googleTest 源码主线框架性分析——TDD 01

TDD，测试驱动开发，英文全称Test-Driven Development，简称TDD，是一种不同于传统软件开发流程的新型的开发方法。它要求在编写某个功能的代码之前先编写测试代码，然后只编写使测试通过的功能代码，通过测试来推动整个开发的进行。这有助于编写简洁可用和高质量的代码，并加速开发过程。

简言之TDD是通过设计 Test 来完成软件设计的一种高效可行的软件开发模式。

为何更丰富地达成测试的目的，googletes是绕不过去的，本文备忘主要关注 googletest 主体的分析过程和结论，即，googleTest框架中是如何通过相关的测试宏的，实现测试的目的。

TEST TEST_F TEST_P 等等

1，googleTest 环境与简单示例

1.1 下载 googletest 并编译

下载：

$ git clone https://github.com/google/googletest.git
$ git checkout release-1.10.0

编译：

$ mkdir build
$ cd build/
$ export CXXFLAGS="-Wno-error=maybe-uninitialized"
$ cmake ..
$ make -j
$ ls lib/

默认为 release，若debug版本则须：

$ cmake .. -DCMAKE_BUILD_TYPE=Debug

成果：

1.2 示例1 验证函数 add

源码

#include <iostream>
#include "gtest/gtest.h"int add_int_int(int a, int b){return a+b;
}TEST(SumFuncTest, twoNumbers){EXPECT_EQ(add_int_int(3,4),7);EXPECT_EQ(27, add_int_int(9, 18));
}GTEST_API_ int main(int argc, char** argv) {printf("Running main() from %s\n", __FILE__);testing::InitGoogleTest(&argc, argv);return RUN_ALL_TESTS();
}

运行：

1.3 示例 2

#include <gtest/gtest.h>int Foo(int a, int b)
{if (a == 0 || b == 0){throw "don't do that";}int c = a % b;if (c == 0)return b;return Foo(b, c);
}TEST(FooTest, HandleNoneZeroInput)
{EXPECT_EQ(2, Foo(4, 10));EXPECT_EQ(6, Foo(30, 18));
}

g++ foo.cpp -I ../../googletest/googletest/include -L ../../googletest/build_dbg/lib -lgtest -lgtest_main

编译运行：

1.4 示例3

源码：

#include <iostream>
#include "gtest/gtest.h"// add_util.ccfloat add_from_left(float a, float b, float c, float d,	float e)
{float sum = 0.0;sum += c;sum += a;sum += b;//sum += c;sum += d;sum += e;
/*sum += a;sum += b;sum += c;sum += d;sum += e;
*/printf("add_from_left: sum = %f\n", sum);return sum;
}float add_from_right(float a, float b, float c,	float d, float e)
{float sum = 0.0;sum += e;sum += d;sum += c;sum += b;sum += a;printf("add_from_right: sum = %f\n", sum);return sum;
}int sum(int a, int b){return a+b;
}TEST(AddFuncTest, floatVSfloat) {printf("AddFuncTest: float  sum = %f\n", 1.238f + 3.7f + 0.000353265f + 7898.3f + 12.23209f);printf("AddFuncTest: double sum = %f\n", 12.23209 + 7898.3 + 0.000353265 + 3.7 + 1.238);EXPECT_EQ(1.238f + 3.7f + 0.000353265f + 7898.3f + 12.23209f, add_from_left(1.238, 3.7, 0.000353265, 7898.3, 12.23209));EXPECT_EQ(1.238f + 3.7f + 0.000353265f + 7898.3f + 12.23209f, add_from_right(1.238, 3.7, 0.000353265, 7898.3, 12.23209));
//
}TEST(AddFuncTest, doubleVSfloat) {printf("AddFuncTest: float  sum = %f\n", 1.238f + 3.7f + 0.000353265f + 7898.3f + 12.23209f);printf("AddFuncTest: double sum = %f\n", 12.23209 + 7898.3 + 0.000353265 + 3.7 + 1.238);EXPECT_EQ(1.238f + 3.7f + 0.000353265f + 7898.3f + 12.23209f, add_from_left(1.238, 3.7, 0.000353265, 7898.3, 12.23209));EXPECT_EQ(1.238 + 3.7 + 0.000353265 + 7898.3 + 12.23209, add_from_right(1.238, 3.7, 0.000353265, 7898.3, 12.23209));
//
}TEST(SumFuncTest, twoNumbers){EXPECT_EQ(sum(3,4),7);EXPECT_EQ(27, sum(9, 18));
}GTEST_API_ int main(int argc, char** argv) {printf("Running main() from %s\n", __FILE__);testing::InitGoogleTest(&argc, argv);return RUN_ALL_TESTS();
}

