mosesdecoder/moses/Hypothesis.cpp

// $Id$
// vim:tabstop=2
/***********************************************************************
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
***********************************************************************/

#include <iostream>
#include <limits>
#include <vector>
#include <algorithm>

#include "TranslationOption.h"
#include "TranslationOptionCollection.h"
#include "Hypothesis.h"
#include "Util.h"
#include "SquareMatrix.h"
#include "StaticData.h"
#include "InputType.h"
#include "Manager.h"
#include "IOWrapper.h"
#include "moses/FF/FFState.h"
#include "moses/FF/StatefulFeatureFunction.h"
#include "moses/FF/StatelessFeatureFunction.h"

#include <boost/foreach.hpp>

using namespace std;

namespace Moses
{

#ifdef USE_HYPO_POOL
  ObjectPool<Hypothesis> Hypothesis::s_objectPool("Hypothesis", 300000);
#endif

  Hypothesis::
  Hypothesis(Manager& manager, InputType const& source, const TranslationOption &initialTransOpt)
    : m_prevHypo(NULL)
    , m_sourceCompleted(source.GetSize(), manager.GetSource().m_sourceCompleted)
    , m_sourceInput(source)
    , m_currSourceWordsRange(
			     m_sourceCompleted.GetFirstGapPos()>0 ? 0 : NOT_FOUND,
			     m_sourceCompleted.GetFirstGapPos()>0 ? m_sourceCompleted.GetFirstGapPos()-1 : NOT_FOUND)
    , m_currTargetWordsRange(NOT_FOUND, NOT_FOUND)
    , m_wordDeleted(false)
    , m_totalScore(0.0f)
    , m_futureScore(0.0f)
    , m_ffStates(StatefulFeatureFunction::GetStatefulFeatureFunctions().size())
    , m_arcList(NULL)
    , m_transOpt(initialTransOpt)
    , m_manager(manager)
    , m_id(m_manager.GetNextHypoId())
  {
    // used for initial seeding of trans process
    // initialize scores
    //_hash_computed = false;
    //s_HypothesesCreated = 1;
    const vector<const StatefulFeatureFunction*>& ffs = StatefulFeatureFunction::GetStatefulFeatureFunctions();
    for (unsigned i = 0; i < ffs.size(); ++i)
      m_ffStates[i] = ffs[i]->EmptyHypothesisState(source);
    m_manager.GetSentenceStats().AddCreated();
  }

  /***
   * continue prevHypo by appending the phrases in transOpt
   */
  Hypothesis::
  Hypothesis(const Hypothesis &prevHypo, const TranslationOption &transOpt)
    : m_prevHypo(&prevHypo)
    , m_sourceCompleted(prevHypo.m_sourceCompleted )
    , m_sourceInput(prevHypo.m_sourceInput)
    , m_currSourceWordsRange(transOpt.GetSourceWordsRange())
    , m_currTargetWordsRange(prevHypo.m_currTargetWordsRange.GetEndPos() + 1,
			     prevHypo.m_currTargetWordsRange.GetEndPos() 
			     + transOpt.GetTargetPhrase().GetSize())
    , m_wordDeleted(false)
    , m_totalScore(0.0f)
    , m_futureScore(0.0f)
    , m_ffStates(prevHypo.m_ffStates.size())
    , m_arcList(NULL)
    , m_transOpt(transOpt)
    , m_manager(prevHypo.GetManager())
    , m_id(m_manager.GetNextHypoId())
  {
    m_currScoreBreakdown.PlusEquals(transOpt.GetScoreBreakdown());

    // assert that we are not extending our hypothesis by retranslating something
    // that this hypothesis has already translated!
    assert(!m_sourceCompleted.Overlap(m_currSourceWordsRange));

    //_hash_computed = false;
    m_sourceCompleted.SetValue(m_currSourceWordsRange.GetStartPos(), m_currSourceWordsRange.GetEndPos(), true);
    m_wordDeleted = transOpt.IsDeletionOption();
    m_manager.GetSentenceStats().AddCreated();
  }

