宽字符和窄字符之间的转换，以及对中文的处理问题总汇

  1 /*实现宽字节和窄字符转换以及中文文件的输入输出*/
  2 
  3 #include <iostream>
  4 #include <string>
  5 #include <fstream>
  6 #include <locale>
  7 #include <codecvt>
  8 #include <windows.h>
  9 #include <tchar.h>
 10 using namespace std;
 11 
 12 //-----------------------------------------------------------------------------
 13 //转换宽窄字符串时可以成功，并且支持中文转换
 14 const std::wstring s2ws(const std::string& s)
 15 {
 16     std::locale old_loc = std::locale::global(std::locale(""));
 17     const char* src_str = s.c_str();
 18     const size_t buffer_size = s.size() + 1;
 19     wchar_t* dst_wstr = new wchar_t[buffer_size];
 20     wmemset(dst_wstr, 0, buffer_size);
 21     mbstowcs(dst_wstr, src_str, buffer_size);
 22     std::wstring result = dst_wstr;
 23     delete []dst_wstr;
 24     std::locale::global(old_loc);
 25     return result;
 26 }
 27 const std::string ws2s(const std::wstring& ws)
 28 {
 29     std::locale old_loc = std::locale::global(std::locale(""));
 30     const wchar_t* src_wstr = ws.c_str();
 31     size_t buffer_size = ws.size() * 4 + 1;
 32     char* dst_str = new char[buffer_size];
 33     memset(dst_str, 0, buffer_size);
 34     wcstombs(dst_str ,src_wstr, buffer_size);
 35     std::string result = dst_str;
 36     delete []dst_str;
 37     std::locale::global(old_loc);
 38     return result;
 39 }
 40 
 41 //-----------------------------------------------------------------------------
 42 //转换宽窄字符串时可以成功，并且支持中文转换
 43 std::string WChar2Ansi(LPCWSTR pwszSrc)
 44 {
 45     int nLen = WideCharToMultiByte(CP_ACP, 0, pwszSrc, -1, NULL, 0, NULL, NULL);
 46 
 47     if (nLen<= 0) return std::string("");
 48 
 49     char* pszDst = new char[nLen];
 50     if (NULL == pszDst) return std::string("");
 51     ZeroMemory(pszDst, nLen);
 52 
 53     WideCharToMultiByte(CP_ACP, 0, pwszSrc, -1, pszDst, nLen, NULL, NULL);
 54     pszDst[nLen -1] = 0;
 55 
 56     std::string strTemp(pszDst);
 57     delete [] pszDst;
 58 
 59     return strTemp;
 60 }
 61 std::wstring Ansi2WChar(LPCSTR pszSrc, int nLen)
 62 
 63 {
 64     int nSize = MultiByteToWideChar(CP_ACP, 0, (LPCSTR)pszSrc, nLen, 0, 0);
 65     if(nSize <= 0) return NULL;
 66 
 67     WCHAR *pwszDst = new WCHAR[nSize+1];
 68     if( NULL == pwszDst) return NULL;
 69     ZeroMemory(pwszDst, nSize);
 70 
 71     MultiByteToWideChar(CP_ACP, 0, pszSrc, nLen, pwszDst, nSize);
 72     pwszDst[nSize] = 0;
 73 
 74     if( pwszDst[0] == 0xFEFF) // skip Oxfeff
 75         for(int i = 0; i < nSize; i ++)
 76             pwszDst[i] = pwszDst[i+1];
 77 
 78     wstring wcharString(pwszDst);
 79     delete pwszDst;
 80 
 81     return wcharString;
 82 }
 83 
 84 //-----------------------------------------------------------------------------
 85 //虽然可以转换宽窄字符串，但是不能解决中文乱码问题
 86 std::wstring strtowstr(const std::string &i_str)
 87 {
 88     std::wstring_convert<std::codecvt_utf8<wchar_t>> conv;
 89     std::wstring l_wstr = conv.from_bytes(i_str);
 90     return l_wstr;
 91 }
 92 std::string wstrtostr(const std::wstring &i_wstr)
 93 {
 94     std::wstring_convert<std::codecvt_utf8<wchar_t>> conv;
 95     std::string l_str = conv.to_bytes(i_wstr);
 96     return l_str;
 97 }
 98 
