Test Driven Development (TDD) simply means that you write the tests first. Before writing a line of business logic, you can pre-set expectations for correct code. TDD not only helps ensure that your code is correct, but it also helps you write smaller functions, refactor your code without breaking functionality, and better understand your problem.
In this article, I will introduce some concepts of TDD by building a small utility program. We'll also cover some practical scenarios where TDD will make your life easier.
We will gradually build a simple HTTP client that abstracts various HTTP verbs. To make refactoring go smoothly, we will follow TDD practices. We will use Jasmine, Sinon and Karma for testing. First, copy package.json, karma.conf.js, and webpack.test.js from the sample project, or clone the sample project directly from the GitHub repository.
It will help if you understand how the new Fetch API works, but these examples should be easy to understand. For newbies, Fetch API is a better alternative to XMLHttpRequest. It simplifies network interaction and works well with Promises.
First, create an empty file at src/http.js and an accompanying test file at src/__tests__/http-test.js.
Let's set up a test environment for this service.
import * as http from "../http.js";
import sinon from "sinon";
import * as fetch from "isomorphic-fetch";
describe("TestHttpService", () => {
describe("Test success scenarios", () => {
beforeEach(() => {
stubedFetch = sinon.stub(window, "fetch");
window.fetch.returns(Promise.resolve(mockApiResponse()));
function mockApiResponse(body = {}) {
return new window.Response(JSON.stringify(body), {
status: 200,
headers: { "Content-type": "application/json" }
});
}
});
});
});
We are using Jasmine and Sinon here - Jasmine defines test scenarios, Sinon assertions and monitoring objects. (Jasmine has its own way of monitoring and stubbing tests, but I prefer Sinon's API.)
The code above is self-explanatory. Before each test run, we hijack the call to the Fetch API since there is no server available and return a mock Promise object. The goal here is to unit test whether the Fetch API is called with the correct parameters and see if the wrapper handles any network errors correctly.
Let’s start with the failing test case:
describe("Test get requests", () => {
it("should make a GET request", done => {
http.get(url).then(response => {
expect(stubedFetch.calledWith(`${url}`)).toBeTruthy();
expect(response).toEqual({});
done();
});
});
});
Start the test runner by calling karma start. Now the test will obviously fail because there is no get method in http. Let's correct this problem.
const status = response => {
if (response.ok) {
return Promise.resolve(response);
}
return Promise.reject(new Error(response.statusText));
};
export const get = (url, params = {}) => {
return fetch(url)
.then(status);
};
If you run the test now, you will see the failed response, showing Expected [object Response] equal to Object({ }). The response is a Stream object. As the name suggests, a stream object is a data stream. To get data from a stream, you need to first read the stream using some of its helper methods. Now, we can assume the stream is JSON and deserialize it by calling response.json().
const deserialize = response => response.json();
export const get = (url, params = {}) => {
return fetch(url)
.then(status)
.then(deserialize)
.catch(error => Promise.reject(new Error(error)));
};
Our test suite should now be green.
So far, the get method just makes a simple call without any query parameters. Let's write a failing test and see how it handles query parameters. If we pass { users: [1, 2], limit: 50, isDetailed: false } as query parameters, our HTTP client should make a network call to /api/v1/users/ ?users=1&users=2&limit=50&isDetailed=false.
it("should serialize array parameter", done => {
const users = [1, 2];
const limit = 50;
const isDetailed = false;
const params = { users, limit, isDetailed };
http
.get(url, params)
.then(response => {
expect(stubedFetch.calledWith(`${url}?isDetailed=false&limit=50&users=1&users=2/`)).toBeTruthy();
done();
})
});
Now that we have the test set up, let's extend the get method to handle the query parameters.
import { stringify } from "query-string";
export const get = (url, params) => {
const prefix = url.endsWith('/') ? url : `${url}/`;
const queryString = params ? `?${stringify(params)}/` : '';
return fetch(`${prefix}${queryString}`)
.then(status)
.then(deserializeResponse)
.catch(error => Promise.reject(new Error(error)));
};
If the parameter is present, we construct a query string and append it to the URL.
Here, I'm using the Query String Library - a nice little helper library that helps with various query parameter scenarios.
GET is probably the simplest HTTP method to implement. GET is idempotent and should not be used for any mutations. POST usually means updating some records in the server. This means that POST requests require some safeguards by default, such as CSRF tokens. More on this in the next section.
