binary.go

Documentation: encoding/binary

  // 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 binary implements simple translation between numbers and byte
  // sequences and encoding and decoding of varints.
  //
  // Numbers are translated by reading and writing fixed-size values.
  // A fixed-size value is either a fixed-size arithmetic
  // type (bool, int8, uint8, int16, float32, complex64, ...)
  // or an array or struct containing only fixed-size values.
  //
  // The varint functions encode and decode single integer values using
  // a variable-length encoding; smaller values require fewer bytes.
  // For a specification, see
  // https://developers.google.com/protocol-buffers/docs/encoding.
  //
  // This package favors simplicity over efficiency. Clients that require
  // high-performance serialization, especially for large data structures,
  // should look at more advanced solutions such as the encoding/gob
  // package or protocol buffers.
  package binary
  
  import (
  	"errors"
  	"io"
  	"math"
  	"reflect"
  )
  
  // A ByteOrder specifies how to convert byte sequences into
  // 16-, 32-, or 64-bit unsigned integers.
  type ByteOrder interface {
  	Uint16([]byte) uint16
  	Uint32([]byte) uint32
  	Uint64([]byte) uint64
  	PutUint16([]byte, uint16)
  	PutUint32([]byte, uint32)
  	PutUint64([]byte, uint64)
  	String() string
  }
  
  // LittleEndian is the little-endian implementation of ByteOrder.
  var LittleEndian littleEndian
  
  // BigEndian is the big-endian implementation of ByteOrder.
  var BigEndian bigEndian
  
  type littleEndian struct{}
  
  func (littleEndian) Uint16(b []byte) uint16 {
  	_ = b[1] // bounds check hint to compiler; see golang.org/issue/14808
  	return uint16(b[0]) | uint16(b[1])<<8
  }
  
  func (littleEndian) PutUint16(b []byte, v uint16) {
  	_ = b[1] // early bounds check to guarantee safety of writes below
  	b[0] = byte(v)
  	b[1] = byte(v >> 8)
  }
  
  func (littleEndian) Uint32(b []byte) uint32 {
  	_ = b[3] // bounds check hint to compiler; see golang.org/issue/14808
  	return uint32(b[0]) | uint32(b[1])<<8 | uint32(b[2])<<16 | uint32(b[3])<<24
  }
  
  func (littleEndian) PutUint32(b []byte, v uint32) {
  	_ = b[3] // early bounds check to guarantee safety of writes below
  	b[0] = byte(v)
  	b[1] = byte(v >> 8)
  	b[2] = byte(v >> 16)
  	b[3] = byte(v >> 24)
  }
  
  func (littleEndian) Uint64(b []byte) uint64 {
  	_ = b[7] // bounds check hint to compiler; see golang.org/issue/14808
  	return uint64(b[0]) | uint64(b[1])<<8 | uint64(b[2])<<16 | uint64(b[3])<<24 |
  		uint64(b[4])<<32 | uint64(b[5])<<40 | uint64(b[6])<<48 | uint64(b[7])<<56
  }
  
  func (littleEndian) PutUint64(b []byte, v uint64) {
  	_ = b[7] // early bounds check to guarantee safety of writes below
  	b[0] = byte(v)
  	b[1] = byte(v >> 8)
  	b[2] = byte(v >> 16)
  	b[3] = byte(v >> 24)
  	b[4] = byte(v >> 32)
  	b[5] = byte(v >> 40)
  	b[6] = byte(v >> 48)
  	b[7] = byte(v >> 56)
  }
  
  func (littleEndian) String() string { return "LittleEndian" }
  
  func (littleEndian) GoString() string { return "binary.LittleEndian" }
  
  type bigEndian struct{}
  
  func (bigEndian) Uint16(b []byte) uint16 {
  	_ = b[1] // bounds check hint to compiler; see golang.org/issue/14808
  	return uint16(b[1]) | uint16(b[0])<<8
  }
  
  func (bigEndian) PutUint16(b []byte, v uint16) {
  	_ = b[1] // early bounds check to guarantee safety of writes below
  	b[0] = byte(v >> 8)
  	b[1] = byte(v)
  }
  
  func (bigEndian) Uint32(b []byte) uint32 {
  	_ = b[3] // bounds check hint to compiler; see golang.org/issue/14808
  	return uint32(b[3]) | uint32(b[2])<<8 | uint32(b[1])<<16 | uint32(b[0])<<24
  }
  
  func (bigEndian) PutUint32(b []byte, v uint32) {
  	_ = b[3] // early bounds check to guarantee safety of writes below
  	b[0] = byte(v >> 24)
  	b[1] = byte(v >> 16)
  	b[2] = byte(v >> 8)
  	b[3] = byte(v)
  }
  
