path: root/content/blog/2024-03-29-org-blog.org

#+date: <2024-03-29>
#+title: Blogging in Org-Mode 
#+description: 
#+slug: org-blog

First and foremost, apologies to those who subscribe via RSS as I know that my
feed duplicated itself when I moved this blog over to org-mode last night.

This post focuses specifically on the configuration and tools I use to blog from
Emacs with Org-Mode and does not focus on Emacs or Org-Mode themselves. Refer to
the post I wrote about [[https://cmc.pub/blog/doom-emacs-org-mode.html][Doom Emacs & Org-Mode]] for more information about my base
Emacs configuration.

* Weblorg

The first step in blogging with Org-Mode is to choose a method to convert the
source files to HTML and publish them. The Worg site maintains a nice list of
[[https://orgmode.org/worg/org-blog-wiki.html][Blogs and Wikis with Org]], but the tools are inevitably different and
opinionated, so you'll need to find what works for you.

I tried using Jekyll, Hugo, ox-hugo, Nikola, Blorg, org-static-blog, and the
native org-publish functions before finally settling on Weblorg. For one reason
or another, the other solutions were a drastic step down from my previous
workflow that used [[https://www.getzola.org/][Zola]] with Markdown content.

[[https://github.com/emacs-love/weblorg][Weblorg]] is a static site generator for [[https://orgmode.org/][org-mode]], built for use within [[https://www.gnu.org/software/emacs/][Emacs]].
Since it's written in Emacs Lisp, there's no need to install other languages or
frameworks to get started. More than that, you can write in any editor you
please and simply invoke the Emacs build process with the =--script= parameter
instead of requiring you to blog inside Emacs.

** Installation

The [[https://emacs.love/weblorg/doc/index.html][Getting Started]] page details broad installation requirements. I am using
Doom Emacs on macOS, which requires you to add the package to the
=~/.doom.d/packages.el= file and configure the =publish.el= file slightly
differently.

To start, add the =htmlize= and =weblorg= packages to Doom, sync the changes,
and reload.

#+begin_src sh
nano ~/.doom.d/packages.el
#+end_src

#+begin_src lisp
(package! htmlize)
(package! weblorg)
#+end_src

#+begin_src sh
doom sync
#+end_src

Either re-open Emacs or hit =SPC h r r= to reload the changes.

** Configuration

Now that I've installed weblorg, I need to configure the project. I'll start by
navigating to my site's source code and creating a =publish.el= file.

#+begin_src sh
cd ~/Source/cmc.pub && nano publish.el
#+end_src

Since I'm using Doom, Emacs will not automatically load the packages I need
later in the build process. To compensate, my =publish.el= file needs to
explicitly tell Emacs where Doom stores the =htmlize=, =weblorg=, and
=templatel= packages.

#+begin_src lisp
;; explicity load packages since I'm using Doom Emacs
(add-to-list 'load-path "~/.emacs.d/.local/straight/repos/emacs-htmlize")
(add-to-list 'load-path "~/.emacs.d/.local/straight/repos/weblorg")
(add-to-list 'load-path "~/.emacs.d/.local/straight/repos/templatel")
(require 'htmlize)
(require 'weblorg)

;; defaults to http://localhost:8000
;; To build with the custom URL below, call:
;;;; ENV=prod emacs --script publish.el
(if (string= (getenv "ENV") "prod")
 (setq weblorg-default-url "https://cmc.pub"))

;; site metadata
(weblorg-site
 :theme nil
 :template-vars '(("site_name" . "cmc.pub")
                  ("site_owner" . "hello@cmc.pub")
                  ("site_description" . "Just a blip of ones and zeroes.")))

