完成pubmed txt文件的处理，将cerr改成cout，因为这样在matlab里调用的时候，可以在命令行窗口显示错误信息

CommonLib/matlab_io.cpp

@@ -26,7 +26,7 @@ bool ReadMtxString(const string& filePath, const string& mtxName,
 
 	pMatFile = matOpen(filePath.c_str(), "r"); //打开.mat文件
 	if (pMatFile == nullptr) {
-		cerr << "filePath is error!" << endl;
+		cout << "filePath is error!" << endl;
 		return false;
 	}
 	pMxArray = matGetVariable(pMatFile, mtxName.c_str()); //获取.mat文件里面名为matrixName的矩阵
@@ -38,7 +38,7 @@ bool ReadMtxString(const string& filePath, const string& mtxName,
 		for (int j = 0; j < colNum; ++j) {
 			pCell = mxGetCell(pMxArray, j * rowNum + i);
 			if (mxGetString(pCell, strBuf, STRING_BUF_SIZE) != 0) {
-				cerr << "String is too large to fit in the buffer! " << i + 1 << '\t' << j + 1 << endl;
+				cout << "String is too large to fit in the buffer! " << i + 1 << '\t' << j + 1 << endl;
 				return false;
 			}
 			vStr[i * colNum + j] = strBuf;
@@ -57,7 +57,7 @@ T* ReadMtxDouble(const string& filePath, const string& mtxName, int* pRowNum, in
 	double* matData;
 	pMatFile = matOpen(filePath.c_str(), "r"); //打开.mat文件
 	if (pMatFile == nullptr) {
-		cerr << "filePath is error!" << endl;
+		cout << "filePath is error!" << endl;
 		return nullptr;
 	}
 	pMxArray = matGetVariable(pMatFile, mtxName.c_str()); //获取.mat文件里面名为matrixName的矩阵
@@ -90,7 +90,7 @@ bool SaveMtxDouble(T* src, MATFile* pMatFile, string matrixName, int rowNum, int
 	// memset(mtxData, 0, datasize * sizeof(double));
 	if (pMatFile == nullptr)
 	{
-		cerr << "mat file pointer is error!" << endl;
+		cout << "mat file pointer is error!" << endl;
 		return false;
 	}
 	for (int i = 0; i < rowNum; i++)

CppRun/main.cpp

@@ -17,7 +17,7 @@ int main(int argc, const char** argv) {
 	clock_t begin, finish;
 	begin = clock();
 	if (argc < 2) {
-		cerr << "This program take at least 1 arguments(CMD; [Options])!" << endl;
+		cout << "This program take at least 1 arguments(CMD; [Options])!" << endl;
 		return 1;
 	}
 

CppRun/process_pubmed_txt.cpp

@@ -19,24 +19,52 @@ using namespace std;
 
 /* 将结果写入mat文件 */
 /* 将数据写入mat文件中，用给定的名称命名 */
-bool SavePubmed(const string& matPath, 
+bool SavePubmed(const string& matPath,
-	const vector<string> &vTgName, 
+	const vector<string>& vTgName,
-	const vector<unordered_map<string, string> >& vumPaperTagVal)
+	vector<unordered_map<string, string> >& vumPaperTagVal)
 {
-	MATFile* pMatFile = matOpen(matPath.c_str(), "r"); //打开.mat文件
+	MATFile* pMatFile = matOpen(matPath.c_str(), "w"); //打开.mat文件
 	if (pMatFile == nullptr) {
-		cerr << "filePath is error! " << matPath << endl;
+		cout << "filePath is error! " << matPath << endl;
 		return false;
 	}
 
 	vector<const char*> vTgChars;
-	for (auto strTg : vTgName) {
+	for (auto& strTg : vTgName) {
 		vTgChars.push_back(strTg.c_str());
 	}
 
 	// 创建结构体数据
 	mxArray* mxStruct = mxCreateStructMatrix(1, 1, vTgName.size(), vTgChars.data());
+	// 创建cell matrix
+	unordered_map<string, mxArray*> ummxCellMtx;
+	for (auto & tgName : vTgName) {
+		ummxCellMtx[tgName] = mxCreateCellMatrix(1, vumPaperTagVal.size());
+	}
 
+	// 遍历每一篇文章
+	for (int i = 0; i < vumPaperTagVal.size(); ++i) {
+		auto& umTagVal = vumPaperTagVal[i];
+		// 遍历文章的每一个tag
+		for (auto& tgName : vTgName) {
+			mxArray* mxStr = mxCreateString(umTagVal[tgName].c_str());
+			mxArray* pMxArr = ummxCellMtx[tgName];
+			mxSetCell(pMxArr, i, mxStr);
+		}
+	}
+
+	// 将cell matrix赋值给struct matrix
+	for (auto& tgName : vTgName) {
+		mxArray* pMxArr = ummxCellMtx[tgName];
+		mxSetField(mxStruct, 0, tgName.c_str(), pMxArr);
+	}
+
+	// 将结构体写入mat，并命名为Tx
+	matPutVariable(pMatFile, "Tx", mxStruct);
+	// 将abstract信息写入mat，并命名为abs1
+	matPutVariable(pMatFile, "abs1", ummxCellMtx["AB"]);
+
+	matClose(pMatFile);
 
 	return true;
 }
@@ -129,7 +157,7 @@ void ProcessPubmedTxt(int argc, const char** argv) {
 		vumPaperTagVal.push_back(umTagContent);
 	}
 
-	cout << "文件个数：" << vumPaperTagVal.size() << endl;
+	// cout << "文件个数：" << vumPaperTagVal.size() << endl;
 
 	/* 去除没有摘要的文章 */
 	const string abstractTag = "AB";
@@ -147,7 +175,7 @@ void ProcessPubmedTxt(int argc, const char** argv) {
 	const string pmidTag = "PMID";
 	for (auto itr = vumPaperTagVal.begin(); itr != vumPaperTagVal.end(); ) {
 		if (umPMID.find((*itr)[pmidTag]) != umPMID.end()) {
-			cout << "duplicate " << (*itr)[pmidTag] << endl;
+			// out << "duplicate " << (*itr)[pmidTag] << endl;
 			itr = vumPaperTagVal.erase(itr);
 		}
 		else {
@@ -172,7 +200,7 @@ void ProcessPubmedTxt(int argc, const char** argv) {
 		testOfs << vumPaperTagVal[i][abstractTag] << endl;
 	}
 	testOfs.close();
-	cout << "文件个数：" << vumPaperTagVal.size() << endl;
+	// cout << "文件个数：" << vumPaperTagVal.size() << endl;
 	// for (auto num : vPaperStartIdx) cout << num << endl;
 	ifsPubmedTxt.close();
 

GMM/main.cpp

@@ -183,7 +183,7 @@ void ThreadProcessData(const ThreadParam& param) {
 int main(int argc, const char** argv) {
 
 	if (argc != 5) {
-		cerr << "This program should take 4 arguments(1.parrent Dir; 2. mat file suffix; 3. out mat filename; 4. thread number)!" << endl;
+		cout << "This program should take 4 arguments(1.parrent Dir; 2. mat file suffix; 3. out mat filename; 4. thread number)!" << endl;
 		return 1;
 	}
 	string parrentDir(argv[1]); // 知识颗粒的父目录名称