Let’s first build a test for a basic POST request:
describe(`Test post requests`, () => {
it("should send request with custom headers", done => {
const postParams = {
users: [1, 2]
};
http.post(url, postParams, { contentType: http.HTTP_HEADER_TYPES.text })
.then(response => {
const [uri, params] = [...stubedFetch.getCall(0).args];
expect(stubedFetch.calledWith(`${url}`)).toBeTruthy();
expect(params.body).toEqual(JSON.stringify(postParams));
expect(params.headers.get("Content-Type")).toEqual(http.HTTP_HEADER_TYPES.text);
done();
});
});
});
The signature of POST is very similar to that of GET. It requires an options attribute where you define the header, body, and most importantly the method. The method describes the HTTP verb, in this case "post".
Now, let's assume the content type is JSON and start implementing the POST request.
export const HTTP_HEADER_TYPES = {
json: "application/json",
text: "application/text",
form: "application/x-www-form-urlencoded",
multipart: "multipart/form-data"
};
export const post = (url, params) => {
const headers = new Headers();
headers.append("Content-Type", HTTP_HEADER_TYPES.json);
return fetch(url, {
headers,
method: "post",
body: JSON.stringify(params),
});
};
At this point, our post method is very primitive. It doesn't support anything other than JSON requests.
让我们允许调用者决定内容类型，并将 CSRF 令牌投入战斗。根据您的要求，您可以将 CSRF 设为可选。在我们的用例中，我们将假设这是一个选择加入功能，并让调用者确定是否需要在标头中设置 CSRF 令牌。
为此，首先将选项对象作为第三个参数传递给我们的方法。
it("should send request with CSRF", done => {
const postParams = {
users: [1, 2 ]
};
http.post(url, postParams, {
contentType: http.HTTP_HEADER_TYPES.text,
includeCsrf: true
}).then(response => {
const [uri, params] = [...stubedFetch.getCall(0).args];
expect(stubedFetch.calledWith(`${url}`)).toBeTruthy();
expect(params.body).toEqual(JSON.stringify(postParams));
expect(params.headers.get("Content-Type")).toEqual(http.HTTP_HEADER_TYPES.text);
expect(params.headers.get("X-CSRF-Token")).toEqual(csrf);
done();
});
});
当我们提供 options 和 {contentType: http.HTTP_HEADER_TYPES.text,includeCsrf: true 时，它应该相应地设置内容标头和 CSRF 标头。让我们更新 post 函数以支持这些新选项。
export const post = (url, params, options={}) => {
const {contentType, includeCsrf} = options;
const headers = new Headers();
headers.append("Content-Type", contentType || HTTP_HEADER_TYPES.json());
if (includeCsrf) {
headers.append("X-CSRF-Token", getCSRFToken());
}
return fetch(url, {
headers,
method: "post",
body: JSON.stringify(params),
});
};
const getCsrfToken = () => {
//This depends on your implementation detail
//Usually this is part of your session cookie
return 'csrf'
}
请注意，获取 CSRF 令牌是一个实现细节。通常，它是会话 cookie 的一部分，您可以从那里提取它。我不会在本文中进一步讨论它。
您的测试套件现在应该很满意。
我们的 post 方法现在已经成型，但是在发送正文时仍然很简单。您必须针对每种内容类型以不同的方式处理数据。处理表单时，我们应该在通过网络发送数据之前将数据编码为字符串。
it("should send a form-encoded request", done => {
const users = [1, 2];
const limit = 50;
const isDetailed = false;
const postParams = { users, limit, isDetailed };
http.post(url, postParams, {
contentType: http.HTTP_HEADER_TYPES.form,
includeCsrf: true
}).then(response => {
const [uri, params] = [...stubedFetch.getCall(0).args];
expect(stubedFetch.calledWith(`${url}`)).toBeTruthy();
expect(params.body).toEqual("isDetailed=false&limit=50&users=1&users=2");
expect(params.headers.get("Content-Type")).toEqual(http.HTTP_HEADER_TYPES.form);
expect(params.headers.get("X-CSRF-Token")).toEqual(csrf);
done();
});
});
让我们提取一个小辅助方法来完成这项繁重的工作。基于 contentType，它对数据的处理方式有所不同。
const encodeRequests = (params, contentType) => {
switch (contentType) {
case HTTP_HEADER_TYPES.form: {
return stringify(params);
}
default:
return JSON.stringify(params);
}
}
export const post = (url, params, options={}) => {
const {includeCsrf, contentType} = options;
const headers = new Headers();
