DEV Community


Posted on • Updated on • Originally published at

Customizing pandoc to generate beautiful pdf and epub from markdown

Table of Contents

Either you've already heard of pandoc or if you have searched online for markdown to pdf or similar, you are sure to come across pandoc. This tutorial will help you use pandoc to generate pdf and epub from a GitHub style markdown file. The main motivation for this blog post is to highlight what customizations I did to generate pdf and epub versions for self-publishing my ebooks. It wasn't easy to arrive at the set-up I ended up with, so I hope this will be useful for those looking to use pandoc to generate pdf and epub formats. This guide is specifically aimed at technical books that has code snippets.


I use Ubuntu, as far as I remember, the below steps are enough to work for the demos in this tutorial. If you get an error or warning, search that issue online and you'll likely find what else has to be installed.

I first downloaded deb file from pandoc: releases and installed it. Followed by packages needed for pdf generation.

$ # latest pandoc version as of 21 July 2020 is 2.10
$ sudo gdebi ~/Downloads/pandoc-2.10-1-amd64.deb

$ # note that download size is 750+ MB
$ sudo apt install texlive-xetex
$ sudo apt install librsvg2-bin
$ sudo apt install texlive-math-extra
Enter fullscreen mode Exit fullscreen mode

For more details and guide for other OS, refer to pandoc: installation

Minimal example

Once pandoc is working on your system, try generating a sample pdf without any customization.

See repo for all the input and output files referred in this tutorial.

$ pandoc -f gfm -o sample_1.pdf
Enter fullscreen mode Exit fullscreen mode

Here is input markdown file and -f is used to specify that the input format is GitHub style markdown. The -o option specifies the output file type based on extension. The default output is probably good enough. But I wished to customize hyperlinks, inline code style, add page breaks between chapters, etc. This blog post will discuss these customizations one by one.

pandoc has its own flavor of markdown with many useful extensions — see pandoc: pandocs-markdown for details. GitHub style markdown is recommended if you wish to use the same source (or with minor changes) in multiple places.

It is advised to use markdown headers in order without skipping — for example, H1 for chapter heading and H2 for chapter sub-section, etc is fine. H1 for chapter heading and H3 for sub-section is not. Using the former can give automatic index navigation on ebook readers.

On Evince reader, the index navigation for above sample looks like this:

chapter index

Chapter breaks

As observed from previous demo, by default there are no chapter breaks. Searching for a solution online, I got this piece of tex code:

Enter fullscreen mode Exit fullscreen mode

This can be added using -H option. From pandoc manual,

-H FILE, --include-in-header=FILE

Include contents of FILE, verbatim, at the end of the header. This
can be used, for example, to include special CSS or JavaScript in
HTML documents. This option can be used repeatedly to include multiple
files in the header. They will be included in the order specified.
Implies --standalone.

The pandoc invocation now looks like:

$ pandoc -f gfm -H chapter_break.tex -o sample_1_chapter_break.pdf
Enter fullscreen mode Exit fullscreen mode

You can add further customization to headings, for example use \sectionfont{\underline\clearpage} to underline chapter names or \sectionfont{\LARGE\clearpage} to allow chapter names to get even bigger. Here's some more links to read about various customizations:

Changing settings via -V option

-V KEY[=VAL], --variable=KEY[:VAL]

Set the template variable KEY to the value VAL when rendering the
document in standalone mode. This is generally only useful when the
--template option is used to specify a custom template, since pandoc
automatically sets the variables used in the default templates. If
no VAL is specified, the key will be given the value true.

The -V option allows to change variable values to customize settings like page size, font, link color, etc. As more settings are changed, better to use a simple script to call pandoc instead of typing the whole command on terminal.


pandoc "$1" \
    -f gfm \
    --include-in-header chapter_break.tex \
    -V linkcolor:blue \
    -V geometry:a4paper \
    -V geometry:margin=2cm \
    -V mainfont="DejaVu Serif" \
    -V monofont="DejaVu Sans Mono" \
    --pdf-engine=xelatex \
    -o "$2"
Enter fullscreen mode Exit fullscreen mode
  • mainfont is for normal text
  • monofont is for code snippets
  • geometry for page size and margins
  • linkcolor to set hyperlink color
  • to increase default font size, use -V fontsize=12pt

Using xelatex as the pdf-engine allows to use any font installed in the system. One reason I chose DejaVu was because it supported Greek and other Unicode characters that were causing error with other fonts. See tex.stackexchange: Using XeLaTeX instead of pdfLaTeX for some more details.

The pandoc invocation is now through a script:

$ chmod +x
$ ./ sample_1_settings.pdf
Enter fullscreen mode Exit fullscreen mode

Do compare the pdf generated side by side with previous output before proceeding.

On my system, DejaVu Serif did not have italic variation installed, so I had to use sudo apt install ttf-dejavu-extra to get it.

Syntax highlighting

One option to customize syntax highlighting for code snippets is to save one of the pandoc themes and editing it. See stackoverflow: What are the available syntax highlighters? for available themes and more details (as a good practice on stackoverflow, go through all answers and comments — the linked/related sections on sidebar are useful as well).

