URL to PDF Microservice

Web page PDF rendering done right. Microservice for rendering receipts, invoices, or any content. Packaged to an easy API.

⚠️ WARNING ⚠️ Don't serve this API publicly to the internet unless you are aware of the risks. It allows API users to run any JavaScript code inside a Chrome session on the server. It's fairly easy to expose the contents of files on the server. You have been warned!. See alvarcarto#12 for background.

⭐️ Features:

• Converts any URL or HTML content to a PDF file or an image (PNG/JPEG)
• Rendered with Headless Chrome, using Puppeteer. The PDFs should match to the ones generated with a desktop Chrome.
• Sensible defaults but everything is configurable.
• Single-page app (SPA) support. Waits until all network requests are finished before rendering.
• Easy deployment to Heroku. We love Lambda but...Deploy to Heroku button.
• Renders lazy loaded elements. (scrollPage option)
• Supports optional x-api-key authentication. (API_TOKENS env var)

Usage is as simple as https://url-to-pdf-api.herokuapp.com/api/render?url=http://google.com. There's also a POST /api/render if you prefer to send options in the body.

🔍 Why?

This microservice is useful when you need to automatically produce PDF files for whatever reason. The files could be receipts, weekly reports, invoices, or any content.

PDFs can be generated in many ways, but one of them is to convert HTML+CSS content to a PDF. This API does just that.

🚀 Shortcuts:

• Examples
• API
• I want to run this myself

How it works

Local setup is identical except Express API is running on your machine and requests are direct connections to it.

Good to know

• By default, page's @media print CSS rules are ignored. We set Chrome to emulate @media screen to make the default PDFs look more like actual sites. To get results closer to desktop Chrome, add &emulateScreenMedia=false query parameter. See more at Puppeteer API docs.

• Chrome is launched with --no-sandbox --disable-setuid-sandbox flags to fix usage in Heroku. See this issue.

• Heavy pages may cause Chrome to crash if the server doesn't have enough RAM.

Examples

⚠️ Restrictions ⚠️:

• For security reasons the urls have been restricted and HTML rendering is disabled. For full demo, run this app locally or deploy to Heroku.
• The demo Heroku app runs on a free dyno which sleep after idle. A request to sleeping dyno may take even 30 seconds.

The most minimal example, render google.com

https://url-to-pdf-api.herokuapp.com/api/render?url=http://google.com

The most minimal example, render google.com as PNG image

https://url-to-pdf-api.herokuapp.com/api/render?output=screenshot&url=http://google.com

Use the default @media print instead of @media screen.

https://url-to-pdf-api.herokuapp.com/api/render?url=http://google.com&emulateScreenMedia=false

Use scrollPage=true which tries to reveal all lazy loaded elements. Not perfect but better than without.

https://url-to-pdf-api.herokuapp.com/api/render?url=http://www.andreaverlicchi.eu/lazyload/demos/lazily_load_lazyLoad.html&scrollPage=true

Render only the first page.

https://url-to-pdf-api.herokuapp.com/api/render?url=https://en.wikipedia.org/wiki/Portable_Document_Format&pdf.pageRanges=1

Render A5-sized PDF in landscape.

https://url-to-pdf-api.herokuapp.com/api/render?url=http://google.com&pdf.format=A5&pdf.landscape=true

Add 2cm margins to the PDF.

https://url-to-pdf-api.herokuapp.com/api/render?url=http://google.com&pdf.margin.top=2cm&pdf.margin.right=2cm&pdf.margin.bottom=2cm&pdf.margin.left=2cm

Wait for extra 1000ms before render.

https://url-to-pdf-api.herokuapp.com/api/render?url=http://google.com&waitFor=1000

Download the PDF with a given attachment name

https://url-to-pdf-api.herokuapp.com/api/render?url=http://google.com&attachmentName=google.pdf

Wait for an element matching the selector input appears.

https://url-to-pdf-api.herokuapp.com/api/render?url=http://google.com&waitFor=input

Render HTML sent in JSON body

NOTE: Demo app has disabled html rendering for security reasons.

curl -o html.pdf -XPOST -d'{"html": "<body>test</body>"}' -H"content-type: application/json" http://localhost:9000/api/render

Render HTML sent as text body

NOTE: Demo app has disabled html rendering for security reasons.

curl -o html.pdf -XPOST -d@test/resources/large.html -H"content-type: text/html" http://localhost:9000/api/render

API

To understand the API options, it's useful to know how Puppeteer is internally used by this API. The render code is quite simple, check it out. Render flow:

1. page.setViewport(options) where options matches viewport.*.

