[es6] support rest parameters in arrow functions

src/scanner.cc

Index: src/scanner.cc
diff --git a/src/scanner.cc b/src/scanner.cc
index 9e90868131cc7469b6d88d79e2d409e561915dce..6b9b8134455359228a388f5828869a35a414c8ae 100644
--- a/src/scanner.cc
+++ b/src/scanner.cc
@@ -52,7 +52,8 @@ void Scanner::Initialize(Utf16CharacterStream* source) {
   // after a newline and scan first token.
   has_line_terminator_before_next_ = true;
   SkipWhiteSpace();
-  Scan();
+  peek_count_ = 1;
+  Scan(&next_[0]);
 }
 
 
@@ -228,21 +229,26 @@ static const byte one_char_tokens[] = {
 
 
 Token::Value Scanner::Next() {
-  current_ = next_;
+  current_ = next_[0];
+  if (peek_count_ > 1) {
+    std::memmove(&next_[0], &next_[1], (peek_count_ - 1) * sizeof(TokenDesc));
+    next_[--peek_count_].token = Token::ILLEGAL;
+    return current_.token;
+  }
   has_line_terminator_before_next_ = false;
   has_multiline_comment_before_next_ = false;
   if (static_cast<unsigned>(c0_) <= 0x7f) {
     Token::Value token = static_cast<Token::Value>(one_char_tokens[c0_]);
     if (token != Token::ILLEGAL) {
       int pos = source_pos();
-      next_.token = token;
-      next_.location.beg_pos = pos;
-      next_.location.end_pos = pos + 1;
+      next_[0].token = token;
+      next_[0].location.beg_pos = pos;
+      next_[0].location.end_pos = pos + 1;
       Advance();
       return current_.token;
     }
   }
-  Scan();
+  Scan(&next_[0]);
   return current_.token;
 }
 
@@ -420,13 +426,35 @@ Token::Value Scanner::ScanHtmlComment() {
 }
 
 
-void Scanner::Scan() {
-  next_.literal_chars = NULL;
-  next_.raw_literal_chars = NULL;
+Token::Value Scanner::peek(int n) {
+  if (n == 0) return peek();
+  PeekScan(n);
+  return next_[n].token;
+}
+
+
+Scanner::Location Scanner::peek_location(int n) {
+  if (n == 0) return peek_location();
+  PeekScan(n);
+  return next_[n].location;
+}
+
+
+void Scanner::PeekScan(int count) {
+  DCHECK(count > 0 && count < kMaxLookahead);
+  while (peek_count_ <= count) {
+    Scan(&next_[peek_count_++]);
+  }
+}
+
+
+void Scanner::Scan(TokenDesc* next) {
+  next->literal_chars = NULL;
+  next->raw_literal_chars = NULL;
   Token::Value token;
   do {
     // Remember the position of the next token
-    next_.location.beg_pos = source_pos();
+    next->location.beg_pos = source_pos();
 
     switch (c0_) {
       case ' ':
@@ -680,17 +708,21 @@ void Scanner::Scan() {
     // whitespace.
   } while (token == Token::WHITESPACE);
 
-  next_.location.end_pos = source_pos();
-  next_.token = token;
+  next->location.end_pos = source_pos();
+  next->token = token;
 }
 
 
 void Scanner::SeekForward(int pos) {
+  // Not supported when there is multiple lookahead tokens

[marja, 2015/03/10 09:11:19]

... what prevents this from happening?

[caitp (gmail), 2015/03/10 14:47:51]

it just seemed complicated to make it work with multiple lookaheads

+  DCHECK(peek_count_ < 2);
+
   // After this call, we will have the token at the given position as
   // the "next" token. The "current" token will be invalid.
-  if (pos == next_.location.beg_pos) return;
+  TokenDesc* next = &next_[0];
+  if (pos == next->location.beg_pos) return;
   int current_pos = source_pos();
-  DCHECK_EQ(next_.location.end_pos, current_pos);
+  DCHECK_EQ(next->location.end_pos, current_pos);
   // Positions inside the lookahead token aren't supported.
   DCHECK(pos >= current_pos);
   if (pos != current_pos) {
@@ -702,12 +734,12 @@ void Scanner::SeekForward(int pos) {
     has_line_terminator_before_next_ = false;
     has_multiline_comment_before_next_ = false;
   }
-  Scan();
+  Scan(next);
 }
 
 
 template <bool capture_raw, bool in_template_literal>
-bool Scanner::ScanEscape() {
+bool Scanner::ScanEscape(TokenDesc* next) {
   uc32 c = c0_;
   Advance<capture_raw>();
 
