如何使用 gmock 在具体类中模拟非虚拟方法? [英] How to mock Non-virtual Methods in concrete classes using gmock?

问题描述

我以某种方式从 donsoft.io 的示例扩展了 gmock 测试用例，并按如下方式制作:

coinflipper/├── 建设├── 工作区├── coinflipper.cc├── coinflipper.h├── rng.cc└── rng.h

好吧，我将 Rng 类作为 CoinFlipper 构造函数的参数，而是在 CoinFlipper::flipCoin() 方法.

我想知道在这种情况下如何从 Rng 模拟 generate() ?

coinflipper.cc

#include "coinflipper.h";#include CoinFlipper::CoinFlipper() {}CoinFlipper::Result CoinFlipper::flipCoin() const {Rng d_rng;const double val = d_rng.generate(0.0, 1.0);返回 (val <0.5) ?头部:尾巴；}int main(int argc, char** argv) {CoinFlipper 参见；CoinFlipper::Result res = cf.flipCoin();如果(res == 0){std::cout <<头"<<std::endl;} 别的 {std::cout <<尾巴"<<std::endl;}返回0；}

coinflipper.h

#include "rng.h";类 CoinFlipper {民众:枚举结果 { HEADS = 0, TAILS = 1 };显式 CoinFlipper();结果 flipCoin() const;};

rng.h

#ifndef RNG_H#define RNG_H类 Rng {民众:Rng() {};~Rng() {};双生成(双最小，双最大)；};#万一

rng.cc

#include "rng.h";double Rng::generate(double min, double max) {返回 0.75；//只是为了测试}

构建

load("@rules_cc//cc:defs.bzl", "cc_binary", "cc_library")cc_图书馆(名称 = "rng",srcs = [rng.cc"],hdrs = [rng.h"],)cc_binary(name = "coinflipper",srcs = [coinflipper.cc"，coinflipper.h"]，深度 = [":rng",],)cc_test(name = "coinflipper_test",大小=小"，srcs = [coinflipper_test.cc"，rng.h"，coinflipper.h"，coinflipper.cc"]，deps = ["@com_google_googletest//:gtest_main"],)

coinflipper_test.cc

#include "coinflipper.h";#include "gmock/gmock.h";#include "gtest/gtest.h";类 MockRng {民众:MOCK_METHOD2(生成，双(双，双))；};测试(CoinFlipper，ShouldReturnHeadsIfRandValueIsLessThanProbability){MockRng;EXPECT_CALL(rng, generate(::testing::DoubleEq(0.0),::testing::DoubleEq(1.0))).Times(::testing::Exactly(1)).WillOnce(::testing::Return(0.25));CoinFlipper coinFlipper；自动结果 = coinFlipper.flipCoin(rng);EXPECT_EQ(CoinFlipper::HEADS，结果)；}

巴泽尔输出:

