Remove InputBuffer

R=yangguo@chromium.org
BUG=

Review URL: https://chromiumcodereview.appspot.com/11727004
Patch from Dan Carney <dcarney@google.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@13298 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

src/objects.h

@@ -7313,18 +7313,6 @@ class String: public HeapObject {
   const uc16* GetTwoByteData();
   const uc16* GetTwoByteData(unsigned start);
 
-  // Support for StringInputBuffer
-  static const unibrow::byte* ReadBlock(String* input,
-                                        unibrow::byte* util_buffer,
-                                        unsigned capacity,
-                                        unsigned* remaining,
-                                        unsigned* offset);
-  static const unibrow::byte* ReadBlock(String** input,
-                                        unibrow::byte* util_buffer,
-                                        unsigned capacity,
-                                        unsigned* remaining,
-                                        unsigned* offset);
-
   // Helper function for flattening strings.
   template <typename sinkchar>
   static void WriteToFlat(String* source,
@@ -7383,33 +7371,6 @@ class String: public HeapObject {
                            int32_t type,
                            unsigned length);
 
- protected:
-  class ReadBlockBuffer {
-   public:
-    ReadBlockBuffer(unibrow::byte* util_buffer_,
-                    unsigned cursor_,
-                    unsigned capacity_,
-                    unsigned remaining_) :
-      util_buffer(util_buffer_),
-      cursor(cursor_),
-      capacity(capacity_),
-      remaining(remaining_) {
-    }
-    unibrow::byte* util_buffer;
-    unsigned       cursor;
-    unsigned       capacity;
-    unsigned       remaining;
-  };
-
-  static inline const unibrow::byte* ReadBlock(String* input,
-                                               ReadBlockBuffer* buffer,
-                                               unsigned* offset,
-                                               unsigned max_chars);
-  static void ReadBlockIntoBuffer(String* input,
-                                  ReadBlockBuffer* buffer,
-                                  unsigned* offset_ptr,
-                                  unsigned max_chars);
-
  private:
   // Try to flatten the top level ConsString that is hiding behind this
   // string.  This is a no-op unless the string is a ConsString.  Flatten
@@ -7485,14 +7446,6 @@ class SeqOneByteString: public SeqString {
   // Q.v. String::kMaxLength which is the maximal size of concatenated strings.
   static const int kMaxLength = (kMaxSize - kHeaderSize);
 
-  // Support for StringInputBuffer.
-  inline void SeqOneByteStringReadBlockIntoBuffer(ReadBlockBuffer* buffer,
-                                                unsigned* offset,
-                                                unsigned chars);
-  inline const unibrow::byte* SeqOneByteStringReadBlock(unsigned* remaining,
-                                                      unsigned* offset,
-                                                      unsigned chars);
-
   DECLARE_VERIFIER(SeqOneByteString)
 
  private:
@@ -7537,11 +7490,6 @@ class SeqTwoByteString: public SeqString {
   // Q.v. String::kMaxLength which is the maximal size of concatenated strings.
   static const int kMaxLength = (kMaxSize - kHeaderSize) / sizeof(uint16_t);
 
-  // Support for StringInputBuffer.
-  inline void SeqTwoByteStringReadBlockIntoBuffer(ReadBlockBuffer* buffer,
-                                                  unsigned* offset_ptr,
-                                                  unsigned chars);
-
  private:
   DISALLOW_IMPLICIT_CONSTRUCTORS(SeqTwoByteString);
 };
@@ -7584,14 +7532,6 @@ class ConsString: public String {
   static const int kSecondOffset = kFirstOffset + kPointerSize;
   static const int kSize = kSecondOffset + kPointerSize;
 
-  // Support for StringInputBuffer.
-  inline const unibrow::byte* ConsStringReadBlock(ReadBlockBuffer* buffer,
-                                                  unsigned* offset_ptr,
-                                                  unsigned chars);
-  inline void ConsStringReadBlockIntoBuffer(ReadBlockBuffer* buffer,
-                                            unsigned* offset_ptr,
-                                            unsigned chars);
-
   // Minimum length for a cons string.
   static const int kMinLength = 13;
 
@@ -7636,13 +7576,6 @@ class SlicedString: public String {
   static const int kOffsetOffset = kParentOffset + kPointerSize;
   static const int kSize = kOffsetOffset + kPointerSize;
 
-  // Support for StringInputBuffer
-  inline const unibrow::byte* SlicedStringReadBlock(ReadBlockBuffer* buffer,
-                                                    unsigned* offset_ptr,
-                                                    unsigned chars);
-  inline void SlicedStringReadBlockIntoBuffer(ReadBlockBuffer* buffer,
-                                              unsigned* offset_ptr,
-                                              unsigned chars);
   // Minimum length for a sliced string.
   static const int kMinLength = 13;
 