@@ -757,7 +789,7 @@ bool Scanner::ScanEscape() {
   // According to ECMA-262, section 7.8.4, characters not covered by the
   // above cases should be illegal, but they are commonly handled as
   // non-escaped characters by JS VMs.
-  AddLiteralChar(c);
+  AddLiteralChar(next, c);
   return true;
 }
 
@@ -796,6 +828,7 @@ Token::Value Scanner::ScanString() {
   Advance<false, false>();  // consume quote
 
   LiteralScope literal(this);
+  TokenDesc* next = literal.next_;
   while (true) {
     if (c0_ > kMaxAscii) {
       HandleLeadSurrogate();
@@ -810,7 +843,7 @@ Token::Value Scanner::ScanString() {
     uc32 c = c0_;
     if (c == '\\') break;
     Advance<false, false>();
-    AddLiteralChar(c);
+    AddLiteralChar(next, c);
   }
 
   while (c0_ != quote && c0_ >= 0
@@ -818,9 +851,9 @@ 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>(next)) return Token::ILLEGAL;
     } else {
-      AddLiteralChar(c);
+      AddLiteralChar(next, c);
     }
   }
   if (c0_ != quote) return Token::ILLEGAL;
@@ -844,9 +877,9 @@ Token::Value Scanner::ScanTemplateSpan() {
   // A TEMPLATE_SPAN should always be followed by an Expression, while a
   // TEMPLATE_TAIL terminates a TemplateLiteral and does not need to be
   // followed by an Expression.
-
   Token::Value result = Token::TEMPLATE_SPAN;
   LiteralScope literal(this);
+  TokenDesc* next = literal.next_;
   StartRawLiteral();
   const bool capture_raw = true;
   const bool in_template_literal = true;
@@ -856,11 +889,11 @@ Token::Value Scanner::ScanTemplateSpan() {
     Advance<capture_raw>();
     if (c == '`') {
       result = Token::TEMPLATE_TAIL;
-      ReduceRawLiteralLength(1);
+      ReduceRawLiteralLength(next, 1);
       break;
     } else if (c == '$' && c0_ == '{') {
       Advance<capture_raw>();  // Consume '{'
-      ReduceRawLiteralLength(2);
+      ReduceRawLiteralLength(next, 2);
       break;
     } else if (c == '\\') {
       if (c0_ > 0 && unicode_cache_->IsLineTerminator(c0_)) {
@@ -869,14 +902,14 @@ Token::Value Scanner::ScanTemplateSpan() {
         uc32 lastChar = c0_;
         Advance<capture_raw>();
         if (lastChar == '\r') {
-          ReduceRawLiteralLength(1);  // Remove \r
+          ReduceRawLiteralLength(next, 1);  // Remove \r
           if (c0_ == '\n') {
             Advance<capture_raw>();  // Adds \n
           } else {
-            AddRawLiteralChar('\n');
+            AddRawLiteralChar(next, '\n');
           }
         }
-      } else if (!ScanEscape<capture_raw, in_template_literal>()) {
+      } else if (!ScanEscape<capture_raw, in_template_literal>(next)) {
         return Token::ILLEGAL;
       }
     } else if (c < 0) {
@@ -888,42 +921,43 @@ Token::Value Scanner::ScanTemplateSpan() {
       // The TRV of LineTerminatorSequence :: <CR><LF> is the sequence
       // consisting of the CV 0x000A.
       if (c == '\r') {
-        ReduceRawLiteralLength(1);  // Remove \r
+        ReduceRawLiteralLength(next, 1);  // Remove \r
         if (c0_ == '\n') {
           Advance<capture_raw>();  // Adds \n
         } else {
-          AddRawLiteralChar('\n');
+          AddRawLiteralChar(next, '\n');
         }
         c = '\n';
       }
-      AddLiteralChar(c);
+      AddLiteralChar(next, c);
     }
   }
   literal.Complete();
-  next_.location.end_pos = source_pos();
-  next_.token = result;
+  next->location.end_pos = source_pos();
+  next->token = result;
   return result;
 }
 
 
 Token::Value Scanner::ScanTemplateStart() {
   DCHECK(c0_ == '`');
-  next_.location.beg_pos = source_pos();
+  TokenDesc* next = PeekTokenDesc();
+  next->location.beg_pos = source_pos();
   Advance();  // Consume `
   return ScanTemplateSpan();
 }
 
