add MaxNonTerm and MaxHieroNonTerm

6 files changed, 26 insertions, 0 deletions

diff --git a/contrib/other-builds/extract-mixed-syntax/Main.cpp b/contrib/other-builds/extract-mixed-syntax/Main.cpp
index 48bd7aab4..105224c3d 100644
--- a/contrib/other-builds/extract-mixed-syntax/Main.cpp
+++ b/contrib/other-builds/extract-mixed-syntax/Main.cpp
@@ -28,6 +28,8 @@ int main(int argc, char** argv)
     ("GlueGrammar", po::value<string>()->default_value(params.gluePath), "Output glue grammar to here")
     ("SentenceOffset", po::value<long>()->default_value(params.sentenceOffset), "Starting sentence id. Not used")
     ("GZOutput", "Compress extract files")
+    ("MaxNonTerm", po::value<int>()->default_value(params.maxNonTerm), "Maximum number of non-terms allowed per rule")
+    ("MaxHieroNonTerm", po::value<int>()->default_value(params.maxHieroNonTerm), "Maximum number of Hiero non-term. Usually, --MaxNonTerm is the normal constraint")
 
     ("SourceSyntax", "Source sentence is a parse tree")
     ("TargetSyntax", "Target sentence is a parse tree")
@@ -62,6 +64,8 @@ int main(int argc, char** argv)
   if (vm.count("GZOutput")) params.gzOutput = true;
   if (vm.count("GlueGrammar")) params.gluePath = vm["GlueGrammar"].as<string>();
   if (vm.count("SentenceOffset")) params.sentenceOffset = vm["SentenceOffset"].as<long>();
+  if (vm.count("MaxNonTerm")) params.maxNonTerm = vm["MaxNonTerm"].as<int>();
+  if (vm.count("MaxHieroNonTerm")) params.maxHieroNonTerm = vm["MaxHieroNonTerm"].as<int>();
 
   if (vm.count("SourceSyntax")) params.sourceSyntax = true;
   if (vm.count("TargetSyntax")) params.targetSyntax = true;
diff --git a/contrib/other-builds/extract-mixed-syntax/NonTerm.cpp b/contrib/other-builds/extract-mixed-syntax/NonTerm.cpp
index 0118709f9..e9c6d74d0 100644
--- a/contrib/other-builds/extract-mixed-syntax/NonTerm.cpp
+++ b/contrib/other-builds/extract-mixed-syntax/NonTerm.cpp
@@ -56,3 +56,8 @@ bool NonTerm::IsHiero(Moses::FactorDirection direction, const Parameter &params)
 	const std::string &label = NonTerm::GetLabel(direction);
 	return label == params.hieroNonTerm;
 }
+
+bool NonTerm::IsHiero(const Parameter &params) const
+{
+ return IsHiero(Moses::Input, params) && IsHiero(Moses::Output, params);
+}
diff --git a/contrib/other-builds/extract-mixed-syntax/NonTerm.h b/contrib/other-builds/extract-mixed-syntax/NonTerm.h
index 452014da5..3565c5750 100644
--- a/contrib/other-builds/extract-mixed-syntax/NonTerm.h
+++ b/contrib/other-builds/extract-mixed-syntax/NonTerm.h
@@ -36,6 +36,7 @@ public:
 
   const std::string &GetLabel(Moses::FactorDirection direction) const;
   bool IsHiero(Moses::FactorDirection direction, const Parameter &params) const;
+  bool IsHiero(const Parameter &params) const;
 
 protected:
 	const ConsistentPhrase *m_consistentPhrase;
diff --git a/contrib/other-builds/extract-mixed-syntax/Parameter.cpp b/contrib/other-builds/extract-mixed-syntax/Parameter.cpp
index 39901ea17..77aa8436b 100644
--- a/contrib/other-builds/extract-mixed-syntax/Parameter.cpp
+++ b/contrib/other-builds/extract-mixed-syntax/Parameter.cpp
@@ -9,6 +9,7 @@
 Parameter::Parameter()
 :maxSpan(10)
 ,maxNonTerm(2)
+,maxHieroNonTerm(999)
 ,maxSymbolsTarget(999)
 ,maxSymbolsSource(5)
 ,minHoleSource(2)
diff --git a/contrib/other-builds/extract-mixed-syntax/Parameter.h b/contrib/other-builds/extract-mixed-syntax/Parameter.h
index 491565744..ead94b7f9 100644
--- a/contrib/other-builds/extract-mixed-syntax/Parameter.h
+++ b/contrib/other-builds/extract-mixed-syntax/Parameter.h
@@ -16,6 +16,7 @@ public:
 
   int maxSpan;
   int maxNonTerm;
+  int maxHieroNonTerm;
   int maxSymbolsTarget;
   int maxSymbolsSource;
   int minHoleSource;
diff --git a/contrib/other-builds/extract-mixed-syntax/Rule.cpp b/contrib/other-builds/extract-mixed-syntax/Rule.cpp
index 1fd21fefe..b8fa7e385 100644
--- a/contrib/other-builds/extract-mixed-syntax/Rule.cpp
+++ b/contrib/other-builds/extract-mixed-syntax/Rule.cpp
@@ -191,10 +191,16 @@ void Rule::Prevalidate(const Parameter &params)
 
   // check number of non-terms
   int numNonTerms = 0;
+  int numHieroNonTerms = 0;
   for (size_t i = 0; i < m_source.GetSize(); ++i) {
 	  const RuleSymbol *arc = m_source[i];
 	  if (arc->IsNonTerm()) {
 		  ++numNonTerms;
+		  const NonTerm &nonTerm = *static_cast<const NonTerm*>(arc);
+		  bool isHiero = nonTerm.IsHiero(params);
+		  if (isHiero) {
+			  ++numHieroNonTerms;
+		  }
 	  }
   }
 
@@ -206,6 +212,14 @@ void Rule::Prevalidate(const Parameter &params)
 	  }
   }
 
+  if (numHieroNonTerms >= params.maxHieroNonTerm) {
+	  m_canRecurse = false;
+	  if (numHieroNonTerms > params.maxHieroNonTerm) {
+		  m_isValid = false;
+		  return;
+	  }
+  }
+
   // check if 2 consecutive non-terms in source
   if (!params.nonTermConsecSource && m_nonterms.size() >= 2) {
 	  const NonTerm &lastNonTerm = *m_nonterms.back();