<locale>: Repair std::collate<unsigned short> (#5361)

muellerj2 · StephanTLavavej · web-flow · commit 42cee97f2f4b · 2025-04-22T14:39:56.000-07:00
Co-authored-by: Stephan T. Lavavej &lt;stl@microsoft.com&gt;
diff --git a/stl/inc/locale b/stl/inc/locale
@@ -22,6 +22,78 @@ _STL_DISABLE_CLANG_WARNINGS
 #undef new
 
 _STD_BEGIN
+template <class _Elem>
+int __CRTDECL _LStrcoll(const _Elem* _First1, const _Elem* _Last1, const _Elem* _First2, const _Elem* _Last2,
+    const _Locinfo::_Collvec*) { // perform locale-specific comparison of _Elem sequences
+    for (; _First1 != _Last1 && _First2 != _Last2; ++_First1, ++_First2) {
+        if (*_First1 < *_First2) {
+            return -1; // [_First1, _Last1) < [_First2, _Last2)
+        } else if (*_First2 < *_First1) {
+            return +1; // [_First1, _Last1) > [_First2, _Last2)
+        }
+    }
+
+    return _First2 != _Last2 ? -1 : _First1 != _Last1 ? +1 : 0;
+}
+
+template <>
+inline int __CRTDECL _LStrcoll(const char* _First1, const char* _Last1, const char* _First2, const char* _Last2,
+    const _Locinfo::_Collvec* _Vector) { // perform locale-specific comparison of char sequences
+    return _Strcoll(_First1, _Last1, _First2, _Last2, _Vector);
+}
+
+template <>
+inline int __CRTDECL _LStrcoll(const wchar_t* _First1, const wchar_t* _Last1, const wchar_t* _First2,
+    const wchar_t* _Last2,
+    const _Locinfo::_Collvec* _Vector) { // perform locale-specific comparison of wchar_t sequences
+    return _Wcscoll(_First1, _Last1, _First2, _Last2, _Vector);
+}
+
+#ifdef _CRTBLD
+template <>
+inline int __CRTDECL _LStrcoll(const unsigned short* _First1, const unsigned short* _Last1,
+    const unsigned short* _First2, const unsigned short* _Last2,
+    const _Locinfo::_Collvec* _Vector) { // perform locale-specific comparison of unsigned short sequences
+    return _Wcscoll(reinterpret_cast<const wchar_t*>(_First1), reinterpret_cast<const wchar_t*>(_Last1),
+        reinterpret_cast<const wchar_t*>(_First2), reinterpret_cast<const wchar_t*>(_Last2), _Vector);
+}
+#endif // defined(_CRTBLD)
+
+template <class _Elem>
+size_t __CRTDECL _LStrxfrm(_Elem* _First1, _Elem* _Last1, const _Elem* _First2, const _Elem* _Last2,
+    const _Locinfo::_Collvec*) { // perform locale-specific transform of _Elems [_First1, _Last1)
+    const ptrdiff_t _Count = _Last2 - _First2;
+    if (_Count <= _Last1 - _First1) {
+        _CSTD memcpy(_First1, _First2, _Count * sizeof(_Elem));
+    }
+
+    return _Count;
+}
+
+template <>
+inline size_t __CRTDECL _LStrxfrm(_Out_writes_(_Last1 - _First1) _Post_readable_size_(return) char* _First1,
+    _In_z_ char* _Last1, const char* _First2, const char* _Last2,
+    const _Locinfo::_Collvec* _Vector) { // perform locale-specific transform of chars [_First1, _Last1)
+    return _Strxfrm(_First1, _Last1, _First2, _Last2, _Vector);
+}
+
+template <>
+inline size_t __CRTDECL _LStrxfrm(_Out_writes_(_Last1 - _First1) _Post_readable_size_(return) wchar_t* _First1,
+    _In_z_ wchar_t* _Last1, const wchar_t* _First2, const wchar_t* _Last2,
+    const _Locinfo::_Collvec* _Vector) { // perform locale-specific transform of wchar_ts [_First1, _Last1)
+    return _Wcsxfrm(_First1, _Last1, _First2, _Last2, _Vector);
+}
+
+#ifdef _CRTBLD
+template <>
+inline size_t __CRTDECL _LStrxfrm(_Out_writes_(_Last1 - _First1) _Post_readable_size_(return) unsigned short* _First1,
+    _In_z_ unsigned short* _Last1, const unsigned short* _First2, const unsigned short* _Last2,
+    const _Locinfo::_Collvec* _Vector) { // perform locale-specific transform of unsigned shorts [_First1, _Last1)
