This article explains in detail the process of setting up a bare-bones website using Hugo and Org mode. My goal in writing this is to provide readers with a superior understanding of the fundamentals of this workflow. It is by no means an exhaustive explanation of Org mode or Emacs, but should give readers of any skill level a strong foundation to apply their own knowledge and techniques to the Emacs-Hugo toolchain.
I assume only beginner-level knowledge of Emacs.
Intro & Setup
Kaushal Modi created ox-hugo on top of his ox-blackfriday package, providing an impressive amount of features for organizing blog text and linked data with Hugo. He maintains great documentation and ample examples for using the package. I will explain my own workflow here, but for an exhaustive (though terse) reference, I highly recommend Modi’s test site and post source Org file, which contain demonstrations and tests for all of ox-hugo’s features.
After issuing the Emacs command
M-x package-install RET ox-hugo RET, you’ll
require it. You can do this by running
M-: (require 'ox-hugo), but
you’ll want to add it to your configuration as explained here. Once this is
done, using ox-hugo is just a matter of making an Org file and writing
content. Org’s format is very straightforward, and is designed to make sense to
the reader even if they’re unfamiliar with the formal syntax. For instance,
* My food | Where's My Food? | Fridge | Counter | Mouth | Total | | Oranges | 1 | 3 | 0 | :=vsum($2..$4) | | Marshmallows | 0 | 100 | 20 | :=vsum($2..$4) | | Brussel Sprouts | 32 | 4 | 0 | :=vsum($2..$4) |
Produces a dynamic spreadsheet table in Org mode that exports to HTML like this:
|Where’s My Food?||Fridge||Counter||Mouth||Total|
If you’re already familiar with Org mode, the benefits are obvious and creating content is fairly trivial. Org mode is, however, a complex and expansive program with many features, and its learning curve can appear daunting at first glance. Using ox-hugo is a great way to learn the format, since it gives the author a command-center view of their entire content hierarchy, much like a traditional database, but in a flat format that’s much easier to read and understand. Org features present themselves naturally, and the author can easily visualize the correspondence between the Org format and the output on their webpage.
Just take a look at the Org file for this webpage. Search for “ox-hugo is super cool!” and you should find this very paragraph.
Eventually you’ll want to read the manual, though. You may access it in Emacs
Making a New Blog
Compared to a generic Org file, the only necessary data that ox-hugo needs to
properly export to Hugo is an
:EXPORT_FILE_NAME: property in the
:PROPERTIES: block of an Org heading.
:PROPERTIES: blocks are common in Org
for defining arbitrary metadata about sections, and ox-hugo uses them to
generate Hugo’s front matter (used for associating a title, header, or other
custom data with the page it generates). Providing an
definition signals to ox-hugo that a particular heading is available for export
to Hugo. For example, the
:PROPERTIES: block of the page you’re currently
reading looks like this:
:PROPERTIES: :EXPORT_FILE_NAME: ox-hugo-tutorial :EXPORT_DESCRIPTION: Exporting to Hugo's Blackfriday Markdown from Orgmode :EXPORT_HUGO_IMAGES: /img/org.png :END:
:EXPORT_DESCRIPTION: variables are optional
definitions allowed by the Speedy theme of this website, but the filename is the
only required property for ox-hugo. Our goal here is to organize the structure
of our website as a tree using Org headers. So, as a minimal example, here’s
what a new site might look like in its entirety:
The Org file can be placed in any directory so long as
identifies the Hugo project’s root directory. This path definition is required
for any valid ox-hugo file, and in the example above uses
to specify that the base directory will be the same path as this Org file. If
you saved this file as hugotest.org, exported it with Org’s exporter
and selected the Hugo output
H and the All Subtrees To Files option
wind up with the following files in your directory:
. ├── content │ ├── bad-night.md │ └── _index.md └── hugotest.org
Most sites will be more than a blog, though, and will want multiple sections. In fact, many sites are made up of nothing but a slew of sections that users navigate between with some built-in menu. So a more functional minimal example would be the following:
Which yields the following files on export:
. ├── content │ ├── _index.md │ └── posts │ ├── bad-night.md │ └── _index.md └── hugotest.org
As you might expect if you’re already familiar with Hugo, this structure adheres to the Hugo content management scheme. Additionally, the index files have been marked with menu metadata, which allows Hugo themes to automatically generate navigation menus from the markdown files. Hereafter, making new blog posts is as simple as adding new sub-headings under the “Blog Posts” heading, and exporting. As you can see, this is suitable for defining the hierarchical structure of any general website, not just blogs. Org mode and Hugo just make creating new pages so simple and well-structured that providing content is all that’s required for a new page, blog entry, or entirely new site section. If you can blog with ox-hugo, you can deftly deploy any manner of web content, or even develop entire websites as naturally as you make blog posts. Any tool that can turn blogging and web development into the same task is quite an achievement!
