Weightfile Class Reference

The Weightfile class serializes all information about a training into an xml tree. More...

#include <Weightfile.h>

Collaboration diagram for Weightfile:

Public Member Functions
	Weightfile ()
	Construct an empty weightfile.
	~Weightfile ()
	Destructor (removes temporary files associated with this weightfiles)
void	addFeatureImportance (const std::map< std::string, float > &importance)
	Add variable importance.
std::map< std::string, float >	getFeatureImportance () const
	Get feature importance.
void	addOptions (const Options &options)
	Add an Option object to the xml tree.
void	getOptions (Options &options) const
	Fills an Option object from the xml tree.
void	addSignalFraction (float signal_fraction)
	Saves the signal fraction in the xml tree.
float	getSignalFraction () const
	Loads the signal fraction frm the xml tree.
std::string	generateFileName (const std::string &suffix="")
	Returns a temporary filename with the given suffix.
void	addFile (const std::string &identifier, const std::string &custom_weightfile)
	Add a file (mostly a weightfile from a MVA library) to our Weightfile.
void	addStream (const std::string &identifier, std::istream &in)
	Add a stream to our weightfile.
template<class T >
void	addElement (const std::string &identifier, const T &element)
	Add an element to the xml tree.
template<class T >
void	addVector (const std::string &identifier, const std::vector< T > &vector)
	Add a vector to the xml tree.
void	getFile (const std::string &identifier, const std::string &custom_weightfile)
	Creates a file from our weightfile (mostly this will be a weightfile of an MVA library)
std::string	getStream (const std::string &identifier) const
	Returns the content of a stored stream as string.
template<class T >
T	getElement (const std::string &identifier) const
	Returns a stored element from the xml tree.
bool	containsElement (const std::string &identifier) const
	Returns true if given element is stored in the property tree.
template<class T >
T	getElement (const std::string &identifier, const T &default_value) const
	Returns a stored element from the xml tree.
template<class T >
std::vector< T >	getVector (const std::string &identifier) const
	Returns a stored vector from the xml tree.
void	setRemoveTemporaryDirectories (bool remove_temporary_directories)
	Set the deletion behaviour of the weightfile object for temporary directories For debugging it can be useful to not delete temporary directories.
const boost::property_tree::ptree &	getXMLTree () const
	Get xml tree.

Static Public Member Functions
static void	save (Weightfile &weightfile, const std::string &filename, const Belle2::IntervalOfValidity &iov=Belle2::IntervalOfValidity(0, 0, -1, -1))
	Static function which saves a Weightfile to a file.
static void	saveToROOTFile (Weightfile &weightfile, const std::string &filename)
	Static function which saves a Weightfile to a ROOT file.
static void	saveToXMLFile (Weightfile &weightfile, const std::string &filename)
	Static function which saves a Weightfile to a XML file.
static void	saveToStream (Weightfile &weightfile, std::ostream &stream)
	Static function which serializes a Weightfile to a stream.
static Weightfile	load (const std::string &filename, const Belle2::EventMetaData &emd=Belle2::EventMetaData(0, 0, 0))
	Static function which loads a Weightfile from a file or from the database.
static Weightfile	loadFromFile (const std::string &filename)
	Static function which loads a Weightfile from a file.
static Weightfile	loadFromROOTFile (const std::string &filename)
	Static function which loads a Weightfile from a ROOT file.
static Weightfile	loadFromXMLFile (const std::string &filename)
	Static function which loads a Weightfile from a XML file.
static Weightfile	loadFromStream (std::istream &stream)
	Static function which deserializes a Weightfile from a stream.
static void	saveToDatabase (Weightfile &weightfile, const std::string &identifier, const Belle2::IntervalOfValidity &iov=Belle2::IntervalOfValidity(0, 0, -1, -1))
	Static function which saves a Weightfile in the basf2 condition database.
static void	saveArrayToDatabase (const std::vector< Weightfile > &weightfiles, const std::string &identifier, const Belle2::IntervalOfValidity &iov=Belle2::IntervalOfValidity(0, 0, -1, -1))
	Static function which saves an array of Weightfile objects in the basf2 condition database.
static Weightfile	loadFromDatabase (const std::string &identifier, const Belle2::EventMetaData &emd=Belle2::EventMetaData(0, 0, 0))
	Static function which loads a Weightfile from the basf2 condition database.

