Automate Headless Chrome -- start/stop Chrome instances, open & close tabs, and communicate with the target page.

Compatibility

• You must use version >= 59 of Chrome (currently that means Chrome Beta) or use Chrome Canary.
• Canary isn't supported on Linux platform.

Install

npm install chromate
npm run sample

Use

let {Chrome, Tab} = require('chromate');

// start a headless Chrome process
Chrome.start().then(chrome => {
  let tab = new Tab({
    verbose: true,
    failonerror: false
  });
  
  tab.open(targetUrl)
  .then(() => tab.evaluate('testResults'))
  .then(res => console.log) // results...
  .then(() => tab.close())
  .then(() => {
    Chrome.kill(chrome);
    process.exit(0);
  });
});

Page events

Handle events, including any chrome-remote-interface events.

new Tab(options)
 .on('ready', (tab) => console.log('tab is ready', tab.client.target.id))
 .on('load', () => console.log('page loaded'))
 .on('console', (args) => console.log('console.* called', args))
 .on('Network.requestWillBeSent', param => console.log('Getting resource', param.request.url))
 .once('Runtime.consoleAPICalled', param => console.log('Runtime.consoleAPICalled called', param))
 .open(targetUrl);

The ready event is fired once when the target client is ready (this overrides the CDP ready event). The target may fire any number of custom events.

Custom events

A target page may communicate back to the controlling process by calling console.debug(message), where message is {event, data}. This is useful for running automated tests, such as for replacing PhantomJS. See phantom-menace for example.

// useful for short messages (< 100 chars)
console.debug({
  event: 'done',
  data: JSON.stringify({foo: 1}) // must be stringify'd
});

// then, in runner process
new Tab()
  .on('done', res => console.log); // {event: 'done', data: foo:1}}

A special function __chromate(message) is injected in target page to facilitate this, so that the above can be replaced by:

// in target (useful for any length message)
if (window.__chromate) __chromate({event: 'done', data: {foo:1}});;

The format of the message is flexible, but should be sensible. If no event property is found in the message, a 'data' event is triggered.

// in target
__chromate('foo');
console.debug({a:1});

// in runner
tab.on('data', res => console.log); // 'foo' and {a:1}

Script injection

Often it's useful for the running script to inject custom JS into the target page. This can be done thorough Page.addScriptToEvaluateOnLoad() or the Runtime.evaluate() method. Two helper methods are provided: tab.evaluate() and tab.execute():

new Tab()
  .on('done', (res, tab) => {
     tab.evaluate('JSON.stringify(__coverage__)')
       .then(result => console.log(result))
       .then(() => tab.close());
  })
  .open(targetUrl);

Or execute a (named) function in target.

new Tab()
  .open(targetUrl)
  .then(tab => {
    tab.execute('getResult').then(res => console.log);
  })
  
// in target
function getResult() {
  return JSON.stringify(result);
}

tab.execute() takes additional parameters to pass as arguments to the function. If the function is expected to return a Promise, pass a {awaitPromise: true} as the last argument.

API

See API docs.

Simple CLI

Usage:

$ chromate
Usage: chromate start [<chrome flags>] | list | kill <pid> ... | killall | version | open <url> | 
    list-tabs | close <tabId> | close-tabs  [--canary | --verbose | -v]

Chrome process control

$ chromate start --window-size=800x600 --canary
42706: /Applications/Google Chrome Canary.app/Contents/MacOS/Google Chrome Canary
  --remote-debugging-port=9222 --headless --disable-gpu
  --user-data-dir=/var/folders/jl/zr54cdxs08l1djw8s4ws2t540000gn/T/chrome-PdvzfF 
  --window-size=800x600 --canary --disable-translate --disable-extensions
  --disable-background-networking --safebrowsing-disable-auto-update --disable-sync
  --metrics-recording-only --disable-default-apps --no-first-run 
  --disable-background-timer-throttling --disable-renderer-backgrounding
  --disable-device-discovery-notifications