2. Possibly page.emulateMedia('screen') if emulateScreenMedia=true is set.

3. Render url or html.

   If url is defined, page.goto(url, options) is called and options match goto.*. Otherwise page.setContent(html, options) is called where html is taken from request body, and options match goto.*.

4. Possibly page.waitFor(numOrStr) if e.g. waitFor=1000 is set.

5. Possibly Scroll the whole page to the end before rendering if e.g. scrollPage=true is set.

   Useful if you want to render a page which lazy loads elements.

6. Render the output

• If output is pdf rendering is done with page.pdf(options) where options matches pdf.*.
• Else if output is screenshot rendering is done with page.screenshot(options) where options matches screenshot.*.

GET /api/render

All options are passed as query parameters. Parameter names match Puppeteer options.

These options are exactly the same as its POST counterpart, but options are expressed with the dot notation. E.g. ?pdf.scale=2 instead of { pdf: { scale: 2 }}.

The only required parameter is url.

Parameter	Type	Default	Description
url	string	-	URL to render as PDF. (required)
output	string	pdf	Specify the output format. Possible values: pdf , screenshot or html.
emulateScreenMedia	boolean	true	Emulates @media screen when rendering the PDF.
enableGPU	boolean	false	When set, enables chrome GPU. For windows user, this will always return false. See https://developers.google.com/web/updates/2017/04/headless-chrome
ignoreHttpsErrors	boolean	false	Ignores possible HTTPS errors when navigating to a page.
scrollPage	boolean	false	Scroll page down before rendering to trigger lazy loading elements.
waitFor	number or string	-	Number in ms to wait before render or selector element to wait before render.
attachmentName	string	-	When set, the content-disposition headers are set and browser will download the PDF instead of showing inline. The given string will be used as the name for the file.
viewport.width	number	1600	Viewport width.
viewport.height	number	1200	Viewport height.
viewport.deviceScaleFactor	number	1	Device scale factor (could be thought of as dpr).
viewport.isMobile	boolean	false	Whether the meta viewport tag is taken into account.
viewport.hasTouch	boolean	false	Specifies if viewport supports touch events.
viewport.isLandscape	boolean	false	Specifies if viewport is in landscape mode.
cookies[0][name]	string	-	Cookie name (required)
cookies[0][value]	string	-	Cookie value (required)
cookies[0][url]	string	-	Cookie url
cookies[0][domain]	string	-	Cookie domain
cookies[0][path]	string	-	Cookie path
cookies[0][expires]	number	-	Cookie expiry in unix time
cookies[0][httpOnly]	boolean	-	Cookie httpOnly
cookies[0][secure]	boolean	-	Cookie secure
cookies[0][sameSite]	string	-	Strict or Lax
goto.timeout	number	30000	Maximum navigation time in milliseconds, defaults to 30 seconds, pass 0 to disable timeout.
goto.waitUntil	string	networkidle0	When to consider navigation succeeded. Options: load, domcontentloaded, networkidle0, networkidle2. load - consider navigation to be finished when the load event is fired. domcontentloaded - consider navigation to be finished when the DOMContentLoaded event is fired. networkidle0 - consider navigation to be finished when there are no more than 0 network connections for at least 500 ms. networkidle2 - consider navigation to be finished when there are no more than 2 network connections for at least 500 ms.
pdf.scale	number	1	Scale of the webpage rendering.
pdf.printBackground	boolean	false	Print background graphics.
pdf.displayHeaderFooter	boolean	false	Display header and footer.
pdf.headerTemplate	string	-	HTML template to use as the header of each page in the PDF. Currently Puppeteer basically only supports a single line of text and you must use pdf.margins+CSS to make the header appear! See alvarcarto#77.
pdf.footerTemplate	string	-	HTML template to use as the footer of each page in the PDF. Currently Puppeteer basically only supports a single line of text and you must use pdf.margins+CSS to make the footer appear! See alvarcarto#77.
pdf.landscape	boolean	false	Paper orientation.
pdf.pageRanges	string	-	Paper ranges to print, e.g., '1-5, 8, 11-13'. Defaults to the empty string, which means print all pages.
pdf.format	string	A4	Paper format. If set, takes priority over width or height options.
pdf.width	string	-	Paper width, accepts values labeled with units.
pdf.height	string	-	Paper height, accepts values labeled with units.
pdf.fullPage	boolean	-	Create PDF in a single page
pdf.margin.top	string	-	Top margin, accepts values labeled with units.
pdf.margin.right	string	-	Right margin, accepts values labeled with units.
pdf.margin.bottom	string	-	Bottom margin, accepts values labeled with units.
pdf.margin.left	string	-	Left margin, accepts values labeled with units.
screenshot.fullPage	boolean	true	When true, takes a screenshot of the full scrollable page.
screenshot.type	string	png	Screenshot image type. Possible values: png, jpeg
screenshot.quality	number	-	The quality of the JPEG image, between 0-100. Only applies when screenshot.type is jpeg.
screenshot.omitBackground	boolean	false	Hides default white background and allows capturing screenshots with transparency.
screenshot.clip.x	number	-	Specifies x-coordinate of top-left corner of clipping region of the page.
screenshot.clip.y	number	-	Specifies y-coordinate of top-left corner of clipping region of the page.
screenshot.clip.width	number	-	Specifies width of clipping region of the page.
screenshot.clip.height	number	-	Specifies height of clipping region of the page.
screenshot.selector	string	-	Specifies css selector to clip the screenshot to.

