From bb7ff0c0c2d20ce795feaf9e9647b30ad739367a Mon Sep 17 00:00:00 2001
From: Robin Linden <dev@robinlinden.eu>
Date: Tue, 21 Jan 2025 21:55:33 +0100
Subject: [PATCH] css2: Handle escape sequences more correctly

---
 css2/tokenizer.cpp      | 11 +++++++++++
 css2/tokenizer.h        |  1 +
 css2/tokenizer_test.cpp | 14 ++++++++++++++
 3 files changed, 26 insertions(+)

diff --git a/css2/tokenizer.cpp b/css2/tokenizer.cpp
index 2b2b7ada..0f03a3e8 100644
--- a/css2/tokenizer.cpp
+++ b/css2/tokenizer.cpp
@@ -56,6 +56,8 @@ std::string_view to_string(ParseError e) {
             return "EofInEscapeSequence";
         case ParseError::EofInString:
             return "EofInString";
+        case ParseError::InvalidEscapeSequence:
+            return "InvalidEscapeSequence";
         case ParseError::NewlineInString:
             return "NewlineInString";
     }
@@ -177,6 +179,15 @@ void Tokenizer::run() {
                     case '[':
                         emit(OpenSquareToken{});
                         continue;
+                    case '\\':
+                        if (is_valid_escape_sequence('\\', peek_input(0))) {
+                            reconsume_in(State::IdentLike);
+                            continue;
+                        }
+
+                        emit(ParseError::InvalidEscapeSequence);
+                        emit(DelimToken{'\\'});
+                        continue;
                     case ']':
                         emit(CloseSquareToken{});
                         continue;
diff --git a/css2/tokenizer.h b/css2/tokenizer.h
index d58b9115..d144a687 100644
--- a/css2/tokenizer.h
+++ b/css2/tokenizer.h
@@ -35,6 +35,7 @@ enum class ParseError : std::uint8_t {
     EofInComment,
     EofInEscapeSequence,
     EofInString,
+    InvalidEscapeSequence,
     NewlineInString,
 };
 
diff --git a/css2/tokenizer_test.cpp b/css2/tokenizer_test.cpp
index 464333f9..c9a39924 100644
--- a/css2/tokenizer_test.cpp
+++ b/css2/tokenizer_test.cpp
@@ -316,6 +316,7 @@ int main() {
     s.add_test("at keyword start, but with bad escape", [](etest::IActions &a) {
         auto output = run_tokenizer(a, "@\\\n");
         expect_token(output, DelimToken{'@'});
+        expect_error(output, ParseError::InvalidEscapeSequence);
         expect_token(output, DelimToken{'\\'});
         expect_token(output, WhitespaceToken{});
     });
@@ -638,6 +639,19 @@ int main() {
     s.add_test("hash token: invalid escape", [](etest::IActions &a) {
         auto output = run_tokenizer(a, "#\\\n");
         expect_token(output, DelimToken{'#'});
+        expect_error(output, ParseError::InvalidEscapeSequence);
+        expect_token(output, DelimToken{'\\'});
+        expect_token(output, WhitespaceToken{});
+    });
+
+    s.add_test("\\: ident-like", [](etest::IActions &a) {
+        auto output = run_tokenizer(a, "\\Hallo");
+        expect_token(output, IdentToken{"Hallo"});
+    });
+
+    s.add_test("\\: invalid escape", [](etest::IActions &a) {
+        auto output = run_tokenizer(a, "\\\n");
+        expect_error(output, ParseError::InvalidEscapeSequence);
         expect_token(output, DelimToken{'\\'});
         expect_token(output, WhitespaceToken{});
     });