@@ -7719,14 +7652,6 @@ class ExternalAsciiString: public ExternalString {
   template<typename StaticVisitor>
   inline void ExternalAsciiStringIterateBody();
 
-  // Support for StringInputBuffer.
-  const unibrow::byte* ExternalAsciiStringReadBlock(unsigned* remaining,
-                                                    unsigned* offset,
-                                                    unsigned chars);
-  inline void ExternalAsciiStringReadBlockIntoBuffer(ReadBlockBuffer* buffer,
-                                                     unsigned* offset,
-                                                     unsigned chars);
-
  private:
   DISALLOW_IMPLICIT_CONSTRUCTORS(ExternalAsciiString);
 };
@@ -7767,12 +7692,6 @@ class ExternalTwoByteString: public ExternalString {
   template<typename StaticVisitor>
   inline void ExternalTwoByteStringIterateBody();
 
-
-  // Support for StringInputBuffer.
-  void ExternalTwoByteStringReadBlockIntoBuffer(ReadBlockBuffer* buffer,
-                                                unsigned* offset_ptr,
-                                                unsigned chars);
-
  private:
   DISALLOW_IMPLICIT_CONSTRUCTORS(ExternalTwoByteString);
 };
@@ -7819,24 +7738,6 @@ class FlatStringReader : public Relocatable {
 };
 
 
-// Note that StringInputBuffers are not valid across a GC!  To fix this
-// it would have to store a String Handle instead of a String* and
-// AsciiStringReadBlock would have to be modified to use memcpy.
-//
-// StringInputBuffer is able to traverse any string regardless of how
-// deeply nested a sequence of ConsStrings it is made of.  However,
-// performance will be better if deep strings are flattened before they
-// are traversed.  Since flattening requires memory allocation this is
-// not always desirable, however (esp. in debugging situations).
-class StringInputBuffer: public unibrow::InputBuffer<String, String*, 1024> {
- public:
-  virtual void Seek(unsigned pos);
-  inline StringInputBuffer(): unibrow::InputBuffer<String, String*, 1024>() {}
-  explicit inline StringInputBuffer(String* backing):
-      unibrow::InputBuffer<String, String*, 1024>(backing) {}
-};
-
-
 // A ConsStringOp that returns null.
 // Useful when the operation to apply on a ConsString
 // requires an expensive data structure.

src/scanner.h

@@ -430,10 +430,6 @@ class Scanner {
   // be empty).
   bool ScanRegExpFlags();
 
-  // Tells whether the buffer contains an identifier (no escapes).
-  // Used for checking if a property name is an identifier.
-  static bool IsIdentifier(unibrow::CharacterStream* buffer);
-
  private:
   // The current and look-ahead token.
   struct TokenDesc {

