Replace usage of old unicode API removed in Py3.12/PEP 623

Pythonwin/win32util.cpp

@@ -1137,8 +1137,8 @@ CString GetReprText(PyObject *objectUse)
     PyObject *s;
     CString csRet;
     s = PyObject_Str(objectUse);
-    if (s) {
-        csRet = CString(PyUnicode_AsUnicode(s));
+    if (s) if (TmpWCHAR ts=s) {
+        csRet = CString(ts);
         Py_DECREF(s);
         return csRet;
     }
@@ -1152,7 +1152,12 @@ CString GetReprText(PyObject *objectUse)
 
     // repr() should always return a unicode string, but for hysterical raisens we check if it is bytes.
     if (PyUnicode_Check(s))
-        csRet = CString(PyUnicode_AS_UNICODE(s));
+        if (TmpWCHAR ts=s)
+            csRet = ts;
+        else {
+            PyErr_Clear();
+            csRet = L"??? wide string allocation error ???";
+        }
     else if (PyBytes_Check(s))
         csRet = CString(PyBytes_AS_STRING(s));
     else

Pythonwin/win32virt.cpp

@@ -117,11 +117,9 @@ BOOL CVirtualHelper::do_call(PyObject *args)
             if (obRepr) {
                 if (PyBytes_Check(obRepr))
                     szRepr = PyBytes_AS_STRING(obRepr);
-                else if (PyUnicode_Check(obRepr))
-                    szRepr = W2A(PyUnicode_AS_UNICODE(obRepr));
+                else if (TmpWCHAR tmpw=obRepr)
+                    szRepr = W2A(tmpw);
             }
-            else
-                PyErr_Clear();
 
             LPTSTR HandlerName = csHandlerName.GetBuffer(csHandlerName.GetLength());
             snprintf(msg, sizeof(msg) / sizeof(msg[0]), "%s() virtual handler (%s) raised an exception",

com/win32com/src/extensions/PySTGMEDIUM.cpp

@@ -50,6 +50,7 @@ PyObject *PySet(PyObject *self, PyObject *args)
         }
         case TYMED_HGLOBAL: {
             const void *buf = NULL;
+            TmpWCHAR tmpw;
             Py_ssize_t cb = 0;
             PyWinBufferView pybuf;
             // In py3k, unicode objects don't support the buffer
@@ -62,8 +63,8 @@ PyObject *PySet(PyObject *self, PyObject *args)
                 buf = (void *)PyBytes_AS_STRING(ob);
             }
             else if (PyUnicode_Check(ob)) {
-                cb = PyUnicode_GET_DATA_SIZE(ob) + sizeof(Py_UNICODE);
-                buf = (void *)PyUnicode_AS_UNICODE(ob);
+                buf = tmpw = ob;  if (!tmpw) return NULL;
+                cb = (tmpw.length + 1) * sizeof(WCHAR);
             }
             else {
                 if (!pybuf.init(ob))

isapi/src/PyExtensionObjects.cpp

@@ -25,10 +25,12 @@
 
 //#define PY_SSIZE_T_CLEAN  // defined by isapi\src\StdAfx.h
 #include "stdafx.h"
+#include "pywintypes.h"
 #include "Utils.h"
 #include "PyExtensionObjects.h"
 #include "PythonEng.h"
 
+
 // Asynch IO callbacks are a little tricky, as we never know how many
 // callbacks a single connection might make (often each callback will trigger
 // another IO request.) So we keep the Python objects used by the callback
@@ -174,7 +176,7 @@ PyObject *PyVERSION_INFO::getattro(PyObject *self, PyObject *obname)
     PyVERSION_INFO *me = (PyVERSION_INFO *)self;
     if (!me->m_pvi)
         return PyErr_Format(PyExc_RuntimeError, "VERSION_INFO structure no longer exists");
-    TCHAR *name = PYISAPI_ATTR_CONVERT(obname);
+    TmpWCHAR name = obname;  if (!name) return NULL;
     if (_tcscmp(name, _T("ExtensionDesc")) == 0) {
         return PyBytes_FromString(me->m_pvi->lpszExtensionDesc);
     }
@@ -184,7 +186,7 @@ PyObject *PyVERSION_INFO::getattro(PyObject *self, PyObject *obname)
 int PyVERSION_INFO::setattro(PyObject *self, PyObject *obname, PyObject *v)
 {
     PyVERSION_INFO *me = (PyVERSION_INFO *)self;
-    TCHAR *name = PYISAPI_ATTR_CONVERT(obname);
+    TmpWCHAR name = obname;  if (!name) return -1;
     if (!me->m_pvi) {
         PyErr_Format(PyExc_RuntimeError, "VERSION_INFO structure no longer exists");
         return -1;
@@ -343,7 +345,7 @@ PyECB::~PyECB()
 
