word.cpp

来自「对中科院分词程序的简化版本」· C++ 代码 · 共 64 行

// RRR.cpp : Defines the entry point for the DLL application.
//

#include "stdafx.h"
#include  <io.h>
#include <stdio.h>
#include <string.h>
#include "Utility\\Utility.h"
#include "Result\\Result.h"

extern "C" _declspec(dllexport) bool IsDataExists()
{	//中科院程序运行必需DATA文件夹,此函数判断是否存在
	char sDataFiles[][40]={"data\\BigramDict.dct",
		                   "data\\coreDict.dct",
						   "data\\lexical.ctx",
		                   "data\\nr.dct",
						   "data\\nr.ctx",
		                   "data\\ns.dct",
						   "data\\ns.ctx",
		                   "data\\tr.dct",
						   "data\\tr.ctx",
						   ""

	};
	int i=0;
	while(sDataFiles[i][0]!=0)
	{
		if((_access( sDataFiles[i], 0 ))==-1)
			return false;
		i++;
	}
	return true;
}



extern "C" _declspec(dllexport) char* RunFenci(char m_file[]) 
{		
	char *sResult;
	CResult m_ICTCLAS;

	if(strlen(m_file)==0)
	{
		sResult="待分析内容为空!";
		return sResult;
	}
	
	int m_OperateType=2;//设定默认的输出格式为北大标准
	int m_OutputFormat=0;
	m_ICTCLAS.m_nOutputFormat=m_OutputFormat;
	m_ICTCLAS.m_nOperateType=m_OperateType;


	if(m_OutputFormat!=2)
		sResult=new char [(strlen((char *)(LPCTSTR)m_file)+13)*3];
	else
		sResult=new char [(strlen((char *)(LPCTSTR)m_file)+13)*50];
	if(m_ICTCLAS.ParagraphProcessing((char *)(LPCTSTR)m_file,sResult))		
	{return  sResult;	}
	else
	{return "";}
}