xcode

moses-chart-cmd/src/Main.cpp

@@ -48,6 +48,7 @@ POSSIBILITY OF SUCH DAMAGE.
 #include "Sentence.h"
 #include "ConfusionNet.h"
 #include "WordLattice.h"
+#include "WordLattice2.h"
 #include "TreeInput.h"
 #include "TranslationAnalysis.h"
 #include "mbr.h"
@@ -158,6 +159,9 @@ bool ReadInput(IOWrapper &ioWrapper, InputTypeEnum inputType, InputType*& source
   case TreeInputType:
     source = ioWrapper.GetInput(new TreeInput(Input));
     break;
+  case WordLatticeInput2:
+      source = ioWrapper.GetInput(new WordLattice2);
+      break;
   default:
     TRACE_ERR("Unknown input type: " << inputType << "\n");
   }

moses/moses.xcodeproj/project.pbxproj

@@ -60,6 +60,10 @@
 		1E2E163D132A892800ED4085 /* RuleCubeQueue.h in Headers */ = {isa = PBXBuildFile; fileRef = 1E2E1636132A892800ED4085 /* RuleCubeQueue.h */; };
 		1E2E163E132A892800ED4085 /* ThreadPool.cpp in Sources */ = {isa = PBXBuildFile; fileRef = 1E2E1637132A892800ED4085 /* ThreadPool.cpp */; };
 		1E2E163F132A892800ED4085 /* ThreadPool.h in Headers */ = {isa = PBXBuildFile; fileRef = 1E2E1638132A892800ED4085 /* ThreadPool.h */; };
+		1E3CBD3A145D755D00AF72FC /* WordLattice2.cpp in Sources */ = {isa = PBXBuildFile; fileRef = 1E3CBD38145D755D00AF72FC /* WordLattice2.cpp */; };
+		1E3CBD3B145D755D00AF72FC /* WordLattice2.h in Headers */ = {isa = PBXBuildFile; fileRef = 1E3CBD39145D755D00AF72FC /* WordLattice2.h */; };
+		1E3CBD3F145D75F300AF72FC /* WeightedDirectedGraph.cpp in Sources */ = {isa = PBXBuildFile; fileRef = 1E3CBD3D145D75F300AF72FC /* WeightedDirectedGraph.cpp */; };
+		1E3CBD40145D75F300AF72FC /* WeightedDirectedGraph.h in Headers */ = {isa = PBXBuildFile; fileRef = 1E3CBD3E145D75F300AF72FC /* WeightedDirectedGraph.h */; };
 		1E46B5A613BA5C7F0084F898 /* RuleCubeItem.cpp in Sources */ = {isa = PBXBuildFile; fileRef = 1E46B5A413BA5C7F0084F898 /* RuleCubeItem.cpp */; };
 		1E46B5A713BA5C7F0084F898 /* RuleCubeItem.h in Headers */ = {isa = PBXBuildFile; fileRef = 1E46B5A513BA5C7F0084F898 /* RuleCubeItem.h */; };
 		1E474E12145575CA00178AD5 /* RuleTableLoader.h in Headers */ = {isa = PBXBuildFile; fileRef = 1E474E11145575CA00178AD5 /* RuleTableLoader.h */; };
@@ -322,6 +326,10 @@
 		1E2E1636132A892800ED4085 /* RuleCubeQueue.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; name = RuleCubeQueue.h; path = src/RuleCubeQueue.h; sourceTree = "<group>"; };
 		1E2E1637132A892800ED4085 /* ThreadPool.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = ThreadPool.cpp; path = src/ThreadPool.cpp; sourceTree = "<group>"; };
 		1E2E1638132A892800ED4085 /* ThreadPool.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; name = ThreadPool.h; path = src/ThreadPool.h; sourceTree = "<group>"; };
+		1E3CBD38145D755D00AF72FC /* WordLattice2.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = WordLattice2.cpp; path = src/WordLattice2.cpp; sourceTree = "<group>"; };
+		1E3CBD39145D755D00AF72FC /* WordLattice2.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; name = WordLattice2.h; path = src/WordLattice2.h; sourceTree = "<group>"; };
+		1E3CBD3D145D75F300AF72FC /* WeightedDirectedGraph.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = WeightedDirectedGraph.cpp; path = src/WeightedDirectedGraph.cpp; sourceTree = "<group>"; };
