添加一些单元测试

qdj0511 · Mar 28, 2017 · 3011d98 · 3011d98
1 parent 4fb1332
commit 3011d98
Show file tree

Hide file tree

Showing 4 changed files with 162 additions and 6 deletions.
diff --git a/Makefile b/Makefile
@@ -2,24 +2,28 @@ dst_dir=.
 include_dir=include
 src_dir=src
 bin_dir=.
+test_dir=test
 thulac=g++ -O3 -march=native -I $(include_dir)
 
 # all: $(bin_dir)/thulac_test $(bin_dir)/train_c $(bin_dir)/thulac
-all: $(bin_dir)/thulac $(bin_dir)/train_c
+all: $(bin_dir)/thulac $(bin_dir)/train_c $(bin_dir)/thulac_test
 
 $(bin_dir)/thulac: $(src_dir)/thulac.cc $(include_dir)/*.h
 	$(thulac) $(src_dir)/thulac.cc -o $(bin_dir)/thulac
 
 $(bin_dir)/train_c: $(src_dir)/train_c.cc $(include_dir)/*.h
 	$(thulac) -o $(bin_dir)/train_c $(src_dir)/train_c.cc
 
-$(bin_dir)/thulac_test: $(src_dir)/thulac_test.cc $(include_dir)/*.h
-	$(thulac) -o $(bin_dir)/thulac_test $(src_dir)/thulac_test.cc	
+$(bin_dir)/thulac_test: $(test_dir)/test_case.cpp $(include_dir)/*.h
+	$(thulac) -o $(bin_dir)/thulac_test $(test_dir)/test_case.cpp
+
+# $(bin_dir)/thulac_test: $(src_dir)/thulac_test.cc $(include_dir)/*.h
+# 	$(thulac) -o $(bin_dir)/thulac_test $(src_dir)/thulac_test.cc	
 
 clean:
 	rm -f $(bin_dir)/thulac 
 	rm -f $(bin_dir)/train_c 
-	rm -f $(bin_dir)/thulac_test 
+	rm -f $(bin_dir)/thulac_test
 
 pack:
 	tar -czvf THULAC_lite_c++_v1.tar.gz src Makefile doc README.md
diff --git a/include/thulac.h b/include/thulac.h
@@ -22,7 +22,8 @@ class THULAC {
 public:
 	int init(const char* model_path = NULL, const char* user_path = NULL, int just_seg = 0, int t2s = 0, int ufilter = 0, char separator = '_');
 	void deinit();
-	int cut(const std::string&, THULAC_result&);	
+	int cut(const std::string&, THULAC_result&);
+    std::string toString(const THULAC_result&);
 	THULAC() {
 		user_specified_dict_name=NULL;
 		model_path_char=NULL;
@@ -247,7 +248,9 @@ int THULAC::cut(const std::string &in, THULAC_result& result) {
 	                for(int j = 0; j < tagged.size(); j++) {
 	                	ous.str("");
 						ous << tagged[j].word;
-	                	result.push_back(std::make_pair<std::string, std::string>(ous.str(), tagged[j].tag));
+                        std::string s = tagged[j].tag;
+//                        char * ss = s;
+	                	result.push_back(std::make_pair<std::string, std::string>(ous.str(), s.c_str()));
 	                }
 
 				}
@@ -269,5 +272,16 @@ int THULAC::cut(const std::string &in, THULAC_result& result) {
 }
 
 
+std::string THULAC::toString(const THULAC_result& result) {
+    std::string output = "";
+    for(auto i : result) {
+        if(i.first == "\n") continue;
+        if(seg_only) output += i.first + i.second + " ";
+        else output += i.first + char(separator) + i.second + " ";
+    }
+    output.erase(output.size() - 1, 1);
+    return output;
+}
+
 
