#include <FeatureExtractionModelTransformer.h>

Inheritance diagram for FeatureExtractionModelTransformer:

[legend]

Collaboration diagram for FeatureExtractionModelTransformer:

[legend]

Public Types
typedef std::vector< Model * >	Models

enum	Type { RTThread, NonRTThread }

Public Slots
void	start ()

Public Member Functions
	FeatureExtractionModelTransformer (Input input, const Transform &transform)

	FeatureExtractionModelTransformer (Input input, const Transforms &relatedTransforms)

virtual	~FeatureExtractionModelTransformer ()

Models	getAdditionalOutputModels ()
	Return any additional models that were created during processing. More...

bool	willHaveAdditionalOutputModels ()
	Return true if the current transform is one that may produce additional models (to be retrieved through getAdditionalOutputModels above). More...

void	abandon ()
	Hint to the processing thread that it should give up, for example because the process is going to exit or we want to get rid of the input model. More...

Model *	getInputModel ()
	Return the input model for the transform. More...

int	getInputChannel ()
	Return the input channel spec for the transform. More...

Models	getOutputModels ()
	Return the set of output models created by the transform or transforms. More...

Models	detachOutputModels ()
	Return the set of output models, also detaching them from the transformer so that they will not be deleted when the transformer is. More...

virtual Models	detachAdditionalOutputModels ()
	Return the set of additional models, also detaching them from the transformer. More...

QString	getMessage () const
	Return a warning or error message. More...

Protected Types
typedef std::map< int, std::map< int, SparseTimeValueModel * > >	AdditionalModelMap

Protected Member Functions
bool	initialise ()

virtual void	run ()

void	createOutputModels (int n)

SparseTimeValueModel *	getAdditionalModel (int transformNo, int binNo)

void	addFeature (int n, int blockFrame, const Vamp::Plugin::Feature &feature)

void	setCompletion (int, int)

void	getFrames (int channelCount, long startFrame, long size, float **buffer)

DenseTimeValueModel *	getConformingInput ()

template<typename ModelClass >
bool	isOutput (int n)

template<typename ModelClass >
ModelClass *	getConformingOutput (int n)

Protected Attributes
Vamp::Plugin *	m_plugin

std::vector< Vamp::Plugin::OutputDescriptor * >	m_descriptors

std::vector< int >	m_fixedRateFeatureNos

std::vector< int >	m_outputNos

std::map< int, bool >	m_needAdditionalModels

AdditionalModelMap	m_additionalModels

Transforms	m_transforms

Input	m_input

Models	m_outputs

bool	m_detached

bool	m_detachedAdd

bool	m_abandoned

QString	m_message

Detailed Description

Definition at line 31 of file FeatureExtractionModelTransformer.h.

Member Typedef Documentation

◆ AdditionalModelMap

typedef std::map<int, std::map<int, SparseTimeValueModel *> > FeatureExtractionModelTransformer::AdditionalModelMap

protected

Definition at line 64 of file FeatureExtractionModelTransformer.h.

◆ Models

typedef std::vector<Model *> ModelTransformer::Models

inherited

Definition at line 43 of file ModelTransformer.h.

Member Enumeration Documentation

◆ Type

enum Thread::Type

inherited

Enumerator
RTThread
NonRTThread

Definition at line 29 of file Thread.h.

Constructor & Destructor Documentation

◆ FeatureExtractionModelTransformer() [1/2]

FeatureExtractionModelTransformer::FeatureExtractionModelTransformer	(	Input	input,
		const Transform &	transform
	)

Definition at line 42 of file FeatureExtractionModelTransformer.cpp.

References initialise().

◆ FeatureExtractionModelTransformer() [2/2]

FeatureExtractionModelTransformer::FeatureExtractionModelTransformer	(	Input	input,
		const Transforms &	relatedTransforms
	)

Definition at line 52 of file FeatureExtractionModelTransformer.cpp.

References initialise().

◆ ~FeatureExtractionModelTransformer()

FeatureExtractionModelTransformer::~FeatureExtractionModelTransformer ( )

virtual

Definition at line 476 of file FeatureExtractionModelTransformer.cpp.

References m_descriptors, and m_plugin.

Member Function Documentation

◆ getAdditionalOutputModels()

FeatureExtractionModelTransformer::Models FeatureExtractionModelTransformer::getAdditionalOutputModels ( )

virtual

Return any additional models that were created during processing.

This might happen if, for example, a transform was configured to split a multi-bin output into separate single-bin models as it processed. These should not be queried until after the transform has completed.

Reimplemented from ModelTransformer.

Definition at line 486 of file FeatureExtractionModelTransformer.cpp.

References m_additionalModels.

