skip escaped characters in special comment scanner

ethereum · Jul 1, 2024 · 96c162f · 96c162f
1 parent fc80f66
commit 96c162f
Show file tree

Hide file tree

Showing 3 changed files with 22 additions and 41 deletions.
diff --git a/liblangutil/Scanner.cpp b/liblangutil/Scanner.cpp
@@ -759,13 +759,12 @@ void Scanner::scanToken()
  m_tokens[NextNext].extendedTokenInfo = std::make_tuple(m, n);
 }
 
-bool Scanner::scanEscape(bool const _rejectInvalidEscapes)
+bool Scanner::scanEscape()
 {
  char c = m_char;
 
- // Normally we ignore the slash just before a newline since it's meaningless.
- // In the case of not rejecting invalid escapes, though, we preserve it.
- if (_rejectInvalidEscapes && tryScanEndOfLine())
+ // Skip escaped newlines.
+ if (tryScanEndOfLine())
  return true;
  advance();
 
@@ -787,42 +786,17 @@ bool Scanner::scanEscape(bool const _rejectInvalidEscapes)
  case 'u':
  {
  if (auto const codepoint = scanUnicode(); codepoint.has_value())
- {
  addUnicodeAsUTF8(*codepoint);
- return true;
- }
- else if (_rejectInvalidEscapes)
- return false;
  else
- {
- addLiteralChar('\\');
- addLiteralChar(c);
- return true;
- }
+ return false;
+ return true;
  }
  case 'x':
- if (_rejectInvalidEscapes)
- {
- if (!scanHexByte(c))
- return false;
- }
- else
- if (!scanHexByte(c))
- {
- addLiteralChar('\\');
- addLiteralChar(c);
- return true;
- }
+ if (!scanHexByte(c))
+ return false;
  break;
  default:
- if (_rejectInvalidEscapes)
- return false;
- else
- {
- addLiteralChar('\\');
- addLiteralChar(c);
- return true;
- }
+ return false;
  }
 
  addLiteralChar(c);
@@ -864,9 +838,7 @@ Token Scanner::scanString(bool const _isUnicode)
  {
  if (isSourcePastEndOfInput())
  return setError(ScannerError::IllegalEscapeSequence);
- bool const validEscape = scanEscape(false /* _rejectInvalidEscapes */);
- // there are no invalid escapes in special comments except unterminated backslash at eos
- solAssert(validEscape);
+ advance();
  }
  else
  addLiteralChar(c);

diff --git a/liblangutil/Scanner.h b/liblangutil/Scanner.h
@@ -246,7 +246,7 @@ class Scanner
  /// Scans an escape-sequence which is part of a string and adds the
  /// decoded character to the current literal. Returns true if a pattern
  /// is scanned.
- bool scanEscape(bool _rejectInvalidEscapes = true);
+ bool scanEscape();
 
  /// @returns true iff we are currently positioned at a unicode line break.
  bool isUnicodeLinebreak();

diff --git a/test/liblangutil/Scanner.cpp b/test/liblangutil/Scanner.cpp
@@ -1029,7 +1029,7 @@ BOOST_AUTO_TEST_CASE(yul_function_with_whitespace)
 BOOST_AUTO_TEST_CASE(special_comment_with_invalid_escapes)
 {
  std::string input(R"("test\x\f\g\u\g\a\u\g\a12\uö\xyoof")");
- std::string expectedOutput(R"(test\x\f\g\u\g\a\u\g\a12\uö\xyoof)");
+ std::string expectedOutput(R"(test12öyoof)");
  CharStream stream(input, "");
  Scanner scanner(stream, ScannerKind::SpecialComment);
  BOOST_REQUIRE(scanner.currentToken() == Token::StringLiteral);
@@ -1039,8 +1039,7 @@ BOOST_AUTO_TEST_CASE(special_comment_with_invalid_escapes)
 BOOST_AUTO_TEST_CASE(special_comment_with_valid_and_invalid_escapes)
 {
  std::string input(R"("test\n\x61\t\u01A9test\f")");
- std::string expectedOutput(R"(test
-a Ʃtest\f)");
+ std::string expectedOutput(R"(test6101A9test)");
  CharStream stream(input, "");
  Scanner scanner(stream, ScannerKind::SpecialComment);
  BOOST_REQUIRE(scanner.currentToken() == Token::StringLiteral);
@@ -1050,11 +1049,21 @@ a Ʃtest\f)");
 BOOST_AUTO_TEST_CASE(special_comment_with_unterminated_escape_sequence_at_eos)
 {
  CharStream stream(R"("test\)", "");
+ std::string expectedOutput(R"(test6101A9test)");
  Scanner scanner(stream, ScannerKind::SpecialComment);
  BOOST_REQUIRE(scanner.currentToken() == Token::Illegal);
  BOOST_REQUIRE(scanner.currentError() == ScannerError::IllegalEscapeSequence);
 }
 
+BOOST_AUTO_TEST_CASE(special_comment_with_escaped_quotes)
+{
+ CharStream stream(R"("test\\\"")", "");
+ std::string expectedOutput(R"(test)");
+ Scanner scanner(stream, ScannerKind::SpecialComment);
+ BOOST_REQUIRE(scanner.currentToken() == Token::StringLiteral);
+ BOOST_REQUIRE(scanner.currentLiteral() == expectedOutput);
+}
+
 BOOST_AUTO_TEST_CASE(special_comment_with_unterminated_string)
 {
  CharStream stream(R"("test)", "");