  Hypothesis::
  ~Hypothesis()
  {
    for (unsigned i = 0; i < m_ffStates.size(); ++i)
      delete m_ffStates[i];

    if (m_arcList) {
      ArcList::iterator iter;
      for (iter = m_arcList->begin() ; iter != m_arcList->end() ; ++iter) {
	FREEHYPO(*iter);
      }
      m_arcList->clear();

      delete m_arcList;
      m_arcList = NULL;
    }
  }

  void 
  Hypothesis::
  AddArc(Hypothesis *loserHypo)
  {
    if (!m_arcList) {
      if (loserHypo->m_arcList) { // we don't have an arcList, but loser does
	this->m_arcList = loserHypo->m_arcList;  // take ownership, we'll delete
	loserHypo->m_arcList = 0;                // prevent a double deletion
      } else {
	this->m_arcList = new ArcList();
      }
    } else {
      if (loserHypo->m_arcList) {  // both have an arc list: merge. delete loser
	size_t my_size = m_arcList->size();
	size_t add_size = loserHypo->m_arcList->size();
	this->m_arcList->resize(my_size + add_size, 0);
	std::memcpy(&(*m_arcList)[0] + my_size, &(*loserHypo->m_arcList)[0], add_size * sizeof(Hypothesis *));
	delete loserHypo->m_arcList;
	loserHypo->m_arcList = 0;
      } else { // loserHypo doesn't have any arcs
	// DO NOTHING
      }
    }
    m_arcList->push_back(loserHypo);
  }

  /***
   * return the subclass of Hypothesis most appropriate to the given translation option
   */
  Hypothesis* 
  Hypothesis::
  CreateNext(const TranslationOption &transOpt) const
  {
    return Create(*this, transOpt);
  }

  /***
   * return the subclass of Hypothesis most appropriate to the given translation option
   */
  Hypothesis* 
  Hypothesis::
  Create(const Hypothesis &prevHypo, const TranslationOption &transOpt)
  {

#ifdef USE_HYPO_POOL
    Hypothesis *ptr = s_objectPool.getPtr();
    return new(ptr) Hypothesis(prevHypo, transOpt);
#else
    return new Hypothesis(prevHypo, transOpt);
#endif
  }
  /***
   * return the subclass of Hypothesis most appropriate to the given target phrase
   */

  Hypothesis* 
  Hypothesis::
  Create(Manager& manager, InputType const& m_source, 
	 const TranslationOption &initialTransOpt)
  {
#ifdef USE_HYPO_POOL
    Hypothesis *ptr = s_objectPool.getPtr();
    return new(ptr) Hypothesis(manager, m_source, initialTransOpt);
#else
    return new Hypothesis(manager, m_source, initialTransOpt);
#endif
  }

  /** check, if two hypothesis can be recombined.
      this is actually a sorting function that allows us to
      keep an ordered list of hypotheses. This makes recombination
      much quicker.
  */
  int 
  Hypothesis::
  RecombineCompare(const Hypothesis &compare) const
  {
    // -1 = this < compare
    // +1 = this > compare
    // 0	= this ==compare
    int comp = m_sourceCompleted.Compare(compare.m_sourceCompleted);
    if (comp != 0)
      return comp;

    for (unsigned i = 0; i < m_ffStates.size(); ++i) {
      if (m_ffStates[i] == NULL || compare.m_ffStates[i] == NULL) {
	comp = m_ffStates[i] - compare.m_ffStates[i];
      } else {
	comp = m_ffStates[i]->Compare(*compare.m_ffStates[i]);
      }
      if (comp != 0) return comp;
    }

    return 0;
  }

  void 
  Hypothesis::
  EvaluateWhenApplied(StatefulFeatureFunction const& sfff,
		      int state_idx)
  {
    const StaticData &staticData = StaticData::Instance();
    if (! staticData.IsFeatureFunctionIgnored( sfff )) 
      {
	m_ffStates[state_idx] 
	  = sfff.EvaluateWhenApplied
	  (*this, m_prevHypo ? m_prevHypo->m_ffStates[state_idx] : NULL,
	   &m_currScoreBreakdown);
    }
  }

  void 
  Hypothesis::
  EvaluateWhenApplied(const StatelessFeatureFunction& slff)
  {
    const StaticData &staticData = StaticData::Instance();
    if (! staticData.IsFeatureFunctionIgnored( slff )) {
      slff.EvaluateWhenApplied(*this, &m_currScoreBreakdown);
    }
  }

