文本编码的自动检测

来源：互联网发布：2017淘宝入驻条件费用编辑：程序博客网时间：2024/04/30 19:21

void CDetectCodeDlg::OnButton1()
{
// TODO: Add your control notification handler code here

::CoInitializeEx(NULL,COINIT_MULTITHREADED); // init COM

IMultiLanguage2* pIMultiLanguage2   =   NULL;
HRESULT hr   =   S_OK;

hr   =   CoCreateInstance(CLSID_CMultiLanguage,   NULL,   CLSCTX_INPROC_SERVER,   IID_IMultiLanguage2,   (void**)&pIMultiLanguage2);
if   (hr   ==   S_OK)
{
  DetectEncodingInfo   encoding[10];


  CFile file;
  if (!file.Open(_T("5.txt"),CFile::modeRead | CFile::shareDenyNone, NULL))
  {
   return ;
  }

  int nFileLen = (int)file.GetLength();
//  char *pBuf = new char[nFileLen];
    char *pBuf = new char[nFileLen+1];
    memset(pBuf, 0, nFileLen +1);
//  memset(pBuf, 0, nFileLen );


  file.SeekToBegin();
  file.Read(pBuf,nFileLen);

  file.Close();
  //  char*   str1="中华人民共和国万岁";
  char*   str1=pBuf;
  int   nScores   =   10;
  int   nBufLen   =   strlen(str1);
//  int   nBufLen   =   266465;

  memset(encoding,   0,   sizeof(DetectEncodingInfo)*10);
  hr   =   pIMultiLanguage2->DetectInputCodepage(MLDETECTCP_NONE,   0,   str1,   &nBufLen,   encoding,   &nScores);
  if   (hr   ==   S_OK)
  {
   TRACE(_T("%d,%d/n"),   encoding[0].nLangID,   encoding[0].nCodePage);
  }
  pIMultiLanguage2->Release();
  delete []pBuf;
}
::CoUninitialize();

}

PC上运行正常.WINCE上如果是Imultilanguage2创建失败,需要在image中添加Internet Explorer Multiple-Language Full API组件. 但是在wince上在执行DetectInputCodePage时仍是不正常的,除了中文会返回932,英文及其它语言的都会返回65001(UTF-8),而西里尔文会直接返回 E_FAIL,执行失败.在WINCE上应该是缺少正确的支持.