<string>: Fix handling for char_traits<unicorn> (#4951)

frederick-vs-ja · StephanTLavavej · web-flow · commit 5af09e7b142a · 2024-09-19T13:46:46.000-07:00
Co-authored-by: Stephan T. Lavavej &lt;stl@nuwen.net&gt;
diff --git a/stl/inc/__msvc_string_view.hpp b/stl/inc/__msvc_string_view.hpp
@@ -331,6 +331,17 @@ template <>
 struct char_traits<unsigned short> : _WChar_traits<unsigned short> {};
 #endif // defined(_CRTBLD)
 
+// signed char and other unsigned integral types are supported as an extension.
+// Use of other arithmetic types and nullptr_t should be rejected.
+template <class _Ty>
+constexpr bool _Is_implementation_handled_char_like_type = is_arithmetic_v<_Ty> || is_null_pointer_v<_Ty>;
+
+template <class>
+constexpr bool _Is_implementation_handled_char_traits = false;
+template <class _Elem>
+constexpr bool _Is_implementation_handled_char_traits<char_traits<_Elem>> =
+    _Is_implementation_handled_char_like_type<_Elem>;
+
 #if defined(__cpp_char8_t) && !defined(__clang__) && !defined(__EDG__)
 #define _HAS_U8_INTRINSICS 1
 #else // ^^^ Use intrinsics for char8_t / don't use said intrinsics vvv
@@ -675,9 +686,9 @@ class _String_bitmap { // _String_bitmap for character types
 template <class _Elem>
 class _String_bitmap<_Elem, false> { // _String_bitmap for wchar_t/unsigned short/char16_t/char32_t/etc. types
 public:
-    static_assert(is_unsigned_v<_Elem>,
-        "Standard char_traits is only provided for char, wchar_t, char16_t, and char32_t. See N4950 [char.traits]. "
-        "Visual C++ accepts other unsigned integral types as an extension.");
+    static_assert(is_unsigned_v<_Elem>, "Standard char_traits is only provided for char, wchar_t, char8_t, char16_t, "
+                                        "and char32_t. See N4988 [char.traits]. "
+                                        "Visual C++ accepts other unsigned integral types as an extension.");
 
     constexpr bool _Mark(const _Elem* _First, const _Elem* const _Last) noexcept {
         // mark this bitmap such that the characters in [_First, _Last) are intended to match
@@ -702,7 +713,7 @@ class _String_bitmap<_Elem, false> { // _String_bitmap for wchar_t/unsigned shor
     bool _Matches[256] = {};
 };
 
-template <class _Traits, bool _Special = _Is_specialization_v<_Traits, char_traits>>
+template <class _Traits, bool _Special = _Is_implementation_handled_char_traits<_Traits>>
 constexpr size_t _Traits_find_first_of(_In_reads_(_Hay_size) const _Traits_ptr_t<_Traits> _Haystack,
     const size_t _Hay_size, const size_t _Start_at, _In_reads_(_Needle_size) const _Traits_ptr_t<_Traits> _Needle,
     const size_t _Needle_size) noexcept {
@@ -764,7 +775,7 @@ constexpr size_t _Traits_find_first_of(_In_reads_(_Hay_size) const _Traits_ptr_t
     return static_cast<size_t>(-1); // no match
 }
 
-template <class _Traits, bool _Special = _Is_specialization_v<_Traits, char_traits>>
+template <class _Traits, bool _Special = _Is_implementation_handled_char_traits<_Traits>>
 constexpr size_t _Traits_find_last_of(_In_reads_(_Hay_size) const _Traits_ptr_t<_Traits> _Haystack,
     const size_t _Hay_size, const size_t _Start_at, _In_reads_(_Needle_size) const _Traits_ptr_t<_Traits> _Needle,
     const size_t _Needle_size) noexcept {
@@ -802,7 +813,7 @@ constexpr size_t _Traits_find_last_of(_In_reads_(_Hay_size) const _Traits_ptr_t<
     return static_cast<size_t>(-1); // no match
 }
 
