Stubbing out of Scorer.

Scorer test data.


git-svn-id: https://mosesdecoder.svn.sourceforge.net/svnroot/mosesdecoder/trunk@1646 1f5c12ca-751b-0410-a591-d2e778427230

mert/BleuScorer.cpp

@@ -0,0 +1,5 @@
+#include "BleuScorer.h"
+
+/**
+ Implementation of bleu scoring for use in mert
+ **/

mert/BleuScorer.h

@@ -0,0 +1,6 @@
+#ifndef __BLEU_SCORER_H__
+#define __BLEU_SCORER_H__
+
+#include "Scorer.h"
+
+#endif //__BLEU_SCORER_H__

mert/Makefile

@@ -1,13 +1,13 @@
 OBJS= Util.o Timer.o Parameter.o \
 ScoreStats.o ScoreArray.o ScoreData.o \
-FeatureStats.o FeatureArray.o FeatureData.o Optimizer.o Point.o
+Optimizer.o FeatureStats.o FeatureArray.o FeatureData.o  Point.o BleuScorer.o
 
 CFLAGS=-O3 -DTRACE_ENABLE
 GCC=g++
 LDFLAGS=
 LDLIBS=-lm 
 
-all: main feature_extractor
+all: main feature_extractor test_scorer
 
 clean: 
 	rm -f *.o
@@ -21,3 +21,6 @@ feature_extractor: $(OBJS) feature_extractor.cpp
 main: $(OBJS) main.cpp
 	$(GCC) $(CFLAGS) $(OBJS) $(LDLIBS) -o $@ $@.cpp
 
+test_scorer: $(OBJS) test_scorer.cpp
+	$(GCC) $(CFLAGS) $(OBJS) $(LDLIBS) -o $@ $@.cpp
+

mert/ScoreData.cpp

@@ -14,8 +14,7 @@
 ScoreData::ScoreData(Scorer& ptr):
 bufLen_(0), theScorer(&ptr)
 {
-	score_type = "BLEU4";
-//	score_type = theScorer.name();
+	score_type = theScorer->getName();
 };
 
 void ScoreData::savetxt(std::ofstream& outFile)
@@ -107,7 +106,7 @@ void ScoreData::loadnbest(const std::string &file)
  * theStatistics will contain the statistics (as a string)
  * coming from the actual Scorer
  */
-//		theStatistics = theScorer.getStatistics(sentence_index, theSentence);
+		theScorer->prepareStats(sentence_index, theSentence,entry);
 		entry.set(theStatistics);
 
 		add(entry,sentence_index);

mert/Scorer.h

@@ -1,35 +1,63 @@
 #ifndef __SCORER_H__
 #define __SCORER_H__
 
+#include <algorithm>
 #include <iostream>
+#include <stdexcept>
 #include <string>
 #include <vector>
 
-using namespace std;
+
+class ScoreData;
+class ScoreStats;
 
 class Scorer {
-	/**
-	  * Extract initial statistics from the nbestfile and write them to the stats
-      * file. For example, for bleu these are the ngram counts and length.
-	  **/
+
 	public:
 
-		Scorer(const string& statsfile): _statsfile(statsfile) {}
+		Scorer(const std::string& name): _name(name), _scoreData(0)  {}
+
+		const std::string& getName() const {return _name;}
+
+		/**
+		  * set the reference files. This must be called before prepareStats.
+		  **/
+		void setReferenceFiles(const std::vector<std::string>& referenceFiles) {
+			_referenceFiles.clear();
+			_referenceFiles.resize(referenceFiles.size());
+			std::copy(referenceFiles.begin(),referenceFiles.end(),_referenceFiles.begin());
+		}
 
-		virtual void prepare(const vector<string>& referencefiles, const string& nbestfile) {
-			//dummy impl
+		/**
+		 * Process the given guessed text, corresponding to the given reference sindex
+         * and add the appropriate statistics to the entry.
+		**/
+		virtual void prepareStats(int sindex, const std::string& text, ScoreStats& entry) {
+			//std::cerr << text << std::endl;
 		}
-	/**
-	  * Calculate the score of the sentences corresponding to the list of candidate
-      * indices. Each index indicates the 1-best choice from the n-best list.
-	  **/
-		virtual float score(const vector<unsigned int>& candidates) {
-			//dummy impl
+
+		/**
+		  * Set the score data, prior to scoring.
+		  **/
+		void setScoreData(ScoreData* scoreData) {
+			_scoreData = scoreData;
+		}
+
+		/**
+		  * Calculate the score of the sentences corresponding to the list of candidate
+		  * indices. Each index indicates the 1-best choice from the n-best list.
+		  **/
+		virtual float score(const std::vector<unsigned int>& candidates) {
+			if (!_scoreData) {
+				throw std::runtime_error("score data not loaded");
+			}
 			return 0;
 		}
 
-	protected:
-		string _statsfile;
+		private:
+			std::string _name;
+			std::vector<std::string> _referenceFiles;
+			ScoreData* _scoreData;
 
 };
 

mert/test_scorer.cpp

@@ -0,0 +1,18 @@
+#include <iostream>
+#include <vector>
+
+#include "ScoreData.h"
+#include "BleuScorer.h"
+
+using namespace std;
+
+int main(int argc, char** argv) {
+	cout << "Testing the scorer" << endl;	
+	//BleuScorer bs("test-scorer-data/cppstats.feats.opt");;
+	vector<string> references;
+	references.push_back("test_scorer_data/reference.txt");
+	//bs.prepare(references, "test-scorer-data/nbest.out");
+	Scorer scorer("test");
+	ScoreData sd(scorer);
+	sd.loadnbest("test_scorer_data/nbest.out");
+}