fn_load_dataset_from_file.hpp

Go to the documentation of this file.

// Copyright (C) 2011 the authors listed below
// http://ecocpak.sourceforge.net
// 
// Authors:
// - Dimitrios Bouzas (bouzas at ieee dot org)
// - Nikolaos Arvanitopoulos (niarvani at ieee dot org)
// - Anastasios Tefas (tefas at aiia dot csd dot auth dot gr)
// 
// This file is part of the ECOC PAK C++ library. It is 
// provided without any warranty of fitness for any purpose.
//
// You can redistribute this file and/or modify it under 
// the terms of the GNU Lesser General Public License (LGPL) 
// as published by the Free Software Foundation, either 
// version 3 of the License or (at your option) any later 
// version.
// (see http://www.opensource.org/licenses for more info)





void 
load_dataset_from_file
  (
  const char* filename,
  mat& samples,
  icolvec& labels,
  u32& n_classes
  )
  {
  // open file
  FILE *fp = fopen(filename, "r");

  // in case the file opening failed exit with error message
  if(fp == 0) 
    {
    #ifdef NCURSES_OUTPUT
    endwin();
    #endif
    
    fprintf(stderr, "error: can't open input file %s\n\n", filename);
    
    #ifdef ECOCPAK_PROGRAM
    exit_with_help();
    #endif
    
    exit(1);
    }
  
  // temporary attributes
  u32 tmp_attr = 0;
  
  // final attributes
  u32 attr = 0;

  // number of samples in dataset
  u32 n_samples = 0;
  
  // number of elements in dataset
  u32 elements = 0;
  
  // initialize number of classes
  n_classes = 0;

  // temporary vector for parsing the number of dataset's attributes
  vector<int> tmp_vec;

  while(true) 
    {
    int c = fgetc(fp);
    switch(c) 
      {
      case '\n':
        {
        ++n_samples;
          
        tmp_attr = 0;
        // fall through,
        // count the '-1' element
        break;
        }
      case ':':
        {       
        int i = 1;
        int j = tmp_vec.size() - 1;
        tmp_attr = 0;

                                // until we find a space character
        while(!isspace(tmp_vec[j]))
          {
          // compute the number of the current attribute
          tmp_attr += ((tmp_vec[j] - 48) * i);
          i *= 10;
          j--;
          }

                                // update the number of attributes
        if(tmp_attr > attr)
          {
          attr = tmp_attr;
          }
        
        // clear temporary vector
        tmp_vec.clear();
        
        // increase the number of elements
        ++elements;
        break;
        }
      
      // end of file reached
      case EOF:
        {
        goto out;
        }
        
      default:
        {
        tmp_vec.push_back(c);
        }
        
      }
      
   }

  out: //
  
  // rewind file
  rewind(fp);

  attr;
  labels.zeros(n_samples);
  samples.zeros(n_samples, attr);
  
  vector<int> distinct_labels;

  for(u32 i = 0; i < n_samples; i++) 
    {
    // read label
    fscanf(fp, "%d", &labels[i]);

    // if distinct_labels vector is empty insert its first element
    if(distinct_labels.empty() == true)
      {
      distinct_labels.push_back(labels[i]);
      n_classes++;
      }
    else
      {
      bool found = false;       
      
      // check wether current label has been stored in distinct_labels vector
      for(u32 j = 0; j < distinct_labels.size(); j++)
        {
        if(distinct_labels[j] == labels[i])
          {
          found = true;
          }
          
        }

      // curent label is seen for first time store it and increase classes number
      if(found == false)
        {
        distinct_labels.push_back(labels[i]);
        n_classes++;
        }
        
      }
    
    // read rest of line which contains the sample vector
    while(true)
      {
      int c;
      do 
        {
        c = getc(fp);
        if(c == '\n') 
          {
          // line end reached go to next line
          goto out2;
          }
        } 
      while(isspace(c));

      ungetc(c, fp);

      u32 cur_index;
      double cur_value;
      if(fscanf(fp, "%d:%lf", &cur_index, &cur_value) < 2 ) 
        {
        #ifdef NCURSES_OUTPUT
        endwin();
        #endif
        
        fprintf(stderr, "Wrong input format at line %d\n", i + 1);
        exit(1);
        }

