历史溯源
由于历史原因,我们目前看到的大部分的网络协议都是基于ASCII码这种纯文本方式,也就是基于字符串的命令行方式,比如HTTP、FTP、POP3、SMTP、Telnet等。早期操作系统UNIX(或DOS),用户操作界面就是控制台,控制台的输入输出方式就决定了用户只能通过敲击键盘的方式将协议命令输入到网络,这也就导致了回车换行"\r\n"会作为一次命令结束的标识。
比如HTTP协议,与主机建立连接后,输入"GET / HTTP/1.1\r\n"即可获取网站的主页。
比如email协议,早期的电子邮件协议只支持ASCII码这种纯文本传输,但随着全世界人民对物质文化生活的不断向往,这种落后的传输方式,已经无法满足世界人民对美好生活的追求,比如图像、视频、音频、Office文件如何在邮件中展现?不同国家(非英语国家)字符集该如何传输和展现?
换句话说,就是这种非ASCII的二进制富文本,该如何传输和呈现?
MIME的诞生
此时MIME标准诞生了,MIME的出现更多的是一种向下兼容的无奈,而不是革命。通过对二进制数据或非ASCII码数据进行base64或quoted-printable编码,来实现纯ASCII码的传输。显然这种方式会让你的邮件体变大,传输效率下降。尤其附件很多时,通过MIME的boundary来解析邮件的附件也是一笔额外的负担。
同时MIME的标准也被HTTP协议所采用,我们可以通过content-type指定传输的内容是什么类型,通过MIME的boundary来对Form-Data数据进行扩展,让我们Post数据时也能够在“表格”数据中插入文件,从而达到上传文件的效果。
显然这种方式不如二进制简洁,但却非常的直观,所见即所得,一眼就能看明白。但就传输效率上不如二进制方式。
又比如websocket协议虽然建立会话时采用的是HTTP协议,但后续的数据帧格式却是一个二进制格式。如下:
在这种格式下,为了表示每帧数据长度,就一定会有一个“数据长度”项,比如上面的payload len,当该值小于126时,直接表示数据区(payload data)长度;为126时用后面的2个字节表示数据区长度,为127时用后面的8个字节表示数据区长度。此时就涉及到了网络字节序和主机字节序的转换,如果数据区是一个二进制内容的话,我们就很难使用string的操作方式将整个数据报文拼接起来(可以用memcpy来拼接)。当然,我们这篇文章不是对websocket协议的讲解,而是通过该协议的数据区引出二进制数据流封装的必要性。如果是文本协议,各种开发语言对string的封装已经足够强大,已经没有封装的必要。除非你想重新改造字符串操作来提升效率或其它目的,比如我的前一篇文章:
为何写服务器程序需要自己管理内存,从改造std::string字符串操作说起。。。
话不多说,下面是一个简单的数据流的封装类CDataStream,非常简单。
.h头文件
#include <windows.h>
// 数据流
class CDataStream
{
public:
CDataStream(BOOL bNetworkOrder = FALSE);
virtual ~CDataStream();
// 关联一块stream
void Attach(const BYTE* pStream, int iStreamSize){
m_pStream = (BYTE*)pStream;
m_iStreamSize = iStreamSize;
m_iCurrPos = 0;
}
// 解除关联
void Detach(){
m_pStream = NULL;
m_iStreamSize = 0;
m_iCurrPos = 0;
}
void Reset(){
m_iCurrPos = 0;
}
// 获取流数据
const BYTE* GetStreamData(){
return m_pStream;
}
int GetStreamSize(){
return m_iCurrPos;
}
// 在当前位置上移动iDistance距离
int Offset(int iDistance);
// 移动到新位置
int MoveTo(int iNewPos);
void MoveToBegin(){
m_iCurrPos = 0;
}
void MoveToEnd(){
m_iCurrPos = m_iStreamSize;
}
// 读写字节
void WriteByte(BYTE byValue);
BYTE ReadByte();
// 读写WORD
void WriteWord(WORD wValue);
WORD ReadWord();
// 读写DWORD
void WriteDWord(DWORD dwValue);
DWORD ReadDWord();
// 读写int64
void WriteInt64(__int64 i64Value);
__int64 ReadInt64();
// 读写Float
void WriteFloat(float fValue);
float ReadFloat();
// 读写double
void WriteDouble(double dValue);
double ReadDouble();
// 读写数据流
void WriteData(unsigned char* pData, int iDataLen);
BYTE* ReadData(int iDataLen);
// 读写字符串
void WriteString(const char* pszValue);
const char* ReadString();
// =============运算符重载=============
CDataStream& operator<<(BYTE byValue) { WriteByte(byValue); return *this; }
