property.cc

Amis - A maximum entropy estimator 一个最大熵模型统计工具

//////////////////////////////////////////////////////////////////////
//
//  Copyright (c) 2000, Yusuke Miyao
//  You may distribute under the terms of the Artistic License.
//
//  Property.cc - Program property manager
//  $Id: Property.cc,v 1.13 2003/05/16 05:11:08 yusuke Exp $
//
//////////////////////////////////////////////////////////////////////

#include <amis/Property.h>
#include <amis/Tokenizer.h>
#include <amis/ArgumentParser.h>

AMIS_NAMESPACE_BEGIN

// Constants
const std::string Property::DEFAULT_PROPERTY_FILE( "amis.conf" );

// Static members
Initializer< OptionManager* >* PropertyItemBase::OPTION_QUEUE = NULL;
OptionManager Property::option_manager( PropertyItemBase::OPTION_QUEUE );
bool Property::initialized = false;

//////////////////////////////////////////////////////////////////////

// Launchers

ModelLauncher Property::model_launcher;
EventSpaceLauncher Property::event_space_launcher;
ModelFormatLauncher Property::model_format_launcher;
EventFormatLauncher Property::event_format_launcher;
EstimatorLauncher Property::estimator_launcher;
GaussianPriorLauncher Property::gaussian_prior_launcher;
ModelStatLauncher Property::model_stat_launcher;

//////////////////////////////////////////////////////////////////////

// Options

PropertyItemToggle Property::help( "", "", "-h", false, "Print help messages", true );
PropertyItemToggle Property::heavy_help( "", "--help", "-hh", false, "Print detailed help messages", true );
PropertyItemToggle Property::show_option_status( "", "--show-option-status", "-sos", false, "Show option status", true );
PropertyItemToggle Property::show_version( "", "--version", "-v", false, "Show version", true );

PropertyItem< std::string > Property::feature_type( "FEATURE_TYPE", "--feature-type", "-f", "binary", "The type of features" );
class FeatureWeightHash : public StringHash< FeatureWeightType > {
public:
  FeatureWeightHash( void ) {
    (*this)[ "alpha" ] = ALPHA;
    (*this)[ "lambda" ] = LAMBDA;
  }
};
namespace { FeatureWeightHash feature_weight_hash; };
PropertyItemNamedInt< FeatureWeightType > Property::feature_weight_type( "FEATURE_WEIGHT_TYPE", "--feature-weight-type", "-w", feature_weight_hash, ALPHA, "The type of weights of features" );

PropertyItem< std::string > Property::root_dir( "ROOT_DIR", "--root-dir", "-R", ".", "Root directory" );
PropertyItemVector< std::string > Property::model_file_list( "MODEL_FILE", "--model-file", "-m", std::vector< std::string >( 1, "amis.model" ), "Model file name(s)" );
PropertyItemVector< std::string > Property::event_file_list( "EVENT_FILE", "--event-file", "-e", std::vector< std::string >( 1, "amis.event" ), "Event file name(s)" );
PropertyItem< std::string > Property::output_file( "OUTPUT_FILE", "--output-file", "-o", "amis.output", "The name of the output file" );
PropertyItem< std::string > Property::log_file( "LOG_FILE", "--log-file", "-l", "amis.log", "The name of the log file" );
PropertyItem< std::string > Property::model_stat_file( "MODEL_STAT_FILE", "--model-stat-file", "", "amis.stat", "The name of the file of the model statistics" );
PropertyItem< std::string > Property::data_format( "DATA_FORMAT", "--data-format", "-d", "Amis", "Data format type (Amis, AmisTree, AmisFix, AmisFixKernel, or CMEE)" );
PropertyItem< std::string > Property::estimation_algorithm( "ESTIMATION_ALGORITHM", "--estimation-algorithm", "-a", "GIS", "Estimation algorithm (GIS, IIS, BFGS, BFGSMAP, LMVM, LMVMMAP or LMVMVBC)" );
PropertyItem< int > Property::num_iterations( "NUM_ITERATIONS", "--num-iterations", "-i", 200, "The number of iterations" );
PropertyItem< int > Property::report_interval( "REPORT_INTERVAL", "--report-interval", "-r", 1, "The interval of interation report" );
PropertyItem< int > Property::prec( "PRECISION", "--precision", "-p", 6, "The precision of the estimation" );
PropertyItem< bool > Property::event_on_file( "EVENT_ON_FILE", "--event-on-file", "", false, "Store events on file during the estimation" );
PropertyItem< std::string > Property::event_on_file_name( "EVENT_ON_FILE_NAME", "--event-on-file-name", "", "amis.event.tmp", "The name of temp event file used by EVENT_ON_FILE" );
PropertyItem< int > Property::num_threads( "NUM_THREADS", "--num-threads", "-j", 1, "The level of parallelism of estimation" );
PropertyItem< bool > Property::till_convergence( "TILL_CONVERGENCE", "--till-convergence", "-tc", false, "Run estimation till convergence (When using LMVM, LMVMMAP, and LMVMBC, TILL_CONVERGENCE is true always)" );
PropertyItem< bool > Property::suppress_message( "SUPPRESS_MESSAGE", "--suppress-message", "", false, "Suppress profiling messages" );
PropertyItem< std::string > Property::variance_type( "VARIANCE_TYPE", "--variance-type", "", "single", "Type of the variances used for MAP estimation" );
PropertyItem< bool > Property::model_stat( "MODEL_STAT", "--model-stat", "", false, "Output the statistics of the model" );

//////////////////////////////////////////////////////////////////////

void Property::parseArguments( int argc, const char** argv ) {
  ArgumentParser arguments( argc, argv );
  bool property_file_initialized = false;
  while ( ! arguments.empty() ) {
    std::string arg = arguments.nextArgument( "Too few arguments" );
    if ( arg[0] != '-' ) {
      if ( property_file_initialized ) {
	throw IllegalOptionError( "More than one arguments specified: " + arg );
      }
      property_file = arg;
      property_file_initialized = true;
    } else {
      OptionBase* option = option_manager.findOptionLongName( arg );
      if ( option == NULL ) {
	option = option_manager.findOptionShortName( arg );
	if ( option == NULL ) {
	  throw IllegalOptionError( "Unknown option found: " + arg );
	}
      }
      int num_args = option->getNumArguments();
      if ( num_args < 0 ) {
	// any number of arguments can be specified
	while ( ! arguments.empty() ) {
	  std::string arg = arguments.nextArgument( "Too few arguments" );
	  if ( arg[0] == '-' ) {
	    arguments.windBack();
	    break;
	  }
	  option->setValueFromString( arg );
	}
      } else if ( num_args == 0 ) {
	option->setValueFromString( "" );
      } else {
	for ( int i = 0; i < num_args; i++ ) {
	  if ( arguments.empty() ) {
	    OStringStream os;
	    os << option->getNameDescription() << " requires " << num_args << " arguments.";
	    throw IllegalOptionError( os.str() );
	  }
	  arg = arguments.nextArgument( "Too few arguments" );
	  option->setValueFromString( arg );
	}
      }
    }
  }
}

void Property::importProperty( void ) {
  std::ifstream pf( property_file.c_str() );
  if ( ! pf ) {
    throw IllegalPropertyError( "Can't open property file: " + property_file );
  }
  option_manager.resetToInit();

  Tokenizer t( pf );
  std::string attr, value, dummy;
  while ( ! t.endOfStream() ) {
    if ( ! t.nextToken( attr ) ) continue;  // empty line
    OptionBase* option = option_manager.findOption( attr );
    if ( option == NULL ) {
      throw IllegalPropertyError( "Unknown property: " + attr );
    }
    int num_args = option->getNumArguments();
    if ( num_args < 0 ) {
      while( t.nextToken( value ) ) {
	option->setValueFromString( value );
      }
    } else if ( num_args == 0 ) {
      // toggle-type property
      // Allow the following way
      // PROP1   ==> !default_val
      // PROP1 true  ==> true
      // PROP2 false ==> false
      option->setValueFromString( "" );
      while ( t.nextToken( value ) ) {
	option->setValueFromString( value );
      }
    } else {
      for ( int i = 0; i < num_args; i++ ) {
	if ( ! t.nextToken( value ) ) {
	  OStringStream os;
	  os << "Property: " << attr << " requires " << num_args << " arguments.";
	  throw IllegalPropertyError( os.str() );
	}
	option->setValueFromString( value );
      }
    }
  }
}

//////////////////////////////////////////////////////////////////////

/*
Property::Property() {

	// MODEL_STAT_FILE
	opt.setOpt( new Opt< std::string >( "MODEL_STAT_FILE", "--model-stat-file", "-msf", &model_stat_file, "amis.model.stat", "Model statiscs are written to this file." ) );
	
	opt.setOpt( new OptToggle( "WRITE_HEAVY_STAT", "--write-heavy-stat", "-whs", &write_heavy_stat, false, "Write heavy stats. Costs much file strage" ) );
	
	// FEATURE_COUNT_HASH
	opt.setOpt( new Opt<bool>( "FEATURE_COUNT_HASH", "--feature-count-hash", "-fch", &feature_count_hash, false, "Whether to use the feature count hash" ) );
	
	// FEATURE_WEIGHT_TYPE