Changeset 597 for trunk/se/lu/thep/wenni

Timestamp:

Feb 28, 2008, 12:36:46 AM (16 years ago)

Author:

Jari Häkkinen

Message:

Improved memory usage of NNIFileConverter.

Location:

trunk/se/lu/thep/wenni

Files:

• bin/NNIFileConverter/NNIFileConverter.cc (modified) (3 diffs)
• lib/c++_tools/gslapi/vector.cc (modified) (1 diff)
• lib/c++_tools/gslapi/vector.h (modified) (4 diffs)

trunk/se/lu/thep/wenni/bin/NNIFileConverter/NNIFileConverter.cc

@@ -2 +2 @@
 
 /*
-  Copyright (C) 2005, 2006 Jari Häkkinen
+  Copyright (C) 2005, 2006, 2008 Jari Häkkinen
 
   This file is part of WeNNI,
@@ -28 +28 @@
 #include <cmath>
 #include <fstream>
+#include <string>
 
 #include <c++_tools/gslapi/matrix.h>
@@ -35 +36 @@
 {
   theplu::wenni::nnifileconverter::Parameter option(argc,argv);
-  std::ifstream is(option.fg1().c_str());
-  theplu::gslapi::matrix fg1(is);
-  is.close();
-  is.open(option.bgstd1().c_str());
-  theplu::gslapi::matrix bgstd1(is);
-  is.close();
-  is.open(option.fg2().c_str());
-  theplu::gslapi::matrix fg2(is);
-  is.close();
-  is.open(option.bgstd2().c_str());
-  theplu::gslapi::matrix bgstd2(is);
-  is.close();
+  std::ifstream is_f1(option.fg1().c_str());
+  std::ifstream is_f2(option.fg2().c_str());
+  std::ifstream is_s1(option.bgstd1().c_str());
+  std::ifstream is_s2(option.bgstd2().c_str());
+  std::ifstream* is_b1=NULL;
+  std::ifstream* is_b2=NULL;
+  if (option.datatype()==std::string("raw")) {
+    is_b1=new std::ifstream(option.bg1().c_str());
+    is_b2=new std::ifstream(option.bg2().c_str());
+  }
 
-  theplu::gslapi::matrix bg1;
-  theplu::gslapi::matrix bg2;
-  if (option.datatype()==std::string("raw")) {
-    is.open(option.bg1().c_str());
-    bg1=theplu::gslapi::matrix(is);
-    is.close();
-    is.open(option.bg2().c_str());
-    bg2=theplu::gslapi::matrix(is);
-    is.close();
-  }
-  else
-    bg1=bg2=theplu::gslapi::matrix(fg1.rows(),fg1.columns(),0.0);
-
-  // Calculating log2ratio
-  theplu::gslapi::matrix log2ratio(fg1.rows(),fg1.columns(),0.0);
-  double log2=log(2);
-  for (size_t i=0; i<log2ratio.rows(); i++) 
-    for (size_t j=0; j<log2ratio.columns(); j++)
-      if (fg1(i,j)>bg1(i,j) && fg2(i,j)>bg2(i,j))
-        log2ratio(i,j)=log( (fg1(i,j)-bg1(i,j)) / (fg2(i,j)-bg2(i,j)) ) / log2;
-
-  // calculating weights
-  theplu::gslapi::matrix weight(fg1.rows(),fg1.columns());
-  for (size_t i=0; i<weight.rows(); i++)
-    for (size_t j=0; j<weight.columns(); j++) {
-      using namespace theplu::wenni::weight;
-      weight(i,j)=weight_SNR2(get_snr(fg1(i,j),bg1(i,j),bgstd1(i,j)),
-                              get_snr(fg2(i,j),bg2(i,j),bgstd2(i,j)),
-                              option.beta());
+  std::ofstream os_r(option.logratio().c_str());
+  std::ofstream os_w(option.weight().c_str());
+  std::string f1str;
+  while (std::getline(is_f1,f1str)) {
+    std::string f2str;
+    std::string s1str;
+    std::string s2str;
+    std::getline(is_f2,f2str);
+    std::getline(is_s1,s1str);
+    std::getline(is_s2,s2str);
+    if (is_f1.fail() || is_f2.fail() ||is_s1.fail() ||is_s2.fail()) {
+      std::cerr << "Failed to read at least one file. Length problem?\n"
+                << "   (fg1, fg2, bgstd1, or bgstd2)" << std::endl;
+      exit(-1);
+    }
+    theplu::gslapi::vector f1(f1str);
+    theplu::gslapi::vector f2(f2str);
+    theplu::gslapi::vector s1(s1str);
+    theplu::gslapi::vector s2(s2str);
+    theplu::gslapi::vector b1(f1.size(),0.0);
+    theplu::gslapi::vector b2(f2.size(),0.0);
+    if (is_b1) {
+      std::string b1str;
+      std::getline(*is_b1,b1str);
+      if (is_b1->fail()) {
+        std::cerr << "Failed to read at bg1 file. Length problem?"
+                  << std::endl;
+        exit(-1);
+      }
+      b1=theplu::gslapi::vector(b1str);
+    }
+    if (is_b2) {
+      std::string b2str;
+      std::getline(*is_b2,b2str);
+      if (is_b2->fail()) {
+        std::cerr << "Failed to read at bg2 file. Length problem?"
+                  << std::endl;
+        exit(-1);
+      }
+      b2=theplu::gslapi::vector(b2str);
     }
 
