Question

我创建了一个轮询服务，该服务以递归方式调用api，并且在满足特定条件的情况下，如果api成功，将继续进行轮询。

/**
   * start a timer with the interval specified by the user || default interval
   * we are using setTimeout and not setinterval because a slow back end server might take more time than our interval time and that would lead to
   * a queue of ajax requests with no response at all.
   * -----------------------------------------
   * This function would call the api first time and only on the success response of the api we would poll again after the interval
   */
  runPolling() {
    const { url, onSuccess, onFailure, interval } = this.config;
    const _this = this;
    this.poll = setTimeout(() => {
      /* onSuccess would be handled by the user of service which would either return true or false
      * true - This means we need to continue polling
      * false - This means we need to stop polling
      */
      api
        .request(url)
        .then(response => {
          console.log('I was called', response);
          onSuccess(response);
        })
        .then(continuePolling => {
          _this.isPolling && continuePolling ? _this.runPolling() : _this.stopPolling();
        })
        .catch(error => {
          if (_this.config.shouldRetry && _this.config.retryCount > 0) {
            onFailure && onFailure(error);
            _this.config.retryCount--;
            _this.runPolling();
          } else {
            onFailure && onFailure(error);
            _this.stopPolling();
          }
        });
    }, interval);
  }

在尝试为其编写测试用例时，我不确定如何模拟假计时器和axios api响应。

这是我到目前为止所拥有的

import PollingService from '../PollingService';
import { statusAwaitingProduct } from '@src/__mock_data__/getSessionStatus';
import mockAxios from 'axios';

describe('timer events for runPoll', () => {
    let PollingObject,
    pollingInterval = 3000,
    url = '/session/status',
    onSuccess = jest.fn(() => {
      return false;
    });
    beforeAll(() => {
      PollingObject = new PollingService({
        url: url,
        interval: pollingInterval,
        onSuccess: onSuccess
      });
    });
    beforeEach(() => {
      jest.useFakeTimers();
    });
    test('runPolling should be called recursively when onSuccess returns true', async () => {
      expect.assertions(1);
      const mockedRunPolling = jest.spyOn(PollingObject, 'runPolling');
      const mockedOnSuccess = jest.spyOn(PollingObject.config, 'onSuccess');
      mockAxios.request.mockImplementation(
        () =>
          new Promise(resolve => {
            resolve(statusAwaitingProduct);
          })
      );

      PollingObject.startPolling();
      expect(mockedRunPolling).toHaveBeenCalledTimes(1);
      expect(setTimeout).toHaveBeenCalledTimes(1);
      expect(mockAxios.request).toHaveBeenCalledTimes(0);
      expect(setTimeout).toHaveBeenLastCalledWith(expect.any(Function), pollingInterval);

      jest.runAllTimers();
      expect(mockAxios.request).toHaveBeenCalledTimes(1);
      expect(mockedOnSuccess).toHaveBeenCalledTimes(1);
      expect(PollingObject.isPolling).toBeTruthy();
      expect(mockedRunPolling).toHaveBeenCalledTimes(2);
    });
  });
});

即使调用了mockedOnsuccess，但只是期望调用失败，因为它被调用了0次而不是被调用了1次。

有人可以帮忙吗？谢谢

Answer 1

问题

您的测试也可能有其他问题，但是我将解决您所提出的关于expect(mockedOnSuccess).toHaveBeenCalledTimes(1);失败0 times的特定问题：

jest.runAllTimers将同步运行任何待处理的计时器回调，直到没有剩余的为止。这将执行在setTimeout中用runPolling安排的匿名函数。当匿名函数执行时，它将调用api.request(url)，但是这就是所有发生的事情。匿名函数中的其他所有内容都包含在get queued in the PromiseJobs Jobs Queue introduced with ES6的then回调中。到jest.runAllTimers返回时，这些作业都没有执行，测试将继续进行。

expect(mockAxios.request).toHaveBeenCalledTimes(1);然后由于执行api.request(url)而通过。

expect(mockedOnSuccess).toHaveBeenCalledTimes(1);然后失败，因为可以调用它的then回调仍在PromiseJobs队列中并且尚未执行。

解决方案

解决方案是在断言PromiseJobs被调用之前，确保mockedOnSuccess中排队的作业有运行的机会。

幸运的是，很容易允许PromiseJobs中的所有待处理作业在async的{{1}}测试中运行，只需调用Jest。这实际上将其余测试排在await Promise.resolve();的末尾，并允许队列中的所有未决作业先执行：

PromiseJobs

请注意，理想情况下，异步函数将返回Promise，然后测试可以等待。在您的情况下，您已安排了test('runPolling should be called recursively when onSuccess returns true', async () => { ... jest.runAllTimers(); await Promise.resolve(); // allow any pending jobs in PromiseJobs to execute expect(mockAxios.request).toHaveBeenCalledTimes(1); expect(mockedOnSuccess).toHaveBeenCalledTimes(1); // SUCCESS ... }进行的回调，因此无法返回Promise来等待测试。

还要注意，您有多个链接的setTimeout回调，因此您可能需要在测试期间多次then中等待挂起的作业。

有关虚假计时器和承诺如何互动here的更多详细信息。

用玩笑和假计时器测试递归轮询功能

1 个答案:

问题

解决方案