+		1E3CBD3E145D75F300AF72FC /* WeightedDirectedGraph.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; name = WeightedDirectedGraph.h; path = src/WeightedDirectedGraph.h; sourceTree = "<group>"; };
 		1E46B5A413BA5C7F0084F898 /* RuleCubeItem.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = RuleCubeItem.cpp; path = src/RuleCubeItem.cpp; sourceTree = "<group>"; };
 		1E46B5A513BA5C7F0084F898 /* RuleCubeItem.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; name = RuleCubeItem.h; path = src/RuleCubeItem.h; sourceTree = "<group>"; };
 		1E474E11145575CA00178AD5 /* RuleTableLoader.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; name = RuleTableLoader.h; path = src/RuleTableLoader.h; sourceTree = "<group>"; };
@@ -557,6 +565,10 @@
 		08FB7795FE84155DC02AAC07 /* Source */ = {
 			isa = PBXGroup;
 			children = (
+				1E3CBD3D145D75F300AF72FC /* WeightedDirectedGraph.cpp */,
+				1E3CBD3E145D75F300AF72FC /* WeightedDirectedGraph.h */,
+				1E3CBD38145D755D00AF72FC /* WordLattice2.cpp */,
+				1E3CBD39145D755D00AF72FC /* WordLattice2.h */,
 				1E474E11145575CA00178AD5 /* RuleTableLoader.h */,
 				1E16D086144DAA3F00B60B4F /* LM */,
 				1ED0FD4C124BB9380029177F /* AlignmentInfo.cpp */,
@@ -1000,6 +1012,8 @@
 				1E16D0BC144DAA6C00B60B4F /* SingleFactor.h in Headers */,
 				1E16D0BE144DAA6C00B60B4F /* SRI.h in Headers */,
 				1E474E12145575CA00178AD5 /* RuleTableLoader.h in Headers */,
+				1E3CBD3B145D755D00AF72FC /* WordLattice2.h in Headers */,
+				1E3CBD40145D75F300AF72FC /* WeightedDirectedGraph.h in Headers */,
 			);
 			runOnlyForDeploymentPostprocessing = 0;
 		};
@@ -1175,6 +1189,8 @@
 				1E16D0B5144DAA6C00B60B4F /* ParallelBackoff.cpp in Sources */,
 				1E16D0BB144DAA6C00B60B4F /* SingleFactor.cpp in Sources */,
 				1E16D0BD144DAA6C00B60B4F /* SRI.cpp in Sources */,
+				1E3CBD3A145D755D00AF72FC /* WordLattice2.cpp in Sources */,
+				1E3CBD3F145D75F300AF72FC /* WeightedDirectedGraph.cpp in Sources */,
 			);
 			runOnlyForDeploymentPostprocessing = 0;
 		};

moses/src/Makefile.am

@@ -125,8 +125,10 @@ libmoses_la_HEADERS = \
         UniqueObject.h \
         UserMessage.h \
         Util.h \
+        WeightedDirectedGraph.h \
         Word.h \
         WordLattice.h \
+        WordLattice2.h \
         WordsBitmap.h \
         WordsRange.h \
         XmlOption.h
@@ -270,8 +272,10 @@ libmoses_la_SOURCES = \
         TrellisPathCollection.cpp \
         UserMessage.cpp \
         Util.cpp \
+        WeightedDirectedGraph.cpp \
         Word.cpp \
         WordLattice.cpp \
+        WordLattice2.cpp \
         WordsBitmap.cpp \
         WordsRange.cpp \
         XmlOption.cpp

moses/src/PCNTools.cpp

@@ -165,5 +165,11 @@ CN parsePCN(const std::string& in)
   return res;
 }
 
+Moses::WeightedDirectedGraph parseLattice(const std::string& in)
+{
+  return Moses::WeightedDirectedGraph();
+  
+}
+
 }
 

moses/src/PCNTools.h

@@ -26,6 +26,7 @@ Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA
 #include <string>
 #include <utility>
 #include <cstdlib>
+#include "WeightedDirectedGraph.h"
 