Makefile

EXE := hello_gtest_ex hello_gtest_add_int_int
all: $(EXE)%: %.cppg++ -O0 -fno-toplevel-reorder $< -o $@ $(INC) $(LD_FLAGS)INC := -I../googletest/googletest/include/
LD_FLAGS := -L../googletest/build/lib/ -lgtest -lgtest_main.PHONY: clean
clean:-rm -rf $(EXE)

2，示例与源码分析

使用最简单的测试示例，聚焦googletest本身的代码逻辑

观察点，main 函数如何调用到 TEST(...){...} 这种结构中的代码

两种方式互相印证：

方式1，通过编译器的预编译指令 g++ -E ... 生成展开代码;

方式2，通过跟踪源代码，来份些TEST等的展开结果

2.1 TEST

示例代码如上：

simple_gtest.cpp


#include "gtest/gtest.h"int add_int_int(int a, int b){return a+b;
}TEST(SumFuncTest, twoNumbers){EXPECT_EQ(add_int_int(3,4),7);
}

方式1:

g++ -E simple_gtest.cpp -o simple_gtest.i

展开后，simple_gtest.i文件有8W多行，但是其中对我们理解有意义的也就最尾巴上的几行：


int add_int_int(int a, int b){return a+b;
}static_assert(sizeof("SumFuncTest") > 1, "test_suite_name must not be empty"); 
static_assert(sizeof("twoNumbers") > 1, "test_name must not be empty"); class SumFuncTest_twoNumbers_Test : public ::testing::Test { 
public: SumFuncTest_twoNumbers_Test() {} 
private:virtual void TestBody(); static ::testing::TestInfo* const test_info_ __attribute__ ((unused)); SumFuncTest_twoNumbers_Test(SumFuncTest_twoNumbers_Test const &) = delete; void operator=(SumFuncTest_twoNumbers_Test const &) = delete; 
}; ::testing::TestInfo* const SumFuncTest_twoNumbers_Test::test_info_ = ::testing::internal::MakeAndRegisterTestInfo( "SumFuncTest", "twoNumbers", nullptr, nullptr, ::testing::internal::CodeLocation("simple_gtest.cpp", 8), (::testing::internal::GetTestTypeId()), ::testing::internal::SuiteApiResolver< ::testing::Test>::GetSetUpCaseOrSuite("simple_gtest.cpp", 8), ::testing::internal::SuiteApiResolver< ::testing::Test>::GetTearDownCaseOrSuite("simple_gtest.cpp", 8), new ::testing::internal::TestFactoryImpl<SumFuncTest_twoNumbers_Test>); void SumFuncTest_twoNumbers_Test::TestBody()
{switch (0) case 0: default: if (const ::testing::AssertionResult gtest_ar = (::testing::internal::EqHelper::Compare("add_int_int(3,4)", "7", add_int_int(3,4), 7))) ; else ::testing::internal::AssertHelper(::testing::TestPartResult::kNonFatalFailure, "simple_gtest.cpp", 9, gtest_ar.failure_message()) = ::testing::Message();
}

分析这段代码会发现，

TEST被展开成为了一个 class SumFuncTest_twoNumbers_Test

它有一个成员函数 TestBody(){....}

观察上述代码中最后一个函数体：void SumFuncTest_twoNumbers_Test::TestBody()

