use open nlp chunker

7 files changed, 89 insertions, 24 deletions

diff --git a/contrib/other-builds/manual-label/DeEn.cpp b/contrib/other-builds/manual-label/DeEn.cpp
index 03aa57e92..460bee367 100644
--- a/contrib/other-builds/manual-label/DeEn.cpp
+++ b/contrib/other-builds/manual-label/DeEn.cpp
@@ -1,6 +1,6 @@
 #include <list>
 #include "DeEn.h"
-#include "manual-label.h"
+#include "Main.h"
 #include "moses/Util.h"
 
 using namespace std;
diff --git a/contrib/other-builds/manual-label/DeEn.h b/contrib/other-builds/manual-label/DeEn.h
index 531b56721..c24ce0079 100644
--- a/contrib/other-builds/manual-label/DeEn.h
+++ b/contrib/other-builds/manual-label/DeEn.h
@@ -1,5 +1,5 @@
 #pragma once
 
-#include "manual-label.h"
+#include "Main.h"
 
 void LabelDeEn(const Phrase &source, std::ostream &out);
diff --git a/contrib/other-builds/manual-label/EnOpenNLPChunker.cpp b/contrib/other-builds/manual-label/EnOpenNLPChunker.cpp
new file mode 100644
index 000000000..7ac5b49b1
--- /dev/null
+++ b/contrib/other-builds/manual-label/EnOpenNLPChunker.cpp
@@ -0,0 +1,24 @@
+/*
+ * EnApacheChunker.cpp
+ *
+ *  Created on: 28 Feb 2014
+ *      Author: hieu
+ */
+
+#include "EnOpenNLPChunker.h"
+
+EnOpenNLPChunker::EnOpenNLPChunker(const std::string &openNLPPath)
+:m_openNLPPath(openNLPPath)
+{
+	// TODO Auto-generated constructor stub
+
+}
+
+EnOpenNLPChunker::~EnOpenNLPChunker() {
+	// TODO Auto-generated destructor stub
+}
+
+void EnOpenNLPChunker::Process(std::istream &in, std::ostream &out)
+{
+
+}
diff --git a/contrib/other-builds/manual-label/EnOpenNLPChunker.h b/contrib/other-builds/manual-label/EnOpenNLPChunker.h
new file mode 100644
index 000000000..773a5017a
--- /dev/null
+++ b/contrib/other-builds/manual-label/EnOpenNLPChunker.h
@@ -0,0 +1,21 @@
+/*
+ * EnApacheChunker.h
+ *
+ *  Created on: 28 Feb 2014
+ *      Author: hieu
+ */
+
+#pragma once
+
+#include <string>
+#include <iostream>
+
+class EnOpenNLPChunker {
+public:
+	EnOpenNLPChunker(const std::string &openNLPPath);
+	virtual ~EnOpenNLPChunker();
+	void Process(std::istream &in, std::ostream &out);
+protected:
+	const std::string m_openNLPPath;
+};
+
diff --git a/contrib/other-builds/manual-label/EnPhrasalVerb.h b/contrib/other-builds/manual-label/EnPhrasalVerb.h
index 0f184257b..4cb5f7348 100644
--- a/contrib/other-builds/manual-label/EnPhrasalVerb.h
+++ b/contrib/other-builds/manual-label/EnPhrasalVerb.h
@@ -1,7 +1,8 @@
 #pragma once
 
-#include "manual-label.h"
+#include "Main.h"
 
+// roll your own identification of phrasal verbs
 void EnPhrasalVerb(const Phrase &source, int revision, std::ostream &out);
 
 bool Exist(const Phrase &source, int start, int end, int factor, const std::string &str);
diff --git a/contrib/other-builds/manual-label/manual-label.cpp b/contrib/other-builds/manual-label/Main.cpp
index d81b7c1e1..e9018ffe5 100644
--- a/contrib/other-builds/manual-label/manual-label.cpp
+++ b/contrib/other-builds/manual-label/Main.cpp
@@ -2,9 +2,10 @@
 #include <cstdlib>
 #include <boost/program_options.hpp>
 #include "moses/Util.h"
-#include "manual-label.h"
+#include "Main.h"
 #include "DeEn.h"
 #include "EnPhrasalVerb.h"
+#include "EnOpenNLPChunker.h"
 
 using namespace std;
 
@@ -23,7 +24,11 @@ int main(int argc, char** argv)
     ("add", "additional options")
     ("source-language,s", po::value<string>()->required(), "Source Language")
     ("target-language,t", po::value<string>()->required(), "Target Language")
-    ("revision,r", po::value<int>()->default_value(0), "Revision");
+    ("revision,r", po::value<int>()->default_value(0), "Revision")
+
+    ("opennlp-path", po::value<int>()->default_value(0), "Path to Apache OpenNLP toolkit")
+
+    ;
 
   po::variables_map vm;
   try
@@ -56,27 +61,41 @@ int main(int argc, char** argv)
 
   cerr << sourceLang << " " << targetLang << " " << revision << endl;
 
-  string line;
-  size_t lineNum = 1;
-
-  while (getline(cin, line)) {
-    //cerr << lineNum << ":" << line << endl;
-    if (lineNum % 1000 == 0) {
-      cerr << lineNum << " ";
-    }
-
-    Phrase source = Tokenize(line);
 
-    if (sourceLang == "de" && targetLang == "en") {
-    	LabelDeEn(source, cout);
-    }
-    else if (sourceLang == "en") {
-    	EnPhrasalVerb(source, revision, cout);
-    }
-
-    ++lineNum;
+  if (sourceLang == "en" && revision == 2) {
+	string openNLPPath = vm["opennlp-path"].as<string>();
+  	EnOpenNLPChunker chunker(openNLPPath);
+  	chunker.Process(cin, cout);
+  }
+  else {
+	  // process line-by-line
+	  string line;
+	  size_t lineNum = 1;
+
+	  while (getline(cin, line)) {
+		//cerr << lineNum << ":" << line << endl;
+		if (lineNum % 1000 == 0) {
+		  cerr << lineNum << " ";
+		}
+
+		Phrase source = Tokenize(line);
+
+		if (sourceLang == "de" && targetLang == "en") {
+			LabelDeEn(source, cout);
+		}
+		else if (sourceLang == "en") {
+			if (revision == 0 || revision == 1) {
+				EnPhrasalVerb(source, revision, cout);
+			}
+			else if (revision == 2) {
+				  string openNLPPath = vm["opennlp-path"].as<string>();
+				  EnOpenNLPChunker chunker(openNLPPath);
+			}
+		}
+
+		++lineNum;
+	  }
   }
-
 
 
   cerr << "Finished" << endl;
diff --git a/contrib/other-builds/manual-label/manual-label.h b/contrib/other-builds/manual-label/Main.h
index bc93fe003..bc93fe003 100644
--- a/contrib/other-builds/manual-label/manual-label.h
+++ b/contrib/other-builds/manual-label/Main.h