Private Attributes
boost::property_tree::ptree	m_pt
	xml tree containing all the saved information of this weightfile
std::vector< std::string >	m_filenames
	generated temporary filenames, which will be removed in the destructor of this class
bool	m_remove_temporary_directories = true
	remove all temporary directories in the destructor of this class

Detailed Description

The Weightfile class serializes all information about a training into an xml tree.

It can save and load this information from: a stream, a file, the database

Definition at line 38 of file Weightfile.h.

Constructor & Destructor Documentation

Weightfile()

Weightfile ( )

inline

Construct an empty weightfile.

Definition at line 44 of file Weightfile.h.

{};

~Weightfile()

~Weightfile

(

)

Destructor (removes temporary files associated with this weightfiles)

Definition at line 52 of file Weightfile.cc.

    {
      for (auto& filename : m_filenames) {
        if (fs::exists(filename)) {
          if (m_remove_temporary_directories)
            fs::remove_all(filename);
        }
      }
    }

Member Function Documentation

addElement()

void addElement ( const std::string &  identifier, const T &  element  )

inline

Add an element to the xml tree.

Parameters

identifier	identifier of the element
element	the value

Definition at line 114 of file Weightfile.h.

      {
        m_pt.put(identifier, element);
      }

addFeatureImportance()

void addFeatureImportance

(

const std::map< std::string, float > &

importance

)

Add variable importance.

Parameters

importance

Definition at line 72 of file Weightfile.cc.

    {
      m_pt.put("number_of_importance_vars", importance.size());
      unsigned int i = 0;
      for (auto& pair : importance) {
        m_pt.put(std::string("importance_key") + std::to_string(i), pair.first);
        m_pt.put(std::string("importance_value") + std::to_string(i), pair.second);
        ++i;
      }
    }

addFile()

void addFile	(	const std::string &	identifier,
		const std::string &	custom_weightfile
	)

Add a file (mostly a weightfile from a MVA library) to our Weightfile.

Parameters

identifier	of this weightfile
custom_weightfile	the filename

Definition at line 115 of file Weightfile.cc.

    {
      // TODO Test if file is valid
      std::ifstream in(custom_weightfile, std::ios::in | std::ios::binary);
      addStream(identifier, in);
      in.close();
    }

addOptions()

void addOptions

(

const Options &

options

)

Add an Option object to the xml tree.

Parameters

options

to add

Definition at line 62 of file Weightfile.cc.

    {
      options.save(m_pt);
    }

addSignalFraction()

void addSignalFraction

(

float

signal_fraction

)

Saves the signal fraction in the xml tree.

Parameters

signal_fraction

Definition at line 95 of file Weightfile.cc.

    {
      m_pt.put("signal_fraction", signal_fraction);
    }

addStream()

void addStream	(	const std::string &	identifier,
		std::istream &	in
	)

Add a stream to our weightfile.

Parameters

identifier	of the stream
in	the stream

Definition at line 123 of file Weightfile.cc.

    {
      using base64_t =  boost::archive::iterators::base64_from_binary <
                        boost::archive::iterators::transform_width<std::string::const_iterator, 6, 8 >>;
 
      std::string contents;
      in.seekg(0, std::ios::end);
      contents.resize(in.tellg());
      in.seekg(0, std::ios::beg);
      in.read(&contents[0], contents.size());
      std::string enc(base64_t(contents.begin()), base64_t(contents.end()));
 
      m_pt.put(identifier, enc);
    }

addVector()

void addVector ( const std::string &  identifier, const std::vector< T > &  vector  )

inline

Add a vector to the xml tree.

Parameters

identifier	identifier of the vector
vector	the std::vector

Definition at line 125 of file Weightfile.h.

      {
        m_pt.put(identifier + "_size", vector.size());
        for (unsigned int i = 0; i < vector.size(); ++i) {
          m_pt.put(identifier + std::to_string(i), vector[i]);
        }
      }

containsElement()

bool containsElement ( const std::string &  identifier ) const

inline

Returns true if given element is stored in the property tree.

Parameters

identifier

of the stored element

Definition at line 160 of file Weightfile.h.

      {
        return m_pt.count(identifier) > 0;
      }

generateFileName()

std::string generateFileName

(

const std::string &

suffix = ""

)

Returns a temporary filename with the given suffix.