其中出现了被测试的函数等。

这说明，这个函数体中的代码才是是被测试内容，而其外围都是框架。

框架部分只需要把这中类的一个实例添加到某个链表中，然后依次迭代执行每个类的 TestBody成员函数，既可以完成测试任务。

本例中的 class 如下：

通过方法2.来验证一下展开的结果：

第一部分，class 宏

关联TEST宏，我们可以找到如下内容：


#define TEST(test_suite_name, test_name) GTEST_TEST(test_suite_name, test_name)#define GTEST_TEST(test_suite_name, test_name)             \GTEST_TEST_(test_suite_name, test_name, ::testing::Test, \::testing::internal::GetTestTypeId())// Expands to the name of the class that implements the given test.
#define GTEST_TEST_CLASS_NAME_(test_suite_name, test_name) \test_suite_name##_##test_name##_Test// Helper macro for defining tests.
#define GTEST_TEST_(test_suite_name, test_name, parent_class, parent_id)       \static_assert(sizeof(GTEST_STRINGIFY_(test_suite_name)) > 1,                 \"test_suite_name must not be empty");                          \static_assert(sizeof(GTEST_STRINGIFY_(test_name)) > 1,                       \"test_name must not be empty");                                \class GTEST_TEST_CLASS_NAME_(test_suite_name, test_name)                     \: public parent_class {                                                  \public:                                                                     \GTEST_TEST_CLASS_NAME_(test_suite_name, test_name)() = default;            \~GTEST_TEST_CLASS_NAME_(test_suite_name, test_name)() override = default;  \GTEST_TEST_CLASS_NAME_(test_suite_name, test_name)                         \(const GTEST_TEST_CLASS_NAME_(test_suite_name, test_name) &) = delete;     \GTEST_TEST_CLASS_NAME_(test_suite_name, test_name) & operator=(            \const GTEST_TEST_CLASS_NAME_(test_suite_name,                          \test_name) &) = delete; /* NOLINT */      \GTEST_TEST_CLASS_NAME_(test_suite_name, test_name)                         \(GTEST_TEST_CLASS_NAME_(test_suite_name, test_name) &&) noexcept = delete; \GTEST_TEST_CLASS_NAME_(test_suite_name, test_name) & operator=(            \GTEST_TEST_CLASS_NAME_(test_suite_name,                                \test_name) &&) noexcept = delete; /* NOLINT */  \\private:                                                                    \void TestBody() override;                                                  \static ::testing::TestInfo* const test_info_ GTEST_ATTRIBUTE_UNUSED_;      \};                                                                           \\::testing::TestInfo* const GTEST_TEST_CLASS_NAME_(test_suite_name,           \test_name)::test_info_ =   \::testing::internal::MakeAndRegisterTestInfo(                            \#test_suite_name, #test_name, nullptr, nullptr,                      \::testing::internal::CodeLocation(__FILE__, __LINE__), (parent_id),  \::testing::internal::SuiteApiResolver<                               \parent_class>::GetSetUpCaseOrSuite(__FILE__, __LINE__),          \::testing::internal::SuiteApiResolver<                               \parent_class>::GetTearDownCaseOrSuite(__FILE__, __LINE__),       \new ::testing::internal::TestFactoryImpl<GTEST_TEST_CLASS_NAME_(     \test_suite_name, test_name)>);                                   \void GTEST_TEST_CLASS_NAME_(test_suite_name, test_name)::TestBody()

其中的如下两行：

class GTEST_TEST_CLASS_NAME_(test_suite_name, test_name) \

: public parent_class { \

TEST 的宏充分展开后，会根据TEST(X,Y) 括号中的X、Y字串定义一个完整的类，并且包含成员函数:
TestBody()

但是展开的内容中，没有这个函数的函数体。

这个函数体正好就是TEST(X,Y){Z} 中，{Z}的这个部分，即，

TestBody(){Z}

只需要在整个测试系统中，讲上面展开生成的class的一个实例，insert进一个链表中，并依次迭代执行链表的每一对象的成员函数 TestBody(){Z}，即可达到测试目的。

第二部分，函数体中的宏

关于 EXPECT_EQ，我们会发现如下定义：

#define EXPECT_EQ(val1, val2) \EXPECT_PRED_FORMAT2(::testing::internal::EqHelper::Compare, val1, val2)

而其中的 EXPECT_PRED_FORMAT2 又被展开为如下：

// Binary predicate assertion macros.
#define EXPECT_PRED_FORMAT2(pred_format, v1, v2) \GTEST_PRED_FORMAT2_(pred_format, v1, v2, GTEST_NONFATAL_FAILURE_)

又 GTEST_PRED_FORMAT2_ 被定义为：

#define GTEST_PRED_FORMAT2_(pred_format, v1, v2, on_failure) \GTEST_ASSERT_(pred_format(#v1, #v2, v1, v2), on_failure)

