more LM clean-up

git-svn-id: https://mosesdecoder.svn.sourceforge.net/svnroot/mosesdecoder/trunk@144 1f5c12ca-751b-0410-a591-d2e778427230

6 files changed, 53 insertions, 30 deletions

diff --git a/moses/src/LanguageModel.cpp b/moses/src/LanguageModel.cpp
index 987962e90..3513810ae 100755
--- a/moses/src/LanguageModel.cpp
+++ b/moses/src/LanguageModel.cpp
@@ -38,6 +38,7 @@ using namespace std;
 const LmId LanguageModel::UNKNOWN_LM_ID(0);
 
 LanguageModel::LanguageModel() {}
+LanguageModel::~LanguageModel() {}
 
 void LanguageModel::CalcScore(const Phrase &phrase
 														, float &fullScore
diff --git a/moses/src/LanguageModel.h b/moses/src/LanguageModel.h
index e78cd61c5..5170363e1 100755
--- a/moses/src/LanguageModel.h
+++ b/moses/src/LanguageModel.h
@@ -43,6 +43,7 @@ public:
 	static const LmId UNKNOWN_LM_ID;
 
 	LanguageModel();
+	virtual ~LanguageModel();
 	virtual void Load(size_t id
 					, const std::string &fileName
 					, FactorCollection &factorCollection
@@ -83,5 +84,9 @@ public:
 		return m_id;
 	}
 	virtual float GetValue(const std::vector<const Factor*> &contextFactor) const = 0;
+
+  // one of the following should probably be made available
+  // virtual LmId GetLmID( const Factor *factor )  const = 0;
+  // virtual LmId GetLmID( const std::string &factor )  const = 0;
 };
 
diff --git a/moses/src/LanguageModel_Internal.cpp b/moses/src/LanguageModel_Internal.cpp
index 4dcbf89d4..3cca42784 100755
--- a/moses/src/LanguageModel_Internal.cpp
+++ b/moses/src/LanguageModel_Internal.cpp
@@ -235,3 +235,8 @@ float LanguageModel_Internal::GetValue(const Factor *factor0, const Factor *fact
 
 }
 
+LmId LanguageModel_Internal::GetLmID( const Factor *factor )  const
+{
+	return factor->GetLmId();
+}
+
diff --git a/moses/src/LanguageModel_Internal.h b/moses/src/LanguageModel_Internal.h
index b13037228..e445845b2 100755
--- a/moses/src/LanguageModel_Internal.h
+++ b/moses/src/LanguageModel_Internal.h
@@ -57,10 +57,7 @@ protected:
 	float GetValue(const Factor *factor0, const Factor *factor1, const Factor *factor2) const;
 
 public:
-	LmId GetLmID( const Factor *factor )  const
-	{
-		return factor->GetLmId();
-	}
-	
+	LmId GetLmID( const Factor *factor )  const;
+
 };
 
diff --git a/moses/src/LanguageModel_SRI.cpp b/moses/src/LanguageModel_SRI.cpp
index d1fea8832..341e94957 100755
--- a/moses/src/LanguageModel_SRI.cpp
+++ b/moses/src/LanguageModel_SRI.cpp
@@ -24,7 +24,8 @@ Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA
 #include <iostream>
 #include <fstream>
 
-#include "NGramNode.h"
+#include "Ngram.h"
+#include "Vocab.h"
 
 #include "LanguageModel_SRI.h"
 #include "TypeDef.h"
@@ -35,10 +36,15 @@ Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA
 using namespace std;
 
 LanguageModel_SRI::LanguageModel_SRI()
-:m_srilmVocab()
-,m_srilmModel(m_srilmVocab, 3)
+:m_srilmVocab(0)
+,m_srilmModel(0)
 {
-	m_srilmModel.skipOOVs() = false;
+}
+
+LanguageModel_SRI::~LanguageModel_SRI()
+{
+  delete m_srilmModel;
+  delete m_srilmVocab;
 }
 
 void LanguageModel_SRI::Load(size_t id
@@ -48,13 +54,17 @@ void LanguageModel_SRI::Load(size_t id
 												, float weight
 												, size_t nGramOrder)
 {
+	m_srilmVocab  = new Vocab();
+  m_srilmModel	= new Ngram(*m_srilmVocab, nGramOrder);
 	m_id					= id;
 	m_factorType 	= factorType;
 	m_weight			= weight;
 	m_nGramOrder	= nGramOrder;
 
