<regex>: Limit backreference parsing to single digit for basic regular expressions (#5167)

muellerj2 · StephanTLavavej · web-flow · commit bbd15badf3d9 · 2025-01-13T23:44:08.000-08:00
Co-authored-by: Stephan T. Lavavej &lt;stl@microsoft.com&gt;
diff --git a/stl/inc/regex b/stl/inc/regex
@@ -1187,8 +1187,6 @@ _NODISCARD bool operator!=(const match_results<_BidIt, _Alloc>& _Left, const mat
 }
 #endif // !_HAS_CXX20
 
-_INLINE_VAR constexpr unsigned int _BRE_MAX_GRP = 9U;
-
 _INLINE_VAR constexpr unsigned int _Bmp_max   = 256U; // must fit in an unsigned int
 _INLINE_VAR constexpr unsigned int _Bmp_shift = 3U;
 _INLINE_VAR constexpr unsigned int _Bmp_chrs  = 1U << _Bmp_shift; // # of bits to be stored in each char
@@ -1705,7 +1703,7 @@ private:
 
     // parsing
     int _Do_digits(int _Base, int _Count, regex_constants::error_type _Error_type);
-    bool _DecimalDigits(regex_constants::error_type _Error_type);
+    bool _DecimalDigits2(regex_constants::error_type _Error_type, int _Count = INT_MAX);
     void _HexDigits(int);
     bool _OctalDigits();
     void _Do_ex_class(_Meta_type);
@@ -3943,9 +3941,9 @@ int _Parser<_FwdIt, _Elem, _RxTraits>::_Do_digits(
 }
 
 template <class _FwdIt, class _Elem, class _RxTraits>
-bool _Parser<_FwdIt, _Elem, _RxTraits>::_DecimalDigits(
-    regex_constants::error_type _Error_type) { // check for decimal value
-    return _Do_digits(10, INT_MAX, _Error_type) != INT_MAX;
+bool _Parser<_FwdIt, _Elem, _RxTraits>::_DecimalDigits2(
+    const regex_constants::error_type _Error_type, const int _Count /* = INT_MAX */) { // check for decimal value
+    return _Do_digits(10, _Count, _Error_type) != _Count;
 }
 
 template <class _FwdIt, class _Elem, class _RxTraits>
@@ -4034,7 +4032,7 @@ _Prs_ret _Parser<_FwdIt, _Elem, _RxTraits>::_ClassEscape2() { // check for class
         return _Prs_chr;
     } else if ((_L_flags & _L_esc_wsd) && _CharacterClassEscape(false)) {
         return _Prs_set;
-    } else if (_DecimalDigits(regex_constants::error_escape)) { // check for invalid value
+    } else if (_DecimalDigits2(regex_constants::error_escape)) { // check for invalid value
         if (_Val != 0) {
             _Error(regex_constants::error_escape);
         }
@@ -4332,15 +4330,17 @@ bool _Parser<_FwdIt, _Elem, _RxTraits>::_CharacterEscape() { // check for valid
 
 template <class _FwdIt, class _Elem, class _RxTraits>
 void _Parser<_FwdIt, _Elem, _RxTraits>::_AtomEscape() { // check for valid atom escape
-    if ((_L_flags & _L_bckr) && _DecimalDigits(regex_constants::error_backref)) { // check for valid back reference
+    constexpr int _Bre_max_backref_digits = 1;
+    if ((_L_flags & _L_bckr)
+        && _DecimalDigits2(regex_constants::error_backref,
+            (_L_flags & _L_lim_bckr) ? _Bre_max_backref_digits : INT_MAX)) { // check for valid back reference
         if (_Val == 0) { // handle \0
             if (!(_L_flags & _L_bzr_chr)) {
                 _Error(regex_constants::error_escape);
             } else {
                 _Nfa._Add_char(static_cast<_Elem>(_Val));
             }
-        } else if (((_L_flags & _L_lim_bckr) && _BRE_MAX_GRP < static_cast<size_t>(_Val))
-                   || _Grp_idx < static_cast<size_t>(_Val) || _Finished_grps.size() <= static_cast<size_t>(_Val)
+        } else if (_Grp_idx < static_cast<size_t>(_Val) || _Finished_grps.size() <= static_cast<size_t>(_Val)
                    || !_Finished_grps[static_cast<size_t>(_Val)]) {
             _Error(regex_constants::error_backref);
         } else {
@@ -4364,7 +4364,7 @@ void _Parser<_FwdIt, _Elem, _RxTraits>::_Quantifier() { // check for quantifier
             _Max = 1;
         } else if (_Mchar == _Meta_lbr) { // check for valid bracketed value
             _Next();
-            if (!_DecimalDigits(regex_constants::error_badbrace)) {
+            if (!_DecimalDigits2(regex_constants::error_badbrace)) {
                 _Error(regex_constants::error_badbrace);
             }
 
@@ -4374,7 +4374,7 @@ void _Parser<_FwdIt, _Elem, _RxTraits>::_Quantifier() { // check for quantifier
             } else { // check for decimal constant following comma
                 _Next();
                 if (_Mchar != _Meta_rbr) {
-                    if (!_DecimalDigits(regex_constants::error_badbrace)) {
+                    if (!_DecimalDigits2(regex_constants::error_badbrace)) {
                         _Error(regex_constants::error_badbrace);
                     }
 
diff --git a/tests/std/tests/VSO_0000000_regex_use/test.cpp b/tests/std/tests/VSO_0000000_regex_use/test.cpp
@@ -669,6 +669,20 @@ void test_gh_5160() {
     neg_regex.should_search_fail(L"xxxYxx\x2009xxxZxxx"); // U+2009 THIN SPACE
 }
 
+void test_gh_5167() {
+    // GH-5167: Limit backreference parsing to single digit for basic regular expressions
+    g_regexTester.should_match("abab0", R"(\(ab*\)\10)", basic);
+    g_regexTester.should_match("abab0", R"(\(ab*\)\10)", grep);
+    g_regexTester.should_match("abbcdccdc5abb8", R"(\(ab*\)\([cd]*\)\25\18)", basic);
+    g_regexTester.should_match("abbcdccdc5abb8", R"(\(ab*\)\([cd]*\)\25\18)", grep);
+    g_regexTester.should_not_match("abbcdccdc5abb8", R"(\(ab*\)\([cd]*\)\15\28)", basic);
+    g_regexTester.should_not_match("abbcdccdc5abb8", R"(\(ab*\)\([cd]*\)\15\28)", grep);
+    g_regexTester.should_throw(R"(abc\1d)", error_backref, basic);
+    g_regexTester.should_throw(R"(abc\1d)", error_backref, grep);
+    g_regexTester.should_throw(R"(abc\10)", error_backref, basic);
+    g_regexTester.should_throw(R"(abc\10)", error_backref, grep);
+}
+
 int main() {
     test_dev10_449367_case_insensitivity_should_work();
     test_dev11_462743_regex_collate_should_not_disable_regex_icase();
@@ -699,6 +713,7 @@ int main() {
     test_gh_4995();
     test_gh_5058();
     test_gh_5160();
+    test_gh_5167();
 
     return g_regexTester.result();
 }