而且其中的 GTEST_ASSERT_ 被展开为：

#define GTEST_ASSERT_(expression, on_failure)                   \GTEST_AMBIGUOUS_ELSE_BLOCKER_                                 \if (const ::testing::AssertionResult gtest_ar = (expression)) \;                                                           \else                                                          \on_failure(gtest_ar.failure_message())

其中 GTEST_AMBIGUOUS_ELSE_BLOCKER_ 展开为：

#define GTEST_AMBIGUOUS_ELSE_BLOCKER_ \switch (0)                          \case 0:                             \default:  // NOLINT

于是得到函数体为：

总之，只需要调用这个 TestBody() 函数，即可完成测试任务。

2.2 TEST_F

2.2.1 小示例编译与运行

保持关于 TEST 宏分析的记忆，我们以一个简单的示例来分析 TEST_F 宏，

#include <gtest/gtest.h>class SampleTestWithFixture : public ::testing::Test {
protected:void SetUp() override {a_ = 1;b_ = 2;}int a_;int b_;
};TEST_F(SampleTestWithFixture, Case2) {a_ = 3;EXPECT_EQ(a_ + b_, 5);
}

Makefile:

EXE := hello_gtest_f
all: $(EXE)%: %.cppg++ -g -fno-toplevel-reorder $< -o $@ $(INC) $(LD_FLAGS)# g++ -E hello_gtest_f.cpp  -I ../googletest/googletest/include/ -o hello_gtest_f.i
#g++ -g -fno-toplevel-reorder $< -o $@ $(INC) $(LD_FLAGS)INC := -I../../googletest/googletest/include/
LD_FLAGS := -L../../googletest/build_dbg/lib/ -lgtest -lgtest_main.PHONY: clean
clean:-rm -rf $(EXE)

编译，确保能够正确运行：

$ make

$ ./hello_gtest_f

2.2.2 TEST_F 宏展开分析

$ g++ -E hello_gtest_f.cpp  -I ../googletest/googletest/include/ -o hello_gtest_f.i

生成的预处理后的文件 hello_gtest_f.i 主要内容还是在文件的尾巴上，摘录调整格式如下：

class SampleTestWithFixture : public ::testing::Test {
protected:void SetUp() override {a_ = 1;b_ = 2;}int a_;int b_;
};static_assert(sizeof("SampleTestWithFixture") > 1, "test_suite_name must not be empty");
static_assert(sizeof("Case2") > 1, "test_name must not be empty");class SampleTestWithFixture_Case2_Test : public SampleTestWithFixture{
public:SampleTestWithFixture_Case2_Test() = default;~SampleTestWithFixture_Case2_Test() override = default;SampleTestWithFixture_Case2_Test(const SampleTestWithFixture_Case2_Test&) = delete;SampleTestWithFixture_Case2_Test& operator=( const SampleTestWithFixture_Case2_Test&) = delete;SampleTestWithFixture_Case2_Test(SampleTestWithFixture_Case2_Test &&) noexcept = delete;SampleTestWithFixture_Case2_Test& operator=(SampleTestWithFixture_Case2_Test &&) noexcept = delete;private: void TestBody() override;static ::testing::TestInfo* const test_info_ __attribute__((unused));
};::testing::TestInfo* const SampleTestWithFixture_Case2_Test::test_info_=::testing::internal::MakeAndRegisterTestInfo("SampleTestWithFixture","Case2",nullptr,nullptr,::testing::internal::CodeLocation("hello_gtest_f.cpp", 27),(::testing::internal::GetTypeId<SampleTestWithFixture>()),::testing::internal::SuiteApiResolver< SampleTestWithFixture>::GetSetUpCaseOrSuite("hello_gtest_f.cpp", 27),::testing::internal::SuiteApiResolver< SampleTestWithFixture>::GetTearDownCaseOrSuite("hello_gtest_f.cpp", 27),new ::testing::internal::TestFactoryImpl<SampleTestWithFixture_Case2_Test>);void SampleTestWithFixture_Case2_Test::TestBody()
{a_ = 3;switch (0)case 0:default:if (const ::testing::AssertionResult gtest_ar = (::testing::internal::EqHelper::Compare( "a_ + b_" ,  "5" ,  a_ + b_ ,  5 )));else::testing::internal::AssertHelper(::testing::TestPartResult::kNonFatalFailure, "hello_gtest_f.cpp", 29, gtest_ar.failure_message())= ::testing::Message();
}

跟 TEST 宏的展开类似，组合 TEST_F(X,Y) 的两个参数，构成一个新的类

class X_Y_Test :public SampleTestWithFixture{

...

