ViterbiPosTagger.h

Go to the documentation of this file.

// Copyright 2002-2013 CEA LIST
// SPDX-FileCopyrightText: 2022 CEA LIST <gael.de-chalendar@cea.fr>
//
// SPDX-License-Identifier: MIT
 
/***************************************************************************
 *   Copyright (C) 2004-2012 by CEA LIST                               *
 *                                                                         *
 ***************************************************************************/
#ifndef LIMA_LINGUISTICPROCESSING_POSTAGGER_VITERBIPOSTAGGER_H
#define LIMA_LINGUISTICPROCESSING_POSTAGGER_VITERBIPOSTAGGER_H
 
#include "PosTaggerExport.h"
#include "viterbiUtils.h"
#include "frequencyCost.h"
 
#include "common/XMLConfigurationFiles/xmlConfigurationFileExceptions.h"
#include "common/MediaticData/mediaticData.h"
#include "common/time/timeUtilsController.h"
#include "linguisticProcessing/common/annotationGraph/AnnotationData.h"
 
#include "common/MediaProcessors/MediaProcessUnit.h"
#include "linguisticProcessing/core/LinguisticAnalysisStructure/AnalysisGraph.h"
#include "linguisticProcessing/core/LinguisticAnalysisStructure/LinguisticGraph.h"
#include "linguisticProcessing/core/LinguisticAnalysisStructure/MorphoSyntacticData.h"
#include "linguisticProcessing/core/LinguisticAnalysisStructure/MorphoSyntacticDataUtils.h"
#include "linguisticProcessing/client/LinguisticProcessingException.h"
 
#include <list>
#include <vector>
#include <set>
#include <map>
#include <queue>
 
#include <sstream>
 
namespace Lima
{
 
namespace LinguisticProcessing
{
 
namespace PosTagger
{
 
#define VITERBIPOSTAGGER_CLASSID "ViterbiPosTagger"
 
template<typename Cost,typename CostFunction>
class ViterbiPosTagger : public MediaProcessUnit
{
public:
  ViterbiPosTagger(const CostFunction& costFunction) :
      m_costFunction(costFunction),
      m_defaultCateg(),
      m_stopCategories(),
      m_microAccessor() {}
 
  virtual ~ViterbiPosTagger() {}
 
  void init(
    Common::XMLConfigurationFiles::GroupConfigurationStructure& unitConfiguration,
    Manager* manager) override;
 
  LimaStatusCode process(AnalysisContent& analysis) const override;
 
private:
 
  struct PredData
  {
    PredData() :
        m_cost(),
        m_predMicro(),
        m_predIndex(),
    m_predPredMicros() {}
    Cost m_cost;
    LinguisticCode m_predMicro;
    uint64_t m_predIndex;
    std::vector<LinguisticCode> m_predPredMicros;
    // this is used to insert PredData in ordered containers
    inline bool operator<(const PredData& pd) const { return m_predMicro<pd.m_predMicro; }
  };
 
  typedef std::map< LinguisticCode, std::vector<PredData> > MicroCatDataMap;
  typedef typename std::map< LinguisticCode, std::vector<PredData> >::iterator MicroCatDataMapItr;
  typedef typename std::vector<PredData>::iterator PredDataVectorItr;
  typedef typename std::vector<PredData>::const_iterator PredDataVectorCItr;
 
  struct StepData
  {
    LinguisticGraphVertex m_srcVertex; // associated vertex in the graph
    std::vector<uint64_t> m_predStepIndexes; // direct ancestors in the StepData vector
    MicroCatDataMap m_microCatsData; // the micro categories associated to the node
  };
 
  typedef std::vector<StepData> StepDataVector;
  typedef typename std::vector<StepData>::iterator StepDataVectorItr;
 
  struct TargetVertexId
  {
    LinguisticGraphVertex m_sourceVx;
    LinguisticCode m_categ;
    std::vector<LinguisticCode> m_preds;
    bool operator<(const TargetVertexId& tvi) const
    {
      if (m_sourceVx!=tvi.m_sourceVx) return m_sourceVx<tvi.m_sourceVx;
      if (m_categ!=tvi.m_categ) return m_categ<tvi.m_categ;
      return m_preds<tvi.m_preds;
    }
  };
 
  const CostFunction m_costFunction;
  LinguisticCode m_defaultCateg;
  std::list<LinguisticCode> m_stopCategories;
  const Common::PropertyCode::PropertyAccessor* m_microAccessor;
  MediaId m_language;
 
    void initializeStepDataFromGraph(
    const LinguisticGraph* srcgraph,
    LinguisticGraphVertex start,
    LinguisticCode startMicro,
    const std::list< LinguisticCode >& predCats,
    LinguisticGraphVertex end,
    StepDataVector& stepData) const;
 
  void performViterbiOnStepData(StepDataVector& stepData) const;
 
    LinguisticGraphVertex reportPathsInGraph(
    LinguisticGraph* srcgraph,
    LinguisticGraph* resultgraph,
    LinguisticGraphVertex startVertex,
    StepDataVector& stepData,
    Common::AnnotationGraphs::AnnotationData* annotationData) const;
 
};
 
class LIMA_POSTAGGER_EXPORT ViterbiPosTaggerFactory : public InitializableObjectFactory<MediaProcessUnit>
{
public:
 
  virtual std::shared_ptr<MediaProcessUnit> create(
    Common::XMLConfigurationFiles::GroupConfigurationStructure& unitConfiguration,
    MediaProcessUnit::Manager* manager) const override ;
 
private:
  ViterbiPosTaggerFactory(const std::string& id);
  static std::unique_ptr< ViterbiPosTaggerFactory > s_instance;
 
};
 
 
}
 
}
 
}
 
#include "ViterbiPosTagger.tcc"
 
#endif