_Locinfo: Use _wsetlocale to query and restore locales

stl/inc/xlocinfo

@@ -134,6 +134,24 @@ private:
     void* _Timeptr; // pointer to time information
 };
 
+template <class _Elem>
+_NODISCARD _Elem* _Ntcts_dup_dbg(const _Elem* const _Ptr) noexcept {
+    _STL_INTERNAL_STATIC_ASSERT(_Is_any_of_v<_Elem, char, wchar_t, unsigned short>);
+
+    const size_t _Bytes = (char_traits<_Elem>::length(_Ptr) + 1) * sizeof(_Elem);
+
+#ifdef _DEBUG
+    const auto _Result = static_cast<_Elem*>(_CSTD _malloc_dbg(_Bytes, _CRT_BLOCK, __FILE__, __LINE__));
+#else
+    const auto _Result = static_cast<_Elem*>(_CSTD malloc(_Bytes));
+#endif
+    if (_Result) {
+        _CSTD memcpy(_Result, _Ptr, _Bytes);
+    }
+
+    return _Result;
+}
+
 extern "C++" template <class _Elem>
 class _CRTIMP2_PURE_IMPORT _Yarn { // wrap a NTCTS
 public:
@@ -156,26 +174,22 @@ public:
             _Tidy();
 
             if (_Right) { // new is not empty, copy it
-                const _Elem* _Ptr = _Right;
-                while (*_Ptr != _Elem{}) {
-                    ++_Ptr;
-                }
+                _Myptr = _STD _Ntcts_dup_dbg(_Right);
+            }
+        }
 
-                const auto _Count = (++_Ptr - _Right) * sizeof(_Elem);
+        return *this;
+    }
 
-#ifdef _DEBUG
-                _Myptr = static_cast<_Elem*>(_malloc_dbg(_Count, _CRT_BLOCK, __FILE__, __LINE__));
-#else
-                _Myptr = static_cast<_Elem*>(_CSTD malloc(_Count));
-#endif
+    template <class _Elem2 = _Elem, enable_if_t<is_same_v<_Elem2, char>, int> = 0>
+    void _From_wide(const wchar_t* const _Right) noexcept {
+        if (reinterpret_cast<wchar_t*>(_Myptr) != _Right) { // new value, discard old and copy new
+            _Tidy();
 
-                if (_Myptr) {
-                    _CSTD memcpy(_Myptr, _Right, _Count);
-                }
+            if (_Right) { // new is not empty, copy it
+                _Myptr = reinterpret_cast<char*>(_STD _Ntcts_dup_dbg(_Right));
             }
         }
-
-        return *this;
     }
 
     __CLR_OR_THIS_CALL ~_Yarn() noexcept {
@@ -382,6 +396,7 @@ private:
     _Yarn<char> _Months; // month names
     _Yarn<wchar_t> _W_Days; // wide weekday names
     _Yarn<wchar_t> _W_Months; // wide month names
+    // TRANSITION, ABI, `_Oldlocname._Myptr` is reinterpreted as `wchar_t*`. We should use `_Yarn<wchar_t>`.
     _Yarn<char> _Oldlocname; // old locale name to revert to on destruction
     _Yarn<char> _Newlocname; // new locale name for this object
 };

stl/src/locale.cpp

@@ -137,9 +137,7 @@ void __CLRCALL_PURE_OR_CDECL locale::_Locimp::_Locimp_Addfac(
 
 void __CLRCALL_PURE_OR_CDECL _Locinfo::_Locinfo_ctor(
     _Locinfo* pLocinfo, int cat, const char* locname) { // capture a named locale
-    const char* oldlocname = setlocale(LC_ALL, nullptr);
-
-    pLocinfo->_Oldlocname = oldlocname == nullptr ? "" : oldlocname;
+    pLocinfo->_Oldlocname._From_wide(_wsetlocale(LC_ALL, nullptr));
     _Locinfo_Addcats(pLocinfo, cat, locname);
 }
 

