@@ -8,21 +8,13 @@

 // [[Rcpp::export]]

 Rcpp::List cogapsFromFile_cpp(const std::string D)

 {

-    CsvParser csv(D);

-/*

-    while (csv.hasNext())

-    {

-        MatrixElement m = csv.getNext();

-        Rcpp::Rcout << m.row << "," << m.col << "," << m.value << '\n';

-    }

+    // CsvParser csv(D);

+    //Rcpp::Rcout << csv.nRow() << "," << csv.nCol() << '\n';

-    for (unsigned i = 0; i < csv.mRowNames.size(); ++i)

-        Rcpp::Rcout << csv.mRowNames[i] << '\n';

-

-    for (unsigned i = 0; i < csv.mColNames.size(); ++i)

-        Rcpp::Rcout << csv.mColNames[i] << '\n';

-*/

-    Rcpp::Rcout << csv.nRow() << "," << csv.nCol() << '\n';

+    //RowMatrix mat(D);

+    RowMatrix mat(D);

+    Rcpp::Rcout << mat.nRow() << "," << mat.nCol() << '\n';

+    //Rcpp::Rcout << gaps::algo::sum(mat) << '\n';

 }

 // [[Rcpp::export]]

@@ -1,5 +1,6 @@

 #include "Matrix.h"

 #include "../file_parser/CsvParser.h"

+#include "../file_parser/MatrixElement.h"

 template<class GenericMatrix>

 static Rcpp::NumericMatrix convertToRMatrix(const GenericMatrix &mat)

@@ -53,15 +54,24 @@ RowMatrix::RowMatrix(const Rcpp::NumericMatrix &rmat)

 RowMatrix::RowMatrix(const std::string &path)

 {

-    CsvParser csv(path);

-

-    while (csv.hasNextRow())

+    // get matrix dimensions

+    MatrixDimensions dim(CsvParser::getDimensions(path));

+    mNumRows = dim.nRow;

+    mNumCols = dim.nCol;

+        

+    // allocate matrix

+    for (unsigned i = 0; i < mNumRows; ++i)

     {

-        mRows.push_back(Vector(csv.getNextRow()));

+        mRows.push_back(Vector(mNumCols));

     }

-    mNumRows = mRows.size();

-    mNumCols = mRows[0].size();

+    // populate matrix

+    CsvParser csv(path);

+    while (csv.hasNext())

+    {

+        MatrixElement e(csv.getNext());

+        this->operator(e.row, e.col) = e.value;

+    }

 }

 void RowMatrix::operator=(const RowMatrix &mat)

@@ -1,5 +1,13 @@

 #include "Vector.h"

+Vector::Vector(const std::vector<float> &v) : mValues(v.size())

+{

+    for (unsigned i = 0; i < v.size(); ++i)

+    {

+        mValues[i] = v[i];

+    }

+}

+

 void Vector::concat(const Vector& vec)

 {

     mValues.insert(mValues.end(), vec.mValues.begin(), vec.mValues.end());

@@ -18,7 +18,7 @@ private:

 public:

     Vector(unsigned size) : mValues(aligned_vector(size, 0.f)) {}

-    Vector(std::vector<float> v) : mValues(aligned_vector(v)) {}

+    Vector(const std::vector<float> &v);

     const float* ptr() const {return &mValues[0];}

     float* ptr() {return &mValues[0];}

@@ -1,78 +1,77 @@

 #include "CsvParser.h"

+#include "../math/Algorithms.h"

 #include <iostream>

-// read through whole file once, store row/col names - gives dimensions

-// open file, read column names

-CsvParser::CsvParser(const std::string &path) : mCurrentRow(0), mCurrentCol(0)

+// get the number of rows and cols in a csv file

+MatrixDimension CsvParser::getDimensions(const std::string &path)

 {

-    mFile.open(path.c_str());

+    // initialize struct that holds dimensions

+    MatrixDimension dim(0,0);

+

+    // open file stream

+    std::ifstream str(path);

+    // read first entry (blank)

     std::string line;

-    std::getline(mFile, line, ','); // read first entry (blank)

+    std::getline(mFile, line, ',');

+    // get col size

     std::size_t pos;

-    std::getline(mFile, line, ',');

-    while ((pos = line.find('\n')) == std::string::npos)

+    do

     {

-        mColNames.push_back(line);

         std::getline(mFile, line, ',');

+        dim.nCol++;

     }

-    mColNames.push_back(line.substr(0,pos));

+    while ((pos = line.find('\n')) == std::string::npos);

-    mRowNames.push_back(line.substr(pos+1));

+    // get row size

+    dim.nRow++; // acount for current row

     while (mFile.peek() != EOF)

     {

-        while ((pos = line.find('\n')) == std::string::npos)

+        // throw away data

+        do

         {

             std::getline(mFile, line, ',');

