Markpdf: a best in class Markdown -> PDF tool

Offtopic
1

Markpdf is currently part of GitHub - ralsina/markterm · GitHub (along with markterm and markmark)

It’s a markdown->PDF converter based on litehtml and libharu

It parses markdown using markd then takes the HTML and uses libhtml to do layout and finally libharu to create a PDF.

Some of the highlights:

  • Can create a 600 A4 page PDF of Brothers Karamazov in under 5 seconds.
  • Tables
  • Footnotes
  • Annotations
  • Fenced code blocks with syntax highlighting
  • Inline and display math (cheating via libtexprintf
  • Fully justified with hyphenation (includes implementation of the TeX hyphenation algorithm)
  • Outline based on headings
  • Multiple image types, including svg (requires external tool)
  • Inline HTML rendered
  • HTML blocks rendered
  • emoji, accents, CJK
  • Page breaking with widow heading prevention
  • Pageless mode: all your PDF is a single page. Yes, you can do a page with all of Brothers Karamazov in it.
  • CSS stylesheets (a few included, user can add customized)
  • Color schemes

There isn’t AFAIK, a tool with this much feature coverage that doesn’t involve installing chrome or TeX or Typst. Markpdf is a 9MB static binary.

Here is an example of how the output looks:

image
image680×552 97.6 KB

2

As someone who uses Pandoc with Docker, I’m somewhat interested in this tool, but I’m not sure if it’s right for me. A website where you can view sample PDFs created with markpdf might be helpful for people who are on the fence about trying it out.

3

There will be, at some point.

4

There is now. Demo with full server-side rendering, can tweak the contents, look, templates, whatever.

5

Great hat-tip to @sdogruyol and the kemal project, which makes this sort of thing easy and enjoyable to create

6

I made a similar tool many years ago: GitHub - straight-shoota/md2pdf: Convert markdown files to pdf with flexible template and stylesheet · GitHub
It’s built on Crystal with markd as well, so I suppose it ticks most of the above features.
But I used wkhtmltopdf for rendering. So not the entire Chrome browser, but the webkit engine. I don’t think that exists anymore. It was perhaps a bit easier to just delegate layout and rendering to a single library.

On top, it also integrates Crinja for templating. Maybe that could be an inspiration for markpdf? :wink:

7

Emojis are broken :(

image
image608×130 15.7 KB

8

Probably lacking a font in the docker image :-)

I’ll look into it

9

I am half decided to support short codes (which are a way to embed crinja templates) but honestly it was so much work getting it to this stage I am hesitant.

10

Fixed!

11

Just noticed, this is where I got the idea! Thanks @hugopl !!!

12

Thank you for implementing something I wanted but would never have time to do so… at least until my kids grow.

13

Mine’s 19, so keep hope! :slight_smile:

14

imagen
imagen641×190 8.22 KB

The footnote is not on the foot :blush: but maybe is the issue with a text that span just a tiny bit over a page?

15

They are technically endnotes :-)

16

This command-line tool is handy, and I like the concept behind it. However, I’ve run into a few minor issues when trying to use it on macOS or when working with Japanese text, so I might submit a few pull requests in the future.

17

There is no text shaping so support for some scripts is … iffy.

18

Now with decent math:

image
image828×1150 37.4 KB

And without TeX!

Ok, it links in MicroTeX, which is a C++ library and does a bridge to libharu so the output looks nice in the PDF

19

@ralsina is it possible to use it with a set of formatting rules, in my case for scientific / academic papers, when generating the PDF from a markdown?

20

It supports CSS (within limits)