◆ willHaveAdditionalOutputModels()

bool FeatureExtractionModelTransformer::willHaveAdditionalOutputModels ( )

virtual

Return true if the current transform is one that may produce additional models (to be retrieved through getAdditionalOutputModels above).

Reimplemented from ModelTransformer.

Definition at line 502 of file FeatureExtractionModelTransformer.cpp.

References m_needAdditionalModels.

◆ initialise()

bool FeatureExtractionModelTransformer::initialise ( )

protected

!!! hang on, this isn't right – we're modifying a copy

Definition at line 71 of file FeatureExtractionModelTransformer.cpp.

References areTransformsSimilar(), createOutputModels(), Transform::getBlockSize(), DenseTimeValueModel::getChannelCount(), getConformingInput(), TransformFactory::getInstance(), Transform::getPluginIdentifier(), Transform::getPluginVersion(), Model::getSampleRate(), Transform::getStepSize(), FeatureExtractionPluginFactory::instanceFor(), FeatureExtractionPluginFactory::instantiatePlugin(), m_descriptors, m_fixedRateFeatureNos, ModelTransformer::m_message, m_outputNos, m_plugin, ModelTransformer::m_transforms, TransformFactory::makeContextConsistentWithPlugin(), Transform::setBlockSize(), TransformFactory::setPluginParameters(), Transform::setStepSize(), and SVDEBUG.

Referenced by FeatureExtractionModelTransformer().

◆ run()

void FeatureExtractionModelTransformer::run ( )

protectedvirtual

!! need a better way to handle this – previously we were using a QMessageBox but that isn't an appropriate thing to do here either

Implements Thread.

Definition at line 561 of file FeatureExtractionModelTransformer.cpp.

References addFeature(), Transform::getBlockSize(), ModelTransformer::Input::getChannel(), DenseTimeValueModel::getChannelCount(), getConformingInput(), Transform::getDuration(), Model::getEndFrame(), getFrames(), ModelTransformer::Input::getModel(), Model::getSampleRate(), Model::getStartFrame(), Transform::getStartTime(), Transform::getStepSize(), Transform::getWindowType(), FFTModel::isOK(), Model::isReady(), ModelTransformer::m_abandoned, ModelTransformer::m_input, ModelTransformer::m_message, m_outputNos, ModelTransformer::m_outputs, m_plugin, ModelTransformer::m_transforms, StorageAdviser::PrecisionCritical, RealTime::realTime2Frame(), FFTModel::resume(), and setCompletion().

◆ createOutputModels()

void FeatureExtractionModelTransformer::createOutputModels ( int n )

protected

!! SV doesn't actually support display of models that have !! different underlying rates together – so we always set !! the model rate to be the input model's rate, and adjust !! the resolution appropriately. We can't properly display !! data with a higher resolution than the base model at all

Definition at line 223 of file FeatureExtractionModelTransformer.cpp.

References EditableDenseThreeDimensionalModel::BasicMultirateCompression, getConformingInput(), PluginRDFDescription::getOutputEventTypeURI(), PluginRDFDescription::getOutputSignalTypeURI(), Model::getSampleRate(), m_descriptors, m_needAdditionalModels, m_outputNos, ModelTransformer::m_outputs, m_plugin, ModelTransformer::m_transforms, EditableDenseThreeDimensionalModel::setBinNames(), Model::setRDFTypeURI(), SparseValueModel< PointType >::setScaleUnits(), and Model::setSourceModel().

Referenced by initialise().

◆ getAdditionalModel()

SparseTimeValueModel * FeatureExtractionModelTransformer::getAdditionalModel	(	int	transformNo,
		int	binNo
	)

protected

Definition at line 513 of file FeatureExtractionModelTransformer.cpp.

References Model::getRDFTypeURI(), SparseModel< PointType >::getResolution(), SparseModel< PointType >::getSampleRate(), SparseValueModel< PointType >::getScaleUnits(), SparseValueModel< PointType >::getValueMaximum(), SparseValueModel< PointType >::getValueMinimum(), m_additionalModels, m_needAdditionalModels, Model::setRDFTypeURI(), and SparseValueModel< PointType >::setScaleUnits().

Referenced by addFeature().

◆ addFeature()

void FeatureExtractionModelTransformer::addFeature	(	int	n,
		int	blockFrame,
		const Vamp::Plugin::Feature &	feature
	)

protected

Definition at line 812 of file FeatureExtractionModelTransformer.cpp.

References SparseValueModel< PointType >::addPoint(), SparseModel< PointType >::addPoint(), RegionModel::addPoint(), getAdditionalModel(), ModelTransformer::Input::getModel(), EditableDenseThreeDimensionalModel::getResolution(), Model::getSampleRate(), m_descriptors, m_fixedRateFeatureNos, ModelTransformer::m_input, m_needAdditionalModels, EditableDenseThreeDimensionalModel::setColumn(), SVDEBUG, and RealTime::toDouble().

