twirls/RandSim/rand_sim.cpp

#include <iostream>
#include <fstream>
#include <algorithm>
#include <random>
#include <unordered_map>
#include <omp.h>
#include <time.h>
#include <string>
#ifdef _WIN32
#include <io.h>
#include <process.h>
#define F_OK 0
#else
#include <unistd.h>
#endif
#include "rand_sim.h"
using namespace std;

void txtReaderVvs(vector<vector<string>>& vecVecStr, string cf) {
	filebuf fb;
	string fileName = cf;
	if (fb.open(fileName.c_str(), ios::in) == NULL) {
		cout << "error" << endl;
	}
	istream is(&fb);
	string lineInfo;
	while (getline(is, lineInfo)) {
		int i = 0;
		vector<string> vecStr;
		string tmp;
		while (i < lineInfo.length()) {
			while (i < lineInfo.length() && lineInfo[i] != ' ') {
				tmp += lineInfo[i++];
			}
			if (!tmp.empty()) {
				transform(tmp.begin(), tmp.end(), tmp.begin(), ::tolower);
				vecStr.push_back(tmp);
			}
			tmp.clear();
			++i;
		}
		vecVecStr.push_back(vecStr);
	}
	fb.close();
}
template <class AUTO>
void txtWriterVvs(vector<vector<AUTO>>& vecVecStr, string cf) {
	ofstream os;
	string fileName = cf;
	os.open(fileName, ios::out);

	for (auto vecStr : vecVecStr) {
		for (auto s : vecStr)
			os << s << " ";
		os << "" << endl;
	}
}

void txtReaderVs(vector<string>& vecStr, string cf) {
	filebuf fb;
	string fileName = cf;
	if (fb.open(fileName.c_str(), ios::in) == NULL) {
		cout << "error" << endl;
	}
	istream is(&fb);
	string lineInfo;
	while (getline(is, lineInfo)) {
		transform(lineInfo.begin(), lineInfo.end(), lineInfo.begin(), ::tolower);
		vecStr.push_back(lineInfo);
	}
	fb.close();
}
void txtWriterVs(vector<string>& vecStr, string cf) {
	ofstream os;
	string fileName = cf;
	os.open(fileName, ios::out);

	for (auto s : vecStr) {
		os << s << endl;
	}
}

/* run this program using the console pauser or add your own getch, system("pause") or input loop */
int main(int argc, char** argv) {
	//input : dicr, wd2, wd1s;
	vector<string> dicr;
	vector<vector<string>> wd2;
	vector<vector<string>> wd1s;

	string dicrName = "F:\\myWork\\20191129\\dicr.txt";
	string wd2Name = "F:\\myWork\\20191129\\wd2.txt";
	string wd1sName = "F:\\myWork\\20191129\\wd1s.txt";
	string ZrName = "F:\\myWork\\20191129\\Zr.txt";
	int loopNum;

	if (argc != 6) {
		cout << "Unexpected count of input arguments! Please input paths of dicr, wd2, wd1s, Zr and loop num! eg: F:\\myWork\\dicr.txt or 1000" << endl;
		return -1;
	}

	dicrName = argv[1];
	wd2Name = argv[2];
	wd1sName = argv[3];
	ZrName = argv[4];
	loopNum = atoi(argv[5]);

	if (access(dicrName.c_str(), F_OK) == -1) {
		cout << "Invalid dicr file path! Please input a right path!" << endl;
		return -1;
	}

	if (access(wd2Name.c_str(), F_OK) == -1) {
		cout << "Invalid wd2 file path! Please input a right path!" << endl;
		return -1;
	}

	if (access(wd1sName.c_str(), F_OK) == -1) {
		cout << "Invalid wd1s file path! Please input a right path!" << endl;
		return -1;
	}

	txtReaderVs(dicr, dicrName);
	txtReaderVvs(wd2, wd2Name);
	txtReaderVvs(wd1s, wd1sName);
	// cout<<"/*************   Input file read finished!   ***************/"<<endl;
	//cout<<dicr.size()<<" "<<wd2.size()<<" "<<wd1s.size()<<endl;
	//for(int i=0;i<50;++i)
	//	cout<<wd1s[i].size()<<endl;
	unordered_map<string, int> dicrHashMap;
	for (int i = 0;i < dicr.size();++i)
		dicrHashMap[dicr[i]] = i;
	//output : Zr;
	vector<vector<int>> Zr(loopNum, vector<int>(dicr.size(), 0));
	vector<int> randNums(wd2.size(), 0);
	for (int i = 0; i < wd2.size(); ++i) {
		randNums[i] = i;
	}
	//cout<<"12345"<<endl;
	clock_t start, end;
	start = clock();
	//cout<<dicr.size()<<endl;
	//cout<<wd2.size()<<" "<<wd2[0].size()<<endl;
	//cout<<wd1s.size()<<" "<<wd1s[0].size()<<endl; 
	// #pragma omp parallel for
	for (int i = 0; i < loopNum; ++i) {
		clock_t begin = clock();
		//#pragma omp atomic
		/*vector<int> randNums(wd2.size(),0);
		for(int i = 0; i < wd2.size(); ++i){
			randNums[i] = i;
		}*/
		std::random_device rd;
		std::shuffle(randNums.begin(), randNums.end(), std::default_random_engine(rd()));
		//cout<<i<<endl;
		for (int j = 0; j < wd1s.size(); ++j) {
			vector<string> wd2rCell = wd2[randNums[j]];
			//cout<<"step 1"<<endl;
			vector<int> f(dicr.size(), 0);
			//cout<<"step 2"<<endl;
			for (int k = 0;k < wd2rCell.size();++k) {
				string tmp = wd2rCell[k];
				//cout<<tmp<<" "<<k<<endl;
				if (dicrHashMap.find(tmp) != dicrHashMap.end()) {
					//cout<<dicrHashMap[tmp]<<endl;
					f[dicrHashMap[tmp]] = 1;
				}
			}
			//cout<<"step 3"<<endl;
			for (int k = 0;k < dicr.size();++k) {
				if (f[k] == 1)
					Zr[i][k]++;
			}
			//cout<<"step 4"<<endl;
		}
		clock_t finish = clock();
		double loopTime = (double)(finish - begin) / CLOCKS_PER_SEC;
		// cout<<"Loop : "<<i<<"  Time consuming :"<<loopTime<<"s"<<endl;
	}
	end = clock();
	double endtime = (double)(end - start) / CLOCKS_PER_SEC;
	cout<<"RandSim Total time:"<<endtime << " s" << endl;
	txtWriterVvs(Zr, ZrName);
	return 0;
}