languageData.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
 
/***************************************************************************
 *   Copyright (C) 2004 by CEA - LIST                                      *
 *                                                                         *
 ***************************************************************************/
 
 
 
#ifndef LIMA_COMMONS_LINGUISTICDATA_LANGUAGEDATA_H
#define LIMA_COMMONS_LINGUISTICDATA_LANGUAGEDATA_H
 
#include "common/LimaCommon.h"
#include "linguisticProcessing/LinguisticProcessingCommon.h"
#include "common/MediaticData/mediaData.h"
 
#include <stdexcept>
#include <list>
#include <vector>
#include <set>
 
namespace Lima
{
 
namespace Common
{
  namespace PropertyCode {
    class PropertyCodeManager;
  }
  namespace XMLConfigurationFiles {
    class XMLConfigurationFileParser;
  }
namespace MediaticData
{
#define LANGUAGEDATA_CLASSID "LanguageData"
 
#define ID_NONE_1 L_NONE
#define SYMBOLIC_NONE_1 "NONE_1"
 
typedef enum {NO_CHAIN_TYPE=0, NOMINAL, VERBAL} ChainsType;
 
typedef uint64_t SyntacticRelationId;
 
class LanguageDataPrivate;
class LIMA_LINGUISTICDATA_EXPORT LanguageData : public MediaData
{
public:
 
  LanguageData();
 
  virtual ~LanguageData();
 
  virtual void initialize(
    MediaId lang,
    const std::string& resourcesPath,
    XMLConfigurationFiles::XMLConfigurationFileParser& conf) override;
 
  const PropertyCode::PropertyCodeManager& getPropertyCodeManager() const;
 
  const std::vector<std::set<LinguisticCode> >& getMicrosForNENormalization() const;
 
  bool isAnEmptyMacroCategory(LinguisticCode id) const ;
 
  bool isAnEmptyMicroCategory(LinguisticCode id) const ;
 
  bool isAConjugatedVerb(LinguisticCode id) const;
 
  bool isAPropositionIntroductor(LinguisticCode id) const;
 
  bool isTypeARelationForChain(ChainsType, uint64_t) const;
 
  LinguisticCode compoundTense(LinguisticCode mode, LinguisticCode auxTense) const;
  LinguisticCode compoundTense(std::string& mode, std::string& auxTense) const;
 
  std::set< LinguisticCode >&  getNounPhraseHeadMicroCategories();
 
  const std::set< LinguisticCode >&  getNounPhraseHeadMicroCategories() const;
  const std::set< LinguisticCode >&  getDefiniteMicroCategories() const;
  const std::set< LinguisticCode >&  getConjugatedVerbs() const;
 
  typedef std::map<std::string,uint64_t> EntityTypes;
  typedef std::map<uint64_t,std::string> EntityNames;
 
  LinguisticCode getEntityType(const EntityTypes& entityTypes,
                               const std::string& typeName) const;
  LinguisticCode getEntityType(const std::string& entity,
                               const std::string& typeName) const;
 
  // same but using common types identifiers (defined in linguisticData)
  LinguisticCode getEntityCommonType(const std::string& entity,
                                     const std::string& typeName) const;
 
  const std::string& getEntityName(const std::string& entity,
                                   LinguisticCode type) const;
 
  const EntityTypes& getEntityTypes(const std::string& entity) const;
  // not a const reference because the result object is built
  // inside the function
  EntityNames getEntityNames(const std::string& entity) const;
 
  const std::string& getSyntacticRelationName(SyntacticRelationId id) const;
  SyntacticRelationId getSyntacticRelationId(const std::string& name) const;
 
  const std::map<QString, QString>& getLimaToLanguageCodeMapping() const;
  QString getLimaToLanguageCodeMappingValue(const QString& code) const;
private:
  LanguageData(const LanguageData& ld);
  LanguageData& operator=(const LanguageData& ld);
  LanguageDataPrivate* m_d;
};
 
} // MediaticData
} // Common
} // Lima
 
#endif