Skip to content

faq.md

Latest commit

 

History

History
149 lines (117 loc) · 6.76 KB

faq.md

File metadata and controls

149 lines (117 loc) · 6.76 KB

FAQ

My tests time out in Protractor, but everything's working fine when running manually. What's up?

Protractor attempts to wait until the page is completely loaded before performing any action (such as finding an element or sending a command to an element). If your application continuously polls $timeout or $http, it will never be registered as completely loaded. You should use the $interval service for anything that polls continuously (introduced in Angular 1.2rc3). Further discussion is in issue 49.

You may also be running into a timeout because your page is slow to load or perform actions. By default, Protractor sets the timeout for actions to 11 seconds. You can change this in your config with the allScriptsTimeout options.

  exports.config = {
    // Override the timeout for webdriver to 20 seconds.
    allScriptsTimeout: 20000,
  }

Check out the debugging doc for more information.

Why both Karma and Protractor? When do I use which?

Karma is a great tool for unit testing, and Protractor is intended for end to end or integration testing. This means that small tests for the logic of your individual controllers, directives, and services should be run using Karma. Big tests in which you have a running instance of your entire application should be run using Protractor. Protractor is intended to run tests from a user's point of view - if your test could be written down as instructions for a human interacting with your application, it should be an end to end test written with Protractor.

Here's a great blog post with more info.

Angular can't be found on my page

Protractor supports angular 1.0.6/1.1.4 and higher - check that your version of Angular is upgraded.

The angular variable is expected to be available in the global context. Try opening chrome devtools or firefox and see if angular is defined.

How do I deal with my log-in page?

If your app needs log-in, there are a couple ways to deal with it. If your login page is not written with Angular, you'll need to interact with it via unwrapped webdriver, which can be accessed like browser.driver.get().

You can put your log-in code into an onPrepare function, which will be run once before any of your tests. See this example.

Which browsers are supported?

The last two major versions of Chrome, Firefox, IE, and Safari. See details at browser support.

The result of getText from an input element is always empty

This is a webdriver quirk. <input> and <textarea> elements always have empty getText values. Instead, try element.getAttribute('value').

How can I interact directly with the JavaScript running in my app?

In general, the design of WebDriver tests is to interact with the page as a user would, so it gets a little tricky if you want to interact with the JavaScript directly. You should avoid it unless you have a good reason. However, there are ways of doing it.

You can use the evaluate function on a WebElement to get the value of an Angular expression in the scope of that element. e.g.

by.css('.foo').evaluate('bar')

would return whatever {{bar}} is in the scope of the element with class 'foo'.

You can also execute arbitrary JavaScript in the browser with

browser.executeScript('your script as a string')

How can I get hold of the browser's console?

In your test:

browser.manage().logs().get('browser').then(function(browserLog) {
  console.log('log: ' + require('util').inspect(browserLog));
});

This will output logs from the browser console. Note that logs below the set logging level will be ignored. WebDriver does not currently support changing the logging level for browser logs.

See an example of using this API to fail tests if the console has errors.

How can I get screenshots of failures?

First, this is how you can take a screenshot:

browser.takeScreenshot().then(function(png) {
  var stream = fs.createWriteStream("/tmp/screenshot.png");
  stream.write(new Buffer(png, 'base64'));
  stream.end();
});

The method to take a screenshot automatically on failure would depend on the type of failure.

  • For failures of entire specs (such as timeout or an expectation within the spec failed), you can add a reporter as such:
jasmine.getEnv().addReporter(new function() {
  this.reportSpecResults = function(spec) {
    if (!spec.results().passed()) {
      //take screenshot
    }
  };
});

Note, you can also choose to take a screenshot in AfterEach. However, because Jasmine does not execute AfterEach for timeouts, those would not produce screenshots

  • For failures of individual expectations, you can override jasmine's addMatcherResult function as such:
var originalAddMatcherResult = jasmine.Spec.prototype.addMatcherResult;
jasmine.Spec.prototype.addMatcherResult = function() {
  if (!arguments[0].passed()) {
    //take screenshot
  }
  return originalAddMatcherResult.apply(this, arguments);
};

See an example of taking screenshot on spec failures.

How do I produce an XML report of my test results?

Use jasmine-reporters and add a JUnit XML Reporter. Check out this example.

How can I catch errors such as ElementNotFound?

WebDriver throws errors when commands cannot be completed - e.g. not being able to click on an element which is obscured by another element. If you need to retry these actions, try using webdriverjs-retry. If you would just like to catch the error, do so like this

elm.click().then(function() { /* passing case */}, function(err) { /* error handling here */})