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.
5 #ifndef V8_SCANNER_CHARACTER_STREAMS_H_
6 #define V8_SCANNER_CHARACTER_STREAMS_H_
8 #include "src/scanner.h"
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 {
18 BufferedUtf16CharacterStream();
19 virtual ~BufferedUtf16CharacterStream();
21 virtual void PushBack(uc32 character);
24 static const unsigned kBufferSize = 512;
25 static const unsigned kPushBackStepSize = 16;
27 virtual unsigned SlowSeekForward(unsigned delta);
28 virtual bool ReadBlock();
29 virtual void SlowPushBack(uc16 character);
31 virtual unsigned BufferSeekForward(unsigned delta) = 0;
32 virtual unsigned FillBuffer(unsigned position) = 0;
34 const uc16* pushback_limit_;
35 uc16 buffer_[kBufferSize];
39 // Generic string stream.
40 class GenericStringUtf16CharacterStream: public BufferedUtf16CharacterStream {
42 GenericStringUtf16CharacterStream(Handle<String> data,
43 unsigned start_position,
44 unsigned end_position);
45 virtual ~GenericStringUtf16CharacterStream();
48 virtual unsigned BufferSeekForward(unsigned delta);
49 virtual unsigned FillBuffer(unsigned position);
51 Handle<String> string_;
56 // Utf16 stream based on a literal UTF-8 string.
57 class Utf8ToUtf16CharacterStream: public BufferedUtf16CharacterStream {
59 Utf8ToUtf16CharacterStream(const byte* data, unsigned length);
60 virtual ~Utf8ToUtf16CharacterStream();
62 static unsigned CopyChars(uint16_t* dest, unsigned length, const byte* src,
63 unsigned* src_pos, unsigned src_length);
66 virtual unsigned BufferSeekForward(unsigned delta);
67 virtual unsigned FillBuffer(unsigned char_position);
68 void SetRawPosition(unsigned char_position);
70 const byte* raw_data_;
71 unsigned raw_data_length_; // Measured in bytes, not characters.
72 unsigned raw_data_pos_;
73 // The character position of the character at raw_data[raw_data_pos_].
74 // Not necessarily the same as pos_.
75 unsigned raw_character_position_;
79 // ExternalStreamingStream is a wrapper around an ExternalSourceStream (see
80 // include/v8.h) subclass implemented by the embedder.
81 class ExternalStreamingStream : public BufferedUtf16CharacterStream {
83 ExternalStreamingStream(ScriptCompiler::ExternalSourceStream* source_stream,
84 v8::ScriptCompiler::StreamedSource::Encoding encoding)
85 : source_stream_(source_stream),
88 current_data_offset_(0),
89 current_data_length_(0),
90 utf8_split_char_buffer_length_(0) {}
92 virtual ~ExternalStreamingStream() { delete[] current_data_; }
94 virtual unsigned BufferSeekForward(unsigned delta) OVERRIDE {
95 // We never need to seek forward when streaming scripts. We only seek
96 // forward when we want to parse a function whose location we already know,
97 // and when streaming, we don't know the locations of anything we haven't
103 virtual unsigned FillBuffer(unsigned position) OVERRIDE;
106 void HandleUtf8SplitCharacters(unsigned* data_in_buffer);
108 ScriptCompiler::ExternalSourceStream* source_stream_;
109 v8::ScriptCompiler::StreamedSource::Encoding encoding_;
110 const uint8_t* current_data_;
111 unsigned current_data_offset_;
112 unsigned current_data_length_;
113 // For converting UTF-8 characters which are split across two data chunks.
114 uint8_t utf8_split_char_buffer_[4];
115 unsigned utf8_split_char_buffer_length_;
119 // UTF16 buffer to read characters from an external string.
120 class ExternalTwoByteStringUtf16CharacterStream: public Utf16CharacterStream {
122 ExternalTwoByteStringUtf16CharacterStream(Handle<ExternalTwoByteString> data,
125 virtual ~ExternalTwoByteStringUtf16CharacterStream();
127 virtual void PushBack(uc32 character) {
128 DCHECK(buffer_cursor_ > raw_data_);
134 virtual unsigned SlowSeekForward(unsigned delta) {
135 // Fast case always handles seeking.
138 virtual bool ReadBlock() {
139 // Entire string is read at start.
142 Handle<ExternalTwoByteString> source_;
143 const uc16* raw_data_; // Pointer to the actual array of characters.
146 } } // namespace v8::internal
148 #endif // V8_SCANNER_CHARACTER_STREAMS_H_