 
 Token::Value Scanner::ScanTemplateContinuation() {
-  DCHECK_EQ(next_.token, Token::RBRACE);
-  next_.location.beg_pos = source_pos() - 1;  // We already consumed }
+  TokenDesc* next = PeekTokenDesc();
+  DCHECK_EQ(next->token, Token::RBRACE);
+  next->location.beg_pos = source_pos() - 1;  // We already consumed }
   return ScanTemplateSpan();
 }
 
 
-void Scanner::ScanDecimalDigits() {
-  while (IsDecimalDigit(c0_))
-    AddLiteralCharAdvance();
+void Scanner::ScanDecimalDigits(TokenDesc* next) {
+  while (IsDecimalDigit(c0_)) AddLiteralCharAdvance(next);
 }
 
 
@@ -933,50 +967,51 @@ Token::Value Scanner::ScanNumber(bool seen_period) {
   enum { DECIMAL, HEX, OCTAL, IMPLICIT_OCTAL, BINARY } kind = DECIMAL;
 
   LiteralScope literal(this);
+  TokenDesc* next = literal.next_;
   bool at_start = !seen_period;
   if (seen_period) {
     // we have already seen a decimal point of the float
-    AddLiteralChar('.');
-    ScanDecimalDigits();  // we know we have at least one digit
+    AddLiteralChar(next, '.');
+    ScanDecimalDigits(next);  // we know we have at least one digit
 
   } else {
     // if the first character is '0' we must check for octals and hex
     if (c0_ == '0') {
       int start_pos = source_pos();  // For reporting octal positions.
-      AddLiteralCharAdvance();
+      AddLiteralCharAdvance(next);
 
       // either 0, 0exxx, 0Exxx, 0.xxx, a hex number, a binary number or
       // an octal number.
       if (c0_ == 'x' || c0_ == 'X') {
         // hex number
         kind = HEX;
-        AddLiteralCharAdvance();
+        AddLiteralCharAdvance(next);
         if (!IsHexDigit(c0_)) {
           // we must have at least one hex digit after 'x'/'X'
           return Token::ILLEGAL;
         }
         while (IsHexDigit(c0_)) {
-          AddLiteralCharAdvance();
+          AddLiteralCharAdvance(next);
         }
       } else if (harmony_numeric_literals_ && (c0_ == 'o' || c0_ == 'O')) {
         kind = OCTAL;
-        AddLiteralCharAdvance();
+        AddLiteralCharAdvance(next);
         if (!IsOctalDigit(c0_)) {
           // we must have at least one octal digit after 'o'/'O'
           return Token::ILLEGAL;
         }
         while (IsOctalDigit(c0_)) {
-          AddLiteralCharAdvance();
+          AddLiteralCharAdvance(next);
         }
       } else if (harmony_numeric_literals_ && (c0_ == 'b' || c0_ == 'B')) {
         kind = BINARY;
-        AddLiteralCharAdvance();
+        AddLiteralCharAdvance(next);
         if (!IsBinaryDigit(c0_)) {
           // we must have at least one binary digit after 'b'/'B'
           return Token::ILLEGAL;
         }
         while (IsBinaryDigit(c0_)) {
-          AddLiteralCharAdvance();
+          AddLiteralCharAdvance(next);
         }
       } else if ('0' <= c0_ && c0_ <= '7') {
         // (possible) octal number
@@ -992,7 +1027,7 @@ Token::Value Scanner::ScanNumber(bool seen_period) {
             octal_pos_ = Location(start_pos, source_pos());
             break;
           }
-          AddLiteralCharAdvance();
+          AddLiteralCharAdvance(next);
         }
       }
     }
@@ -1006,10 +1041,10 @@ Token::Value Scanner::ScanNumber(bool seen_period) {
 
           uc32 first_char = c0_;
           Advance<false, false>();
-          AddLiteralChar(first_char);
+          AddLiteralChar(next, first_char);
         }
 
-        if (next_.literal_chars->one_byte_literal().length() < 10 &&
+        if (next->literal_chars->one_byte_literal().length() < 10 &&
             c0_ != '.' && c0_ != 'e' && c0_ != 'E') {
           smi_value_ = value;
           literal.Complete();
@@ -1020,10 +1055,10 @@ Token::Value Scanner::ScanNumber(bool seen_period) {
         HandleLeadSurrogate();
       }
 
