这是 Maximilian Schwarzmüller 讲授的《JavaScript 单元测试 - 实用指南》课程的总结。
深入研究自动化测试可能是一个好主意,因为与手动测试相比,它更具可预测性和一致性。
让我们考虑一个场景,您添加新功能或更改代码的某些部分。
您不一定知道或意识到代码中真正受到影响的每个部分或所有部分。
通过手动测试,我们必须测试(或尝试)整个应用程序,或者至少测试可能受更改影响的部分。为什么?因为我们需要看看一切是否仍然正常工作,并且那个微小的变化或新功能没有破坏某处的代码。
所以,正如我们想象的那样,有很多工作。
此外,不能保证需要测试的所有内容都会被测试,并且每次发生变化时,都不能保证以相同的方式进行测试。
通过自动化测试,我们有初步的努力,但后来会带来很多好处。
在当今我们拥有的各种自动化测试中,这里我们将讨论单元测试和集成测试,但主要是单元测试。
让我们将单元视为代码的最小部分,例如函数或类。因此,在单元测试中,顾名思义,我们将测试应用程序的每个单元。因此,如果我们有 50 个函数,我们将为所有或大多数函数创建测试。这是因为我们希望保证每个部分、每个单元都按照最初的预期正常工作。
另一方面,集成测试将关注一起测试这些单元,或者更好地说,它们如何一起工作以及它们是否可以很好地一起工作。这是因为即使我们单独测试这些单元,也不能保证它们能够协同工作或按应有的方式工作。
我们还应该了解 TDD,即测试驱动开发的简称。
TDD 是一个框架/哲学,它引导我们考虑首先编写失败的测试,然后实现使测试成功的代码。然后重构,作为一个循环的事情。
让你的测试保持简单!
当其他人需要阅读你的代码,甚至在未来的场景中需要阅读你的代码时,重要的是不要花太长时间来理解。这必须是一项简单的任务。
对于本指南,我们将使用 Vitest,这是一个基于著名的 Jest 应用程序的工具,用于测试等。在这里,它并不是要深入 Vitest 语法或所有功能,而是作为理解测试核心的工具。
如果您想了解或查看 Vitest 可以帮助我们的所有内容,请访问以下链接的文档:Vitest 文档
Vitest 的工作方式与 Webpack 捆绑工具类似,因此在使用 ES 模块表示法时,我们实际上不需要显式告知要导入的文件的扩展名。例如:
从“./math/Math”导入数学
相反的是:
从 './math/Math.js' 导入数学
这里有一个小指南,可引导您在编写测试例程中进行良好实践。
单元和集成测试可能非常有用,但前提是编写得好。为此,我们可以遵循以下将探讨的一系列“良好实践”:
当我们谈论测试我们自己的代码时,这意味着第三方代码不是我们测试的责任。嗯,它是供第三方代码编写者使用的,以确保其正常工作,因此测试它是他们的责任。
而且因为你无法更改它,所以测试它没有用。
您不会通过客户端代码隐式测试服务器端代码。
您将测试客户端对不同响应和错误的反应。
因此,将您的测试、前端开发测试和后端开发测试分开。
这里有一个 Vitest 代码示例:
import { expect, it } from 'vitest'; import { toSum } from './math'; it('should sum two numbers', () => { //Arrange const num1 = 1; const num2 = 2; //Act const result = toSum(num1, num2); //Assert const expectedResult = num1 + num2; expect(result).toBe(expectedResult); });
只是为了澄清代码的作用,以防您不知道。
首先,我们从 vitest 导入“expect”和“it”,这是我们需要的功能,以及“toSum”函数,这是为示例构建的函数,但位于另一个文件中。
The "it" works as the scope for our test; it receives a string that behaves as the identifier and a function that will run the test code. Here is very simple; we are saying that it should sum two numbers, that's our expectation for the function and for the test.
In the arrange part we create the variables which will be passed to the "toSum" function. Then, in the act part, we create a constant that will receive the result of the function. Finally, in the assert, we will store the expected result, which would be the sum of the two numbers and use "expect" from Vitest to make the assertion. Basically, we are saying that we expect the result to be the same as the expected result.
There are many assertion possibilities, so please do check the documentation for further study.
Note: "it" is an alias for "test"
Also, it's very important the following line:
const expectedResult = num1 + num2;
Imagine if we've done it like this:
const expectedResult = 3;
It's okay for the current test because we are indeed expecting 3 as the result.
But, imagine in the future, someone changes "num1" or "num2", and forgets to change the "expectedResult"; it would not work if the result of the sum was not 3.
If you, for example, created a function that is going to receive an array of numbers as an argument and you need to test if it actually received an array of numbers in your test file, you just need to reference an array, for example:
const items = [1, 2];
You don't need to create a bigger array, for example:
const items = [1, 2, 3, 4, 5, 6, 7, 8];
It's unnecessary and redundant. Keep it short, simple and concise, so you will make sure that you are only testing what needs to be tested or what is important for the function.
You can think of one thing as one feature or one behavior. For example, if you have a function that will sum two numbers (the same example above) and you need to make sure that is summing two numbers (because indeed that's the main functionality) but also that the output is of type number, then, you can separate it into two assertions, for example:
import { describe, expect, it } from 'vitest'; import { toSum } from './math'; describe('toSum()', () => { it('should sum two numbers', () => { const num1 = 1; const num2 = 2; const result = toSum(num1, num2); const expectedResult = num1 + num2; expect(result).toBe(expectedResult); }); it('should output a result of type number', () => { const num1 = 1; const num2 = 2; const result = toSum(num1, num2); const expectedResult = num1 + num2; expect(result).toBe(expectedResult); }); })
If you're wondering what describe does, it help us to create suites. As many suites as we want, like dividing blocks of tests. It keeps our code organized, clear, and easier to understand the outputting.
