Easily test your Electron apps using ChromeDriver and WebdriverIO.

For given versions of Electron you must depend on a very specific version range of Spectron. Below is a version mapping table between Spectron version and Electron version.

Learn more from this presentation.

🚨 Upgrading from 1.x to 2.x/3.x? Read the changelog.

npm install --save-dev spectron

Spectron works with any testing framework but the following example uses mocha:

To get up and running from your command line:

# Install mocha locally as a dev dependency.
npm i mocha -D

# From the project root, create a folder called test, in that directory, create a file called 'spec.js'
touch test/spec.js

# Change directory to test
cd test

Then simply include the following in your first spec.js.

const Application = require('spectron').Application
const assert = require('assert')
const electronPath = require('electron') // Require Electron from the binaries included in node_modules.
const path = require('path')

describe('Application launch', function () {
  this.timeout(10000)

  beforeEach(function () {
    this.app = new Application({
      // Your electron path can be any binary
      // i.e for OSX an example path could be '/Applications/MyApp.app/Contents/MacOS/MyApp'
      // But for the sake of the example we fetch it from our node_modules.
      path: electronPath,

      // Assuming you have the following directory structure

      //  |__ my project
      //     |__ ...
      //     |__ main.js
      //     |__ package.json
      //     |__ index.html
      //     |__ ...
      //     |__ test
      //        |__ spec.js  <- You are here! ~ Well you should be.

      // The following line tells spectron to look and use the main.js file
      // and the package.json located 1 level above.
      args: [path.join(__dirname, '..')]
    })
    return this.app.start()
  })

  afterEach(function () {
    if (this.app && this.app.isRunning()) {
      return this.app.stop()
    }
  })

  it('shows an initial window', function () {
    return this.app.client.getWindowCount().then(function (count) {
      assert.equal(count, 1)
      // Please note that getWindowCount() will return 2 if `dev tools` are opened.
      // assert.equal(count, 2)
    })
  })
})

Create an npm task in your package.json file

"scripts": {
  "test": "mocha"
}

And from the root of your project, in your command-line simply run:

npm test

By default, mocha searches for a folder with the name test ( which we created before ). For more information on how to configure mocha, please visit mocha.

As stated in issue #19, Spectron will not be able to start if your Electron app is launched using the remote-debugging-port command-line switch (i.e. app.commandLine.appendSwitch('remote-debugging-port', <debugging-port-number>);). Please make sure to include the necessary logic in your app's code to disable the switch during tests.

Spectron exports an Application class that when configured, can start and stop your Electron application.

Create a new application with the following options:

• path - Required. String path to the Electron application executable to launch. Note: If you want to invoke electron directly with your app's main script then you should specify path as electron via electron-prebuilt and specify your app's main script path as the first argument in the args array.
• args - Array of arguments to pass to the Electron application.
• chromeDriverArgs - Array of arguments to pass to ChromeDriver. See here for details on the Chrome arguments.
• cwd- String path to the working directory to use for the launched application. Defaults to process.cwd().
• env - Object of additional environment variables to set in the launched application.
• host - String host name of the launched chromedriver process. Defaults to 'localhost'.
• port - Number port of the launched chromedriver process. Defaults to 9515.
• nodePath - String path to a node executable to launch ChromeDriver with. Defaults to process.execPath.
• connectionRetryCount - Number of retry attempts to make when connecting to ChromeDriver. Defaults to 10 attempts.
• connectionRetryTimeout - Number in milliseconds to wait for connections to ChromeDriver to be made. Defaults to 30000 milliseconds.
• quitTimeout - Number in milliseconds to wait for application quitting. Defaults to 1000 milliseconds.
• requireName - Custom property name to use when requiring modules. Defaults to require. This should only be used if your application deletes the main window.require function and assigns it to another property name on window.
• startTimeout - Number in milliseconds to wait for ChromeDriver to start. Defaults to 5000 milliseconds.
• waitTimeout - Number in milliseconds to wait for calls like waitUntilTextExists and waitUntilWindowLoaded to complete. Defaults to 5000 milliseconds.
• debuggerAddress - String address of a Chrome debugger server to connect to.
• chromeDriverLogPath - String path to file to store ChromeDriver logs in. Setting this option enables --verbose logging when starting ChromeDriver.
• webdriverLogPath - String path to a directory where Webdriver will write logs to. Setting this option enables verbose logging from Webdriver.
• webdriverOptions - Object of additional options for Webdriver

The Electron helpers provided by Spectron require accessing the core Electron APIs in the renderer processes of your application. So, either your Electron application has nodeIntegration set to true or you'll need to expose a require window global to Spectron so it can access the core Electron APIs.

You can do this by adding a preload script that does the following:

if (process.env.NODE_ENV === 'test') {
  window.electronRequire = require
}

Then create the Spectron Application with the requireName option set to 'electronRequire' and then runs your tests via NODE_ENV=test npm test.

Note: This is only required if your tests are accessing any Electron APIs. You don't need to do this if you are only accessing the helpers on the client property which do not require Node integration.

