多线程处理给定的文件夹下的所有子目录，进行高斯混合模型拟合

GMM/main.cpp

@@ -10,6 +10,8 @@
 #include <string>
 #include <vector>
 #include <queue>
+#include <filesystem>
+#include <thread>
 #ifdef _WIN32
 #include <io.h>
 #include <process.h>
@@ -22,6 +24,7 @@
 using namespace std;
 using std::cout;
 using std::vector;
+namespace fs = std::filesystem;
 
 /* 从mat文件中读取给定名称的矩阵数据，并获取矩阵的行列数值 */
 template<typename T>
@@ -39,7 +42,7 @@ T* ReadMatlabMat(const string &filePath, const string &mtxName, int *pRowNum, in
 	pMxArray = matGetVariable(pMatFile, mtxName.c_str()); //获取.mat文件里面名为matrixName的矩阵
 	rowNum = mxGetM(pMxArray);
 	colNum = mxGetN(pMxArray);
-	cout << rowNum << " " << colNum << endl;
+	// cout << rowNum << " " << colNum << endl;
 	matData = (double*)mxGetData(pMxArray); //获取指针
 
 	dst = new T[rowNum * colNum];
@@ -145,10 +148,10 @@ void GMMToFactorEY(GMM& gmm, double binWidth, vector<double> &vYBin, vector<doub
 	for (int i = 0; i < topM; ++i) {
 		pair<double, double> topEle = pqTopM.top();
 		pqTopM.pop();
-		cout << topEle.first << '\t' << topEle.second << endl;
+		// cout << topEle.first << '\t' << topEle.second << endl;
 		zoomFactorSum += topEle.first / topEle.second;
 	}
-	cout << endl;
+	// cout << endl;
 
 	double zoomFactor = zoomFactorSum / topM;
 
@@ -311,15 +314,84 @@ void processTxtData(const string& filePath) {
 		ifs.close();
 }
 
+/* 处理一个知识颗粒 */
+struct ThreadParam {
+	fs::path matFilePath;
+	fs::path outFilePath;
+};
+void ThreadProcessData(const ThreadParam& param) {
+	const fs::path& matFilePath = param.matFilePath;
+	const fs::path& outFilePath = param.outFilePath;
+	// cout << parrentPath.string() << '\t' << matFilePath.filename().string() << endl;
+	cout << outFilePath.string() << endl;
+	double* hs = nullptr;
+	int rowNum = 0;
+	int colNum = 0;
+
+	hs = ReadMatlabMat<double>(matFilePath.string(), "hs", &rowNum, &colNum);
+
+	vector<double>vXBin;
+	vector<double>vYBin;
+	vector<double>vEY;
+	vector<double>vFactor;
+	/* 用来保存数据，存入mat文件 */
+	vector<double>vDist(rowNum);
+	vector<double>vFactorAll;
+	for (int i = 0; i < rowNum; ++i) {
+		PutXtoBin(hs + i * colNum, colNum, 0.2, vXBin, vYBin);
+
+		GMM gmm(1, 2); // 1维， 2个高斯模型
+		gmm.Train(vXBin.data(), vXBin.size());
+
+		GMMToFactorEY(gmm, 0.2, vYBin, vFactor, vEY);
+		vDist[i] = CorrelationDistance(vYBin, vEY);
+		vFactorAll.insert(vFactorAll.end(), vFactor.begin(), vFactor.end());
+	}
+	/* 写入matlab文件 */
+	MATFile* pMatFile = matOpen(outFilePath.string().c_str(), "w");
+	SaveMatrix<double>(vFactorAll.data(), pMatFile, "factor", rowNum, 6);
+	SaveMatrix<double>(vDist.data(), pMatFile, "correlation", rowNum, 1);
+	matClose(pMatFile);
+
+	delete[] hs;
+}
+
 int main(int argc, char** argv) {
 
-	//if (argc != 2) 
-	//	cerr << "This program should take one argument(input data file)!" << endl;
-
+	if (argc != 4) {
+		cerr << "This program should take 3 arguments(1.parrent Dir; 2. mat file suffix; 3. out mat filename)!" << endl;
+		return 1;
+	}
+	string parrentDir(argv[1]); // 知识颗粒的父目录名称
+	string hsMatSuffix(argv[2]); // hs矩阵对应的mat文件的后缀名（可以是全文件名，可以是文件名后缀，必须保证唯一）
+	fs::path outFileName(argv[3]);
+	vector<thread> vThread;
+	clock_t begin, finish;
+	begin = clock();
 
+	/* 遍历所有的知识颗粒目录，注意进行处理 */
+	for (auto& childDir : fs::directory_iterator(parrentDir)) {
+		// cout << childDir.path().string() << endl;
+		fs::path outFilePath = childDir / outFileName;
+		for (auto& file : fs::directory_iterator(childDir)) {
+			// cout << file.path().filename().string() << endl;
+			const string& fileName = file.path().filename().string();
+			auto rPos = fileName.rfind(hsMatSuffix);
+			if (rPos != string::npos && fileName.size() - rPos == hsMatSuffix.size()) {
+				ThreadParam tParam = { file, outFilePath };
+				vThread.push_back(thread(ThreadProcessData, tParam));
+				// ThreadProcessData(tParam);
+			}
+		}
+	}
+	for (auto& thread : vThread) {
+		thread.join();
+	}
+	finish = clock();
+	cout << "Total time:" << (double)(finish - begin) / CLOCKS_PER_SEC << endl;
 	// processMatData(argv[1]);
-	processMatData("D:\\Twirls\\runtime\\ALS_test\\1775\\twirls_id_abs2class_hs.mat");
-	processTxtData("D:\\Twirls\\backup\\xy.txt");
+	// processMatData("D:\\Twirls\\runtime\\ALS_test\\1775\\twirls_id_abs2class_hs.mat");
+	// processTxtData("D:\\Twirls\\backup\\xy.txt");
 
 	return 0;
 }