Span is a static-site generator based on Pandoc. It generates a static site from a folder of markdown sources. You can also easily generate a RSS feed or other non-HTML generated files using Span.

Currently, the only way to install this is using cabal:

git clone https://github.com/PerpetualCreativity/span.git
cabal install --overwrite-policy=always

And make sure to have ~/.cabal/bin in your path.

./
    content/
    templates/
    output/
    span.yml

The content of the website goes in content/; your paths are a direct reflection of the folder structure (e.g. /content/blog/interesting.md becomes output/content/blog/interesting). Span supports using DocTemplate syntax within a content file; however, due to limitations in Pandoc, Span employs a workaround: inside content files, use % for DocTemplate syntax instead of $.

Templates are Pandoc templates used to render your content.

Span puts the resulting files in output/.

When you specify a template, Span uses the following logic for figuring out which template to use.

Consider the following site structure:

./
    content/
        index.md
        about.md
        blog/
            interesting.md
            awesome.md
            cool.md
        notes/
            contents.md
            2021/
                note.md
            2022/
                contents.md
                note1.md
                note2.md
    snippets/
        navbar.html
        blog/
            navbar.html
        notes/
            author.html
    templates/
        default.html
        blog/
            default.html
        notes/
            default.html
            contents.html

Span will use the closest matching template. Span considers files in the closest matching directory, looking in the parent only if necessary, and using a matching name (without extension), or default.html templates when possible. Consider notes/2022/note1.md. Span looks under templates, but can't find templates/notes/2022/, so it jumps to the parent, templates/notes, and looks there. This has a default.html, so Span will use this template to render notes/2022/note1.md. Below is a table of sample files and what template it will use:

Thus, you must have at least a default.html in templates/ for Span to work. Use pandoc -D html > templates/default.html to use the default pandoc template (though I recommend using a custom template).

If you want to use a partial in your templates, put them in the same directory as the template:

partials will be sought in the directory containing the main template, and will be assumed to have the extension of the main template

• Pandoc DocTemplates documentation

The config file is by default span.yml (you can specify another name through the command-line options):

# Ignore files in contents/ that match these glob expressions.
ignore:
- drafts/**

# Pass-through (just copy, don't run pandoc) on files in contents/ that
# match these glob expressions.
passthrough:
- games/**
- experiments/**

# Passed to every template.
globals:
  name: Span
  cool: Yeah!

# Passed to the corresponding template when this file is run.
# Equivalent to Pandoc metadata, except these can be passed to multiple files
# The actual metadata in the files (if any) override these values.
variables:
  posts:
    files:
    - blog.html
    - rss.xml
    data:
    - title: Hello
      description: Cool stuff here
      path: posts/asdf.html
    - title: Very interesting?
      description: More cool stuff here
      path: posts/asdf2.html
  news:
    files:
    - news.html
    - news.xml
    data:
    - message: Awesome news!
      date: 1970-01-01
    - message: Very cool news!
      date: 1970-01-02

# Span, by default, treats files with the `.md` extension as markdown, and
# files with the `.html` extension as html. If you want to add recognition
# for more extensions, fill this out.
# These strings are the recognized input names:
# https://hackage.haskell.org/package/pandoc-3.1.6/docs/src/Text.Pandoc.Readers.html#readers
extensions:
  md: markdown
  html: html
  # you can leave the above two out if you're just using the default values.
  xml: html # useful for RSS feeds
  typ: typst

# Run filters against files in contents/ that match these globs.
# Relative paths to filters are relative to the directory span is run in.
filters:
  "**/":
  - Lua: filters/commonfilter
  - Lua: filters/anotherfilter
  notes/*/math/**/:
  - JSON: ~/path/to/pandoc-asciimath2tex

This is the glob syntax.

TODO. I'd like to put an RSS feed template here, or even a "full" example.