All temporary files are removed when the Weightfile class is destroyed

Parameters

suffix

of the temporary filename

Returns

Definition at line 105 of file Weightfile.cc.

    {
      char* directory_template = strdup((fs::temp_directory_path() / "Basf2MVA.XXXXXX").c_str());
      auto directory = mkdtemp(directory_template);
      std::string tmpfile = std::string(directory) + std::string("/weightfile") + suffix;
      m_filenames.emplace_back(directory);
      free(directory_template);
      return tmpfile;
    }

getElement() [1/2]

T getElement ( const std::string &  identifier ) const

inline

Returns a stored element from the xml tree.

Parameters

identifier

of the stored element

Definition at line 151 of file Weightfile.h.

      {
        return m_pt.get<T>(identifier);
      }

getElement() [2/2]

T getElement ( const std::string &  identifier, const T &  default_value  ) const

inline

Returns a stored element from the xml tree.

Parameters

identifier	of the stored element
default_value	default value of the stored element

Definition at line 171 of file Weightfile.h.

      {
        return m_pt.get<T>(identifier, default_value);
      }

getFeatureImportance()

std::map< std::string, float > getFeatureImportance

(

)

const

Get feature importance.

Definition at line 83 of file Weightfile.cc.

    {
      std::map<std::string, float> importance;
      unsigned int numberOfImportanceVars = m_pt.get<unsigned int>("number_of_importance_vars", 0);
      for (unsigned int i = 0; i < numberOfImportanceVars; ++i) {
        auto key = m_pt.get<std::string>(std::string("importance_key") + std::to_string(i));
        auto value = m_pt.get<float>(std::string("importance_value") + std::to_string(i));
        importance[key] = value;
      }
      return importance;
    }

getFile()

void getFile	(	const std::string &	identifier,
		const std::string &	custom_weightfile
	)

Creates a file from our weightfile (mostly this will be a weightfile of an MVA library)

Parameters

identifier	of the file
custom_weightfile	the filename which is created

Definition at line 138 of file Weightfile.cc.

    {
      std::ofstream out(custom_weightfile, std::ios::out | std::ios::binary);
      out << getStream(identifier);
    }

getOptions()

void getOptions

(

Options &

options

)

const

Fills an Option object from the xml tree.

Parameters

options

Definition at line 67 of file Weightfile.cc.

    {
      options.load(m_pt);
    }

getSignalFraction()

float getSignalFraction

(

)

const

Loads the signal fraction frm the xml tree.

Returns

Definition at line 100 of file Weightfile.cc.

    {
      return m_pt.get<float>("signal_fraction");
    }

getStream()

std::string getStream

(

const std::string &

identifier

)

const

Returns the content of a stored stream as string.

Parameters

identifier

of the stored stream

Definition at line 144 of file Weightfile.cc.

    {
      using binary_t = boost::archive::iterators::transform_width <
                       boost::archive::iterators::binary_from_base64<std::string::const_iterator>, 8, 6 >;
 
      auto contents = m_pt.get<std::string>(identifier);
      std::string dec(binary_t(contents.begin()), binary_t(contents.end()));
      return dec;
    }

getVector()

std::vector< T > getVector ( const std::string &  identifier ) const

inline

Returns a stored vector from the xml tree.

Parameters

identifier

of the stored std::vector

Definition at line 181 of file Weightfile.h.

      {
        std::vector<T> vector;
        vector.resize(m_pt.get<size_t>(identifier + "_size"));
        for (unsigned int i = 0; i < vector.size(); ++i) {
          vector[i] = m_pt.get<T>(identifier + std::to_string(i));
        }
        return vector;
      }

getXMLTree()

const boost::property_tree::ptree & getXMLTree ( ) const

inline

Get xml tree.

Definition at line 287 of file Weightfile.h.

{ return m_pt; };

load()

Weightfile load ( const std::string &  filename, const Belle2::EventMetaData &  emd = Belle2::EventMetaData(0, 0, 0)  )

static

Static function which loads a Weightfile from a file or from the database.

Parameters

filename	of the file containing our serialized Weightfile
emd	EventMetaData

Definition at line 195 of file Weightfile.cc.

