Some experimental work on using Pandoc, LaTeX and Markdown to create court documents from version-controllable plain text sources.
Build the Markdown document in foo.md:
> ./build foo
You'll either get a nice foo.pdf, or an incomprehensible LaTeX error message.
build is just a short shell script which calls pandoc and xelatex, so
you'll need those on your PATH.
You can modify build so that pandoc is responsible for choosing its own
LaTeX engine, but I wanted to use nice fonts and couldn't get pandoc to call
XeTeX.
Use order.md or subs.md as a starting point. You can write anything that's
supported in Pandoc-flavoured Markdown, including any extensions enabled in the
build script. I'm using yaml_metadata_block, citations and footnotes.
All documents are assumed to be court documents relating to the same action, defined
in header.yaml. The structure of this YAML file is pretty self-explanatory; Pandoc
feeds this YAML to the data you add in the block at the top of your Markdown file into
courtdoc.template.tex, so the allowable keywords are all variables in that template.
The template is defined in courtdoc.template.tex and is based on the style
used in Western Australia. You can see how the keys in header.yaml and the
YAML metadata block at the start of each document are fed into variables like
$foo$. There are also a few special template keywords like $if$, $for$
and $sep$. This code isn't really LaTeX, it's a
little-documented
Pandoc DSL which is compiled to an intermediate .tex file before build
invokes xelatex.
This templating language is fairly limited. A more sophisticated system might
involve Haskell scripts which parse custom Markdown features using
Text.Pandoc.JSON. See the Pandoc scripting
documentation.