  func (bigEndian) Uint64(b []byte) uint64 {
  	_ = b[7] // bounds check hint to compiler; see golang.org/issue/14808
  	return uint64(b[7]) | uint64(b[6])<<8 | uint64(b[5])<<16 | uint64(b[4])<<24 |
  		uint64(b[3])<<32 | uint64(b[2])<<40 | uint64(b[1])<<48 | uint64(b[0])<<56
  }
  
  func (bigEndian) PutUint64(b []byte, v uint64) {
  	_ = b[7] // early bounds check to guarantee safety of writes below
  	b[0] = byte(v >> 56)
  	b[1] = byte(v >> 48)
  	b[2] = byte(v >> 40)
  	b[3] = byte(v >> 32)
  	b[4] = byte(v >> 24)
  	b[5] = byte(v >> 16)
  	b[6] = byte(v >> 8)
  	b[7] = byte(v)
  }
  
  func (bigEndian) String() string { return "BigEndian" }
  
  func (bigEndian) GoString() string { return "binary.BigEndian" }
  
  // Read reads structured binary data from r into data.
  // Data must be a pointer to a fixed-size value or a slice
  // of fixed-size values.
  // Bytes read from r are decoded using the specified byte order
  // and written to successive fields of the data.
  // When decoding boolean values, a zero byte is decoded as false, and
  // any other non-zero byte is decoded as true.
  // When reading into structs, the field data for fields with
  // blank (_) field names is skipped; i.e., blank field names
  // may be used for padding.
  // When reading into a struct, all non-blank fields must be exported
  // or Read may panic.
  //
  // The error is EOF only if no bytes were read.
  // If an EOF happens after reading some but not all the bytes,
  // Read returns ErrUnexpectedEOF.
  func Read(r io.Reader, order ByteOrder, data interface{}) error {
  	// Fast path for basic types and slices.
  	if n := intDataSize(data); n != 0 {
  		var b [8]byte
  		var bs []byte
  		if n > len(b) {
  			bs = make([]byte, n)
  		} else {
  			bs = b[:n]
  		}
  		if _, err := io.ReadFull(r, bs); err != nil {
  			return err
  		}
  		switch data := data.(type) {
  		case *bool:
  			*data = b[0] != 0
  		case *int8:
  			*data = int8(b[0])
  		case *uint8:
  			*data = b[0]
  		case *int16:
  			*data = int16(order.Uint16(bs))
  		case *uint16:
  			*data = order.Uint16(bs)
  		case *int32:
  			*data = int32(order.Uint32(bs))
  		case *uint32:
  			*data = order.Uint32(bs)
  		case *int64:
  			*data = int64(order.Uint64(bs))
  		case *uint64:
  			*data = order.Uint64(bs)
  		case []bool:
  			for i, x := range bs { // Easier to loop over the input for 8-bit values.
  				data[i] = x != 0
  			}
  		case []int8:
  			for i, x := range bs {
  				data[i] = int8(x)
  			}
  		case []uint8:
  			copy(data, bs)
  		case []int16:
  			for i := range data {
  				data[i] = int16(order.Uint16(bs[2*i:]))
  			}
  		case []uint16:
  			for i := range data {
  				data[i] = order.Uint16(bs[2*i:])
  			}
  		case []int32:
  			for i := range data {
  				data[i] = int32(order.Uint32(bs[4*i:]))
  			}
  		case []uint32:
  			for i := range data {
  				data[i] = order.Uint32(bs[4*i:])
  			}
  		case []int64:
  			for i := range data {
  				data[i] = int64(order.Uint64(bs[8*i:]))
  			}
  		case []uint64:
  			for i := range data {
  				data[i] = order.Uint64(bs[8*i:])
  			}
  		}
  		return nil
  	}
  
  	// Fallback to reflect-based decoding.
  	v := reflect.ValueOf(data)
  	size := -1
  	switch v.Kind() {
  	case reflect.Ptr:
  		v = v.Elem()
  		size = dataSize(v)
  	case reflect.Slice:
  		size = dataSize(v)
  	}
  	if size < 0 {
  		return errors.New("binary.Read: invalid type " + reflect.TypeOf(data).String())
  	}
  	d := &decoder{order: order, buf: make([]byte, size)}
  	if _, err := io.ReadFull(r, d.buf); err != nil {
  		return err
  	}
  	d.value(v)
  	return nil
  }
  