-  std::ofstream os(option.logratio().c_str());
-  os << log2ratio;
-  os.close();
-  os.open(option.weight().c_str());
-  os << weight;
-  os.close();
+    // Calculating log2ratio
+    theplu::gslapi::vector log2ratio(f1.size());
+    double log2=log(2);
+    for (size_t i=0; i<log2ratio.size(); i++) 
+      if (f1(i)>b1(i) && f2(i)>b2(i))
+        log2ratio(i)=log( (f1(i)-b1(i)) / (f2(i)-b2(i)) ) / log2;
+
+    // calculating weights
+    theplu::gslapi::vector weight(f1.size());
+    for (size_t i=0; i<weight.size(); i++) {
+      using namespace theplu::wenni::weight;
+      weight(i)=weight_SNR2(get_snr(f1(i),b1(i),s1(i)),
+                            get_snr(f2(i),b2(i),s2(i)),option.beta());
+    }
+
+    os_r << log2ratio << '\n';
+    os_w << weight << '\n';
+  }
+
+  is_f1.close();
+  is_s1.close();
+  is_f2.close();
+  is_s2.close();
+  if (is_b1) {
+    is_b1->close();
+    delete is_b1;
+  }
+  if (is_b2) {
+    is_b2->close();
+    delete is_b2;
+  }
+  os_r.close();
+  os_w.close();
 
   return 0;

trunk/se/lu/thep/wenni/lib/c++_tools/gslapi/vector.cc

@@ -165 +165 @@
 
 
+  vector::vector(std::string& line, char sep)
+    throw (utility::IO_error,std::exception)
+    : view_(NULL), const_view_(NULL)
+  {
+    // Empty line
+    if (!line.size())
+      return;
+
+    std::vector<double> v;
+    v.reserve(line.length()/2);
+    std::string element;
+    std::stringstream ss(line);
+    bool ok=true;
+    while(ok) {
+      if(sep=='\0')
+        ok=(ss>>element);
+      else 
+        ok=getline(ss, element, sep);
+      if(!ok)
+        break;
+
+      if(utility::is_double(element)) {
+        v.push_back(atof(element.c_str()));
+      }
+      else if (!element.size() || utility::is_nan(element)) {
+        v.push_back(std::numeric_limits<double>::quiet_NaN());
+      }
+    }
+    if (sep!='\0' && line[line.size()-1]==sep) // add NaN for final separator
+      v.push_back(std::numeric_limits<double>::quiet_NaN());
+
+    // convert the data to a gsl vector
+    v_ = gsl_vector_alloc(v.size());
+    size_t n=0;
+    for (size_t i=0; i<v.size(); i++)
+      gsl_vector_set(v_, n++, v[i]);
+  }
+
+
+
   vector::~vector(void)
   {

trunk/se/lu/thep/wenni/lib/c++_tools/gslapi/vector.h

@@ +1 @@
+#ifndef _theplu_gslapi_vector_
+#define _theplu_gslapi_vector_
+
 // $Id$
 
@@ -5 +8 @@
   Copyright (C) 2004 Jari Häkkinen, Peter Johansson
   Copyright (C) 2005 Jari Häkkinen, Peter Johansson, Markus Ringnér
-  Copyright (C) 2006 Jari Häkkinen
+  Copyright (C) 2006, 2008 Jari Häkkinen
 
   This file is part of the thep c++ tools library,
@@ -26 +29 @@
 */
 
-#ifndef _theplu_gslapi_vector_
-#define _theplu_gslapi_vector_
-
 #include <c++_tools/utility/Exception.h>
 
 #include <iostream>
+#include <string>
+#include <utility>
 #include <vector>
-#include <utility>
 
 #include <gsl/gsl_vector.h>
@@ -179 +180 @@
     ///
     explicit vector(std::istream &, char sep='\0') throw (utility::IO_error,std::exception);
+
+    ///
+    /// The string constructor.
+    ///
+    /// Either elements should be separated 
+    /// with white space characters (default), or elements should be separated
+    /// by the delimiter \a sep. When delimiter \a sep is used empty elements
+    /// are stored as NaN's (except that empty lines are ignored). The
+    /// end of input to the vector is at end of file marker.
+    ///
+    explicit vector(std::string &, char sep='\0') throw (utility::IO_error,std::exception);