  /***
   * calculate the logarithm of our total translation score (sum up components)
   */
  void 
  Hypothesis::
  EvaluateWhenApplied(const SquareMatrix &futureScore)
  {
    IFVERBOSE(2) {
      m_manager.GetSentenceStats().StartTimeOtherScore();
    }
    // some stateless score producers cache their values in the translation
    // option: add these here
    // language model scores for n-grams completely contained within a target
    // phrase are also included here

    // compute values of stateless feature functions that were not
    // cached in the translation option
    const vector<const StatelessFeatureFunction*>& sfs =
      StatelessFeatureFunction::GetStatelessFeatureFunctions();
    for (unsigned i = 0; i < sfs.size(); ++i) {
      const StatelessFeatureFunction &ff = *sfs[i];
      EvaluateWhenApplied(ff);
    }

    const vector<const StatefulFeatureFunction*>& ffs =
      StatefulFeatureFunction::GetStatefulFeatureFunctions();
    for (unsigned i = 0; i < ffs.size(); ++i) {
      const StatefulFeatureFunction &ff = *ffs[i];
      const StaticData &staticData = StaticData::Instance();
      if (! staticData.IsFeatureFunctionIgnored(ff)) {
	m_ffStates[i] = ff.EvaluateWhenApplied(*this,
					       m_prevHypo ? m_prevHypo->m_ffStates[i] : NULL,
					       &m_currScoreBreakdown);
      }
    }

    IFVERBOSE(2) {
      m_manager.GetSentenceStats().StopTimeOtherScore();
      m_manager.GetSentenceStats().StartTimeEstimateScore();
    }

    // FUTURE COST
    m_futureScore = futureScore.CalcFutureScore( m_sourceCompleted );

    // TOTAL
    m_totalScore = m_currScoreBreakdown.GetWeightedScore() + m_futureScore;
    if (m_prevHypo) m_totalScore += m_prevHypo->GetScore();

    IFVERBOSE(2) {
      m_manager.GetSentenceStats().StopTimeEstimateScore();
    }
  }

  const Hypothesis* Hypothesis::GetPrevHypo()const
  {
    return m_prevHypo;
  }

  /**
   * print hypothesis information for pharaoh-style logging
   */
  void 
  Hypothesis::
  PrintHypothesis() const
  {
    if (!m_prevHypo) {
      TRACE_ERR(endl << "NULL hypo" << endl);
      return;
    }
    TRACE_ERR(endl << "creating hypothesis "<< m_id <<" from "<< m_prevHypo->m_id<<" ( ");
    int end = (int)(m_prevHypo->GetCurrTargetPhrase().GetSize()-1);
    int start = end-1;
    if ( start < 0 ) start = 0;
    if ( m_prevHypo->m_currTargetWordsRange.GetStartPos() == NOT_FOUND ) {
      TRACE_ERR( "<s> ");
    } else {
      TRACE_ERR( "... ");
    }
    if (end>=0) {
      WordsRange range(start, end);
      TRACE_ERR( m_prevHypo->GetCurrTargetPhrase().GetSubString(range) << " ");
    }
    TRACE_ERR( ")"<<endl);
    TRACE_ERR( "\tbase score "<< (m_prevHypo->m_totalScore - m_prevHypo->m_futureScore) <<endl);
    TRACE_ERR( "\tcovering "<<m_currSourceWordsRange.GetStartPos()<<"-"<<m_currSourceWordsRange.GetEndPos()
	       <<": " << m_transOpt.GetInputPath().GetPhrase() << endl);

    TRACE_ERR( "\ttranslated as: "<<(Phrase&) GetCurrTargetPhrase()<<endl); // <<" => translation cost "<<m_score[ScoreType::PhraseTrans];