$ chromate list
[ { pid: '42706',
    command: '/Applications/Google Chrome Canary.app/Contents/MacOS/Google Chrome Canary',
    arguments:
     [ '--remote-debugging-port=9222',
       '--headless',
       '--disable-gpu',
       '--user-data-dir=/var/folders/jl/zr54cdxs08l1djw8s4ws2t540000gn/T/chrome-PdvzfF',
       '--window-size=800x600',
       '--canary',
       '--disable-translate',
       '--disable-extensions',
       '--disable-background-networking',
       '--safebrowsing-disable-auto-update',
       '--disable-sync',
       '--metrics-recording-only',
       '--disable-default-apps',
       '--no-first-run',
       '--disable-background-timer-throttling',
       '--disable-renderer-backgrounding',
       '--disable-device-discovery-notifications' ],
    ppid: '1' } ]
    
$ chromate killall 
2

killall returns the number of processes (including sub-processes) killed.

For list of Chrome Headless flags, see here. Of course, any Chrome flag can be specified.

To use a custom Chrome path and/or port, use:

$ CHROME_BIN=/path/to/chrome CHROME_PORT=9226 chromate start

Chrome tab control

$ chromate open https://github.com
{ description: '',
  devtoolsFrontendUrl: '/devtools/inspector.html?ws=localhost:9222/devtools/page/904ddfa4-2344-4e45-a625-8261ffbee251',
  id: '904ddfa4-2344-4e45-a625-8261ffbee251',
  title: '',
  type: 'page',
  url: 'about:blank',
  webSocketDebuggerUrl: 'ws://localhost:9222/devtools/page/904ddfa4-2344-4e45-a625-8261ffbee251' }

$ chromate list-tabs
[ { description: '',
    devtoolsFrontendUrl: '/devtools/inspector.html?ws=localhost:9222/devtools/page/e97b0e1e-1fb5-41be-83d2-bdb9fbc406bc',
    id: 'e97b0e1e-1fb5-41be-83d2-bdb9fbc406bc',
    title: 'The world's leading software development platform · GitHub',
    type: 'page',
    url: 'https://github.com/',
    webSocketDebuggerUrl: 'ws://localhost:9222/devtools/page/e97b0e1e-1fb5-41be-83d2-bdb9fbc406bc' },
  { description: '',
    devtoolsFrontendUrl: '/devtools/inspector.html?ws=localhost:9222/devtools/page/e4c16358-7670-4deb-8b2e-29f802e599a3',
    id: 'e4c16358-7670-4deb-8b2e-29f802e599a3',
    title: 'about:blank',
    type: 'page',
    url: 'about:blank',
    webSocketDebuggerUrl: 'ws://localhost:9222/devtools/page/e4c16358-7670-4deb-8b2e-29f802e599a3' } ]

$ chromate close-tabs
2

Test

# npm i -g mocha  (if you don't already have it)
npm test

Thanks and references

• This library is based on the awesome work of chrome-remote-interface
• Idea for Chrome.ready() taken from Lighthouse
• DevTools Protocol Viewer complete reference.
• Getting Started with Headless Chrome

Change log

• v0.3.5 - Readme update.
• v0.3.4 - Added Chrome.settings.userDataDir. By default a temporary user data dir is used and cleaned up.
• v0.3.3 - fixed 'ps-node' reference
• v0.3.2 - fixed internal print() method
• v0.3.1 - added events 'abort', 'exception', and 'console.*'. Export chromate.version.
• v0.3.0 - tab.open(url) takes url rather than constructor. tab.execute can take a local function. Use ps-moos with fix for spaces in path.
• v0.2.0 - Added expression and function evaluation and __chromate global for general message passing. Events now get complete message, not just the data part. (May 2017)
• v0.1.x - Initial version (May 2017)

License

MIT