Testing with Playwright Since 2.5
Playwright is a Node library to automate the Chromium (opens new window), WebKit (opens new window) and Firefox (opens new window) browsers with a single API. It enables cross-browser web automation that is ever-green, capable, reliable and fast.
Playwright was built similarly to Puppeteer (opens new window), using its API and so is very different in usage. However, Playwright has cross browser support with better design for test automaiton.
Take a look at a sample test:
I.amOnPage('https://github.com'); I.click('Sign in', '//html/body/div[1]/header'); I.see('Sign in to GitHub', 'h1'); I.fillField('Username or email address', '[email protected]'); I.fillField('Password', '123456'); I.click('Sign in'); I.see('Incorrect username or password.', '.flash-error');
It's readable and simple and working using Playwright API!
Setup
To start you need CodeceptJS with Playwright packages installed
npm install codeceptjs playwright --save
Or see alternative installation options (opens new window)
If you already have CodeceptJS project, just install
playwright
package and enable a helper it in config.
And a basic project initialized
npx codeceptjs init
You will be asked for a Helper to use, you should select Playwright and provide url of a website you are testing.
Configuring
Make sure Playwright
helper is enabled in codecept.conf.js
config:
{ // .. helpers: { Playwright: { url: "http://localhost", show: true, browser: 'chromium' } } // .. }
Turn off the
show
option if you want to run test in headless mode. If you don't specify the browser here,chromium
will be used. Possible browsers are:chromium
,firefox
andwebkit
Playwright uses different strategies to detect if a page is loaded. In configuration use waitForNavigation
option for that:
When to consider navigation succeeded, defaults to load
. Given an array of event strings, navigation is considered to be successful after all events have been fired. Events can be either:
-
load
- consider navigation to be finished when the load event is fired. -
domcontentloaded
- consider navigation to be finished when the DOMContentLoaded event is fired. -
networkidle
- consider navigation to be finished when there are no network connections for at least 500 ms.
helpers: { Playwright: { url: "http://localhost", show: true, browser: 'chromium', waitForNavigation: "networkidle0" } }
When a test runs faster than application it is recommended to increase waitForAction
config value. It will wait for a small amount of time (100ms) by default after each user action is taken.
▶ More options are listed in helper reference (opens new window).
Writing Tests
Additional CodeceptJS tests should be created with gt
command:
npx codeceptjs gt
As an example we will use ToDoMvc
app for testing.
Actions
Tests consist with a scenario of user's action taken on a page. The most widely used ones are:
-
amOnPage
- to open a webpage (accepts relative or absolute url) -
click
- to locate a button or link and click on it -
fillField
- to enter a text inside a field -
selectOption
,checkOption
- to interact with a form -
wait*
to wait for some parts of page to be fully rendered (important for testing SPA) -
grab*
to get values from page sources -
see
,dontSee
- to check for a text on a page -
seeElement
,dontSeeElement
- to check for elements on a page
ℹ All actions are listed in Playwright helper reference (opens new window).*
All actions which interact with elements can use CSS or XPath locators (opens new window). Actions like click
or fillField
can locate elements by their name or value on a page:
// search for link or button I.click('Login'); // locate field by its label I.fillField('Name', 'Miles'); // we can use input name I.fillField('user[email]','[email protected]');
You can also specify the exact locator type with strict locators:
I.click({css: 'button.red'}); I.fillField({name: 'user[email]'},'[email protected]'); I.seeElement({xpath: '//body/header'});
Interactive Pause
It's easy to start writing a test if you use interactive pause. Just open a web page and pause execution.
Feature('Sample Test'); Scenario('open my website', ({ I }) => { I.amOnPage('http://todomvc.com/examples/react/'); pause(); });
This is just enough to run a test, open a browser, and think what to do next to write a test case.
When you execute such test with codeceptjs run
command you may see the browser is started
npx codeceptjs run --steps
After a page is opened a full control of a browser is given to a terminal. Type in different commands such as click
, see
, fillField
to write the test. A successful commands will be saved to ./output/cli-history
file and can be copied into a test.