CDataStream& operator<<(WORD wValue) { WriteWord(wValue); return *this; }
CDataStream& operator<<(DWORD dwValue) { WriteDWord(dwValue); return *this; }
CDataStream& operator<<(__int64 i64Value) { WriteInt64(i64Value); return *this; }
CDataStream& operator<<(float fValue) { WriteFloat(fValue); return *this; }
CDataStream& operator<<(double dValue) { WriteDouble(dValue); return *this; }
CDataStream& operator<<(const char* pszValue) { WriteString(pszValue); return *this; }
CDataStream& operator>>(BYTE& byValue) { byValue = ReadByte(); return *this; }
CDataStream& operator>>(WORD& wValue) { wValue = ReadWord(); return *this; }
CDataStream& operator>>(DWORD& dwValue) { dwValue = ReadDWord(); return *this; }
CDataStream& operator>>(__int64& i64Value) { i64Value = ReadInt64(); return *this; }
CDataStream& operator>>(float& fValue) { fValue = ReadFloat(); return *this; }
CDataStream& operator>>(double& dValue) { dValue = ReadDouble(); return *this; }
CDataStream& operator>>(const char*& pszValue) { pszValue = ReadString(); return *this; }
public:
// WORD值反序
static WORD Swap(WORD wValue){
WORD wRet = 0;
((BYTE*)&wRet)[0] = ((BYTE*)&wValue)[1];
((BYTE*)&wRet)[1] = ((BYTE*)&wValue)[0];
return wRet;
}
// DWORD反序
static DWORD Swap(DWORD dwValue){
DWORD dwRet = 0;
((BYTE*)&dwRet)[0] = ((BYTE*)&dwValue)[3];
((BYTE*)&dwRet)[1] = ((BYTE*)&dwValue)[2];
((BYTE*)&dwRet)[2] = ((BYTE*)&dwValue)[1];
((BYTE*)&dwRet)[3] = ((BYTE*)&dwValue)[0];
return dwRet;
}
// i64(long long)反序
static __int64 Swap(__int64 i64Value){
__int64 i64Ret = 0;
((BYTE*)&i64Ret)[0] = ((BYTE*)&i64Value)[7];
((BYTE*)&i64Ret)[1] = ((BYTE*)&i64Value)[6];
((BYTE*)&i64Ret)[2] = ((BYTE*)&i64Value)[5];
((BYTE*)&i64Ret)[3] = ((BYTE*)&i64Value)[4];
((BYTE*)&i64Ret)[4] = ((BYTE*)&i64Value)[3];
((BYTE*)&i64Ret)[5] = ((BYTE*)&i64Value)[2];
((BYTE*)&i64Ret)[6] = ((BYTE*)&i64Value)[1];
((BYTE*)&i64Ret)[7] = ((BYTE*)&i64Value)[0];
return i64Ret;
}
// 下面的函数也是将64位长整形反序,但比较难理解,不如上面的函数简单、粗暴和直观
// 即使你现在能整明白,下次未必能“见字如面”
static __int64 Swap64(__int64 i64Value)
{
return i64Value >> 56|
(i64Value & 0x00ff000000000000) >> 40 |
(i64Value & 0x0000ff0000000000) >> 24 |
(i64Value & 0x000000ff00000000) >> 8 |
(i64Value & 0x00000000ff000000) << 8 |
(i64Value & 0x0000000000ff0000) << 24 |
(i64Value & 0x000000000000ff00) << 40 |
i64Value << 56;
}
// 浮点型按照IEEE745标准不存在网络字节序和机器字节序,这里只是给出实现方法
// float反序
static float Swap(float fValue){
float fRet = fValue;
Swap((BYTE*)&fRet, sizeof(float));
return fRet;
}
