Fix out-of-bound source code accesses in Lexer

Signed-off-by: HyukWoo Park <[email protected]>

src/parser/Lexer.cpp

@@ -1003,7 +1003,7 @@ void Scanner::scanPunctuator(Scanner::ScannerResult* token, char16_t ch)
 
     case '.':
         kind = Period;
-        if (this->peekChar() == '.' && this->sourceCharAt(this->index + 1) == '.') {
+        if (this->peekChar() == '.' && this->peekChar(this->index + 1) == '.') {
             // Spread operator "..."
             this->index += 2;
             kind = PeriodPeriodPeriod;
@@ -2354,8 +2354,16 @@ void Scanner::lex(Scanner::ScannerResult* token)
     }
 
     if (isIdentifierStart(cp)) {
-        goto ScanID;
+        this->scanIdentifier(token, cp);
+        return;
+    }
+
+    // common cases : `(` and `)` and `;`
+    if (cp == 0x28 || cp == 0x29 || cp == 0x3B) {
+        this->scanPunctuator(token, cp);
+        return;
     }
+
     // String literal starts with single quote (U+0027) or double quote (U+0022).
     if (cp == 0x27 || cp == 0x22) {
         this->scanStringLiteral(token);
@@ -2364,8 +2372,12 @@ void Scanner::lex(Scanner::ScannerResult* token)
 
     // Dot (.) U+002E can also start a floating-point number, hence the need
     // to check the next character.
-    if (UNLIKELY(cp == 0x2E) && isDecimalDigit(this->sourceCharAt(this->index + 1))) {
-        this->scanNumericLiteral(token);
+    if (cp == 0x2E) {
+        if (UNLIKELY(isDecimalDigit(this->peekChar(this->index + 1)))) {
+            this->scanNumericLiteral(token);
+            return;
+        }
+        this->scanPunctuator(token, cp);
         return;
     }
 
@@ -2374,21 +2386,19 @@ void Scanner::lex(Scanner::ScannerResult* token)
         return;
     }
 
-    if (UNLIKELY(cp == '`')) {
+    if (UNLIKELY(cp == 0x60)) {
         ++this->index;
         this->scanTemplate(token, true);
         return;
     }
+
     // Possible identifier start in a surrogate pair.
-    if (UNLIKELY(cp >= 0xD800 && cp < 0xDFFF) && isIdentifierStart(this->codePointAt(this->index))) {
-        goto ScanID;
+    if (UNLIKELY(cp >= 0xD800 && cp < 0xDFFF && isIdentifierStart(this->codePointAt(this->index)))) {
+        this->scanIdentifier(token, cp);
+        return;
     }
 
     this->scanPunctuator(token, cp);
     return;
-
-ScanID:
-    this->scanIdentifier(token, cp);
-    return;
 }
 } // namespace Escargot

src/parser/Lexer.h

@@ -583,6 +583,7 @@ class Scanner {
 
     ALWAYS_INLINE char16_t sourceCharAt(const size_t idx) const
     {
+        ASSERT(idx < this->length);
         return sourceCodeAccessData.charAt(idx);
     }
 
@@ -597,20 +598,20 @@ class Scanner {
     {
         bool start = (this->index == 0);
         while (LIKELY(!this->eof())) {
-            char16_t ch = this->sourceCharAt(this->index);
+            char16_t ch = this->peekCharWithoutEOF();
 
             if (isWhiteSpace(ch)) {
                 ++this->index;
             } else if (isLineTerminator(ch)) {
                 ++this->index;
-                if (ch == 0x0D && this->sourceCharAt(this->index) == 0x0A) {
+                if (ch == 0x0D && this->peekChar(this->index) == 0x0A) {
                     ++this->index;
                 }
                 ++this->lineNumber;
                 this->lineStart = this->index;
                 start = true;
             } else if (ch == 0x2F) { // U+002F is '/'
-                ch = this->sourceCharAt(this->index + 1);
+                ch = this->peekChar(this->index + 1);
                 if (ch == 0x2F) {
                     this->index += 2;
                     this->skipSingleLineComment();
@@ -621,29 +622,24 @@ class Scanner {
                 } else {
                     break;
                 }
-            } else if (start && ch == 0x2D) { // U+002D is '-'
+            } else if (start && ch == 0x2D && !this->isModule) { // U+002D is '-'
                 // U+003E is '>'
-                if (!this->isModule && (this->sourceCharAt(this->index + 1) == 0x2D) && (this->sourceCharAt(this->index + 2) == 0x3E)) {
+                if ((this->peekChar(this->index + 1) == 0x2D) && (this->peekChar(this->index + 2) == 0x3E)) {
                     // '-->' is a single-line comment
                     this->index += 3;
                     this->skipSingleLineComment();
                 } else {
                     break;
                 }
-            } else if (ch == 0x3C) { // U+003C is '<'
-                if (!this->isModule && this->length > this->index + 4) {
-                    if (this->sourceCharAt(this->index + 1) == '!'
-                        && this->sourceCharAt(this->index + 2) == '-'
-                        && this->sourceCharAt(this->index + 3) == '-') {
-                        this->index += 4; // `<!--`
-                        this->skipSingleLineComment();
-                    } else {
-                        break;
-                    }
+            } else if (ch == 0x3C && !this->isModule) { // U+003C is '<'
+                if (this->peekChar(this->index + 1) == '!'
+                    && this->peekChar(this->index + 2) == '-'
+                    && this->peekChar(this->index + 3) == '-') {
+                    this->index += 4; // `<!--`
+                    this->skipSingleLineComment();
                 } else {
                     break;
                 }
-
             } else {
                 break;
             }
@@ -690,6 +686,7 @@ class Scanner {
 private:
     ALWAYS_INLINE char16_t peekCharWithoutEOF()
     {
+        ASSERT(!this->eof());
         return this->sourceCharAt(this->index);
     }
 
@@ -698,6 +695,12 @@ class Scanner {
         return UNLIKELY(this->eof()) ? 0 : this->sourceCharAt(this->index);
     }
 
+    ALWAYS_INLINE char16_t peekChar(size_t idx)
+    {
+        // check EOF
+        return UNLIKELY(idx >= this->length) ? 0 : this->sourceCharAt(idx);
+    }
+
     char32_t scanHexEscape(char prefix);
     char32_t scanUnicodeCodePointEscape();