src/unicode-inl.h

@@ -137,109 +137,6 @@ unsigned Utf8::Length(uchar c, int previous) {
   }
 }
 
-uchar CharacterStream::GetNext() {
-  uchar result = DecodeCharacter(buffer_, &cursor_);
-  if (remaining_ == 1) {
-    cursor_ = 0;
-    FillBuffer();
-  } else {
-    remaining_--;
-  }
-  ASSERT(BoundsCheck(cursor_));
-  return result;
-}
-
-#if __BYTE_ORDER == __LITTLE_ENDIAN
-#define IF_LITTLE(expr) expr
-#define IF_BIG(expr)    ((void) 0)
-#elif __BYTE_ORDER == __BIG_ENDIAN
-#define IF_LITTLE(expr) ((void) 0)
-#define IF_BIG(expr)    expr
-#else
-#warning Unknown byte ordering
-#endif
-
-bool CharacterStream::EncodeAsciiCharacter(uchar c, byte* buffer,
-    unsigned capacity, unsigned& offset) {
-  if (offset >= capacity) return false;
-  buffer[offset] = c;
-  offset += 1;
-  return true;
-}
-
-bool CharacterStream::EncodeNonAsciiCharacter(uchar c, byte* buffer,
-    unsigned capacity, unsigned& offset) {
-  unsigned aligned = (offset + 0x3) & ~0x3;
-  if ((aligned + sizeof(uchar)) > capacity)
-    return false;
-  if (offset == aligned) {
-    IF_LITTLE(*reinterpret_cast<uchar*>(buffer + aligned) = (c << 8) | 0x80);
-    IF_BIG(*reinterpret_cast<uchar*>(buffer + aligned) = c | (1 << 31));
-  } else {
-    buffer[offset] = 0x80;
-    IF_LITTLE(*reinterpret_cast<uchar*>(buffer + aligned) = c << 8);
-    IF_BIG(*reinterpret_cast<uchar*>(buffer + aligned) = c);
-  }
-  offset = aligned + sizeof(uchar);
-  return true;
-}
-
-bool CharacterStream::EncodeCharacter(uchar c, byte* buffer, unsigned capacity,
-    unsigned& offset) {
-  if (c <= Utf8::kMaxOneByteChar) {
-    return EncodeAsciiCharacter(c, buffer, capacity, offset);
-  } else {
-    return EncodeNonAsciiCharacter(c, buffer, capacity, offset);
-  }
-}
-
-uchar CharacterStream::DecodeCharacter(const byte* buffer, unsigned* offset) {
-  byte b = buffer[*offset];
-  if (b <= Utf8::kMaxOneByteChar) {
-    (*offset)++;
-    return b;
-  } else {
-    unsigned aligned = (*offset + 0x3) & ~0x3;
-    *offset = aligned + sizeof(uchar);
-    IF_LITTLE(return *reinterpret_cast<const uchar*>(buffer + aligned) >> 8);
-    IF_BIG(return *reinterpret_cast<const uchar*>(buffer + aligned) &
-                    ~(1 << 31));
-  }
-}
-
-#undef IF_LITTLE
-#undef IF_BIG
-
-template <class R, class I, unsigned s>
-void InputBuffer<R, I, s>::FillBuffer() {
-  buffer_ = R::ReadBlock(input_, util_buffer_, s, &remaining_, &offset_);
-}
-
-template <class R, class I, unsigned s>
-void InputBuffer<R, I, s>::Rewind() {
-  Reset(input_);
-}
-
-template <class R, class I, unsigned s>
-void InputBuffer<R, I, s>::Reset(unsigned position, I input) {
-  input_ = input;
-  remaining_ = 0;
-  cursor_ = 0;
-  offset_ = position;
-  buffer_ = R::ReadBlock(input_, util_buffer_, s, &remaining_, &offset_);
-}
-
-template <class R, class I, unsigned s>
-void InputBuffer<R, I, s>::Reset(I input) {
-  Reset(0, input);
-}
-
-template <class R, class I, unsigned s>
-void InputBuffer<R, I, s>::Seek(unsigned position) {
-  offset_ = position;
-  buffer_ = R::ReadBlock(input_, util_buffer_, s, &remaining_, &offset_);
-}
-
 Utf8DecoderBase::Utf8DecoderBase()
   : unbuffered_start_(NULL),
     utf16_length_(0),

src/unicode.cc

@@ -277,33 +277,6 @@ uchar Utf8::CalculateValue(const byte* str,
 }
 
 
-unsigned CharacterStream::Length() {
-  unsigned result = 0;
-  while (has_more()) {
-    result++;
-    GetNext();
-  }
-  Rewind();
-  return result;
-}
-
-unsigned CharacterStream::Utf16Length() {
-  unsigned result = 0;
-  while (has_more()) {
-    uchar c = GetNext();
-    result += c > Utf16::kMaxNonSurrogateCharCode ? 2 : 1;
-  }
-  Rewind();
-  return result;
-}
-
-void CharacterStream::Seek(unsigned position) {
-  Rewind();
-  for (unsigned i = 0; i < position; i++) {
-    GetNext();
-  }
-}
-
 void Utf8DecoderBase::Reset(uint16_t* buffer,
                             unsigned buffer_length,
                             const uint8_t* stream,

src/unicode.h

@@ -100,21 +100,6 @@ class UnicodeData {
   static const uchar kMaxCodePoint;
 };
 
-// --- U t f   8   a n d   16 ---
-
-template <typename Data>
-class Buffer {
- public:
-  inline Buffer(Data data, unsigned length) : data_(data), length_(length) { }
-  inline Buffer() : data_(0), length_(0) { }
-  Data data() { return data_; }
-  unsigned length() { return length_; }
- private:
-  Data data_;
-  unsigned length_;
-};
-
-
 class Utf16 {
  public:
   static inline bool IsLeadSurrogate(int code) {
@@ -173,72 +158,6 @@ class Utf8 {
                               unsigned* cursor);
 };
 
