bowToken.h

Go to the documentation of this file.

// Copyright 2002-2020 CEA LIST
// SPDX-FileCopyrightText: 2022 CEA LIST <gael.de-chalendar@cea.fr>
//
// SPDX-License-Identifier: MIT
 
/************************************************************************
 * @file     bowToken.h
 * @author   Besancon Romaric
 * @date     Tue Oct  7 17:13:26 2003
 * copyright Copyright (C) 2003 by CEA LIST
 *
 * @brief
 *
 ***********************************************************************/
 
#ifndef BOWTOKEN_H
#define BOWTOKEN_H
 
#include "linguisticProcessing/common/BagOfWords/AbstractBoWElement.h"
#include "common/LimaCommon.h"
#include "linguisticProcessing/LinguisticProcessingCommon.h"
#include "common/Data/LimaString.h"
#include "linguisticProcessing/common/misc/positionLengthList.h"
#include <iostream>
#include <string>
#include <stdexcept>
#include <map>
#include <set>
#include <memory>
 
namespace Lima {
namespace Common {
namespace PropertyCode {
  class PropertyManager;
}
namespace BagOfWords {
 
class BoWTokenPrivate;
class LIMA_BOW_EXPORT BoWToken : public AbstractBoWElement
{
  friend class BoWTokenPrivate;
public:
 
  BoWToken(const Lima::LimaString& lemma,
           const LinguisticCode category,
           const uint64_t position,
           const uint64_t length);
  BoWToken(const Lima::LimaString& str,
           const uint64_t position=0,
           const uint64_t length=0);
  BoWToken(const BoWToken& tok);
  BoWToken();
 
  virtual ~BoWToken();
 
  virtual void clear() override {}
 
  virtual BoWToken* clone() const override;
//   virtual BoWToken* clone(const std::map<BoWToken*,BoWToken*>&) const;
 
  virtual Lima::LimaString getLemma(void) const;
  virtual Lima::LimaString getInflectedForm(void) const;
  LinguisticCode getCategory(void) const;
  uint64_t getPosition(void) const override;
  uint64_t getLength(void) const override;
 
  virtual uint64_t size(void) const override;
 
  void setLemma(const Lima::LimaString& lemma);
  void setInflectedForm(const Lima::LimaString& inflectedForm);
  void setCategory(LinguisticCode c);
  void setPosition(const uint64_t pos) override;
  void setLength(const uint64_t len) override;
 
  virtual Lima::LimaString getString(void) const override;
 
  static void setInternalSeparator(const LimaChar c);
  static void setCompoundSeparator(const LimaChar c);
 
  static LimaChar getInternalSeparator();
  static LimaChar getCompoundSeparator();
 
  virtual BoWType getType() const override;
 
  uint64_t getVertex() const;
  void setVertex(uint64_t vertex);
 
  virtual std::set< uint64_t > getVertices() const override;
 
  virtual std::string getOutputUTF8String(const Common::PropertyCode::PropertyManager* macroManager = 0) const override;
 
  virtual std::string getIdUTF8String() const override;
 
  friend LIMA_BOW_EXPORT std::ostream& operator << (std::ostream&, const BoWToken&);
  friend LIMA_BOW_EXPORT QDebug& operator << (QDebug&, const BoWToken&);
 
  bool operator==(const BoWToken&) const;
 
  bool operator!=(const BoWToken&) const;
 
 
  virtual Common::Misc::PositionLengthList getPositionLengthList() const;
 
  virtual void addToPosition(const uint64_t offset);
 
  friend LIMA_BOW_EXPORT std::ostream& operator << (std::ostream& os,
                                                    const Common::Misc::PositionLengthList& p);
  friend LIMA_BOW_EXPORT QDebug& operator << (QDebug& os,
                                                    const Common::Misc::PositionLengthList& p);
 
  static void setUseOnlyLemma(const bool b);
  static bool getUseOnlyLemma();
 
  friend LIMA_BOW_EXPORT std::ostream& operator << (std::ostream& os,
                                                    const std::map<BoWToken*,uint64_t>& refMap);
  friend LIMA_BOW_EXPORT QDebug& operator << (QDebug& os,
                                                    const std::map<BoWToken*,uint64_t>& refMap);
 
protected:
  BoWToken(BoWTokenPrivate& d);
  BoWToken& operator=(const BoWToken& tok);
  bool operator==(const BoWToken& tok);
  BoWTokenPrivate* m_d;
};
//**************************************************************
// exceptions
//**************************************************************
class LIMA_BOW_EXPORT InputErrorException : public std::runtime_error {
 public:
  InputErrorException(const std::string& mes):std::runtime_error(mes){};
};
 
//**************************************************************
// functions
//**************************************************************
} // namespace BagOfWords
} // namespace Common
} // namespace Lima
 
#endif