2006-07-04 22:04:38 +04:00
|
|
|
// $Id$
|
|
|
|
|
|
|
|
/***********************************************************************
|
|
|
|
Moses - factored phrase-based language decoder
|
|
|
|
Copyright (c) 2006 University of Edinburgh
|
|
|
|
All rights reserved.
|
|
|
|
|
|
|
|
Redistribution and use in source and binary forms, with or without modification,
|
|
|
|
are permitted provided that the following conditions are met:
|
|
|
|
|
|
|
|
* Redistributions of source code must retain the above copyright notice,
|
|
|
|
this list of conditions and the following disclaimer.
|
|
|
|
* Redistributions in binary form must reproduce the above copyright notice,
|
|
|
|
this list of conditions and the following disclaimer in the documentation
|
|
|
|
and/or other materials provided with the distribution.
|
|
|
|
* Neither the name of the University of Edinburgh nor the names of its contributors
|
|
|
|
may be used to endorse or promote products derived from this software
|
|
|
|
without specific prior written permission.
|
|
|
|
|
|
|
|
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
|
|
|
|
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
|
|
|
|
THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
|
|
|
|
PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS
|
|
|
|
BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
|
|
|
|
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
|
|
|
|
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
|
|
|
|
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER
|
|
|
|
IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
|
|
|
|
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
|
|
|
|
POSSIBILITY OF SUCH DAMAGE.
|
|
|
|
***********************************************************************/
|
|
|
|
|
|
|
|
// example file on how to use moses library
|
|
|
|
|
|
|
|
#include <iostream>
|
|
|
|
#include "TypeDef.h"
|
|
|
|
#include "Util.h"
|
2006-11-22 02:06:30 +03:00
|
|
|
#include "IOStream.h"
|
2006-07-04 22:04:38 +04:00
|
|
|
#include "Hypothesis.h"
|
2006-07-25 03:48:18 +04:00
|
|
|
#include "WordsRange.h"
|
2007-05-16 17:08:55 +04:00
|
|
|
#include "TrellisPathList.h"
|
2006-07-28 05:15:09 +04:00
|
|
|
#include "StaticData.h"
|
2006-07-29 05:11:45 +04:00
|
|
|
#include "DummyScoreProducers.h"
|
2006-11-22 02:06:30 +03:00
|
|
|
#include "InputFileStream.h"
|
2006-07-04 22:04:38 +04:00
|
|
|
|
|
|
|
using namespace std;
|
|
|
|
|
2006-11-22 02:06:30 +03:00
|
|
|
IOStream::IOStream(
|
2006-08-02 00:02:39 +04:00
|
|
|
const vector<FactorType> &inputFactorOrder
|
|
|
|
, const vector<FactorType> &outputFactorOrder
|
2006-08-11 00:42:34 +04:00
|
|
|
, const FactorMask &inputFactorUsed
|
2006-07-04 22:04:38 +04:00
|
|
|
, size_t nBestSize
|
|
|
|
