<regex>: Correct characters not matched by special character dot (#…

…5192) Co-authored-by: Stephan T. Lavavej <[email protected]>
microsoft · Jan 14, 2025 · 247c51f · 247c51f
1 parent 3b6d29c
commit 247c51f
Show file tree

Hide file tree

Showing 3 changed files with 60 additions and 13 deletions.
diff --git a/stl/inc/regex b/stl/inc/regex
@@ -79,6 +79,8 @@ enum _Meta_type : int { // meta character representations for parser
     _Meta_nl    = '\n',
     _Meta_cr    = '\r',
     _Meta_bsp   = '\b',
+    _Meta_ls    = L'\u2028',
+    _Meta_ps    = L'\u2029',
     _Meta_chr   = 0,
 
     _Esc_bsl      = '\\',
@@ -3523,10 +3525,23 @@ bool _Matcher<_BidIt, _Elem, _RxTraits, _It>::_Match_pat(_Node_base* _Nx) { // c
             break;
 
         case _N_dot:
-            if (_Tgt_state._Cur == _End || *_Tgt_state._Cur == _Meta_nl || *_Tgt_state._Cur == _Meta_cr) {
+            if (_Tgt_state._Cur == _End) {
                 _Failed = true;
             } else {
-                ++_Tgt_state._Cur;
+                const _Elem _Ch = *_Tgt_state._Cur;
+                if (_Sflags
+                    & (regex_constants::basic | regex_constants::extended | regex_constants::grep
+                        | regex_constants::egrep | regex_constants::awk)) {
+                    if (_Ch == _Elem()) {
+                        _Failed = true;
+                    }
+                } else if (_Ch == _Meta_nl || _Ch == _Meta_cr || _Ch == _Meta_ls || _Ch == _Meta_ps) { // ECMAScript
+                    _Failed = true;
+                }
+
+                if (!_Failed) {
+                    ++_Tgt_state._Cur;
+                }
             }
 
             break;

diff --git a/tests/std/tests/VSO_0000000_regex_use/test.cpp b/tests/std/tests/VSO_0000000_regex_use/test.cpp
@@ -683,6 +683,42 @@ void test_gh_5167() {
     g_regexTester.should_throw(R"(abc\10)", error_backref, grep);
 }
 
+void test_gh_5192() {
+    // GH-5192: Correct characters not matched by special character dot
+    for (const syntax_option_type option : {
+             regex_constants::basic,
+             regex_constants::extended,
+             regex_constants::awk,
+             regex_constants::grep,
+             regex_constants::egrep,
+         }) {
+        const test_regex caretDotStar(&g_regexTester, "^.*", option);
+        caretDotStar.should_search_match("abc\nd\re\0f"s, "abc\nd\re"s);
+        caretDotStar.should_search_match("abcd\re\ngh\0i"s, "abcd\re\ngh"s);
+
+        const test_wregex wCaretDotStar(&g_regexTester, L"^.*", option);
+        wCaretDotStar.should_search_match(L"abc\nd\re\0f"s, L"abc\nd\re"s);
+        wCaretDotStar.should_search_match(L"abcd\re\ngh\0i"s, L"abcd\re\ngh"s);
+        wCaretDotStar.should_search_match(L"abc\u2028d\ne\0f"s, L"abc\u2028d\ne"s); // U+2028 LINE SEPARATOR
+        wCaretDotStar.should_search_match(L"abc\u2029d\ne\0f"s, L"abc\u2029d\ne"s); // U+2029 PARAGRAPH SEPARATOR
+    }
+
+    for (const syntax_option_type option : {
+             regex_constants::ECMAScript,
+             syntax_option_type(),
+         }) {
+        const test_regex caretDotStar(&g_regexTester, "^.*", option);
+        caretDotStar.should_search_match("ab\0c\nd\re\0f"s, "ab\0c"s);
+        caretDotStar.should_search_match("ab\0cd\re\ngh\0i"s, "ab\0cd"s);
+
+        const test_wregex wCaretDotStar(&g_regexTester, L"^.*", option);
+        wCaretDotStar.should_search_match(L"abc\0\nd\re\0f"s, L"abc\0"s);
+        wCaretDotStar.should_search_match(L"ab\0cd\re\ngh\0i"s, L"ab\0cd"s);
+        wCaretDotStar.should_search_match(L"ab\0c\u2028d\ne\0f"s, L"ab\0c"s); // U+2028 LINE SEPARATOR
+        wCaretDotStar.should_search_match(L"a\0bc\u2029d\ne\0f"s, L"a\0bc"s); // U+2029 PARAGRAPH SEPARATOR
+    }
+}
+
 int main() {
     test_dev10_449367_case_insensitivity_should_work();
     test_dev11_462743_regex_collate_should_not_disable_regex_icase();
@@ -714,6 +750,7 @@ int main() {
     test_gh_5058();
     test_gh_5160();
     test_gh_5167();
+    test_gh_5192();
 
     return g_regexTester.result();
 }
diff --git a/tests/tr1/tests/regex3/test.cpp b/tests/tr1/tests/regex3/test.cpp
@@ -321,17 +321,12 @@ static void test_uncoveredgrammar() {
     STDString str2(T("Prime number"));
     STDString str3(T("aaaqxzbbb"));
     STDString str4(T("aaa\nxzbbb"));
-    static const STD regex_constants::syntax_option_type flag[6] = {
-        STD regex_constants::ECMAScript,
-        STD regex_constants::basic,
-        STD regex_constants::extended,
-        STD regex_constants::awk,
-        STD regex_constants::grep,
-        STD regex_constants::egrep,
-    };
-    for (int i = 0; i <= 5; i++) {
-        rx1.assign(T(".*"), flag[i]);
-        rx2.assign(T("aaa...bbb"), flag[i]);
+    for (const STD regex_constants::syntax_option_type flag : {
+             STD regex_constants::ECMAScript,
+             STD regex_constants::syntax_option_type(),
+         }) {
+        rx1.assign(T(".*"), flag);
+        rx2.assign(T("aaa...bbb"), flag);
         CHECK(!STD regex_match(str1, rx1));
         CHECK(STD regex_match(str2, rx1));
         CHECK(STD regex_match(str3, rx2));