table.cpp

来自「某个实验事编写粗糙集智能信息处理的程序」· C++ 代码 · 共 493 行

// Table.cpp : implementation file
//

#include "stdafx.h"
#include "Table.h"
#include "fstream.h"
#ifndef MAX
#define MAX 30
#endif
#ifdef _DEBUG
#define new DEBUG_NEW
#undef THIS_FILE
static char THIS_FILE[] = __FILE__;
#endif

/////////////////////////////////////////////////////////////////////////////
// CTable



CTable::CTable()
{
	m_AttrName=NULL;
	m_AttrType=NULL;
	m_Record=NULL;
	m_strValue=NULL;
	m_nCount=0;
	m_nConAttrNum=0;
	m_bHaveCuts=false;
	m_Cuts=NULL;
	m_bStr=NULL;
	m_bHaveStr=false;
	m_bAttr=NULL;
	m_bObj=NULL;
}

CTable::~CTable()
{
	int i,j;
	if(m_Record!=NULL)
	{
    	for(i=0;i<m_nCount;i++)
    		delete []m_Record[i];
    	delete []m_Record;
	}
	if(m_AttrName!=NULL)
	{
    	for(i=0;i<m_nConAttrNum+1;i++)
	    	delete []m_AttrName[i];
    	delete []m_AttrName;
    }
	if(m_AttrType!=NULL)
	{
		for(i=0;i<m_nConAttrNum+1;i++)
			delete []m_AttrType[i];
		delete []m_AttrType;
	}
	if(m_strValue!=NULL)
	{
		for(i=0;i<m_nConAttrNum+1;i++)
		{
			if(m_bStr[i])
			{
				for(j=0;j<m_nCount;j++)
					delete []m_strValue[i][j];
				delete []m_strValue[i];
			}
		}
		delete []m_strValue;
	}
	if(m_Cuts)
	{
		for(i=0;i<m_nConAttrNum+1;i++)
		{
			if(m_Cuts[i])
				delete m_Cuts[i];
		}
		delete []m_Cuts;
	}
	if(m_bAttr)
		delete []m_bAttr;
	if(m_bObj)
		delete []m_bObj;
	if(m_bStr)
		delete []m_bStr;
}
int CTable::GetConAttrNum()
{
	return m_nConAttrNum;
}
int CTable::GetCount()
{
	return m_nCount;
}

bool CTable::StrToInt(int col)
{
	int i,j;                                //循环变量
	int att_num=0;                          //统计字符串的种类
	bool find=FALSE;
	char ** val;
	try
	{
		val=new char * [m_nCount];		//count为单元个数
	}
	catch(CMemoryException * e)
	{
		::MessageBeep(MB_ICONHAND);
		AfxMessageBox("Out of the memory!",MB_OK|MB_ICONSTOP);
		e->Delete();
		return 0;
	} 

	for(i=0;i<m_nCount;i++)				//
	{
		try
		{
			val[i]=new char[MAX];//用来存放信息表每个属性值得中间信息表
		}
		catch(CMemoryException * e)
		{
	    	::MessageBeep(MB_ICONHAND);
	       	AfxMessageBox("Out of the memory!",MB_OK|MB_ICONSTOP);
			e->Delete();
	    	return 0;
		} 
	}

	strcpy(val[0],m_strValue[col][0]);//全部拷贝
	att_num++;//开始att_num=0   //赋为1
	for(i=1;i<m_nCount;i++)
	{								//和val中已有的数据比较
		for(j=0;j<att_num;j++)
			if(strcmp(m_strValue[col][i],val[j])==0)
			{//第一次时肯定为真?
				find=TRUE;
				break;
			}
		if(find==FALSE)//此时把strValuepcol][1[赋予val[1]
			strcpy(val[att_num++],m_strValue[col][i]);
		find=FALSE;
	}
	for(i=0;i<m_nCount;i++)
		for(j=0;j<att_num;j++)
			if(strcmp(m_strValue[col][i],val[j])==0)
			{
				m_Record[i][col]=j;//(float)j;
				break;
			}

	for(i=0;i<m_nCount;i++)
		delete []val[i];
    delete []val;
	return TRUE;
}

bool CTable::InitTable(CString infotable)
{
	
	int i,j;                                    //循环变量
	FILE * fp;                                  //文件指针
	fp=fopen(infotable,"r");
	if(!fp)
	{
		::MessageBeep(MB_ICONHAND);
		AfxMessageBox("Some error happen,File can't be opened!",
			          MB_OK|MB_ICONSTOP);
		return false;
	}
	fscanf(fp,"Style: %s\n",m_strStyle);		//
	fscanf(fp,"Stage: %d\n",&m_nStage);			//
	if(stricmp(m_strStyle,"train")!=0) 
	{
    	if(AfxMessageBox("file input is not correct, are you sure to continue?",
			MB_YESNO|MB_ICONINFORMATION|MB_DEFBUTTON2)==IDNO)
		{	fclose(fp);
			return false;
		}
	}
	fscanf(fp,"Condition attributes number:%d\n",&m_nConAttrNum);
	fscanf(fp,"Records number:%d\n",&m_nCount);

