mixed syntax type

contrib/other-builds/extract-mixed-syntax/AlignedSentence.cpp

@@ -95,7 +95,7 @@ std::vector<int> AlignedSentence::GetSourceAlignmentCount() const
 void AlignedSentence::Create(const Parameter &params)
 {
 	CreateConsistentPhrases(params);
-	m_consistentPhrases.AddHieroNonTerms(false);
+	m_consistentPhrases.AddHieroNonTerms();
 }
 
 void AlignedSentence::CreateConsistentPhrases(const Parameter &params)

contrib/other-builds/extract-mixed-syntax/AlignedSentenceSyntax.cpp

@@ -28,24 +28,33 @@ AlignedSentenceSyntax::~AlignedSentenceSyntax() {
 
 void AlignedSentenceSyntax::Create(const Parameter &params)
 {
-	m_sourceTree.SetDefaultLabel("[X]");
-	m_targetTree.SetDefaultLabel("[X]");
-
 	// parse source and target string
 	if (params.sourceSyntax) {
 		m_sourceStr = "<xml>" + m_sourceStr + "</xml>";
 		XMLParse(m_source, m_sourceTree, m_sourceStr, params);
+
+		if (params.mixedSyntaxType != 0) {
+			// mixed syntax. Always add [X] where there isn't 1
+			m_sourceTree.SetDefaultLabel("[X]");
+		}
 	}
 	else {
 		PopulateWordVec(m_source, m_sourceStr);
+		m_sourceTree.SetDefaultLabel("[X]");
 	}
 
 	if (params.targetSyntax) {
 		m_targetStr = "<xml>" + m_targetStr + "</xml>";
 		XMLParse(m_target, m_targetTree, m_targetStr, params);
+
+		if (params.mixedSyntaxType != 0) {
+			// mixed syntax. Always add [X] where there isn't 1
+			m_targetTree.SetDefaultLabel("[X]");
+		}
 	}
 	else {
 		PopulateWordVec(m_target, m_targetStr);
+		m_targetTree.SetDefaultLabel("[X]");
 	}
 
 	PopulateAlignment(m_alignmentStr);

contrib/other-builds/extract-mixed-syntax/ConsistentPhrases.cpp

@@ -84,7 +84,7 @@ std::string ConsistentPhrases::Debug() const
 	return out.str();
 }
 
-void ConsistentPhrases::AddHieroNonTerms(bool onlyIfNoNonTerm)
+void ConsistentPhrases::AddHieroNonTerms()
 {
 	// add [X] labels everywhere
 	for (int i = 0; i < m_coll.size(); ++i) {
@@ -94,10 +94,6 @@ void ConsistentPhrases::AddHieroNonTerms(bool onlyIfNoNonTerm)
 			ConsistentPhrases::Coll::iterator iter;
 			for (iter = coll.begin(); iter != coll.end(); ++iter) {
 				ConsistentPhrase &cp = **iter;
-
-				if (onlyIfNoNonTerm && cp.GetNonTerms().size()) {
-					continue;
-				}
 				cp.AddNonTerms("[X]", "[X]");
 			}
 		}

contrib/other-builds/extract-mixed-syntax/ConsistentPhrases.h

@@ -25,7 +25,7 @@ public:
 	void Add(int sourceStart, int sourceEnd,
 			int targetStart, int targetEnd);
 
-	void AddHieroNonTerms(bool onlyIfNoNonTerm);
+	void AddHieroNonTerms();
 
 	const Coll &GetColl(int sourceStart, int sourceEnd) const;
 	Coll &GetColl(int sourceStart, int sourceEnd);

contrib/other-builds/extract-mixed-syntax/Main.cpp

@@ -25,7 +25,8 @@ int main(int argc, char** argv)
     ("help", "Print help messages")
     ("MaxSpan", po::value<int>()->default_value(params.maxSpan), "Max (source) span of a rule. ie. number of words in the source")
     ("SourceSyntax", "Source sentence is a parse tree")
-    ("TargetSyntax", "Target sentence is a parse tree");
+    ("TargetSyntax", "Target sentence is a parse tree")
+    ("MixedSyntaxType", po::value<int>()->default_value(params.mixedSyntaxType), "Hieu's Mixed syntax type. 0(default)=no mixed syntax, 1=add [X] only if no syntactic label. 2=add [X] everywhere");
 
   po::variables_map vm;
   try
@@ -52,9 +53,10 @@ int main(int argc, char** argv)
     return EXIT_FAILURE;
   }
 
-  if (vm.count("maxSpan")) params.maxSpan = vm["maxSpan"].as<int>();
+  if (vm.count("MaxSpan")) params.maxSpan = vm["MaxSpan"].as<int>();
   if (vm.count("SourceSyntax")) params.sourceSyntax = true;
   if (vm.count("TargetSyntax")) params.targetSyntax = true;
+  if (vm.count("MixedSyntaxType")) params.mixedSyntaxType = vm["MixedSyntaxType"].as<int>();
 
   // input files;
   string pathTarget = argv[1];

contrib/other-builds/extract-mixed-syntax/Parameter.cpp

@@ -18,6 +18,7 @@ Parameter::Parameter()
 
 ,sourceSyntax(false)
 ,targetSyntax(false)
+,mixedSyntaxType(0)
 {}
 
 Parameter::~Parameter() {

contrib/other-builds/extract-mixed-syntax/Parameter.h

@@ -24,6 +24,7 @@ public:
   bool fractionalCounting;
 
   bool sourceSyntax, targetSyntax;
+  int mixedSyntaxType;
 
 };