  // Write writes the binary representation of data into w.
  // Data must be a fixed-size value or a slice of fixed-size
  // values, or a pointer to such data.
  // Boolean values encode as one byte: 1 for true, and 0 for false.
  // Bytes written to w are encoded using the specified byte order
  // and read from successive fields of the data.
  // When writing structs, zero values are written for fields
  // with blank (_) field names.
  func Write(w io.Writer, order ByteOrder, data interface{}) error {
  	// Fast path for basic types and slices.
  	if n := intDataSize(data); n != 0 {
  		var b [8]byte
  		var bs []byte
  		if n > len(b) {
  			bs = make([]byte, n)
  		} else {
  			bs = b[:n]
  		}
  		switch v := data.(type) {
  		case *bool:
  			if *v {
  				b[0] = 1
  			} else {
  				b[0] = 0
  			}
  		case bool:
  			if v {
  				b[0] = 1
  			} else {
  				b[0] = 0
  			}
  		case []bool:
  			for i, x := range v {
  				if x {
  					bs[i] = 1
  				} else {
  					bs[i] = 0
  				}
  			}
  		case *int8:
  			b[0] = byte(*v)
  		case int8:
  			b[0] = byte(v)
  		case []int8:
  			for i, x := range v {
  				bs[i] = byte(x)
  			}
  		case *uint8:
  			b[0] = *v
  		case uint8:
  			b[0] = v
  		case []uint8:
  			bs = v
  		case *int16:
  			order.PutUint16(bs, uint16(*v))
  		case int16:
  			order.PutUint16(bs, uint16(v))
  		case []int16:
  			for i, x := range v {
  				order.PutUint16(bs[2*i:], uint16(x))
  			}
  		case *uint16:
  			order.PutUint16(bs, *v)
  		case uint16:
  			order.PutUint16(bs, v)
  		case []uint16:
  			for i, x := range v {
  				order.PutUint16(bs[2*i:], x)
  			}
  		case *int32:
  			order.PutUint32(bs, uint32(*v))
  		case int32:
  			order.PutUint32(bs, uint32(v))
  		case []int32:
  			for i, x := range v {
  				order.PutUint32(bs[4*i:], uint32(x))
  			}
  		case *uint32:
  			order.PutUint32(bs, *v)
  		case uint32:
  			order.PutUint32(bs, v)
  		case []uint32:
  			for i, x := range v {
  				order.PutUint32(bs[4*i:], x)
  			}
  		case *int64:
  			order.PutUint64(bs, uint64(*v))
  		case int64:
  			order.PutUint64(bs, uint64(v))
  		case []int64:
  			for i, x := range v {
  				order.PutUint64(bs[8*i:], uint64(x))
  			}
  		case *uint64:
  			order.PutUint64(bs, *v)
  		case uint64:
  			order.PutUint64(bs, v)
  		case []uint64:
  			for i, x := range v {
  				order.PutUint64(bs[8*i:], x)
  			}
  		}
  		_, err := w.Write(bs)
  		return err
  	}
  
  	// Fallback to reflect-based encoding.
  	v := reflect.Indirect(reflect.ValueOf(data))
  	size := dataSize(v)
  	if size < 0 {
  		return errors.New("binary.Write: invalid type " + reflect.TypeOf(data).String())
  	}
  	buf := make([]byte, size)
  	e := &encoder{order: order, buf: buf}
  	e.value(v)
  	_, err := w.Write(buf)
  	return err
  }
  
  // Size returns how many bytes Write would generate to encode the value v, which
  // must be a fixed-size value or a slice of fixed-size values, or a pointer to such data.
  // If v is neither of these, Size returns -1.
  func Size(v interface{}) int {
  	return dataSize(reflect.Indirect(reflect.ValueOf(v)))
  }
  
  // dataSize returns the number of bytes the actual data represented by v occupies in memory.
  // For compound structures, it sums the sizes of the elements. Thus, for instance, for a slice
  // it returns the length of the slice times the element size and does not count the memory
  // occupied by the header. If the type of v is not acceptable, dataSize returns -1.
  func dataSize(v reflect.Value) int {
  	if v.Kind() == reflect.Slice {
  		if s := sizeof(v.Type().Elem()); s >= 0 {
  			return s * v.Len()
  		}
  		return -1
  	}
  	return sizeof(v.Type())
  }
  