// double反序
static double Swap(double dValue){
double dRet = dValue;
Swap((BYTE*)&dRet, sizeof(double));
return dRet;
}
// 内存数据反序
static void Swap(BYTE* pData, int iDataLen);
// 内存反序后返回新内存
static BYTE* SwapClone(BYTE* pData, int iDataLen);
protected:
BOOL m_bNetworkOrder; // 数据流是否为网络字节序,缺省为FALSE
BYTE *m_pStream; // stream缓存
int m_iStreamSize; // 缓存大小
int m_iCurrPos; // 当前数据位置
};
.cpp实现文件
#include "DataStream.h"
#include <assert.h>
#include <stdlib.h>
// 将一块内存反序
void CDataStream::Swap(BYTE* pData, int iDataLen)
{
if(NULL == pData || iDataLen <= 0)
return;
for(int i = 0 ; i < iDataLen / 2; i++)
{
BYTE temp = pData[i];
pData[i] = pData[iDataLen - i - 1];
pData[iDataLen - i - 1] = temp;
}
}
// 将一块内存反序后返回新内存
BYTE* CDataStream::SwapClone(BYTE* pData, int iDataLen)
{
if(NULL == pData || iDataLen <= 0)
return NULL;
BYTE* pSwap = (BYTE*)malloc(iDataLen);
int j = 0;
for(int i = iDataLen-1; i >= 0; i--)
{
pSwap[j] = pData[i];
j++;
}
return pSwap;
}
//////////////////////////////////////////////////////////////////////
// Construction/Destruction
//////////////////////////////////////////////////////////////////////
CDataStream::CDataStream(BOOL bNetworkOrder)
{
m_bNetworkOrder = bNetworkOrder;
m_pStream = NULL;
m_iStreamSize = 0;
m_iCurrPos = 0;
}
CDataStream::~CDataStream()
{
m_pStream = NULL;
m_iStreamSize = 0;
m_iCurrPos = 0;
}
// 在当前位置上移动iDistance距离
int CDataStream::Offset(int iDistance)
{
int iNewPos = m_iCurrPos+iDistance;
if(iNewPos < 0)
m_iCurrPos = 0;
else if(iNewPos > m_iStreamSize)
m_iCurrPos = m_iStreamSize;
else
m_iCurrPos = iNewPos;
return m_iCurrPos;
}
// 移动到新位置
int CDataStream::MoveTo(int iNewPos)
{
if(iNewPos < 0)
m_iCurrPos = 0;
else if(iNewPos > m_iStreamSize)
m_iCurrPos = m_iStreamSize;
else
m_iCurrPos = iNewPos;
return m_iCurrPos;
}
// 读写字节
void CDataStream::WriteByte(BYTE byValue)
{
assert(m_iCurrPos+1 <= m_iStreamSize); // 越界断言
if(m_iCurrPos+1 > m_iStreamSize)
return;
*(m_pStream+m_iCurrPos) = byValue;
m_iCurrPos++;
}
BYTE CDataStream::ReadByte()
{
assert(m_iCurrPos+1 <= m_iStreamSize); // 越界断言
if(m_iCurrPos+1 > m_iStreamSize)
return 0;
BYTE byValue = *(m_pStream+m_iCurrPos);
m_iCurrPos++;
return byValue;
}
// 读写WORD
void CDataStream::WriteWord(WORD wValue)
{
assert(m_iCurrPos+2 <= m_iStreamSize); // 越界断言
if(m_iCurrPos+2 > m_iStreamSize)
return;
// 如果是网络字节流则反序
if(m_bNetworkOrder)
wValue = Swap(wValue);
*(WORD*)(m_pStream+m_iCurrPos) = wValue;
m_iCurrPos += 2;
}
WORD CDataStream::ReadWord()
{
assert(m_iCurrPos+2 <= m_iStreamSize); // 越界断言
