Optimized scanner to avoid virtual calls for every character read.

src/scanner.h

Index: src/scanner.h
diff --git a/src/scanner.h b/src/scanner.h
index adeea9b23a00d50d7f07a9203dcc311fd50cebe4..d23d7c736697f2e8f1c75b3bf181ff6e446522d8 100644
--- a/src/scanner.h
+++ b/src/scanner.h
@@ -35,67 +35,95 @@
 namespace v8 {
 namespace internal {
 
-// UTF16 buffer to read characters from a character stream.
-class CharacterStreamUTF16Buffer: public UTF16Buffer {
+// A buffered character stream based on a random access character
+// source (ReadBlock can be called with pos_ pointing to any position,
+// even positions before the current).
+class BufferedUC16CharacterStream: public UC16CharacterStream {
  public:
-  CharacterStreamUTF16Buffer();
-  virtual ~CharacterStreamUTF16Buffer() {}
-  void Initialize(Handle<String> data,
-                  unibrow::CharacterStream* stream,
-                  int start_position,
-                  int end_position);
-  virtual void PushBack(uc32 ch);
-  virtual uc32 Advance();
-  virtual void SeekForward(int pos);
-
- private:
-  List<uc32> pushback_buffer_;
-  uc32 last_;
-  unibrow::CharacterStream* stream_;
-
-  List<uc32>* pushback_buffer() { return &pushback_buffer_; }
+  BufferedUC16CharacterStream();
+  virtual ~BufferedUC16CharacterStream();
+
+  virtual void PushBack(uc16 character);

[Erik Corry, 2010/12/07 12:27:30]

Missing blank line.

[Lasse Reichstein, 2010/12/07 14:05:54]

Done.

+ protected:
+  static const unsigned kBufferSize = 512;
+  static const unsigned kPushBackStepSize = 16;
+
+  virtual unsigned SlowSeekForward(unsigned delta);
+  virtual bool ReadBlock();
+  virtual void SlowPushBack(uc16 character);
+
+  virtual unsigned BufferSeekForward(unsigned delta) = 0;
+  virtual unsigned FillBuffer(unsigned position, unsigned length) = 0;
+
+  const uc16* pushback_limit_;
+  uc16 buffer_[kBufferSize];
 };
 
 
-// UTF16 buffer to read characters from an external string.
-template <typename StringType, typename CharType>
-class ExternalStringUTF16Buffer: public UTF16Buffer {
+// Generic string stream.
+class GenericStringUC16CharacterStream: public BufferedUC16CharacterStream {
+ public:
+  GenericStringUC16CharacterStream(Handle<String> data,
+                                   unsigned start_position,
+                                   unsigned end_position);
+  virtual ~GenericStringUC16CharacterStream();
+
+ protected:
+  virtual unsigned BufferSeekForward(unsigned delta);
+  virtual unsigned FillBuffer(unsigned position, unsigned length);
+
+  Handle<String> string_;
+  unsigned start_position_;
+  unsigned length_;
+};
+
+
+// UC16 stream based on a literal UTF-8 string.
+class Utf8ToUC16CharacterStream: public BufferedUC16CharacterStream {
  public:
-  ExternalStringUTF16Buffer();
-  virtual ~ExternalStringUTF16Buffer() {}
-  void Initialize(Handle<StringType> data,
-                  int start_position,
-                  int end_position);
-  virtual void PushBack(uc32 ch);
-  virtual uc32 Advance();
-  virtual void SeekForward(int pos);
-
- private:
-  const CharType* raw_data_;  // Pointer to the actual array of characters.
+  Utf8ToUC16CharacterStream(const byte* data, unsigned length);
+  virtual ~Utf8ToUC16CharacterStream();

[Erik Corry, 2010/12/07 12:27:30]

Missing blank line

[Lasse Reichstein, 2010/12/07 14:05:54]

Done.

+ protected:
+  virtual unsigned BufferSeekForward(unsigned delta);
+  virtual unsigned FillBuffer(unsigned char_position, unsigned length);
+  void SetRawPosition(unsigned char_position);
+
+  const byte* raw_data_;
+  unsigned raw_data_length_;  // Not the number of characters!

[Erik Corry, 2010/12/07 12:27:30]

Measured in bytes?

[Lasse Reichstein, 2010/12/07 14:05:54]

Yes. Reworded to say so.

