🤖 pdf-bot

Easily create a microservice for generating PDFs using headless Chrome.

pdf-bot is installed on a server and will receive URLs to turn into PDFs through its API or CLI. pdf-bot will manage a queue of PDF jobs. Once a PDF job has run it will notify you using a webhook so you can fetch the API. pdf-bot supports storing PDFs on S3 out of the box. Failed PDF generations and Webhook pings will be retried after a configurable decaying schedule.

pdf-bot uses html-pdf-chrome under the hood and supports all the settings that it supports. Major thanks to @westy92 for making this possible.

How does it work?

Imagine you have an app that creates invoices. You want to save those invoices as PDF. You install pdf-bot on a server as an API. Your app server sends the URL of the invoice to the pdf-bot server. A cronjob on the pdf-bot server keeps checking for new jobs, generates a PDF using headless Chrome and sends the location back to the application server using a webhook.

Prerequisites

• Node.js v6 or later

Installation

$ npm install -g pdf-bot
$ pdf-bot install

Make sure the node path is in your $PATH

pdf-bot install will prompt for some basic configurations and then create a storage folder where your database and pdf files will be saved.

Configuration

pdf-bot comes packaged with sensible defaults. At the very minimum you must have a config file in the same folder from which you are executing pdf-bot with a storagePath given. However, in reality what you probably want to do is use the pdf-bot install command to generate a configuration file and then use an alias ALIAS pdf-bot = "pdf-bot -c /home/pdf-bot.config.js"

pdf-bot.config.js

var htmlPdf = require('html-pdf-chrome')

module.exports = {
  api: {
    token: 'crazy-secret'
  },
  generator: {
    completionTrigger: new htmlPdf.CompletionTrigger.Timer(1000) // 1 sec timeout
  },
  storagePath: 'storage'
}

$ pdf-bot -c ./pdf-bot.config.js push https://esbenp.github.io

See a full list of the available configuration options.

Usage guide

Structure and concept

pdf-bot is meant to be a microservice that runs a server to generate PDFs for you. That usually means you will send requests from your application server to the PDF server to request an url to be generated as a PDF. pdf-bot will manage a queue and retry failed generations. Once a job is successfully generated a path to it will be sent back to your application server.

Let us check out the flow for an app that generates PDF invoices.

1. (App server): An invoice is created ----> Send URL to invoice to pdf-bot server
2. (pdf-bot server): Put the URL in the queue
3. (pdf-bot server): PDF is generated using headless Chrome
4. (pdf-bot server): (if failed try again using 1 min, 3 min, 10 min, 30 min, 60 min delay)
5. (pdf-bot server): Upload PDF to storage (e.g. Amazon S3)
6. (pdf-bot server): Send S3 location of PDF back to the app server
7. (App server): Receive S3 location of PDF -> Check signature sum matches for security
8. (App server): Handle PDF however you see fit (move it, download it, save it etc.)

You can send meta data to the pdf-bot server that will be sent back to the application. This can help you identify what PDF you are receiving.

Setup

On your pdf-bot server start by creating a config file pdf-bot.config.js. You can see an example file here

pdf-bot.config.js

module.exports = {
  api: {
    port: 3000,
    token: 'api-token'
  },
  storage: {
    's3': createS3Config({
      bucket: '',
      accessKeyId: '',
      region: '',
      secretAccessKey: ''
    })
  },
  webhook: {
    secret: '1234',
    url: 'http://localhost:3000/webhooks/pdf'
  }
}

As a minimum you should configure an access token for your API. This will be used to authenticate jobs sent to your pdf-bot server. You also need to add a webhook configuration to have pdf notifications sent back to your application server. You should add a secret that will be used to generate a signature used to check that the request has not been tampered with during transfer.

Start your API using

pdf-bot -c ./pdf-bot.config.js api

This will start an express server that listens for new jobs on port 3000.

Setting up Chrome

pdf-bot uses html-pdf-chrome which in turns uses chrome-launcher to launch chrome. You should check out those two resources on how to properly setup Chrome. However, with chrome-launcher Chrome should be started automatically. Otherwise, html-pdf-chrome has a small guide on how to have it running as a process using pm2.

