How can I automate HTML-to-PDF conversions?

Question

I've been using htmldoc for a while, but I've run into some fairly serious limitations. I need the end solution to work on a Linux box. I'll be calling this library/utility/application from a Perl app, so any Perl interfaces would be a bonus.

Answer 1

Sorry to unearth this old post, but it came out first in my search for the best HTML/PDF conversion tool. On Linux wkhtmltopdf is very good (takes into account CSS, among others) and GPL.

Answer 2

WeasyPrint produces nice PDFs with selectable text and hyperlinks.

weasyprint input.html output.pdf

If you use wkhtmltopdf instead, try the following options:

wkhtmltopdf --margin-bottom 20mm --margin-top 20mm --minimum-font-size 16 ...

Answer 3

Update 2019-05

The whole process has thankfully been packed into a docker image by TheCodingMachine: https://github.com/thecodingmachine/gotenberg

This makes maintenance and usage of chrome based pdf generation in production environments really smooth and hassle free.

---

There is a new headless mode since Chrome 59. As all the other solutions really struggle with newer (or not so new anymore) CSS features like flexbox, this was in my case the only solution to produce a proper PDF output.

To create a pdf from a local html file just use the following command: chrome --headless --disable-gpu --print-to-pdf file:///path/to/myfile.html.

For Mac OS substitue chrome with /Applications/Google\ Chrome.app/Contents/MacOS/Google\ Chrome.

The only downside I noticed so far is that (currently) you can not pass the html via stdin, but creating a temporary file is not that much of an issue.

For more information see https://developers.google.com/web/updates/2017/04/headless-chrome#create_a_pdf_dom

Update: As it turns out, the chrome guys will most likely provide some kind of node module for this task, which would eventually deprecate the headless mode (https://bugs.chromium.org/p/chromium/issues/detail?id=719921).

The best bet would be to use the node based approach using the puppeteer module as documented under https://developers.google.com/web/updates/2017/04/headless-chrome#node and print the page via the Page.printToPDF command, which enables some additional configuration, too.

Of course, you can connect to the debug console websocket from any other environment than node (i.e. PHP script), too.

Answer 4

NOTE: This answer is from 2008 and is probably now incorrect; please check the other answers

PrinceXML is the best one I've seen (it parses regular HTML as well as XML/XHTML). How is it the best? Well, it passes the acid2 test which I thought was pretty darn impressive

It is however, quite expensive

Answer 5

I did a bit of googling for you and came up with two options. There may be more, my google strategy was to try "webkit command-line pdf" and "gecko command-line pdf", basically looking for commandline programs that embed the two popular open-source rendering engines in command-line renderers. Here's what I found:

Firefox command-line printer - outputs to pdf and png

wkpdf - while this is for mac, it's probably pretty portable.

Answer 6

I wont claim this is the "best" solution but it is "a" solution i have used.

HTML Input --> HTML 2 PS --> PS 2 PDF --> PDF Output

Answer 7

You can install the free Calibre, and use the ebook-convert command line utility it has, to convert many html documents into a single epub, or pdf.

https://manual.calibre-ebook.com/generated/en/ebook-convert.html

Idea comes from here

I haven't used it, but this npm module wraps this process up like my following bash script, but probably better ;-)

For me, on my mac, I use the following bash script to convert a local html website to a PDF:

convert_html_to_pdf.sh

function show_help()
{
  ME=$(basename $0)
  IT=$(cat <<EOF
  
  Converts an html file to pdf, epub, mobi or more if you look!

  usage: input.html output.{pdf|epub|mobi}
  
  e.g. 
  
  $ME index.html output.pdf 

  Note: Requires Calibre be installed. more info here: https://ebooks.stackexchange.com/a/6285
EOF
  )
  echo "$IT"
  exit
}

if [ "$1" == "help" ]
then
  show_help
fi
if [ "$1" == "--help" ]
then
  show_help
fi

/Applications/calibre.app/Contents/MacOS/ebook-convert $1 $2 --max-levels=1

Answer 8

This would be total overkill, but you could download and install mirth. It is a message routing engine, but it has the ability to convert html to pdf, so you could set it up to pick up an html file in a folder, convert to pdf, and drop the pdf in the same or other folder. Like I said, overkill, a bit of a learning curve, but it's free, and java so you can run it on linux if you like. And all your perl app would have to do is drop the html to a file.

