PhantomJS integration module for NodeJS

phantom - Fast NodeJS API for PhantomJS


NPM Version NPM Downloads Linux Build Dependencies

Super easy to use

var phantom = require('phantom');

var sitepage = null;
var phInstance = null;
    .then(instance => {
        phInstance = instance;
        return instance.createPage();
    .then(page => {
        sitepage = page;
    .then(status => {
    .then(content => {
    .catch(error => {

See examples folder for more ways to use this module.


$ npm install phantom --save

How does it work?

v1.0.x used to use dnode to communicate between nodejs and phantomjs. This approach raised a lot of security restrictions and did not work well when using cluster or pm2.

v2.0.x has been completely rewritten to use sysin and sysout pipes to communicate with the phantomjs process. It works out of the box with cluster and pm2. If you want to see the messages that are sent try adding DEBUG=true to your execution, ie. DEBUG=true node path/to/test.js. The new code is much cleaner and simpler. PhantomJS is started with shim.js which proxies all messages to the page or phantom object.

Migrating from 1.0.x

Version 2.0.x is not backward compatible with previous versions. Most notability, method calls do not take a callback function anymore. Since node supports Promise, each of the methods return a promise. Instead of writing, function(){}) you would have to write{}).

The API is much more consistent now. All properties can be read with and settings can be read with page.setting(key). See below for more example.

phantom object API


To create a new instance of phantom use phantom.create() which returns a Promise which should resolve with a phantom object. If you want add parameters to the phantomjs process you can do so by doing:

var phantom = require('phantom');
phantom.create(['--ignore-ssl-errors=yes', '--load-images=no']).then(...)

You can also explicitly set phantomjs path to use by passing it in cofig object:

var phantom = require('phantom');
phantom.create([], {phantomPath: '/path/to/phantomjs'}).then(...)


To create a new page, you have to call createPage():

var phantom = require('phantom');
phantom.create().then(function(ph) {
    ph.createPage().then(function(page) {
        // use page


Sends an exit call to phantomjs process.

Make sure to call it on the phantom instance to kill the phantomjs process. Otherwise, the process will never exit.


Kills the underlying phantomjs process (by sending SIGKILL to it).

It may be a good idea to register handlers to SIGTERM and SIGINT signals with #kill().

However, be aware that phantomjs process will get detached (and thus won't exit) if node process that spawned it receives SIGKILL!

page object API

The page object that is returned with #createPage is a proxy that sends all methods to phantom. Most method calls should be identical to PhantomJS API. You must remember that each method returns a Promise.


page.settings can be accessed via page.setting(key) or set via page.setting(key, value). Here is an example to read javascriptEnabled property.



Page properties can be read using the #property(key) method.'plainText').then(function(content) {

Page properties can be set using the #property(key, value) method.'viewportSize', {width: 800, height: 600}).then(function() {

When setting values, using then() is optional. But beware that the next method to phantom will block until it is ready to accept a new message.

You can set events using #property() because they are property members of page.'onResourceRequested', function(requestData, networkRequest) {

It is important to understand that the function above executes in the PhantomJS process. PhantomJS does not share any memory or variables with node. So using closures in javascript to share any variables outside of the function is not possible. Variables can be passed to #property instead. So for example, let's say you wanted to pass process.env.DEBUG to onResourceRequested method above. You could do this by:'onResourceRequested', function(requestData, networkRequest, debug) {
      // do something with it
}, process.env.DEBUG);

Even if it is possible to set the events using this way, we recommend you use #on() for events (see below).

You can return data to NodeJS by using #createOutObject(). This is a special object that let's you write data in PhantomJS and read it in NodeJS. Using the example above, data can be read by doing:

var outObj = phInstance.createOutObject();
outObj.urls = [];'onResourceRequested', function(requestData, networkRequest, out) {
}, outObj);

// after call to'urls').then(function(urls){


By using on(event, [runOnPhantom=false],listener, args*), you can listen to the events the events the page emits.

var urls = [];

page.on('onResourceRequested', function (requestData, networkRequest) {
    urls.push(requestData.url); // this would push the url into the urls array above
    networkRequest.abort(); // This will fail, because the params are a serialized version of what was provided


As you see, using on you have access to the closure variables and all the node goodness using this function ans in contrast of setting and event with property, you can set as many events as you want.

If you want to register a listener to run in phantomjs runtime (and thus, be able to cancel the request lets say), you can make it by passing the optional param runOnPhantom as true;

var urls = [];

page.on('onResourceRequested', true, function (requestData, networkRequest) {
    urls.push(requestData.url); // now this wont work, because this function would exercute in phantom runtime and thus wont have acces to the closure.
    networkRequest.abort(); // This would work, because you are accessing to the non serialized networkRequest.


The same as in property, you can pass additional params to the function in the same way, and even use the object created by #createOutObject().

You cannot use #property() and #on() at the same time, because it would conflict. Property just sets the function in phantomjs, while #on() manages the event in a different way.


#off(event) is usefull to remove all the event listeners set by #on() for ans specific event.


Using #evaluate() is similar to passing a function above. For example, to return HTML of an element you can do:

page.evaluate(function() {
    return document.getElementById('foo').innerHTML;


Evaluate a function contained in a string. It is similar to #evaluate(), but the function can't take any arguments. This example does the same thing as the example of #evaluate():

page.evaluateJavaScript('function() { return document.getElementById(\'foo\').innerHTML; }').then(function(html){


Switch to the frame specified by a frame name or a frame position:

page.switchToFrame(framePositionOrName).then(function() {
    // now the context of `page` will be the iframe if frame name or position exists


Switch to the main frame of the page:

page.switchToMainFrame().then(function() {
    // now the context of `page` will the main frame


A method can be defined using the #defineMethod(name, definition) method.

page.defineMethod('getZoomFactor', function() {
    return this.zoomFactor;


An asynchronous method can be invoked using the #invokeAsyncMethod(method, arg1, arg2, arg3...) method.

page.invokeAsyncMethod('open', '').then(function(status) {


A method can be invoked using the #invokeMethod(method, arg1, arg2, arg3...) method.

page.invokeMethod('evaluate', function() {
    return document.title;
}).then(function(title) {
page.invokeMethod('evaluate', function(selector) {
    return document.querySelector(selector) !== null;
}, '#element').then(function(exists) {


To run the test suite, first install the dependencies, then run npm test:

$ npm install
$ npm test


This package is under development. Pull requests are welcomed. Please make sure tests are added for new functionalities and that your build does pass in TravisCI.


The current lead maintainer is Amir Raminfar

List of all contributors