Referenced by run().

◆ setCompletion()

void FeatureExtractionModelTransformer::setCompletion	(	int	n,
		int	completion
	)

protected

!!m_context.updates);

Definition at line 1011 of file FeatureExtractionModelTransformer.cpp.

References ModelTransformer::abandon(), Model::isAbandoning(), SparseModel< PointType >::setCompletion(), and EditableDenseThreeDimensionalModel::setCompletion().

Referenced by run().

◆ getFrames()

void FeatureExtractionModelTransformer::getFrames	(	int	channelCount,
		long	startFrame,
		long	size,
		float **	buffer
	)

protected

Definition at line 752 of file FeatureExtractionModelTransformer.cpp.

References ModelTransformer::Input::getChannel(), DenseTimeValueModel::getChannelCount(), getConformingInput(), DenseTimeValueModel::getData(), and ModelTransformer::m_input.

Referenced by run().

◆ getConformingInput()

DenseTimeValueModel * FeatureExtractionModelTransformer::getConformingInput ( )

protected

Definition at line 548 of file FeatureExtractionModelTransformer.cpp.

References ModelTransformer::getInputModel(), and SVDEBUG.

Referenced by createOutputModels(), getFrames(), initialise(), and run().

◆ isOutput()

template<typename ModelClass >

bool FeatureExtractionModelTransformer::isOutput ( int n )

inlineprotected

Definition at line 81 of file FeatureExtractionModelTransformer.h.

References ModelTransformer::m_outputs.

◆ getConformingOutput()

template<typename ModelClass >

ModelClass* FeatureExtractionModelTransformer::getConformingOutput ( int n )

inlineprotected

Definition at line 85 of file FeatureExtractionModelTransformer.h.

References ModelTransformer::m_outputs.

◆ abandon()

void ModelTransformer::abandon ( )

inlineinherited

Hint to the processing thread that it should give up, for example because the process is going to exit or we want to get rid of the input model.

Caller should still wait() and/or delete the transform before assuming its input and output models are no longer required.

Definition at line 68 of file ModelTransformer.h.

References ModelTransformer::m_abandoned.

Referenced by ModelTransformerFactory::modelAboutToBeDeleted(), and setCompletion().

◆ getInputModel()

Model* ModelTransformer::getInputModel ( )

inlineinherited

Return the input model for the transform.

Definition at line 73 of file ModelTransformer.h.

References ModelTransformer::Input::getModel(), and ModelTransformer::m_input.

Referenced by RealTimeEffectModelTransformer::getConformingInput(), getConformingInput(), and ModelTransformerFactory::modelAboutToBeDeleted().

◆ getInputChannel()

int ModelTransformer::getInputChannel ( )

inlineinherited

Return the input channel spec for the transform.

Definition at line 78 of file ModelTransformer.h.

References ModelTransformer::Input::getChannel(), and ModelTransformer::m_input.

◆ getOutputModels()

Models ModelTransformer::getOutputModels ( )

inlineinherited

Return the set of output models created by the transform or transforms.

Returns an empty list if any transform could not be initialised; an error message may be available via getMessage() in this situation.

Definition at line 86 of file ModelTransformer.h.

References ModelTransformer::m_outputs.

Referenced by ModelTransformer::detachOutputModels(), ModelTransformerFactory::modelAboutToBeDeleted(), and ModelTransformer::~ModelTransformer().

◆ detachOutputModels()

Models ModelTransformer::detachOutputModels ( )

inlineinherited

Return the set of output models, also detaching them from the transformer so that they will not be deleted when the transformer is.

The caller takes ownership of the models.

Definition at line 93 of file ModelTransformer.h.

References ModelTransformer::getOutputModels(), and ModelTransformer::m_detached.

Referenced by ModelTransformerFactory::transformMultiple().

◆ detachAdditionalOutputModels()

virtual Models ModelTransformer::detachAdditionalOutputModels ( )

inlinevirtualinherited

Return the set of additional models, also detaching them from the transformer.

The caller takes ownership of the models.

Definition at line 118 of file ModelTransformer.h.

References ModelTransformer::getAdditionalOutputModels(), and ModelTransformer::m_detachedAdd.

Referenced by ModelTransformerFactory::transformerFinished().

◆ getMessage()

QString ModelTransformer::getMessage ( ) const

inlineinherited

Return a warning or error message.

If getOutputModel returned a null pointer, this should contain a fatal error message for the transformer; otherwise it may contain a warning to show to the user about e.g. suboptimal block size or whatever.

Definition at line 129 of file ModelTransformer.h.

References ModelTransformer::m_message.

Referenced by ModelTransformerFactory::transformMultiple().