... TestBody()

}

宏展开的新类中也有一个成员函数 TestBody()；

其中 SampleTestWithFixture 是自己定义的类，会被 X_Y_Test 类共有继承走。

而 TEST_F(...){body} 的类似函数体的部分 {body}，也同样被安排成为了 TestBody函数的函数体。

接下来，gtest框架会通过成员 X_Y_Test::test_info_ 的静态赋值过程，将本测试用例挂进系统的代运行链表，届时依次迭代调用 X_Y_Test::TestBody(); 实现测试感兴趣代码的目的。

2.2.3 总结 TEST_F

TEST_F的意图：

TEST_F的目的是为了把关系密切的测试问题汇总到一个class中来进行测试，可以共用同一个类的对象的上下文成员数据。

TEST_F 中，成员函数的执行顺序：

那么，成员函数 Setup( ) 在什么时候执行呢？

先说答案：

1 X_Y_Test() 构造函数；//c++ 语法

2 Setup(); //数据预备，资源申请

3 TestBody(); //测试部分

4 TearDown(); //资源释放

5 X_Y_Test() 析构函数；//c++ 语法

改造刚才的示例：

#include <gtest/gtest.h>class SampleTestWithFixture : public ::testing::Test {
public:
SampleTestWithFixture(){std::cout<<"construct_STWF"<<std::endl;}
~SampleTestWithFixture(){std::cout<<"destruct_STWF"<<std::endl;}
protected:void SetUp() override {std::cout <<"Hello setupupup()000"<<std::endl;a_ = 1;b_ = 2;std::cout <<"Hello setupupup()111"<<std::endl;}void TearDown() override {std::cout <<"Hello teardownnn()000"<<std::endl;a_ = 4;b_ = 5;std::cout <<"Hello teardownnn()111"<<std::endl;}int a_;int b_;
};TEST_F(SampleTestWithFixture, Case2) {std::cout <<"test_f Casess222"<<std::endl;a_ = 3;EXPECT_EQ(a_ + b_, 5);
}

编译运行：

TEST_F函数体中的部分的一些宏，跟TEST中的一样，展开成为一些比较语句。

2.3 TEST_P

2.3.1 可运行示例

#include "gtest/gtest.h"namespace TTT
{
namespace testing
{
int g_env_switch = 0;class BasicTestFixture : public ::testing::TestWithParam<int>
{
public:BasicTestFixture() {}void SetUp(){g_env_switch = GetParam();std::cout<<"BTF_SetUp() ges="<<g_env_switch<<std::endl;}void TearDown(){}
};#define OK 0
#define FAIL 1int envCheckFunc(void)
{if(g_env_switch > 0) {return OK;}else {return FAIL;}
}TEST_P(BasicTestFixture, BasicTest)
{ASSERT_EQ(envCheckFunc(), OK);
}INSTANTIATE_TEST_SUITE_P(configSwitch, BasicTestFixture, ::testing::Values(1, 0));
//INSTANTIATE_TEST_SUITE_P(failSwitch, BasicTestFixture, ::testing::Values(2, 3));}
}

Makefile：

EXE := hello_gtest_pall: $(EXE)%: %.cppg++ -g -fno-toplevel-reorder $< -o $@ $(INC) $(LD_FLAGS)# g++ -E hello_gtest_p.cpp  -I ../googletest/googletest/include/ -o hello_gtest_p.i
#g++ -g -fno-toplevel-reorder $< -o $@ $(INC) $(LD_FLAGS)INC := -I../../googletest/googletest/include/
LD_FLAGS := -L../../googletest/build_dbg/lib/ -lgtest -lgtest_main.PHONY: clean
clean:-rm -rf $(EXE)

