[parser] implement error reporting for Scanner errors

src/parsing/scanner.cc

Index: src/parsing/scanner.cc
diff --git a/src/parsing/scanner.cc b/src/parsing/scanner.cc
index a197829b35b69a6ecbe7d9b7521554eeb8c8c8c1..e1019594f5292404ecf52327cc1fd30c0c3f2601 100644
--- a/src/parsing/scanner.cc
+++ b/src/parsing/scanner.cc
@@ -62,15 +62,19 @@ void Scanner::Initialize(Utf16CharacterStream* source) {
   Scan();
 }
 
-
-template <bool capture_raw>
+template <bool capture_raw, bool unicode>
 uc32 Scanner::ScanHexNumber(int expected_length) {
   DCHECK(expected_length <= 4);  // prevent overflow
 
+  int begin = source_pos() - 2;
   uc32 x = 0;
   for (int i = 0; i < expected_length; i++) {
     int d = HexValue(c0_);
     if (d < 0) {
+      ReportScannerError(Location(begin, begin + expected_length + 2),
+                         unicode
+                             ? MessageTemplate::kInvalidUnicodeEscapeSequence
+                             : MessageTemplate::kInvalidHexEscapeSequence);
       return -1;
     }
     x = x * 16 + d;
@@ -80,20 +84,21 @@ uc32 Scanner::ScanHexNumber(int expected_length) {
   return x;
 }
 
-
 template <bool capture_raw>
-uc32 Scanner::ScanUnlimitedLengthHexNumber(int max_value) {
+uc32 Scanner::ScanUnlimitedLengthHexNumber(int max_value, int beg_pos) {
   uc32 x = 0;
   int d = HexValue(c0_);
-  if (d < 0) {

[adamk, 2016/03/21 19:44:31]

I think you still need this early return...

[caitp (gmail), 2016/03/21 19:51:23]

you're correct, `\u{}` parses successfully with this :(

-    return -1;
-  }
   while (d >= 0) {
     x = x * 16 + d;
-    if (x > max_value) return -1;
+    if (x > max_value) {
+      ReportScannerError(Location(beg_pos, source_pos() + 1),
+                         MessageTemplate::kUndefinedUnicodeCodePoint);
+      return -1;
+    }
     Advance<capture_raw>();
     d = HexValue(c0_);
   }
+
   return x;
 }
 
@@ -856,7 +861,9 @@ Token::Value Scanner::ScanString() {
     uc32 c = c0_;
     Advance();
     if (c == '\\') {
-      if (c0_ < 0 || !ScanEscape<false, false>()) return Token::ILLEGAL;
+      if (c0_ < 0 || !ScanEscape<false, false>()) {
+        return Token::ILLEGAL;
+      }
     } else {
       AddLiteralChar(c);
     }
@@ -888,7 +895,6 @@ Token::Value Scanner::ScanTemplateSpan() {
   StartRawLiteral();
   const bool capture_raw = true;
   const bool in_template_literal = true;
-
   while (true) {
     uc32 c = c0_;
     Advance<capture_raw>();
@@ -1108,18 +1114,22 @@ uc32 Scanner::ScanUnicodeEscape() {
   // Accept both \uxxxx and \u{xxxxxx}. In the latter case, the number of
   // hex digits between { } is arbitrary. \ and u have already been read.
   if (c0_ == '{') {
+    int begin = source_pos() - 2;
     Advance<capture_raw>();
-    uc32 cp = ScanUnlimitedLengthHexNumber<capture_raw>(0x10ffff);
+    uc32 cp = ScanUnlimitedLengthHexNumber<capture_raw>(0x10ffff, begin);
     if (cp < 0) {

[adamk, 2016/03/21 19:44:31]

...and then report an error here, right?

[caitp (gmail), 2016/03/21 19:51:23]

Done.

       return -1;
     }
     if (c0_ != '}') {
+      ReportScannerError(source_pos(),
+                         MessageTemplate::kInvalidUnicodeEscapeSequence);
       return -1;
     }
     Advance<capture_raw>();
     return cp;
   }
-  return ScanHexNumber<capture_raw>(4);
+  const bool unicode = true;
+  return ScanHexNumber<capture_raw, unicode>(4);
 }