    {
      if (boost::ends_with(filename, ".root")) {
        return loadFromROOTFile(filename);
      } else if (boost::ends_with(filename, ".xml")) {
        return loadFromXMLFile(filename);
      } else {
        return loadFromDatabase(filename, emd);
      }
    }

loadFromDatabase()

Weightfile loadFromDatabase ( const std::string &  identifier, const Belle2::EventMetaData &  emd = Belle2::EventMetaData(0, 0,                                         0)  )

static

Static function which loads a Weightfile from the basf2 condition database.

Parameters

identifier	in the database
emd	event meta data containing the current valid event which is checked against the interval of validity of the stored weightfiles.

Definition at line 281 of file Weightfile.cc.

    {
      auto pair = Belle2::Database::Instance().getData(emd, makeSaveForDatabase(identifier));
 
      if (pair.first == 0) {
        throw std::runtime_error("Given identifier cannot be loaded from the database: " + identifier);
      }
 
      DatabaseRepresentationOfWeightfile database_representation_of_weightfile = *static_cast<DatabaseRepresentationOfWeightfile*>
          (pair.first);
      std::stringstream ss(database_representation_of_weightfile.m_data);
      return Weightfile::loadFromStream(ss);
    }

loadFromFile()

Weightfile loadFromFile ( const std::string &  filename )

static

Static function which loads a Weightfile from a file.

Parameters

filename

of the file containing our serialized Weightfile

Definition at line 206 of file Weightfile.cc.

    {
      if (boost::ends_with(filename, ".root")) {
        return loadFromROOTFile(filename);
      } else if (boost::ends_with(filename, ".xml")) {
        return loadFromXMLFile(filename);
      } else {
        throw std::runtime_error("Cannot load file " + filename + " because file extension is not supported");
      }
    }

loadFromROOTFile()

Weightfile loadFromROOTFile ( const std::string &  filename )

static

Static function which loads a Weightfile from a ROOT file.

Parameters

filename

of the file containing our serialized Weightfile

Definition at line 217 of file Weightfile.cc.

    {
 
      if (not fs::exists(filename)) {
        throw std::runtime_error("Given filename does not exist: " + filename);
      }
 
      TFile file(filename.c_str(), "READ");
      if (file.IsZombie() or not file.IsOpen()) {
        throw std::runtime_error("Error during open of ROOT file named " + filename);
      }
 
      DatabaseRepresentationOfWeightfile* database_representation_of_weightfile = nullptr;
      file.GetObject("Weightfile", database_representation_of_weightfile);
 
      if (database_representation_of_weightfile == nullptr) {
        throw std::runtime_error("The provided ROOT file " + filename + " does not contain a valid MVA weightfile.");
      }
      std::stringstream ss(database_representation_of_weightfile->m_data);
      delete database_representation_of_weightfile;
      return Weightfile::loadFromStream(ss);
    }

loadFromStream()

Weightfile loadFromStream ( std::istream &  stream )

static

Static function which deserializes a Weightfile from a stream.

Parameters

stream

containing our serialized Weightfile

Definition at line 251 of file Weightfile.cc.

    {
      Weightfile weightfile;
      boost::property_tree::xml_parser::read_xml(stream, weightfile.m_pt);
      return weightfile;
    }

loadFromXMLFile()

Weightfile loadFromXMLFile ( const std::string &  filename )

static

Static function which loads a Weightfile from a XML file.

Parameters

filename

of the file containing our serialized Weightfile

Definition at line 240 of file Weightfile.cc.

    {
      if (not fs::exists(filename)) {
        throw std::runtime_error("Given filename does not exist: " + filename);
      }
 
      Weightfile weightfile;
      boost::property_tree::xml_parser::read_xml(filename, weightfile.m_pt);
      return weightfile;
    }

save()

void save ( Weightfile &  weightfile, const std::string &  filename, const Belle2::IntervalOfValidity &  iov = Belle2::IntervalOfValidity(0, 0, -1, -1)  )

static

Static function which saves a Weightfile to a file.

Parameters

weightfile	to store
filename	of the file or identifier of the database
iov	IntervalOfValidity of the Weightfile

Definition at line 154 of file Weightfile.cc.

    {
      if (boost::ends_with(filename, ".root")) {
        return saveToROOTFile(weightfile, filename);
      } else if (boost::ends_with(filename, ".xml")) {
        return saveToXMLFile(weightfile, filename);
      } else {
        return saveToDatabase(weightfile, filename, iov);
      }
    }

saveArrayToDatabase()

void saveArrayToDatabase ( const std::vector< Weightfile > &  weightfiles, const std::string &  identifier, const Belle2::IntervalOfValidity &  iov = Belle2::IntervalOfValidity(0, 0, -1, -1)  )

static

Static function which saves an array of Weightfile objects in the basf2 condition database.