stl/src/locale0.cpp

@@ -221,9 +221,8 @@ void __CLRCALL_PURE_OR_CDECL locale::_Locimp::_Locimp_dtor(_Locimp* _This) { //
 
 void __CLRCALL_PURE_OR_CDECL _Locinfo::_Locinfo_ctor(
     _Locinfo* pLocinfo, const char* locname) { // switch to a named locale
-    const char* oldlocname = setlocale(LC_ALL, nullptr);
+    pLocinfo->_Oldlocname._From_wide(_wsetlocale(LC_ALL, nullptr));
 
-    pLocinfo->_Oldlocname = oldlocname == nullptr ? "" : oldlocname;
     if (locname != nullptr) {
         locname = setlocale(LC_ALL, locname);
     }
@@ -232,8 +231,12 @@ void __CLRCALL_PURE_OR_CDECL _Locinfo::_Locinfo_ctor(
 }
 
 void __CLRCALL_PURE_OR_CDECL _Locinfo::_Locinfo_dtor(_Locinfo* pLocinfo) { // destroy a _Locinfo object, revert locale
-    if (!pLocinfo->_Oldlocname._Empty()) {
-        setlocale(LC_ALL, pLocinfo->_Oldlocname._C_str());
+    if (pLocinfo->_Oldlocname._Empty()) {
+        // `pLocinfo->_Oldlocname._C_str()` points to a single `char` of value 0 in this case,
+        // so reinterpret_cast is not reliable.
+        _wsetlocale(LC_ALL, L"");
+    } else {
+        _wsetlocale(LC_ALL, reinterpret_cast<const wchar_t*>(pLocinfo->_Oldlocname._C_str()));
     }
 }
 _STD_END

tests/std/test.lst

@@ -277,6 +277,7 @@ tests\GH_005504_avoid_function_call_wrapping
 tests\GH_005546_containers_size_type_cast
 tests\GH_005553_regex_character_translation
 tests\GH_005768_pow_accuracy
+tests\GH_005780_non_ascii_locales
 tests\GH_005800_stable_sort_large_alignment
 tests\GH_005968_headers_provide_begin_end
 tests\LWG2381_num_get_floating_point

tests/std/tests/GH_005780_non_ascii_locales/env.lst

@@ -0,0 +1,4 @@
+# Copyright (c) Microsoft Corporation.
+# SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+
+RUNALL_INCLUDE ..\usual_matrix.lst

tests/std/tests/GH_005780_non_ascii_locales/test.cpp

@@ -0,0 +1,44 @@
+// Copyright (c) Microsoft Corporation.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+
+#include <algorithm>
+#include <cassert>
+#include <clocale>
+#include <iomanip>
+#include <iostream>
+#include <locale>
+#include <string>
+using namespace std;
+
+string set_locale(const string& locale_name) {
+    const char* const ret = setlocale(LC_ALL, locale_name.c_str());
+    assert(ret != nullptr);
+    return ret;
+}
+
+string query_locale() {
+    const char* const ret = setlocale(LC_ALL, nullptr);
+    assert(ret != nullptr);
+    return ret;
+}
+
+void assert_string_non_ascii(const string& str) {
+    const auto char_not_ascii = [](const char c) { return (c & 0x80) != 0; };
+    assert(any_of(str.begin(), str.end(), char_not_ascii));
+}
+
+// GH-5780 <iostream>: Printing with std::cout changes the global locale permanently
+void test_gh_5780() {
+    // https://learn.microsoft.com/en-us/cpp/c-runtime-library/language-strings#supported-language-strings
+    const string locale_name = set_locale("norwegian-bokmal.437");
+    assert_string_non_ascii(locale_name);
+
+    cerr.imbue(locale::classic());
+    cerr << setprecision(2) << 0.1 << endl;
+
+    assert(query_locale() == locale_name);
+}
+
+int main() {
+    test_gh_5780();
+}