      // Copy value to dataMat
      samples(i, cur_index - 1) = cur_value;
      
      }

    out2:;
    }
  
  fclose(fp);
  }
 
 
  
void 
load_dataset_from_file
  (
  const char* filename,
  mat& samples,
  icolvec& labels
  )
  {
  u32 n_classes = 0;
  
  // open file
  FILE *fp = fopen(filename, "r");

  // in case the file opening failed exit with error message
  if(fp == 0) 
    {
    #ifdef NCURSES_OUTPUT
    endwin();
    #endif
    
    fprintf(stderr, "error: can't open input file %s\n\n", filename);
    
    #ifdef ECOCPAK_PROGRAM
    exit_with_help();
    #endif
    
    exit(1);
    }
  
  // temporary attributes
  u32 tmp_attr = 0;
  
  // final attributes
  u32 attr = 0;

  // number of samples in dataset
  u32 n_samples = 0;
  
  // number of elements in dataset
  u32 elements = 0;
  
  // initialize number of classes
  n_classes = 0;

  // temporary vector for parsing the number of dataset's attributes
  vector<int> tmp_vec;

  while(true) 
    {
    int c = fgetc(fp);
    switch(c) 
      {
      case '\n':
        {
        ++n_samples;
          
        tmp_attr = 0;
        // fall through,
        // count the '-1' element
        break;
        }
      case ':':
        {       
        int i = 1;
        int j = tmp_vec.size() - 1;
        tmp_attr = 0;

                                // until we find a space character
        while(!isspace(tmp_vec[j]))
          {
          // compute the number of the current attribute
          tmp_attr += ((tmp_vec[j] - 48) * i);
          i *= 10;
          j--;
          }

                                // update the number of attributes
        if(tmp_attr > attr)
          {
          attr = tmp_attr;
          }
        
        // clear temporary vector
        tmp_vec.clear();
        
        // increase the number of elements
        ++elements;
        break;
        }
      
      // end of file reached
      case EOF:
        {
        goto out;
        }
        
      default:
        {
        tmp_vec.push_back(c);
        }
        
      }
      
   }

  out: //
  
  // rewind file
  rewind(fp);

  attr;
  labels.zeros(n_samples);
  samples.zeros(n_samples, attr);
  
  vector<int> distinct_labels;

  for(u32 i = 0; i < n_samples; i++) 
    {
    // read label
    fscanf(fp, "%d", &labels[i]);

    // if distinct_labels vector is empty insert its first element
    if(distinct_labels.empty() == true)
      {
      distinct_labels.push_back(labels[i]);
      n_classes++;
      }
    else
      {
      bool found = false;       
      
      // check wether current label has been stored in distinct_labels vector
      for(u32 j = 0; j < distinct_labels.size(); j++)
        {
        if(distinct_labels[j] == labels[i])
          {
          found = true;
          }
          
        }

      // curent label is seen for first time store it and increase classes number
      if(found == false)
        {
        distinct_labels.push_back(labels[i]);
        n_classes++;
        }
        
      }
    
    // read rest of line which contains the sample vector
    while(true)
      {
      int c;
      do 
        {
        c = getc(fp);
        if(c == '\n') 
          {
          // line end reached go to next line
          goto out2;
          }
        } 
      while(isspace(c));

      ungetc(c, fp);

      u32 cur_index;
      double cur_value;
      if(fscanf(fp, "%d:%lf", &cur_index, &cur_value) < 2 ) 
        {
        #ifdef NCURSES_OUTPUT
        endwin();
        #endif
        
        fprintf(stderr, "Wrong input format at line %d\n", i + 1);
        exit(1);
        }

      // Copy value to dataMat
      samples(i, cur_index - 1) = cur_value;
      
      }

    out2:;
    }
  
  fclose(fp);
  }