$ pandoc --print-highlight-style=pygments > pygments.theme
Enter fullscreen mode Exit fullscreen mode

Edit the above file to customize the theme. Use sites like colorhexa to help with color choices, hex values, etc. For this demo, the below settings are changed:

# by default, background is same as normal text
# change it to a shade of gray to easily distinguish code and text
"background-color": "#f8f8f8",

# change italic to false, messes up comments with slashes
# change comment text-color to yet another shade of gray
"Comment": {
    "text-color": "#9c9c9c",
    "background-color": null,
    "bold": false,
    "italic": false,
    "underline": false
Enter fullscreen mode Exit fullscreen mode

Inline code

Similar to changing background color for code snippets, I found a solution online to change background color for inline code snippets.



Enter fullscreen mode Exit fullscreen mode

Add --highlight-style pygments.theme and --include-in-header inline_code.tex to the script and generate the pdf again.

With pandoc -f gfm -o sample_2.pdf the output would be:

Default syntax highlighting

With ./ sample_2_syn.pdf the output is:

Customized syntax highlighting

For my Python re(gex)? book, by chance I found that using ruby instead of python for REPL code snippets syntax highlighting was better. Snapshot from ./ sample_3.pdf result is shown below. For python directive, string output gets treated as a comment and color for boolean values isn't easy to distinguish from string values. The ruby directive treats string value as expected and boolean values are easier to spot.

REPL syntax highlighting

Bullet styling

This stackoverflow Q&A helped for bullet styling.


% level one
% level two
% level three
Enter fullscreen mode Exit fullscreen mode

Comparing pandoc -f gfm -o sample_4.pdf vs ./ sample_4_bullet.pdf gives:

Bullet styling

PDF properties

This tex.stackexchange Q&A helped to change metadata. See also pspdfkit: What’s Hiding in Your PDF? and discussion on HN.


  pdftitle={My awesome book},
Enter fullscreen mode Exit fullscreen mode

./ sample_4_bullet_prop.pdf gives:

pdf properties

Adding table of contents

There's a handy option --toc to automatically include table of contents at top of the generated pdf. You can control number of levels using --toc-depth option, the default is 3 levels. You can also change the default string Contents to something else using -V toc-title option.

./ sample_1_toc.pdf gives:

table of contents

Adding cover image

To add something prior to table of contents, cover image for example, you can use a tex file and include it verbatim. Create a tex file (named as cover.tex here) with content as shown below:

Enter fullscreen mode Exit fullscreen mode

Then, modify the previous script by adding --include-before-body cover.tex and tada — you get the cover image before table of contents. \thispagestyle{empty} helps to avoid page number on the cover page, see also tex.stackexchange: clear page.

The bash script invocation is now ./ sample_5.pdf.

You'll need at least one image in input markdown file, otherwise settings won't apply to the cover image and you may end up with weird output. used in the command above includes an image. And be careful to use escapes if the image path can contain tex metacharacters.

Stylish blockquote

By default, blockquotes (lines starting with > in markdown) are just indented in the pdf output. To make them standout, tex.stackexchange: change the background color and border of blockquote helped.

Create quote.tex with the contents as shown below. You can change the colors to suit your own preferred style.

\newtcolorbox{myquote}{colback=red!5!white, colframe=red!75!black}
Enter fullscreen mode Exit fullscreen mode

The bash script invocation is now ./ sample_5_quote.pdf. The difference between default and styled blockquote is shown below.

styled blockquote

Customizing epub

For a long time, I thought epub didn't make sense for programming books. Turned out, I wasn't using the right ebook readers. FBReader is good for novels but not ebooks with code snippets. When I used atril and calibre ebook-viewer, the results were good.

I didn't know how to use css before trying to generate the epub version. Somehow, I managed to take the default epub.css provided by pandoc and customize it as close as possible to the pdf version. The modified epub.css is available from the repo. The bash script to generate the epub is shown below and invoked as ./ sample_5.epub. Note that pygments.theme is same as the pdf customization discussed before.


pandoc  "$1" \
        -f gfm \
        --toc \
        --standalone \
        --top-level-division=chapter \
        --highlight-style pygments.theme \
        --css epub.css \
        --metadata=title:"My awesome book" \
        --metadata=author:"learnbyexample" \
        --metadata=lang:"en-US" \
        --metadata=cover-image:"cover.png" \
        -o "$2"
Enter fullscreen mode Exit fullscreen mode

Resource links

More options and workflows for generating ebooks:


Top comments (3)

sassbalint profile image
Sass Bálint

Dear Sundeep, this article is extra useful! Keep up the good work! :)

learnbyexample profile image
Sundeep • Edited

Hey, sorry for the late reply, I haven't visited this site for a while.

Thanks a lot for the feedback :)

I recently updated this blog post on my site, see for details.

shikharsaxena profile image
Shikhar Saxena

Very useful thanks!!