code clean - move creating wordgraph file to iostream class

git-svn-id: https://mosesdecoder.svn.sourceforge.net/svnroot/mosesdecoder/trunk@1581 1f5c12ca-751b-0410-a591-d2e778427230

moses-cmd/src/IOStream.cpp

@@ -57,22 +57,11 @@ IOStream::IOStream(
 ,m_inputFile(NULL)
 ,m_inputStream(&std::cin)
 ,m_nBestStream(NULL)
+,m_outputWordGraphStream(NULL)
 {
-	m_surpressSingleBestOutput = false;
-	if (nBestSize > 0)
-	{
-		if (nBestFilePath == "-")
-		{
-			m_nBestStream = &std::cout;
-			m_surpressSingleBestOutput = true;
-		} 
-		else 
-		{
-			std::ofstream *nBestFile = new std::ofstream;
-			m_nBestStream = nBestFile;
-			nBestFile->open(nBestFilePath.c_str());
-		}
-	}
+	Initialization(inputFactorOrder, outputFactorOrder
+								, inputFactorUsed
+								, nBestSize, nBestFilePath);
 }
 
 IOStream::IOStream(const std::vector<FactorType>	&inputFactorOrder
@@ -87,24 +76,13 @@ IOStream::IOStream(const std::vector<FactorType>	&inputFactorOrder
 ,m_inputFilePath(inputFilePath)
 ,m_inputFile(new InputFileStream(inputFilePath))
 ,m_nBestStream(NULL)
+,m_outputWordGraphStream(NULL)
 {
-	m_surpressSingleBestOutput = false;
-	m_inputStream = m_inputFile;
+	Initialization(inputFactorOrder, outputFactorOrder
+								, inputFactorUsed
+								, nBestSize, nBestFilePath);
 
-	if (nBestSize > 0)
-	{
-		if (nBestFilePath == "-")
-		{
-			m_nBestStream = &std::cout;
-			m_surpressSingleBestOutput = true;
-		} 
-		else 
-		{
-			std::ofstream *nBestFile = new std::ofstream;
-			m_nBestStream = nBestFile;
-			nBestFile->open(nBestFilePath.c_str());
-		}
-	}
+	m_inputStream = m_inputFile;
 }
 
 IOStream::~IOStream()
@@ -115,6 +93,47 @@ IOStream::~IOStream()
 	{ // outputting n-best to file, rather than stdout. need to close file and delete obj
 		delete m_nBestStream;
 	}
+	if (m_outputWordGraphStream != NULL)
+	{
+		delete m_outputWordGraphStream;
+	}
+}
+
+void IOStream::Initialization(const std::vector<FactorType>	&inputFactorOrder
+														, const std::vector<FactorType>			&outputFactorOrder
+														, const FactorMask							&inputFactorUsed
+														, size_t												nBestSize
+														, const std::string							&nBestFilePath)
+{
+	const StaticData &staticData = StaticData::Instance();
+
+	// n-best
+	m_surpressSingleBestOutput = false;
+	if (nBestSize > 0)
+	{
+		if (nBestFilePath == "-")
+		{
+			m_nBestStream = &std::cout;
+			m_surpressSingleBestOutput = true;
+		} 
+		else 
+		{
+			std::ofstream *file = new std::ofstream;
+			m_nBestStream = file;
+			file->open(nBestFilePath.c_str());
+		}
+	}
+
+	// wordgraph output
+	if (staticData.GetOutputWordGraph())
+	{
+		string fileName = staticData.GetParam("output-word-graph")[0];
+		bool outputNBest = Scan<bool>(staticData.GetParam("output-word-graph")[1]);
+
+		std::ofstream *file = new std::ofstream;
+		m_outputWordGraphStream  = file;
+		file->open(fileName.c_str());
+	}
 }
 