编译执行：

因为故意藏了一个逻辑错误，所以第二个参数时，会测试失败：

2.3.2 无实例化宏的预编译

注释掉程序中的所有 INSTANTIATE_TEST_SUITE_P 的行，不厌其烦地再贴一次：

#include "gtest/gtest.h"namespace TTT
{
namespace testing
{
int g_env_switch = 0;class BasicTestFixture : public ::testing::TestWithParam<int>
{
public:BasicTestFixture() {}void SetUp(){g_env_switch = GetParam();std::cout<<"BTF_SetUp() ges="<<g_env_switch<<std::endl;}void TearDown(){}
};#define OK 0
#define FAIL 1int envCheckFunc(void)
{if(g_env_switch > 0) {return OK;}else {return FAIL;}
}TEST_P(BasicTestFixture, BasicTest)
{ASSERT_EQ(envCheckFunc(), OK);
}//INSTANTIATE_TEST_SUITE_P(configSwitch, BasicTestFixture, ::testing::Values(1, 0));
//INSTANTIATE_TEST_SUITE_P(failSwitch, BasicTestFixture, ::testing::Values(2, 3));}
}

g++ -E hello_gtest_p.cpp -I ../googletest/googletest/include/ -o hello_gtest_p.i

这时候预编译后生成的代码如下：

namespace TTT
{
namespace testing
{
int g_env_switch = 0;class BasicTestFixture : public ::testing::TestWithParam<int>
{
public:BasicTestFixture() {}void SetUp(){g_env_switch = GetParam();std::cout<<"BTF_SetUp() ges="<<g_env_switch<<std::endl;}void TearDown(){}
};int envCheckFunc(void)
{if(g_env_switch > 0) {return 0;}else {return 1;}
}class BasicTestFixture_BasicTest_Test : public BasicTestFixture, private ::testing::internal::GTestNonCopyable {
public:BasicTestFixture_BasicTest_Test() {}void TestBody() override;
private:static int AddToRegistry(){::testing::UnitTest::GetInstance()->parameterized_test_registry().GetTestSuitePatternHolder<BasicTestFixture>( "BasicTestFixture", ::testing::internal::CodeLocation("hello_gtest_p.cpp", 35))->AddTestPattern( "BasicTestFixture","BasicTest",new ::testing::internal::TestMetaFactory<BasicTestFixture_BasicTest_Test>(),::testing::internal::CodeLocation("hello_gtest_p.cpp", 35));return 0;}static int gtest_registering_dummy_ __attribute__((unused));
};int BasicTestFixture_BasicTest_Test::gtest_registering_dummy_ = BasicTestFixture_BasicTest_Test::AddToRegistry();void BasicTestFixture_BasicTest_Test::TestBody()
{switch (0)case 0:default:if (const ::testing::AssertionResult gtest_ar = (::testing::internal::EqHelper::Compare("envCheckFunc()", "0", envCheckFunc(), 0)));elsereturn ::testing::internal::AssertHelper(::testing::TestPartResult::kFatalFailure, "hello_gtest_p.cpp", 37, gtest_ar.failure_message()) = ::testing::Message();
}

没有实例化宏的代码，展开到被测试类的 TestBody()函数后就停止了，如上代码中最后一个函数的定义。

TEST_P(X,Y){...}

展开为

X_Y_Test，且包含一个成员函数 TestBody(){...}，TEST_P的函数体也被TestBody() 接了过来。

2.3.3 有实例化宏的预编译

多了这两句：

INSTANTIATE_TEST_SUITE_P(configSwitch, BasicTestFixture, ::testing::Values(1, 0));

INSTANTIATE_TEST_SUITE_P(failSwitch, BasicTestFixture, ::testing::Values(2, 3));