$ bazel test --test_output=all//:coinflipper_test调试:规则rules_cc"表明可以通过修改参数 sha256 =56ac9633c13d74cb71e0546f103ce1c58810e4a76aa8325da593ca4277908d7"来获得规范的可复制形式调试:存储库 rules_cc 在以下位置实例化:/Users/pvd/Downloads/toys/cpp/bazelgtest/coinflipper/WORKSPACE:9:13: 在 <toplevel>存储库规则 http_archive 定义在:/private/var/tmp/_bazel_pvd/3077b447e558b1418694504407cbcb45/external/bazel_tools/tools/build_defs/repo/http.bzl:336:31:在<toplevel>调试:规则com_google_googletest"表明可以通过修改参数 sha256 = 5cf189eb6847b4f8fc603a3ffff3b0771c08eec7dd4bd961bfd45477dd13eb73"来获得规范的可重现形式调试:存储库 com_google_googletest 在以下位置实例化:/Users/pvd/Downloads/toys/cpp/bazelgtest/coinflipper/WORKSPACE:3:13: 在 <toplevel>存储库规则 http_archive 定义在:/private/var/tmp/_bazel_pvd/3077b447e558b1418694504407cbcb45/external/bazel_tools/tools/build_defs/repo/http.bzl:336:31:在<toplevel>信息:分析目标//:coinflipper_test(加载了 0 个包，配置了 0 个目标).信息:找到 1 个测试目标...失败://:coinflipper_test(见/private/var/tmp/_bazel_pvd/3077b447e558b1418694504407cbcb45/execroot/__main__/bazel-out/darwin-fastbuild/testlogs/coinflipper_test/test.log)信息:来自测试//:coinflipper_test:====================//:coinflipper_test 的测试输出:dyld:懒惰的符号绑定失败:找不到符号:__ZN3Rng8generateEdd参考自:/private/var/tmp/_bazel_pvd/3077b447e558b1418694504407cbcb45/sandbox/darwin-sandbox/72/execroot/__main__/bazel-out/darwin-fastbuild/bin/coinflipper/__lip__main_runfiles预期在:平面命名空间dyld:找不到符号:__ZN3Rng8generateEdd参考自:/private/var/tmp/_bazel_pvd/3077b447e558b1418694504407cbcb45/sandbox/darwin-sandbox/72/execroot/__main__/bazel-out/darwin-fastbuild/bin/coinflipper/__lip__main_runfiles预期在:平面命名空间================================================================================目标//:coinflipper_test 最新:bazel-bin/coinflipper_test信息:经过时间:0.429s，关键路径:0.14s信息:2 个进程:2 个达尔文沙箱.信息:构建完成，1 次测试失败，总共 2 次操作//:coinflipper_test 在 0.1 秒内失败/private/var/tmp/_bazel_pvd/3077b447e558b1418694504407cbcb45/execroot/__main__/bazel-out/darwin-fastbuild/testlogs/coinflipper_test/test.log信息:构建完成，1 次测试失败，总共 2 次操作

解决方案

定义依赖项(这里的随机生成器)因为局部变量不推荐，做依赖项注入要困难得多(否则不可能)，所以我将函数 Rng_t 更改为模板函数并将 Rng 作为参数传递.

在实践中构造随机代可能是繁重的工作，需要初始化它的内部状态，每次调用flipCoin函数都构造它是浪费.

非虚函数可以mock，一种最常用的策略是使用模板，这里我们把类CoinFlipper的成员函数作为模板函数，然后我们可以用我们的MockRng测试依赖代码>.

注意，对于模板函数，我们需要在头文件中定义成员函数.

coinflipper.h:

#pragma once#include "rng.h";类 CoinFlipper {民众:枚举结果 { HEADS = 0, TAILS = 1 };模板 结果翻转硬币(Rng_t& rng){const double val = rng.generate(0.0, 1.0);返回 (val <0.5) ?头部:尾巴；}};

测试文件部分，MockRng 现在没有继承任何东西.我们在这里使用的测试成员函数的类型为 CoinFlipper::flipCoin.对于生产代码:我们使用类型 CoinFlipper::flipCoin

//#include "mockrng.h";#include "coinflipper.h";#include "gmock/gmock.h";#include "gtest/gtest.h";类 MockRng {民众:MOCK_METHOD2(生成，双(双，双))；};测试(CoinFlipper，ShouldReturnHeadsIfRandValueIsLessThanProbability){MockRng;EXPECT_CALL(rng, generate(::testing::DoubleEq(0.0),::testing::DoubleEq(1.0))).Times(::testing::Exactly(1)).WillOnce(::testing::Return(0.25));CoinFlipper coinFlipper；自动结果 = coinFlipper.flipCoin(rng);EXPECT_EQ(CoinFlipper::HEADS，结果)；}

在此处查看相关问题:

模拟非虚拟方法 C++ (gmock)

官方文档:

https://chromium.googlesource.com/external/github.com/google/googletest/+/refs/tags/release-1.8.0/googlemock/docs/CookBook.md>

I somehow extended the gmock test case from donsoft.io's example, and made it as follows:

coinflipper/
├── BUILD
├── WORKSPACE
├── coinflipper.cc
├── coinflipper.h
├── rng.cc
└── rng.h

Well, instead put the Rng class as a parameter of the constructor of the CoinFlipper, I made it initialized inside the CoinFlipper::flipCoin() method.

