Source file src/bufio/bufio.go

Documentation: bufio

  // Copyright 2009 The Go Authors. All rights reserved.
  // Use of this source code is governed by a BSD-style
  // license that can be found in the LICENSE file.
  
  // Package bufio implements buffered I/O. It wraps an io.Reader or io.Writer
  // object, creating another object (Reader or Writer) that also implements
  // the interface but provides buffering and some help for textual I/O.
  package bufio
  
  import (
  	"bytes"
  	"errors"
  	"io"
  	"unicode/utf8"
  )
  
  const (
  	defaultBufSize = 4096
  )
  
  var (
  	ErrInvalidUnreadByte = errors.New("bufio: invalid use of UnreadByte")
  	ErrInvalidUnreadRune = errors.New("bufio: invalid use of UnreadRune")
  	ErrBufferFull        = errors.New("bufio: buffer full")
  	ErrNegativeCount     = errors.New("bufio: negative count")
  )
  
  // Buffered input.
  
  // Reader implements buffering for an io.Reader object.
  type Reader struct {
  	buf          []byte
  	rd           io.Reader // reader provided by the client
  	r, w         int       // buf read and write positions
  	err          error
  	lastByte     int
  	lastRuneSize int
  }
  
  const minReadBufferSize = 16
  const maxConsecutiveEmptyReads = 100
  
  // NewReaderSize returns a new Reader whose buffer has at least the specified
  // size. If the argument io.Reader is already a Reader with large enough
  // size, it returns the underlying Reader.
  func NewReaderSize(rd io.Reader, size int) *Reader {
  	// Is it already a Reader?
  	b, ok := rd.(*Reader)
  	if ok && len(b.buf) >= size {
  		return b
  	}
  	if size < minReadBufferSize {
  		size = minReadBufferSize
  	}
  	r := new(Reader)
  	r.reset(make([]byte, size), rd)
  	return r
  }
  
  // NewReader returns a new Reader whose buffer has the default size.
  func NewReader(rd io.Reader) *Reader {
  	return NewReaderSize(rd, defaultBufSize)
  }
  
  // Reset discards any buffered data, resets all state, and switches
  // the buffered reader to read from r.
  func (b *Reader) Reset(r io.Reader) {
  	b.reset(b.buf, r)
  }
  
  func (b *Reader) reset(buf []byte, r io.Reader) {
  	*b = Reader{
  		buf:          buf,
  		rd:           r,
  		lastByte:     -1,
  		lastRuneSize: -1,
  	}
  }
  
  var errNegativeRead = errors.New("bufio: reader returned negative count from Read")
  
  // fill reads a new chunk into the buffer.
  func (b *Reader) fill() {
  	// Slide existing data to beginning.
  	if b.r > 0 {
  		copy(b.buf, b.buf[b.r:b.w])
  		b.w -= b.r
  		b.r = 0
  	}
  
  	if b.w >= len(b.buf) {
  		panic("bufio: tried to fill full buffer")
  	}
  
  	// Read new data: try a limited number of times.
  	for i := maxConsecutiveEmptyReads; i > 0; i-- {
  		n, err := b.rd.Read(b.buf[b.w:])
  		if n < 0 {
  			panic(errNegativeRead)
  		}
  		b.w += n
  		if err != nil {
  			b.err = err
  			return
  		}
  		if n > 0 {
  			return
  		}
  	}
  	b.err = io.ErrNoProgress
  }
  
  func (b *Reader) readErr() error {
  	err := b.err
  	b.err = nil
  	return err
  }
  
  // Peek returns the next n bytes without advancing the reader. The bytes stop
  // being valid at the next read call. If Peek returns fewer than n bytes, it
  // also returns an error explaining why the read is short. The error is
  // ErrBufferFull if n is larger than b's buffer size.
  func (b *Reader) Peek(n int) ([]byte, error) {
  	if n < 0 {
  		return nil, ErrNegativeCount
  	}
  
  	for b.w-b.r < n && b.w-b.r < len(b.buf) && b.err == nil {
  		b.fill() // b.w-b.r < len(b.buf) => buffer is not full
  	}
  
  	if n > len(b.buf) {
  		return b.buf[b.r:b.w], ErrBufferFull
  	}
  
  	// 0 <= n <= len(b.buf)
  	var err error
  	if avail := b.w - b.r; avail < n {
  		// not enough data in buffer
  		n = avail
  		err = b.readErr()
  		if err == nil {
  			err = ErrBufferFull
  		}
  	}
  	return b.buf[b.r : b.r+n], err
  }
  