, const string &nBestFilePath)
|
2006-08-02 00:02:39 +04:00
|
|
|
:m_inputFactorOrder(inputFactorOrder)
|
|
|
|
,m_outputFactorOrder(outputFactorOrder)
|
2006-07-04 22:04:38 +04:00
|
|
|
,m_inputFactorUsed(inputFactorUsed)
|
2006-11-22 02:06:30 +03:00
|
|
|
,m_inputFile(NULL)
|
|
|
|
,m_inputStream(&std::cin)
|
2007-01-21 16:38:03 +03:00
|
|
|
,m_nBestStream(NULL)
|
2006-07-04 22:04:38 +04:00
|
|
|
{
|
2007-01-08 17:46:48 +03:00
|
|
|
m_surpressSingleBestOutput = false;
|
2006-07-04 22:04:38 +04:00
|
|
|
if (nBestSize > 0)
|
|
|
|
{
|
2007-01-08 17:46:48 +03:00
|
|
|
if (nBestFilePath == "-")
|
|
|
|
{
|
2007-01-21 16:38:03 +03:00
|
|
|
m_nBestStream = &std::cout;
|
2007-01-08 17:46:48 +03:00
|
|
|
m_surpressSingleBestOutput = true;
|
2007-01-21 16:38:03 +03:00
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
std::ofstream *nBestFile = new std::ofstream;
|
|
|
|
m_nBestStream = nBestFile;
|
|
|
|
nBestFile->open(nBestFilePath.c_str());
|
2007-01-08 17:46:48 +03:00
|
|
|
}
|
2006-07-04 22:04:38 +04:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2006-11-22 02:06:30 +03:00
|
|
|
IOStream::IOStream(const std::vector<FactorType> &inputFactorOrder
|
|
|
|
, const std::vector<FactorType> &outputFactorOrder
|
|
|
|
, const FactorMask &inputFactorUsed
|
|
|
|
, size_t nBestSize
|
|
|
|
, const std::string &nBestFilePath
|
|
|
|
, const std::string &inputFilePath)
|
|
|
|
:m_inputFactorOrder(inputFactorOrder)
|
|
|
|
,m_outputFactorOrder(outputFactorOrder)
|
|
|
|
,m_inputFactorUsed(inputFactorUsed)
|
|
|
|
,m_inputFilePath(inputFilePath)
|
|
|
|
,m_inputFile(new InputFileStream(inputFilePath))
|
2007-01-21 16:38:03 +03:00
|
|
|
,m_nBestStream(NULL)
|
2006-11-22 02:06:30 +03:00
|
|
|
{
|
2007-01-08 17:46:48 +03:00
|
|
|
m_surpressSingleBestOutput = false;
|
2006-11-22 02:06:30 +03:00
|
|
|
m_inputStream = m_inputFile;
|
|
|
|
|
|
|
|
if (nBestSize > 0)
|
|
|
|
{
|
2007-01-08 17:46:48 +03:00
|
|
|
if (nBestFilePath == "-")
|
|
|
|
{
|
2007-01-21 16:38:03 +03:00
|
|
|
m_nBestStream = &std::cout;
|
2007-01-08 17:46:48 +03:00
|
|
|
m_surpressSingleBestOutput = true;
|
2007-01-21 16:38:03 +03:00
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
std::ofstream *nBestFile = new std::ofstream;
|
|
|
|
m_nBestStream = nBestFile;
|
|
|
|
nBestFile->open(nBestFilePath.c_str());
|
2007-01-08 17:46:48 +03:00
|
|
|
}
|
2006-11-22 02:06:30 +03:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
IOStream::~IOStream()
|
|
|
|
{
|
|
|
|
if (m_inputFile != NULL)
|
|
|
|
delete m_inputFile;
|
2007-01-21 16:38:03 +03:00
|
|
|
if (m_nBestStream != NULL && !m_surpressSingleBestOutput)
|
|
|
|
{ // outputting n-best to file, rather than stdout. need to close file and delete obj
|
|
|
|
delete m_nBestStream;
|
|
|
|
}
|