I was wondering how to mock the generate() from Rng in this case?

coinflipper.cc

#include "coinflipper.h"
#include <iostream>

CoinFlipper::CoinFlipper() {}

CoinFlipper::Result CoinFlipper::flipCoin() const {
  Rng d_rng;
  const double val = d_rng.generate(0.0, 1.0);
  return (val < 0.5) ? HEADS : TAILS;
}

int main(int argc, char** argv) {
  CoinFlipper cf;
  CoinFlipper::Result res = cf.flipCoin();
  if (res == 0) {
      std::cout << "head" << std::endl;
  } else {
      std::cout << "tail" << std::endl;
  }
  return 0;
}

coinflipper.h

#include "rng.h"

class CoinFlipper {
 public:
  enum Result { HEADS = 0, TAILS = 1 };

  explicit CoinFlipper();
  Result flipCoin() const;
};

rng.h

#ifndef RNG_H
#define RNG_H
class Rng {
 public:
  Rng() {};
  ~Rng() {};
  double generate(double min, double max);
};
#endif

rng.cc

#include "rng.h"

double Rng::generate(double min, double max) {    
    return 0.75; //Just for test
}

BUILD

load("@rules_cc//cc:defs.bzl", "cc_binary", "cc_library")

cc_library(
    name = "rng",
    srcs = ["rng.cc"],
    hdrs = ["rng.h"],
)

cc_binary(
    name = "coinflipper",
    srcs = ["coinflipper.cc", "coinflipper.h"],    
    deps = [
        ":rng",
    ],
)

cc_test(
  name = "coinflipper_test",
  size = "small",
  srcs = ["coinflipper_test.cc", "rng.h","coinflipper.h","coinflipper.cc"],
  deps = ["@com_google_googletest//:gtest_main"],
)

coinflipper_test.cc

#include "coinflipper.h"
#include "gmock/gmock.h"
#include "gtest/gtest.h"

class MockRng {
 public:  
  MOCK_METHOD2(generate, double(double, double));
};

TEST(CoinFlipper, ShouldReturnHeadsIfRandValueIsLessThanProbability) {
  MockRng rng;
  EXPECT_CALL(rng, generate(::testing::DoubleEq(0.0), ::testing::DoubleEq(1.0)))
      .Times(::testing::Exactly(1))
      .WillOnce(::testing::Return(0.25));

  CoinFlipper coinFlipper;
  auto result = coinFlipper.flipCoin<MockRng>(rng);
  EXPECT_EQ(CoinFlipper::HEADS, result);
}

Bazel output:

$ bazel test --test_output=all //:coinflipper_test
DEBUG: Rule 'rules_cc' indicated that a canonical reproducible form can be obtained by modifying arguments sha256 = "56ac9633c13d74cb71e0546f103ce1c58810e4a76aa8325da593ca4277908d72"
DEBUG: Repository rules_cc instantiated at:
  /Users/pvd/Downloads/toys/cpp/bazelgtest/coinflipper/WORKSPACE:9:13: in <toplevel>
Repository rule http_archive defined at:
  /private/var/tmp/_bazel_pvd/3077b447e558b1418694504407cbcb45/external/bazel_tools/tools/build_defs/repo/http.bzl:336:31: in <toplevel>
DEBUG: Rule 'com_google_googletest' indicated that a canonical reproducible form can be obtained by modifying arguments sha256 = "5cf189eb6847b4f8fc603a3ffff3b0771c08eec7dd4bd961bfd45477dd13eb73"
DEBUG: Repository com_google_googletest instantiated at:
  /Users/pvd/Downloads/toys/cpp/bazelgtest/coinflipper/WORKSPACE:3:13: in <toplevel>
Repository rule http_archive defined at:
  /private/var/tmp/_bazel_pvd/3077b447e558b1418694504407cbcb45/external/bazel_tools/tools/build_defs/repo/http.bzl:336:31: in <toplevel>
INFO: Analyzed target //:coinflipper_test (0 packages loaded, 0 targets configured).
INFO: Found 1 test target...
FAIL: //:coinflipper_test (see /private/var/tmp/_bazel_pvd/3077b447e558b1418694504407cbcb45/execroot/__main__/bazel-out/darwin-fastbuild/testlogs/coinflipper_test/test.log)
INFO: From Testing //:coinflipper_test:
==================== Test output for //:coinflipper_test:
dyld: lazy symbol binding failed: Symbol not found: __ZN3Rng8generateEdd
  Referenced from: /private/var/tmp/_bazel_pvd/3077b447e558b1418694504407cbcb45/sandbox/darwin-sandbox/72/execroot/__main__/bazel-out/darwin-fastbuild/bin/coinflipper_test.runfiles/__main__/coinflipper_test
  Expected in: flat namespace