if(m_iCurrPos+2 > m_iStreamSize)
return 0;
WORD wValue = *(WORD*)(m_pStream+m_iCurrPos);
m_iCurrPos += 2;
// 如果是网络字节流则反序
if(m_bNetworkOrder)
wValue = Swap(wValue);
return wValue;
}
// 读写DWORD
void CDataStream::WriteDWord(DWORD dwValue)
{
assert(m_iCurrPos+4 <= m_iStreamSize); // 越界断言
if(m_iCurrPos+4 > m_iStreamSize)
return;
// 如果是网络字节流则反序
if(m_bNetworkOrder)
dwValue = Swap(dwValue);
*(DWORD*)(m_pStream+m_iCurrPos) = dwValue;
m_iCurrPos += 4;
}
DWORD CDataStream::ReadDWord()
{
assert(m_iCurrPos+4 <= m_iStreamSize); // 越界断言
if(m_iCurrPos+4 > m_iStreamSize)
return 0;
DWORD dwValue = *(DWORD*)(m_pStream+m_iCurrPos);
m_iCurrPos += 4;
// 如果是网络字节流则反序
if(m_bNetworkOrder)
dwValue = Swap(dwValue);
return dwValue;
}
// 读写int64
void CDataStream::WriteInt64(__int64 i64Value)
{
assert(m_iCurrPos+8 <= m_iStreamSize); // 越界断言
if(m_iCurrPos+8 > m_iStreamSize)
return;
// 如果是网络字节流则反序
if(m_bNetworkOrder)
i64Value = Swap(i64Value);
*(__int64*)(m_pStream+m_iCurrPos) = i64Value;
m_iCurrPos += 8;
}
__int64 CDataStream::ReadInt64()
{
assert(m_iCurrPos+8 <= m_iStreamSize); // 越界断言
if(m_iCurrPos+8 > m_iStreamSize)
return 0;
__int64 i64Value = *(__int64*)(m_pStream+m_iCurrPos);
m_iCurrPos += 8;
// 如果是网络字节流则反序
if(m_bNetworkOrder)
i64Value = Swap(i64Value);
return i64Value;
}
// 读写float
void CDataStream::WriteFloat(float fValue)
{
int iFloatSize = sizeof(float);
assert(m_iCurrPos+iFloatSize <= m_iStreamSize);
if(m_iCurrPos+iFloatSize > m_iStreamSize)
return;
*(float*)(m_pStream+m_iCurrPos) = fValue;
m_iCurrPos += iFloatSize;
}
float CDataStream::ReadFloat()
{
int iFloatSize = sizeof(float);
assert(m_iCurrPos+iFloatSize <= m_iStreamSize);
if(m_iCurrPos+iFloatSize > m_iStreamSize)
return 0;
float fValue = *(float*)(m_pStream+m_iCurrPos);
m_iCurrPos += iFloatSize;
return fValue;
}
// 读写double
void CDataStream::WriteDouble(double dValue)
{
int iDoubleSize = sizeof(double);
assert(m_iCurrPos+iDoubleSize <= m_iStreamSize);
if(m_iCurrPos+iDoubleSize > m_iStreamSize)
return;
*(double*)(m_pStream+m_iCurrPos) = dValue;
m_iCurrPos += iDoubleSize;
}
double CDataStream::ReadDouble()
{
int iDoubleSize = sizeof(double);
assert(m_iCurrPos+iDoubleSize <= m_iStreamSize);
if(m_iCurrPos+iDoubleSize > m_iStreamSize)
return 0;
double dValue = *(double*)(m_pStream+m_iCurrPos);
m_iCurrPos += iDoubleSize;
return dValue;
}
// 读写数据流
void CDataStream::WriteData(unsigned char* pData, int iDataLen)
{
if(NULL == pData || iDataLen <= 0)
return;
assert(m_iCurrPos + iDataLen <= m_iStreamSize); // 越界断言