Example:

curl -o google.pdf https://url-to-pdf-api.herokuapp.com/api/render?url=http://google.com

POST /api/render - (JSON)

All options are passed in a JSON body object. Parameter names match Puppeteer options.

These options are exactly the same as its GET counterpart.

Body

The only required parameter is url.

{
  // Url to render. Either url or html is required
  url: "https://google.com",

  // Either "pdf" or "screenshot"
  output: "pdf",

  // HTML content to render. Either url or html is required
  html: "<html><head></head><body>Your content</body></html>",

  // If we should emulate @media screen instead of print
  emulateScreenMedia: true,

  // If we should ignore HTTPS errors
  ignoreHttpsErrors: false,

  // If true, page is scrolled to the end before rendering
  // Note: this makes rendering a bit slower
  scrollPage: false,

  // Passed to Puppeteer page.waitFor()
  waitFor: null,

  // Passsed to Puppeteer page.setCookies()
  cookies: [{ ... }]

  // Passed to Puppeteer page.setViewport()
  viewport: { ... },

  // Passed to Puppeteer page.goto() as the second argument after url
  goto: { ... },

  // Passed to Puppeteer page.pdf()
  pdf: { ... },

  // Passed to Puppeteer page.screenshot()
  screenshot: { ... },
}

Example:

curl -o google.pdf -XPOST -d'{"url": "http://google.com"}' -H"content-type: application/json" http://localhost:9000/api/render

curl -o html.pdf -XPOST -d'{"html": "<body>test</body>"}' -H"content-type: application/json" http://localhost:9000/api/render

POST /api/render - (HTML)

HTML to render is sent in body. All options are passed in query parameters. Supports exactly the same query parameters as GET /api/render, except url paremeter.

Remember that relative links do not work.

Example:

curl -o receipt.html https://rawgit.com/wildbit/postmark-templates/master/templates_inlined/receipt.html
curl -o html.pdf -XPOST [email protected] -H"content-type: text/html" http://localhost:9000/api/render?pdf.scale=1

Development

To get this thing running, you have two options: run it in Heroku, or locally.

The code requires Node 8+ (async, await).

1. Heroku deployment

Scroll this readme up to the Deploy to Heroku -button. Click it and follow instructions.

WARNING: Heroku dynos have a very low amount of RAM. Rendering heavy pages may cause Chrome instance to crash inside Heroku dyno. 512MB should be enough for most real-life use cases such as receipts. Some news sites may need even 2GB of RAM.

2. Local development

First, clone the repository and cd into it.

• cp .env.sample .env

• Fill in the blanks in .env

• npm install

• npm start Start express server locally

• Server runs at http://localhost:9000 or what $PORT env defines

Techstack

• Node 8+ (async, await), written in ES7
• Express.js app with a nice internal architecture, based on these conventions.
• Hapi-style Joi validation with express-validation
• Heroku + Puppeteer buildpack
• Puppeteer to control Chrome