  // Discard skips the next n bytes, returning the number of bytes discarded.
  //
  // If Discard skips fewer than n bytes, it also returns an error.
  // If 0 <= n <= b.Buffered(), Discard is guaranteed to succeed without
  // reading from the underlying io.Reader.
  func (b *Reader) Discard(n int) (discarded int, err error) {
  	if n < 0 {
  		return 0, ErrNegativeCount
  	}
  	if n == 0 {
  		return
  	}
  	remain := n
  	for {
  		skip := b.Buffered()
  		if skip == 0 {
  			b.fill()
  			skip = b.Buffered()
  		}
  		if skip > remain {
  			skip = remain
  		}
  		b.r += skip
  		remain -= skip
  		if remain == 0 {
  			return n, nil
  		}
  		if b.err != nil {
  			return n - remain, b.readErr()
  		}
  	}
  }
  
  // Read reads data into p.
  // It returns the number of bytes read into p.
  // The bytes are taken from at most one Read on the underlying Reader,
  // hence n may be less than len(p).
  // At EOF, the count will be zero and err will be io.EOF.
  func (b *Reader) Read(p []byte) (n int, err error) {
  	n = len(p)
  	if n == 0 {
  		return 0, b.readErr()
  	}
  	if b.r == b.w {
  		if b.err != nil {
  			return 0, b.readErr()
  		}
  		if len(p) >= len(b.buf) {
  			// Large read, empty buffer.
  			// Read directly into p to avoid copy.
  			n, b.err = b.rd.Read(p)
  			if n < 0 {
  				panic(errNegativeRead)
  			}
  			if n > 0 {
  				b.lastByte = int(p[n-1])
  				b.lastRuneSize = -1
  			}
  			return n, b.readErr()
  		}
  		// One read.
  		// Do not use b.fill, which will loop.
  		b.r = 0
  		b.w = 0
  		n, b.err = b.rd.Read(b.buf)
  		if n < 0 {
  			panic(errNegativeRead)
  		}
  		if n == 0 {
  			return 0, b.readErr()
  		}
  		b.w += n
  	}
  
  	// copy as much as we can
  	n = copy(p, b.buf[b.r:b.w])
  	b.r += n
  	b.lastByte = int(b.buf[b.r-1])
  	b.lastRuneSize = -1
  	return n, nil
  }
  
  // ReadByte reads and returns a single byte.
  // If no byte is available, returns an error.
  func (b *Reader) ReadByte() (byte, error) {
  	b.lastRuneSize = -1
  	for b.r == b.w {
  		if b.err != nil {
  			return 0, b.readErr()
  		}
  		b.fill() // buffer is empty
  	}
  	c := b.buf[b.r]
  	b.r++
  	b.lastByte = int(c)
  	return c, nil
  }
  
  // UnreadByte unreads the last byte. Only the most recently read byte can be unread.
  func (b *Reader) UnreadByte() error {
  	if b.lastByte < 0 || b.r == 0 && b.w > 0 {
  		return ErrInvalidUnreadByte
  	}
  	// b.r > 0 || b.w == 0
  	if b.r > 0 {
  		b.r--
  	} else {
  		// b.r == 0 && b.w == 0
  		b.w = 1
  	}
  	b.buf[b.r] = byte(b.lastByte)
  	b.lastByte = -1
  	b.lastRuneSize = -1
  	return nil
  }
  
  // ReadRune reads a single UTF-8 encoded Unicode character and returns the
  // rune and its size in bytes. If the encoded rune is invalid, it consumes one byte
  // and returns unicode.ReplacementChar (U+FFFD) with a size of 1.
  func (b *Reader) ReadRune() (r rune, size int, err error) {
  	for b.r+utf8.UTFMax > b.w && !utf8.FullRune(b.buf[b.r:b.w]) && b.err == nil && b.w-b.r < len(b.buf) {
  		b.fill() // b.w-b.r < len(buf) => buffer is not full
  	}
  	b.lastRuneSize = -1
  	if b.r == b.w {
  		return 0, 0, b.readErr()
  	}
  	r, size = rune(b.buf[b.r]), 1
  	if r >= utf8.RuneSelf {
  		r, size = utf8.DecodeRune(b.buf[b.r:b.w])
  	}
  	b.r += size
  	b.lastByte = int(b.buf[b.r-1])
  	b.lastRuneSize = size
  	return r, size, nil
  }
  