2006-11-22 02:06:30 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
InputType*IOStream::GetInput(InputType* inputType)
|
2006-07-04 22:04:38 +04:00
|
|
|
{
|
2007-02-16 21:08:37 +03:00
|
|
|
if(inputType->Read(*m_inputStream, m_inputFactorOrder))
|
2006-11-22 02:06:30 +03:00
|
|
|
{
|
2007-04-18 18:08:46 +04:00
|
|
|
if (long x = inputType->GetTranslationId()) { if (x>=m_translationId) m_translationId = x+1; }
|
|
|
|
else inputType->SetTranslationId(m_translationId++);
|
|
|
|
|
2006-11-22 02:06:30 +03:00
|
|
|
return inputType;
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
delete inputType;
|
|
|
|
return NULL;
|
|
|
|
}
|
2006-07-04 22:04:38 +04:00
|
|
|
}
|
|
|
|
|
2006-08-12 01:04:38 +04:00
|
|
|
/***
|
|
|
|
* print surface factor only for the given phrase
|
|
|
|
*/
|
2006-08-02 00:02:39 +04:00
|
|
|
void OutputSurface(std::ostream &out, const Phrase &phrase, const std::vector<FactorType> &outputFactorOrder, bool reportAllFactors)
|
2006-07-04 22:04:38 +04:00
|
|
|
{
|
2006-08-02 00:12:58 +04:00
|
|
|
assert(outputFactorOrder.size() > 0);
|
2006-08-21 18:24:09 +04:00
|
|
|
if (reportAllFactors == true)
|
2006-08-02 00:02:39 +04:00
|
|
|
{
|
|
|
|
out << phrase;
|
2006-08-21 18:24:09 +04:00
|
|
|
}
|
2006-08-02 00:02:39 +04:00
|
|
|
else
|
|
|
|
{
|
|
|
|
size_t size = phrase.GetSize();
|
|
|
|
for (size_t pos = 0 ; pos < size ; pos++)
|
|
|
|
{
|
2006-08-02 00:12:58 +04:00
|
|
|
const Factor *factor = phrase.GetFactor(pos, outputFactorOrder[0]);
|
|
|
|
out << *factor;
|
|
|
|
|
|
|
|
for (size_t i = 1 ; i < outputFactorOrder.size() ; i++)
|
2006-08-02 00:02:39 +04:00
|
|
|
{
|
|
|
|
const Factor *factor = phrase.GetFactor(pos, outputFactorOrder[i]);
|
2006-08-02 00:12:58 +04:00
|
|
|
out << "|" << *factor;
|
2006-08-02 00:02:39 +04:00
|
|
|
}
|
2006-08-02 00:12:58 +04:00
|
|
|
out << " ";
|
2006-08-02 00:02:39 +04:00
|
|
|
}
|
2006-07-04 22:04:38 +04:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2006-08-02 00:02:39 +04:00
|
|
|
void OutputSurface(std::ostream &out, const Hypothesis *hypo, const std::vector<FactorType> &outputFactorOrder
|
2006-08-30 23:51:07 +04:00
|
|
|
,bool reportSegmentation, bool reportAllFactors)
|
2006-07-04 22:04:38 +04:00
|
|
|
{
|
|
|
|
if ( hypo != NULL)
|
|
|
|
{
|
2006-08-30 23:51:07 +04:00
|
|
|
OutputSurface(out, hypo->GetPrevHypo(), outputFactorOrder, reportSegmentation, reportAllFactors);
|
2007-02-12 14:05:13 +03:00
|
|
|
OutputSurface(out, hypo->GetCurrTargetPhrase(), outputFactorOrder, reportAllFactors);
|
2006-07-25 03:48:18 +04:00
|
|
|
|
2006-08-30 23:51:07 +04:00
|
|
|
if (reportSegmentation == true
|
2007-02-12 14:05:13 +03:00
|
|
|
&& hypo->GetCurrTargetPhrase().GetSize() > 0) {
|
2006-08-21 18:24:09 +04:00
|
|
|
out << "|" << hypo->GetCurrSourceWordsRange().GetStartPos()
|
|
|
