MarkdownLectures

Convert a single markdown file into lectures, slides, and outlines with key terms.

An explanation of the philosophy is here, although the lecture templates there no longer work. I've packaged a script in src that converts from the format there (around code blocks) to the new one (around fenced divs).

This code does not create the slide deck directly; rather, it writes to the pandoc AST in a format that should create slides in any of Pandoc's built-in slide formats. I generally use it with reveal-js slides; certain options (such as the handling of fullscreen slides) will only work with reveal-js classes.

This repo is periodically updated to work with the latest build of Pandoc, which has a frequently-changing API. The release numbers are tied to the version of pandoc I expect this to run with.

Quickstart

From the main directory, the following commands will take the demonstration lecture at lecture.md and create a website you can open at slides.html; and a PDF outline at outline.pdf. I don't know if this is possible without a working Haskell installation locally.

pandoc --filter src/lectureToSlidedeck.hs -t revealjs lecture.md --standalone -o slides.html -V revealjs-url=http://lab.hakim.se/reveal-js
pandoc --filter src/lectureToOutline.hs lecture.md -o outline.pdf

Installation

If running this frequently, it's faster to compile the filters to executables.

As of version 2.7.1 of this project, I do so use using 'stack.' With this directory located locally, just enter 'stack install' to install the binaries locally. (To make these universally accessible, I add the following line to .stack/config.yaml: local-bin-path: /usr/local/bin/.

Writing format

1. Lectures are written as Markdown.

2. Anything in a fenced code div classed as a 'slide' will become a slide. For example.

   These are lecturer notes.
   
   :::slide
   This will appear on the *screen*.
   :::
   
   

3. Any standalone images (i.e., images that take up a paragraph on their own) will become a fullpage slide: the image caption will be turned into large text on the screen.

   
   ![Big white label](image.png)
   
   ![Big, red text](image.png){.red}
   
   

   (In practice, the way this works is by turning these into level 2 headers, so styling is inherited from them.) Image attributes are passed to the underlying text; so the special class '.red' on the bottom image makes the text be read. Colors supported are '.red', '.black', and '.blue'. (No color leads the text to be white.) Leave the image description blank for no label.

4. Any level 1 headers (# Introduction) are kept in the outline, and also used as the title for a slide. I use reveal's two-level setting, so these act as the column heads.

Compiling

If you do this a lot, you may want to compile the pandoc filters rather than run the scripts, which greatly increases the speed of the conversions. The makefile will install them to a source on disk (by default, /usr/local/bin/lectureToSlidedeck and /usr/local/bin/lectureToOutline). Then you just use those locations instead of the ones above.

Making more elaborate slides.

Pandoc now allows cool things like two column slides. Those will work with this: just insert more fenced divs inside your slide iv.