  // sizeof returns the size >= 0 of variables for the given type or -1 if the type is not acceptable.
  func sizeof(t reflect.Type) int {
  	switch t.Kind() {
  	case reflect.Array:
  		if s := sizeof(t.Elem()); s >= 0 {
  			return s * t.Len()
  		}
  
  	case reflect.Struct:
  		sum := 0
  		for i, n := 0, t.NumField(); i < n; i++ {
  			s := sizeof(t.Field(i).Type)
  			if s < 0 {
  				return -1
  			}
  			sum += s
  		}
  		return sum
  
  	case reflect.Bool,
  		reflect.Uint8, reflect.Uint16, reflect.Uint32, reflect.Uint64,
  		reflect.Int8, reflect.Int16, reflect.Int32, reflect.Int64,
  		reflect.Float32, reflect.Float64, reflect.Complex64, reflect.Complex128:
  		return int(t.Size())
  	}
  
  	return -1
  }
  
  type coder struct {
  	order ByteOrder
  	buf   []byte
  }
  
  type decoder coder
  type encoder coder
  
  func (d *decoder) bool() bool {
  	x := d.buf[0]
  	d.buf = d.buf[1:]
  	return x != 0
  }
  
  func (e *encoder) bool(x bool) {
  	if x {
  		e.buf[0] = 1
  	} else {
  		e.buf[0] = 0
  	}
  	e.buf = e.buf[1:]
  }
  
  func (d *decoder) uint8() uint8 {
  	x := d.buf[0]
  	d.buf = d.buf[1:]
  	return x
  }
  
  func (e *encoder) uint8(x uint8) {
  	e.buf[0] = x
  	e.buf = e.buf[1:]
  }
  
  func (d *decoder) uint16() uint16 {
  	x := d.order.Uint16(d.buf[0:2])
  	d.buf = d.buf[2:]
  	return x
  }
  
  func (e *encoder) uint16(x uint16) {
  	e.order.PutUint16(e.buf[0:2], x)
  	e.buf = e.buf[2:]
  }
  
  func (d *decoder) uint32() uint32 {
  	x := d.order.Uint32(d.buf[0:4])
  	d.buf = d.buf[4:]
  	return x
  }
  
  func (e *encoder) uint32(x uint32) {
  	e.order.PutUint32(e.buf[0:4], x)
  	e.buf = e.buf[4:]
  }
  
  func (d *decoder) uint64() uint64 {
  	x := d.order.Uint64(d.buf[0:8])
  	d.buf = d.buf[8:]
  	return x
  }
  
  func (e *encoder) uint64(x uint64) {
  	e.order.PutUint64(e.buf[0:8], x)
  	e.buf = e.buf[8:]
  }
  
  func (d *decoder) int8() int8 { return int8(d.uint8()) }
  
  func (e *encoder) int8(x int8) { e.uint8(uint8(x)) }
  
  func (d *decoder) int16() int16 { return int16(d.uint16()) }
  
  func (e *encoder) int16(x int16) { e.uint16(uint16(x)) }
  
  func (d *decoder) int32() int32 { return int32(d.uint32()) }
  
  func (e *encoder) int32(x int32) { e.uint32(uint32(x)) }
  
  func (d *decoder) int64() int64 { return int64(d.uint64()) }
  
  func (e *encoder) int64(x int64) { e.uint64(uint64(x)) }
  
  func (d *decoder) value(v reflect.Value) {
  	switch v.Kind() {
  	case reflect.Array:
  		l := v.Len()
  		for i := 0; i < l; i++ {
  			d.value(v.Index(i))
  		}
  
  	case reflect.Struct:
  		t := v.Type()
  		l := v.NumField()
  		for i := 0; i < l; i++ {
  			// Note: Calling v.CanSet() below is an optimization.
  			// It would be sufficient to check the field name,
  			// but creating the StructField info for each field is
  			// costly (run "go test -bench=ReadStruct" and compare
  			// results when making changes to this code).
  			if v := v.Field(i); v.CanSet() || t.Field(i).Name != "_" {
  				d.value(v)
  			} else {
  				d.skip(v)
  			}
  		}
  
  	case reflect.Slice:
  		l := v.Len()
  		for i := 0; i < l; i++ {
  			d.value(v.Index(i))
  		}
  
  	case reflect.Bool:
  		v.SetBool(d.bool())
  
  	case reflect.Int8:
  		v.SetInt(int64(d.int8()))
  	case reflect.Int16:
  		v.SetInt(int64(d.int16()))
  	case reflect.Int32:
  		v.SetInt(int64(d.int32()))
  	case reflect.Int64:
  		v.SetInt(d.int64())
  
  	case reflect.Uint8:
  		v.SetUint(uint64(d.uint8()))
  	case reflect.Uint16:
  		v.SetUint(uint64(d.uint16()))
  	case reflect.Uint32:
  		v.SetUint(uint64(d.uint32()))
  	case reflect.Uint64:
  		v.SetUint(d.uint64())
  