dyld: Symbol not found: __ZN3Rng8generateEdd
  Referenced from: /private/var/tmp/_bazel_pvd/3077b447e558b1418694504407cbcb45/sandbox/darwin-sandbox/72/execroot/__main__/bazel-out/darwin-fastbuild/bin/coinflipper_test.runfiles/__main__/coinflipper_test
  Expected in: flat namespace

================================================================================
Target //:coinflipper_test up-to-date:
  bazel-bin/coinflipper_test
INFO: Elapsed time: 0.429s, Critical Path: 0.14s
INFO: 2 processes: 2 darwin-sandbox.
INFO: Build completed, 1 test FAILED, 2 total actions
//:coinflipper_test                                                      FAILED in 0.1s
  /private/var/tmp/_bazel_pvd/3077b447e558b1418694504407cbcb45/execroot/__main__/bazel-out/darwin-fastbuild/testlogs/coinflipper_test/test.log

INFO: Build completed, 1 test FAILED, 2 total actions

解决方案

Define dependencies(The random generator here) as local variables are not recommended, it's much harder to do dependencies injection(Or it won't be possible), so I change the functions Rng_t into template function and pass the Rng as a parameter.

In practice to construct a random generation may be heavy work, it needs to initialize its internal status, to construct it every time we call the function flipCoin is waste.

The non-virtual function can be mocked, one most commonly used strategy is to use the template, here we make the class CoinFlipper's member function as a template function, then we can test the dependency with our MockRng.

Be aware that for the template function, we need to define the member function in the header file.

coinflipper.h:

#pragma once
#include "rng.h"

class CoinFlipper {
 public:
  enum Result { HEADS = 0, TAILS = 1 };

  template <typename Rng_t>
  Result flipCoin(Rng_t& rng) {
    const double val = rng.generate(0.0, 1.0);
    return (val < 0.5) ? HEADS : TAILS;
  }
};

The test file part, MockRng doesn't inherit anything now. And the test member function we use here has the type CoinFlipper::flipCoin<MockRng>. For production code: we use the type CoinFlipper::flipCoin<Rng>

//#include "mockrng.h"

#include "coinflipper.h"
#include "gmock/gmock.h"
#include "gtest/gtest.h"

class MockRng {
 public:
  MOCK_METHOD2(generate, double(double, double));
};

TEST(CoinFlipper, ShouldReturnHeadsIfRandValueIsLessThanProbability) {
  MockRng rng;
  EXPECT_CALL(rng, generate(::testing::DoubleEq(0.0), ::testing::DoubleEq(1.0)))
      .Times(::testing::Exactly(1))
      .WillOnce(::testing::Return(0.25));

  CoinFlipper coinFlipper;
  auto result = coinFlipper.flipCoin<MockRng>(rng);
  EXPECT_EQ(CoinFlipper::HEADS, result);
}

See related question here:

Mock non-virtual method C++ (gmock)

The official document:

https://chromium.googlesource.com/external/github.com/google/googletest/+/refs/tags/release-1.8.0/googlemock/docs/CookBook.md

这篇关于如何使用 gmock 在具体类中模拟非虚拟方法?的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！