如何在Protractor中计算各步骤执行耗时并设置超时失败规则？

Hey there! Let's break down how to solve your problem of measuring test step execution time (like how long it takes for a page to load and be ready for the next step after clicking a link) and failing the test if it exceeds 2 seconds.

First off, I notice you tried protractor-perf—but that tool is built for tracking browser performance metrics (like frame rates, style/JS execution times) rather than measuring the exact duration of a single user action + page readiness. That's why the data you got felt unhelpful for your specific need. Let's use a more direct approach instead.

This method lets you precisely measure the time between starting an action and the page being fully ready for the next step.

Example Code #

describe('Test Step Duration Check', () => {
  it('Measure time to load after clicking a link', async () => {
    // 1. Capture start timestamp right before the action
    const startTime = Date.now();

    // 2. Perform your test action (e.g., click the target link)
    await element(by.linkText('Your Target Link')).click();

    // 3. Wait for the page to be truly ready (adjust this to your page's key indicator)
    // Use ExpectedConditions to wait for a critical element that signals "ready"
    await browser.wait(
      protractor.ExpectedConditions.elementToBeClickable(element(by.id('critical-element-after-load'))),
      10000 // Max wait time to avoid hanging indefinitely
    );

    // 4. Calculate elapsed time
    const elapsedTime = Date.now() - startTime;
    console.log(`Time taken after click: ${elapsedTime}ms`);

    // 5. Fail the test if it takes over 2 seconds
    expect(elapsedTime).toBeLessThanOrEqual(2000);
  });
});

Key Notes #

• Don't skip the wait step: Clicking a link doesn't mean the page is ready immediately. Always wait for a clear signal (like a key element being visible/clickable, or an API request finishing) to ensure you're measuring the actual "ready" time.
• Adjust ExpectedConditions: Use visibilityOf, presenceOf, or elementToBeClickable based on what defines "ready" for your page. elementToBeClickable is often the best choice for "can execute next step".

The data you got from protractor-perf relates to browser performance, not action duration. Here's a quick breakdown of what those metrics mean:

• Styles/Javascript: Time spent processing stylesheets and JavaScript (in ms—0 here likely means your action didn't trigger heavy style/JS work)
• numAnimationFrames/numFramesSentToScreen: Total animation frames generated by the browser vs. those actually rendered to the screen
• droppedFrameCount: Number of frames that failed to render (high numbers mean page lag, but this doesn't directly tell you how long the action took)

If you want to use protractor-perf to complement timing checks (e.g., ensure the page doesn't lag during loading), you can do this:

const PerfRunner = require('protractor-perf');

describe('Performance + Timing Check', () => {
  let perfRunner;

  beforeAll(() => {
    perfRunner = new PerfRunner(protractor);
  });

  it('Check load time and frame rate after click', async () => {
    const startTime = Date.now();
    
    await perfRunner.start();
    await element(by.linkText('Your Target Link')).click();
    await browser.wait(protractor.ExpectedConditions.elementToBeClickable(element(by.id('critical-element'))), 10000);
    
    const elapsedTime = Date.now() - startTime;
    await perfRunner.stop();

    // Check duration first
    expect(elapsedTime).toBeLessThanOrEqual(2000);

    // Optional: Check for excessive frame drops (e.g., less than 10% drop rate)
    const frameDropRate = (perfRunner.getStats().droppedFrameCount / perfRunner.getStats().numFramesSentToScreen) * 100;
    expect(frameDropRate).toBeLessThan(10);
  });
});

• Wait for async requests: If your page loads data via API calls, use browser.executeAsyncScript to wait for all XHR requests to finish before ending the timer.
• Average multiple runs: To avoid flakiness from one-off slowdowns, run the test 3-5 times and check the average elapsed time instead of a single run.

内容的提问来源于stack exchange，提问作者ASB