Sparse matrix fix (#42)

* Fix for sparse matrix clustering. Had to remove the sim = true flag, and other values were not being set properly

* Fix for column reader reading the last line twice.

src/Adapters/DistanceFileReader.h

@@ -18,7 +18,7 @@ class DistanceFileReader {
     // We need to deduce type, the easy way to do that is to see if there is a number for the first item read.
     // Phylip files have a number of sequences located at the top. We can use that to our advantage.
     virtual bool Read(const std::string& filePath) {return false;}
-    DistanceFileReader(SparseDistanceMatrix*, ListVector*);
+    DistanceFileReader(SparseDistanceMatrix*, ListVector*, bool);
     DistanceFileReader() = default;
     // Phylip files do not need a count table
     SparseDistanceMatrix* GetSparseMatrix() const {return new SparseDistanceMatrix(*sparseMatrix);}

src/ColumnDistanceMatrixReader.cpp

@@ -41,19 +41,15 @@ bool ColumnDistanceMatrixReader::Read(const std::string& filePath) {
 	fileHandle >> firstName;
 	fileHandle >> secondName;
 	fileHandle >> dist;
+
 	if(nameToIndexMap.find(firstName) != nameToIndexMap.end() ||
 			nameToIndexMap.find(secondName) != nameToIndexMap.end()) {
 			fileHandle.clear();
 			fileHandle.seekg(0, std::ifstream::beg);
 	}
-	while(fileHandle && lt == 1){  //let's assume it's a triangular matrix...
-
-		fileHandle >> firstName;
-        fileHandle >> secondName;
-        fileHandle >> distance;	// get the row and column names and distance
+	while(fileHandle >> firstName >> secondName >> distance && lt == 1){  //let's assume it's a triangular matrix...
         int itA;
 		int itB;
-
 		try {
 			itA = nameToIndexMap.at(firstName);
 			itB = nameToIndexMap.at(secondName);
@@ -68,7 +64,6 @@ bool ColumnDistanceMatrixReader::Read(const std::string& filePath) {
 			}
 			util.CheckForDistanceFileError(container);
 		}
-        // std::map<std::string,int>::iterator itB = nameMap->find(secondName);
 
 		if (util.isEqual(distance, -1)) { distance = 1000000; }
 		else if (sim) { distance = 1 - distance;  }  //user has entered a sim matrix that we need to convert.
@@ -108,19 +103,14 @@ bool ColumnDistanceMatrixReader::Read(const std::string& filePath) {
 		}
 	}
 
-	if(lt == 0){  // oops, it was square
+	if(lt == 0) {  // oops, it was square
 		fileHandle.close();  //let's start over
 		sparseMatrix->clear();  //let's start over
 		fileHandle.open(filePath); //let's start over
 
-		while(fileHandle){
-			fileHandle >> firstName;
-            fileHandle >> secondName;
-            fileHandle >> distance;	// get the row and column names and distance
-
+		while(fileHandle >> firstName >> secondName >> distance){
 			int itA;
 			int itB;
-
 			try {
 				itA = nameToIndexMap.at(firstName);
 				itB = nameToIndexMap.at(secondName);
@@ -145,7 +135,6 @@ bool ColumnDistanceMatrixReader::Read(const std::string& filePath) {
 			}
 		}
 	}
-
 	fileHandle.close();
 	list->setLabel("0");
 	return true;

src/DistanceFileReader.cpp

@@ -6,7 +6,7 @@
 
 
 DistanceFileReader::DistanceFileReader(SparseDistanceMatrix *sparseDistanceMatrix,
-    ListVector *listVector):sparseMatrix(sparseDistanceMatrix), list(listVector) {}
+    ListVector *listVector, const bool isSim):sparseMatrix(sparseDistanceMatrix), list(listVector), sim(isSim){}
 