-      ScanDecimalDigits();  // optional
+      ScanDecimalDigits(next);  // optional
       if (c0_ == '.') {
-        AddLiteralCharAdvance();
-        ScanDecimalDigits();  // optional
+        AddLiteralCharAdvance(next);
+        ScanDecimalDigits(next);  // optional
       }
     }
   }
@@ -1033,14 +1068,13 @@ Token::Value Scanner::ScanNumber(bool seen_period) {
     DCHECK(kind != HEX);  // 'e'/'E' must be scanned as part of the hex number
     if (kind != DECIMAL) return Token::ILLEGAL;
     // scan exponent
-    AddLiteralCharAdvance();
-    if (c0_ == '+' || c0_ == '-')
-      AddLiteralCharAdvance();
+    AddLiteralCharAdvance(next);
+    if (c0_ == '+' || c0_ == '-') AddLiteralCharAdvance(next);
     if (!IsDecimalDigit(c0_)) {
       // we must have at least one decimal digit after 'e'/'E'
       return Token::ILLEGAL;
     }
-    ScanDecimalDigits();
+    ScanDecimalDigits(next);
   }
 
   // The source character immediately following a numeric literal must
@@ -1218,12 +1252,13 @@ bool Scanner::IdentifierIsFutureStrictReserved(
 
 Token::Value Scanner::ScanIdentifierOrKeyword() {
   DCHECK(unicode_cache_->IsIdentifierStart(c0_));
+  TokenDesc* next = PeekTokenDesc();
   LiteralScope literal(this);
   if (IsInRange(c0_, 'a', 'z')) {
     do {
       uc32 first_char = c0_;
       Advance<false, false>();
-      AddLiteralChar(first_char);
+      AddLiteralChar(next, first_char);
     } while (IsInRange(c0_, 'a', 'z'));
 
     if (IsDecimalDigit(c0_) || IsInRange(c0_, 'A', 'Z') || c0_ == '_' ||
@@ -1231,11 +1266,11 @@ Token::Value Scanner::ScanIdentifierOrKeyword() {
       // Identifier starting with lowercase.
       uc32 first_char = c0_;
       Advance<false, false>();
-      AddLiteralChar(first_char);
+      AddLiteralChar(next, first_char);
       while (IsAsciiIdentifier(c0_)) {
         uc32 first_char = c0_;
         Advance<false, false>();
-        AddLiteralChar(first_char);
+        AddLiteralChar(next, first_char);
       }
       if (c0_ <= kMaxAscii && c0_ != '\\') {
         literal.Complete();
@@ -1244,7 +1279,7 @@ Token::Value Scanner::ScanIdentifierOrKeyword() {
     } else if (c0_ <= kMaxAscii && c0_ != '\\') {
       // Only a-z+: could be a keyword or identifier.
       literal.Complete();
-      Vector<const uint8_t> chars = next_.literal_chars->one_byte_literal();
+      Vector<const uint8_t> chars = next->literal_chars->one_byte_literal();
       return KeywordOrIdentifierToken(chars.start(), chars.length(),
                                       harmony_scoping_, harmony_modules_,
                                       harmony_classes_);
@@ -1255,7 +1290,7 @@ Token::Value Scanner::ScanIdentifierOrKeyword() {
     do {
       uc32 first_char = c0_;
       Advance<false, false>();
-      AddLiteralChar(first_char);
+      AddLiteralChar(next, first_char);
     } while (IsAsciiIdentifier(c0_));
 
     if (c0_ <= kMaxAscii && c0_ != '\\') {
@@ -1273,12 +1308,12 @@ Token::Value Scanner::ScanIdentifierOrKeyword() {
         !unicode_cache_->IsIdentifierStart(c)) {
       return Token::ILLEGAL;
     }
-    AddLiteralChar(c);
+    AddLiteralChar(next, c);
     return ScanIdentifierSuffix(&literal);
   } else {
     uc32 first_char = c0_;
     Advance();
-    AddLiteralChar(first_char);
+    AddLiteralChar(next, first_char);
   }
 
   // Scan the rest of the identifier characters.
@@ -1286,7 +1321,7 @@ Token::Value Scanner::ScanIdentifierOrKeyword() {
     if (c0_ != '\\') {
       uc32 next_char = c0_;
       Advance();
-      AddLiteralChar(next_char);
+      AddLiteralChar(next, next_char);
       continue;
     }
     // Fallthrough if no longer able to complete keyword.
@@ -1295,8 +1330,8 @@ Token::Value Scanner::ScanIdentifierOrKeyword() {
 
   literal.Complete();
 
