boolRS.cpp~

#include"irs.h"
void readallfile()
{
	struct dirent *pDirEntry=NULL;
	DIR *pDir=NULL;
	int len=50000000;
	FILE *fp;
	string dir="./input/";
	if((pDir=opendir("./input/"))==NULL)
	{
		cout<<"open dir failed!"<<endl;
	}
	else
	{
		int i=1;
		while((pDirEntry=readdir(pDir))!=NULL)
		{
			if(!strcmp(pDirEntry->d_name,"..")||!strcmp(pDirEntry->d_name,"."))
				continue;
		//	cout<<i<<" Read OK : "<<pDirEntry->d_name<<endl;
			i++;
			dir=dir+pDirEntry->d_name;
		//	cout<<dir<<endl;
			if((fp=fopen(dir.c_str(),"r"))==NULL)
			{
				cout<<"open file error"<<endl;
			}
			else
			{
				fread(buf,1,len,fp);
		//		cout<<buf<<endl;
				fclose(fp);
				doc pdoc;
				pdoc.fname=pDirEntry->d_name;
				pdoc.fcontent=buf;
				docs.push_back(pdoc);
			//	cout<<"Save : "<<pDirEntry->d_name<<endl;
			
			}
			dir="./input/";
			memset(buf,0,sizeof(buf));
		}
		closedir(pDir);
	}
//	cout<<docs[30].fcontent<<endl;
	cout<<docs[30].fname<<endl;
	cout<<docs[30].fcontent.length()<<endl;
	cout<<docs[42].fname<<endl;
	cout<<docs[42].fcontent.length()<<endl;
}
void genIndex()
{
	const char *spilt="[]\t\r\n ,()-+.|\"':?~!;";
	for(int i = 0;i < docs.size(); i++)
	{   
	//	cout<<i<<endl;
		//cout<<docs[0].fcontent.length()<<endl;
		char *str=(char *)malloc(docs[i].fcontent.length()+10);
		if(str==NULL)
		{
			cout<<"malloc error!"<<endl;
			return;
		}
		//cout<<sizeof(str)<<endl;
		strcpy(str,docs[i].fcontent.c_str());
		char *str2=str;
		char *p=strtok(str,spilt);
		while(p)
		{
//			p=trim(p);
			string temp=p;
			temp=strtolow(temp);
			if(myindex.find(temp)==myindex.end())
			{
				list<int> listtemp;
				listtemp.push_back(i);
				myindex.insert(pair<string,list<int> >(temp,listtemp));
			}
			else if(myindex.at(temp).back()!=i)
			{
				myindex.at(temp).push_back(i);
			}
	//		cout<<"--"<<p<<"--"<<endl;
			p=strtok(NULL,spilt);
		}
		free(str2);
	}
	//add the length of postings lists	
	for(map<string,list<int> >::iterator it=myindex.begin();it!=myindex.end();it++)
	{
		it->second.push_front(it->second.size());
	}
}
void printIndex()
{
	for(map<string,list<int> >::iterator it=myindex.begin();it!=myindex.end();it++)
	{
		cout<<it->first;
		for(list<int>::iterator its=it->second.begin();its!=it->second.end();its++)
			cout<<" "<<*its;
		cout<<endl;
	}
}
list<int> queryOne(string str)
{
	list<int> l;
	map<string,list<int> >::iterator it = myindex.find(str);
	if(it!=myindex.end())
	{
		l = it->second;
		cout<<l.back()<<endl;
	}
	return l;
}
void printList(list<int> l)
{
	list<int>::iterator it=l.begin();
	cout<<"nums: "<<*it<<endl;
	for(;it!=l.end();it++)
	{
		cout<<*it<<" "<<docs[*it].fname<<endl;
	}
	cout<<endl;
}
int main()
{
	map<string,int> index;
	index["hello"]=9;
	index["world"]=7;
	map<string,int>::iterator it=index.begin();
	while (it!=index.end())
	{
		cout<<it->first<<"   "<<it->second<<endl;
		it++;
	}
	readallfile();
	char *c=(char*)malloc(4);
	char ch[]="zfx hello, zhang fei (xue) h - + gh";
	c=ch;
	cout<<c<<endl;
	c[0]='F';
	cout<<c<<endl;
	char chh[]="zfx ubuntu:irs g++ . boo;hellp";
	cout<<"------------------------"<<endl;
	genIndex();
	printIndex();
	list<int> myq=queryOne("caesar");
	list<int> m2=queryOne("brutus");
	list<int> m3=queryOne("calpurnia");
	if(myq.empty())
		cout<<"NOT FIND!"<<endl;
	else
	{
		printList(myq);
		printList(m2);
		printList(m3);
	}
}