if(m_iCurrPos + iDataLen > m_iStreamSize)
return;
memcpy(m_pStream+m_iCurrPos, pData, iDataLen);
m_iCurrPos += iDataLen;
}
BYTE* CDataStream::ReadData(int iDataLen)
{
if(iDataLen <= 0 || m_iCurrPos >= m_iStreamSize)
return NULL;
assert(m_iCurrPos + iDataLen <= m_iStreamSize); // 越界断言
if(m_iCurrPos + iDataLen > m_iStreamSize)
return NULL;
BYTE* pData = m_pStream+m_iCurrPos;
m_iCurrPos += iDataLen;
return pData;
}
// 读写字符串
void CDataStream::WriteString(const char* pszValue)
{
if(NULL == pszValue)
return ;
int iStrLen = strlen(pszValue)+1; // 末尾0
assert(m_iCurrPos+iStrLen <= m_iStreamSize); // 越界断言
if(m_iCurrPos+iStrLen > m_iStreamSize)
return;
memcpy(m_pStream+m_iCurrPos, pszValue, iStrLen);
m_iCurrPos += iStrLen;
}
const char* CDataStream::ReadString()
{
if(m_iCurrPos >= m_iStreamSize)
return NULL;
int iCurrPos = m_iCurrPos;
char* psz = (char*)(m_pStream+m_iCurrPos); // 字符串位置
while(iCurrPos < m_iStreamSize)
{
if(!m_pStream[iCurrPos]) // 字符串最后一个字符为0
{
m_iCurrPos = iCurrPos;
break;
}
iCurrPos++;
}
// 判断是否合法
if(m_iCurrPos < m_iStreamSize)
{
m_iCurrPos++; // skip 0
return psz;
}
assert(FALSE); // 越界断言
return NULL;
}
测试代码
void TestDataStream()
{
// 1、测试数据流,写入数据
BYTE szBuff[1024] = {0};
CDataStream ds;
ds.Attach(szBuff, 1024);
ds.WriteByte(1);
ds.WriteWord(2);
ds.WriteDWord(1000);
ds.WriteInt64(5678);
ds.WriteData((BYTE*)"ASDF\0", 5);
ds.WriteFloat(1234567890.12f);
ds.WriteDouble(1234567890.123);
ds.WriteString("Hello word!");
// 读取数据流
ds.Reset(); // 指向流的头
BYTE byValue = ds.ReadByte();
WORD wValue = ds.ReadWord();
DWORD dwValue = ds.ReadDWord();
__int64 i64Value = ds.ReadInt64();
BYTE* pData = ds.ReadData(5);
float fValue = ds.ReadFloat();
double dValue = ds.ReadDouble();
const char* psz = ds.ReadString();
printf("CDataStream读写测试:\r\n");
printf("BYTE=%d, WORD=%d, DWORD = %d, INT64 = %I64u, FLOAT = %f, DOUBLE = %f, %s\r\n",
byValue, wValue, dwValue, i64Value, fValue, dValue, psz);
printf("pData = %s\r\n", (char*)pData);
// 2、测试数据流,重载运算符(<<,>>)的测试
// 测试运算符重载
CDataStream dds;
dds.Attach(szBuff, 1024);
BYTE byRet = 0;
WORD wRet = 0;
DWORD dwRet = 0;
__int64 i64Ret = 0;
float fRet = 0;
double dRet = 0;
char* pszRet;
dds << (BYTE)1 << (WORD)2 << (DWORD)3 << (__int64)100 << 30.1f << 128.12 << "Hello word!";
dds.Reset();
dds >> byRet >> wRet >> dwRet >> i64Ret >> fRet >> dRet >> pszRet;
printf("CDataStream测试,运算符重载:\r\n");
printf("by1 = %d, WORD = %d, DWORD = %d, INT64 = %I64u, FLOAT = %f, DOUBLE = %f, %s\r\n",
byRet, wRet, dwRet, i64Ret, fRet, dRet, pszRet);
}
输出结果