 PyObject *PyECB::getattro(PyObject *self, PyObject *obname)
 {
-    TCHAR *name = PYISAPI_ATTR_CONVERT(obname);
+    TmpWCHAR name = obname;  if (!name) return NULL;
 
     if (_tcscmp(name, _T("softspace")) == 0)  // help 'print' semantics.
         return PyLong_FromLong(1);
@@ -383,7 +385,7 @@ int PyECB::setattro(PyObject *self, PyObject *obname, PyObject *v)
         PyErr_SetString(PyExc_AttributeError, "can't delete ECB attributes");
         return -1;
     }
-    TCHAR *name = PYISAPI_ATTR_CONVERT(obname);
+    TmpWCHAR name = obname;  if (!name) return -1;
 
     if (_tcscmp(name, _T("HttpStatusCode")) == 0) {
         PyECB *pecb = (PyECB *)self;
@@ -681,8 +683,9 @@ PyObject *PyECB::GetAnonymousToken(PyObject *self, PyObject *args)
         Py_END_ALLOW_THREADS
     }
     else if (PyUnicode_Check(obStr)) {
+        TmpWCHAR tmpw = obStr;  if (!tmpw) return NULL;
         Py_BEGIN_ALLOW_THREADS bRes = ecb->ServerSupportFunction(ecb->ConnID, HSE_REQ_GET_UNICODE_ANONYMOUS_TOKEN,
-                                                                 PyUnicode_AS_UNICODE(obStr), (DWORD *)&handle, NULL);
+                                                                 tmpw, (DWORD *)&handle, NULL);
         Py_END_ALLOW_THREADS
     }
     else

isapi/src/PyFilterObjects.cpp

@@ -25,6 +25,7 @@
 
 //#define PY_SSIZE_T_CLEAN  // defined by isapi\src\StdAfx.h
 #include "stdafx.h"
+#include "pywintypes.h"
 #include "Utils.h"
 #include "pyFilterObjects.h"
 
@@ -68,7 +69,7 @@ PyObject *PyFILTER_VERSION::getattro(PyObject *self, PyObject *obname)
     PyFILTER_VERSION *me = (PyFILTER_VERSION *)self;
     if (!me->m_pfv)
         return PyErr_Format(PyExc_RuntimeError, "FILTER_VERSION structure no longer exists");
-    TCHAR *name = PYISAPI_ATTR_CONVERT(obname);
+    TmpWCHAR name = obname;  if (!name) return NULL;
     // @prop int|ServerFilterVersion|(read-only)
     if (_tcscmp(name, _T("ServerFilterVersion")) == 0) {
         return PyLong_FromLong(me->m_pfv->dwServerFilterVersion);
@@ -85,7 +86,7 @@ PyObject *PyFILTER_VERSION::getattro(PyObject *self, PyObject *obname)
     if (_tcscmp(name, _T("FilterDesc")) == 0) {
         return PyBytes_FromString(me->m_pfv->lpszFilterDesc);
     }
-    return PyErr_Format(PyExc_AttributeError, "PyFILTER_VERSION has no attribute '%s'", name);
+    return PyErr_Format(PyExc_AttributeError, "PyFILTER_VERSION has no attribute '%S'", obname);
 }
 
 int PyFILTER_VERSION::setattro(PyObject *self, PyObject *obname, PyObject *v)
@@ -99,7 +100,7 @@ int PyFILTER_VERSION::setattro(PyObject *self, PyObject *obname, PyObject *v)
         PyErr_SetString(PyExc_AttributeError, "can't delete FILTER_VERSION attributes");
         return -1;
     }
-    TCHAR *name = PYISAPI_ATTR_CONVERT(obname);
+    TmpWCHAR name = obname;  if (!name) return -1;
     if (_tcscmp(name, _T("FilterVersion")) == 0) {
         if (!PyLong_Check(v)) {
             PyErr_Format(PyExc_ValueError, "FilterVersion must be an int (got %s)", v->ob_type->tp_name);
@@ -415,7 +416,7 @@ PyHFC::~PyHFC()
 
 PyObject *PyHFC::getattro(PyObject *self, PyObject *obname)
 {
-    TCHAR *name = PYISAPI_ATTR_CONVERT(obname);
+    TmpWCHAR name = obname;  if (!name) return NULL;
 