Of course, themes to style this content are another can of worms entirely, but we’ll get to that soon. It is sufficient for now to mention that Hugo makes using themes as easy as downloading one and specifying it in Hugo’s config file.
One question you may ask is why the blog’s homepage is not defined in the Blog
Posts heading. This is a fair question! Any heading with an
:EXPORT_FILE_NAME: property will export all of that heading’s content,
including subheadings beneath it. This allows Org headings to be used as part
of the content of a post, where they will be exported as markdown heading
levels, which translate to HTML heading elements
Furthermore, properties other than
:EXPORT_FILE_NAME: are inherited by
sub-headings, including the
:EXPORT_HUGO_MENU: properties. A
:EXPORT_HUGO_MENU: property at the section root would cause all exported files
within that section to be added to the menu specified. This might be intended by
the content creator, but most likely you don’t want every single post you make
to be in the main menu. So it makes sense to define all your pages, including
the index, as a sub-heading of the section definition (which merely specifies
which sub-directory the content will output to).
To illustrate, let’s assume you want to extend the previous site definition with a section about fishsticks. We’ll do this the “wrong way” first to show how Org handles inheritence:
In this example, we’ve defined the main homepage of the section inside the tier-1 heading for Fishsticks. This is technically valid, and produces the expected file output:
. ├── content │ ├── fishsticks │ │ ├── gortons.md │ │ ├── _index.md │ │ └── van-de-camps.md │ ├── _index.md │ └── posts │ ├── bad-night.md │ └── _index.md └── hugotest.org
But on inspection of the gortons.md file, we find the anomoly mentioned above:
Uh oh! Not only did these fishsticks give us herpes, they are now part of the main menu. Worse, when the index page was exported, each of the subsequent posts became part of its content:
This explains the flexibility of ox-hugo’s straightforward parsing
rules. Specifically, that any headings with an
:EXPORT_FILE_NAME: tag will
export everything beneath them as content. The content organization in this
erroneous example duplicates data, but might still be useful if you wanted to
create, for instance, an “all_content” page for the section. In general, though,
be sure to put your index pages in subheadings (just as you do with normal
pages) so that the tier-1 heading can be used for “global” definitions that
affect all of the pages. A correct section for fishsticks should look like
Now the homepage for the fishsticks section has a heading all its own, just like
any other page. That’s better! Now our homepage will output the content only
from its subheading, and the other pages don’t inherit the homepage’s
properties. All pages inherit the
:EXPORT_HUGO_SECTION: fishsticks property
though, which is what we want to ensure that these pages are exported to the
At this point, setting up Hugo and publishing is simple. Installing Hugo is pretty straightforward on any Unix-like system with a package manager; it is available on most distributions at this point. Windows installation is a bigger pain in the ass, but you should be used to that if you’re still in the stone-age.
hugo new site . on the command-line will create a new hugo site in the
current directory, but
hugo expects to be creating a new directory with this
command and will complain if it already exists. It also provides the
option to allow creating a new site in an extant directory, but this too will
fail if the content subdirectory already exists (which ox-hugo will create
when you export).
So you have three choices:
hugo new site /path/to/some-new-dirand move your Org file to this new directory
rm -Rf content/to remove the content directory ox-hugo created, then run
hugo new site --force .
- don’t even bother with the
hugo new sitecommand, and make a config.toml file manually (the only file really required for Hugo to run).
It’s convenient to do this through the
hugo command because it will create
Hugo-specific subdirectories like archetypes, layouts, themes, etcetera, in
addition to populating a basic config.toml file. The subdirectories it creates
aren’t necessary, but help illustrate Hugo’s structure. In any case, you’ll want
to wind up with a directory structure something like this (created with option 2
above, extending from previous examples):
. ├── archetypes │ └── default.md ├── config.toml ├── content ├── data ├── hugotest.org ├── layouts ├── static └── themes
Exporting with ox-hugo using
C-c C-e H A again will, as expected, fill the
content directory with our content.