|
<< "-" << hypo->GetCurrSourceWordsRange().GetEndPos() << "| ";
|
|
|
|
}
|
2006-07-04 22:04:38 +04:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2006-11-22 02:06:30 +03:00
|
|
|
void IOStream::Backtrack(const Hypothesis *hypo){
|
2006-07-15 01:51:05 +04:00
|
|
|
|
2006-07-17 19:05:00 +04:00
|
|
|
if (hypo->GetPrevHypo() != NULL) {
|
2006-10-17 15:07:17 +04:00
|
|
|
VERBOSE(3,hypo->GetId() << " <= ");
|
2006-07-15 01:51:05 +04:00
|
|
|
Backtrack(hypo->GetPrevHypo());
|
|
|
|
}
|
|
|
|
}
|
2007-05-16 00:54:39 +04:00
|
|
|
|
|
|
|
void IOStream::OutputBestHypo(const std::vector<const Factor*>& mbrBestHypo, long /*translationId*/, bool reportSegmentation, bool reportAllFactors)
|
|
|
|
{
|
|
|
|
for (size_t i = 0 ; i < mbrBestHypo.size() ; i++)
|
|
|
|
{
|
|
|
|
const Factor *factor = mbrBestHypo[i];
|
2008-01-16 23:25:01 +03:00
|
|
|
if (i>0) cout << " ";
|
|
|
|
cout << factor->GetString();
|
2007-05-16 00:54:39 +04:00
|
|
|
}
|
2008-01-16 23:25:01 +03:00
|
|
|
cout << endl;
|
2007-05-16 00:54:39 +04:00
|
|
|
}
|
2006-07-15 01:51:05 +04:00
|
|
|
|
2007-09-28 20:43:33 +04:00
|
|
|
void OutputInput(std::vector<const Phrase*>& map, const Hypothesis* hypo)
|
|
|
|
{
|
|
|
|
if (hypo->GetPrevHypo())
|
|
|
|
{
|
|
|
|
OutputInput(map, hypo->GetPrevHypo());
|
|
|
|
map[hypo->GetCurrSourceWordsRange().GetStartPos()] = hypo->GetSourcePhrase();
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
void OutputInput(std::ostream& os, const Hypothesis* hypo)
|
|
|
|
{
|
|
|
|
size_t len = StaticData::Instance().GetInput()->GetSize();
|
|
|
|
std::vector<const Phrase*> inp_phrases(len, 0);
|
|
|
|
OutputInput(inp_phrases, hypo);
|
|
|
|
for (size_t i=0; i<len; ++i)
|
|
|
|
if (inp_phrases[i]) os << *inp_phrases[i];
|
|
|
|
}
|
|
|
|
|
2006-11-22 02:06:30 +03:00
|
|
|
void IOStream::OutputBestHypo(const Hypothesis *hypo, long /*translationId*/, bool reportSegmentation, bool reportAllFactors)
|
2006-07-04 22:04:38 +04:00
|
|
|
{
|
|
|
|
if (hypo != NULL)
|
|
|
|
{
|
2007-02-23 02:44:38 +03:00
|
|
|
VERBOSE(1,"BEST TRANSLATION: " << *hypo << endl);
|
2006-08-30 23:51:07 +04:00
|
|
|
VERBOSE(3,"Best path: ");
|
2006-07-17 19:05:00 +04:00
|
|
|
Backtrack(hypo);
|
2006-08-30 23:51:07 +04:00
|
|
|
VERBOSE(3,"0" << std::endl);
|
2006-07-15 01:51:05 +04:00
|
|
|
|
2007-01-08 17:46:48 +03:00
|
|
|
if (!m_surpressSingleBestOutput)
|
|
|
|
{
|
2007-09-28 20:43:33 +04:00
|
|
|
if (StaticData::Instance().IsPathRecoveryEnabled()) {
|
|
|
|
OutputInput(cout, hypo);
|
|
|
|
cout << "||| ";
|
|
|
|
}
|
2007-01-08 17:46:48 +03:00
|
|
|
OutputSurface(cout, hypo, m_outputFactorOrder, reportSegmentation, reportAllFactors);
|
|
|
|
cout << endl;
|
|
|
|
}
|
2006-07-04 22:04:38 +04:00
|
|
|
}
|
|
|
|
else
|
2006-07-14 22:13:50 +04:00
|
|
|
{
|