A complete ToDo-MVC test may look like:
Feature('ToDo'); Scenario('create todo item', ({ I }) => { I.amOnPage('http://todomvc.com/examples/react/'); I.dontSeeElement('.todo-count'); I.fillField('What needs to be done?', 'Write a guide'); I.pressKey('Enter'); I.see('Write a guide', '.todo-list'); I.see('1 item left', '.todo-count'); });
Grabbers
If you need to get element's value inside a test you can use grab*
methods. They should be used with await
operator inside async
function:
const assert = require('assert'); Scenario('get value of current tasks', async ({ I }) => { I.createTodo('do 1'); I.createTodo('do 2'); let numTodos = await I.grabTextFrom('.todo-count strong'); assert.equal(2, numTodos); });
Within
In case some actions should be taken inside one element (a container or modal window or iframe) you can use within
block to narrow the scope. Please take a note that you can't use within inside another within in Playwright helper:
within('.todoapp', () => { I.createTodo('my new item'); I.see('1 item left', '.todo-count'); I.click('.todo-list input.toggle'); }); I.see('0 items left', '.todo-count');
CodeceptJS allows you to implement custom actions like I.createTodo
or use PageObjects. Learn how to improve your tests in PageObjects (opens new window) guide.
Multi Session Testing
TO launch additional browser context (or incognito window) use session
command.
Scenario('I try to open this site as anonymous user', () => { I.amOnPage('/'); I.dontSee('Agree to cookies'); session('anonymous user', () => { I.amOnPage('/'); I.see('Agree to cookies'); }); })
ℹ Learn more about multi-session testing
Device Emulation
Playwright can emulate browsers of mobile devices. Instead of paying for expensive devices for mobile tests you can adjust Playwright settings so it could emulate mobile browsers on iPhone, Samsung Galaxy, etc.
Device emulation can be enabled in CodeceptJS globally in a config or per session.
Playwright contains a list of predefined devices (opens new window) to emulate, for instance this is how you can enable iPhone 6 emulation for all tests:
const { devices } = require('playwright'); helpers: { Playwright: { // regular config goes here emulate: devices['iPhone 6'], } }
To adjust browser settings you can pass custom options (opens new window)
helpers: { Playwright: { // regular config goes here // put on mobile device emulate: { isMobile: true, deviceScaleFactor: 2 } } }
To enable device emulation for a specific test, create an additional browser session and pass in config as a second parameter:
const { devices } = require('playwright'); Scenario('website looks nice on iPhone', () => { session('mobile user', devices['iPhone 6'], () => { I.amOnPage('/'); I.see('Hello, iPhone user!') }) });
Configuring CI
GitHub Actions
Playwright can be added to GitHub Actions using official action (opens new window). Use it before starting CodeceptJS tests to install all dependencies. It is important to run tests in headless mode (otherwise you will need to enable xvfb to emulate desktop (opens new window)).
# from workflows/tests.yml - uses: microsoft/playwright-github-action@v1 - name: run CodeceptJS tests run: npx codeceptjs run
Accessing Playwright API
To get Playwright API (opens new window) inside a test use I.usePlaywrightTo
method with a callback. To keep test readable provide a description of a callback inside the first parameter.
I.usePlaywrightTo('emulate offline mode', async ({ browser, context, page }) => { // use browser, page, context objects inside this function await context.setOffline(true); });
Playwright commands are asynchronous so a callback function must be async.
A Playwright helper is passed as argument for callback, so you can combine Playwrigth API with CodeceptJS API:
I.usePlaywrightTo('emulate offline mode', async (Playwright) => { // access internal objects browser, page, context of helper await Playwright.context.setOffline(true); // call a method of helper, await is required here await Playwright.click('Reload'); });
Extending Helper
To create custom I.*
commands using Playwright API you need to create a custom helper.
Start with creating an MyPlaywright
helper using generate:helper
or gh
command:
npx codeceptjs gh
Then inside a Helper you can access Playwright
helper of CodeceptJS. Let's say you want to create I.grabDimensionsOfCurrentPage
action. In this case you need to call evaluate
method of page
object
// inside a MyPlaywright helper async grabDimensionsOfCurrentPage() { const { page } = this.helpers.Playwright; await page.goto('https://www.example.com/'); return page.evaluate(() => { return { width: document.documentElement.clientWidth, height: document.documentElement.clientHeight, deviceScaleFactor: window.devicePixelRatio } }) }
The same way you can also access browser
object to implement more actions or handle events. For instance, you want to set the permissions, you can approach it with:
// inside a MyPlaywright helper async setPermissions() { const { browser } = this.helpers.Playwright; const context = browser.defaultContext() return context.setPermissions('https://html5demos.com', ['geolocation']); }
© 2015 DavertMik <[email protected]> (http://codegyre.com)
Licensed under the MIT License.
https://codecept.io/playwright/