fixed a bug related to the handling of feature names

git-svn-id: https://mosesdecoder.svn.sourceforge.net/svnroot/mosesdecoder/trunk@1821 1f5c12ca-751b-0410-a591-d2e778427230

mert/Data.cpp

@@ -62,33 +62,31 @@ void Data::loadnbest(const std::string &file)
 		nextPound = getNextPound(stringBuf, substring, "|||"); //third field
 
 		if (!existsFeatureNames()){
-		std::string stringsupport=substring;
-		// adding feature names
-		std::string tmpname="";
-		size_t tmpidx=0;
-		while (!stringsupport.empty()){
-//			TRACE_ERR("Decompounding: " << substring << std::endl); 
-			nextPound = getNextPound(stringsupport, subsubstring);
-			
-			// string ending with ":" are skipped, because they are the names of the features
-			if ((loc = subsubstring.find(":")) != subsubstring.length()-1){
-				featname2idx_[tmpname+"_"+stringify(tmpidx)]=idx2featname_.size();
-				idx2featname_[idx2featname_.size()]=tmpname+"_"+stringify(tmpidx);
-				tmpidx++;
-			}
-			else{
-				tmpidx=0;
-				tmpname=subsubstring.substr(0,subsubstring.size() - 1);
-			}
-		}
-		std::string features="";
-		for (size_t i=0; i<idx2featname_.size(); i++)
-			features+=idx2featname_[i]+" ";
+			std::string stringsupport=substring;
+			// adding feature names
+			std::string features="";
+			std::string tmpname="";
 
-		NumberOfFeatures(idx2featname_.size());
-		Features(features);
-		TRACE_ERR("number_of_features: " << NumberOfFeatures() << std::endl); 
-		TRACE_ERR("features: " << Features() << std::endl); 
+			size_t tmpidx=0;
+			while (!stringsupport.empty()){
+				//			TRACE_ERR("Decompounding: " << substring << std::endl); 
+				nextPound = getNextPound(stringsupport, subsubstring);
+				
+				// string ending with ":" are skipped, because they are the names of the features
+				if ((loc = subsubstring.find(":")) != subsubstring.length()-1){
+					features+=tmpname+"_"+stringify(tmpidx)+" ";
+					tmpidx++;
+				}
+				else{
+					tmpidx=0;
+					tmpname=subsubstring.substr(0,subsubstring.size() - 1);
+				}
+			}
+			
+			NumberOfFeatures(idx2featname_.size());
+			Features(features);
+		
+			setFeatureMap();
 		}
 		
 // adding features
@@ -103,6 +101,18 @@ void Data::loadnbest(const std::string &file)
 		}
 		featdata->add(featentry,sentence_index);
 	}
-
+	
 	inp.close();
 }
+
+void Data::setFeatureMap()
+{
+	std::string substring, stringBuf;
+	stringBuf=Features();
+	while (!stringBuf.empty()){
+		getNextPound(stringBuf, substring);
+				
+		featname2idx_[substring]=idx2featname_.size();
+		idx2featname_[idx2featname_.size()]=substring;
+	}
+}

mert/Data.h

@@ -54,6 +54,7 @@ public:
   void load(const std::string &featfile,const std::string &scorefile){
 		featdata->load(featfile);
 		scoredata->load(scorefile);
+		setFeatureMap();
   }
 	
 	void save(const std::string &featfile,const std::string &scorefile, bool bin=false){
@@ -62,8 +63,17 @@ public:
 	}
 
 	bool existsFeatureNames(){ return (idx2featname_.size() > 0)?true:false; };
-	std::string getFeatureName(size_t idx){ return idx2featname_[idx]; };
-  size_t getFeatureIndex(const std::string& name){ return featname2idx_[name]; };
+	std::string getFeatureName(size_t idx){
+		if (idx >= idx2featname_.size())
+			throw runtime_error("Error: you required the " + idx + "-th feature, but there are only " + (idx2featname_.size()) + " features");
+		return idx2featname_[idx];
+	};
+  size_t getFeatureIndex(const std::string& name){
+		if (featname2idx_.find(name)!=featname2idx_.end())
+			throw runtime_error("Error: feature is unknown");
+		return featname2idx_[name];
+	};
+  void setFeatureMap();
 };
 
 

mert/FeatureData.cpp

@@ -39,9 +39,7 @@ void FeatureData::load(ifstream& inFile)
 {
   FeatureArray entry;
 
-	int iter=0;
 	while (!inFile.eof()){
-		
 
 		if (!inFile.good()){
 			std::cerr << "ERROR FeatureData::load inFile.good()" << std::endl; 
@@ -50,14 +48,15 @@ void FeatureData::load(ifstream& inFile)
 		entry.clear();
 		entry.load(inFile);
 
-		if (entry.size() == 0){
-			return;
+		if (entry.size() == 0)
+			break;
+
+		if (size() == 0){
+			number_of_features=entry.NumberOfFeatures();
+			features=entry.Features();
 		}
 		add(entry);
-		iter++;
 	}
-	if (size()>0)
-		number_of_features=get(0).NumberOfFeatures();
 }
 
 

mert/ScoreArray.cpp

@@ -63,7 +63,6 @@ void ScoreArray::loadtxt(ifstream& inFile)
   std::string::size_type loc;
 
 
-	TRACE_ERR("starting loadtxt..." << std::endl);
 	std::getline(inFile, stringBuf);
 	if (!inFile.good()){
 		return;
@@ -93,7 +92,7 @@ void ScoreArray::loadtxt(ifstream& inFile)
 	std::getline(inFile, stringBuf);
 	if (!stringBuf.empty()){         
 //		TRACE_ERR("Reading: " << stringBuf << std::endl); 
-                if ((loc = stringBuf.find(SCORES_TXT_END)) != 0){
+		if ((loc = stringBuf.find(SCORES_TXT_END)) != 0){
 			TRACE_ERR("ERROR: ScoreArray::loadtxt(): Wrong footer");
 			return;
 		}

mert/ScoreData.cpp

@@ -46,8 +46,6 @@ void ScoreData::load(ifstream& inFile)
 
 	int iter=0;
 	while (!inFile.eof()){
-		TRACE_ERR("iter " << iter << " size " << size() << std::endl);
-		
 		entry.clear();
 		entry.loadtxt(inFile);