. ├── archetypes │ └── default.md ├── config.toml ├── content │ ├── fishsticks │ │ ├── gortons.md │ │ ├── _index.md │ │ └── van-de-camps.md │ ├── _index.md │ └── posts │ ├── bad-night.md │ └── _index.md ├── data ├── hugotest.org ├── layouts ├── static └── themes
The last thing to do here is to download or create a theme for Hugo. As mentioned before, installing a theme is very simple. This blog uses a custom theme named Speedy that I have been developing to help myself learn Hugo’s internals, but for this example I’ll be using Kaushal Modi’s bare-min theme. The bare-min theme is the best starting place out there for making new themes, and outputs basic HTML pages without any need to mess with CSS or JS. It also provides easy debugging facilities and search features.
So let’s install it! You can download the theme from its github page and extract
it to the themes folder, or much more easily use git to clone it to your themes
git clone https://github.com/kaushalmodi/hugo-bare-min-theme.git
themes/bare-min Then open up your config.toml file, and add the theme.
Be sure that the theme’s name matches the theme directory’s name in the themes/ directory of your project base directory. (e.g. themes/bare-min here). That’s it for installing the theme.
Now, running the command
hugo with no subcommands will invoke the Hugo
generator on the current directory, and output finalized content in the
. ├── archetypes │ └── default.md ├── config.toml ├── content │ ├── fishsticks │ │ ├── gortons.md │ │ ├── _index.md │ │ └── van-de-camps.md │ ├── _index.md │ └── posts │ ├── bad-night.md │ └── _index.md ├── data ├── hugotest.org ├── layouts ├── public │ ├── categories │ │ ├── index.html │ │ └── index.xml │ ├── css │ │ └── github_chroma.css │ ├── fishsticks │ │ ├── gortons │ │ │ └── index.html │ │ ├── index.html │ │ ├── index.xml │ │ └── van-de-camps │ │ └── index.html │ ├── index.html │ ├── index.xml │ ├── js │ │ └── search.js │ ├── page │ │ └── 1 │ │ └── index.html │ ├── posts │ │ ├── bad-night │ │ │ └── index.html │ │ ├── index.html │ │ └── index.xml │ ├── sitemap.xml │ └── tags │ ├── index.html │ └── index.xml ├── static └── themes ...
Hugo, by default, generates xml files that are suitable for RSS feeds. With a theme installed, Hugo will produce more suitable web content (usually HTML) to be served over HTTP. The bare-min theme outputs HTML, provides CSS for doing chroma-based syntax highlighting (in case you include code blocks), and inline styles for basic page formatting. Generated pages also have a lot of useful debugging information. You’ll also notice that Hugo has generated folders for “categories” and “tags”. These are default organization labels for your content called taxonomies.
The taxonomy index pages allow users to browse content by category or tag. These taxonomies correspond to Org mode tags, and ox-hugo will automatically associated tagged headings with the tags taxonomy, or the categories taxonomy if prefixed with an @ symbol. You are free to define your own taxonomies, and even disable the default “tags” and “categories” taxonomies, but since Org mode tags directly translate to the default Hugo taxonomies, it makes sense to just use the default taxonomies for now.
As an example of taxonomies, I’ll add some tags and categories to our hugotest.org file to create a complete blog structure with tags and categories:
Exporting hugotest.org with
C-c C-e H A and generating with
hugo will yield
the same file structure as before, but this time we’ll see that the categories
and tags directories have sections for our newly added taxonomies.
. └── public ├── categories │ ├── index.html │ ├── index.xml │ ├── reviews │ │ ├── index.html │ │ └── index.xml │ └── updates │ ├── index.html │ └── index.xml └── tags ├── fear │ ├── index.html │ └── index.xml ├── herpes │ ├── index.html │ └── index.xml ├── index.html └── index.xml
You can now serve the public/ directory over an HTTP server. Hugo is packaged
with an internal HTTP server to help with testing, which is quite convenient
because it can automatically refresh whenever content in its content/ directory
is updated (so when you export from ox-hugo, you don’t have to run
again). To use it, simply run
hugo server and point your browser at
http://localhost:1313 (1313 is the default
--port argument for
Eventually you’ll want to move on to other themes, or develop your own, but at this point you’ve got a fully functional blog publishing workflow from start to finish that you can view in-browser as you develop.
Attaching Files, Capturing Information & Automation
Once you have a basic site structured in your Org file, you’re ready to start throwing information in it. It is of course sufficient to open the Org file and edit it, but most Org mode users prefer to automate everything, and being able to use Org’s capture feature to instantly populate new blog posts is extremely convenient.
