segtool.h

用字典进行预料的切词

// SegTool.h: interface for the CSegTool class.
//
//////////////////////////////////////////////////////////////////////

#if !defined(AFX_SEGTOOL_H__7595F220_94E2_4B62_804E_4C7977955CD1__INCLUDED_)
#define AFX_SEGTOOL_H__7595F220_94E2_4B62_804E_4C7977955CD1__INCLUDED_

#if _MSC_VER > 1000
#pragma once
#endif // _MSC_VER > 1000

#include<vector>
#include<map>
#include<string>
using namespace std;
#pragma warning (disable:4760)

class CSegTool  
{
public:
	bool InitSegTool(char *psWordDict);
	bool GetWord( int nIndex, char *psWord );
	int SegSentence( int nResNum ,char *psInputSen , char *psOutputSen ,char *&psRes);
	int SegFile( char *psInputFile,char *psOutputFile );
	
protected:
	bool GetOneWord(char *psInput ,char *psOutput);
	int SearchOneWord( char *psKey ); // return index or -1 if not
	int SearchWordHeadRange( char *psKey ,int nHead,int nTail,int &nReHead, int &nReTail);
	int SearchWordHeadTop( char *psKey ,int nHead, int nTail);
	int SearchWordHeadDown( char *psKey ,int nHead, int nTail);
	int BSearch( char *psKey ,int nHead , int nTail );
public:
	CSegTool();
	virtual ~CSegTool();
protected:
	vector<string> m_WordList;

};

#endif // !defined(AFX_SEGTOOL_H__7595F220_94E2_4B62_804E_4C7977955CD1__INCLUDED_)