-template <class _Traits, bool _Special = _Is_specialization_v<_Traits, char_traits>>
+template <class _Traits, bool _Special = _Is_implementation_handled_char_traits<_Traits>>
 constexpr size_t _Traits_find_first_not_of(_In_reads_(_Hay_size) const _Traits_ptr_t<_Traits> _Haystack,
     const size_t _Hay_size, const size_t _Start_at, _In_reads_(_Needle_size) const _Traits_ptr_t<_Traits> _Needle,
     const size_t _Needle_size) noexcept {
@@ -851,7 +862,7 @@ constexpr size_t _Traits_find_not_ch(_In_reads_(_Hay_size) const _Traits_ptr_t<_
     return static_cast<size_t>(-1); // no match
 }
 
-template <class _Traits, bool _Special = _Is_specialization_v<_Traits, char_traits>>
+template <class _Traits, bool _Special = _Is_implementation_handled_char_traits<_Traits>>
 constexpr size_t _Traits_find_last_not_of(_In_reads_(_Hay_size) const _Traits_ptr_t<_Traits> _Haystack,
     const size_t _Hay_size, const size_t _Start_at, _In_reads_(_Needle_size) const _Traits_ptr_t<_Traits> _Needle,
     const size_t _Needle_size) noexcept {
diff --git a/stl/inc/regex b/stl/inc/regex
@@ -193,19 +193,29 @@ struct _Char_traits_lt {
     }
 };
 
+// signed char and other unsigned integral types are supported as an extension.
+template <class _Ty>
+constexpr bool _Is_predefined_char_like_type = _Is_character<_Ty>::value || is_unsigned_v<_Ty>;
+
 // library-provided char_traits::eq behaves like equal_to<_Elem>
-// TRANSITION: This should not be activated for user-defined specializations of char_traits
 template <class _Elem>
 constexpr bool _Can_memcmp_elements_with_pred<_Elem, _Elem, _Char_traits_eq<char_traits<_Elem>>> =
-    _Can_memcmp_elements<_Elem, _Elem>;
+    _Is_predefined_char_like_type<_Elem> && _Can_memcmp_elements<_Elem, _Elem>;
 
-// library-provided char_traits::lt behaves like less<make_unsigned_t<_Elem>>
-// TRANSITION: This should not be activated for user-defined specializations of char_traits
-template <class _Elem>
-struct _Lex_compare_memcmp_classify_pred<_Elem, _Elem, _Char_traits_lt<char_traits<_Elem>>> {
+template <class _Elem, bool = _Is_predefined_char_like_type<_Elem>>
+struct _Lex_compare_memcmp_classify_pred_for_char_traits_lt {
     using _UElem = make_unsigned_t<_Elem>;
     using _Pred  = conditional_t<_Lex_compare_memcmp_classify_elements<_UElem, _UElem>, less<int>, void>;
 };
+template <class _Elem>
+struct _Lex_compare_memcmp_classify_pred_for_char_traits_lt<_Elem, false> {
+    using _Pred = void;
+};
+
+// library-provided char_traits::lt behaves like less<make_unsigned_t<_Elem>>
+template <class _Elem>
+struct _Lex_compare_memcmp_classify_pred<_Elem, _Elem, _Char_traits_lt<char_traits<_Elem>>>
+    : _Lex_compare_memcmp_classify_pred_for_char_traits_lt<_Elem> {};
 
 template <class _RxTraits>
 struct _Cmp_cs { // functor to compare two character values for equality
diff --git a/tests/std/test.lst b/tests/std/test.lst
@@ -250,6 +250,7 @@ tests\GH_004618_mixed_operator_usage_keeps_statistical_properties
 tests\GH_004618_normal_distribution_avoids_resets
 tests\GH_004657_expected_constraints_permissive
 tests\GH_004845_logical_operator_traits_with_non_bool_constant
+tests\GH_004930_char_traits_user_specialization
 tests\LWG2381_num_get_floating_point
 tests\LWG2597_complex_branch_cut
 tests\LWG3018_shared_ptr_function
diff --git a/tests/std/tests/GH_004930_char_traits_user_specialization/env.lst b/tests/std/tests/GH_004930_char_traits_user_specialization/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_004930_char_traits_user_specialization/test.cpp b/tests/std/tests/GH_004930_char_traits_user_specialization/test.cpp