    if (m_wordDeleted) TRACE_ERR( "\tword deleted"<<endl);
    //	TRACE_ERR( "\tdistance: "<<GetCurrSourceWordsRange().CalcDistortion(m_prevHypo->GetCurrSourceWordsRange())); // << " => distortion cost "<<(m_score[ScoreType::Distortion]*weightDistortion)<<endl;
    //	TRACE_ERR( "\tlanguage model cost "); // <<m_score[ScoreType::LanguageModelScore]<<endl;
    //	TRACE_ERR( "\tword penalty "); // <<(m_score[ScoreType::WordPenalty]*weightWordPenalty)<<endl;
    TRACE_ERR( "\tscore "<<m_totalScore - m_futureScore<<" + future cost "<<m_futureScore<<" = "<<m_totalScore<<endl);
    TRACE_ERR(  "\tunweighted feature scores: " << m_currScoreBreakdown << endl);
    //PrintLMScores();
  }

  void 
  Hypothesis::
  CleanupArcList()
  {
    // point this hypo's main hypo to itself
    SetWinningHypo(this);

    if (!m_arcList) return;

    /* keep only number of arcs we need to create all n-best paths.
     * However, may not be enough if only unique candidates are needed,
     * so we'll keep all of arc list if nedd distinct n-best list
     */
    const StaticData &staticData = StaticData::Instance();
    size_t nBestSize = staticData.GetNBestSize();
    bool distinctNBest = (staticData.GetDistinctNBest() || 
			  staticData.GetLatticeSamplesSize() ||  
			  staticData.UseMBR() || 
			  staticData.GetOutputSearchGraph() || 
			  staticData.GetOutputSearchGraphSLF() || 
			  staticData.GetOutputSearchGraphHypergraph() || 
			  staticData.UseLatticeMBR());

    if (!distinctNBest && m_arcList->size() > nBestSize * 5) 
      {
	// prune arc list only if there too many arcs
	NTH_ELEMENT4(m_arcList->begin(), m_arcList->begin() + nBestSize - 1,
		     m_arcList->end(), CompareHypothesisTotalScore());
	
	// delete bad ones
	ArcList::iterator iter;
	for (iter = m_arcList->begin() + nBestSize; iter != m_arcList->end() ; ++iter) 
	  FREEHYPO(*iter);
	m_arcList->erase(m_arcList->begin() + nBestSize, m_arcList->end());
      }
    
    // set all arc's main hypo variable to this hypo
    ArcList::iterator iter = m_arcList->begin();
    for (; iter != m_arcList->end() ; ++iter) {
      Hypothesis *arc = *iter;
      arc->SetWinningHypo(this);
    }
  }

  TargetPhrase const&
  Hypothesis::
  GetCurrTargetPhrase() const
  { return m_transOpt.GetTargetPhrase(); }

  void 
  Hypothesis::
  GetOutputPhrase(Phrase &out) const
  {
    if (m_prevHypo != NULL) 
      m_prevHypo->GetOutputPhrase(out);
    out.Append(GetCurrTargetPhrase());
  }
  
  TO_STRING_BODY(Hypothesis)

  // friend
  ostream& operator<<(ostream& out, const Hypothesis& hypo)
  {
    hypo.ToStream(out);
    // words bitmap
    out << "[" << hypo.m_sourceCompleted << "] ";

    // scores
    out << " [total=" << hypo.GetTotalScore() << "]";
    out << " " << hypo.GetScoreBreakdown();

    // alignment
    out << " " << hypo.GetCurrTargetPhrase().GetAlignNonTerm();

    return out;
  }


  std::string 
  Hypothesis::
  GetSourcePhraseStringRep(const vector<FactorType> factorsToPrint) const
  { return m_transOpt.GetInputPath().GetPhrase().GetStringRep(factorsToPrint); }

  std::string 
  Hypothesis::
  GetTargetPhraseStringRep(const vector<FactorType> factorsToPrint) const
  { return (m_prevHypo ? GetCurrTargetPhrase().GetStringRep(factorsToPrint) : ""); }

  std::string 
  Hypothesis::
  GetSourcePhraseStringRep() const
  {
    vector<FactorType> allFactors(MAX_NUM_FACTORS);
    for(size_t i=0; i < MAX_NUM_FACTORS; i++) 
      allFactors[i] = i;
    return GetSourcePhraseStringRep(allFactors);
  }

  std::string 
  Hypothesis::
  GetTargetPhraseStringRep() const
  {
    vector<FactorType> allFactors(MAX_NUM_FACTORS);
    for(size_t i=0; i < MAX_NUM_FACTORS; i++) 
      allFactors[i] = i;
    return GetTargetPhraseStringRep(allFactors);
  }

  void 
  Hypothesis::
  OutputAlignment(std::ostream &out) const
  {
    std::vector<const Hypothesis *> edges;
    const Hypothesis *currentHypo = this;
    while (currentHypo) {
      edges.push_back(currentHypo);
      currentHypo = currentHypo->GetPrevHypo();
    }
    