+    return _Wcsxfrm(reinterpret_cast<wchar_t*>(_First1), reinterpret_cast<wchar_t*>(_Last1),
+        reinterpret_cast<const wchar_t*>(_First2), reinterpret_cast<const wchar_t*>(_Last2), _Vector);
+}
+#endif // defined(_CRTBLD)
+
 _EXPORT_STD template <class _Elem>
 class collate : public locale::facet { // facet for ordering sequences of elements
 public:
diff --git a/stl/inc/xlocinfo b/stl/inc/xlocinfo
@@ -385,58 +385,6 @@ private:
     _Yarn<char> _Oldlocname; // old locale name to revert to on destruction
     _Yarn<char> _Newlocname; // new locale name for this object
 };
-
-template <class _Elem>
-int __CRTDECL _LStrcoll(const _Elem* _First1, const _Elem* _Last1, const _Elem* _First2, const _Elem* _Last2,
-    const _Locinfo::_Collvec*) { // perform locale-specific comparison of _Elem sequences
-    for (; _First1 != _Last1 && _First2 != _Last2; ++_First1, ++_First2) {
-        if (*_First1 < *_First2) {
-            return -1; // [_First1, _Last1) < [_First2, _Last2)
-        } else if (*_First2 < *_First1) {
-            return +1; // [_First1, _Last1) > [_First2, _Last2)
-        }
-    }
-
-    return _First2 != _Last2 ? -1 : _First1 != _Last1 ? +1 : 0;
-}
-
-template <>
-inline int __CRTDECL _LStrcoll(const char* _First1, const char* _Last1, const char* _First2, const char* _Last2,
-    const _Locinfo::_Collvec* _Vector) { // perform locale-specific comparison of char sequences
-    return _Strcoll(_First1, _Last1, _First2, _Last2, _Vector);
-}
-
-template <>
-inline int __CRTDECL _LStrcoll(const wchar_t* _First1, const wchar_t* _Last1, const wchar_t* _First2,
-    const wchar_t* _Last2,
-    const _Locinfo::_Collvec* _Vector) { // perform locale-specific comparison of wchar_t sequences
-    return _Wcscoll(_First1, _Last1, _First2, _Last2, _Vector);
-}
-
-template <class _Elem>
-size_t __CRTDECL _LStrxfrm(_Elem* _First1, _Elem* _Last1, const _Elem* _First2, const _Elem* _Last2,
-    const _Locinfo::_Collvec*) { // perform locale-specific transform of _Elems [_First1, _Last1)
-    const ptrdiff_t _Count = _Last2 - _First2;
-    if (_Count <= _Last1 - _First1) {
-        _CSTD memcpy(_First1, _First2, _Count * sizeof(_Elem));
-    }
-
-    return _Count;
-}
-
-template <>
-inline size_t __CRTDECL _LStrxfrm(_Out_writes_(_Last1 - _First1) _Post_readable_size_(return) char* _First1,
-    _In_z_ char* _Last1, const char* _First2, const char* _Last2,
-    const _Locinfo::_Collvec* _Vector) { // perform locale-specific transform of chars [_First1, _Last1)
-    return _Strxfrm(_First1, _Last1, _First2, _Last2, _Vector);
-}
-
-template <>
-inline size_t __CRTDECL _LStrxfrm(_Out_writes_(_Last1 - _First1) _Post_readable_size_(return) wchar_t* _First1,
-    _In_z_ wchar_t* _Last1, const wchar_t* _First2, const wchar_t* _Last2,
-    const _Locinfo::_Collvec* _Vector) { // perform locale-specific transform of wchar_ts [_First1, _Last1)
-    return _Wcsxfrm(_First1, _Last1, _First2, _Last2, _Vector);
-}
 _STD_END
 #pragma pop_macro("new")
 _STL_RESTORE_CLANG_WARNINGS
diff --git a/tests/std/test.lst b/tests/std/test.lst
@@ -256,6 +256,7 @@ tests\GH_004929_internal_tag_constructors
 tests\GH_004930_char_traits_user_specialization
 tests\GH_005090_stl_hardening
 tests\GH_005204_regex_collating_ranges
+tests\GH_005236_collate_facet
 tests\GH_005244_regex_escape_sequences
 tests\GH_005315_destructor_tombstones
 tests\LWG2381_num_get_floating_point
diff --git a/tests/std/tests/GH_005204_regex_collating_ranges/test.cpp b/tests/std/tests/GH_005204_regex_collating_ranges/test.cpp
@@ -11,22 +11,16 @@
 #include <test_regex_support.hpp>
 