2007-02-23 02:44:38 +03:00
|
|
|
VERBOSE(1, "NO BEST TRANSLATION" << endl);
|
2007-03-14 22:47:51 +03:00
|
|
|
if (!m_surpressSingleBestOutput)
|
|
|
|
{
|
|
|
|
cout << endl;
|
|
|
|
}
|
2006-07-14 22:13:50 +04:00
|
|
|
}
|
2006-07-04 22:04:38 +04:00
|
|
|
}
|
|
|
|
|
2007-05-16 17:08:55 +04:00
|
|
|
void IOStream::OutputNBestList(const TrellisPathList &nBestList, long translationId)
|
2006-07-04 22:04:38 +04:00
|
|
|
{
|
2007-02-16 21:08:37 +03:00
|
|
|
bool labeledOutput = StaticData::Instance().IsLabeledNBestList();
|
2007-02-26 23:59:41 +03:00
|
|
|
bool includeAlignment = StaticData::Instance().NBestIncludesAlignment();
|
2006-08-13 23:08:39 +04:00
|
|
|
|
2007-05-16 17:08:55 +04:00
|
|
|
TrellisPathList::const_iterator iter;
|
2006-07-04 22:04:38 +04:00
|
|
|
for (iter = nBestList.begin() ; iter != nBestList.end() ; ++iter)
|
|
|
|
{
|
2007-05-16 17:08:55 +04:00
|
|
|
const TrellisPath &path = **iter;
|
2006-07-29 02:11:20 +04:00
|
|
|
const std::vector<const Hypothesis *> &edges = path.GetEdges();
|
2006-07-04 22:04:38 +04:00
|
|
|
|
2006-08-04 08:45:48 +04:00
|
|
|
// print the surface factor of the translation
|
2007-01-21 16:38:03 +03:00
|
|
|
*m_nBestStream << translationId << " ||| ";
|
2006-07-04 22:04:38 +04:00
|
|
|
for (int currEdge = (int)edges.size() - 1 ; currEdge >= 0 ; currEdge--)
|
|
|
|
{
|
2006-07-29 02:11:20 +04:00
|
|
|
const Hypothesis &edge = *edges[currEdge];
|
2007-02-12 14:05:13 +03:00
|
|
|
OutputSurface(*m_nBestStream, edge.GetCurrTargetPhrase(), m_outputFactorOrder, false); // false for not reporting all factors
|
2006-07-04 22:04:38 +04:00
|
|
|
}
|
2007-01-21 16:38:03 +03:00
|
|
|
*m_nBestStream << " ||| ";
|
2006-07-04 22:04:38 +04:00
|
|
|
|
2006-08-04 08:45:48 +04:00
|
|
|
// print the scores in a hardwired order
|
|
|
|
// before each model type, the corresponding command-line-like name must be emitted
|
|
|
|
// MERT script relies on this
|
2006-07-29 05:11:45 +04:00
|
|
|
|
|
|
|
// basic distortion
|
2006-08-13 23:08:39 +04:00
|
|
|
if (labeledOutput)
|
2007-01-21 16:38:03 +03:00
|
|
|
*m_nBestStream << "d: ";
|
2007-02-16 21:08:37 +03:00
|
|
|
*m_nBestStream << path.GetScoreBreakdown().GetScoreForProducer(StaticData::Instance().GetDistortionScoreProducer()) << " ";
|
2006-07-04 22:04:38 +04:00
|
|
|
|
2006-08-13 23:51:17 +04:00
|
|
|
// reordering
|
2007-02-16 21:08:37 +03:00
|
|
|
vector<LexicalReordering*> rms = StaticData::Instance().GetReorderModels();
|
2006-08-13 22:11:52 +04:00
|
|
|
if(rms.size() > 0)
|
|
|
|
{
|
|
|
|
vector<LexicalReordering*>::iterator iter;
|
|
|
|
for(iter = rms.begin(); iter != rms.end(); ++iter)
|
|
|
|
{
|
2006-08-13 23:51:17 +04:00
|
|
|
vector<float> scores = path.GetScoreBreakdown().GetScoresForProducer(*iter);
|
|
|
|
for (size_t j = 0; j<scores.size(); ++j)