    OutputAlignment(out, edges);
    
  }
  
  void 
  Hypothesis::
  OutputAlignment(ostream &out, const vector<const Hypothesis *> &edges)
  {
    size_t targetOffset = 0;
    
    for (int currEdge = (int)edges.size() - 1 ; currEdge >= 0 ; currEdge--) {
      const Hypothesis &edge = *edges[currEdge];
      const TargetPhrase &tp = edge.GetCurrTargetPhrase();
      size_t sourceOffset = edge.GetCurrSourceWordsRange().GetStartPos();
      
      OutputAlignment(out, tp.GetAlignTerm(), sourceOffset, targetOffset);
      
      targetOffset += tp.GetSize();
    }
    // Used by --print-alignment-info, so no endl
  }

  void 
  Hypothesis::
  OutputAlignment(ostream &out, const AlignmentInfo &ai, 
		  size_t sourceOffset, size_t targetOffset)
  {
    typedef std::vector< const std::pair<size_t,size_t>* > AlignVec;
    AlignVec alignments = ai.GetSortedAlignments();

    AlignVec::const_iterator it;
    for (it = alignments.begin(); it != alignments.end(); ++it) {
      const std::pair<size_t,size_t> &alignment = **it;
      out << alignment.first + sourceOffset << "-" << alignment.second + targetOffset << " ";
    }
    
  }

  void 
  Hypothesis::
  OutputInput(std::vector<const Phrase*>& map, const Hypothesis* hypo)
  {
    if (!hypo->GetPrevHypo()) return;
    OutputInput(map, hypo->GetPrevHypo());
    map[hypo->GetCurrSourceWordsRange().GetStartPos()] 
      = &hypo->GetTranslationOption().GetInputPath().GetPhrase();
  }

  void 
  Hypothesis::
  OutputInput(std::ostream& os) const
  {
    size_t len = this->GetInput().GetSize();
    std::vector<const Phrase*> inp_phrases(len, 0);
    OutputInput(inp_phrases, this);
    for (size_t i=0; i<len; ++i)
      if (inp_phrases[i]) os << *inp_phrases[i];
  }
  
  void 
  Hypothesis::
  OutputBestSurface(std::ostream &out, const std::vector<FactorType> &outputFactorOrder,
		    char reportSegmentation, bool reportAllFactors) const
  {
    if (m_prevHypo) 
      { // recursively retrace this best path through the lattice, starting from the end of the hypothesis sentence
	m_prevHypo->OutputBestSurface(out, outputFactorOrder, reportSegmentation, reportAllFactors);
      }
    OutputSurface(out, *this, outputFactorOrder, reportSegmentation, reportAllFactors);
  }

  //////////////////////////////////////////////////////////////////////////
  /***
   * print surface factor only for the given phrase
   */
  void 
  Hypothesis::
  OutputSurface(std::ostream &out, const Hypothesis &edge, 
		const std::vector<FactorType> &outputFactorOrder,
		char reportSegmentation, bool reportAllFactors) const
  {
    UTIL_THROW_IF2(outputFactorOrder.size() == 0,
		   "Must specific at least 1 output factor");
    const TargetPhrase& phrase = edge.GetCurrTargetPhrase();
    bool markUnknown = StaticData::Instance().GetMarkUnknown();
    if (reportAllFactors == true) {
      out << phrase;
    } else {
      FactorType placeholderFactor = StaticData::Instance().GetPlaceholderFactor();

      std::map<size_t, const Factor*> placeholders;
      if (placeholderFactor != NOT_FOUND) {
	// creates map of target position -> factor for placeholders
	placeholders = GetPlaceholders(edge, placeholderFactor);
      }

      size_t size = phrase.GetSize();
      for (size_t pos = 0 ; pos < size ; pos++) {
	const Factor *factor = phrase.GetFactor(pos, outputFactorOrder[0]);

	if (placeholders.size()) {
	  // do placeholders
	  std::map<size_t, const Factor*>::const_iterator iter = placeholders.find(pos);
	  if (iter != placeholders.end()) {
	    factor = iter->second;
	  }
	}