//-------allocate memory for m_bAttr and initialize-----------------
	if((m_bAttr=new bool[m_nConAttrNum+1])==NULL ) //initialize m_bAttr
	{
		//memory error
		AfxMessageBox("out of memory!");
		fclose(fp);
		return false;
	} 
	for( i=0;i<m_nConAttrNum+1;i++)
		m_bAttr[i]=true;

//----------allocate memory for m_bStr and initialize---------------
/*
	m_bStr=new bool[m_nConAttrNum+1];	//initialize m_bStr
	for(i=0;i<m_nConAttrNum+1;i++)
		m_bStr[i]=false;
*/

//-----------allocate memory for m_bObj--------------
	if((m_bObj=new bool[m_nCount])==NULL)
	{		
		AfxMessageBox("out of memory!");//memory error
		fclose(fp);
		return false;
	}
	for( i=0;i<m_nCount;i++)
		m_bObj[i]=true;

//---------------condition attribute name-------------
	if((m_AttrName=new char* [m_nConAttrNum+1])==NULL)
	{
		//memory error
		AfxMessageBox("out of memory!");
		fclose(fp);
		return false;
	}
	for(i=0;i<m_nConAttrNum+1;i++)
	{
		try
		{
    		m_AttrName[i]=new char[MAX];  // 属性名称
		}
		catch(CMemoryException * e)
		{
	    	::MessageBeep(MB_ICONHAND);
	       	AfxMessageBox("Out of the memory!",MB_OK|MB_ICONSTOP);
			e->Delete();
			fclose(fp);
	    	return false;
		} 
		fscanf(fp,"%s",m_AttrName[i]);
	} 

	if((m_AttrType=new char * [m_nConAttrNum+1])==NULL)
	{
		//memory error
		AfxMessageBox("out of memory!");
		fclose(fp);
		return false;
	}
	for(i=0;i<m_nConAttrNum+1;i++)
	{
		if((m_AttrType[i]=new char[10])==NULL)
		{
			//memory error
			AfxMessageBox("out of memory!");
			fclose(fp);
			return false;
		}
		fscanf(fp,"%s",m_AttrType[i]);
	}

/*	m_strValue=new char ** [m_nConAttrNum+1];     //字符串类型的属性值
	for(i=0;i<m_nConAttrNum+1;i++)					//m_nCount为纪录数
	{
		if(!m_bStr[i]) 
			m_strValue[i]=NULL;
		else
		{		
			try
			{
				m_strValue[i]=new char * [m_nCount];
			}
			catch(CMemoryException * e)
			{
				::MessageBeep(MB_ICONHAND);
				AfxMessageBox("Out of the memory!",MB_OK|MB_ICONSTOP);
				e->Delete();
				return false;
			} 
			
			for(j=0;j<m_nCount;j++)
			{
				try
				{
					m_strValue[i][j]=new char[MAX];
				}
				catch(CMemoryException * e)
				{
					::MessageBeep(MB_ICONHAND);
					AfxMessageBox("Out of the memory!",MB_OK|MB_ICONSTOP);
					e->Delete();
					return false;
				} 
			}
		}
	}
*/
/////////////////////alocate memory for  m_Record
	try
	{
		m_Record=new int* [m_nCount];		//create table(int)
	}
	catch(CMemoryException * e)
	{
		::MessageBeep(MB_ICONHAND);
		AfxMessageBox("Out of the memory!",MB_OK|MB_ICONSTOP);
		e->Delete();
		fclose(fp);
		return false;
	} 
	for(i=0;i<m_nCount;i++)
	{
		try
		{
			m_Record[i]=new int[m_nConAttrNum+1];
		}
		catch(CMemoryException * e)
		{
			::MessageBeep(MB_ICONHAND);
			AfxMessageBox("Out of the memory!",MB_OK|MB_ICONSTOP);
			e->Delete();
			fclose(fp);
			return false;
		} 
	}

///////////////////////read data from file input
    CString str_temp;
	for(i=0;i<m_nCount;i++)
	{
		for(j=0;j<m_nConAttrNum+1;j++)
		{ 
			fscanf(fp,"%s",str_temp);
			m_Record[i][j]=atoi(str_temp);
		}
	}
						