  // UnreadRune unreads the last rune. If the most recent read operation on
  // the buffer was not a ReadRune, UnreadRune returns an error.  (In this
  // regard it is stricter than UnreadByte, which will unread the last byte
  // from any read operation.)
  func (b *Reader) UnreadRune() error {
  	if b.lastRuneSize < 0 || b.r < b.lastRuneSize {
  		return ErrInvalidUnreadRune
  	}
  	b.r -= b.lastRuneSize
  	b.lastByte = -1
  	b.lastRuneSize = -1
  	return nil
  }
  
  // Buffered returns the number of bytes that can be read from the current buffer.
  func (b *Reader) Buffered() int { return b.w - b.r }
  
  // ReadSlice reads until the first occurrence of delim in the input,
  // returning a slice pointing at the bytes in the buffer.
  // The bytes stop being valid at the next read.
  // If ReadSlice encounters an error before finding a delimiter,
  // it returns all the data in the buffer and the error itself (often io.EOF).
  // ReadSlice fails with error ErrBufferFull if the buffer fills without a delim.
  // Because the data returned from ReadSlice will be overwritten
  // by the next I/O operation, most clients should use
  // ReadBytes or ReadString instead.
  // ReadSlice returns err != nil if and only if line does not end in delim.
  func (b *Reader) ReadSlice(delim byte) (line []byte, err error) {
  	for {
  		// Search buffer.
  		if i := bytes.IndexByte(b.buf[b.r:b.w], delim); i >= 0 {
  			line = b.buf[b.r : b.r+i+1]
  			b.r += i + 1
  			break
  		}
  
  		// Pending error?
  		if b.err != nil {
  			line = b.buf[b.r:b.w]
  			b.r = b.w
  			err = b.readErr()
  			break
  		}
  
  		// Buffer full?
  		if b.Buffered() >= len(b.buf) {
  			b.r = b.w
  			line = b.buf
  			err = ErrBufferFull
  			break
  		}
  
  		b.fill() // buffer is not full
  	}
  
  	// Handle last byte, if any.
  	if i := len(line) - 1; i >= 0 {
  		b.lastByte = int(line[i])
  		b.lastRuneSize = -1
  	}
  
  	return
  }
  
  // ReadLine is a low-level line-reading primitive. Most callers should use
  // ReadBytes('\n') or ReadString('\n') instead or use a Scanner.
  //
  // ReadLine tries to return a single line, not including the end-of-line bytes.
  // If the line was too long for the buffer then isPrefix is set and the
  // beginning of the line is returned. The rest of the line will be returned
  // from future calls. isPrefix will be false when returning the last fragment
  // of the line. The returned buffer is only valid until the next call to
  // ReadLine. ReadLine either returns a non-nil line or it returns an error,
  // never both.
  //
  // The text returned from ReadLine does not include the line end ("\r\n" or "\n").
  // No indication or error is given if the input ends without a final line end.
  // Calling UnreadByte after ReadLine will always unread the last byte read
  // (possibly a character belonging to the line end) even if that byte is not
  // part of the line returned by ReadLine.
  func (b *Reader) ReadLine() (line []byte, isPrefix bool, err error) {
  	line, err = b.ReadSlice('\n')
  	if err == ErrBufferFull {
  		// Handle the case where "\r\n" straddles the buffer.
  		if len(line) > 0 && line[len(line)-1] == '\r' {
  			// Put the '\r' back on buf and drop it from line.
  			// Let the next call to ReadLine check for "\r\n".
  			if b.r == 0 {
  				// should be unreachable
  				panic("bufio: tried to rewind past start of buffer")
  			}
  			b.r--
  			line = line[:len(line)-1]
  		}
  		return line, true, nil
  	}
  
  	if len(line) == 0 {
  		if err != nil {
  			line = nil
  		}
  		return
  	}
  	err = nil
  
  	if line[len(line)-1] == '\n' {
  		drop := 1
  		if len(line) > 1 && line[len(line)-2] == '\r' {
  			drop = 2
  		}
  		line = line[:len(line)-drop]
  	}
  	return
  }
  
  // ReadBytes reads until the first occurrence of delim in the input,
  // returning a slice containing the data up to and including the delimiter.
  // If ReadBytes encounters an error before finding a delimiter,
  // it returns the data read before the error and the error itself (often io.EOF).
  // ReadBytes returns err != nil if and only if the returned data does not end in
  // delim.
  // For simple uses, a Scanner may be more convenient.
  func (b *Reader) ReadBytes(delim byte) ([]byte, error) {
  	// Use ReadSlice to look for array,
  	// accumulating full buffers.
  	var frag []byte
  	var full [][]byte
  	var err error
  	for {
  		var e error
  		frag, e = b.ReadSlice(delim)
  		if e == nil { // got final fragment
  			break
  		}
  		if e != ErrBufferFull { // unexpected error
  			err = e
  			break
  		}
  
