Changeset 5a84ee for util/src/average.cpp

Timestamp:

Apr 15, 2010, 12:34:20 PM (16 years ago)

Author:

Frederik Heber <heber@…>

Children:

e34098

Parents:

35bab2

Message:

Rewrite of average.cpp and added first UnitTest.

• average.cpp is now called averager.cpp and uses a function AverageColumns().
• new function AverageColumns(): takes an input streamd and a set of column indices for which averages shall be computed
• Introduced unit tests to Utils part of ESPACK:
  • new unit test AverageColumnsUnitTest testing the above.
  • Makefile generation introduced into configure.ac
  • subdir forked in Makefile.am

File:

• util/src/average.cpp (modified) (1 diff)

util/src/average.cpp

@@ +1 @@
+/*
+ * average.cpp
+ *
+ *  Created on: Apr 15, 2010
+ *      Author: heber
+ */
+
 using namespace std;
+
 #include <iostream>
 #include <iomanip>
-#include <fstream>
+#include <string>
 #include <sstream>
-#include <math.h>
-#include <string>
 
-#include "version.h"
+#include "average.hpp"
 
-int main(int argc, char **argv) {
+/** Takes the average over all entries in a list of columns per column.
+ * Note that point in ifstream \a data is returned to its position on beginning.
+ * \param data input stream with entries, columns are white-space separated, rows by new-line.
+ * \param Columns set of indices (zero-based) for which columns to compute the average,
+ *        note that Columns is copied to allow for removal of faulty column indices.
+ * \return pointer to map with column index as key and tuple (mean,std deviation) as value
+ */
+MeanErrorMap *AverageColumns(istream &data, IndexSet Columns)
+{
+  MeanErrorMap *Values = new MeanErrorMap;
+  map<int, int> CountMap;
+  double tmp = 0.;
+  int cols = 0;
+  int lines = 0;
+  string zeile;
+  stringstream line;
 
-        double avg, dev;
-        double tmp;
-        int zaehler;
-  int col;
-        int flag;
-        int skiplines=0;
-        string zeile;
-        stringstream line;
-        
-        cout << ESPACKVersion << endl;
+  // store position of ifstream
+  size_t position = data.tellg();
 
-        if (argc <= 2) {
-                cout << "Usage: " << argv[0] << " <column (zero-based)> <data file> [#skip initial lines]\n";
-                return(1);
-        }
-        {
-    line.str(argv[1]);
-    line >> col;
-    cout << "Using column " << col << "." << endl;
-    line.clear();
-        }
-        if (argc > 3) {
-    line.str(argv[3]);
-    line >> skiplines;
-    cout << "Will skip first " << skiplines << " lines." << endl;
-    line.clear();
-        }
+  // set initial values in maps to zero
+  for (IndexSet::iterator ColRunner = Columns.begin(); ColRunner != Columns.end(); ++ColRunner) {
+    Values->insert( pair<int, pair<double, double> > (*ColRunner, pair<double, double> (0., 0.) ) );
+    CountMap.insert( pair<int, int> (0, 0) );
+  }
 