     // other manual attributes.
     if (_tcscmp(name, _T("FilterContext")) == 0) {
@@ -434,7 +435,7 @@ PyObject *PyHFC::getattro(PyObject *self, PyObject *obname)
 
 int PyHFC::setattro(PyObject *self, PyObject *obname, PyObject *v)
 {
-    TCHAR *name = PYISAPI_ATTR_CONVERT(obname);
+    TmpWCHAR name = obname;  if (!name) return -1;
     if (v == NULL) {
         PyErr_SetString(PyExc_AttributeError, "can't delete ECB attributes");
         return -1;
@@ -536,7 +537,7 @@ PyObject *PyURL_MAP::getattro(PyObject *self, PyObject *obname)
     if (!pMap)
         return NULL;
     // @prop string|URL|
-    TCHAR *name = PYISAPI_ATTR_CONVERT(obname);
+    TmpWCHAR name = obname;  if (!name) return NULL;
     if (_tcscmp(name, _T("URL")) == 0) {
         return PyBytes_FromString(pMap->pszURL);
     }
@@ -552,7 +553,7 @@ int PyURL_MAP::setattro(PyObject *self, PyObject *obname, PyObject *v)
     HTTP_FILTER_URL_MAP *pMap = ((PyURL_MAP *)self)->GetURLMap();
     if (!pMap)
         return NULL;
-    TCHAR *name = PYISAPI_ATTR_CONVERT(obname);
+    TmpWCHAR name = obname;  if (!name) return -1;
     if (_tcscmp(name, _T("PhysicalPath")) == 0) {
         if (!PyBytes_Check(v)) {
             PyErr_Format(PyExc_TypeError, "PhysicalPath must be a string");
@@ -779,7 +780,7 @@ PyObject *PyRAW_DATA::getattro(PyObject *self, PyObject *obname)
     if (!pRD)
         return NULL;
     // @prop string|InData|
-    TCHAR *name = PYISAPI_ATTR_CONVERT(obname);
+    TmpWCHAR name = obname;  if (!name) return NULL;
     if (_tcscmp(name, _T("InData")) == 0) {
         if (pRD->pvInData == NULL) {
             Py_INCREF(Py_None);
@@ -799,7 +800,7 @@ int PyRAW_DATA::setattro(PyObject *self, PyObject *obname, PyObject *v)
     if (!pRD || !pFC)
         return NULL;
 
-    TCHAR *name = PYISAPI_ATTR_CONVERT(obname);
+    TmpWCHAR name = obname;  if (!name) return -1;
     if (_tcscmp(name, _T("InData")) == 0) {
         if (!PyBytes_Check(v)) {
             PyErr_Format(PyExc_TypeError, "InData must be a string (got %s)", v->ob_type->tp_name);
@@ -885,7 +886,7 @@ PyObject *PyAUTHENT::getattro(PyObject *self, PyObject *obname)
     HTTP_FILTER_AUTHENT *pAE = ((PyAUTHENT *)self)->GetAUTHENT();
     if (!pAE)
         return NULL;
-    TCHAR *name = PYISAPI_ATTR_CONVERT(obname);
+    TmpWCHAR name = obname;  if (!name) return NULL;
     // @prop string|User|
     if (_tcscmp(name, _T("User")) == 0) {
         if (pAE->pszUser == NULL) {
@@ -913,7 +914,7 @@ int PyAUTHENT::setattro(PyObject *self, PyObject *obname, PyObject *v)
     if (!pAE || !pFC)
         return NULL;
 
-    TCHAR *name = PYISAPI_ATTR_CONVERT(obname);
+    TmpWCHAR name = obname;  if (!name) return -1;
     if (_tcscmp(name, _T("User")) == 0) {
         if (!PyBytes_Check(v)) {
             PyErr_Format(PyExc_TypeError, "User must be a string (got %s)", v->ob_type->tp_name);
@@ -998,7 +999,7 @@ PyObject *PyFILTER_LOG::getattro(PyObject *self, PyObject *obname)
     HTTP_FILTER_LOG *pLog = ((PyFILTER_LOG *)self)->GetFilterLog();
     if (!pLog)
         return NULL;
-    TCHAR *name = PYISAPI_ATTR_CONVERT(obname);
+    TmpWCHAR name = obname;  if (!name) return NULL;
     // @prop string|ClientHostName|
     if (_tcscmp(name, _T("ClientHostName")) == 0)
         return PyBytes_FromString(pLog->pszClientHostName);
@@ -1063,7 +1064,7 @@ int PyFILTER_LOG::setattro(PyObject *self, PyObject *obname, PyObject *v)
     HTTP_FILTER_LOG *pLog = ((PyFILTER_LOG *)self)->GetFilterLog();
     if (!pLog || !pFC)
         return NULL;
-    TCHAR *name = PYISAPI_ATTR_CONVERT(obname);
+    TmpWCHAR name = obname;  if (!name) return -1;
     CHECK_SET_FILTER_LOG_STRING(ClientHostName)
     CHECK_SET_FILTER_LOG_STRING(ClientUserName)
     CHECK_SET_FILTER_LOG_STRING(ServerName)