Spectron uses WebdriverIO and exposes the managed client property on the created Application instances.

The client API is WebdriverIO's browser object. Documentation can be found here.

Several additional commands are provided specific to Electron.

All the commands return a Promise.

So if you wanted to get the text of an element you would do:

app.client.getText('#error-alert').then(function (errorText) {
  console.log('The #error-alert text content is ' + errorText)
})

The electron property is your gateway to accessing the full Electron API.

Each Electron module is exposed as a property on the electron property so you can think of it as an alias for require('electron') from within your app.

So if you wanted to access the clipboard API in your tests you would do:

app.electron.clipboard.writeText('pasta')
   .electron.clipboard.readText().then(function (clipboardText) {
     console.log('The clipboard text is ' + clipboardText)
   })

The browserWindow property is an alias for require('electron').remote.getCurrentWindow().

It provides you access to the current BrowserWindow and contains all the APIs.

So if you wanted to check if the current window is visible in your tests you would do:

app.browserWindow.isVisible().then(function (visible) {
  console.log('window is visible? ' + visible)
})

It is named browserWindow instead of window so that it doesn't collide with the WebDriver command of that name.

The async capturePage API is supported but instead of taking a callback it returns a Promise that resolves to a Buffer that is the image data of screenshot.

app.browserWindow.capturePage().then(function (imageBuffer) {
  fs.writeFile('page.png', imageBuffer)
})

The webContents property is an alias for require('electron').remote.getCurrentWebContents().

It provides you access to the WebContents for the current window and contains all the APIs.

So if you wanted to check if the current window is loading in your tests you would do:

app.webContents.isLoading().then(function (visible) {
  console.log('window is loading? ' + visible)
})

The async savePage API is supported but instead of taking a callback it returns a Promise that will raise any errors and resolve to undefined when complete.

app.webContents.savePage('/Users/kevin/page.html', 'HTMLComplete')
  .then(function () {
    console.log('page saved')
  }).catch(function (error) {
    console.error('saving page failed', error.message)
  })

The async executeJavaScript API is supported but instead of taking a callback it returns a Promise that will resolve with the result of the last statement of the script.

app.webContents.executeJavaScript('1 + 2')
  .then(function (result) {
    console.log(result) // prints 3
  })

The mainProcess property is an alias for require('electron').remote.process.

It provides you access to the main process's process global.

So if you wanted to get the argv for the main process in your tests you would do:

app.mainProcess.argv().then(function (argv) {
  console.log('main process args: ' + argv)
})

Properties on the process are exposed as functions that return promises so make sure to call mainProcess.env().then(...) instead of mainProcess.env.then(...).

The rendererProcess property is an alias for global.process.

It provides you access to the renderer process's process global.

So if you wanted to get the environment variables for the renderer process in your tests you would do:

app.rendererProcess.env().then(function (env) {
  console.log('renderer process env variables: ' + env)
})

Starts the application. Returns a Promise that will be resolved when the application is ready to use. You should always wait for start to complete before running any commands.

Stops the application. Returns a Promise that will be resolved once the application has stopped.

Stops the application and then starts it. Returns a Promise that will be resolved once the application has started again.

Checks to determine if the application is running or not.

Returns a Boolean.

Get all the configured options passed to the new Application() constructor. This will include the default options values currently being used.

Returns an Object.

Gets the console log output from the main process. The logs are cleared after they are returned.

Returns a Promise that resolves to an array of string log messages

app.client.getMainProcessLogs().then(function (logs) {
  logs.forEach(function (log) {
    console.log(log)
  })
})

Gets the console log output from the render process. The logs are cleared after they are returned.

Returns a Promise that resolves to an array of log objects.

app.client.getRenderProcessLogs().then(function (logs) {
  logs.forEach(function (log) {
    console.log(log.message)
    console.log(log.source)
    console.log(log.level)
  })
})

Get the selected text in the current window.

app.client.getSelectedText().then(function (selectedText) {
  console.log(selectedText)
})

Gets the number of open windows. <webview> tags are also counted as separate windows.

app.client.getWindowCount().then(function (count) {
  console.log(count)
})

Waits until the element matching the given selector contains the given text. Takes an optional timeout in milliseconds that defaults to 5000.

app.client.waitUntilTextExists('#message', 'Success', 10000)

Wait until the window is no longer loading. Takes an optional timeout in milliseconds that defaults to 5000.

app.client.waitUntilWindowLoaded(10000)

Focus a window using its index from the windowHandles() array. <webview> tags can also be focused as a separate window.

app.client.windowByIndex(1)

Focus a window using its URL or title.

// switch via url match
app.client.switchWindow('google.com')

// switch via title match
app.client.switchWindow('Next-gen WebDriver test framework')

Spectron bundles the Accessibility Developer Tools provided by Google and adds support for auditing each window and <webview> tag in your application.

Run an accessibility audit in the focused window with the specified options.