-        // get average
-        avg = 0.;
-        zaehler=0;
-        ifstream test(argv[2]);
-        if (test == NULL) { cout << "Can't open File " << argv[2] << "\n"; return(255); }
-        flag=1;
-        while (getline(test, zeile, '\n')) {
-          if (skiplines) {
-            skiplines--;
-            continue;
-          }
+  /// The average is taken by going through each line, scanning the desired column, adding up
+  /// and keeping count of the number of summands.
+  lines = 0;
+  while (getline(data, zeile, '\n')) {
+    // get next line
     line.clear();
     line.str(zeile);
-    for (int j=0;j<=col;j++)
-      if (!line.eof()) {
-        line >> ws >> tmp;
+    lines++;
+    // go through the columns
+    cols = -1;
+    IndexSet::const_iterator Eraser = Columns.end();
+    for (IndexSet::const_iterator ColRunner = Columns.begin(); ColRunner != Columns.end(); ++ColRunner) {
+      // delete earlier column if necessary
+      if (Eraser != Columns.end()) {
+        Columns.erase(Eraser);
+        Eraser = Columns.end();
+      }
+      // skip to next desired column
+      for(;cols!=*ColRunner;++cols)
+        if (!line.eof()) // check for end of line
+          line >> ws >> tmp;
+        else
+          break;
+      if (cols == *ColRunner) { // if end of line has not been reached
+        (*Values)[*ColRunner].first += tmp;
+        ++CountMap[*ColRunner];
       } else {
-        flag=0;
+        cerr << "Not enough columns in line " << lines << "." << endl;
+        Eraser = ColRunner;
+        break;
       }
-    if (flag) {
-      avg += tmp;
-      zaehler++;
     }
-        }
-        if (!flag) {
-    cerr << "File does not contain " << col << " column(s)." << endl;
-    exit(1);
-        } else
-          if (zaehler != 0) avg /= zaehler;
-        // goto to beginning again
-        test.clear();
-  test.seekg(ios::beg);
+    if (Eraser != Columns.end()) {
+      Columns.erase(Eraser);
+      Eraser = Columns.end();
+    }
+  }
 
-        // get deviation
-        dev = 0.;
-        zaehler=0;
-        flag=1;
-  while (getline(test, zeile, '\n')) {
-    if (skiplines) {
-      skiplines--;
-      continue;
+  // go through each value in Results and take average
+  {
+    MeanErrorMap::iterator Eraser = Values->end();
+    for (MeanErrorMap::iterator Runner = Values->begin(); Runner != Values->end(); ++Runner) {
+      if(Eraser != Values->end()) {
+        Values->erase(Eraser);
+        Eraser = Values->end();
+      }
+      if (CountMap[Runner->first] != 0)
+        Runner->second.first /= CountMap[Runner->first];
+      else {
+        cerr << "For column " << CountMap[Runner->first] << " no entries have been found." << endl;
+        Eraser = Runner;
+      }
     }
+    if(Eraser != Values->end()) {
+      Values->erase(Eraser);
+      Eraser = Values->end();
+    }
+  }
+
+  // goto to beginning again for second sweep
+  data.clear();
+  data.seekg(position);
+
+  /// The average is taken by going through each line, scanning the desired column, adding up
+  /// and keeping count of the number of summands.
+  lines = 0;
+  while (getline(data, zeile, '\n')) {
+    // get next line
     line.clear();
     line.str(zeile);
-    for (int j=0;j<=col;j++)
-      line >> ws >> tmp;
-    if (!line.eof()) {
-      dev += (tmp - avg)*(tmp - avg);
-      zaehler++;
+    lines++;
+    // go through the columns
+    cols = 0;
+    for (IndexSet::const_iterator ColRunner = Columns.begin(); ColRunner != Columns.end(); ++ColRunner) {
+      // skip to next desired column
+      for(;cols!=*ColRunner;++cols)
+        if (!line.eof()) // check for end of line
+          line >> ws >> tmp;
+        else
+          break;
+      if (cols == *ColRunner) { // if end of line has not been reached
+        (*Values)[*ColRunner].second += (tmp - (*Values)[*ColRunner].first)*(tmp - (*Values)[*ColRunner].first);
+      } else {
+        cerr << "Not enough columns in line " << lines << "." << endl;
+        break;
+      }
     }
   }
-        test.close();
-        if (zaehler != 0) dev /= zaehler;
-                        
-        cout << setprecision(8) << avg << "\t" << setprecision(8) << sqrt(dev) << "\n";
-        return(0);
-}
+
+  // go through each value in Results and take std deviation
+  for (MeanErrorMap::iterator Runner = Values->begin(); Runner != Values->end(); ++Runner)
+    if (CountMap[Runner->first] != 0)
+      Runner->second.second /= CountMap[Runner->first];
+    else
+      cerr << "For column " << CountMap[Runner->first] << " no entries have been found." << endl;
+
+  // go back to initial pointer of data
+  data.clear();
+  data.seekg(position);
+
+  return Values;
+};