 /** A couple of utilities to read .pcn files. A python-compatible format
   * for encoding confusion networks and word lattices.
@@ -36,11 +37,12 @@ namespace PCN
 typedef std::pair<std::pair<std::string, std::vector<float> >, size_t> CNAlt;
 typedef std::vector<CNAlt> CNCol;
 typedef std::vector<CNCol> CN;
-
+  
 /** Given a string ((('foo',0.1,1),('bar',0.9,2)),...) representation of a
   * word lattice in PCN format, return a CN object representing the lattice
   */
 CN parsePCN(const std::string& in);
+Moses::WeightedDirectedGraph parseLattice(const std::string& in);
 
 };
 

moses/src/StaticData.cpp

@@ -991,25 +991,31 @@ bool StaticData::LoadPhraseTables()
         // it only work with binrary file. This is a hack
 
         m_numInputScores=m_parameter->GetParam("weight-i").size();
-        for(unsigned k=0; k<m_numInputScores; ++k)
-          weight.push_back(Scan<float>(m_parameter->GetParam("weight-i")[k]));
-
+        
+        if (implementation == Binary)
+        {
+          for(unsigned k=0; k<m_numInputScores; ++k)
+            weight.push_back(Scan<float>(m_parameter->GetParam("weight-i")[k]));
+        }
+        
         if(m_parameter->GetParam("link-param-count").size())
           m_numLinkParams = Scan<size_t>(m_parameter->GetParam("link-param-count")[0]);
 
         //print some info about this interaction:
-        if (m_numLinkParams == m_numInputScores) {
-          VERBOSE(1,"specified equal numbers of link parameters and insertion weights, not using non-epsilon 'real' word link count.\n");
-        } else if ((m_numLinkParams + 1) == m_numInputScores) {
-          VERBOSE(1,"WARN: "<< m_numInputScores << " insertion weights found and only "<< m_numLinkParams << " link parameters specified, applying non-epsilon 'real' word link count for last feature weight.\n");
-        } else {
-          stringstream strme;
-          strme << "You specified " << m_numInputScores
-                << " input weights (weight-i), but you specified " << m_numLinkParams << " link parameters (link-param-count)!";
-          UserMessage::Add(strme.str());
-          return false;
+        if (implementation == Binary) {
+          if (m_numLinkParams == m_numInputScores) {
+            VERBOSE(1,"specified equal numbers of link parameters and insertion weights, not using non-epsilon 'real' word link count.\n");
+          } else if ((m_numLinkParams + 1) == m_numInputScores) {
+            VERBOSE(1,"WARN: "<< m_numInputScores << " insertion weights found and only "<< m_numLinkParams << " link parameters specified, applying non-epsilon 'real' word link count for last feature weight.\n");
+          } else {
+            stringstream strme;
+            strme << "You specified " << m_numInputScores
+                  << " input weights (weight-i), but you specified " << m_numLinkParams << " link parameters (link-param-count)!";
+            UserMessage::Add(strme.str());
+            return false;
+          }
         }
-
+        
       }
       if (!m_inputType) {
         m_numInputScores=0;

moses/src/TypeDef.h

@@ -181,6 +181,8 @@ enum InputTypeEnum {
   ,ConfusionNetworkInput	= 1
   ,WordLatticeInput				= 2
   ,TreeInputType					= 3
+  ,WordLatticeInput2			= 4
+  
 };
 
