c++多字节与宽字节字符串转换(windows平台)

本文转自：http://www.asarea.cn/showdiary.jsp?id=193

 

大家都知道，在标准c++中有string和wstring，前者为多字节ANSI字串，后者为宽字节wide字串(UTF-16)。
在一些应用中常常需要在二者之间以及与UTF-8之间进行字串转换，这里提供一个自己写的类，用于之间的转换。
原理是利用windows的api：WideCharToMultiByte、MultiByteToWideChar。
这两个api的第一个参数就是代码页，其中CP_ACP代表ANSI CODE PAGE，而CP_UTF8自然代表UTF-8。
注：由于用到了windows api，所以只适用windows平台。

 

/*** Author: ATHER Shu 2009.3.20* StringUtil类:用于ANSI、UTF-16、UTF-8之间的字串转换* 功能：* 1.UTF-16转ANSI ws2s* 2.ANSI转UTF-16 s2ws* 3.UTF-16转UTF-8 ws2utf8* 4.UTF-8转UTF-16 utf82ws* http://www.asarea.cn* ATHER Shu(AS)*/#include <string>#include <Windows.h>using namespace std;class StringUtil{public: static string ws2s(const wstring& ws) { const wchar_t* _Source = ws.c_str(); size_t _Dsize = WideCharToMultiByte(CP_ACP, 0, _Source, -1, NULL, 0, NULL, NULL); char *_Dest = new char[_Dsize]; WideCharToMultiByte(CP_ACP, 0, _Source, -1, _Dest, _Dsize, NULL, NULL); string result = _Dest; delete []_Dest; return result; } static wstring s2ws(const string& s) { const char* _Source = s.c_str(); size_t _Dsize = MultiByteToWideChar(CP_ACP, 0, _Source, -1, NULL, 0); wchar_t *_Dest = new wchar_t[_Dsize]; MultiByteToWideChar(CP_ACP, 0, _Source, -1, _Dest, _Dsize); wstring result = _Dest; delete []_Dest; return result; } static string ws2utf8(const wstring& ws) { const wchar_t* _Source = ws.c_str(); size_t _Dsize = WideCharToMultiByte(CP_UTF8, 0, _Source, -1, NULL, 0, NULL, NULL); char *_Dest = new char[_Dsize]; WideCharToMultiByte(CP_UTF8, 0, _Source, -1, _Dest, _Dsize, NULL, NULL); string result = _Dest; delete []_Dest; return result; } static wstring utf82ws(const string& s) { const char* _Source = s.c_str(); size_t _Dsize = MultiByteToWideChar(CP_UTF8, 0, _Source, -1, NULL, 0); wchar_t *_Dest = new wchar_t[_Dsize]; MultiByteToWideChar(CP_UTF8, 0, _Source, -1, _Dest, _Dsize); wstring result = _Dest; delete []_Dest; return result; }};