namespace TTT
{
namespace testing
{
int g_env_switch = 0;class BasicTestFixture : public ::testing::TestWithParam<int>
{
public:BasicTestFixture() {}void SetUp(){g_env_switch = GetParam();std::cout<<"BTF_SetUp() ges="<<g_env_switch<<std::endl;}void TearDown(){}
};int envCheckFunc(void)
{if(g_env_switch > 0) {return 0;}else {return 1;}
}class BasicTestFixture_BasicTest_Test : public BasicTestFixture, private ::testing::internal::GTestNonCopyable {
public:BasicTestFixture_BasicTest_Test() {}void TestBody() override;private:static int AddToRegistry(){::testing::UnitTest::GetInstance()->parameterized_test_registry().GetTestSuitePatternHolder<BasicTestFixture>( "BasicTestFixture", ::testing::internal::CodeLocation("hello_gtest_p.cpp", 35))->AddTestPattern( "BasicTestFixture","BasicTest",new ::testing::internal::TestMetaFactory<BasicTestFixture_BasicTest_Test>(),::testing::internal::CodeLocation("hello_gtest_p.cpp",35));return 0;}static int gtest_registering_dummy_ __attribute__((unused));
};int BasicTestFixture_BasicTest_Test::gtest_registering_dummy_ = BasicTestFixture_BasicTest_Test::AddToRegistry();
void BasicTestFixture_BasicTest_Test::TestBody()
{switch (0)case 0:default:if (const ::testing::AssertionResult gtest_ar = (::testing::internal::EqHelper::Compare( "envCheckFunc()", "0", envCheckFunc(), 0)));elsereturn ::testing::internal::AssertHelper(::testing::TestPartResult::kFatalFailure, "hello_gtest_p.cpp", 37, gtest_ar.failure_message()) = ::testing::Message();
}
//TEST_P(X, Y, ...){}
// 跟 TEST_F的展开比较接近，同样是将 TEST_P(){...} 的{...} 部分，作为TestBody的函数体；
// 但，一个重要不懂的地方在于，TEST_P 展开为被测试模式或模版，真正注册近被执行队列，是INSTANTIATE_TEST_SUITE_P的工作。//INSTANTIATE_TEST_SUITE_P(X, Y, ...)的展开位三个函数
//参数容器函数：gtest_XY_EvaluGenerator_    测试参数值；
//被测试类函数：gtest_XY_EvalGenerateName_  测试参数类型；
//测试类压栈函数：gtest_XY_dummy_           将测试类实例化后 push进带测试队列中，通过调用 AddTestSuiteInstantiation//展开，INSTANTIATE_TEST_SUITE_P(configSwitch, BasicTestFixture, ::testing::Values(1, 0));static ::testing::internal::ParamGenerator<BasicTestFixture::ParamType> gtest_configSwitchBasicTestFixture_EvalGenerator_()
{return::testing::Values(1, 0);
}static ::std::string gtest_configSwitchBasicTestFixture_EvalGenerateName_( const ::testing::TestParamInfo<BasicTestFixture::ParamType>& info)
{if (::testing::internal::AlwaysFalse()) {::testing::internal::TestNotEmpty(::testing::internal::DefaultParamName<BasicTestFixture::ParamType>);auto t = std::make_tuple(::testing::Values(1, 0));static_assert(std::tuple_size<decltype(t)>::value <= 2, "Too Many Args!");}return ((::testing::internal::DefaultParamName<BasicTestFixture::ParamType>))(info);
}static int gtest_configSwitchBasicTestFixture_dummy_ __attribute__((unused))= ::testing::UnitTest::GetInstance()->parameterized_test_registry().GetTestSuitePatternHolder<BasicTestFixture>("BasicTestFixture", ::testing::internal::CodeLocation("hello_gtest_p.cpp", 40))->AddTestSuiteInstantiation("configSwitch",&gtest_configSwitchBasicTestFixture_EvalGenerator_,&gtest_configSwitchBasicTestFixture_EvalGenerateName_,"hello_gtest_p.cpp",40);//展开，INSTANTIATE_TEST_SUITE_P(failSwitch, BasicTestFixture, ::testing::Values(2, 3));static ::testing::internal::ParamGenerator<BasicTestFixture::ParamType> gtest_failSwitchBasicTestFixture_EvalGenerator_()
{return::testing::Values(2, 3);
}static ::std::string gtest_failSwitchBasicTestFixture_EvalGenerateName_( const ::testing::TestParamInfo<BasicTestFixture::ParamType>& info)
{if (::testing::internal::AlwaysFalse()) {::testing::internal::TestNotEmpty(::testing::internal::DefaultParamName<BasicTestFixture::ParamType>);auto t = std::make_tuple(::testing::Values(2, 3));static_assert(std::tuple_size<decltype(t)>::value <= 2, "Too Many Args!");}return ((::testing::internal::DefaultParamName<BasicTestFixture::ParamType>))(info);
}static int gtest_failSwitchBasicTestFixture_dummy_ __attribute__((unused))= ::testing::UnitTest::GetInstance()->parameterized_test_registry().GetTestSuitePatternHolder<BasicTestFixture>("BasicTestFixture", ::testing::internal::CodeLocation("hello_gtest_p.cpp", 41))->AddTestSuiteInstantiation("failSwitch",&gtest_failSwitchBasicTestFixture_EvalGenerator_,&gtest_failSwitchBasicTestFixture_EvalGenerateName_,"hello_gtest_p.cpp",41);}
}

