CStringLocaleHelper : 一个用于UTF8和UNICODE字符串转换的辅助类

PomeloWu原创©，转载请注明出处

微软抛弃MBCS看来是必然的选择。VS2005默认创建的工程都是UNICODE的。但无论是std::string/std::wstring还是CString，都没有提供不同字符集间转换的包装。

写CStringLocaleHelper本意是辅助Linux和Windows通讯时UTF-8和UNICODE的转换，实际上对它稍加改变，就能面向更多的code page，让人从WINDOWS程序中无趣的字符集转换之中解放出来。

CStringLocaleHelper其实就是对调用WideCharToMultiByte/MultiByteToWideChar的封装，可以返回指定字符集（目前是UTF-8或者UNICODE）的字符串或者std::string和std::wstring。此外，重载了针对LPCTSTR、std::string和std::wstring的operator ()，方便类型强制转化；重载了operator =以便增加CStringLocaleHelper的灵活性。

 以下是具体代码：

   
   

    
    /* -------------------------------------------------------------
	@filename: 	StringLocaleHelper.h
	@author:	Pomelo Wu
	@purpose:	A helper string class, handling string 
				conversions between UTF-8 and UNICODE
	@revision history:
	-- Date			Ver.	Modifier	Description--
	2006/03/15		0.1		Pomelo		created
	2006/03/29		0.2		Pomelo		return type modified
	2006/04/03		0.3		Pomelo		cast operator added
// -----------------------------------------------------------*/

#ifndef __STRINGLOCALEHELPER_H__
#define __STRINGLOCALEHELPER_H__

using namespace std;

class CStringLocaleHelper
{
public:
	CStringLocaleHelper();
	CStringLocaleHelper(string str);
	CStringLocaleHelper(wstring str);
	~CStringLocaleHelper(void);

public:
	const char * GetLocaleBuffer() const { return m_strLocale.c_str(); }
	const wchar_t * GetUnicodeBuffer() const { return m_strUnicode.c_str(); }

	const string&  GetLocaleString() const { return m_strLocale; }
	const wstring& GetUnicodeString() const { return m_strUnicode; }

	operator LPCTSTR () const;
	operator string () const;
	operator wstring () const;

	CStringLocaleHelper & operator = (string strSrc);
	CStringLocaleHelper & operator = (wstring strSrc);

protected:
	string  m_strLocale;
	wstring m_strUnicode;
};

#endif // __STRINGLOCALEHELPER_H__

/* -------------------------------------------------------------
	@filename: 	StringLocaleHelper.cpp
	@author:	Pomelo Wu
	@purpose:	The implementation of CStringLocaleHelper
	@revision history:
	-- Date			Ver.	Modifier	Description--
	2006/03/15		0.1		Pomelo		created
	2006/03/29		0.2		Pomelo		return type modified
	2006/04/03		0.3		Pomelo		cast operator added
// -----------------------------------------------------------*/

// ----------------------- Using strings -----------------------
#include <string>
// --------------------- Unicode functions ---------------------
#include <Windows.h>

#include "StringLocaleHelper.h"

//
//
// native functions
int Unicode2Locale(const wchar_t *pSrc, char *pDest = NULL, int nCount = 0)
{
	if (pSrc == NULL)	// invalid parameter
		return -1;

	if (pDest == NULL)	// queries the count
		return WideCharToMultiByte(CP_UTF8, 0, pSrc, -1, NULL, 0, NULL, NULL);

	memset(pDest, 0, sizeof(char) * (nCount + 1));
	return WideCharToMultiByte(CP_UTF8, 0, pSrc, -1, pDest, nCount, NULL, NULL);
}

int Locale2Unicode(const char *pSrc, wchar_t *pDest = NULL, int nCount = 0)
{
	if (pSrc == NULL)	// invalid parameter
		return -1;

	if (pDest == NULL)	// queries the count
		return MultiByteToWideChar(CP_UTF8, NULL, pSrc, -1, NULL, 0);

	memset(pDest, 0, sizeof(wchar_t) * (nCount + 1));
	return MultiByteToWideChar(CP_UTF8, NULL, pSrc, -1, pDest, nCount + 1);
}

//
//
// constructors / destructors
CStringLocaleHelper::CStringLocaleHelper()
{
	m_strLocale.clear();
	m_strUnicode.clear();
}

CStringLocaleHelper::CStringLocaleHelper(string str)
{
	m_strLocale = str;
	int nCount = Locale2Unicode(m_strLocale.c_str());

	wchar_t *pwcText = new wchar_t [nCount + 1];
	if (pwcText)
	{
		nCount = Locale2Unicode(str.c_str(), pwcText, nCount);
		m_strUnicode = pwcText;

		delete [] pwcText;
	}
}

CStringLocaleHelper::CStringLocaleHelper(wstring str)
{
	m_strUnicode = str;
	int nCount = Unicode2Locale(str.c_str());

	char *pBufferAnsi = new char[nCount + 1];
	if (pBufferAnsi)
	{
		nCount = Unicode2Locale(str.c_str(), pBufferAnsi, nCount);
		m_strLocale = pBufferAnsi;

		delete [] pBufferAnsi;
	}
}

CStringLocaleHelper::~CStringLocaleHelper(void)
{

}

//
//
// operators
CStringLocaleHelper::operator LPCTSTR() const
{
#ifdef _UNICODE 
	return m_strUnicode.c_str();
#else
	return m_strLocale.c_str();
#endif
}

CStringLocaleHelper::operator string () const
{
	return m_strLocale;
}

CStringLocaleHelper::operator wstring () const
{
	return m_strUnicode;
}

CStringLocaleHelper & CStringLocaleHelper::operator =(std::string strSrc)
{
	m_strLocale = strSrc;
	int nCount = Locale2Unicode(strSrc.c_str());

	wchar_t *pwcText = new wchar_t [nCount + 1];
	if (pwcText)
	{
		nCount = Locale2Unicode(strSrc.c_str(), pwcText, nCount);
		m_strUnicode = pwcText;

		delete [] pwcText;
	}

	return *this;
}

CStringLocaleHelper & CStringLocaleHelper::operator =(std::wstring strSrc)
{
	m_strUnicode = strSrc;
	int nCount = Unicode2Locale(strSrc.c_str());

	char *pcText = new char [nCount + 1];
	if (pcText)
	{
		nCount = Unicode2Locale(strSrc.c_str(), pcText, nCount);
		m_strLocale = pcText;

		delete [] pcText;
	}

	return *this;
}