<regex>: Make negated character classes not match characters not included in the negated character class (#5214)

Co-authored-by: Stephan T. Lavavej <[email protected]>

Author: muellerj2

stl/inc/regex

@@ -1524,7 +1524,7 @@ public:
     void _Add_class();
     void _Add_char_to_class(_Elem _Ch);
     void _Add_range2(_Elem, _Elem);
-    void _Add_named_class(_Regex_traits_base::char_class_type, bool = false);
+    void _Add_named_class(typename _RxTraits::char_class_type, bool);
     void _Add_equiv(_FwdIt, _FwdIt, _Difft);
     void _Add_coll(_FwdIt, _FwdIt, _Difft);
     _Node_base* _Begin_group();
@@ -1548,7 +1548,7 @@ private:
     bool _Beg_expr(_Node_base*) const;
     void _Add_char_to_bitmap(_Elem _Ch);
     void _Add_char_to_array(_Elem _Ch);
-    void _Add_elts(_Node_class<_Elem, _RxTraits>*, _Regex_traits_base::char_class_type, bool);
+    void _Add_elts(_Node_class<_Elem, _RxTraits>*, typename _RxTraits::char_class_type, bool);
     void _Char_to_elts(_FwdIt, _FwdIt, _Difft, _Sequence<_Elem>**);
 
     _Root_node* _Root;
@@ -2935,7 +2935,7 @@ void _Builder<_FwdIt, _Elem, _RxTraits>::_Add_range2(const _Elem _Arg0, const _E
 
 template <class _FwdIt, class _Elem, class _RxTraits>
 void _Builder<_FwdIt, _Elem, _RxTraits>::_Add_elts(
-    _Node_class<_Elem, _RxTraits>* _Node, _Regex_traits_base::char_class_type _Cl, bool _Negate) {
+    _Node_class<_Elem, _RxTraits>* _Node, typename _RxTraits::char_class_type _Cl, bool _Negate) {
     // add characters in named class to set
     for (unsigned int _Ch = 0; _Ch < _Bmp_max; ++_Ch) { // add elements or their inverse
         bool _Matches = _Traits.isctype(static_cast<_Elem>(_Ch), _Cl);
@@ -2950,12 +2950,12 @@ void _Builder<_FwdIt, _Elem, _RxTraits>::_Add_elts(
 }
 
 template <class _FwdIt, class _Elem, class _RxTraits>
-void _Builder<_FwdIt, _Elem, _RxTraits>::_Add_named_class(typename _Regex_traits_base::char_class_type _Cl,
+void _Builder<_FwdIt, _Elem, _RxTraits>::_Add_named_class(typename _RxTraits::char_class_type _Cl,
     bool _Negate) { // add contents of named class to bracket expression
     _Node_class<_Elem, _RxTraits>* _Node = static_cast<_Node_class<_Elem, _RxTraits>*>(_Current);
     _Add_elts(_Node, _Cl, _Negate);
-    if (_Bmp_max < static_cast<unsigned int>(_STD _Max_limit<_Elem>())) {
-        _Node->_Classes = static_cast<_Regex_traits_base::char_class_type>(_Node->_Classes | _Cl);
+    if (_Bmp_max <= _STD _Max_limit<typename _RxTraits::_Uelem>() && !_Negate) {
+        _Node->_Classes = static_cast<typename _RxTraits::char_class_type>(_Node->_Classes | _Cl);
     }
 }
 
@@ -4023,7 +4023,7 @@ void _Parser<_FwdIt, _Elem, _RxTraits>::_Do_ex_class(
             _Error(regex_constants::error_ctype);
         }
 
-        _Nfa._Add_named_class(_Cls);
+        _Nfa._Add_named_class(_Cls, false);
     } else if (_End_arg == _Meta_equal) { // process =
         if (_Beg == _Pat) {
             _Error(regex_constants::error_collate);

tests/std/tests/VSO_0000000_regex_use/test.cpp

@@ -730,6 +730,24 @@ void test_gh_5192() {
     }
 }
 
+void test_gh_5214() {
+    // GH-5214 makes negated character class escapes not match characters not included in the negated character class
+    {
+        const test_wregex neg_word_regex(&g_regexTester, LR"([\W])");
+        neg_word_regex.should_search_fail(L"\u0100"); // U+0100 LATIN CAPITAL LETTER A WITH MACRON
+    }
+
+    {
+        const test_wregex neg_space_regex(&g_regexTester, LR"([\S])");
+        neg_space_regex.should_search_fail(L"\u2028"); // U+2028 LINE SEPARATOR
+    }
+
+    {
+        const test_wregex neg_digit_regex(&g_regexTester, LR"([\D])");
+        neg_digit_regex.should_search_fail(L"\u0662"); // U+0662 ARABIC-INDIC DIGIT TWO
+    }
+}
+
 int main() {
     test_dev10_449367_case_insensitivity_should_work();
     test_dev11_462743_regex_collate_should_not_disable_regex_icase();
@@ -763,6 +781,7 @@ int main() {
     test_gh_5160();
     test_gh_5167();
     test_gh_5192();
+    test_gh_5214();
 
     return g_regexTester.result();
 }