The ox-hugo documentation provides succinct explanations on how to do this,
including elisp snippets for capture setup, image linking, and automating
exports when you save your Org file (so no more need to
C-c C-e H A every
time, just save the file as usual with
Indexes and Page Resources
You may be wondering why our index pages are exported as _index rather than index. Hugo uses a concept called Page Bundles to organize exported content. The gist of this is that a file named index is known as a “Leaf Node” and cannot have any children. A file named _index is considered a “Branch Node” and allows nesting other bundles beneath it. In other words, an Org heading with an exported file name of index will be treated as a single page with no subfolders. This is useful for single pages, but a section index (e.g. for a blog) with many subpages and other resources will more than likely want to allow nested bundles beneath it.
You may export an Org heading as a Page Bundle by providing the Org property
:EXPORT_HUGO_BUNDLE: with an argument (string) that will become the name of
the folder created. If you do this, you will need to set the
:EXPORT_FILE_NAME: property to either index for Leaf Nodes, or _index for
The capture setup provided by Kaushal Modi above provides methods to automatically create either a normal page, or a leaf node bundle when invoking Org Capture.
Drafts and Automatic Timestamps
By default, Hugo will not build any markdown files whose front-matter properties
draft: true. This is very convenient for in-progress posts that you
leave in your Org file, or in the content/ directory.
Ox-hugo will always fill out the draft property, and by default every exported
header will have its draft property set to false. However, ox-hugo also links
this behavior to the TODO feature of Org. When you cycle a heading’s TODO value
S-<RIGHT> (that’s Shift + Right Arrow Key), you will signal to ox-hugo to
export this heading as a draft (i.e.
draft: true), which will prevent Hugo
from building it into an HTML page.
When a heading is cycled to the DONE state in Org, it will automatically
generate a timestamp for when the heading was closed. Ox-hugo will export DONE
draft: false and, better still, will use Org’s timestamp to fill
out the Date property in the markdown file. This makes it trivial to manage
writing multiple posts at once, and automatically timestamp completion dates.
You may also explicitly set this date parameter with the
property, but the ease of using DONE-state switching is pretty hard to pass up.
Renaming Tags and Other Properties
If a theme you are using has any idiosyncrasies about your naming conventions (e.g. if you export your content to more than one site using more than one theme), ox-hugo provides a convenient way to automatically replace any key values on export. This can be done on a per-heading, or a per-file basis.
To replace keys for the entire file, simply add a property to the top of your Org file. For example:
This will make any
:EXPORT_DESCRIPTION: properties export, instead, to a
“summary” key in the front-matter of your output markdown file. It will also be
able to replace exported values in the Org body:
#+begin_description This is the description, it has multiple lines too! It will export as the Summary value in front-matter #+end_description
To do this on a per-heading basis, simply add the
:EXPORT_HUGO_FRONT_MATTER_KEY_REPLACE: property to a heading’s property block,
and the replacements will only occur within that heading.
Why not use Hugo’s internal Org parser?
It’s true that Hugo has an internal Org parser that is well maintained. It provides this as an alternative to markdown files. You may wonder, then, why someone would use Org mode to export to markdown instead of just letting Hugo parse the Org files itself. The answer is two-fold:
- Hugo’s Org format is currently less feature complete than markdown, so exporting from Org mode to the Hugo Org Format would limit potential output.
- Org mode is a lot more than just a file format, and its integration with your system allows all kinds of benefits you can never get out of a simple Org file parser.
Therefore, supporting the Org format in another tool will give you a superior text format for organizing information, but it will be crippled when compared to an actual Org mode implementation. Ox-hugo gives you the ability to use Org mode itself to generate content.
If Hugo’s Org parser gains parity with, or eclipses, the Blackfriday Markdown format currently used by Hugo, ox-hugo could certainly be used to output those Org-Hugo files instead of the current markdown. This would be nice because it would allow Org mode users to view their output content more easily, but the advantages of ox-hugo and a real, bona fide Org mode would still remain.
So you see, Hugo’s Org parser isn’t really in competition with ox-hugo, it’s in competition with the other Hugo parsers (e.g. markdown).
Thanks to Kaushal Modi, who found this article on the googs within days of me posting it, for reaching out to me and providing thorough feedback and error checking.
And special thanks to me, for once again overcoming Hamhock the Laziness Demon who has possessed me since birth and hates it when I do anything productive.