comments

7 files changed, 9 insertions, 1 deletions

diff --git a/contrib/other-builds/extract-mixed-syntax/Main.cpp b/contrib/other-builds/extract-mixed-syntax/Main.cpp
index f5d12f651..4cb5c92e6 100644
--- a/contrib/other-builds/extract-mixed-syntax/Main.cpp
+++ b/contrib/other-builds/extract-mixed-syntax/Main.cpp
@@ -35,7 +35,8 @@ int main(int argc, char** argv)
     ("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")
     ("MultiLabel", po::value<int>()->default_value(params.multiLabel), "What to do with multiple labels on the same span. 0(default)=keep them all, 1=keep only top-most, 2=keep only bottom-most")
-    ("HieroSourceLHS", "Always use Hiero source LHS? Default = 0");
+    ("HieroSourceLHS", "Always use Hiero source LHS? Default = 0")
+    ("MaxSpanFreeNonTermSource", "Max number of words covered by beginning/end NT. Default = 0 (no limit)");
 
   po::variables_map vm;
   try
@@ -74,6 +75,7 @@ int main(int argc, char** argv)
   if (vm.count("MixedSyntaxType")) params.mixedSyntaxType = vm["MixedSyntaxType"].as<int>();
   if (vm.count("MultiLabel")) params.multiLabel = vm["MultiLabel"].as<int>();
   if (vm.count("HieroSourceLHS")) params.hieroSourceLHS = true;
+  if (vm.count("MaxSpanFreeNonTermSource")) params.maxSpanFreeNonTermSource = vm["MaxSpanFreeNonTermSource"].as<int>();
 
   // input files;
   string pathTarget = argv[1];
diff --git a/contrib/other-builds/extract-mixed-syntax/Parameter.cpp b/contrib/other-builds/extract-mixed-syntax/Parameter.cpp
index 2dd041c10..1a1b213bc 100644
--- a/contrib/other-builds/extract-mixed-syntax/Parameter.cpp
+++ b/contrib/other-builds/extract-mixed-syntax/Parameter.cpp
@@ -27,6 +27,7 @@ Parameter::Parameter()
 ,multiLabel(0)
 ,nonTermConsecSourceMixed(true)
 ,hieroSourceLHS(false)
+,maxSpanFreeNonTermSource(0)
 {}
 
 Parameter::~Parameter() {
diff --git a/contrib/other-builds/extract-mixed-syntax/Parameter.h b/contrib/other-builds/extract-mixed-syntax/Parameter.h
index 49be95a8a..fbed7a1b8 100644
--- a/contrib/other-builds/extract-mixed-syntax/Parameter.h
+++ b/contrib/other-builds/extract-mixed-syntax/Parameter.h
@@ -36,6 +36,7 @@ public:
   int mixedSyntaxType, multiLabel;
   bool nonTermConsecSourceMixed;
   bool hieroSourceLHS;
+  int maxSpanFreeNonTermSource;
 
 };
 
diff --git a/contrib/other-builds/extract-mixed-syntax/Phrase.h b/contrib/other-builds/extract-mixed-syntax/Phrase.h
index 8d52d23b3..13912cb95 100644
--- a/contrib/other-builds/extract-mixed-syntax/Phrase.h
+++ b/contrib/other-builds/extract-mixed-syntax/Phrase.h
@@ -3,6 +3,7 @@
 #include <vector>
 #include "Word.h"
 
+// a vector of terminals
 class Phrase : public std::vector<Word*>
 {
 public:
diff --git a/contrib/other-builds/extract-mixed-syntax/RulePhrase.h b/contrib/other-builds/extract-mixed-syntax/RulePhrase.h
index 382b50d37..0bb170644 100644
--- a/contrib/other-builds/extract-mixed-syntax/RulePhrase.h
+++ b/contrib/other-builds/extract-mixed-syntax/RulePhrase.h
@@ -14,6 +14,7 @@
 
 class RuleSymbol;
 
+// a phrase of terms and non-terms for 1 side of a rule
 class RulePhrase
 {
 public:
diff --git a/contrib/other-builds/extract-mixed-syntax/RuleSymbol.h b/contrib/other-builds/extract-mixed-syntax/RuleSymbol.h
index ef86a5034..c292fcc0d 100644
--- a/contrib/other-builds/extract-mixed-syntax/RuleSymbol.h
+++ b/contrib/other-builds/extract-mixed-syntax/RuleSymbol.h
@@ -11,6 +11,7 @@
 #include <iostream>
 #include <string>
 
+// base class - terminal or non-term
 class RuleSymbol {
 public:
 	RuleSymbol();
diff --git a/contrib/other-builds/extract-mixed-syntax/Word.h b/contrib/other-builds/extract-mixed-syntax/Word.h
index 251cbac4f..46feba07c 100644
--- a/contrib/other-builds/extract-mixed-syntax/Word.h
+++ b/contrib/other-builds/extract-mixed-syntax/Word.h
@@ -10,6 +10,7 @@
 #include <set>
 #include "RuleSymbol.h"
 
+// a terminal
 class Word : public RuleSymbol
 {
 public: