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

#include <Weightfile.h>

Collaboration diagram for Weightfile:

[legend]

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	addContractVersion (int version)
	Store the version of the contract between this weightfile and the code applying it.

bool	hasContractVersion () const
	Returns true if this weightfile records a contract version.

int	getContractVersion (int default_value=-1) const
	Returns the contract version of this weightfile, or the default value.

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

Static Private Attributes
static constexpr const char *	c_contractVersion = "contract_version"
	identifier of the element holding the contract version, shared by all weightfiles

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.

44{};

◆ ~Weightfile()

~Weightfile ( )

Destructor (removes temporary files associated with this weightfiles)

Definition at line 51 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

◆ addContractVersion()

void addContractVersion ( int version )

inline

Store the version of the contract between this weightfile and the code applying it.

The contract version says which code can use this weightfile: how the inputs are built and how the outputs are interpreted. It is bumped by hand whenever such a change makes earlier weightfiles unusable. The applying code has to decide whether it implements that version, provides an older code path for it, or refuses to run. It is therefore meant for weightfiles whose payload names carry no version, where the file itself has to say which code it fits.

Parameters

version contract version of the code that created this weightfile

Definition at line 144 of file Weightfile.h.

      {
        addElement(c_contractVersion, version);
      }

◆ addElement()

template<class T>

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 71 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 114 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 61 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 94 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 122 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()

template<class T>

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 200 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 104 of file Weightfile.cc.

    {
      std::string directory_template = (fs::temp_directory_path() / "Basf2MVA.XXXXXX").string();
      if (mkdtemp(directory_template.data()) == nullptr) {
        throw std::runtime_error(std::string("mkdtemp failed: ") + std::strerror(errno));
      }
      m_filenames.push_back(directory_template);
      return directory_template + "/weightfile" + suffix;
    }

◆ getContractVersion()

int getContractVersion ( int default_value = -1 ) const

inline

Returns the contract version of this weightfile, or the default value.

The default is returned if no contract version is stored, and also if the stored value is not an integer. Use hasContractVersion() to tell the two apart.

Parameters

default_value returned if no usable contract version is stored

Definition at line 167 of file Weightfile.h.

      {
        auto version = m_pt.get_optional<int>(c_contractVersion);
        return version ? *version : default_value;
      }

◆ getElement() [1/2]

template<class T>

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 191 of file Weightfile.h.

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

◆ getElement() [2/2]

template<class T>

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 211 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 82 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 137 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 66 of file Weightfile.cc.

    {
      options.load(m_pt);
    }

◆ getSignalFraction()

float getSignalFraction ( ) const

Loads the signal fraction frm the xml tree.

Returns

Definition at line 99 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 143 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()

template<class T>

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 221 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 327 of file Weightfile.h.

327{ return m_pt; };

◆ hasContractVersion()

bool hasContractVersion ( ) const

inline

Returns true if this weightfile records a contract version.

False for weightfiles written before the contract version was introduced, which the applying code has to treat as the oldest version it knows.

Definition at line 155 of file Weightfile.h.

      {
        return containsElement(c_contractVersion);
      }

◆ 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 194 of file Weightfile.cc.

    {
      if (filename.ends_with(".root")) {
        return loadFromROOTFile(filename);
      } else if (filename.ends_with(".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 280 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 205 of file Weightfile.cc.

    {
      if (filename.ends_with(".root")) {
        return loadFromROOTFile(filename);
      } else if (filename.ends_with(".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 216 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 250 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 239 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 153 of file Weightfile.cc.

    {
      if (filename.ends_with(".root")) {
        return saveToROOTFile(weightfile, filename);
      } else if (filename.ends_with(".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 266 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 257 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 164 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 184 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 174 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 322 of file Weightfile.h.

322{ m_remove_temporary_directories = remove_temporary_directories; }

Member Data Documentation

◆ c_contractVersion

const char* c_contractVersion = "contract_version"

staticconstexprprivate

identifier of the element holding the contract version, shared by all weightfiles

Definition at line 331 of file Weightfile.h.

◆ 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 334 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 333 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 335 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

Public Member Functions

Static Public Member Functions

Private Attributes

Static Private Attributes

Detailed Description

Constructor & Destructor Documentation

◆ Weightfile()

◆ ~Weightfile()

Member Function Documentation

◆ addContractVersion()

◆ addElement()

◆ addFeatureImportance()

◆ addFile()

◆ addOptions()

◆ addSignalFraction()

◆ addStream()

◆ addVector()

◆ containsElement()

◆ generateFileName()

◆ getContractVersion()

◆ getElement() [1/2]

◆ getElement() [2/2]

◆ getFeatureImportance()

◆ getFile()

◆ getOptions()

◆ getSignalFraction()

◆ getStream()

◆ getVector()

◆ getXMLTree()

◆ hasContractVersion()

◆ load()

◆ loadFromDatabase()

◆ loadFromFile()

◆ loadFromROOTFile()

◆ loadFromStream()

◆ loadFromXMLFile()

◆ save()

◆ saveArrayToDatabase()

◆ saveToDatabase()

◆ saveToROOTFile()

◆ saveToStream()

◆ saveToXMLFile()

◆ setRemoveTemporaryDirectories()

Member Data Documentation

◆ c_contractVersion

◆ m_filenames

◆ m_pt

◆ m_remove_temporary_directories