2.3.4 分析 TEST_P 和

TEST_P(X, Y, ...){...}的展开跟 TEST_F的展开比较接近，继承用户自定义类后，定义一个新类，含有成员函数 TestBody，同样是将 TEST_P(){...} 的{...} 部分，作为TestBody的函数体；

但，一个重要不同的地方在于，TEST_P 展开为被测试模式或模版，真正注册近被执行队列，是INSTANTIATE_TEST_SUITE_P的工作。

INSTANTIATE_TEST_SUITE_P(X, Y, ...)的展开位三个函数

1，参数容器函数：gtest_XY_EvaluGenerator_ 测试参数值；

2，被测试类函数：gtest_XY_EvalGenerateName_ 测试参数类型；

3，测试类压栈函数：gtest_XY_dummy_ 将测试类实例化后 push进带测试队列中，通过调用 AddTestSuiteInstantiation，压栈入队，等待main函数依次迭代调用各个类实例的 TestBody()成员函数，完成测试任务。

而其中的SetUp(), TearDown()等成员函数的调用，跟TEST_F的调用时机相同。

2.3.5 一个更一般的 TEST_P 示例的简单说明

这个示例展示了测试类 class CombineTest,如何实现对被测试class Bis 的成员函数 Even 和 Suc 的功能进行组合测试。

multi_parameters_test_p.cpp

#include <gtest/gtest.h>class Bis
{
public:bool Even(int n){if (n % 2 == 0){return true;}else{return false;}};bool Suc(bool bSuc){return bSuc;}
};class CombineTest :public ::testing::TestWithParam< ::testing::tuple<bool, int> >
{
protected:bool checkData() {int n = ::testing::get<1>(GetParam());      // <1> 1st parameter in the tuplebool suc = ::testing::get<0>(GetParam());   // <0> 0th parameter in the tuplereturn bis.Suc(suc) && bis.Even(n) || !bis.Suc(suc) && !bis.Even(n);}private:Bis bis;
};TEST_P(CombineTest, Test) {EXPECT_TRUE(checkData());
}//INSTANTIATE_TEST_CASE_P(TestBisValuesCombine, CombineTest, ::testing::Combine(::testing::Bool(), ::testing::Values(0, 1, 2, 3)));
INSTANTIATE_TEST_CASE_P(TestBisValuesCombine, CombineTest, ::testing::Combine(::testing::Values(false, true), ::testing::Values(0, 1, 2, 3)));
//::testing::Bool() 是指遍历所有的 boolean 值
//::testing::Combine(A,B,C,...) 是由A x B x C x ... 构成的一个笛卡尔积

除了注释中的一点说明，想强化备忘一个点，关于 EXPECT_TRUE(checkData()); 中可以直接调用 checkData()的原因：

首先，

由于函数 checkData() 本身是 protected修饰的；

其次，

由于CombineTest类 public 地继承了模版类：::testing::TestWithParam<>，而且类名出现在了 TEST_P(, )的第一个参数的位置，这导致CombineTest被展的测试类 X_Y_Test 进行了public继承。

所以，X_Y_Test的成员函数包括，

<1> TEST_P 宏展出的成员函数 TestBody()

<2> 继承自 TEST_P的第一个参数的类 CombineTest 的所有成员函数。

所以，TestBody() 的函数体实现中，可以调用CombineTest的成员函数 checkData()。

运行结果：

3. 总结

掌握 googletest的使用，
首先需要理解，TEST???(X, Y ){...}结构中，{...}会被展开为一个测试类?X_Y_Test 的成员函数 TestBody(){...} 的函数体:

?X_Y_Test::TestBody(){...}

其次，掌握Setup,TearDown的调用时机；
然后，对各种判别宏有一定掌握，比如 EXPECT_EQ；
最后，掌握 TEST_P 的参数的各种形式的使用方式