- Fixed infinite loop in lexer when reading an unterminated string

- Added a test to ensure unterminated strings are caught as an error by the lexer

layout/src/gv/parser/lexer.rs

@@ -160,7 +160,7 @@ impl Lexer {
         result
     }
 
-    pub fn read_string(&mut self) -> String {
+    pub fn read_string(&mut self) -> Token {
         let mut result = String::new();
         self.read_char();
         while self.ch != '"' {
@@ -173,11 +173,14 @@ impl Lexer {
                     'l' => '\n',
                     _ => self.ch,
                 }
+            } else if self.ch == '\0' {
+                // Reached EOF without completing the string
+                return Token::Error(self.pos);
             }
             result.push(self.ch);
             self.read_char();
         }
-        result
+        Token::Identifier(result)
     }
 
     pub fn next_token(&mut self) -> Token {
@@ -209,8 +212,7 @@ impl Lexer {
                 tok = Token::Comma;
             }
             '"' => {
-                let value = self.read_string();
-                tok = Token::Identifier(value);
+                tok = self.read_string();
             }
             '-' => {
                 self.read_char();

tests/test.rs

@@ -108,6 +108,18 @@ mod tests {
         assert!(matches!(lexer.next_token(), Token::EOF));
     }
 
+    #[test]
+    fn catch_unterminated_str() {
+        let mut lexer = Lexer::from_string("digraph { a -> b; \" } ");
+        assert!(matches!(lexer.next_token(), Token::DigraphKW));
+        assert!(matches!(lexer.next_token(), Token::OpenBrace));
+        assert!(matches!(lexer.next_token(), Token::Identifier(_)));
+        assert!(matches!(lexer.next_token(), Token::ArrowRight));
+        assert!(matches!(lexer.next_token(), Token::Identifier(_)));
+        assert!(matches!(lexer.next_token(), Token::Semicolon));
+        assert!(matches!(lexer.next_token(), Token::Error(_)));
+    }
+
     #[test]
     fn lex_program() {
         let program = get_sample_program2();