fix: regex parsing with \/

avast · Dec 14, 2024 · b8554dc · b8554dc
1 parent dcb256b
commit b8554dc
Show file tree

Hide file tree

Showing 2 changed files with 36 additions and 2 deletions.
diff --git a/src/parser/parser_driver.cpp b/src/parser/parser_driver.cpp
@@ -402,6 +402,9 @@ void ParserDriver::defineTokens()
 	_parser.token(R"([^\\\[\(\)\|\$\.\^\+\+*\?])").states("$regexp").symbol("REGEXP_CHAR").description("regexp character").action([](std::string_view str) -> Value {
 		return std::string{str};
 	});
+	_parser.token(R"(\\\/)").states("$regexp").symbol("REGEXP_CHAR").description("regexp character").action([](std::string_view str) -> Value {
+		return std::string{str};
+	});
 	_parser.token(R"(\\w)").states("$regexp").symbol("REGEXP_WORD_CHAR").description("regexp \\w").action([](std::string_view) -> Value { return {};});
 	_parser.token(R"(\\W)").states("$regexp").symbol("REGEXP_NON_WORD_CHAR").description("regexp \\W").action([](std::string_view) -> Value { return {};});
 	_parser.token(R"(\\s)").states("$regexp").symbol("REGEXP_SPACE").description("regexp \\s").action([](std::string_view) -> Value { return {};});

diff --git a/tests/cpp/parser_tests.cpp b/tests/cpp/parser_tests.cpp
@@ -1139,8 +1139,8 @@ rule regexp_with_just_chars
 	EXPECT_EQ("regexp_with_just_chars", rule->getName());
 	EXPECT_EQ(Rule::Modifier::None, rule->getModifier());
 
-	auto strings = rule->getStrings()
-;	ASSERT_EQ(1u, strings.size());
+	auto strings = rule->getStrings();
+	ASSERT_EQ(1u, strings.size());
 
 	auto regexp0 = strings[0];
 	EXPECT_TRUE(regexp0->isRegexp());
@@ -1150,6 +1150,37 @@ rule regexp_with_just_chars
 	EXPECT_EQ(input_text, driver.getParsedFile().getTextFormatted());
 }
 
+TEST_F(ParserTests,
+RegexpWithEscapeSequences) {
+	prepareInput(
+R"(
+rule regexp_with_escape_sequences
+{
+	strings:
+		$1 = /http:\/\/someone\.doingevil\.com/
+	condition:
+		true
+}
+)");
+
+	EXPECT_TRUE(driver.parse(input));
+	ASSERT_EQ(1u, driver.getParsedFile().getRules().size());
+
+	const auto& rule = driver.getParsedFile().getRules()[0];
+	EXPECT_EQ("regexp_with_escape_sequences", rule->getName());
+	EXPECT_EQ(Rule::Modifier::None, rule->getModifier());
+
+	auto strings = rule->getStrings();
+	ASSERT_EQ(1u, strings.size());
+
+	auto regexp0 = strings[0];
+	EXPECT_TRUE(regexp0->isRegexp());
+	EXPECT_EQ("$1", regexp0->getIdentifier());
+	EXPECT_EQ("/http:\\/\\/someone\\.doingevil\\.com/", regexp0->getText());
+
+	EXPECT_EQ(input_text, driver.getParsedFile().getTextFormatted());
+}
+
 TEST_F(ParserTests,
 MultipleRegexpsWithJustCharsWorks) {
 	prepareInput(