isapi/src/PythonEng.h

@@ -79,6 +79,7 @@ class CPythonHandler {
 void ExtensionError(CControlBlock *pcb, const char *errmsg);
 void FilterError(CFilterContext *pfc, const char *errmsg);
 
+#ifndef __PYWINTYPES_H__
 class CEnterLeavePython {
    public:
     CEnterLeavePython() : state(PyGILState_Ensure()) { ; }
@@ -87,5 +88,6 @@ class CEnterLeavePython {
    protected:
     PyGILState_STATE state;
 };
+#endif
 
 #endif  // __PythonEngine_H

isapi/src/StdAfx.h

@@ -54,7 +54,7 @@
 
 // Macros to handle PyObject layout changes in Py3k
 #define PYISAPI_OBJECT_HEAD PyVarObject_HEAD_INIT(NULL, 0)
-#define PYISAPI_ATTR_CONVERT PyUnicode_AsUnicode
+////#define PYISAPI_ATTR_CONVERT PyUnicode_AsUnicode  // removed in Py3.12+
 
 // A helper that on py3k takes a str or unicode as input and returns a
 // string - exactly how the 's#' PyArg_ParseTuple format string does...

win32/src/PyUnicode.cpp

@@ -34,8 +34,8 @@ BOOL PyWinObject_AsPfnAllocatedWCHAR(PyObject *stringObject, void *(*pfnAllocato
     }
     else if (PyUnicode_Check(stringObject)) {
         // copy the value, including embedded NULLs
-        WCHAR *v = (WCHAR *)PyUnicode_AS_UNICODE(stringObject);
-        Py_ssize_t cch = PyUnicode_GET_SIZE(stringObject);
+        TmpWCHAR v = stringObject;  if (!v) return FALSE;
+        Py_ssize_t cch = v.length;
         *ppResult = (WCHAR *)pfnAllocator((cch + 1) * sizeof(WCHAR));
         if (*ppResult)
             memcpy(*ppResult, v, (cch + 1) * sizeof(WCHAR));
@@ -102,11 +102,7 @@ BOOL PyWinObject_AsChars(PyObject *stringObject, char **pResult, BOOL bNoneOK /*
     // Convert the string if a WIDE string.
     if (PyUnicode_Check(stringObject)) {
         // PyUnicode_EncodeMBCS was removed in Py 3.11.
-        PyObject *unicode = PyUnicode_FromWideChar(PyUnicode_AS_UNICODE(stringObject), -1);
-        if (unicode == NULL)
-            return FALSE;
-        stringObject = tempObject = PyUnicode_EncodeCodePage(CP_ACP, unicode, NULL);
-        Py_DECREF(unicode);
+        stringObject = tempObject = PyUnicode_AsMBCSString(stringObject);
         if (!stringObject)
             return FALSE;
     }
@@ -184,22 +180,11 @@ BOOL PyWinObject_AsBstr(PyObject *stringObject, BSTR *pResult, BOOL bNoneOK /*=
     // sane b/w compat reason to support that any more.
     if (PyUnicode_Check(stringObject)) {
         // copy the value, including embedded NULLs
-        Py_ssize_t nchars = PyUnicode_GET_SIZE(stringObject);
-        *pResult = SysAllocStringLen(NULL, nchars);
-        if (*pResult) {
-#define PUAWC_TYPE PyObject *
-            if (PyUnicode_AsWideChar((PUAWC_TYPE)stringObject, *pResult, nchars) == -1) {
-                rc = FALSE;
-            }
-            else {
-                // The SysAllocStringLen docs indicate that nchars+1 bytes are allocated,
-                // and that normally a \0 is appened by the function.  It also states
-                // the \0 is not necessary!  While it seems to work fine without it,
-                // we do copy it, as the previous code, which used SysAllocStringLen
-                // with a non-NULL arg is documented clearly as appending the \0.
-                (*pResult)[nchars] = 0;
-            }
-        }
+        // Py3.12+: only conversion yields the correct number of wide chars (incl. surrogate pairs).
+        // For simplicity we use a temp buffer.
+        TmpWCHAR tw = stringObject;  if (!tw) return FALSE;
+        // SysAllocStringLen allocates length+1 wchars (and puts a \0 at end); like PyUnicode_AsWideCharString
+        *pResult = SysAllocStringLen(tw, tw.length);
     }
     else if (stringObject == Py_None) {
         if (bNoneOK) {
@@ -228,23 +213,19 @@ void PyWinObject_FreeBstr(BSTR str) { SysFreeString(str); }
 
