most used keywords in codebase

analyse_code.clj

(ns cnd.analyse-code
  
  "Use rewrite-clj zipper to walk code.
  
  Niggles
  * Doesn't use the (ns...) aliases to resolve aliases to full paths.
 
  "
  (:require [clojure.java.io :as io]
            [clojure.zip :as zip]
            [rewrite-clj.zip :as z]))

(defn find-all
  [zloc pred]
  (let [*ks (atom nil)]
    (-> (zip/up zloc)
        (z/postwalk
          (fn select [zloc]
            (let [expr (z/sexpr zloc)]
              (when (pred expr)
                (swap! *ks conj expr)))
            nil)))
    @*ks))

(defn file? [f] (.isFile f))
(defn get-path [f] (.getPath f))

(defn all-source-files
  [dir-path]
  (->> (io/file dir-path)
       (file-seq)
       (filter file?)
       (map get-path)))

(comment (all-source-files "src"))

(defn all-keywords
  [file-paths]
  (mapcat (fn [file-path]
            (try
              (find-all (z/of-file file-path) qualified-keyword?)
              (catch Exception e
                ;(println file-path)
                ;(println e)
                nil)))
          file-paths))

(defn most-used-keywords
  [dir-path]
  (->> (all-source-files dir-path)
       all-keywords
       (frequencies)
       (sort-by val)
       (reverse)
       (take 20)))

(comment (most-used-keywords "src/cljs"))