 99 //-----------------------------------------------------------------------------
100 bool OutputFile(const string &i_strFileName, const string &i_strExt)
101 {
102     std::locale old_loc = std::locale::global(std::locale(""));
103     ofstream ofs(i_strFileName, ifstream::out|ifstream::binary);
104     if (ofs.rdstate()!=ios_base::goodbit)
105     {
106         cout << "open file error." << endl;
107         return false;
108     }
109     ofs.write(i_strExt.c_str(), i_strExt.length());
110     ofs.clear();
111     ofs.close();
112     std::locale::global(old_loc);
113     return true;
114 }
115 bool OutputFileW(const string &i_strFileName, const wstring &i_wstrExt)
116 {
117     std::locale old_loc = std::locale::global(std::locale(""));
118     wofstream wofs(i_strFileName, ifstream::out|ifstream::binary);
119     if (wofs.rdstate()!=ios_base::goodbit)
120     {
121         wcout << L"open file error." << endl;
122         return false;
123     }
124     wofs.write(i_wstrExt.c_str(), i_wstrExt.length());
125     wofs.clear();
126     wofs.close();
127     std::locale::global(old_loc);
128     return true;
129 }
130 //bool OutputFile2(const string &i_strFileName, const string &i_strExt)
131 //{
132 //    //1
133 //    //std::locale loc("");
134 //    //typedef codecvt<wchar_t, char, std::mbstate_t> facet_type;
135 //    //const std::locale::facet &myfacet = use_facet<facet_type>(loc);
136 //    //std::locale curloc(loc, &myfacet);
137 //
138 //    //std::locale loc("");
139 //    //std::codecvt_utf8_utf16<wchar_t> *pconv = new std::codecvt_utf8_utf16<wchar_t> ();
140 //    //std::locale curloc(loc, pconv);
141 //
142 //    ofstream ofs(i_strFileName, ifstream::out|ifstream::binary);
143 //    if (ofs.rdstate()!=ios_base::goodbit)
144 //    {
145 //        cout << "open file error." << endl;
146 //        return false;
147 //    }
148 //    ofs.imbue(std::locale(std::locale(""), new std::codecvt_utf8_utf16<wchar_t>));
149 //    //ofs.imbue(curloc);
150 //    ofs.write(i_strExt.c_str(), i_strExt.length());
151 //    ofs.clear();
152 //    ofs.close();
153 //
154 //    return true;
155 //}
156 
157 //测试以上代码部分
158 void main()
159 {
160     string strExt = "这是一个窄字符串";
161     wstring wstrExt = L"这是一个宽字符串";
162 
163     //string strRet;
164     //wstring wstrRet;
165     //wstrRet = s2ws(strExt);
166     //strRet = ws2s(wstrExt);
167     //cout << strRet << endl;
168     //wcout << wstrRet << endl;
169 
170     //string strFileName1 = "F:\\Desktop\\20140411\\测试\\output.txt";
171     //bool f1 = OutputFile(strFileName1, strExt);
172     //string strFileName2 = "F:\\Desktop\\20140411\\测试\\outputw.txt";
173     //bool f2 = OutputFileW(strFileName2, wstrExt);
174 
175     //LPCWSTR pwstr = _TEXT("windows宽字符串");
176     //std::string strResult = WChar2Ansi(pwstr);
177     //cout << strResult << endl;
178     //LPCSTR pstr = "windows窄字符串";
179     ////size_t cnlen = lstrlenA(pstr);
180     //size_t cblen = lstrlenA(pstr)*sizeof(CHAR);
181     //std::wstring wstrResult = Ansi2WChar(pstr, cblen);
182     //wcout << wstrResult << endl;
183 
184     //std::wstring wstr = L"wchar_t宽字符";
185     //std::string strResult = wstrtostr(wstr);
186     //cout << "宽字符转窄字符：" << strResult << endl;
187 
188     cout << endl;
189 }