unit1.~cpp

处理语料库信息的小程序

//---------------------------------------------------------------------------

#include <vcl.h>
#include <stdio.h>
#pragma hdrstop
#include <stdlib.h>
#include "Unit1.h"
#include <sys\stat.h>
//---------------------------------------------------------------------------
#pragma package(smart_init)
#pragma link "AMONGLib_OCX"
#pragma link "TntStdCtrls"
#pragma link "TntStdCtrls"
#pragma resource "*.dfm"
TForm1 *Form1;
//---------------------------------------------------------------------------
__fastcall TForm1::TForm1(TComponent* Owner)
        : TForm(Owner)
{
}
//==============================================================================
FILE *File_Open(WideString dir,WideString mode)
{
    FILE *file;
    file=_wfopen(dir,mode);
    if(file!=NULL)
    {
      fseek(file,2L,SEEK_SET);
    }
    else
    {
      MessageBox(NULL,"Could not open Dict file.", "Wornning", MB_OK);
    }
    return file;
}

//==============================================================================
WideString Read_File(FILE* file,int bytes,wchar_t c)
{
   int i=0;
   WideString s;
   wchar_t ch[150],mc[1];

   fread(mc,bytes,1,file);
   for(;;fread(mc,bytes,1,file))
   {
      if(mc[0]==c)
      {
          s=WideString(ch).SubString(1,i);
          break;
      }
      else if(mc[0]==13||mc[0]==10)
      {
          continue;
      }
      else
      {
          ch[i++]=mc[0];
      }

   }
   return s;
}
//==============================================================================
void File_Processor(WideString s)
{
   WideString cond,str;

   word.eword=Trim(s.SubString(1,s.Pos(".")-1));
   word.gender=Trim(s.SubString(s.Pos(".")+1,s.Pos("/")-s.Pos(".")-1));
   word.mword=Trim(s.SubString(s.Pos("/")+1,s.Length()-s.Pos("/")));

   cond="Select * from mt_lab where citiao='"+Trim(word.eword)+"' and cixing='"+Trim(word.gender)+"'";
   Form1->Query1->SQL->Clear();
   Form1->Query1->SQL->Add(cond);
   Form1->Query1->Open();
   Form1->Query1->First();
   if(Form1->Query1->Eof)
   {
       str=WideString("INSERT INTO mt_lab(citiao,cixing,mcat,gender,yixiang,yuyilei) VALUES('")+
            word.eword+WideString("','")+word.gender+WideString("','N','MALE','")+
            word.mword+WideString("',' ')");
       Form1->ADOCommand1->CommandText=str;
       Form1->ADOCommand1->Execute();
       Form1->TntEdit1->Text=Form1->ADOCommand1->CommandText;  //ShowMessage(str);
   }

}

//==============================================================================
void __fastcall TForm1::Button1Click(TObject *Sender)
{
  FILE *file;
  WideString s;
  struct stat statbuf;

  file=File_Open(L"words\\words.n.1",L"rb");
  fstat(fileno(file), &statbuf);
  Form1->ProgressBar1->Max=statbuf.st_size;
  Form1->ProgressBar1->Step=(Form1->ProgressBar1->Max)/10000;
  s=Read_File(file,2,L'=');
  while(s.SubString(1,1)!=WideString("#"))
  {
     File_Processor(s);
     s=Read_File(file,2,L'=');
     Form1->ProgressBar1->StepIt();
  }
  fclose(file);
}
//==============================================================================
void __fastcall TForm1::FormCreate(TObject *Sender)
{
    ADOConnection1->ConnectionString="Provider=Microsoft.Jet.OLEDB.4.0;Data Source=emmt.mdb;Persist Security Info=False";
}
//==============================================================================