fix: use _get_buffer_or_str in tp_init for consistent str error

Parse data as PyObject* with O format instead of y*, then validate
via _get_buffer_or_str. This ensures both positional and keyword
str give 'Strings must be encoded before hashing' on all platforms
including PyPy where tp_vectorcall falls back to tp_init.

Author: ifduyue

src/_xxhash.c

@@ -550,18 +550,21 @@ static PyObject *PYXXH32_new(PyTypeObject *type, PyObject *args, PyObject *kwarg
 static int PYXXH32_init(PYXXH32Object *self, PyObject *args, PyObject *kwargs)
 {
     XXH32_hash_t seed = 0;
+    PyObject *data_obj = NULL;
     char *keywords[] = {"data", "seed", NULL};
     Py_buffer buf;
 
     buf.buf = buf.obj = NULL;
 
-    if (PyTuple_GET_SIZE(args) >= 1 && PyUnicode_Check(PyTuple_GET_ITEM(args, 0))) {
-        PyErr_SetString(PyExc_TypeError,
-            "Strings must be encoded before hashing");
+    if (!PyArg_ParseTupleAndKeywords(args, kwargs, "|OI:__init__", keywords, &data_obj, &seed)) {
         return -1;
     }
-    if (!PyArg_ParseTupleAndKeywords(args, kwargs, "|y*I:__init__", keywords, &buf, &seed)) {
-        return -1;
+
+    if (data_obj && data_obj != Py_None) {
+        PyObject *buf_owner;
+        if (_get_buffer_or_str(data_obj, &buf, &buf_owner) < 0)
+            return -1;
+        Py_XDECREF(buf_owner);
     }
 
     self->seed = seed;
@@ -910,18 +913,21 @@ static PyObject *PYXXH64_new(PyTypeObject *type, PyObject *args, PyObject *kwarg
 static int PYXXH64_init(PYXXH64Object *self, PyObject *args, PyObject *kwargs)
 {
     XXH64_hash_t seed = 0;
+    PyObject *data_obj = NULL;
     char *keywords[] = {"data", "seed", NULL};
     Py_buffer buf;
 
     buf.buf = buf.obj = NULL;
 
-    if (PyTuple_GET_SIZE(args) >= 1 && PyUnicode_Check(PyTuple_GET_ITEM(args, 0))) {
-        PyErr_SetString(PyExc_TypeError,
-            "Strings must be encoded before hashing");
+    if (!PyArg_ParseTupleAndKeywords(args, kwargs, "|OK:__init__", keywords, &data_obj, &seed)) {
         return -1;
     }
-    if (!PyArg_ParseTupleAndKeywords(args, kwargs, "|y*K:__init__", keywords, &buf, &seed)) {
-        return -1;
+
+    if (data_obj && data_obj != Py_None) {
+        PyObject *buf_owner;
+        if (_get_buffer_or_str(data_obj, &buf, &buf_owner) < 0)
+            return -1;
+        Py_XDECREF(buf_owner);
     }
 
     self->seed = seed;
@@ -1269,18 +1275,21 @@ static PyObject *PYXXH3_64_new(PyTypeObject *type, PyObject *args, PyObject *kwa
 static int PYXXH3_64_init(PYXXH3_64Object *self, PyObject *args, PyObject *kwargs)
 {
     XXH64_hash_t seed = 0;
+    PyObject *data_obj = NULL;
     char *keywords[] = {"data", "seed", NULL};
     Py_buffer buf;
 
     buf.buf = buf.obj = NULL;
 
-    if (PyTuple_GET_SIZE(args) >= 1 && PyUnicode_Check(PyTuple_GET_ITEM(args, 0))) {
-        PyErr_SetString(PyExc_TypeError,
-            "Strings must be encoded before hashing");
+    if (!PyArg_ParseTupleAndKeywords(args, kwargs, "|OK:__init__", keywords, &data_obj, &seed)) {
         return -1;
     }
-    if (!PyArg_ParseTupleAndKeywords(args, kwargs, "|y*K:__init__", keywords, &buf, &seed)) {
-        return -1;
+
+    if (data_obj && data_obj != Py_None) {
+        PyObject *buf_owner;
+        if (_get_buffer_or_str(data_obj, &buf, &buf_owner) < 0)
+            return -1;
+        Py_XDECREF(buf_owner);
     }
 
     self->seed = seed;
@@ -1637,18 +1646,21 @@ static PyObject *PYXXH3_128_new(PyTypeObject *type, PyObject *args, PyObject *kw
 static int PYXXH3_128_init(PYXXH3_128Object *self, PyObject *args, PyObject *kwargs)
 {
     XXH64_hash_t seed = 0;
+    PyObject *data_obj = NULL;
     char *keywords[] = {"data", "seed", NULL};
     Py_buffer buf;
 
     buf.buf = buf.obj = NULL;
 
-    if (PyTuple_GET_SIZE(args) >= 1 && PyUnicode_Check(PyTuple_GET_ITEM(args, 0))) {
-        PyErr_SetString(PyExc_TypeError,
-            "Strings must be encoded before hashing");
+    if (!PyArg_ParseTupleAndKeywords(args, kwargs, "|OK:__init__", keywords, &data_obj, &seed)) {
         return -1;
     }
-    if (!PyArg_ParseTupleAndKeywords(args, kwargs, "|y*K:__init__", keywords, &buf, &seed)) {
-        return -1;
+
+    if (data_obj && data_obj != Py_None) {
+        PyObject *buf_owner;
+        if (_get_buffer_or_str(data_obj, &buf, &buf_owner) < 0)
+            return -1;
+        Py_XDECREF(buf_owner);
     }
 
     self->seed = seed;