<regex>: Permit escaping of closing brackets in POSIX regexes (#5399)

muellerj2 · web-flow · commit c5ed364866df · 2025-04-22T14:03:02.000-07:00
diff --git a/stl/inc/regex b/stl/inc/regex
@@ -4403,6 +4403,7 @@ bool _Parser<_FwdIt, _Elem, _RxTraits>::_IsIdentityEscape(bool _In_character_cla
         return true;
     case _Meta_dot:
     case _Meta_lsq:
+    case _Meta_rsq:
     case _Meta_star:
     case _Meta_caret:
     case _Meta_dlr:
diff --git a/tests/std/tests/GH_005244_regex_escape_sequences/test.cpp b/tests/std/tests/GH_005244_regex_escape_sequences/test.cpp
@@ -470,6 +470,13 @@ void test_gh_5244_atomescape_posix_common(syntax_option_type option) {
     check_atomescape_identityescape("^", option);
     check_atomescape_identityescape("$", option);
 
+    // Even though [ is special, ] is not,
+    // so the interpretation of the escape sequence \] is undefined
+    // according to the POSIX standard referenced in the C++ standard.
+    // But we treat \] as an identity escape in line with
+    // more recent versions of the POSIX standard.
+    check_atomescape_identityescape("]", option);
+
     // Sections on "BRE Special Characters" and "ERE Special Characters":
     // escaping ordinary characters is undefined -> reject
     g_regexTester.should_throw(R"(\B)", error_escape, option);
@@ -483,9 +490,6 @@ void test_gh_5244_atomescape_posix_common(syntax_option_type option) {
     g_regexTester.should_throw(R"(\W)", error_escape, option);
     g_regexTester.should_throw(R"(\s)", error_escape, option);
     g_regexTester.should_throw(R"(\S)", error_escape, option);
-
-    // while [ is special, ] is not
-    g_regexTester.should_throw(R"(\])", error_escape, option);
 }
 
 void test_gh_5244_atomescape_posix_not_awk(syntax_option_type option) {
@@ -524,8 +528,6 @@ void test_gh_5244_atomescape_basic_or_grep(syntax_option_type option) {
 }
 
 void test_gh_5244_atomescape_extended_egrep_awk(syntax_option_type option) {
-    test_gh_5244_atomescape_posix_common(option);
-
     // check that the parser accepts escaped characters
     // that are only special in extended regexes
     check_atomescape_identityescape("+", option);
@@ -557,6 +559,7 @@ void test_gh_5244_atomescape_extended_or_egrep(syntax_option_type option) {
 
 void test_gh_5244_atomescape_awk() {
     test_gh_5244_atomescape_extended_egrep_awk(awk);
+    test_gh_5244_atomescape_posix_common(awk);
 
     // awk-only escapes
     check_atomescape_controlescape("\a", "a", awk);