//	if(m_bHaveStr)				//deal with string column
//		for(i=0;i<m_nConAttrNum+1;i++)
//		{	if(m_bStr[i])
//				StrToInt(i);
//		}
//处理断点
	char temp[20];
	if(fscanf(fp,"%s\n",temp)==EOF) 
	{
		TRACE("EOF");
	}
	else
	{
		if(stricmp(temp,"[Cuts]")==0)
		{
			m_bHaveCuts=true;
			if((m_Cuts=new cuts* [m_nConAttrNum+1])==NULL)
			{
				//memory error
				AfxMessageBox("out of memory!");
				fclose(fp);
				return false;
			}
			for(i=0;i<m_nConAttrNum+1;i++)
				m_Cuts[i]=NULL;			//初始化
			int col=0;		//该列有断点 第0例
			int n;			//有n个断点
			int m;			//离散化为m
			char tp1[100];
			fscanf(fp,"%d\n",&col);
			do
			{
				fscanf(fp,"%d\n",&n);
				m_Cuts[col]=new cuts(col,n);
				TRACE("%d\n%d\n",col,n);
				for(int k=0;k<n;k++)
				{
					fscanf(fp,"%s %d\n",tp1,&m);
					m_Cuts[col]->data[k]=new char[strlen(tp1)+1];
					strcpy(m_Cuts[col]->data[k],tp1);
					m_Cuts[col]->m[k]=m;
					TRACE("%s %d\n",tp1,m);
				}
			}
			while(fscanf(fp,"%d\n",&col)!=EOF);
		}
	}
//	else TRACE("no cuts\n");
	fclose(fp);	
	return TRUE;
}

bool CTable::CompObj(int i,int j)//compare object i and j
{//仅仅比较条件属性,全部相等返回为true,否则为false
	if(!m_bObj[j])//被删除
       return false;
	for (int k=0;k<m_nConAttrNum;k++)
	{
		if(!m_bAttr[k]) continue;//当条件属性为false时候,不比较
		if(m_Record[i][k]!=m_Record[j][k])
			return false;
	}
	return true;
}

void CTable::deleteDup()
{//删除重复记录
	for(int i=0;i<m_nCount-1;i++)
	{
		if(!m_bObj[i])  continue;
		for(int j=i+1;j<m_nCount;j++)
		{//比较后面的记录,看是否有相同的条件属性和决策,则删除,
			if( m_bObj[j] &&CompObj(i,j) && m_Record[i][m_nConAttrNum]==m_Record[j][m_nConAttrNum]) 
				m_bObj[j]=false;
		}
	}
}

bool CTable::WriteFile(CString name)
{
	TRACE(" Kylin CTable::writefile;\n");
//	deleteDup();
	int i;
//	AfxMessageBox("completed!");
	int nConRed=0;				
	for(i=0;i<m_nConAttrNum;i++)//计算约见后的条件属性个数
		if(m_bAttr[i]) nConRed++;
//	int nRecRed=0;				//计算约简后的单元个数
//	for(i=0;i<m_nCount;i++)	
//		if(m_bObj[i])  nRecRed++;
	ofstream out(name,ios::out);	
	if(! out)
		return false;
//	out<<"Style: "<<m_strStyle<<"\n";			//写文件头
	out<<"Style:train\n";
	out<<"Stage:3\n";
	out<<"Condition attributes number:";
	 out<<nConRed<<'\n';
	out<<"The Number of Condition attributes deleted:";
	out<<m_nConAttrNum-nConRed<<endl;
	out<<"The position of Condition attributes deleted:";
	if((m_nConAttrNum-nConRed)>0)
		for(i=0;i<m_nConAttrNum;i++)
			if(!m_bAttr[i]) out<<' '<<i;
	out<<"\n";
	out<<"Records number:";
//	out<<nRecRed<<"\n";
	out<<m_nCount<<"\n";
	for(i=0;i<m_nConAttrNum;i++)	//写属性名称
		if(m_bAttr[i])  
			out<<m_AttrName[i]<<"  ";
	out<<m_AttrName[m_nConAttrNum]<<'\n';
	for(i=0;i<m_nConAttrNum;i++)	//写数据类型
		if(m_bAttr[i])  
			out<<m_AttrType[i]<<"  ";
	out<<m_AttrType[m_nConAttrNum]<<'\n';
	for(i=0;i<m_nCount;i++)
	{
		for(int j=0;j<m_nConAttrNum;j++)
		{
			if(m_bAttr[j])						//写条件属性
			{
				out<<m_Record[i][j]<<"  ";
			}
		}
		out<<m_Record[i][m_nConAttrNum];
		out<<"\n";
	}
	//deal with cuts
//	m_bHaveCuts=false;
	if(m_bHaveCuts)
		if(nConRed==0)//表明在约简属性集中不存在属性,故不需要写断点
		   m_bHaveCuts=false;
	if(!m_bHaveCuts)
	{
		out.close();
	     return true;
	}
	out<<"[Cuts]"<<'\n';
	int mmm=0;
	for(i=0;i<m_nConAttrNum+1;i++)
	{
		if(!m_bAttr[i]){ mmm++; continue;}
		if(!m_Cuts[i]) continue;
		out<<m_Cuts[i]->col-mmm<<"\n"<<m_Cuts[i]->n<<"\n";
		for(int j=0;j<m_Cuts[i]->n;j++)
			out<<m_Cuts[i]->data[j]<<' '<<m_Cuts[i]->m[j]<<'\n';
	}
	out.close();
	return true;
}

bool CTable::Perform(const char* filein,const char* fileout)
{
	if(!InitTable(filein))
		return false;
	if(! reduct() )
		return false;
	if(!WriteFile(fileout))
		return false;
	return true;
}