;
+    }
+
+    Py_buffer view;
+
+    if (PyUnicode_GetBuffer(unicode, &view, kind) < 0) {
+        return NULL;
+    }
+
+    return PyMemoryView_FromBuffer(&view);
+}
+
 static PyMethodDef TestMethods[] = {
     {"codec_incrementalencoder", codec_incrementalencoder,       METH_VARARGS},
     {"codec_incrementaldecoder", codec_incrementaldecoder,       METH_VARARGS},
@@ -2093,6 +2112,7 @@ static PyMethodDef TestMethods[] = {
     {"unicode_contains",         unicode_contains,               METH_VARARGS},
     {"unicode_isidentifier",     unicode_isidentifier,           METH_O},
     {"unicode_copycharacters",   unicode_copycharacters,         METH_VARARGS},
+    {"unicode_get_buffer",       unicode_get_buffer,             METH_VARARGS},
     {NULL},
 };
 

Objects/unicodeobject.c

@@ -14705,6 +14705,79 @@ _PyUnicode_ExactDealloc(PyObject *op)
     unicode_dealloc(op);
 }
 
+int PyUnicode_GetBuffer(PyObject *unicode, Py_buffer *view, int kind)
+{
+    int itemsize;
+    view->obj = NULL;
+    void *data = PyUnicode_DATA(unicode);
+    bool needs_dealloc = 0;
+
+    if (kind == PyUnicode_KIND(unicode)) {
+        switch (PyUnicode_KIND(unicode)) {
+            case PyUnicode_1BYTE_KIND:
+                itemsize = 1;
+                break;
+            case PyUnicode_2BYTE_KIND:
+                itemsize = 2;
+                break;
+            case PyUnicode_4BYTE_KIND:
+                itemsize = 4;
+                break;
+            default:
+                PyErr_BadInternalCall();
+                return -1;
+        }
+    } else if (kind == PyUnicode_4BYTE_KIND) {
+        needs_dealloc = 1;
+        itemsize = 4;
+        data = PyUnicode_AsUCS4Copy(unicode);
+        if (data == NULL) {
+            return -1;
+        }
+    } else {
+        PyErr_SetString(PyExc_ValueError,
+                        "str contents cannot encode to requested kind");
+        return -1;
+    }
+
+    Py_ssize_t length = PyUnicode_GET_LENGTH(unicode) * itemsize;
+
+    int res = PyBuffer_FillInfo(
+        view,
+        unicode,
+        data,
+        length,
+        1,
+        0
+    );
+
+    if (res < 0) {
+        if (needs_dealloc) {
+            PyMem_Free(data);
+        }
+        return res;
+    }
+
+    view->itemsize = itemsize;
+    view->internal = (void *) needs_dealloc;
+
+    return 0;
+}
+
+static void unicode_release_buffer(PyObject *unicode, Py_buffer *view)
+{
+    bool needs_dealloc = (bool) view->internal;
+    if (needs_dealloc) {
+        PyMem_Free(view->buf);
+    }
+}
+
+
+static PyBufferProcs unicode_as_buffer = {
+//    .bf_getbuffer = unicode_get_buffer,
+   .bf_releasebuffer = unicode_release_buffer,
+};
+
 PyDoc_STRVAR(unicode_doc,
 "str(object='') -> str\n\
 str(bytes_or_buffer[, encoding[, errors]]) -> str\n\
@@ -14739,7 +14812,7 @@ PyTypeObject PyUnicode_Type = {
     (reprfunc) unicode_str,       /* tp_str */
     PyObject_GenericGetAttr,      /* tp_getattro */
     0,                            /* tp_setattro */
-    0,                            /* tp_as_buffer */
+    &unicode_as_buffer,           /* tp_as_buffer */
     Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE |
         Py_TPFLAGS_UNICODE_SUBCLASS |
         _Py_TPFLAGS_MATCH_SELF, /* tp_flags */