|
|
|
|
{
|
2007-01-21 16:38:03 +03:00
|
|
|
*m_nBestStream << scores[j] << " ";
|
2006-08-13 23:51:17 +04:00
|
|
|
}
|
2006-08-13 22:11:52 +04:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2006-07-04 22:04:38 +04:00
|
|
|
// lm
|
2007-02-16 21:08:37 +03:00
|
|
|
const LMList& lml = StaticData::Instance().GetAllLM();
|
2006-08-04 08:45:48 +04:00
|
|
|
if (lml.size() > 0) {
|
2006-08-13 23:08:39 +04:00
|
|
|
if (labeledOutput)
|
2007-01-21 16:38:03 +03:00
|
|
|
*m_nBestStream << "lm: ";
|
2006-08-04 08:45:48 +04:00
|
|
|
LMList::const_iterator lmi = lml.begin();
|
|
|
|
for (; lmi != lml.end(); ++lmi) {
|
2007-01-21 16:38:03 +03:00
|
|
|
*m_nBestStream << path.GetScoreBreakdown().GetScoreForProducer(*lmi) << " ";
|
2006-08-04 08:45:48 +04:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
// translation components
|
2007-04-06 04:24:25 +04:00
|
|
|
if (StaticData::Instance().GetInputType()==SentenceInput){
|
2006-11-16 13:43:57 +03:00
|
|
|
// translation components for text input
|
2007-02-16 21:08:37 +03:00
|
|
|
vector<PhraseDictionary*> pds = StaticData::Instance().GetPhraseDictionaries();
|
2006-11-16 13:43:57 +03:00
|
|
|
if (pds.size() > 0) {
|
|
|
|
if (labeledOutput)
|
2007-01-21 16:38:03 +03:00
|
|
|
*m_nBestStream << "tm: ";
|
2006-11-16 13:43:57 +03:00
|
|
|
vector<PhraseDictionary*>::iterator iter;
|
|
|
|
for (iter = pds.begin(); iter != pds.end(); ++iter) {
|
|
|
|
vector<float> scores = path.GetScoreBreakdown().GetScoresForProducer(*iter);
|
|
|
|
for (size_t j = 0; j<scores.size(); ++j)
|
2007-01-21 16:38:03 +03:00
|
|
|
*m_nBestStream << scores[j] << " ";
|
2006-11-16 13:43:57 +03:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
else{
|
|
|
|
// translation components for Confusion Network input
|
|
|
|
// first translation component has GetNumInputScores() scores from the input Confusion Network
|
|
|
|
// at the beginning of the vector
|
2007-02-16 21:08:37 +03:00
|
|
|
vector<PhraseDictionary*> pds = StaticData::Instance().GetPhraseDictionaries();
|
2006-11-16 13:43:57 +03:00
|
|
|
if (pds.size() > 0) {
|
|
|
|
vector<PhraseDictionary*>::iterator iter;
|
|
|
|
|
|
|
|
iter = pds.begin();
|
|
|
|
vector<float> scores = path.GetScoreBreakdown().GetScoresForProducer(*iter);
|
|
|
|
|
|
|
|
size_t pd_numinputscore = (*iter)->GetNumInputScores();
|
|
|
|
|
|
|
|
if (pd_numinputscore){
|
|
|
|
|
|
|
|
if (labeledOutput)
|
2007-01-21 16:38:03 +03:00
|
|
|
*m_nBestStream << "I: ";
|
2006-11-16 13:43:57 +03:00
|
|
|
|
|
|
|
for (size_t j = 0; j < pd_numinputscore; ++j)
|
2007-01-21 16:38:03 +03:00
|
|
|
*m_nBestStream << scores[j] << " ";
|
2006-11-16 13:43:57 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
for (iter = pds.begin() ; iter != pds.end(); ++iter) {
|
|
|
|