 // skip collation tests when linking to the DLL in case of
-// * undefined _NATIVE_WCHAR_T_DEFINED due to GH-5236
-// * _ITERATOR_DEBUG_LEVEL mismatch between code and linked DLL
+// _ITERATOR_DEBUG_LEVEL mismatch between code and linked DLL
 #ifdef _DEBUG
 #define DEFAULT_IDL_SETTING 2
 #else
 #define DEFAULT_IDL_SETTING 0
 #endif
 
-#ifdef _DLL
-#ifndef _NATIVE_WCHAR_T_DEFINED // TRANSITION, GH-212 or GH-5236
+#if defined(_DLL) && _ITERATOR_DEBUG_LEVEL != DEFAULT_IDL_SETTING
 #define SKIP_COLLATE_TESTS
-#elif _ITERATOR_DEBUG_LEVEL != DEFAULT_IDL_SETTING
-#define SKIP_COLLATE_TESTS
-#endif // !defined(_NATIVE_WCHAR_T_DEFINED) || _ITERATOR_DEBUG_LEVEL != DEFAULT_IDL_SETTING
-#endif // defined(_DLL)
-
+#endif // defined(_DLL) && _ITERATOR_DEBUG_LEVEL != DEFAULT_IDL_SETTING
 
 using namespace std;
 using namespace std::regex_constants;
diff --git a/tests/std/tests/GH_005236_collate_facet/env.lst b/tests/std/tests/GH_005236_collate_facet/env.lst
@@ -0,0 +1,4 @@
+# Copyright (c) Microsoft Corporation.
+# SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+
+RUNALL_INCLUDE ..\usual_matrix.lst
diff --git a/tests/std/tests/GH_005236_collate_facet/test.cpp b/tests/std/tests/GH_005236_collate_facet/test.cpp
@@ -0,0 +1,43 @@
+// Copyright (c) Microsoft Corporation.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+
+#include <cassert>
+#include <locale>
+#include <string>
+
+// skip collate::transform() tests when linking to the DLL in case of
+// _ITERATOR_DEBUG_LEVEL mismatch between code and linked DLL
+#ifdef _DEBUG
+#define DEFAULT_IDL_SETTING 2
+#else
+#define DEFAULT_IDL_SETTING 0
+#endif
+
+#if defined(_DLL) && _ITERATOR_DEBUG_LEVEL != DEFAULT_IDL_SETTING
+#define SKIP_COLLATE_TRANSFORM_TESTS
+#endif // defined(_DLL) && _ITERATOR_DEBUG_LEVEL != DEFAULT_IDL_SETTING
+
+using namespace std;
+
+// GH-5236 "std::collate<wchar_t> does not respect collation order when compiled with /MD(d) /Zc:wchar_t-"
+void test_gh_5236() {
+    const wchar_t Ue = L'\u00DC'; // U+00DC LATIN CAPITAL LETTER U WITH DIARESIS
+    const wchar_t U  = L'U';
+    const wchar_t V  = L'V';
+
+    // German phonebook order: "U+00DC" is sorted between U and V in collation order
+    const locale loc("de-DE_phoneb");
+    const auto& coll = use_facet<collate<wchar_t>>(loc);
+
+    assert(coll.compare(&U, &U + 1, &Ue, &Ue + 1) < 0);
+    assert(coll.compare(&V, &V + 1, &Ue, &Ue + 1) > 0);
+
+#ifndef SKIP_COLLATE_TRANSFORM_TESTS
+    assert(coll.transform(&U, &U + 1) < coll.transform(&Ue, &Ue + 1));
+    assert(coll.transform(&V, &V + 1) > coll.transform(&Ue, &Ue + 1));
+#endif // !defined(SKIP_COLLATE_TRANSFORM_TESTS)
+}
+
+int main() {
+    test_gh_5236();
+}
diff --git a/tests/std/tests/VSO_0000000_instantiate_iterators_misc/test.compile.pass.cpp b/tests/std/tests/VSO_0000000_instantiate_iterators_misc/test.compile.pass.cpp
@@ -810,7 +810,7 @@ void limits_test() {
 void locale_test() {
     char c{};
     locale loc{};
-    // need all collates to instantiate xlocinfo _Lstrcoll and _Lstrxfrm
+    // need all collates to instantiate <locale> _Lstrcoll and _Lstrxfrm
     auto cc   = has_facet<collate<char>>(loc);
     auto cw   = has_facet<collate<wchar_t>>(loc);
     auto cbnc = has_facet<collate_byname<char>>(loc);