         }

-        mRowNames.push_back(line.substr(pos+1));

-    }

+        while ((pos = line.find('\n')) == std::string::npos);

-    for (unsigned i = 0; i < mRowNames.size(); ++i)

-    {

-        std::cout << mRowNames[i] << '\n';

-    }

-    for (unsigned j = 0; j < mColNames.size(); ++j)

-    {

-        std::cout << mRowNames[j] << '\n';

+        // increment row number, ignore last newline in file

+        if (pos + 1 < line.size())

+        {

+            dim.nRow++;

+        }

     }

+    return dim;

 }

-bool hasNextRow()

-{

-

-}

-

-std::vector<float> getNextRow()

+// read through whole file once, store row/col names - gives dimensions

+// open file, read column names

+CsvParser::CsvParser(const std::string &path) : mCurrentRow(0), mCurrentCol(0)

 {

-

+    mFile.open(path);

 }

-void skipNextRow()

+bool CsvParser::hasNext()

 {

-

+    return mFile.peek() != EOF;

 }

-/*MatrixElement CsvParser::getNext()

+MatrixElement CsvParser::getNext()

 {

     std::string line;

-    std::getline(mFile, line, ',');

-

     std::size_t pos;

-    if ((pos = line.find('\n')) != std::string::npos)

+    std::getline(mFile, line, ',');

+    if ((pos = line.find('\n')) != std::string::npos) // end of line

     {

-        if (pos + 1 < line.size())

-        {

-            mRowNames.push_back(line.substr(pos + 1));

-        }

-        unsigned col = mCurrentCol;

-        mCurrentCol = 0;

-        return MatrixElement(mCurrentRow++, col, line.substr(0, pos));

+        return MatrixElement(mCurrentRow, mCurrentCol, line.substr(0, pos));

     }

-    else

+    else if (std::isdigit(line[0])) // data

     {

-        return MatrixElement(mCurrentRow, mCurrentCol++, line);

+        return MatrixElement(mCurrentRow, mCurrentCol, line);

     }

-}*/

\ No newline at end of file

+    else // row/col name

+    {

+        return getNext();

+    }

+}

@@ -1,6 +1,8 @@

 #ifndef __COGAPS_CSV_PARSER_H__

 #define __COGAPS_CSV_PARSER_H__

+#include "MatrixElement.h"

+

 #include <fstream>

 #include <vector>

 #include <string>

@@ -11,27 +13,17 @@ private:

     std::ifstream mFile;

-    std::vector<std::string> mRowNames;

-    std::vector<std::string> mColNames;

-

     unsigned mCurrentRow;

     unsigned mCurrentCol;

 public:

-    // read through whole file once, store row/col names - gives dimensions

     CsvParser(const std::string &path);

-    unsigned nRow() const { return mRowNames.size(); }

-    unsigned nCol() const { return mColNames.size(); }

-

-    bool hasNextRow();

-    std::vector<float> getNextRow(); 

-    void skipNextRow();

+    bool hasNext();

+    MatrixElement getNext();

-    bool hasNextCol();

-    std::vector<float> getNextCol();

-    void skipNextCol();

+    static MatrixDimension getDimensions(const std::string &path);

 };

 #endif

\ No newline at end of file

new file mode 100644

@@ -0,0 +1,33 @@

+#ifndef __COGAPS_MATRIX_ELEMENT_H__

+#define __COGAPS_MATRIX_ELEMENT_H__

+

+#include <sstream>

+#include <string>

+

+struct MatrixElement

+{

+    unsigned row;

+    unsigned col;

+    float value;

+

+    MatrixElement(unsigned r, unsigned c, float v)

+        : row(r), col(c), value(v)

+    {}

+

+    MatrixElement(unsigned r, unsigned c, const std::string &s)

+        : row(r), col(c), value(0.f)

+    {

+        std::stringstream ss(s);

+        ss >> value;

+    }

+};

+

+struct MatrixDimension

+{

+    unsigned nRow;

+    unsigned nCol;

+

+    MatrixDimension(unsigned nr, unsigned nc) : nRow(nr), nCol(nc) {}

+};

+

+#endif

\ No newline at end of file

@@ -3,6 +3,16 @@

 #include "SIMD.h"

 #include <algorithm>

+#include <string>

+#include <sstream>

+

+float gaps::algo::stringToFloat(const std::string &s)

+{

+    std::stringstream ss(s);

+    float f;

+    ss >> f;

+    return f;

+}

 float gaps::algo::sum(const Vector &vec)

 {

@@ -29,6 +29,8 @@ namespace algo

     const float epsilon = 1.0e-10f;

     const float pi = 3.14159265358979323846264f;

+    float stringToFloat(const std::string &s);

+

     bool isVectorZero(const float *vec, unsigned size);

     // vector algorithms