Consultancy - Research Lab - Incubator

based in beautiful Zürich, Switzerland

- 325 words - 2 min read
#tooling

TL;DR Use codegraph to visualize dependencies within a Clojure file. Codegraph applied to itself will for example generate this graph:

example codegraph graph

Here is how it works.

The other day I wished I had a visualization of the dependencies in a piece of ClojureScript code that over the course of the year has gotten a bit unwieldy. I did some thinking and some codeing and it turns out it's quite easy. Here are some of the highlights in code & images, but mostly code.

Reading a string and have it evaluates in Clojure is done with read-string. Reading a string and have it not evaluated is as easy as wrapping the string in an additional set of square brackets.

(read-string (str "[" (slurp path) "]")))

This returns the content of the file given by path as a Clojure data structure. In less homoiconic languages you would probably call this an Abstract Syntax Tree. Homoiconicity for the win!

The data structure is a tree that we want to walk to find dependencies in the code. Here clojure.walk/prewalk comes in handy. We'll pass it a function which will be called for every node in the tree. I went with a multimethod since I initially expected that different typo of nodes would require different action, but it turns out the only thing we're actually interested in are symbols.

If the symbol is member of a given set, namely def, defn, defn-, defmulti, and defmethod, we will pay special attention to the next symbol, because it will be the dependant of the upcoming dependencies that we will record. Every subsequent occurrence of a symbol will then constitute a dependency.

From there the only thing left to do is to limit the dependencies to those where the dependee is also the dependant of another dependency; and render the remainder in dot syntax to be visualized by graphviz.

Check out the code for more details.