	UTIL_THROW_IF2(factor == NULL,
		       "No factor 0 at position " << pos);

	//preface surface form with UNK if marking unknowns
	const Word &word = phrase.GetWord(pos);
	if(markUnknown && word.IsOOV()) {
	  out << "UNK" << *factor;
	} else {
	  out << *factor;
	}

	for (size_t i = 1 ; i < outputFactorOrder.size() ; i++) {
	  const Factor *factor = phrase.GetFactor(pos, outputFactorOrder[i]);
	  UTIL_THROW_IF2(factor == NULL,
			 "No factor " << i << " at position " << pos);

	  out << "|" << *factor;
	}
	out << " ";
      }
    }

    // trace ("report segmentation") option "-t" / "-tt"
    if (reportSegmentation > 0 && phrase.GetSize() > 0) {
      const WordsRange &sourceRange = edge.GetCurrSourceWordsRange();
      const int sourceStart = sourceRange.GetStartPos();
      const int sourceEnd = sourceRange.GetEndPos();
      out << "|" << sourceStart << "-" << sourceEnd;    // enriched "-tt"
      if (reportSegmentation == 2) {
	out << ",wa=";
	const AlignmentInfo &ai = edge.GetCurrTargetPhrase().GetAlignTerm();
	Hypothesis::OutputAlignment(out, ai, 0, 0);
	out << ",total=";
	out << edge.GetScore() - edge.GetPrevHypo()->GetScore();
	out << ",";
	ScoreComponentCollection scoreBreakdown(edge.GetScoreBreakdown());
	scoreBreakdown.MinusEquals(edge.GetPrevHypo()->GetScoreBreakdown());
	scoreBreakdown.OutputAllFeatureScores(out);
      }
      out << "| ";
    }
  }

  std::map<size_t, const Factor*> 
  Hypothesis::
  GetPlaceholders(const Hypothesis &hypo, FactorType placeholderFactor) const
  {
    const InputPath &inputPath = hypo.GetTranslationOption().GetInputPath();
    const Phrase &inputPhrase = inputPath.GetPhrase();
    
    std::map<size_t, const Factor*> ret;
    
    for (size_t sourcePos = 0; sourcePos < inputPhrase.GetSize(); ++sourcePos) {
      const Factor *factor = inputPhrase.GetFactor(sourcePos, placeholderFactor);
      if (factor) {
	std::set<size_t> targetPos = hypo.GetTranslationOption().GetTargetPhrase().GetAlignTerm().GetAlignmentsForSource(sourcePos);
	UTIL_THROW_IF2(targetPos.size() != 1,
		       "Placeholder should be aligned to 1, and only 1, word");
	ret[*targetPos.begin()] = factor;
      }
    }
    
    return ret;
  }

#ifdef HAVE_XMLRPC_C
  void
  Hypothesis::
  OutputLocalWordAlignment(vector<xmlrpc_c::value>& dest) const
  {
    using namespace std;
    WordsRange const& src = this->GetCurrSourceWordsRange();
    WordsRange const& trg = this->GetCurrTargetWordsRange();
    
    vector<pair<size_t,size_t> const* > a 
      = this->GetCurrTargetPhrase().GetAlignTerm().GetSortedAlignments();
    typedef pair<size_t,size_t> item;
    map<string, xmlrpc_c::value> M;
    BOOST_FOREACH(item const* p, a)
      {
	M["source-word"] = xmlrpc_c::value_int(src.GetStartPos() + p->first);
	M["target-word"] = xmlrpc_c::value_int(trg.GetStartPos() + p->second);
	dest.push_back(xmlrpc_c::value_struct(M));
      }
  }

  void 
  Hypothesis::
  OutputWordAlignment(vector<xmlrpc_c::value>& out) const
  {
    vector<Hypothesis const*> tmp;
    for (Hypothesis const* h = this; h; h = h->GetPrevHypo())
      tmp.push_back(h);
    for (size_t i = tmp.size(); i-- > 0;)
      tmp[i]->OutputLocalWordAlignment(out);
  }

#endif
  
  
}