add HMMsegment into demo but with bug unsolved.

demo/Makefile

@@ -24,7 +24,6 @@ SRCDIR = ../src
 SRCLIB = $(SRCDIR)/libcppjieba.a
 
 # remove the objs after compilation
-.INTERMEDIATE:  $(OBJS) 
 .PHONY: clean $(SRCLIB)
 
 # Main Targets

demo/segment_demo.cpp

@@ -5,7 +5,8 @@
 using namespace CppJieba;
 
 Segment seg;
-bool init(const char * const filePath)
+HMMSegment hmmseg;
+bool init(const char * const dictPath, const char * const modelPath)
 {
 	if(!seg.init())
 	{
@@ -13,11 +14,16 @@ bool init(const char * const filePath)
 		return false;
 	}
 
-	if(!seg.loadSegDict(filePath))
+	if(!seg.loadSegDict(dictPath))
 	{
 		cout<<"seg loadDict failed."<<endl;
 		return false;
 	}
+	if(!hmmseg.loadModel(modelPath))
+	{
+		cout<<"hmmseg loadModel failed."<<endl;
+		return false;
+	}
 	return true;
 }
 
@@ -37,6 +43,22 @@ void run(const char * const filePath)
 	}
 }
 
+void hmmrun(const char * const filePath)
+{
+	ifstream ifile(filePath);
+	vector<string> res;
+	string line;
+	while(getline(ifile, line))
+	{
+		res.clear();
+		if(!line.empty())
+		{
+			hmmseg.cut(line, res);
+			cout<<line<<"\n"<<joinStr(res,"/")<<endl;
+		}
+	}
+}
+
 bool dispose()
 {
 	if(!seg.dispose())
@@ -48,22 +70,21 @@ bool dispose()
 }
 
 const char * const DEFAULT_DICTPATH = "../dicts/jieba.dict.utf8";
+const char * const DEFAULT_MODELPATH = "../dicts/hmm_model.utf8";
 
 int main(int argc, char ** argv)
 {
-	/*map<string, string> mpss;
-	getArgvMap(argc, argv, mpss);
-	string enc = getMap<string, string>(mpss, "--encoding", "");
-	string dictPath = getMap<string, string>(mpss, "--dictpath", "../dicts/jieba.dict.gbk");
-	*/
 	if(argc < 2)
 	{
 		cout<<"usage: \n\t"<<argv[0]<<"[options] <filename>\n"
 		    <<"options:\n"
-		    <<"\t--dictpath\tIf is not specified, the default is ../dicts/jieba.dict.utf8\n"
+			<<"\t--algorithm\tSupported encoding methods are [cutDAG, cutHMM] for now. \n\t\t\tIf is not specified, the default is cutDAG\n"
+		    <<"\t--dictpath\tIf is not specified, the default is "<<DEFAULT_DICTPATH<<'\n'
+		    <<"\t--modelpath\tIf is not specified, the default is "<<DEFAULT_MODELPATH<<'\n'
 		    <<"\t--encoding\tSupported encoding methods are [gbk, utf-8] for now. \n\t\t\tIf is not specified, the default is utf8.\n"
 			<<"example:\n"
 			<<"\t"<<argv[0]<<" testlines.utf8 --encoding utf-8 --dictpath ../dicts/jieba.dict.utf8\n"
+			<<"\t"<<argv[0]<<" testlines.utf8 --modelpath ../dicts/hmm_model.utf8 --algorithm cutHMM\n"
 			<<"\t"<<argv[0]<<" testlines.gbk --encoding gbk --dictpath ../dicts/jieba.dict.gbk\n"
 			<<endl;
 		
@@ -71,11 +92,17 @@ int main(int argc, char ** argv)
 	}
 	ArgvContext arg(argc, argv);
 	string dictPath = arg["--dictpath"];
+	string modelPath = arg["--modelpath"];
 	string encoding = arg["--encoding"];
-	if("" == dictPath)
+	string algorithm = arg["--algorithm"];
+	if(dictPath.empty())
 	{
 		dictPath = DEFAULT_DICTPATH;
 	}
+	if(modelPath.empty())
+	{
+		modelPath = DEFAULT_MODELPATH;
+	}
 	if("gbk" == encoding)
 	{
 		TransCode::setGbkEnc();
@@ -85,8 +112,19 @@ int main(int argc, char ** argv)
 		TransCode::setUtf8Enc();
 	}
 
-	init(dictPath.c_str());
+	if(!init(dictPath.c_str(), modelPath.c_str()))
-	run(arg[1].c_str());
+	{
+		LogError("init failed.");
+		return -1;
+	}
+	if("cutHMM" == algorithm)
+	{
+		hmmrun(arg[1].c_str());
+	}
+	else
+	{
+		run(arg[1].c_str());
+	}
 	dispose();
 	return 0;
 }

src/HMMSegment.cpp

@@ -33,6 +33,7 @@ namespace CppJieba
 
 	bool HMMSegment::loadModel(const char* const filePath)
 	{
+		LogInfo(string_format("loadModel [%s] start ...", filePath));
 		ifstream ifile(filePath);
 		string line;
 		vector<string> tmp;
@@ -98,6 +99,8 @@ namespace CppJieba
 			return false;
 		}
 
+		LogInfo(string_format("loadModel [%s] end.", filePath));
+
 		return true;
 	}
 
@@ -117,6 +120,7 @@ namespace CppJieba
 			return false;
 		}
 
+		cout<<__FILE__<<__LINE__<<endl;
 		if(!viterbi(unico, status))
 		{
 			LogError("viterbi failed.");
@@ -183,6 +187,7 @@ namespace CppJieba
 			weight[0 + y * X] = _startProb[y] + _getEmitProb(_emitProbVec[y], unico[0], MIN_DOUBLE);
 			path[0 + y * X] = -1;
 		}
+		cout<<__FILE__<<__LINE__<<endl;
 
 		//process
 		for(uint x = 1; x < X; x++)
@@ -191,6 +196,7 @@ namespace CppJieba
 			{
 				now = x + y*X;
 				weight[now] = MIN_DOUBLE;
+				path[now] = E;
 				for(uint preY = 0; preY < Y; preY++)
 				{
 					old = x - 1 + preY * X;
@@ -209,8 +215,6 @@ namespace CppJieba
 				//cout<<x<<":"<<y<<":"<<weight[now]<<endl;
 				//getchar();
 			}
-			//cout<<_getEmitProb(_emitProbB, unico[x], MIN_DOUBLE)<<endl;
-			//getchar();
 		}
 
 		endE = weight[X-1+E*X];
@@ -224,14 +228,19 @@ namespace CppJieba
 		{
 			stat = S;
 		}
+		cout<<__FILE__<<__LINE__<<endl;
 		
 		status.assign(X, 0);
+		cout<<__FILE__<<__LINE__<<endl;
 		for(int x = X -1 ; x >= 0; x--)
 		{
 			status[x] = stat;
+		cout<<__FILE__<<__LINE__<<endl;
+		cout<<stat<<endl;
 			stat = path[x + stat*X];
 		}
 
+		cout<<__FILE__<<__LINE__<<endl;
 		delete [] path;
 		delete [] weight;
 		return true;
@@ -330,7 +339,7 @@ int main()
 	HMMSegment hmm;
 	hmm.loadModel("../dicts/hmm_model.utf8");
 	vector<string> res;
-	hmm.cut("小明硕士毕业于北邮网络研究院", res);
+	hmm.cut("小明硕士毕业于北邮网络研究院，然", res);
 	cout<<joinStr(res, "/")<<endl;