This Python script generates directed graphs from LaTeX documents to show dependencies between theorems.

Originally, it was written for use in my diploma thesis (which is finished now), but I may continue development if there is interest.

The documentation is taken from the appendix of my thesis and currently only available in German. But looking at the source code, you should be able to quickly see what happens.

HOW TO USE:

First, look at the configuration options in the beginning of the script file.

Make sure GraphViz is installed, then execute: generate-graphs.py your-file.tex

Also, you will have to do some changes to your LaTeX document, to produce good and readable results. Define the following commands and use them in your theorem environments to fine-tune the graph:

\newcommand{\graphref}[1]{\unskip\vphantom{\ref{#1}}} - Adds a \ref which only shows up in the graph \newcommand{\graphcaption}[1]{\unskip} - Sets the caption of the corresponding node \newcommand{\invref}[1]{\unskip\vphantom{\ref{#1}}} - Adds an invisible edge to the graph (can be used to influence the layout)

What you should know before usage:

• Numbers in the graph will only be correct if you have organized your file into chapters and use chapter-level numbering of theorems. Otherwise you will have to change the code related to numbering.
• \includes are being followed, but only one level. Deeper includes won't currently work.
• When you run the script for the first time, expect no miracles. It can take some time to make a grap look good.