Answer 9

You should have a look at http://phantomjs.org/

Conversion can be done by a small script rasterize.js and then issuing

phantomjs rasterize.js 'http://en.wikipedia.org/w/index.php?title=Jakarta&printable=yes' jakarta.pdf

Answer 10

I have found Electroshot to be supportive of modern CSS features, particularly layout. This was after struggling with wkhtmltopdf showing its age in not supporting things like CSS3.

From Electroshot's features description:

Electroshot uses Electron, which offers the most recent stable version of Chrome (rather than one from years ago); this means that pages render as they would in a browser...

I've been able to use Bootstrap 4 to design a page, and then use Electroshot to render a PDF very closely resembling the HTML/CSS.

Answer 11

An alternative solution that hasn't been answered here is to use an API.

That advantage of them is that you externalize the resources needed for the job and have an up-to-date service that implements the recent features (no needs to update the code or install bugfixes).

For instance, with PDFShift, you can do that with a single POST request at:

POST https://api.pdfshift.io/v2/convert/

And passing the "source" (either an URL or a raw HTML code), and you'll get back a PDF in binary. (Disclaimer: I work at PDFShift).

Here's a code sample in Python:

import requests

response = requests.post(
    'https://api.pdfshift.io/v2/convert/',
    auth=('user_api_key', ''),
    json={"source": "https://en.wikipedia.org/wiki/PDF", "landscape": False, "use_print": False}
)

response.raise_for_status()

with open('wikipedia.pdf', 'wb') as f:
    f.write(response.content)

And your PDF will be located at ./wikipedia.pdf

Answer 12

Here is a nice easy-to-install version of headless Chrome:

https://www.npmjs.com/package/chrome-headless-render-pdf

Unlike "standard" headless chrome, this does not show the annoying auto-generated headers and footers!

Or there is unoconv (which uses LibreOffice behind the scenes) can make pdfs from html:

unoconv -f pdf mypage.html

You can install it on most Linux flavours via the package manager, e.g. apt-get install unoconv

That's nice and easy for simple files. If you need javascript of css support, then use headless Chrome.

Answer 13

I have started to put together a tool to provide a simplified interface to common actions.

You can convert an HTML to a PDF like this:

$ npm install @lancejpollard/act -g
$ act convert tmp/index.html -o tmp/index.pdf -w 2000px -h 3000px

This will create a new PDF for the HTML file.

If nothing else check out the source and see how to write your own script to do this in JavaScript.

Answer 14

wkhtmltopdf and wkhtmltoimage are open source (LGPLv3) command line tools to render HTML into PDF and various image formats using the Qt WebKit rendering engine. These run entirely "headless" and do not require a display or display service.

How to use it?

1. Download a precompiled binary or build from source https://wkhtmltopdf.org/downloads.html

   https://github.com/wkhtmltopdf/wkhtmltopdf

2. Create your HTML document that you want to turn into a PDF (or image)

3. Run your HTML document through the tool.

Usage: wkhtmltopdf input.html output_name.pdf

Answer 15

I often get very good results when using the ebook-convert command line tool that ships with Calibre.

ebook-convert <input.html> <output.pdf>

Check the numerous options for tweaking in the manual. For example, it is possible to automatically generate a table of contents based on H1/H2/... headlines (or anything using XPath expressions, basically).

Please note: Calibre focuses on digital documents and I don't know how well ebook-convert works for very complicated HTML. Worth a try though. :-)

Answer 16

You might want to check out 'Document Conversion Service' by Peernet (at http://www.peernet.com/conversion-software/batch-document-converter/). This runs as a service on a Windows Desktop or Windows Server machine. It opens HTML documents in a web browser, then prints them through a print driver to create PDF documents, so that the PDF document produced looks exactly as if you had printed the HTML document from the browser.