Render a webpage and get the image as a stream
Last updated 4 years ago by bubenshchykov .
Repository · Bugs · Original npm · Tarball · package.json
$ cnpm install html-convert 
SYNC missed versions from official npm registry.


Render a webpage and get the image as a stream.

npm install html-convert

Build Status

It uses a pool of phantom processes so it doesn't need to spawn a new process for each website. New requests are added to the pool member with the shortest queue length.


This module depends on the phantomjs-prebuilt module, which will install PhantomJS for you if you don't already have it.

var htmlConvert = require('html-convert');
var fs = require('fs');

var convert = htmlConvert();

// convert a website url


// or as a transform stream


You can also pass some options:

var convert = htmlConvert({
  pool        : 5,           // Change the pool size. Defaults to 1
  queueTimeout     : 1000,   // Kills all worker jobs if queue is not empty within queueTimeout ms. Defaults to 30 seconds.
  fetchTimeout:     : 1000,  // Page url fetch timeout. Defaults to 30 seconds.
  renderTimeout     : 1000,  // Render timeout in milliseconds. Defaults to 30 seconds.
  strictRender:false, // Flag to fail render on any js errors, resource errors or timeouts. Defaults to false.
  tmp         : '/tmp',      // Set the tmp where tmp data is stored when communicating with the phantom process.
                             //   Defaults to /tmp if it exists, or os.tmpDir()
  format      : 'jpeg',      // The default output format. Defaults to png
  quality     : 100,         // The default image quality. Defaults to 100. Only relevant for jpeg format.
  width       : 1280,        // Changes the width size. Defaults to 1280
  height      : 800,         // Changes the height size. Defaults to 960
  paperFormat : 'A4',        // Defaults to A4. Also supported: 'A3', 'A4', 'A5', 'Legal', 'Letter', 'Tabloid'.
  orientation : 'portrait',  // Defaults to portrait. 'landscape' is also valid
  margin      : '0cm',       // Defaults to 0cm. Supported dimension units are: 'mm', 'cm', 'in', 'px'. No unit means 'px'.
  userAgent   : '',          // No default.
  headers     : {Foo:'bar'}, // Additional headers to send with each upstream HTTP request
  paperSize:  : null,        // Defaults to the paper format, orientation, and margin.
  crop        : false,       // Defaults to false. Set to true or {top:5, left:5} to add margin
  printMedia  : false,       // Defaults to false. Force the use of a print stylesheet.
  maxErrors   : 3,           // Number errors phantom process is allowed to throw before killing it. Defaults to 3.
  expects     : 'something', // No default. Do not render until window.renderable is set to 'something'
  retries     : 1,           // How many times to try a render before giving up. Defaults to 1.
  phantomFlags: ['--ignore-ssl-errors=true'] // Defaults to []. Command line flags passed to PhantomJS
  maxRenders  : 500,          // How many renders can a phantom process make before being restarted. Defaults to 500

  injectJs    : ['./includes/my-polyfill.js'] // Array of paths to polyfill components or external scripts that will be injected when the page is initialized

Or override the options for each render stream

convert(myUrl, {format:'jpeg', quality: 100, width: 1280, height: 960}).pipe(...)

Supported output formats

We support the output formats that PhantomJS's render method supports. At the time of this writing these are:

  • png
  • gif
  • jpg
  • pdf


Since the interface is just a stream you can pipe the web site anywhere! Try installing picture-tube and run the following example

var htmlConvert = require('html-convert');
var pictureTube = require('picture-tube');
var convert = htmlConvert();


Deferred convert

If you need your page to do something before phantom renders it you just need to immediately set window.renderable to false. If that is set when the page is opened the module will wait for window.renderable to be set to true and when this happens the render will occur.

Here is an example to illustrate it better.

<html lang="en">
  <script type="text/javascript">window.renderable = false</script>
  <meta charset="UTF-8">

<script type="text/javascript">
  doSomeAjaxLoading(function() {
    window.renderable = true;

Adding Cookies

You can add any special cookies at render time. For format, see Example:

var convert = htmlConvert({
  pool: 5,
  format: 'pdf'
  // other opts

convert('', {
  cookies: [{
    'name'     : 'Valid-Cookie-Name',   /* required property */
    'value'    : 'Valid-Cookie-Value',  /* required property */
    'domain'   : 'localhost',
    'path'     : '/foo',                /* required property */
    'httponly' : true,
    'secure'   : false,
    'expires'  : (new Date()).getTime() + (1000 * 60 * 60)   /* <-- expires in 1 hour */

That will use that cookie for that particular render job. You probably want to set the expires property to something fairly short, as there may not be a guarantee that a pooled phantom process won't pick up the cookie for a particular render job, and you may want that session to only be valid for an individual job run.

Injecting JavaScript

Sometimes you need to inject polyfills, e.g. PhantomJS Date.parse is broken. You can add paths to local files to polyfill broken / missing features of PhantomJS using the opts.injectJs property. Example:

var convert = htmlConvert({
  injectJs: ['./includes/my-date-polyfill.js']

Obviously, make sure the path './includes/my-date-polyfill.js' is resolvable from the project root, or pass in an absolute path. When the page is initialized, any scripts you listed there will be injected before any rendering happens.

Extra Dependencies

For rendering, PhantomJS requires the fontconfig library, which may be missing if you're using Ubuntu Server. To install on Ubuntu:

sudo apt-get install libfontconfig


Render stream emits "log" event with useful debug details coming from onError (JS error), onConsoleMessage, onResourceError, onResourceTimeout webpage hooks.

var convert = htmlConvert();

  .on('log', function(log) {
    // {type: 'error', data: {msg: 'ReferenceError: Can\'t find variable: a', trace: [..]}}

Also, some additional debugging output may be enabled by running your app with a DEBUG environment variable set as follows:

DEBUG=phantom-render-stream  node ./your-script.js

If you are getting undefined error codes and responses when attempting to render, it's likely a connection issue of some sort. If the URL uses SSL, adding --ignore-ssl-errors=true to phantomFlags may help. You also try adding --debug=true to the phantomFlags array.

See Also

  • wkhtmltopdf is a Node module that uses wkhtmltopdf to convert HTML to PDF. It is similar in that it uses Webkit and produces output as a stream, and different in that it doesn't use PhantomJS. Also, wkhtmotopdf only supports PDF output.



Current Tags

  • 2.1.7                                ...           latest (4 years ago)

6 Versions

  • 2.1.7                                ...           4 years ago
  • 2.1.6                                ...           4 years ago
  • 2.1.5                                ...           4 years ago
  • 2.1.4                                ...           4 years ago
  • 2.1.3                                ...           4 years ago
  • 2.1.2                                ...           4 years ago
Maintainers (1)
Today 0
This Week 1
This Month 1
Last Day 1
Last Week 0
Last Month 14
Dependencies (16)
Dev Dependencies (1)
Dependents (1)

Copyright 2014 - 2016 © |