---

SprIOTestApp.cc

Go to the documentation of this file.

//$Id: SprIOTestApp.cc,v 1.3 2007/12/01 01:29:41 narsky Exp $

#include "PhysicsTools/StatPatternRecognition/interface/SprExperiment.hh"
#include "PhysicsTools/StatPatternRecognition/interface/SprAbsFilter.hh"
#include "PhysicsTools/StatPatternRecognition/interface/SprData.hh"
#include "PhysicsTools/StatPatternRecognition/interface/SprAbsReader.hh"
#include "PhysicsTools/StatPatternRecognition/interface/SprAbsWriter.hh"
#include "PhysicsTools/StatPatternRecognition/interface/SprDataFeeder.hh"
#include "PhysicsTools/StatPatternRecognition/interface/SprRWFactory.hh"
#include "PhysicsTools/StatPatternRecognition/interface/SprStringParser.hh"

#include <stdlib.h>
#include <unistd.h>
#include <iostream>
#include <vector>
#include <set>
#include <string>
#include <memory>

using namespace std;


void help(const char* prog) 
{
  cout << "Usage:  " << prog 
       << " training_data_file" << endl;
  cout << "\t Options: " << endl;
  cout << "\t-h --- help                                        " << endl;
  cout << "\t-o output Tuple file                               " << endl;
  cout << "\t-a input ascii file mode (see SprSimpleReader.hh)  " << endl;
  cout << "\t-A save output data in ascii instead of Root       " << endl;
  cout << "\t-y list of input classes (see SprAbsFilter.hh)     " << endl;
  cout << "\t-v verbose level (0=silent default,1,2)            " << endl;
  cout << "\t-V include only these input variables              " << endl;
  cout << "\t-z exclude input variables from the list           " << endl;
  cout << "\t-Z exclude input variables from the list, "
       << "but put them in the output file " << endl;
  cout << "\t\t Variables must be listed in quotes and separated by commas." 
       << endl;
  cout << "\t\t Variables must be listed in quotes and separated by commas." 
       << endl;
}


int main(int argc, char ** argv)
{
  // check command line
  if( argc < 2 ) {
    help(argv[0]);
    return 1;
  }

  // init
  string tupleFile;
  int readMode = 0;
  SprRWFactory::DataType writeMode = SprRWFactory::Root;
  int verbose = 0;
  string outFile;
  string includeList, excludeList;
  string inputClassesString;
  string stringVarsDoNotFeed;

  // decode command line
  int c;
  extern char* optarg;
  //  extern int optind;
  while( (c = getopt(argc,argv,"ho:a:Ay:v:V:z:Z:")) != EOF ) {
    switch( c )
      {
      case 'h' :
        help(argv[0]);
        return 1;
      case 'o' :
        tupleFile = optarg;
        break;
      case 'a' :
        readMode = (optarg==0 ? 0 : atoi(optarg));
        break;
      case 'A' :
        writeMode = SprRWFactory::Ascii;
        break;
      case 'y' :
        inputClassesString = optarg;
        break;
      case 'v' :
        verbose = (optarg==0 ? 0 : atoi(optarg));
        break;
      case 'V' :
        includeList = optarg;
        break;
      case 'z' :
        excludeList = optarg;
        break;
      case 'Z' :
        stringVarsDoNotFeed = optarg;
        break;
      }
  }

  // There has to be 1 argument after all options.
  string trFile = argv[argc-1];
  if( trFile.empty() ) {
    cerr << "No training file is specified." << endl;
    return 1;
  }

  // make reader
  SprRWFactory::DataType inputType 
    = ( readMode==0 ? SprRWFactory::Root : SprRWFactory::Ascii );
  auto_ptr<SprAbsReader> reader(SprRWFactory::makeReader(inputType,readMode));