Here's an example using the toSum function:
As you can see in the image above, it will show us the file path, and after that the "describe" name, and then the "it" name. It's a good idea to keep the describer name short and referencing the function to what the tests are about.
And you could have describers inside describers to organize even further, it's up to you.
when we create our tests following good practices, we are creating tests that will actually help us on what's needed to be tested. And also, testing forces us to write better code. For that, we can write good functions that will hold only the logic of that function so it'll be easier to test what's need to be tested.
It's important to understand also that coverage doesn't mean good testing or testing that is useful and meaningful for the application. Well, you could cover 100% of your code with meaningless tests after all, or, missing important tests that you didn't think of.
Don't see a high amount of code coverage as the ultimate goal!
You will want to try and test cover the majority of the units (functions or classes) in your application, because that's what unit testing is about, but, there is some code that doesn't need to be tested.
Vitest comes with a built-in functionality to help us measure the code coverage; you can access in the following link: Vitest coverage tool
As callbacks and async functions exhibit specific behavior in Vitest, this section is dedicated to exploring them superficially.
When testing for a callback, keep in mind that Vitest does not wait for the response or for the callback to be executed. Therefore, we need to use the "done" argument.
Consider the following test as an example:
import { expect, it } from 'vitest'; import { generateToken } from './async-example'; it('should generate a token value', (done) => { const email = 'test@mail.com'; generateToken(email, (err, token) => { expect(token).toBeDefined(); done() }) })
Now, we are working with a callback function. Notice that there's a parameter being passed. The "done".
Vitest now needs to wait until the done function is called.
What would happen if we didn't use the "done" argument? The "expect" wouldn't be executed.
Still in that function, imagine if we changed toBeDefined to toBe, as in the image below:
import { expect, it } from 'vitest'; import { generateToken } from './async-example'; it('should generate a token value', (done) => { const email = 'test@mail.com'; generateToken(email, (err, token) => { expect(token).toBe(2); done(); }); })
By default, in Vitest, the "toBe" function throws an error each time something doesn't go as expected, in this case, if the token returned wasn't 2.
However, as we are working with a callback, we will need to add an exception handling syntax, such as "try and catch", because if we don't do so, the test will timeout.
import { expect, it } from 'vitest'; import { generateToken } from './async-example'; it('should generate a token value', (done) => { const email = 'test@mail.com'; try { generateToken(email, (err, token) => { expect(token).toBe(2); }); } catch (error) { done(error); } })
Since we are dealing with an error, we also need to pass this error to the "done" function.
Now, when working with promises, it's a bit easier, or should we say, simpler.
import { expect, it } from 'vitest'; import { generateTokenPromise } from './async-example'; it('should generate a token value', () => { const email = 'test@mail.com'; return expect(generateTokenPromise(email)).resolves.toBeDefined(); //or return expect(generateTokenPromise(email)).rejects.toBe(); });
Here we have two possibilities: resolves and rejects
The "return" statement guarantees Vitest waits for the promise to be resolved.
Alternatively, we have:
import { expect, it } from 'vitest'; import { generateTokenPromise } from './async-example'; it('should generate a token value', async () => { const email = 'test@mail.com'; const token = await generateTokenPromise(email); expect(token).resolves.toBeDefined(); // or expect(token).rejects.toBe(); })
Here we don't need to use "return" because "async/await" is being used (since a function annotated with "async" returns a promise implicitly).
Here, we are going to explore a little bit of these important functionalities that Vitest provides to us.
Imagine working with a bunch of tests that use the same variable, and you don't want to initialize it every single time in every single test because it's the same.
Hooks can help us in this case because you can use functions provided by it that allow us to reuse this variable.
Functions available: "beforeAll", "beforeEach", "afterEach", "afterAll".
Here goes a simple example just to show how it works:
import { beforeEach } from 'vitest'; let myVariable; beforeEach(() => { myVariable = ""; }); it('sentence', () => { myVariable = "Hello"; }); it('another sentence', () => { myVariable += 2; });
Now, imagine the same but without the hook:
let myVariable; it('sentence', () => { myVariable = "Hello"; }); it('another sentence', () => { myVariable += 2; });
As we can see, when using "beforeEach" and a global variable, before each test starts to execute, the variable will be "cleaned". This allows the tests to use the variable as if it were fresh.
But, without using the hook and using a global variable, in some cases, things would be tricky. In the example of the test "another sentence," if we didn't clean the variable, it would be holding "Hello" because the "sentence" test is run first. And that's not what we want.
Mocks and spies are mainly to handle side effects and external dependencies.
We could say that spies help us deal with side effects in our functions, and mocks help us deal with side effects of external dependencies.
For that, you will have to import "vi" from vitest.
To build a spy, you can use "vi.fn()" and for a mock "vi.mock()". Inside each function, you will pass the name to the other function (your or external).
So, spies and mocks kind of replace the actual functions with other functions or empty functions.
Mocks will be available only for tests of the file you called them and Vitest, behind the scenes, puts them at the start of the file.
In summary, you need to consider what the unit should or should not do. To achieve this, you can utilize the "it" syntax provided by Vitest, which takes a string describing your expectations and a function that will test the given expectations.
The name of the test should be short, simple and easy to understand.
The testing magic lies in thinking about aspects that were not initially considered, leading to code improvement. This process helps prevent errors and promotes a clearer understanding of what the function should do and its expected behaviors.
以上是JS单元测试讲座-实用指南的详细内容。更多信息请关注PHP中文网其他相关文章!