  		// Make a copy of the buffer.
  		buf := make([]byte, len(frag))
  		copy(buf, frag)
  		full = append(full, buf)
  	}
  
  	// Allocate new buffer to hold the full pieces and the fragment.
  	n := 0
  	for i := range full {
  		n += len(full[i])
  	}
  	n += len(frag)
  
  	// Copy full pieces and fragment in.
  	buf := make([]byte, n)
  	n = 0
  	for i := range full {
  		n += copy(buf[n:], full[i])
  	}
  	copy(buf[n:], frag)
  	return buf, err
  }
  
  // ReadString reads until the first occurrence of delim in the input,
  // returning a string containing the data up to and including the delimiter.
  // If ReadString encounters an error before finding a delimiter,
  // it returns the data read before the error and the error itself (often io.EOF).
  // ReadString returns err != nil if and only if the returned data does not end in
  // delim.
  // For simple uses, a Scanner may be more convenient.
  func (b *Reader) ReadString(delim byte) (string, error) {
  	bytes, err := b.ReadBytes(delim)
  	return string(bytes), err
  }
  
  // WriteTo implements io.WriterTo.
  // This may make multiple calls to the Read method of the underlying Reader.
  func (b *Reader) WriteTo(w io.Writer) (n int64, err error) {
  	n, err = b.writeBuf(w)
  	if err != nil {
  		return
  	}
  
  	if r, ok := b.rd.(io.WriterTo); ok {
  		m, err := r.WriteTo(w)
  		n += m
  		return n, err
  	}
  
  	if w, ok := w.(io.ReaderFrom); ok {
  		m, err := w.ReadFrom(b.rd)
  		n += m
  		return n, err
  	}
  
  	if b.w-b.r < len(b.buf) {
  		b.fill() // buffer not full
  	}
  
  	for b.r < b.w {
  		// b.r < b.w => buffer is not empty
  		m, err := b.writeBuf(w)
  		n += m
  		if err != nil {
  			return n, err
  		}
  		b.fill() // buffer is empty
  	}
  
  	if b.err == io.EOF {
  		b.err = nil
  	}
  
  	return n, b.readErr()
  }
  
  var errNegativeWrite = errors.New("bufio: writer returned negative count from Write")
  
  // writeBuf writes the Reader's buffer to the writer.
  func (b *Reader) writeBuf(w io.Writer) (int64, error) {
  	n, err := w.Write(b.buf[b.r:b.w])
  	if n < 0 {
  		panic(errNegativeWrite)
  	}
  	b.r += n
  	return int64(n), err
  }
  
  // buffered output
  
  // Writer implements buffering for an io.Writer object.
  // If an error occurs writing to a Writer, no more data will be
  // accepted and all subsequent writes, and Flush, will return the error.
  // After all data has been written, the client should call the
  // Flush method to guarantee all data has been forwarded to
  // the underlying io.Writer.
  type Writer struct {
  	err error
  	buf []byte
  	n   int
  	wr  io.Writer
  }
  
  // NewWriterSize returns a new Writer whose buffer has at least the specified
  // size. If the argument io.Writer is already a Writer with large enough
  // size, it returns the underlying Writer.
  func NewWriterSize(w io.Writer, size int) *Writer {
  	// Is it already a Writer?
  	b, ok := w.(*Writer)
  	if ok && len(b.buf) >= size {
  		return b
  	}
  	if size <= 0 {
  		size = defaultBufSize
  	}
  	return &Writer{
  		buf: make([]byte, size),
  		wr:  w,
  	}
  }
  
  // NewWriter returns a new Writer whose buffer has the default size.
  func NewWriter(w io.Writer) *Writer {
  	return NewWriterSize(w, defaultBufSize)
  }
  
  // Reset discards any unflushed buffered data, clears any error, and
  // resets b to write its output to w.
  func (b *Writer) Reset(w io.Writer) {
  	b.err = nil
  	b.n = 0
  	b.wr = w
  }
  
  // Flush writes any buffered data to the underlying io.Writer.
  func (b *Writer) Flush() error {
  	if b.err != nil {
  		return b.err
  	}
  	if b.n == 0 {
  		return nil
  	}
  	n, err := b.wr.Write(b.buf[0:b.n])
  	if n < b.n && err == nil {
  		err = io.ErrShortWrite
  	}
  	if err != nil {
  		if n > 0 && n < b.n {
  			copy(b.buf[0:b.n-n], b.buf[n:b.n])
  		}
  		b.n -= n
  		b.err = err
  		return err
  	}
  	b.n = 0
  	return nil
  }
  