 enum XmlInputType {

moses/src/WeightedDirectedGraph.cpp

@@ -0,0 +1,9 @@
+//
+//  WeightedDirectedGraph.cpp
+//  moses
+//
+//  Created by Hieu Hoang on 30/10/2011.
+//  Copyright 2011 __MyCompanyName__. All rights reserved.
+//
+
+#include <iostream>

moses/src/WeightedDirectedGraph.h

@@ -0,0 +1,26 @@
+//
+//  WeightedDirectedGraph.h
+//  moses
+//
+//  Created by Hieu Hoang on 30/10/2011.
+//  Copyright 2011 __MyCompanyName__. All rights reserved.
+//
+
+#ifndef moses_WeightedDirectedGraph_h
+#define moses_WeightedDirectedGraph_h
+
+namespace Moses {
+
+class WeightedDirectedGraphNode
+{
+  
+};
+
+class WeightedDirectedGraph
+{
+  
+  
+};
+
+}
+#endif

moses/src/WordLattice2.cpp

@@ -0,0 +1,60 @@
+//
+//  WordLattice2.cpp
+//  moses
+//
+//  Created by Hieu Hoang on 30/10/2011.
+//  Copyright 2011 __MyCompanyName__. All rights reserved.
+//
+
+#include <iostream>
+#include "WordLattice2.h"
+#include "PCNTools.h"
+
+using namespace std;
+
+namespace Moses
+{
+size_t WordLattice2::GetSize() const
+{
+  
+}
+
+int WordLattice2::Read(std::istream& in,const std::vector<FactorType>& factorOrder)
+{
+  std::string line;
+  if(!getline(in,line)) return 0;
+  
+  PCN::parseLattice(line);
+  
+}
+
+void WordLattice2::Print(std::ostream&) const
+{
+  
+}
+
+//! create trans options specific to this InputType
+TranslationOptionCollection* WordLattice2::CreateTranslationOptionCollection(const TranslationSystem* system) const
+{
+  
+}
+
+//! return substring. Only valid for Sentence class. TODO - get rid of this fn
+Phrase WordLattice2::GetSubString(const WordsRange&) const
+{
+  
+}
+
+//! return substring at a particular position. Only valid for Sentence class. TODO - get rid of this fn
+const Word& WordLattice2::GetWord(size_t pos) const
+{
+  
+}
+
+const NonTerminalSet &WordLattice2::GetLabelSet(size_t startPos, size_t endPos) const
+{
+  
+}
+
+}
+

moses/src/WordLattice2.h

@@ -0,0 +1,48 @@
+//
+//  WordLattice2.h
+//  moses
+//
+//  Created by Hieu Hoang on 30/10/2011.
+//  Copyright 2011 __MyCompanyName__. All rights reserved.
+//
+
+#ifndef moses_WordLattice2_h
+#define moses_WordLattice2_h
+
+#include "InputType.h"
+
+namespace Moses
+{
+
+class WordLattice2 : public InputType
+{
+
+public:
+  InputTypeEnum GetType() const 
+  { return WordLatticeInput2; }
+  
+  virtual size_t GetSize() const;
+  
+  //! populate this InputType with data from in stream
+  virtual int Read(std::istream& in,const std::vector<FactorType>& factorOrder);
+
+  //! Output debugging info to stream out
+  virtual void Print(std::ostream&) const;
+  
+  //! create trans options specific to this InputType
+  virtual TranslationOptionCollection* CreateTranslationOptionCollection(const TranslationSystem* system) const;
+  
+  //! return substring. Only valid for Sentence class. TODO - get rid of this fn
+  virtual Phrase GetSubString(const WordsRange&) const;
+  
+  //! return substring at a particular position. Only valid for Sentence class. TODO - get rid of this fn
+  virtual const Word& GetWord(size_t pos) const;
+
+  virtual const NonTerminalSet &GetLabelSet(size_t startPos, size_t endPos) const;
+
+  
+};
+
+}
+
+#endif