 Rcpp::DataFrame DistanceFileReader::SparseMatrixToDataFrame() const {
     const size_t size = sparseMatrix->seqVec.size();

src/MatrixAdapter.cpp

@@ -83,17 +83,18 @@ SparseDistanceMatrix MatrixAdapter::CreateSparseMatrix() {
         const Utils util;
         util.CheckForDistanceFileError(unknownNames);
     }
-
+    matrixNames = std::vector<std::string>(nameSize);
     for (int i = 0; i < nameSize; i++) {
-        positionsToNames[xPosition[i]] = countTable.GetNameByIndex(i); // Not going to work, I need a way to link my names to the sparse matix indices
+        positionsToNames[i] = countTable.GetNameByIndex(i);// Not going to work, I need a way to link my names to the sparse matix indices
+        matrixNames[i] = positionsToNames[i];
     }
 
 
-    matrixNames = std::vector<std::string>(nameSize);
-    for (int i = 0; i < nameSize; i++) {
-        positionsOfIndexs[xPosition[i]] = i;
-        matrixNames[i] = positionsToNames[xPosition[i]];
-    }
+
+    // for (int i = 0; i < nameSize; i++) {
+    //     positionsOfIndexs[xPosition[i]] = i; // Position of indexes is incorrectly made
+    //     matrixNames[i] = positionsToNames[i];
+    // }
 
 
     for (int i = 0; i < nSeqs;  i++) {
@@ -103,8 +104,8 @@ SparseDistanceMatrix MatrixAdapter::CreateSparseMatrix() {
         if(currentDist < 0) {
             currentDist = 0;
         }
-        const int xIndex = positionsOfIndexs[xPosition[i]]; // Coming from r -> c++, indeces start at 1 in r
-        const int yIndex = positionsOfIndexs[yPosition[i]];
+        const int xIndex = xPosition[i]; // Coming from r -> c++, indeces start at 1 in r
+        const int yIndex = yPosition[i];
 
         // const double currentValueX = dataList[yIndex].rowValues[xIndex];
         // if(currentValueX != 0){ // We already set the value and this is a sparse matrix.

src/SparseMatrixTestFixture.cpp

@@ -40,10 +40,10 @@ bool SparseMatrixTestFixture::TestRemoveCell(const unsigned long row, const unsi
     const bool expectedResult) {
     Setup();
     unsigned long currentSize = 0;
-    if(row < sparseDistanceMatrix->seqVec.max_size())
+    if(row < sparseDistanceMatrix->seqVec.size())
        currentSize = sparseDistanceMatrix->seqVec[row].size();
     sparseDistanceMatrix->rmCell(row, col);
-    const bool result = sparseDistanceMatrix->seqVec[row].size() < currentSize;
+    const bool result = sparseDistanceMatrix->seqVec[row].size() == currentSize - 1;
     TearDown();
     return result == expectedResult;
 }

src/main.cpp

@@ -91,7 +91,7 @@ SEXP ProcessSparseMatrix(const std::vector<int> &xPosition,
     countTableAdapter.CreateDataFrameMap(countTable);
     MatrixAdapter adapter(xPosition, yPosition, data, cutoff, isSim, countTableAdapter);
     auto* read = new DistanceFileReader(new SparseDistanceMatrix(adapter.CreateSparseMatrix()),
-        new ListVector(adapter.CreateListVector()));
+        new ListVector(adapter.CreateListVector()), isSim);
     read->CreateCountTableAdapter(countTable);
     return Rcpp::XPtr<DistanceFileReader>(read);
 }

src/test-sparse_matrix.cpp

@@ -60,7 +60,7 @@ context("Spase Distance Matrix") {
     }
      test_that("RmCells remove cells correctly") {
         SparseMatrixTestFixture fixture;
-        bool result = fixture.TestRemoveCell(1,1,true);
+        bool result = fixture.TestRemoveCell(1,0,true);
         expect_true(result);
     }
     test_that("Test Adding Cells sorted works") {