src/scanner-character-streams.h

   1 // Copyright 2011 the V8 project authors. All rights reserved.
   2 // Use of this source code is governed by a BSD-style license that can be
   3 // found in the LICENSE file.
   4
   5 #ifndef V8_SCANNER_CHARACTER_STREAMS_H_
   6 #define V8_SCANNER_CHARACTER_STREAMS_H_
   7
   8 #include "src/scanner.h"
   9
  10 namespace v8 {
  11 namespace internal {
  12
  13 // A buffered character stream based on a random access character
  14 // source (ReadBlock can be called with pos_ pointing to any position,
  15 // even positions before the current).
  16 class BufferedUtf16CharacterStream: public Utf16CharacterStream {
  17  public:
  18   BufferedUtf16CharacterStream();
  19   virtual ~BufferedUtf16CharacterStream();
  20
  21   virtual void PushBack(uc32 character);
  22
  23  protected:
  24   static const size_t kBufferSize = 512;
  25   static const size_t kPushBackStepSize = 16;
  26
  27   virtual size_t SlowSeekForward(size_t delta);
  28   virtual bool ReadBlock();
  29   virtual void SlowPushBack(uc16 character);
  30
  31   virtual size_t BufferSeekForward(size_t delta) = 0;
  32   virtual size_t FillBuffer(size_t position) = 0;
  33
  34   const uc16* pushback_limit_;
  35   uc16 buffer_[kBufferSize];
  36 };
  37
  38
  39 // Generic string stream.
  40 class GenericStringUtf16CharacterStream: public BufferedUtf16CharacterStream {
  41  public:
  42   GenericStringUtf16CharacterStream(Handle<String> data, size_t start_position,
  43                                     size_t end_position);
  44   virtual ~GenericStringUtf16CharacterStream();
  45
  46  protected:
  47   virtual size_t BufferSeekForward(size_t delta);
  48   virtual size_t FillBuffer(size_t position);
  49
  50   Handle<String> string_;
  51   size_t length_;
  52 };
  53
  54
  55 // Utf16 stream based on a literal UTF-8 string.
  56 class Utf8ToUtf16CharacterStream: public BufferedUtf16CharacterStream {
  57  public:
  58   Utf8ToUtf16CharacterStream(const byte* data, size_t length);
  59   virtual ~Utf8ToUtf16CharacterStream();
  60
  61   static size_t CopyChars(uint16_t* dest, size_t length, const byte* src,
  62                           size_t* src_pos, size_t src_length);
  63
  64  protected:
  65   virtual size_t BufferSeekForward(size_t delta);
  66   virtual size_t FillBuffer(size_t char_position);
  67   void SetRawPosition(size_t char_position);
  68
  69   const byte* raw_data_;
  70   size_t raw_data_length_;  // Measured in bytes, not characters.
  71   size_t raw_data_pos_;
  72   // The character position of the character at raw_data[raw_data_pos_].
  73   // Not necessarily the same as pos_.
  74   size_t raw_character_position_;
  75 };
  76
  77
  78 // ExternalStreamingStream is a wrapper around an ExternalSourceStream (see
  79 // include/v8.h) subclass implemented by the embedder.
  80 class ExternalStreamingStream : public BufferedUtf16CharacterStream {
  81  public:
  82   ExternalStreamingStream(ScriptCompiler::ExternalSourceStream* source_stream,
  83                           v8::ScriptCompiler::StreamedSource::Encoding encoding)
  84       : source_stream_(source_stream),
  85         encoding_(encoding),
  86         current_data_(NULL),
  87         current_data_offset_(0),
  88         current_data_length_(0),
  89         utf8_split_char_buffer_length_(0) {}
  90
  91   virtual ~ExternalStreamingStream() { delete[] current_data_; }
  92
  93   size_t BufferSeekForward(size_t delta) override {
  94     // We never need to seek forward when streaming scripts. We only seek
  95     // forward when we want to parse a function whose location we already know,
  96     // and when streaming, we don't know the locations of anything we haven't
  97     // seen yet.
  98     UNREACHABLE();
  99     return 0;
 100   }
 101
 102   size_t FillBuffer(size_t position) override;
 103
 104  private:
 105   void HandleUtf8SplitCharacters(size_t* data_in_buffer);
 106
 107   ScriptCompiler::ExternalSourceStream* source_stream_;
 108   v8::ScriptCompiler::StreamedSource::Encoding encoding_;
 109   const uint8_t* current_data_;
 110   size_t current_data_offset_;
 111   size_t current_data_length_;
 112   // For converting UTF-8 characters which are split across two data chunks.
 113   uint8_t utf8_split_char_buffer_[4];
 114   size_t utf8_split_char_buffer_length_;
 115 };
 116
 117
 118 // UTF16 buffer to read characters from an external string.
 119 class ExternalTwoByteStringUtf16CharacterStream: public Utf16CharacterStream {
 120  public:
 121   ExternalTwoByteStringUtf16CharacterStream(Handle<ExternalTwoByteString> data,
 122                                             int start_position,
 123                                             int end_position);
 124   virtual ~ExternalTwoByteStringUtf16CharacterStream();
 125
 126   virtual void PushBack(uc32 character) {
 127     DCHECK(buffer_cursor_ > raw_data_);
 128     buffer_cursor_--;
 129     pos_--;
 130   }
 131
 132  protected:
 133   virtual size_t SlowSeekForward(size_t delta) {
 134     // Fast case always handles seeking.
 135     return 0;
 136   }
 137   virtual bool ReadBlock() {
 138     // Entire string is read at start.
 139     return false;
 140   }
 141   Handle<ExternalTwoByteString> source_;
 142   const uc16* raw_data_;  // Pointer to the actual array of characters.
 143 };
 144
 145 } }  // namespace v8::internal
 146
 147 #endif  // V8_SCANNER_CHARACTER_STREAMS_H_