 InputType*IOStream::GetInput(InputType* inputType)

moses-cmd/src/IOStream.h

@@ -53,12 +53,19 @@ protected:
 	const std::vector<FactorType>	&m_inputFactorOrder;
 	const std::vector<FactorType>	&m_outputFactorOrder;
 	const FactorMask							&m_inputFactorUsed;
-	std::ostream 								*m_nBestStream;
+	std::ostream 									*m_nBestStream
+																,*m_outputWordGraphStream;
 	std::string										m_inputFilePath;
 	std::istream									*m_inputStream;
 	InputFileStream								*m_inputFile;
 	bool													m_surpressSingleBestOutput;
 	
+	void Initialization(const std::vector<FactorType>	&inputFactorOrder
+										, const std::vector<FactorType>			&outputFactorOrder
+										, const FactorMask							&inputFactorUsed
+										, size_t												nBestSize
+										, const std::string							&nBestFilePath);
+
 public:
 	IOStream(const std::vector<FactorType>	&inputFactorOrder
 		, const std::vector<FactorType>			&outputFactorOrder
@@ -82,4 +89,8 @@ public:
 
 	void ResetTranslationId() { m_translationId = 0; }
 
+	std::ostream &GetOutputWordGraphStream()
+	{
+		return *m_outputWordGraphStream;
+	}
 };

moses-cmd/src/Main.cpp

@@ -138,7 +138,7 @@ int main(int argc, char* argv[])
 		manager.ProcessSentence();
 		
 		if (staticData.GetOutputWordGraph())
-			manager.GetWordGraph(source->GetTranslationId());
+			manager.GetWordGraph(source->GetTranslationId(), ioStream->GetOutputWordGraphStream());
 
 		// pick best translation (maximum a posteriori decoding)
 		if (! staticData.UseMBR()) {

moses/src/Manager.cpp

@@ -481,7 +481,7 @@ void Manager::CalcDecoderStatistics() const
   }
 }
 
-void OutputWordGraph(std::ofstream &wordGraphFile, const Hypothesis *hypo, size_t &linkId)
+void OutputWordGraph(std::ostream &outputWordGraphStream, const Hypothesis *hypo, size_t &linkId)
 {
 	const StaticData &staticData = StaticData::Instance();
 
@@ -490,7 +490,7 @@ void OutputWordGraph(std::ofstream &wordGraphFile, const Hypothesis *hypo, size_
 			const Phrase &targetPhrase = hypo->GetCurrTargetPhrase();
 
 			
-			wordGraphFile << "J=" << linkId++
+			outputWordGraphStream << "J=" << linkId++
 						<< "\tS=" << prevHypo->GetId()
 						<< "\tE=" << hypo->GetId()
 						<< "\ta=";
@@ -503,16 +503,16 @@ void OutputWordGraph(std::ofstream &wordGraphFile, const Hypothesis *hypo, size_
 				const PhraseDictionary *phraseTable = *iterPhraseTable;
 				vector<float> scores = hypo->GetScoreBreakdown().GetScoresForProducer(phraseTable);
 
-				wordGraphFile << scores[0];
+				outputWordGraphStream << scores[0];
 				vector<float>::const_iterator iterScore;
 				for (iterScore = ++scores.begin() ; iterScore != scores.end() ; ++iterScore)
 				{
-					wordGraphFile << ", " << *iterScore;
+					outputWordGraphStream << ", " << *iterScore;
 				}
 			}
 
 			// language model scores
-			wordGraphFile << "\tl=";
+			outputWordGraphStream << "\tl=";
 			const LMList &lmList = staticData.GetAllLM();
 			LMList::const_iterator iterLM;
 			for (iterLM = lmList.begin() ; iterLM != lmList.end() ; ++iterLM)
@@ -520,18 +520,18 @@ void OutputWordGraph(std::ofstream &wordGraphFile, const Hypothesis *hypo, size_
 				LanguageModel *lm = *iterLM;
 				vector<float> scores = hypo->GetScoreBreakdown().GetScoresForProducer(lm);
 				
