Reapply 3246 and 3247 after fixing valgrind warning.

src/scanner.h

 // Copyright 2006-2008 the V8 project authors. All rights reserved.
 // Redistribution and use in source and binary forms, with or without
 // modification, are permitted provided that the following conditions are
 // met:
 //
 //     * Redistributions of source code must retain the above copyright
 //       notice, this list of conditions and the following disclaimer.
 //     * Redistributions in binary form must reproduce the above
 //       copyright notice, this list of conditions and the following
 //       disclaimer in the documentation and/or other materials provided
@@ skipping 23 matching lines @@
 namespace v8 {
 namespace internal {
 
 
 class UTF8Buffer {
  public:
   UTF8Buffer();
   ~UTF8Buffer();
 
   void AddChar(uc32 c) {
+    ASSERT_NOT_NULL(data_);
     if (cursor_ <= limit_ &&
         static_cast<unsigned>(c) <= unibrow::Utf8::kMaxOneByteChar) {
       *cursor_++ = static_cast<char>(c);
     } else {
       AddCharSlow(c);
     }
   }
 
-  void Reset() { cursor_ = data_; }
-  int pos() const { return cursor_ - data_; }
+  void Reset() {
+    if (data_ == NULL) {
+      data_ = NewArray<char>(kInitialCapacity);
+      limit_ = ComputeLimit(data_, kInitialCapacity);
+    }
+    cursor_ = data_;
+  }
+
+  int pos() const {
+    ASSERT_NOT_NULL(data_);
+    return cursor_ - data_;
+  }
+
   char* data() const { return data_; }
 
  private:
+  static const int kInitialCapacity = 256;
   char* data_;
   char* cursor_;
   char* limit_;
 
   int Capacity() const {
+    ASSERT_NOT_NULL(data_);
     return (limit_ - data_) + unibrow::Utf8::kMaxEncodedSize;
   }
 
   static char* ComputeLimit(char* data, int capacity) {
     return (data + capacity) - unibrow::Utf8::kMaxEncodedSize;
   }
 
   void AddCharSlow(uc32 c);
 };
 
@@ skipping 199 matching lines @@
 
   // Returns the location information for the current token
   // (the token returned by Next()).
   Location location() const  { return current_.location; }
   Location peek_location() const  { return next_.location; }
 
   // Returns the literal string, if any, for the current token (the
   // token returned by Next()). The string is 0-terminated and in
   // UTF-8 format; they may contain 0-characters. Literal strings are
   // collected for identifiers, strings, and numbers.
+  // These functions only give the correct result if the literal
+  // was scanned between calls to StartLiteral() and TerminateLiteral().
   const char* literal_string() const {
-    return &literals_.data()[current_.literal_pos];
+    return current_.literal_buffer->data();
   }
   int literal_length() const {
-    return current_.literal_end - current_.literal_pos;
-  }
-
-  Vector<const char> next_literal() const {
-    return Vector<const char>(next_literal_string(), next_literal_length());
+    // Excluding terminal '\0' added by TerminateLiteral().
+    return current_.literal_buffer->pos() - 1;
   }
 
   // Returns the literal string for the next token (the token that
   // would be returned if Next() were called).
   const char* next_literal_string() const {
-    return &literals_.data()[next_.literal_pos];
+    return next_.literal_buffer->data();
   }
   // Returns the length of the next token (that would be returned if
   // Next() were called).
   int next_literal_length() const {
-    return next_.literal_end - next_.literal_pos;
+    return next_.literal_buffer->pos() - 1;
+  }
+
+  Vector<const char> next_literal() const {
+    return Vector<const char>(next_literal_string(),
+                              next_literal_length());
   }
 
   // Scans the input as a regular expression pattern, previous
   // character(s) must be /(=). Returns true if a pattern is scanned.
   bool ScanRegExpPattern(bool seen_equal);
   // Returns true if regexp flags are scanned (always since flags can
   // be empty).
   bool ScanRegExpFlags();
 
   // Seek forward to the given position.  This operation does not
@@ skipping 21 matching lines @@
  private:
   CharacterStreamUTF16Buffer char_stream_buffer_;
   TwoByteStringUTF16Buffer two_byte_string_buffer_;
 
   // Source.
   UTF16Buffer* source_;
   int position_;
 
   // Buffer to hold literal values (identifiers, strings, numbers)
   // using 0-terminated UTF-8 encoding.
-  UTF8Buffer literals_;
+  UTF8Buffer literal_buffer_1_;
+  UTF8Buffer literal_buffer_2_;
 
   bool stack_overflow_;
   static StaticResource<Utf8Decoder> utf8_decoder_;
 
   // One Unicode character look-ahead; c0_ < 0 at the end of the input.
   uc32 c0_;
 
   // The current and look-ahead token.
   struct TokenDesc {
     Token::Value token;
     Location location;
-    int literal_pos, literal_end;
+    UTF8Buffer* literal_buffer;
   };
 
   TokenDesc current_;  // desc for current token (as returned by Next())
   TokenDesc next_;     // desc for next token (one token look-ahead)
   bool has_line_terminator_before_next_;
   bool is_pre_parsing_;
 
   // Literal buffer support
   void StartLiteral();
   void AddChar(uc32 ch);
@@ skipping 32 matching lines @@
   }
 
   // Decodes a unicode escape-sequence which is part of an identifier.
   // If the escape sequence cannot be decoded the result is kBadRune.
   uc32 ScanIdentifierUnicodeEscape();
 };
 
 } }  // namespace v8::internal
 
 #endif  // V8_SCANNER_H_