+	m_srilmModel->skipOOVs() = false;
+
 	File file( fileName.c_str(), "r" );
-	if (m_srilmModel.read(file))
+	if (m_srilmModel->read(file))
 	{
 	}
 	else
@@ -71,7 +81,7 @@ void LanguageModel_SRI::CreateFactors(FactorCollection &factorCollection)
 	
 	VocabString str;
 	LmId lmId;
-	VocabIter iter(m_srilmVocab);
+	VocabIter iter(*m_srilmVocab);
 	while ( (str = iter.next()) != NULL)
 	{
 		LmId lmId = GetLmID(str);
@@ -85,6 +95,21 @@ void LanguageModel_SRI::CreateFactors(FactorCollection &factorCollection)
 	
 }
 
+LmId LanguageModel_SRI::GetLmID( const std::string &str ) const
+{
+    LanguageModel_SRI *lm = const_cast<LanguageModel_SRI*>(this); // hack. not sure if supposed to cast this
+    LmId res;
+    res.sri = lm->m_srilmVocab->getIndex( str.c_str(), lm->m_srilmVocab->unkIndex() );
+    return res;
+}
+
+float LanguageModel_SRI::GetValue(LmId wordId, VocabIndex *context) const
+{
+	LanguageModel_SRI *lm = const_cast<LanguageModel_SRI*>(this); // hack. not sure if supposed to cast this
+	float p = lm->m_srilmModel->wordProb( wordId.sri, context );
+	return FloorSRIScore(TransformSRIScore(p));  // log10->log
+}
+
 float LanguageModel_SRI::GetValue(const vector<const Factor*> &contextFactor) const
 {
 	// set up context
diff --git a/moses/src/LanguageModel_SRI.h b/moses/src/LanguageModel_SRI.h
index bcd11537d..32e9b1b30 100755
--- a/moses/src/LanguageModel_SRI.h
+++ b/moses/src/LanguageModel_SRI.h
@@ -29,19 +29,22 @@ Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA
 #include "Util.h"
 
 #include "LanguageModel.h"
-#include "Ngram.h"
-#include "Vocab.h"
 
 
 class FactorCollection;
 class Factor;
 class Phrase;
 
+#include "Vocab.h"
+class Vocab; // SRI forward decl
+class Ngram; // SRI forward decl
+
 class LanguageModel_SRI : public LanguageModel
 {
 public:
 	
 	LanguageModel_SRI();
+	~LanguageModel_SRI();
 	void Load(size_t id
 					, const std::string &fileName
 					, FactorCollection &factorCollection
@@ -50,15 +53,10 @@ public:
 					, size_t nGramOrder);
 	
 protected:
-	Vocab m_srilmVocab;  // TODO - make this a ptr, remove #include from header
-	Ngram m_srilmModel;  // "  "
+	Vocab *m_srilmVocab;  // TODO - make this a ptr, remove #include from header
+	Ngram *m_srilmModel;  // "  "
 
-	float GetValue(LmId wordId, VocabIndex *context) const
-	{
-		LanguageModel_SRI *lm = const_cast<LanguageModel_SRI*>(this);	// hack. not sure if supposed to cast this
-		float p = lm->m_srilmModel.wordProb( wordId.sri, context );
-		return FloorSRIScore(TransformSRIScore(p));	 // log10->log
-	}
+	float GetValue(LmId wordId, VocabIndex *context) const;
 
 	LmId GetLmID( const Factor *factor )  const
 	{
@@ -66,15 +64,7 @@ protected:
 	}
 	void CreateFactors(FactorCollection &factorCollection);
 public:
-	static const LmId UNKNOWN_LM_ID;
-	
-	LmId GetLmID( const std::string &str ) const
-	{
-		LanguageModel_SRI *lm = const_cast<LanguageModel_SRI*>(this);	// hack. not sure if supposed to cast this
-		LmId res;
-    res.sri = lm->m_srilmVocab.getIndex( str.c_str(), lm->m_srilmVocab.unkIndex() );
-    return res;
-	}
+	LmId GetLmID( const std::string &str ) const;
 
   virtual float GetValue(const vector<const Factor*> &contextFactor) const;
 };