just parse it as EDN, I'd say

Just to check we are on the same page: Say I have the file foo.clj

(ns foo)

(defn foo
  []
  (inc 1))

[{:arglist-strs ["[]"]
  :col 1
  :defined-by clojure.core/defn
  :defined-by->lint-as clojure.core/defn
  :end-col 11
  :end-row 5
  :filename "/Users/brandonstubbs/projects/x/src/foo.clj"
  :fixed-arities #{0}
  :name foo
  :name-col 7
  :name-end-col 10
  :name-end-row 3
  :name-row 3
  :ns foo
  :row 3}]

Ah sorry, the entire definition. My recommendation is to use rewrite-clj and iterate over the nodes, until the positions match up. And then just use str on the node, this will give you the definition.

here is a similar function, but then the opposite: it removes definitions from the original code https://github.com/borkdude/carve/blob/cb621317ae1582869c8bd8f1a47cf57a3598d803/src/carve/impl.clj#L118

Ok great thank you I will do that, is this something you do not want in scope of this lib?

not in scope for clj-kondo. it throws away all whitespace so it doesn't even know the original node anymore

I would prefer the whitespace removed, what I am looking for is a quick way to say has this definition changed since this file was analysed

why not just look at the substrings based on the entire file

you don't even need rewrite-clj for this

Busy writing something that does a bit of tree shaking for building, similar to turbopack in the js world. But on second thought, I think whitespace should matter as if someone adds a newline that’s important

ok, then you can just focus on the lines and columns that clj-kondo reports and check the contents in the file

just using raw string comparison

Yeah that will actually be great! Thanks for the discussion 😄