Parameters

weightfiles	vector of weightfiles to save
identifier	in the database
iov	interval of validity of this weightfile

Definition at line 267 of file Weightfile.cc.

    {
      DBImportArray<DatabaseRepresentationOfWeightfile> dbArray(makeSaveForDatabase(identifier));
 
      for (auto weightfile : weightfiles) {
        std::stringstream ss;
        Weightfile::saveToStream(weightfile, ss);
        dbArray.appendNew(ss.str());
      }
 
      dbArray.import(iov);
    }

saveToDatabase()

void saveToDatabase ( Weightfile &  weightfile, const std::string &  identifier, const Belle2::IntervalOfValidity &  iov = Belle2::IntervalOfValidity(0, 0, -1, -1)  )

static

Static function which saves a Weightfile in the basf2 condition database.

Parameters

weightfile	to save
identifier	in the database
iov	interval of validity of this weightfile

Definition at line 258 of file Weightfile.cc.

    {
      std::stringstream ss;
      Weightfile::saveToStream(weightfile, ss);
      DatabaseRepresentationOfWeightfile database_representation_of_weightfile;
      database_representation_of_weightfile.m_data = ss.str();
      Belle2::Database::Instance().storeData(makeSaveForDatabase(identifier), &database_representation_of_weightfile, iov);
    }

saveToROOTFile()

void saveToROOTFile ( Weightfile &  weightfile, const std::string &  filename  )

static

Static function which saves a Weightfile to a ROOT file.

Parameters

weightfile	to store
filename	of the file

Definition at line 165 of file Weightfile.cc.

    {
      std::stringstream ss;
      Weightfile::saveToStream(weightfile, ss);
      DatabaseRepresentationOfWeightfile database_representation_of_weightfile;
      database_representation_of_weightfile.m_data = ss.str();
      TFile file(filename.c_str(), "RECREATE");
      file.WriteObject(&database_representation_of_weightfile, "Weightfile");
    }

saveToStream()

void saveToStream ( Weightfile &  weightfile, std::ostream &  stream  )

static

Static function which serializes a Weightfile to a stream.

Parameters

weightfile	to serialize
stream

Definition at line 185 of file Weightfile.cc.

    {
#if BOOST_VERSION < 105600
      boost::property_tree::xml_writer_settings<char> settings('\t', 1);
#else
      boost::property_tree::xml_writer_settings<std::string> settings('\t', 1);
#endif
      boost::property_tree::xml_parser::write_xml(stream, weightfile.m_pt, settings);
    }

saveToXMLFile()

void saveToXMLFile ( Weightfile &  weightfile, const std::string &  filename  )

static

Static function which saves a Weightfile to a XML file.

Parameters

weightfile	to store
filename	of the file

Definition at line 175 of file Weightfile.cc.

    {
#if BOOST_VERSION < 105600
      boost::property_tree::xml_writer_settings<char> settings('\t', 1);
#else
      boost::property_tree::xml_writer_settings<std::string> settings('\t', 1);
#endif
      boost::property_tree::xml_parser::write_xml(filename, weightfile.m_pt, std::locale(), settings);
    }

setRemoveTemporaryDirectories()

void setRemoveTemporaryDirectories ( bool  remove_temporary_directories )

inline

Set the deletion behaviour of the weightfile object for temporary directories For debugging it can be useful to not delete temporary directories.

Definition at line 282 of file Weightfile.h.

{ m_remove_temporary_directories = remove_temporary_directories; }

Member Data Documentation

m_filenames

std::vector<std::string> m_filenames

private

generated temporary filenames, which will be removed in the destructor of this class

Definition at line 291 of file Weightfile.h.

m_pt

boost::property_tree::ptree m_pt

private

xml tree containing all the saved information of this weightfile

Definition at line 290 of file Weightfile.h.

m_remove_temporary_directories

bool m_remove_temporary_directories = true

private

remove all temporary directories in the destructor of this class

Definition at line 292 of file Weightfile.h.

---

The documentation for this class was generated from the following files:

• mva/interface/include/Weightfile.h
• mva/interface/src/Weightfile.cc