vector<float> scores = path.GetScoreBreakdown().GetScoresForProducer(*iter);
|
|
|
|
|
|
|
|
size_t pd_numinputscore = (*iter)->GetNumInputScores();
|
|
|
|
|
|
|
|
if (iter == pds.begin() && labeledOutput)
|
2007-01-21 16:38:03 +03:00
|
|
|
*m_nBestStream << "tm: ";
|
2006-11-16 13:43:57 +03:00
|
|
|
for (size_t j = pd_numinputscore; j < scores.size() ; ++j)
|
2007-01-21 16:38:03 +03:00
|
|
|
*m_nBestStream << scores[j] << " ";
|
2006-11-16 13:43:57 +03:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2006-07-04 22:04:38 +04:00
|
|
|
|
2006-08-04 08:45:48 +04:00
|
|
|
// word penalty
|
2006-08-13 23:08:39 +04:00
|
|
|
if (labeledOutput)
|
2007-01-21 16:38:03 +03:00
|
|
|
*m_nBestStream << "w: ";
|
2007-02-16 21:08:37 +03:00
|
|
|
*m_nBestStream << path.GetScoreBreakdown().GetScoreForProducer(StaticData::Instance().GetWordPenaltyProducer()) << " ";
|
2006-07-04 22:04:38 +04:00
|
|
|
|
|
|
|
// generation
|
2007-02-16 21:08:37 +03:00
|
|
|
vector<GenerationDictionary*> gds = StaticData::Instance().GetGenerationDictionaries();
|
2006-08-04 08:45:48 +04:00
|
|
|
if (gds.size() > 0) {
|
2006-08-13 23:08:39 +04:00
|
|
|
if (labeledOutput)
|
2007-01-21 16:38:03 +03:00
|
|
|
*m_nBestStream << "g: ";
|
2006-08-13 07:03:18 +04:00
|
|
|
vector<GenerationDictionary*>::iterator iter;
|
|
|
|
for (iter = gds.begin(); iter != gds.end(); ++iter) {
|
|
|
|
vector<float> scores = path.GetScoreBreakdown().GetScoresForProducer(*iter);
|
2006-08-04 08:45:48 +04:00
|
|
|
for (size_t j = 0; j<scores.size(); j++) {
|
2007-01-21 16:38:03 +03:00
|
|
|
*m_nBestStream << scores[j] << " ";
|
2006-08-04 08:45:48 +04:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
2006-07-04 22:04:38 +04:00
|
|
|
|
|
|
|
// total
|
2007-02-26 23:59:41 +03:00
|
|
|
*m_nBestStream << "||| " << path.GetTotalScore();
|
|
|
|
if (includeAlignment) {
|
|
|
|
*m_nBestStream << " |||";
|
|
|
|
for (int currEdge = (int)edges.size() - 2 ; currEdge >= 0 ; currEdge--)
|
|
|
|
{
|
|
|
|
const Hypothesis &edge = *edges[currEdge];
|
|
|
|
WordsRange sourceRange = edge.GetCurrSourceWordsRange();
|
|
|
|
WordsRange targetRange = edge.GetCurrTargetWordsRange();
|
|
|
|
*m_nBestStream << " " << sourceRange.GetStartPos();
|
|
|
|
if (sourceRange.GetStartPos() < sourceRange.GetEndPos()) {
|
|
|
|
*m_nBestStream << "-" << sourceRange.GetEndPos();
|
|
|
|
}
|
|
|
|
*m_nBestStream << "=" << targetRange.GetStartPos();
|
|
|
|
if (targetRange.GetStartPos() < targetRange.GetEndPos()) {
|
|
|
|
*m_nBestStream << "-" << targetRange.GetEndPos();
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
*m_nBestStream << endl;
|
2006-07-04 22:04:38 +04:00
|
|
|
}
|
2006-07-29 02:57:31 +04:00
|
|
|
|
2007-01-21 16:38:03 +03:00
|
|
|
*m_nBestStream<<std::flush;
|
2006-07-04 22:04:38 +04:00
|
|
|
}
|