From 92163d50913d3d14ae7796e882bf3f6533575b44 Mon Sep 17 00:00:00 2001
From: Hieu Hoang <hieu@hoang.co.uk>
Date: Tue, 9 Jul 2013 15:48:36 +0100
Subject: [PATCH] prefix subphrase optimization done for confusion networks

---
 moses/InputPath.cpp                           |  8 +--
 ...ranslationOptionCollectionConfusionNet.cpp | 61 ++++++++++++++++++-
 .../TranslationOptionCollectionConfusionNet.h |  5 +-
 3 files changed, 65 insertions(+), 9 deletions(-)

diff --git a/moses/InputPath.cpp b/moses/InputPath.cpp
index 354e48197..43b4207cf 100644
--- a/moses/InputPath.cpp
+++ b/moses/InputPath.cpp
@@ -1,5 +1,6 @@
 #include "InputPath.h"
 #include "ScoreComponentCollection.h"
+
 namespace Moses
 {
 InputPath::InputPath(const Phrase &phrase, const WordsRange &range, const InputPath *prevNode
@@ -7,13 +8,8 @@ InputPath::InputPath(const Phrase &phrase, const WordsRange &range, const InputP
   :m_prevNode(prevNode)
   ,m_phrase(phrase)
   ,m_range(range)
+  ,m_inputScore(inputScore)
 {
-  if (inputScore) {
-    m_inputScore = new ScoreComponentCollection(*inputScore);
-  }
-  else {
-    m_inputScore = NULL;
-  }
 }
 
 InputPath::~InputPath()
diff --git a/moses/TranslationOptionCollectionConfusionNet.cpp b/moses/TranslationOptionCollectionConfusionNet.cpp
index 06e86d7c7..ebe62b0dd 100644
--- a/moses/TranslationOptionCollectionConfusionNet.cpp
+++ b/moses/TranslationOptionCollectionConfusionNet.cpp
@@ -1,11 +1,12 @@
 // $Id$
 
+#include <list>
 #include "TranslationOptionCollectionConfusionNet.h"
 #include "ConfusionNet.h"
 #include "DecodeStep.h"
 #include "DecodeStepTranslation.h"
 #include "FactorCollection.h"
-#include <list>
+#include "FF/InputFeature.h"
 
 using namespace std;
 
@@ -16,7 +17,63 @@ namespace Moses
 TranslationOptionCollectionConfusionNet::TranslationOptionCollectionConfusionNet(
   const ConfusionNet &input
   , size_t maxNoTransOptPerCoverage, float translationOptionThreshold)
-  : TranslationOptionCollection(input, maxNoTransOptPerCoverage, translationOptionThreshold) {}
+  : TranslationOptionCollection(input, maxNoTransOptPerCoverage, translationOptionThreshold)
+{
+  const InputFeature *inputFeature = StaticData::Instance().GetInputFeature();
+  CHECK(inputFeature);
+
+  size_t size = input.GetSize();
+  m_targetPhrasesfromPt.resize(size);
+
+  // 1-word phrases
+  for (size_t startPos = 0; startPos < size; ++startPos) {
+    vector<InputPathList> &vec = m_targetPhrasesfromPt[startPos];
+	vec.push_back(InputPathList());
+	InputPathList &list = vec.back();
+
+	WordsRange range(startPos, startPos);
+
+	const ConfusionNet::Column &col = input.GetColumn(startPos);
+	for (size_t i = 0; i < col.size(); ++i) {
+	  const Word &word = col[i].first;
+	  Phrase subphrase;
+	  subphrase.AddWord(word);
+
+	  const std::vector<float> &scores = col[i].second;
+	  ScoreComponentCollection *inputScore = new ScoreComponentCollection();
+	  inputScore->Assign(inputFeature, scores);
+
+	  InputPath *node = new InputPath(subphrase, range, NULL, inputScore);
+	  list.push_back(node);
+
+	}
+  }
+
+  /*
+  for (size_t phaseSize = 1; phaseSize <= size; ++phaseSize) {
+	for (size_t startPos = 0; startPos < size - phaseSize + 1; ++startPos) {
+	  size_t endPos = startPos + phaseSize -1;
+	  vector<InputPathList> &vec = m_targetPhrasesfromPt[startPos];
+
+	  Phrase subphrase(input.GetSubString(WordsRange(startPos, endPos)));
+	  WordsRange range(startPos, endPos);
+
+	  InputPath *node;
+	  if (range.GetNumWordsCovered() == 1) {
+		node = new InputPath(subphrase, range, NULL, NULL);
+		vec.push_back(node);
+	  } else {
+		const InputPath &prevNode = GetInputPath(startPos, endPos - 1);
+		node = new InputPath(subphrase, range, &prevNode, NULL);
+		vec.push_back(node);
+	  }
+
+	  m_phraseDictionaryQueue.push_back(node);
+	}
+  }
+  */
+
+}
 
 /* forcibly create translation option for a particular source word.
 	* call the base class' ProcessOneUnknownWord() for each possible word in the confusion network
diff --git a/moses/TranslationOptionCollectionConfusionNet.h b/moses/TranslationOptionCollectionConfusionNet.h
index d24f7288c..0284c44c7 100644
--- a/moses/TranslationOptionCollectionConfusionNet.h
+++ b/moses/TranslationOptionCollectionConfusionNet.h
@@ -15,8 +15,11 @@ class ConfusionNet;
  */
 class TranslationOptionCollectionConfusionNet : public TranslationOptionCollection
 {
+public:
+  typedef std::vector< std::vector<InputPathList> > TargetPhraseMatrix;
+
 protected:
-	typedef std::vector< std::vector<InputPathList> > TargetPhraseMatrix;
+  TargetPhraseMatrix	m_targetPhrasesfromPt; /*< contains translation options */
 
 public:
   TranslationOptionCollectionConfusionNet(const ConfusionNet &source, size_t maxNoTransOptPerCoverage, float translationOptionThreshold);