-  if (next_.literal_chars->is_one_byte()) {
-    Vector<const uint8_t> chars = next_.literal_chars->one_byte_literal();
+  if (next->literal_chars->is_one_byte()) {
+    Vector<const uint8_t> chars = next->literal_chars->one_byte_literal();
     return KeywordOrIdentifierToken(chars.start(),
                                     chars.length(),
                                     harmony_scoping_,
@@ -1309,6 +1344,7 @@ Token::Value Scanner::ScanIdentifierOrKeyword() {
 
 Token::Value Scanner::ScanIdentifierSuffix(LiteralScope* literal) {
   // Scan the rest of the identifier characters.
+  TokenDesc* next = literal->next_;
   while (c0_ >= 0 && unicode_cache_->IsIdentifierPart(c0_)) {
     if (c0_ == '\\') {
       uc32 c = ScanIdentifierUnicodeEscape();
@@ -1318,9 +1354,9 @@ Token::Value Scanner::ScanIdentifierSuffix(LiteralScope* literal) {
           !unicode_cache_->IsIdentifierPart(c)) {
         return Token::ILLEGAL;
       }
-      AddLiteralChar(c);
+      AddLiteralChar(next, c);
     } else {
-      AddLiteralChar(c0_);
+      AddLiteralChar(next, c0_);
       Advance();
     }
   }
@@ -1333,26 +1369,27 @@ Token::Value Scanner::ScanIdentifierSuffix(LiteralScope* literal) {
 bool Scanner::ScanRegExpPattern(bool seen_equal) {
   // Scan: ('/' | '/=') RegularExpressionBody '/' RegularExpressionFlags
   bool in_character_class = false;
+  TokenDesc* next = PeekTokenDesc();
 
   // Previous token is either '/' or '/=', in the second case, the
   // pattern starts at =.
-  next_.location.beg_pos = source_pos() - (seen_equal ? 2 : 1);
-  next_.location.end_pos = source_pos() - (seen_equal ? 1 : 0);
+  next->location.beg_pos = source_pos() - (seen_equal ? 2 : 1);
+  next->location.end_pos = source_pos() - (seen_equal ? 1 : 0);
 
   // Scan regular expression body: According to ECMA-262, 3rd, 7.8.5,
   // the scanner should pass uninterpreted bodies to the RegExp
   // constructor.
   LiteralScope literal(this);
   if (seen_equal) {
-    AddLiteralChar('=');
+    AddLiteralChar(next, '=');
   }
 
   while (c0_ != '/' || in_character_class) {
     if (c0_ < 0 || unicode_cache_->IsLineTerminator(c0_)) return false;
     if (c0_ == '\\') {  // Escape sequence.
-      AddLiteralCharAdvance();
+      AddLiteralCharAdvance(next);
       if (c0_ < 0 || unicode_cache_->IsLineTerminator(c0_)) return false;
-      AddLiteralCharAdvance();
+      AddLiteralCharAdvance(next);
       // If the escape allows more characters, i.e., \x??, \u????, or \c?,
       // only "safe" characters are allowed (letters, digits, underscore),
       // otherwise the escape isn't valid and the invalid character has
@@ -1366,7 +1403,7 @@ bool Scanner::ScanRegExpPattern(bool seen_equal) {
     } else {  // Unescaped character.
       if (c0_ == '[') in_character_class = true;
       if (c0_ == ']') in_character_class = false;
-      AddLiteralCharAdvance();
+      AddLiteralCharAdvance(next);
     }
   }
   Advance();  // consume '/'
@@ -1379,17 +1416,18 @@ bool Scanner::ScanRegExpPattern(bool seen_equal) {
 
 bool Scanner::ScanRegExpFlags() {
   // Scan regular expression flags.
+  TokenDesc* next = PeekTokenDesc();
   LiteralScope literal(this);
   while (c0_ >= 0 && unicode_cache_->IsIdentifierPart(c0_)) {
     if (c0_ != '\\') {
-      AddLiteralCharAdvance();
+      AddLiteralCharAdvance(next);
     } else {
       return false;
     }
   }
   literal.Complete();
 
-  next_.location.end_pos = source_pos() - 1;
+  next->location.end_pos = source_pos() - 1;
   return true;
 }