-				wordGraphFile << scores[0];
+				outputWordGraphStream << scores[0];
 				vector<float>::const_iterator iterScore;
 				for (iterScore = ++scores.begin() ; iterScore != scores.end() ; ++iterScore)
 				{
-					wordGraphFile << ", " << *iterScore;
+					outputWordGraphStream << ", " << *iterScore;
 				}
 			}
 
 			// re-ordering
-			wordGraphFile << "\tr=";
+			outputWordGraphStream << "\tr=";
 
-			wordGraphFile << hypo->GetScoreBreakdown().GetScoreForProducer(staticData.GetDistortionScoreProducer());
+			outputWordGraphStream << hypo->GetScoreBreakdown().GetScoreForProducer(staticData.GetDistortionScoreProducer());
 
 			// lexicalised re-ordering
 			const std::vector<LexicalReordering*> &lexOrderings = staticData.GetReorderModels();
@@ -541,30 +541,27 @@ void OutputWordGraph(std::ofstream &wordGraphFile, const Hypothesis *hypo, size_
 				LexicalReordering *lexicalReordering = *iterLexOrdering;
 				vector<float> scores = hypo->GetScoreBreakdown().GetScoresForProducer(lexicalReordering);
 				
-				wordGraphFile << scores[0];
+				outputWordGraphStream << scores[0];
 				vector<float>::const_iterator iterScore;
 				for (iterScore = ++scores.begin() ; iterScore != scores.end() ; ++iterScore)
 				{
-					wordGraphFile << ", " << *iterScore;
+					outputWordGraphStream << ", " << *iterScore;
 				}
 			}
 
 			// words !!
-			wordGraphFile << "\tw=" << hypo->GetCurrTargetPhrase();
+			outputWordGraphStream << "\tw=" << hypo->GetCurrTargetPhrase();
 
-			wordGraphFile << endl;
+			outputWordGraphStream << endl;
 }
 
-void Manager::GetWordGraph(long translationId) const
+void Manager::GetWordGraph(long translationId, std::ostream &outputWordGraphStream) const
 {
 	const StaticData &staticData = StaticData::Instance();
 	string fileName = staticData.GetParam("output-word-graph")[0];
 	bool outputNBest = Scan<bool>(staticData.GetParam("output-word-graph")[1]);
 	
-	std::ofstream wordGraphFile;
-	wordGraphFile.open(fileName.c_str(), ios::app);
-
-	wordGraphFile << "VERSION=1.0" << endl
+	outputWordGraphStream << "VERSION=1.0" << endl
 								<< "UTTERANCE=" << translationId << endl;
 
 	size_t linkId = 0;
@@ -578,7 +575,7 @@ void Manager::GetWordGraph(long translationId) const
 		for (iterHypo = stack.begin() ; iterHypo != stack.end() ; ++iterHypo)
 		{
 			const Hypothesis *hypo = *iterHypo;
-			OutputWordGraph(wordGraphFile, hypo, linkId);
+			OutputWordGraph(outputWordGraphStream, hypo, linkId);
 			
 			if (outputNBest)
 			{
@@ -589,13 +586,11 @@ void Manager::GetWordGraph(long translationId) const
 					for (iterArcList = arcList->begin() ; iterArcList != arcList->end() ; ++iterArcList)
 					{
 						const Hypothesis *loserHypo = *iterArcList;
-						OutputWordGraph(wordGraphFile, loserHypo, linkId);
+						OutputWordGraph(outputWordGraphStream, loserHypo, linkId);
 					}
 				}
 			} //if (outputNBest)
 		} //for (iterHypo
 	} // for (iterStack 
-
-	wordGraphFile.close();		
 }
 

moses/src/Manager.h

@@ -97,7 +97,7 @@ public:
 	const Hypothesis *GetBestHypothesis() const;
 	void CalcNBest(size_t count, TrellisPathList &ret,bool onlyDistinct=0) const;
 	
-	void GetWordGraph(long translationId) const;
+	void GetWordGraph(long translationId, std::ostream &outputWordGraphStream) const;
 
 	/***
 	 * to be called after processing a sentence (which may consist of more than just calling ProcessSentence() )