-// --- C h a r a c t e r   S t r e a m ---
-
-class CharacterStream {
- public:
-  inline uchar GetNext();
-  inline bool has_more() { return remaining_ != 0; }
-  // Note that default implementation is not efficient.
-  virtual void Seek(unsigned);
-  unsigned Length();
-  unsigned Utf16Length();
-  virtual ~CharacterStream() { }
-  static inline bool EncodeCharacter(uchar c, byte* buffer, unsigned capacity,
-      unsigned& offset);
-  static inline bool EncodeAsciiCharacter(uchar c, byte* buffer,
-      unsigned capacity, unsigned& offset);
-  static inline bool EncodeNonAsciiCharacter(uchar c, byte* buffer,
-      unsigned capacity, unsigned& offset);
-  static inline uchar DecodeCharacter(const byte* buffer, unsigned* offset);
-  virtual void Rewind() = 0;
-
- protected:
-  virtual void FillBuffer() = 0;
-  virtual bool BoundsCheck(unsigned offset) = 0;
-  // The number of characters left in the current buffer
-  unsigned remaining_;
-  // The current offset within the buffer
-  unsigned cursor_;
-  // The buffer containing the decoded characters.
-  const byte* buffer_;
-};
-
-// --- I n p u t   B u f f e r ---
-
-/**
- * Provides efficient access to encoded characters in strings.  It
- * does so by reading characters one block at a time, rather than one
- * character at a time, which gives string implementations an
- * opportunity to optimize the decoding.
- */
-template <class Reader, class Input = Reader*, unsigned kSize = 256>
-class InputBuffer : public CharacterStream {
- public:
-  virtual void Rewind();
-  inline void Reset(Input input);
-  void Seek(unsigned position);
-  inline void Reset(unsigned position, Input input);
- protected:
-  InputBuffer() { }
-  explicit InputBuffer(Input input) { Reset(input); }
-  virtual void FillBuffer();
-  virtual bool BoundsCheck(unsigned offset) {
-    return (buffer_ != util_buffer_) || (offset < kSize);
-  }
-
-  // A custom offset that can be used by the string implementation to
-  // mark progress within the encoded string.
-  unsigned offset_;
-  // The input string
-  Input input_;
-  // To avoid heap allocation, we keep an internal buffer to which
-  // the encoded string can write its characters.  The string
-  // implementation is free to decide whether it wants to use this
-  // buffer or not.
-  byte util_buffer_[kSize];
-};
-
 
 class Utf8DecoderBase {
  public:

test/cctest/test-strings.cc

 // Copyright 2012 the V8 project authors. All rights reserved.
 
 // Check that we can traverse very deep stacks of ConsStrings using
-// StringInputBuffer.  Check that Get(int) works on very deep stacks
+// StringCharacterStram.  Check that Get(int) works on very deep stacks
 // of ConsStrings.  These operations may not be very fast, but they
 // should be possible without getting errors due to too deep recursion.
 
@@ -514,23 +514,16 @@ static Handle<String> ConstructBalanced(
 }
 
 
-static StringInputBuffer buffer;
 static ConsStringIteratorOp cons_string_iterator_op_1;
 static ConsStringIteratorOp cons_string_iterator_op_2;
 
 static void Traverse(Handle<String> s1, Handle<String> s2) {
   int i = 0;
-  buffer.Reset(*s1);
   StringCharacterStream character_stream_1(*s1, &cons_string_iterator_op_1);
   StringCharacterStream character_stream_2(*s2, &cons_string_iterator_op_2);
-  StringInputBuffer buffer2(*s2);
-  while (buffer.has_more()) {
-    CHECK(buffer2.has_more());
-    CHECK(character_stream_1.HasMore());
+  while (character_stream_1.HasMore()) {
     CHECK(character_stream_2.HasMore());
-    uint16_t c = buffer.GetNext();
-    CHECK_EQ(c, buffer2.GetNext());
-    CHECK_EQ(c, character_stream_1.GetNext());
+    uint16_t c = character_stream_1.GetNext();
     CHECK_EQ(c, character_stream_2.GetNext());
     i++;
   }
@@ -543,17 +536,11 @@ static void Traverse(Handle<String> s1, Handle<String> s2) {
 
 static void TraverseFirst(Handle<String> s1, Handle<String> s2, int chars) {
   int i = 0;
-  buffer.Reset(*s1);
-  StringInputBuffer buffer2(*s2);
   StringCharacterStream character_stream_1(*s1, &cons_string_iterator_op_1);
   StringCharacterStream character_stream_2(*s2, &cons_string_iterator_op_2);
-  while (buffer.has_more() && i < chars) {
-    CHECK(buffer2.has_more());
-    CHECK(character_stream_1.HasMore());
+  while (character_stream_1.HasMore() && i < chars) {
     CHECK(character_stream_2.HasMore());
-    uint16_t c = buffer.GetNext();
-    CHECK_EQ(c, buffer2.GetNext());
-    CHECK_EQ(c, character_stream_1.GetNext());
+    uint16_t c = character_stream_1.GetNext();
     CHECK_EQ(c, character_stream_2.GetNext());
     i++;
   }