headers.append("Content-Type", contentType || HTTP_HEADER_TYPES.json);
if (includeCsrf) {
headers.append("X-CSRF-Token", getCSRFToken());
}
return fetch(url, {
headers,
method="post",
body: encodeRequests(params, contentType || HTTP_HEADER_TYPES.json)
}).then(deserializeResponse)
.catch(error => Promise.reject(new Error(error)));
};
看看那个！即使在重构核心组件之后，我们的测试仍然可以通过。
另一个常用的 HTTP 动词是 PATCH。现在，PATCH 是一个变异调用，这意味着这两个操作的签名非常相似。唯一的区别在于 HTTP 动词。通过简单的调整，我们可以重用为 POST 编写的所有测试。
['post', 'patch'].map(verb => {
describe(`Test ${verb} requests`, () => {
let stubCSRF, csrf;
beforeEach(() => {
csrf = "CSRF";
stub(http, "getCSRFToken").returns(csrf);
});
afterEach(() => {
http.getCSRFToken.restore();
});
it("should send request with custom headers", done => {
const postParams = {
users: [1, 2]
};
http[verb](url, postParams, { contentType: http.HTTP_HEADER_TYPES.text })
.then(response => {
const [uri, params] = [...stubedFetch.getCall(0).args];
expect(stubedFetch.calledWith(`${url}`)).toBeTruthy();
expect(params.body).toEqual(JSON.stringify(postParams));
expect(params.headers.get("Content-Type")).toEqual(http.HTTP_HEADER_TYPES.text);
done();
});
});
it("should send request with CSRF", done => {
const postParams = {
users: [1, 2 ]
};
http[verb](url, postParams, {
contentType: http.HTTP_HEADER_TYPES.text,
includeCsrf: true
}).then(response => {
const [uri, params] = [...stubedFetch.getCall(0).args];
expect(stubedFetch.calledWith(`${url}`)).toBeTruthy();
expect(params.body).toEqual(JSON.stringify(postParams));
expect(params.headers.get("Content-Type")).toEqual(http.HTTP_HEADER_TYPES.text);
expect(params.headers.get("X-CSRF-Token")).toEqual(csrf);
done();
});
});
it("should send a form-encoded request", done => {
const users = [1, 2];
const limit = 50;
const isDetailed = false;
const postParams = { users, limit, isDetailed };
http[verb](url, postParams, {
contentType: http.HTTP_HEADER_TYPES.form,
includeCsrf: true
}).then(response => {
const [uri, params] = [...stubedFetch.getCall(0).args];
expect(stubedFetch.calledWith(`${url}`)).toBeTruthy();
expect(params.body).toEqual("isDetailed=false&limit=50&users=1&users=2");
expect(params.headers.get("Content-Type")).toEqual(http.HTTP_HEADER_TYPES.form);
expect(params.headers.get("X-CSRF-Token")).toEqual(csrf);
done();
});
});
});
});
类似地，我们可以通过使动词可配置来重用当前的 post 方法，并重命名方法名称以反映通用的内容。
const request = (url, params, options={}, method="post") => {
const {includeCsrf, contentType} = options;
const headers = new Headers();
headers.append("Content-Type", contentType || HTTP_HEADER_TYPES.json);
if (includeCsrf) {
headers.append("X-CSRF-Token", getCSRFToken());
}
return fetch(url, {
headers,
method,
body: encodeRequests(params, contentType)
}).then(deserializeResponse)
.catch(error => Promise.reject(new Error(error)));
};
export const post = (url, params, options = {}) => request(url, params, options, 'post');
现在我们所有的 POST 测试都已通过，剩下的就是为 patch 添加另一个方法。
export const patch = (url, params, options = {}) => request(url, params, options, 'patch');
很简单，对吧？作为练习，尝试自行添加 PUT 或 DELETE 请求。如果您遇到困难，请随时参考该存储库。
社区对此存在分歧。有些程序员一听到 TDD 这个词就逃跑并躲起来，而另一些程序员则靠它生存。只需拥有一个好的测试套件，您就可以实现 TDD 的一些有益效果。这里没有正确的答案。这完全取决于您和您的团队对您的方法是否满意。
根据经验，我使用 TDD 来解决需要更清晰的复杂、非结构化问题。在评估一种方法或比较多种方法时，我发现预先定义问题陈述和边界很有帮助。它有助于明确您的功能需要处理的需求和边缘情况。如果案例数量太多，则表明您的程序可能做了太多事情，也许是时候将其拆分为更小的单元了。如果需求很简单，我会跳过 TDD，稍后添加测试。
关于这个话题有很多噪音，而且很容易迷失方向。如果我能给你一些临别建议的话：不要太担心 TDD 本身，而要关注基本原则。这一切都是为了编写干净、易于理解、可维护的代码。 TDD 是程序员工具带中的一项有用技能。随着时间的推移，您会对何时应用此方法产生直觉。
感谢您的阅读，请在评论部分告诉我们您的想法。
The above is the detailed content of Practical test-driven development methods. For more information, please follow other related articles on the PHP Chinese website!
![[Web front-end] Node.js quick start](https://img.php.cn/upload/course/000/000/067/662b5d34ba7c0227.png)
