Pandoc Bash Blog

Pandoc Bash Blog provides pbb, a simple generator for static blog sites based on Pandoc and Bash. Progress is chronicled at https://www.benjaminwuethrich.dev.

Installation

Use

make install

to install the executable, the tab completion, the stylesheet, the Lua filter, and the syntax highlighting theme. Installation follows the XDG Base Directory Specification; this means:

• ~/.local/bin has to be in in your $PATH (as per systemd file hierarchy)
• Bash completion has to be configured such that it dynamically looks up completions in $XDG_DATA_HOME/bash-completion/completions; $XDG_DATA_HOME defaults to ~/.local/share
• Assets such as the stylesheet are installed to $XDG_DATA_HOME/pbb
• The man page gets installed to $XDG_DATA_HOME/man; make sure your man finds it there
• The Lua filter for dot graphs is installed to the default location at $XDG_DATA_HOME/pandoc/filters
• The syntax highlighting theme is installed to $XDG_DATA_HOME/pandoc

There is an option to create symlinks instead of copying files; this is useful for development so changes to the original are immediately effective. To do so, set the DEVMODE variable:

make install DEVMODE=1

Dependencies

These are the versions I use on my development machine; some things might break for older versions. make install checks if the executables exist, but not their versions.

• Bash 5.0.17
• Pandoc 2.10
• Git 2.30.0
• GNU Coreutils 8.32: cat, cp, mkdir, ln, rm, tac
• GNU Sed 4.7
• ImageMagick 6.9.10-23 (for favicon)
• Python 3.8.6 (for pbb serve)
• inotify-tools 3.14 (for hot-reloading during pbb serve)
• Bats 1.2.1 (for test suite)
• bash-completion 2.11 (for tab completion)
• graphviz 2.43.0 (for dot graphs)

In the Makefile, additionally:

• GNU Make 4.3
• GNU Awk 5.0.1
• column
• GNU Coreutils 8.32: install, rmdir

Usage

The authoritative source for usage instructions is man pbb.

Initialize a new blog with title "My blog" in an empty Git repository:

git init
pbb init 'My blog'

If you later want to change the title, use

pbb title 'My blog with a new title'

pbb init creates a sample blog post. Blog posts are written in Pandoc Markdown (see also my post about it), with filenames formatted like YYYY-MM-DD-post-title.md.

The first heading of a post has to be a level-one heading:

# A blog post

Images must be stored in the images directory.

To build your blog, run pbb build. This extracts all the titles into an index file; the index file links to all files following the naming convention above and lists them in reverse alphabetical order, with the newest post at the top.

To have a look at your freshly built blog, use pbb serve and point your browser to http://localhost:8000. While pbb serve is running, any changes to .md files or files in the images directory trigger a rebuild or copy of that file, allowing to preview a post by just reloading it in the browser instead of building the whole site over and over again.

Once you think your opus magnum is ready to be published, run pbb deploy. This pulls the generated HTML files into the master branch, commits and then pushes them to a remote. This works for GitHub pages deploying the master branch.

You might have to set the Git remote first:

git remote add origin https://github.com/<yourname>/<repo-name>.git

Table of contents

To get a table of contents for a post, start it with a YAML document that sets the toc variable to true:

---
toc: true
...

Math

To use MathJax to render inline and display math, turn the math feature on:

pbb enable math

Bibliography

To use the pandoc-citeproc filter to get citations and bibliographies, enable the bibliography feature:

pbb enable bibliography

dot graphs

Code blocks with class .dot are replaced by the dot graph they describe. To additionally get a caption, use a caption attribute; to include the graph description as an HTML comment in the output, add the .includeSource (or .include-source) class.

``` {.dot .includeSource caption="A dot graph"}
digraph G {
    a -> b
}
```

Like for other fenced code blocks, if .dot is the only class, the opening line can be just ```dot.

Favicon

To get a favicon, pbb build checks the assets directory for a file named favicon.* and resizes it to a 32x32 PNG image. If there is no such file, more than one, or if it is not an image file ImageMagick can handle, pbb warns about this and continues.

Analytics

Pbb integrates with GoatCounter, a pretty awesome simple web statistics solution. Open an account, add your site and set your code with

pbb gccode <yourcode>

Consider paying for a custom domain or sponsoring the author.

Subcommands

There are eight subcommands (nine, if you count pbb help); when properly installed, they should tab-autocomplete.

pbb init <title>

• Creates new source branch and checks it out
• Adds artifacts directory to .gitignore
• Creates the includes, images and assets directories
• Symlinks the stylesheet from $XDG_DATA_HOME/pbb/pbb.css
• Creates header files in includes that are used on every page, for blog title link, favicon, Google web font links and GoatCounter analytics
• Creates an example post

pbb title <new title>

• Sets a new blog title for an existing blog

pbb enable <feature>

• Turns on a feature
• Options are math and bibliography

pbb disable <feature>

• Turns off a feature

pbb gccode <code>

• Includes a snippet with tracking code for GoatCounter on each page, where the code for the account is <code>
• To turn tracking off, set code to empty with pbb gccode ''

pbb build

• Cleans the artifacts directory, then copies the images directory in there
• Checks for a favicon image and, if there is one, generates the favicon from it (see Favicon)
• Generates the index file, index.md
• Converts the markdown files with datestamps in their names and index.md to HTML, copies the results into artifacts

pbb serve

• Serves the blog on http://localhost:8000 to preview
• Listens to file changes in .md files and the images directory and rebuilds/copies the corresponding files as long as pbb serve is running

pbb deploy

• Checks out the master branch
• Deletes everything but the artifacts directory (and the CNAME file, if you have one)
• Copies the contents of artifacts into the repository root directory
• Adds, commits and pushes everything to the remote
• Checks out the previous branch