  	case reflect.Float32:
  		v.SetFloat(float64(math.Float32frombits(d.uint32())))
  	case reflect.Float64:
  		v.SetFloat(math.Float64frombits(d.uint64()))
  
  	case reflect.Complex64:
  		v.SetComplex(complex(
  			float64(math.Float32frombits(d.uint32())),
  			float64(math.Float32frombits(d.uint32())),
  		))
  	case reflect.Complex128:
  		v.SetComplex(complex(
  			math.Float64frombits(d.uint64()),
  			math.Float64frombits(d.uint64()),
  		))
  	}
  }
  
  func (e *encoder) value(v reflect.Value) {
  	switch v.Kind() {
  	case reflect.Array:
  		l := v.Len()
  		for i := 0; i < l; i++ {
  			e.value(v.Index(i))
  		}
  
  	case reflect.Struct:
  		t := v.Type()
  		l := v.NumField()
  		for i := 0; i < l; i++ {
  			// see comment for corresponding code in decoder.value()
  			if v := v.Field(i); v.CanSet() || t.Field(i).Name != "_" {
  				e.value(v)
  			} else {
  				e.skip(v)
  			}
  		}
  
  	case reflect.Slice:
  		l := v.Len()
  		for i := 0; i < l; i++ {
  			e.value(v.Index(i))
  		}
  
  	case reflect.Bool:
  		e.bool(v.Bool())
  
  	case reflect.Int, reflect.Int8, reflect.Int16, reflect.Int32, reflect.Int64:
  		switch v.Type().Kind() {
  		case reflect.Int8:
  			e.int8(int8(v.Int()))
  		case reflect.Int16:
  			e.int16(int16(v.Int()))
  		case reflect.Int32:
  			e.int32(int32(v.Int()))
  		case reflect.Int64:
  			e.int64(v.Int())
  		}
  
  	case reflect.Uint, reflect.Uint8, reflect.Uint16, reflect.Uint32, reflect.Uint64, reflect.Uintptr:
  		switch v.Type().Kind() {
  		case reflect.Uint8:
  			e.uint8(uint8(v.Uint()))
  		case reflect.Uint16:
  			e.uint16(uint16(v.Uint()))
  		case reflect.Uint32:
  			e.uint32(uint32(v.Uint()))
  		case reflect.Uint64:
  			e.uint64(v.Uint())
  		}
  
  	case reflect.Float32, reflect.Float64:
  		switch v.Type().Kind() {
  		case reflect.Float32:
  			e.uint32(math.Float32bits(float32(v.Float())))
  		case reflect.Float64:
  			e.uint64(math.Float64bits(v.Float()))
  		}
  
  	case reflect.Complex64, reflect.Complex128:
  		switch v.Type().Kind() {
  		case reflect.Complex64:
  			x := v.Complex()
  			e.uint32(math.Float32bits(float32(real(x))))
  			e.uint32(math.Float32bits(float32(imag(x))))
  		case reflect.Complex128:
  			x := v.Complex()
  			e.uint64(math.Float64bits(real(x)))
  			e.uint64(math.Float64bits(imag(x)))
  		}
  	}
  }
  
  func (d *decoder) skip(v reflect.Value) {
  	d.buf = d.buf[dataSize(v):]
  }
  
  func (e *encoder) skip(v reflect.Value) {
  	n := dataSize(v)
  	for i := range e.buf[0:n] {
  		e.buf[i] = 0
  	}
  	e.buf = e.buf[n:]
  }
  
  // intDataSize returns the size of the data required to represent the data when encoded.
  // It returns zero if the type cannot be implemented by the fast path in Read or Write.
  func intDataSize(data interface{}) int {
  	switch data := data.(type) {
  	case bool, int8, uint8, *bool, *int8, *uint8:
  		return 1
  	case []int8:
  		return len(data)
  	case []uint8:
  		return len(data)
  	case int16, uint16, *int16, *uint16:
  		return 2
  	case []int16:
  		return 2 * len(data)
  	case []uint16:
  		return 2 * len(data)
  	case int32, uint32, *int32, *uint32:
  		return 4
  	case []int32:
  		return 4 * len(data)
  	case []uint32:
  		return 4 * len(data)
  	case int64, uint64, *int64, *uint64:
  		return 8
  	case []int64:
  		return 8 * len(data)
  	case []uint64:
  		return 8 * len(data)
  	}
  	return 0
  }
View as plain text