+  unsigned raw_data_pos_;
+  // The character position of the character at raw_data[raw_data_pos_].
+  // Not necessarily the same as pos_.
+  unsigned raw_character_position_;
 };
 
 
-// Initializes a UTF16Buffer as input stream, using one of a number
-// of strategies depending on the available character sources.
-class StreamInitializer {
+// UTF16 buffer to read characters from an external string.
+class ExternalTwoByteStringUC16CharacterStream: public UC16CharacterStream {
  public:
-  UTF16Buffer* Init(Handle<String> source,
-                    unibrow::CharacterStream* stream,
-                    int start_position,
-                    int end_position);
- private:
-  // Different UTF16 buffers used to pull characters from. Based on input one of
-  // these will be initialized as the actual data source.
-  CharacterStreamUTF16Buffer char_stream_buffer_;
-  ExternalStringUTF16Buffer<ExternalTwoByteString, uint16_t>
-      two_byte_string_buffer_;
-  ExternalStringUTF16Buffer<ExternalAsciiString, char> ascii_string_buffer_;
-
-  // Used to convert the source string into a character stream when a stream
-  // is not passed to the scanner.
-  SafeStringInputBuffer safe_string_input_buffer_;
+  ExternalTwoByteStringUC16CharacterStream(Handle<ExternalTwoByteString> data,
+                                           int start_position,
+                                           int end_position);
+  virtual ~ExternalTwoByteStringUC16CharacterStream();
+
+  virtual void PushBack(uc16 character) {
+    ASSERT(buffer_cursor_ > raw_data_);
+    buffer_cursor_--;
+    pos_--;
+  }
+ protected:
+  virtual unsigned SlowSeekForward(unsigned delta) {
+    // Fast case always handles seeking.
+    return 0;
+  }
+  virtual bool ReadBlock() {
+    // Entire string is read at start.
+    return false;
+  }
+  Handle<ExternalTwoByteString> source_;
+  const uc16* raw_data_;  // Pointer to the actual array of characters.
 };
 
+
 // ----------------------------------------------------------------------------
 // V8JavaScriptScanner
 // JavaScript scanner getting its input from either a V8 String or a unicode
@@ -103,28 +131,17 @@ class StreamInitializer {
 
 class V8JavaScriptScanner : public JavaScriptScanner {
  public:
-  V8JavaScriptScanner() {}
-
-  // Initialize the Scanner to scan source.
-  void Initialize(Handle<String> source, int literal_flags = kAllLiterals);
-  void Initialize(Handle<String> source,
-                  unibrow::CharacterStream* stream,
+  V8JavaScriptScanner();
+  void Initialize(UC16CharacterStream* source,
                   int literal_flags = kAllLiterals);
-  void Initialize(Handle<String> source,
-                  int start_position, int end_position,
-                  int literal_flags = kAllLiterals);
-
- protected:
-  StreamInitializer stream_initializer_;
 };
 
 
 class JsonScanner : public Scanner {
  public:
-  JsonScanner();
+  explicit JsonScanner();

[Erik Corry, 2010/12/07 12:27:30]

No explicit.

 
-  // Initialize the Scanner to scan source.
-  void Initialize(Handle<String> source);
+  void Initialize(UC16CharacterStream* source);
 
   // Returns the next token.
   Token::Value Next();
@@ -138,7 +155,7 @@ class JsonScanner : public Scanner {
   // Recognizes all of the single-character tokens directly, or calls a function
   // to scan a number, string or identifier literal.
   // The only allowed whitespace characters between tokens are tab,
-  // carrige-return, newline and space.
+  // carriage-return, newline and space.
   void ScanJson();
 
   // A JSON number (production JSONNumber) is a subset of the valid JavaScript
@@ -159,60 +176,8 @@ class JsonScanner : public Scanner {
   // are the only valid JSON identifiers (productions JSONBooleanLiteral,
   // JSONNullLiteral).
   Token::Value ScanJsonIdentifier(const char* text, Token::Value token);
-
-  StreamInitializer stream_initializer_;
 };
 
-
-// ExternalStringUTF16Buffer
-template <typename StringType, typename CharType>
-ExternalStringUTF16Buffer<StringType, CharType>::ExternalStringUTF16Buffer()
-    : raw_data_(NULL) { }
-
-
-template <typename StringType, typename CharType>
-void ExternalStringUTF16Buffer<StringType, CharType>::Initialize(
-     Handle<StringType> data,
-     int start_position,
-     int end_position) {
-  ASSERT(!data.is_null());
-  raw_data_ = data->resource()->data();
-
-  ASSERT(end_position <= data->length());
-  if (start_position > 0) {
-    SeekForward(start_position);
-  }
-  end_ =
-      end_position != kNoEndPosition ? end_position : data->length();
-}
-
-
-template <typename StringType, typename CharType>
-uc32 ExternalStringUTF16Buffer<StringType, CharType>::Advance() {
-  if (pos_ < end_) {
-    return raw_data_[pos_++];
-  } else {
-    // note: currently the following increment is necessary to avoid a
-    // test-parser problem!
-    pos_++;
-    return static_cast<uc32>(-1);
-  }
-}
-
-
-template <typename StringType, typename CharType>
-void ExternalStringUTF16Buffer<StringType, CharType>::PushBack(uc32 ch) {
-  pos_--;
-  ASSERT(pos_ >= Scanner::kCharacterLookaheadBufferSize);
-  ASSERT(raw_data_[pos_ - Scanner::kCharacterLookaheadBufferSize] == ch);
-}
-
-
-template <typename StringType, typename CharType>
-void ExternalStringUTF16Buffer<StringType, CharType>::SeekForward(int pos) {
-  pos_ = pos;
-}
-
 } }  // namespace v8::internal
 
 #endif  // V8_SCANNER_H_