 // String conversions
 // Convert a Python object to a WCHAR - allow embedded NULLs, None, etc.
-// Must be freed with PyWinObject_FreeWCHAR
+// Must be freed with PyWinObject_FreeWCHAR / PyMem_Free
 BOOL PyWinObject_AsWCHAR(PyObject *stringObject, WCHAR **pResult, BOOL bNoneOK /*= FALSE*/,
                          DWORD *pResultLen /*= NULL*/)
 {
     BOOL rc = TRUE;
     Py_ssize_t resultLen = 0;
     // Do NOT accept 'bytes' for any 'WCHAR' API.
     if (PyUnicode_Check(stringObject)) {
-        resultLen = PyUnicode_GET_SIZE(stringObject);
-        size_t cb = sizeof(WCHAR) * (resultLen + 1);
-        *pResult = (WCHAR *)PyMem_Malloc(cb);
+        *pResult = PyUnicode_AsWideCharString(stringObject, &resultLen);
         if (*pResult == NULL) {
-            PyErr_SetString(PyExc_MemoryError, "Allocating WCHAR array");
+            PyErr_SetString(PyExc_MemoryError, "Getting WCHAR string");
             return FALSE;
         }
-        // copy the value, including embedded NULLs
-        memcpy(*pResult, PyUnicode_AsUnicode(stringObject), cb);
     }
     else if (stringObject == Py_None) {
         if (bNoneOK) {

win32/src/PyWinTypes.h

@@ -167,19 +167,35 @@ PYWINTYPES_EXPORT void PyWinObject_FreeChars(char *pResult);
 // Automatically freed WCHAR that can be used anywhere WCHAR * is required
 class TmpWCHAR {
    public:
-    WCHAR *tmp;
+    WCHAR *tmp;  // (NULL after conversion error)
+    Py_ssize_t length;  // only set after successful auto-conversion; w/o trailing \0
     TmpWCHAR() { tmp = NULL; }
     TmpWCHAR(WCHAR *t) { tmp = t; }
+    TmpWCHAR(PyObject *ob) : tmp(NULL) { *this = ob; }
+    WCHAR *operator=(PyObject *ob) {
+        if (tmp)
+            PyMem_Free(tmp);
+        if (ob == NULL)
+            tmp = NULL;  // (exception already has been set in this case)
+        else
+            tmp = PyUnicode_AsWideCharString(ob, &length);
+        return tmp;
+    }
     WCHAR *operator=(WCHAR *t)
     {
-        PyWinObject_FreeWCHAR(tmp);
+        if (tmp)
+            PyMem_Free(tmp);
         tmp = t;
         return t;
     }
     WCHAR **operator&() { return &tmp; }
     boolean operator==(WCHAR *t) { return tmp == t; }
     operator WCHAR *() { return tmp; }
-    ~TmpWCHAR() { PyWinObject_FreeWCHAR(tmp); }
+    ~TmpWCHAR() { if (tmp) PyMem_Free(tmp); }
+   private:
+    // Block unwanted copy construction
+    TmpWCHAR(const TmpWCHAR& o);  // = delete;
+    const TmpWCHAR& operator=(const TmpWCHAR& o);  // = delete;
 };
 
 // More string helpers - how many do we need?

win32/src/PythonService.cpp

@@ -1463,9 +1463,9 @@ int _tmain(int argc, TCHAR **argv)
     // now get the handle to the DLL, and call the main function.
     if (PyBytes_Check(f))
         hmod = GetModuleHandleA(PyBytes_AsString(f));
-    else if (PyUnicode_Check(f))
-        hmod = GetModuleHandleW(PyUnicode_AsUnicode(f));
-    else {
+    else if (TmpWCHAR tw=f) {
+        hmod = GetModuleHandleW(tw);
+    } else {
         PyErr_SetString(PyExc_TypeError, "servicemanager.__file__ is not a string or unicode !");
         goto failed;
     }