mosesdecoder/moses/TargetPhrase.h
Matthias Huck 1740478238 Framework for additional phrase properties in decoding.
Derive your property class from PhraseProperty. Do any expensive string
processing of the property value in there, not in the feature
implementation, and provide methods to access the information in
appropriate data formats. The property value string will thus have to
be processed only once (on loading) rather than each time the respective
phrase is applied and your feature needs to access the property value.
2014-05-19 21:54:08 +01:00

180 lines
5.4 KiB
C++

// $Id$
/***********************************************************************
Moses - factored phrase-based language decoder
Copyright (C) 2006 University of Edinburgh
This library is free software; you can redistribute it and/or
modify it under the terms of the GNU Lesser General Public
License as published by the Free Software Foundation; either
version 2.1 of the License, or (at your option) any later version.
This library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public
License along with this library; if not, write to the Free Software
Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
***********************************************************************/
#ifndef moses_TargetPhrase_h
#define moses_TargetPhrase_h
#include <algorithm>
#include <vector>
#include "TypeDef.h"
#include "Phrase.h"
#include "ScoreComponentCollection.h"
#include "AlignmentInfo.h"
#include "moses/PP/PhraseProperty.h"
#include "util/string_piece.hh"
#include <boost/shared_ptr.hpp>
#ifdef HAVE_PROTOBUF
#include "rule.pb.h"
#endif
namespace Moses
{
class FeatureFunction;
class InputPath;
/** represents an entry on the target side of a phrase table (scores, translation, alignment)
*/
class TargetPhrase: public Phrase
{
private:
friend std::ostream& operator<<(std::ostream&, const TargetPhrase&);
friend void swap(TargetPhrase &first, TargetPhrase &second);
float m_fullScore, m_futureScore;
ScoreComponentCollection m_scoreBreakdown;
const AlignmentInfo* m_alignTerm, *m_alignNonTerm;
const Word *m_lhsTarget;
mutable Phrase *m_ruleSource; // to be set by the feature function that needs it.
std::map<std::string, boost::shared_ptr<PhraseProperty> > m_properties;
public:
TargetPhrase();
TargetPhrase(const TargetPhrase &copy);
explicit TargetPhrase(std::string out_string);
explicit TargetPhrase(const Phrase &targetPhrase);
~TargetPhrase();
// 1st evaluate method. Called during loading of phrase table.
void Evaluate(const Phrase &source, const std::vector<FeatureFunction*> &ffs);
// as above, score with ALL FFs
// Used only for OOV processing. Doesn't have a phrase table connect with it
void Evaluate(const Phrase &source);
// 'inputPath' is guaranteed to be the raw substring from the input. No factors were added or taken away
void Evaluate(const InputType &input, const InputPath &inputPath);
void SetSparseScore(const FeatureFunction* translationScoreProducer, const StringPiece &sparseString);
// used to set translation or gen score
void SetXMLScore(float score);
#ifdef HAVE_PROTOBUF
void WriteToRulePB(hgmert::Rule* pb) const;
#endif
/***
* return the estimated score resulting from our being added to a sentence
* (it's an estimate because we don't have full n-gram info for the language model
* without using the (unknown) full sentence)
*
*/
inline float GetFutureScore() const {
return m_fullScore;
}
inline const ScoreComponentCollection &GetScoreBreakdown() const {
return m_scoreBreakdown;
}
inline ScoreComponentCollection &GetScoreBreakdown() {
return m_scoreBreakdown;
}
void SetTargetLHS(const Word *lhs) {
m_lhsTarget = lhs;
}
const Word &GetTargetLHS() const {
return *m_lhsTarget;
}
void SetAlignmentInfo(const StringPiece &alignString);
void SetAlignTerm(const AlignmentInfo *alignTerm) {
m_alignTerm = alignTerm;
}
void SetAlignNonTerm(const AlignmentInfo *alignNonTerm) {
m_alignNonTerm = alignNonTerm;
}
void SetAlignTerm(const AlignmentInfo::CollType &coll);
void SetAlignNonTerm(const AlignmentInfo::CollType &coll);
const AlignmentInfo &GetAlignTerm() const {
return *m_alignTerm;
}
const AlignmentInfo &GetAlignNonTerm() const {
return *m_alignNonTerm;
}
const Phrase *GetRuleSource() const {
return m_ruleSource;
}
// To be set by the FF that needs it, by default the rule source = NULL
// make a copy of the source side of the rule
void SetRuleSource(const Phrase &ruleSource) const;
void SetProperties(const StringPiece &str);
void SetProperty(const std::string &key, const std::string &value);
bool GetProperty(const std::string &key, boost::shared_ptr<PhraseProperty> &value) const;
void Merge(const TargetPhrase &copy, const std::vector<FactorType>& factorVec);
bool operator< (const TargetPhrase &compare) const; // NOT IMPLEMENTED
bool operator== (const TargetPhrase &compare) const; // NOT IMPLEMENTED
TO_STRING();
};
void swap(TargetPhrase &first, TargetPhrase &second);
std::ostream& operator<<(std::ostream&, const TargetPhrase&);
/**
* Hasher that looks at source and target phrase.
**/
struct TargetPhraseHasher {
inline size_t operator()(const TargetPhrase& targetPhrase) const {
size_t seed = 0;
boost::hash_combine(seed, targetPhrase);
boost::hash_combine(seed, targetPhrase.GetAlignTerm());
boost::hash_combine(seed, targetPhrase.GetAlignNonTerm());
return seed;
}
};
struct TargetPhraseComparator {
inline bool operator()(const TargetPhrase& lhs, const TargetPhrase& rhs) const {
return lhs.Compare(rhs) == 0 &&
lhs.GetAlignTerm() == rhs.GetAlignTerm() &&
lhs.GetAlignNonTerm() == rhs.GetAlignNonTerm();
}
};
}
#endif