I Blog in Text, You Read in Text || HTML

Inspired by “I blog with raw txt” and “I blog in raw HTML”…

The blog you’re reading was rendered at some point in the past as HTML (maybe) but it was written in text with a dash of Markdown flavor (maybe). However, if you’d like to read the original text version, then go up to the location bar in the browswer and replace the “/” or “html” part of the URL with “.md” and hit enter.¹ I recently replaced a borked Wordpress instance that I maintained for 10+ years with a Bash script that calls out to Pandoc and renders my text as HTML while also inserting some fragments along the way for header and footer glue.

What I do is write a new post into a subdirectory in my blog directory and run:²

find ./posts/<thedir> -name "*.md" -exec sh -c 'pandoc -s \\
--css https://blog.fogus.me/styles/reset.css \\
--css https://blog.fogus.me/styles/index.css \\
-A https://blog.fogus.me/styles/footer.html \\
-B https://blog.fogus.me/styles/header.html \\
-H styles/meta.html -o "${1%.md}.html" "$1"' _ {} \;

I then run a fragment of Clojure code that slurps that filename and title from the new file and adds a new entry into an index.md file and a feed.xml file. I then run a similar command as above on the index.md file and upload all of the new files to my server.

At serve time I have a rewrite rule on a request URL and tries to find an HTML file at that address. If one exists then it serves it. If it doesn’t then it tries to find a .md file and serve it. If it doesn’t find that then it looks for a .txt file or a .org file as a last resort. This seems a bit fiddly and I agree that it is, but it gives me the control that I want and is 100x times less fiddly than the Wordpress instance turned out to be – and I consider that a win.

-fogus