 #endif
diff --git a/test/test_case.cpp b/test/test_case.cpp
@@ -0,0 +1,56 @@
+#include <iostream>
+#include <string>
+#include "thulac.h"
+#include "thulac_test.h"
+
+
+int main() {
+    THULAC lac;
+    std::string case_name;
+    {
+
+        case_name = "检查单分词模型模型";
+        lac.init("/program/python/python-git/thulac/models",NULL,1);
+        THULAC_TEST test = THULAC_TEST(&lac, case_name);
+        test.testEqual("我爱北京天安门", "我 爱 北京 天安门");
+        test.testEqual("小明喜欢玩炉石传说", "小明 喜欢 玩 炉石 传说");
+    }
+
+    {
+        case_name = "检查带词性标注的模型";
+        lac.init("/program/python/python-git/thulac/models",NULL,0);
+        THULAC_TEST test = THULAC_TEST(&lac, case_name);
+        test.testEqual("我爱北京天安门", "我_r 爱_v 北京_ns 天安门_ns");
+        test.testEqual("小明喜欢玩炉石传说", "小明_np 喜欢_v 玩_v 炉石_n 传说_n");
+    }
+
+    {
+        case_name = "检查deli分隔符参数";
+        lac.init("/program/python/python-git/thulac/models",NULL,0, 0, 0,'#');
+        THULAC_TEST test = THULAC_TEST(&lac, case_name);
+        test.testEqual("我爱北京天安门", "我#r 爱#v 北京#ns 天安门#ns");
+        test.testEqual("小明喜欢玩炉石传说", "小明#np 喜欢#v 玩#v 炉石#n 传说#n");
+    }
+
+    {
+        case_name = "检查T2S分隔符参数";
+        lac.init("/program/python/python-git/thulac/models",NULL,1, 1);
+        THULAC_TEST test = THULAC_TEST(&lac, case_name);
+        test.testEqual("我愛北京天安門", "我 爱 北京 天安门");
+        test.testEqual("小明喜歡玩爐石傳說", "小明 喜欢 玩 炉石 传说");
+    }
+
+    {
+        case_name = "检查ufilter参数";
+        lac.init("/program/python/python-git/thulac/models",NULL,1, 0, 1);
+        THULAC_TEST test = THULAC_TEST(&lac, case_name);
+        test.testEqual("我可以爱北京天安门", "我 爱 北京 天安门");
+    }
+
+
+    THULAC_TEST::reportAll();
+    return 0;
+}
+
+
+
diff --git a/test/thulac_test.h b/test/thulac_test.h
@@ -0,0 +1,82 @@
+#include <string>
+#include <vector>
+#include <iostream>
+#include <utility>
+#include "../include/thulac.h"
+class THULAC_TEST {
+public:
+    void report(const std::string &s);
+    void testEqual(const std::string &raw, const std::string &standard);
+    THULAC_TEST(THULAC *lac);
+    THULAC_TEST(THULAC *lac, std::string name);
+    static void reportAll();
+private:
+    std::vector<std::string> errorMsg;
+    THULAC *lac;
+    std::string caseName;
+    bool equals(const std::string &result, const std::string &standard);
+    void report_error(const std::string &e);
+    static std::vector<std::pair<std::string, std::string> > allErrorMsg;
+
+};
+
+THULAC_TEST::THULAC_TEST(THULAC *lac) {
+    this->lac = lac;
+
+
+}
+
+THULAC_TEST::THULAC_TEST(THULAC *lac, std::string caseName) {
+    this->lac = lac;
+    this->caseName = caseName;
+}
+
+bool THULAC_TEST::equals(const std::string &result, const std::string &standard) {
+    return result == standard;
+}
+
+void THULAC_TEST::report_error(const std::string &e) {
+    errorMsg.push_back(e);
+    allErrorMsg.push_back(std::make_pair(caseName, e));
+}
+
+void THULAC_TEST::testEqual(const std::string &raw, const std::string &standard) {
+    THULAC_result result;
+    lac->cut(raw, result);
+    std::string s_result = lac->toString(result);
+    if(equals(s_result, standard)) {
+        std::cout << ".";
+        return;
+    }
+    std::cout << "E";
+    std::string error = "不匹配：" + s_result + "  与  " + standard;
+    report_error(error);
+    return;
+}
+
+void THULAC_TEST::report(const std::string &s) {
+    std::cout << std::endl << s << ":" << std::endl;
+    if(errorMsg.size() == 0) {
+        std::cout << "恭喜！所有case通过测试" << std::endl;
+    }
+    else{
+        for(auto i : errorMsg) {
+
+            std::cout << i << std::endl;
+        }
+
+    }
+    return;
+}
+
+void THULAC_TEST::reportAll() {
+    std::cout << "\n测试结果：\n";
+    if(allErrorMsg.size() == 0) std::cout << "恭喜！所有case通过测试" << std::endl;
+    else {
+        for (auto i : allErrorMsg) {
+            std::cout << i.first << "case中：\n" << i.second << std::endl;
+        }
+    }
+}
+
+std::vector<std::pair<std::string, std::string> > THULAC_TEST::allErrorMsg = std::vector<std::pair<std::string, std::string> >();