How to write unit test cases

Hello everyone , I meet you again , I'm your friend, Quan Jun .

　 One 、 unit test The concept of

　　 Generally speaking, a unit refers to a function that implements simple functions . Unit testing is to use only a specific set of inputs ( The test case ) Test whether the function functions normally , And returned the correct output .

　　 Type of coverage tested

　　1. Statement override ： Statement coverage is to design several test cases , Run the program under test , Make each executable statement execute at least once .

　　2. Determine coverage （ Also called branch coverage ）： Design several test cases , Run the program under test , Make each true branch and false branch of judgment in the program execute at least once .

　　3. Conditional coverage ： Design enough test cases , Run the program under test , Make each possible value of each condition of each judgment in the program execute at least once .

　　4. determine —— Conditional coverage ： Design enough test cases , Run the program under test , Make each possible value of each condition of each judgment in the program execute at least once , And each possible judgment result is executed at least once .

　　5. Conditional combination test ： Design enough test cases , Run the program under test , Make the combination of all conditional values of each judgment in the program execute at least once .

　　6. Path test ： Design enough test cases , Run the program under test , To override all possible paths in the program .

　　 The main design schemes of use cases are as follows ： Conditions for testing , Basic path testing , Cycle test . Through the above method, the logic coverage of the test case to the program can be realized , And path coverage .

Two 、 Preparation before starting the test

　　 At the beginning of the test , Let's make a statement first , No matter how many test cases you design , No matter how perfect your test plan is , Can't completely 100% All found BUG, What we need to do is use the least resources , Do the most tests and checks , Find a balance point to ensure the correctness of the program . Exhaustive testing is impossible . Therefore, for unit testing, I choose the basic path testing method commonly used now .

3、 ... and 、 Start testing

　　 Basic path testing ： The designed test cases should ensure that each basic independent path should be executed at least once .

　　 Function description ： When i_flag=0; return i_count+100

　　 When i_flag=1; return i_count*10

　　 Otherwise return to i_count*20

　　 Input parameters ：inti_count,

　　inti_flag

　　 Output parameters ：inti_return;

　　 Code ：

1intTest(inti_count,inti_flag) 　　2{ 　　3inti_temp=1; 　　4while(i_count>0) 　　5{ 　　6if(0==i_flag) 　　7{ 　　8i_temp=i_count+100; 　　9break; 　　10} 　　11else 　　12{ 　　13if(1==i_flag) 　　14{ 　　15i_temp=i_temp*10; 　　16} 　　17else 　　18{ 　　19i_temp=i_temp*20; 　　20} 　　21} 　　22i_count–; 　　23} 　　24returni_temp; 　　25}

　　1. Draw a program control flow chart

　　 legend ：

　　 Case program flow chart ：

　　 The number in the circle represents the line number of the statement , Maybe someone asked why 4,6,13,8…… As nodes , The first 2 That's ok , The first 3 What behavior is not a node , Because the selection of nodes is regular . Let's look at the program ; The first 2 That's ok , The first 3 The rows are executed in order . Until the first 4 Then there is the circular operation . and 2,3 OK, no judgment , Select branch operation , So we put 2,3,4 Merge all into one node . Others are merged according to this rule , Then there is the flow chart above .

　　2. Computational cyclomatic complexity

　　 With the diagram, we need to know how many test cases we have written , To meet the basic path test .

　　 Here's a new concept —— Cyclomatic complexity

　　 Cycle complexity is a software measure that provides quantitative test for program logic complexity . This metric is used to calculate the number of basic independent paths of the program . An upper bound on the number of tests to ensure that all statements are executed at least once .

　　 Formula cycle complexity V（G）=E-N+2,E Is the number of edges in the flow diagram ,N Is the number of nodes in the flow diagram .

　　 Formula cycle complexity V（G）=P+1,P It's a flow chart G Determines the number of nodes .

　　 Generally speaking, circle responsibility is to judge whether the unit is complex , Is it a good test standard . Generally speaking, if the cyclomatic complexity is greater than 20 This means that the testability of this unit is not good , Too complicated （ Maybe some people think it doesn't matter , But if your company implements CMMI5 Words , There are rules for this ）.

　　 We can see from the picture ,

　　V（G）=10 side -8 node +2=4

　　V（G）=3 Two decision nodes +1=4

　　 The circle complex graph in the above figure is 4. What does this result mean to us ？ It means that we only need... At most 4 One test case can achieve basic path coverage .

　　3. Export program basic path .

　　 Now we know, at least write 4 Test cases , But how to design this 4 Test cases ？

　　 Export program basic path , Design test examples according to the basic path of the program .

　　 Program basic path ： The basic independent path is that you can choose any path to traverse from the beginning node to the end of the program , But each path should contain at least one edge that the defined path does not use .（ It seems hard to understand , Let's look at an example ）.

　　 Let's look at the flow chart above ： From the node 4 To 24 There are several paths ？

　　1B（4,24）

　　2C,E,J（4,6,8,24）

　　3C,D,F,H,A,B（4,6,13,15,22,4,24）

　　4C,D,G,I,A,B（4,6,13,19,22,4,24）

　　 anything else ？？

　　5C,D,C,I,A,C,E,J（4,6,13,19,22,4,6,8,24） Is that right ？

　　 not , Why? ？ Because of the above 4 This path already includes all the edges . The first 5 This path no longer contains unused edges . All paths have been traversed .

　　 Okay , Now we have 4 Basic independent path according to which we can design test cases .

　　1B（4,24）

　　 input data ：i_count=0, Or is it i_count<0 A certain value of .

　　 Expected results ：i_temp=0.

　　2C,E,J（4,6,8,24）

　　 input data ：i_count=1;i_flag=0

　　 Expected results ：i_temp=101.

　　3C,D,F,H,A,B（4,6,13,15,22,4,24）

　　 input data ：i_count=1;i_flag=1

　　 Expected results ：i_temp=10.

　　4C,D,G,I,A,B（4,6,13,19,22,4,24）

　　 input data ：i_count=1;i_flag=2

　　 Expected results ：i_temp=20.

The input data here is inferred from the path and program . Note that the expected result is derived from the function description , Cannot export... From program structure .

　　 Why do you say that? ？

　　 Let's look at the second part of the program 3 That's ok .

　　inti_temp=1; If a developer accidentally makes a mistake , Turned into inti_temp=0; The expected result derived from the program will be an incorrect value , But unit testing doesn't work .

　　 Then the unit test is meaningless .

　　 Someone might ask that such a simple function has 4 Test cases , What if it's more complicated ？ Can the above test cases be simplified ？ The answer is yes .

　　 Let's look at the path 1B（4,24） and 4C,D,G,I,A,B（4,6,13,19,22,4,24）, route 1 It's the path 4 The proper subset of , therefore 1 It can be unnecessary . The cycle complexity of the above figure is 4. What does this result mean to us ？ It means that we only need... At most 4 One test case can achieve basic path coverage . So the cyclomatic complexity indicator is the largest number of test cases , Not necessarily 4 Only one test case can . However, one thing to be clear is that the simpler the test cases, the less your tests , The less secure the program is .

　 Four 、 To complete the test

　　 Next, use the tools to test according to the test cases NUNIT,VS2005 Fine .

　　 Next, write a test report according to the test results , Tested by , Time , result , Use cases , Whether to pass , Format a lot of online , The format of each company is different, so don't say .

Publisher ： Full stack programmer stack length , Reprint please indicate the source ：https://javaforall.cn/148660.html Link to the original text ：https://javaforall.cn