The Protractor Page Object Pattern

The Page Object Model pattern was a thing long before Protractor and even Testacular were ever created. POM is a design pattern for Selenium WebDriver, and it is basically a way to structure your e2e test code so that it is more maintainable, readable, and reusable. In Protractor, we code in JavaScript, but what we are doing is very similar to the code of various WebDriver languages; selecting things from the DOM. This DOM selection code comes up quite a lot, and the POM allows you to define variables in another file so they would only need to be changed in at most one place. Let's take a look at an example of protractor tests without the POM:

By the way, you can also find this code on github here.

noPom.spec.js

(function() {
  'use strict';
  var page;
  describe('Button Click Example', function () {

    beforeEach(function() {
      browser.get('http://localhost:3000/');
    })

    it('should find the \"calculator-title\" element', function() {
      expect(element(by.css('.page-title')).isPresent()).toBeTruthy();
    });

    it('should not find elements that don\'t exist', function() {
      browser.get('http://localhost:3000/');

      expect(element(by.css('nonexistent-element')).isPresent()).toBeFalsy();

    });


    it('should not have clicked text visible to start', function() {
      expect(element(by.id('clicked-text')).isPresent()).toBeTruthy();
      expect(element(by.id('clicked-text')).isDisplayed()).toBeFalsy();

    });

    it('should show the text after the button is clicked', function () {
      element(by.css('.my-button')).click();
      expect(element(by.id('clicked-text')).isDisplayed()).toBeTruthy();
    });

    it ('should show correct text after button is pressed once.', function() {
      element(by.css('.my-button')).click();
      expect(element(by.id('clicked-text')).getText()).toEqual('The button has been clicked 1 time.');
    });

    it('should show the correct text after button is clicked 5 times.', function () {
      element(by.css('.my-button')).click();
      element(by.css('.my-button')).click();
      element(by.css('.my-button')).click();
      element(by.css('.my-button')).click();
      element(by.css('.my-button')).click();
      expect(element(by.id('clicked-text')).getText()).toEqual('The button has been clicked 5 times.');
    })

  })


})();

You ca see that in this code there is a whole lot of "element(by." code, and then making any changing to your scss files may mean one or more epic find/replace's need to be done. It's just straight up duplication really, and changing this code to use POM allows you to refactor it away. 

---

What Exactly is POM?

In the page object model pattern, you still have a .spec.js file that is a describe full of it's with some beforeEach's just like any other test, but here you now have an addition .po.js file that sets common DOM selection to variable of one a single "page object". Since Protractor runs on node.js we can make use of require to pull in this page object to our .spec.js class in the beforeEach function. Here's how it might look:

pomExample.po.js

pomExample.spec.js

(function () {
     'use strict';
     var page;
     describe('Button Click Example', function () {

          beforeEach(function () {
               browser.get('http://localhost:3000/');
               page = require('./main.po');
          })

          it('should find the \"calculator-title\" element', function () {
               expect(page.pageTitle.isPresent()).toBeTruthy();
          });

          it('should not find elements that don\'t exist', function () {
               browser.get('http://localhost:3000/');

               expect(page.nonexistentElement.isPresent()).toBeFalsy();

          });

          it('should not have clicked text visible to start', function () {
               expect(page.clickedText.isPresent()).toBeTruthy();
               expect(page.clickedText.isDisplayed()).toBeFalsy();

          });

          it('should show the text after the button is clicked', function () {
               page.myButton.click();
               expect(page.clickedText.isDisplayed()).toBeTruthy();
          });

          it('should show correct text after button is pressed once.', function () {
               page.myButton.click();
               expect(page.clickedText.getText()).toEqual('The button has been clicked 1 time.');
          });

          it('should show the correct text after button is clicked 5 times.', function () {
               page.myButton.click();
               page.myButton.click();
               page.myButton.click();
               page.myButton.click();
               page.myButton.click();
               expect(page.clickedText.getText()).toEqual('The button has been clicked 5 times.');
          })

     })

})();

Withe POM setup changes to the css require only one change (per style changed) in the .po.js file and no changes to the spec file. It also "lightens up" the spec file by taking out the DOM selection code with all its annoying parentheses. Learn to love the page object model for Protractor testing. It'll serve you well!

---

Bonus Video

Here's a nice video that explains the Page Object Model pretty well. Unfortunately, it's actually about selenium WebDriver and not Protractor so the presenter uses all Java examples instead of Angular Javascript. However, the concepts are the same, and it's not all that hard to see how most of his WebDriver automation presentation applies to your Angular e2e tests as well.