• options - An optional Object with the following keys:
  • ignoreWarnings - true to ignore failures with a severity of 'Warning' and only include failures with a severity of 'Severe'. Defaults to false.
  • ignoreRules - Array of String rule code values such as AX_COLOR_01 to ignore failures for. The full list is available here.

Returns an audit Object with the following properties:

• message - A detailed String message about the results
• failed - A Boolean, false when the audit has failures
• results - An array of detail objects for each failed rule. Each object in the array has the following properties:
  • code - A unique String accessibility rule identifier
  • elements - An Array of Strings representing the selector path of each HTML element that failed the rule
  • message - A String message about the failed rule
  • severity - 'Warning' or 'Severe'
  • url - A String URL providing more details about the failed rule

app.client.auditAccessibility().then(function (audit) {
  if (audit.failed) {
    console.error(audit.message)
  }
})

See https://github.com/GoogleChrome/accessibility-developer-tools/wiki/Audit-Rules for more details about the audit rules.

If you are using a <webview> tag in your app and want to audit both the outer page and the <webview>'s page then you will need to do the following:

// Focus main page and audit it
app.client.windowByIndex(0).then(function() {
  app.client.auditAccessibility().then(function (audit) {
    if (audit.failed) {
      console.error('Main page failed audit')
      console.error(audit.message)
    }

    //Focus <webview> tag and audit it
    app.client.windowByIndex(1).then(function() {
      app.client.auditAccessibility().then(function (audit) {
        if (audit.failed) {
          console.error('<webview> page failed audit')
          console.error(audit.message)
        }
      })
    })
  })
})

You will want to add the following to your .travis.yml file when building on Linux:

before_script:
  - "export DISPLAY=:99.0"
  - "sh -e /etc/init.d/xvfb start"
  - sleep 3 # give xvfb some time to start

Check out Spectron's .travis.yml file for a production example.

You will want to add the following to your appveyor.yml file:

os: unstable

Check out Spectron's appveyor.yml file for a production example.

WebdriverIO is promise-based and so it pairs really well with the Chai as Promised library that builds on top of Chai.

Using these together allows you to chain assertions together and have fewer callback blocks. See below for a simple example:

npm install --save-dev chai
npm install --save-dev chai-as-promised

const Application = require('spectron').Application
const chai = require('chai')
const chaiAsPromised = require('chai-as-promised')
const electronPath = require('electron')
const path = require('path')

chai.should()
chai.use(chaiAsPromised)

describe('Application launch', function () {
  this.timeout(10000);

  beforeEach(function () {
    this.app = new Application({
      path: electronPath,
      args: [path.join(__dirname, '..')]
    })
    return this.app.start()
  })

  beforeEach(function () {
    chaiAsPromised.transferPromiseness = this.app.transferPromiseness
  })

  afterEach(function () {
    if (this.app && this.app.isRunning()) {
      return this.app.stop()
    }
  })

  it('opens a window', function () {
    return this.app.client.waitUntilWindowLoaded()
      .getWindowCount().should.eventually.have.at.least(1)
      .browserWindow.isMinimized().should.eventually.be.false
      .browserWindow.isVisible().should.eventually.be.true
      .browserWindow.isFocused().should.eventually.be.true
      .browserWindow.getBounds().should.eventually.have.property('width').and.be.above(0)
      .browserWindow.getBounds().should.eventually.have.property('height').and.be.above(0)
  })
})

Spectron works with AVA, which allows you to write your tests in ES2015+ without doing any extra work.

import test from 'ava';
import {Application} from 'spectron';

test.beforeEach(t => {
  t.context.app = new Application({
    path: '/Applications/MyApp.app/Contents/MacOS/MyApp'
  });

  return t.context.app.start();
});

test.afterEach(t => {
  return t.context.app.stop();
});

test(t => {
  return t.context.app.client.waitUntilWindowLoaded()
    .getWindowCount().then(count => {
      t.is(count, 1);
    }).browserWindow.isMinimized().then(min => {
      t.false(min);
    }).browserWindow.isDevToolsOpened().then(opened => {
      t.false(opened);
    }).browserWindow.isVisible().then(visible => {
      t.true(visible);
    }).browserWindow.isFocused().then(focused => {
      t.true(focused);
    }).browserWindow.getBounds().then(bounds => {
      t.true(bounds.width > 0);
      t.true(bounds.height > 0);
    });
});

AVA has built-in support for async functions, which simplifies async operations:

import test from 'ava';
import {Application} from 'spectron';

test.beforeEach(async t => {
  t.context.app = new Application({
    path: '/Applications/MyApp.app/Contents/MacOS/MyApp'
  });

  await t.context.app.start();
});

test.afterEach.always(async t => {
  await t.context.app.stop();
});

test(async t => {
  const app = t.context.app;
  await app.client.waitUntilWindowLoaded();

  const win = app.browserWindow;
  t.is(await app.client.getWindowCount(), 1);
  t.false(await win.isMinimized());
  t.false(await win.isDevToolsOpened());
  t.true(await win.isVisible());
  t.true(await win.isFocused());

  const {width, height} = await win.getBounds();
  t.true(width > 0);
  t.true(height > 0);
});