2013-11-05 18:37:56 +04:00
|
|
|
// $Id$
|
|
|
|
#pragma once
|
|
|
|
|
|
|
|
#include <vector>
|
2013-12-16 18:17:56 +04:00
|
|
|
#include "Implementation.h"
|
|
|
|
#include "moses/Hypothesis.h"
|
2013-11-05 18:37:56 +04:00
|
|
|
|
2013-11-11 21:19:44 +04:00
|
|
|
namespace DALM
|
|
|
|
{
|
|
|
|
class Logger;
|
|
|
|
class Vocabulary;
|
2013-12-16 18:17:56 +04:00
|
|
|
class State;
|
2013-11-11 21:19:44 +04:00
|
|
|
class LM;
|
2014-06-03 11:13:32 +04:00
|
|
|
union Fragment;
|
|
|
|
class Gap;
|
2013-11-11 23:49:00 +04:00
|
|
|
|
|
|
|
typedef unsigned int VocabId;
|
2013-11-11 21:19:44 +04:00
|
|
|
}
|
|
|
|
|
2013-11-05 18:37:56 +04:00
|
|
|
namespace Moses
|
|
|
|
{
|
2013-11-11 23:49:00 +04:00
|
|
|
class Factor;
|
2014-06-03 11:13:32 +04:00
|
|
|
class DALMChartState;
|
2013-11-05 18:37:56 +04:00
|
|
|
|
2013-12-16 18:17:56 +04:00
|
|
|
class LanguageModelDALM : public LanguageModel
|
2013-11-05 18:37:56 +04:00
|
|
|
{
|
2013-12-16 18:17:56 +04:00
|
|
|
public:
|
|
|
|
LanguageModelDALM(const std::string &line);
|
|
|
|
virtual ~LanguageModelDALM();
|
2014-01-15 19:42:02 +04:00
|
|
|
|
2015-12-10 06:17:36 +03:00
|
|
|
void Load(AllOptions::ptr const& opts);
|
2013-12-16 18:17:56 +04:00
|
|
|
|
|
|
|
virtual const FFState *EmptyHypothesisState(const InputType &/*input*/) const;
|
|
|
|
|
|
|
|
virtual void CalcScore(const Phrase &phrase, float &fullScore, float &ngramScore, size_t &oovCount) const;
|
|
|
|
|
2014-07-10 02:41:08 +04:00
|
|
|
virtual FFState *EvaluateWhenApplied(const Hypothesis &hypo, const FFState *ps, ScoreComponentCollection *out) const;
|
2013-12-16 18:17:56 +04:00
|
|
|
|
2014-07-10 02:54:16 +04:00
|
|
|
virtual FFState *EvaluateWhenApplied(const ChartHypothesis& hypo, int featureID, ScoreComponentCollection *out) const;
|
2013-12-16 18:17:56 +04:00
|
|
|
|
|
|
|
virtual bool IsUseable(const FactorMask &mask) const;
|
|
|
|
|
2014-01-15 19:42:02 +04:00
|
|
|
virtual void SetParameter(const std::string& key, const std::string& value);
|
|
|
|
|
2013-11-05 18:37:56 +04:00
|
|
|
protected:
|
2013-12-16 18:17:56 +04:00
|
|
|
const Factor *m_beginSentenceFactor;
|
|
|
|
|
|
|
|
FactorType m_factorType;
|
|
|
|
|
|
|
|
std::string m_filePath;
|
|
|
|
size_t m_nGramOrder; //! max n-gram length contained in this LM
|
2014-06-08 11:44:59 +04:00
|
|
|
size_t m_ContextSize;
|
2013-12-16 18:17:56 +04:00
|
|
|
|
2014-01-15 19:42:02 +04:00
|
|
|
DALM::Logger *m_logger;
|
|
|
|
DALM::Vocabulary *m_vocab;
|
|
|
|
DALM::LM *m_lm;
|
|
|
|
DALM::VocabId wid_start, wid_end;
|
2013-11-11 23:49:00 +04:00
|
|
|
|
2014-04-28 21:18:38 +04:00
|
|
|
mutable std::vector<DALM::VocabId> m_vocabMap;
|
2013-11-18 17:54:40 +04:00
|
|
|
|
2014-01-15 19:42:02 +04:00
|
|
|
void CreateVocabMapping(const std::string &wordstxt);
|
|
|
|
DALM::VocabId GetVocabId(const Factor *factor) const;
|
2013-11-11 23:49:00 +04:00
|
|
|
|
2013-12-16 18:17:56 +04:00
|
|
|
private:
|
|
|
|
// Convert last words of hypothesis into vocab ids, returning an end pointer.
|
|
|
|
DALM::VocabId *LastIDs(const Hypothesis &hypo, DALM::VocabId *indices) const {
|
|
|
|
DALM::VocabId *index = indices;
|
|
|
|
DALM::VocabId *end = indices + m_nGramOrder - 1;
|
|
|
|
int position = hypo.GetCurrTargetWordsRange().GetEndPos();
|
|
|
|
for (; ; ++index, --position) {
|
|
|
|
if (index == end) return index;
|
|
|
|
if (position == -1) {
|
|
|
|
*index = wid_start;
|
|
|
|
return index + 1;
|
|
|
|
}
|
|
|
|
*index = GetVocabId(hypo.GetWord(position).GetFactor(m_factorType));
|
|
|
|
}
|
|
|
|
}
|
2014-06-03 11:13:32 +04:00
|
|
|
|
2014-06-08 11:44:59 +04:00
|
|
|
void EvaluateTerminal(
|
|
|
|
const Word &word,
|
|
|
|
float &hypoScore,
|
|
|
|
DALMChartState *newState,
|
|
|
|
DALM::State &state,
|
|
|
|
DALM::Fragment *prefixFragments,
|
|
|
|
unsigned char &prefixLength
|
|
|
|
) const;
|
|
|
|
|
|
|
|
void EvaluateNonTerminal(
|
|
|
|
const Word &word,
|
|
|
|
float &hypoScore,
|
|
|
|
DALMChartState *newState,
|
|
|
|
DALM::State &state,
|
|
|
|
DALM::Fragment *prefixFragments,
|
|
|
|
unsigned char &prefixLength,
|
|
|
|
const DALMChartState *prevState,
|
2015-01-12 07:05:27 +03:00
|
|
|
size_t prevTargetPhraseLength
|
2014-06-08 11:44:59 +04:00
|
|
|
) const;
|
2013-11-05 18:37:56 +04:00
|
|
|
};
|
|
|
|
|
|
|
|
}
|
2013-12-16 18:17:56 +04:00
|
|
|
|