;; route for rendering the index page of the website
(weblorg-route
 :name "index"
 :input-pattern "content/index.org"
 :template "index.html"
 :output ".build/index.html"
 :url "/")

;; route for rendering each blog post
(weblorg-route
 :name "blog"
 :input-pattern "content/blog/*.org"
 :template "post.html"
 :output ".build/blog/{{ slug }}.html"
 :url "/blog/{{ slug }}.html")

;; route for rendering the index page of the blog
(weblorg-route
 :name "blog-index"
 :input-pattern "content/blog/*.org"
 :input-aggregate #'weblorg-input-aggregate-all-desc
 :template "blog.html"
 :output ".build/blog/index.html"
 :url "/blog/")

;; route for rendering each wiki post
(weblorg-route
 :name "wiki"
 :input-pattern "content/wiki/*.org"
 :template "post.html"
 :output ".build/wiki/{{ slug }}.html"
 :url "/wiki/{{ slug }}.html")

;; route for rendering the index page of the wiki
(weblorg-route
 :name "wiki-index"
 :input-pattern "content/wiki/*.org"
 :input-aggregate #'weblorg-input-aggregate-all
 :template "wiki.html"
 :output ".build/wiki/index.html"
 :url "/wiki/")

;; routes for rendering all other pages
(weblorg-route
 :name "pages"
 :input-pattern "content/*.org"
 :template "page.html"
 :output ".build/{{ slug }}.html"
 :url "/{{ slug }}.html")

(weblorg-route
 :name "salary"
 :input-pattern "content/salary/*.org"
 :template "page.html"
 :output ".build/salary/{{ slug }}.html"
 :url "/salary/{{ slug }}.html")

(weblorg-route
 :name "services"
 :input-pattern "content/services/*.org"
 :template "page.html"
 :output ".build/services/{{ slug }}.html"
 :url "/services/{{ slug }}.html")

;; RSS Feed
(weblorg-route
 :name "rss"
 :input-pattern "content/blog/*.org"
 :input-aggregate #'weblorg-input-aggregate-all-desc
 :template "feed.xml"
 :output ".build/feed.xml"
 :url "/feed.xml")

;; route for static assets that also copies files to .build directory
(weblorg-copy-static
 :output ".build/{{ file }}"
 :url "/{{ file }}")

;; fire the engine and export all the files declared in the routes above
(weblorg-export)
#+end_src

* Project

** Structure

The project structure for weblorg is highly customizable and the main
restriction is that the =publish.el= file must point to the correct paths.

For my blog, I prefer to keep the blog content out of the top-level directory.
This results in the following structure (shortened for brevity):

#+begin_src txt
.build/
content/
    blog/
        example-blog-post.org
        index.org
    wiki/
        example-wiki-post.org
        index.org
    index.org
    other-example-page.org
theme/
    static/
        styles.css
        robots.txt
    templates/
        base.html
        blog.html
        index.html
        page.html
        post.html
        wiki.html
build.sh
publish.el
#+end_src

This is simply my preferred structure and you can alter it to fit your needs.
The key here really is that you can customize at will, as long as the
=publish.el= file matches.

** Build & Deploy

Once you're content with the status of the project, you're ready to build and
deploy the blog.

My process utilizes a =build.sh= script that combines the steps I take every
time.

#+begin_src sh
touch build.sh && chmod +x build.sh && nano build.sh
#+end_src

Within this script, I do the following:

1. Remove any files within the =.build= directory that I use to store published
   files.
2. Set the environment variable to =prod= to ensure the =base_url= matches my
   configuration in =publish.el=.
3. Build the site with Emacs & =publish.el=.
4. Use =scp= to copy files to my site's public directory on my server.

#+begin_src sh
rm -rf .build/* && \
ENV=prod emacs--script publish.el && \
scp -r .build/* ubuntu:/var/www/cmc.pub/
#+end_src

*** Time to Build

My only current complaints are:

1. Errors messages are not helpful. It takes work to determine what the error is
   and where it's coming from. I generally have to sit and watch the build
   process to see the file that weblorg pubslishes right before the error
   occurred.
2. The build process re-builds every single file on each run, which takes a long
   time for a blog of my size. See below for the last time I measured.

#+begin_src sh
> time ./build.sh

./build.sh  35.46s user 0.59s system 85% cpu 41.965 total
#+end_src

Overall, I have thoroughly enjoyed using weblog and will continue to use it
going forward until I find something better.