  // include variables
  set<string> includeSet;
  if( !includeList.empty() ) {
    vector<vector<string> > includeVars;
    SprStringParser::parseToStrings(includeList.c_str(),includeVars);
    assert( !includeVars.empty() );
    for( int i=0;i<includeVars[0].size();i++ ) 
      includeSet.insert(includeVars[0][i]);
    if( !reader->chooseVars(includeSet) ) {
      cerr << "Unable to include variables in training set." << endl;
      return 2;
    }
    else {
      cout << "Following variables have been included in optimization: ";
      for( set<string>::const_iterator 
             i=includeSet.begin();i!=includeSet.end();i++ )
        cout << "\"" << *i << "\"" << " ";
      cout << endl;
    }
  }

  // exclude variables
  set<string> excludeSet;
  if( !excludeList.empty() ) {
    vector<vector<string> > excludeVars;
    SprStringParser::parseToStrings(excludeList.c_str(),excludeVars);
    assert( !excludeVars.empty() );
    for( int i=0;i<excludeVars[0].size();i++ ) 
      excludeSet.insert(excludeVars[0][i]);
    if( !reader->chooseAllBut(excludeSet) ) {
      cerr << "Unable to exclude variables from training set." << endl;
      return 2;
    }
    else {
      cout << "Following variables have been excluded from optimization: ";
      for( set<string>::const_iterator 
             i=excludeSet.begin();i!=excludeSet.end();i++ )
        cout << "\"" << *i << "\"" << " ";
      cout << endl;
    }
  }

  // read training data from file
  auto_ptr<SprAbsFilter> filter(reader->read(trFile.c_str()));
  if( filter.get() == 0 ) {
    cerr << "Unable to read data from file " << trFile.c_str() << endl;
    return 2;
  }
  vector<string> vars;
  filter->vars(vars);
  cout << "Read data from file " << trFile.c_str() 
       << " for variables";
  for( int i=0;i<vars.size();i++ ) 
    cout << " \"" << vars[i].c_str() << "\"";
  cout << endl;
  cout << "Total number of points read: " << filter->size() << endl;

  // filter training data by class
  vector<SprClass> inputClasses;
  if( !filter->filterByClass(inputClassesString.c_str()) ) {
    cerr << "Cannot choose input classes for string " 
         << inputClassesString << endl;
    return 2;
  }
  filter->classes(inputClasses);
  assert( inputClasses.size() > 1 );
  cout << "Training data filtered by class." << endl;
  for( int i=0;i<inputClasses.size();i++ ) {
    cout << "Points in class " << inputClasses[i] << ":   " 
         << filter->ptsInClass(inputClasses[i]) << endl;
  }

  // make histogram if requested
  if( tupleFile.empty() ) return 0;

  // make a writer
  auto_ptr<SprAbsWriter> tuple(SprRWFactory::makeWriter(writeMode,"training"));
  if( !tuple->init(tupleFile.c_str()) ) {
    cerr << "Unable to open output file " << tupleFile.c_str() << endl;
    return 8;
  }

  // determine if certain variables are to be excluded from usage,
  // but included in the output storage file (-Z option)
  string printVarsDoNotFeed;
  vector<vector<string> > varsDoNotFeed;
  SprStringParser::parseToStrings(stringVarsDoNotFeed.c_str(),varsDoNotFeed);
  vector<unsigned> mapper;
  for( int d=0;d<vars.size();d++ ) {
    if( varsDoNotFeed.empty() ||
        (find(varsDoNotFeed[0].begin(),varsDoNotFeed[0].end(),vars[d])
         ==varsDoNotFeed[0].end()) ) {
      mapper.push_back(d);
    }
    else {
      printVarsDoNotFeed += ( printVarsDoNotFeed.empty() ? "" : ", " );
      printVarsDoNotFeed += vars[d];
    }
  }
  if( !printVarsDoNotFeed.empty() ) {
    cout << "The following variables are not used in the algorithm, " 
         << "but will be included in the output file: " 
         << printVarsDoNotFeed.c_str() << endl;
  }

  // feed
  SprDataFeeder feeder(filter.get(),tuple.get(),mapper);
  if( !feeder.feed(1000) ) {
    cerr << "Cannot feed data into file " << tupleFile.c_str() << endl;
    return 9;
  }

  // exit
  return 0;
}

---