Visualizes Elixir project codebase as a Hierarchichal Edge Bundling graph.

Utilizes Mike Bostock's example implemented in D3.

git clone https://github.com/elixir-ecto/ecto.git /tmp/ecto

docker run --rm -v /tmp/ecto:/project \
    -e MAX_DEPS_DEPTH=1 \
    -e BUILTIN=false \
    -e SHORTEN_MODULE_NAMES=false \
    -e COLOR=lightsteelblue \
    studzien/graf:latest /project \
    > ecto.svg

After it finishes, ecto.svg should look similar to this:

./priv/generate.sh is an entrypoint to the script. It takes Elixir projects directories as arguments and outputs a SVG with Hierarchichal Edge Bundling graph representing structure of these projects.

Generation has two phases:

1. An Elixir script (priv/graf.exs) first generates JSON with relationships between modules from provided projects and for given options;
2. A Node.js script (priv/heb/index.js) then generates a SVG from the JSON.

You can run this script locally, and not in the Docker container, but make sure you have Node available and run make setup first, that will install all JS dependencies.

All configuration of the generation script is currently done via environmental variables.

The available variables are:

• MAX_DEPS_DEPTH (default: 0) - how deep the generator should look into the dependencies when generating graph?
• BUILTIN (default: false) - should built-in modules (like Enum or erlang) be considered when generating the graph?
• SHORTEN_MODULE_NAMES (default: true) - should the module name on the graph be shortened to the last part? (i.e. Graph instead of Graf.Graph)
• COLOR (default: #ccc) - the color of edges on the graph

Since the generator relies on running mix in your project to gather information like compile paths and checks whether it's an umbrella project, it might not work properly if there is some output in the mix run output, like logs.

A workaround for this is increasing log level or turning logging off for the time of graph generation.

If a graph can't be generated for your project or seems to be incorrect please create an issue describing the problem.