  // Available returns how many bytes are unused in the buffer.
  func (b *Writer) Available() int { return len(b.buf) - b.n }
  
  // Buffered returns the number of bytes that have been written into the current buffer.
  func (b *Writer) Buffered() int { return b.n }
  
  // Write writes the contents of p into the buffer.
  // It returns the number of bytes written.
  // If nn < len(p), it also returns an error explaining
  // why the write is short.
  func (b *Writer) Write(p []byte) (nn int, err error) {
  	for len(p) > b.Available() && b.err == nil {
  		var n int
  		if b.Buffered() == 0 {
  			// Large write, empty buffer.
  			// Write directly from p to avoid copy.
  			n, b.err = b.wr.Write(p)
  		} else {
  			n = copy(b.buf[b.n:], p)
  			b.n += n
  			b.Flush()
  		}
  		nn += n
  		p = p[n:]
  	}
  	if b.err != nil {
  		return nn, b.err
  	}
  	n := copy(b.buf[b.n:], p)
  	b.n += n
  	nn += n
  	return nn, nil
  }
  
  // WriteByte writes a single byte.
  func (b *Writer) WriteByte(c byte) error {
  	if b.err != nil {
  		return b.err
  	}
  	if b.Available() <= 0 && b.Flush() != nil {
  		return b.err
  	}
  	b.buf[b.n] = c
  	b.n++
  	return nil
  }
  
  // WriteRune writes a single Unicode code point, returning
  // the number of bytes written and any error.
  func (b *Writer) WriteRune(r rune) (size int, err error) {
  	if r < utf8.RuneSelf {
  		err = b.WriteByte(byte(r))
  		if err != nil {
  			return 0, err
  		}
  		return 1, nil
  	}
  	if b.err != nil {
  		return 0, b.err
  	}
  	n := b.Available()
  	if n < utf8.UTFMax {
  		if b.Flush(); b.err != nil {
  			return 0, b.err
  		}
  		n = b.Available()
  		if n < utf8.UTFMax {
  			// Can only happen if buffer is silly small.
  			return b.WriteString(string(r))
  		}
  	}
  	size = utf8.EncodeRune(b.buf[b.n:], r)
  	b.n += size
  	return size, nil
  }
  
  // WriteString writes a string.
  // It returns the number of bytes written.
  // If the count is less than len(s), it also returns an error explaining
  // why the write is short.
  func (b *Writer) WriteString(s string) (int, error) {
  	nn := 0
  	for len(s) > b.Available() && b.err == nil {
  		n := copy(b.buf[b.n:], s)
  		b.n += n
  		nn += n
  		s = s[n:]
  		b.Flush()
  	}
  	if b.err != nil {
  		return nn, b.err
  	}
  	n := copy(b.buf[b.n:], s)
  	b.n += n
  	nn += n
  	return nn, nil
  }
  
  // ReadFrom implements io.ReaderFrom.
  func (b *Writer) ReadFrom(r io.Reader) (n int64, err error) {
  	if b.Buffered() == 0 {
  		if w, ok := b.wr.(io.ReaderFrom); ok {
  			return w.ReadFrom(r)
  		}
  	}
  	var m int
  	for {
  		if b.Available() == 0 {
  			if err1 := b.Flush(); err1 != nil {
  				return n, err1
  			}
  		}
  		nr := 0
  		for nr < maxConsecutiveEmptyReads {
  			m, err = r.Read(b.buf[b.n:])
  			if m != 0 || err != nil {
  				break
  			}
  			nr++
  		}
  		if nr == maxConsecutiveEmptyReads {
  			return n, io.ErrNoProgress
  		}
  		b.n += m
  		n += int64(m)
  		if err != nil {
  			break
  		}
  	}
  	if err == io.EOF {
  		// If we filled the buffer exactly, flush preemptively.
  		if b.Available() == 0 {
  			err = b.Flush()
  		} else {
  			err = nil
  		}
  	}
  	return n, err
  }
  
  // buffered input and output
  
  // ReadWriter stores pointers to a Reader and a Writer.
  // It implements io.ReadWriter.
  type ReadWriter struct {
  	*Reader
  	*Writer
  }
  
  // NewReadWriter allocates a new ReadWriter that dispatches to r and w.
  func NewReadWriter(r *Reader, w *Writer) *ReadWriter {
  	return &ReadWriter{r, w}
  }
  

View as plain text