From 5600f2b1b16abfbbb91be93ac6641154c7b9d322 Mon Sep 17 00:00:00 2001
From: Sainan <sainan@calamity.gg>
Date: Tue, 2 Jul 2024 06:02:17 +0200
Subject: [PATCH] Regex: Fix result containing optional capturing group that
 did not match

---
 CLI/cli_test.cpp      |  2 ++
 soup/Regex.cpp        | 29 +++++------------------------
 soup/RegexMatcher.hpp | 30 ++++++++++++++++++++++++++++--
 3 files changed, 35 insertions(+), 26 deletions(-)

diff --git a/CLI/cli_test.cpp b/CLI/cli_test.cpp
index e10c31c6..b8801c7a 100644
--- a/CLI/cli_test.cpp
+++ b/CLI/cli_test.cpp
@@ -870,6 +870,8 @@ spanning over multiple lines */
 
 		assert(Regex("a(.*)z").match("az").toString() == R"(0="az", 1="")");
 		assert(Regex("a(.*)z").match("abz").toString() == R"(0="abz", 1="b")");
+
+		assert(Regex("(A)(B)?").match("A").toString() == R"(0="A", 1="A")");
 	});
 
 	test("MessageStream", []
diff --git a/soup/Regex.cpp b/soup/Regex.cpp
index 010c2957..955b00c5 100644
--- a/soup/Regex.cpp
+++ b/soup/Regex.cpp
@@ -84,31 +84,9 @@ NAMESPACE_SOUP
 		while (m.c != nullptr)
 		{
 #if REGEX_DEBUG_MATCH
-			std::cout << m.c->toString() << ": ";
+			std::cout << m.c->toString() << " (g " << m.c->group->index << "): ";
 #endif
 
-			// Insert missing capturing groups
-			for (auto g = m.c->group; g; g = g->parent)
-			{
-				if (g->lookahead_or_lookbehind)
-				{
-					break;
-				}
-				if (g->isNonCapturing())
-				{
-					continue;
-				}
-				//std::cout << "group " << g->index << "; ";
-				while (g->index >= m.result.groups.size())
-				{
-					m.result.groups.emplace_back(std::nullopt);
-				}
-				if (!m.result.groups.at(g->index).has_value())
-				{
-					m.result.groups.at(g->index) = RegexMatchedGroup{ g->name, m.it, m.it };
-				}
-			}
-
 			if (m.c->rollback_transition)
 			{
 #if REGEX_DEBUG_MATCH
@@ -117,6 +95,8 @@ NAMESPACE_SOUP
 				m.saveRollback(m.c->rollback_transition);
 			}
 
+			m.insertMissingCapturingGroups(m.c->group);
+
 			if (reset_capture)
 			{
 				reset_capture = false;
@@ -180,7 +160,7 @@ NAMESPACE_SOUP
 #if REGEX_DEBUG_MATCH
 				std::cout << "; rolling back\n";
 #endif
-				m.restoreRollback();
+				const RegexGroup* g = m.restoreRollback();
 				SOUP_ASSERT(!m.shouldSaveCheckpoint());
 				reset_capture = m.shouldResetCapture();
 				if (m.c == RegexConstraint::ROLLBACK_TO_SUCCESS)
@@ -190,6 +170,7 @@ NAMESPACE_SOUP
 #endif
 					break;
 				}
+				m.insertMissingCapturingGroups(g);
 				continue;
 			}
 
diff --git a/soup/RegexMatcher.hpp b/soup/RegexMatcher.hpp
index 746cac77..e25ad5d2 100644
--- a/soup/RegexMatcher.hpp
+++ b/soup/RegexMatcher.hpp
@@ -14,6 +14,7 @@ NAMESPACE_SOUP
 	{
 		struct RollbackPoint
 		{
+			const RegexGroup* g;
 			const RegexConstraint* c;
 			const char* it;
 			RegexMatchResult result{};
@@ -42,15 +43,17 @@ NAMESPACE_SOUP
 
 		void saveRollback(const RegexConstraint* rollback_transition)
 		{
-			rollback_points.emplace_back(RollbackPoint{ rollback_transition, it, result });
+			rollback_points.emplace_back(RollbackPoint{ c->group, rollback_transition, it, result });
 		}
 
-		void restoreRollback()
+		[[nodiscard]] const RegexGroup* restoreRollback()
 		{
+			const RegexGroup* g = rollback_points.back().g;
 			c = rollback_points.back().c;
 			it = rollback_points.back().it;
 			result = std::move(rollback_points.back().result);
 			rollback_points.pop_back();
+			return g;
 		}
 
 		bool shouldSaveCheckpoint() noexcept
@@ -84,5 +87,28 @@ NAMESPACE_SOUP
 			it = checkpoints.back();
 			checkpoints.pop_back();
 		}
+
+		void insertMissingCapturingGroups(const RegexGroup* g)
+		{
+			for (; g; g = g->parent)
+			{
+				if (g->lookahead_or_lookbehind)
+				{
+					break;
+				}
+				if (g->isNonCapturing())
+				{
+					continue;
+				}
+				while (g->index >= this->result.groups.size())
+				{
+					this->result.groups.emplace_back(std::nullopt);
+				}
+				if (!this->result.groups.at(g->index).has_value())
+				{
+					this->result.groups.at(g->index) = RegexMatchedGroup{ g->name, this->it, this->it };
+				}
+			}
+		}
 	};
 }