You can install chrome on Ubuntu using

sudo apt-get update && apt-get install chromium-browser

If you are testing things on OSX or similar, chrome-launcher should be able to find and automatically startup Chrome for you.

Setting up the receiving API

In the examples folder there is a small example on how the application API could look. Basically, you just have to define an endpoint that will receive the webhook and check that the signature matches.

api.post('/hook', function (req, res) {
  var signature = req.get('X-PDF-Signature', 'sha1=')

  var bodyCrypted = require('crypto')
    .createHmac('sha1', '12345')
    .update(JSON.stringify(req.body))
    .digest('hex')

  if (bodyCrypted !== signature) {
    res.status(401).send()
    return
  }

  console.log('PDF webhook received', JSON.stringify(req.body))

  res.status(204).send()
})

Setup production environment

Follow the guide under production/ to see how to setup pdf-bot using pm2 and nginx

Setup crontab

We setup our crontab to continuously look for jobs that have not yet been completed.

* * * * * node $(npm bin -g)/pdf-bot -c ./pdf-bot.config.js shift:all >> /var/log/pdfbot.log 2>&1
* * * * * node $(npm bin -g)/pdf-bot -c ./pdf-bot.config.js ping:retry-failed >> /var/log/pdfbot.log 2>&1

Quick example using the CLI

Let us assume I want to generate a PDF for https://esbenp.github.io. I can add the job using the pdf-bot CLI.

$ pdf-bot -c ./pdf-bot.config.js push https://esbenp.github.io --meta '{"id":1}'

Next, if my crontab is not setup to run it automatically I can run it using the shift:all command

$ pdf-bot -c ./pdf-bot.config.js shift:all

This will look for the oldest uncompleted job and run it.

How can I generate PDFs for sites that use Javascript?

This is a common issue with PDF generation. Luckily, html-pdf-chrome has a really awesome API for dealing with Javascript. You can specify a timeout in milliseconds, wait for elements or custom events. To add a wait simply configure the generator key in your configuration. Below are a few examples.

Wait for 5 seconds

var htmlPdf = require('html-pdf-chrome')

module.exports = {
  api: {
    token: 'api-token'
  },
  // html-pdf-chrome options
  generator: {
    completionTrigger: new htmlPdf.CompletionTrigger.Timer(5000), // waits for 5 sec
  },
  webhook: {
    secret: '1234',
    url: 'http://localhost:3000/webhooks/pdf'
  }
}

Wait for event

var htmlPdf = require('html-pdf-chrome')

module.exports = {
  api: {
    token: 'api-token'
  },
  // html-pdf-chrome options
  generator: {
    completionTrigger: new htmlPdf.CompletionTrigger.Event(
      'myEvent', // name of the event to listen for
      '#myElement', // optional DOM element CSS selector to listen on, defaults to body
      5000 // optional timeout (milliseconds)
    )
  },
  webhook: {
    secret: '1234',
    url: 'http://localhost:3000/webhooks/pdf'
  }
}

In your Javascript trigger the event when rendering is complete

document.getElementById('myElement').dispatchEvent(new CustomEvent('myEvent'));

Wait for variable

var htmlPdf = require('html-pdf-chrome')

module.exports = {
  api: {
    token: 'api-token'
  },
  // html-pdf-chrome options
  generator: {
    completionTrigger: new htmlPdf.CompletionTrigger.Variable(
      'myVarName', // optional, name of the variable to wait for.  Defaults to 'htmlPdfDone'
      5000 // optional, timeout (milliseconds)
    )
  },
  webhook: {
    secret: '1234',
    url: 'http://localhost:3000/webhooks/pdf'
  }
}

In your Javascript set the variable when the rendering is complete

window.myVarName = true;

You can find more completion triggers in html-pdf-chrome's documentation

API

Below are given the endpoints that are exposed by pdf-server's REST API

Push URL to queue: POST /

key | type | required | description --- | ---- | -------- | ----------- url | string | yes | The URL to generate a PDF from meta | object | | Optional meta data object to send back to the webhook url

Example

curl -X POST -H 'Authorization: Bearer api-token' -H 'Content-Type: application/json' http://pdf-bot.com/ -d '
  {
    "url":"https://esbenp.github.io",
    "meta":{
      "type