Histogram

histogram.clj

(ns histogram.core
  (:gen-class)
  (require [criterium.core :as criterium]
           [clojure.pprint :as pprint]))

(defn do-sort-map [x]
  (into
   (sorted-map-by
    (fn [key1 key2]
      (compare [(get x key2) key2]
               [(get x key1) key1])))
   x))

(defn -main
  [& arg]
  (let [args              (apply array-map arg)
        file-path         (args "-f")
        number-of-threads (Integer/parseInt (args "-t"))
        benchmark         (Integer/parseInt (or (args "-b") "0"))
        text              (slurp file-path)
        part-size         (inc (quot (count text) number-of-threads))
        parts             (mapv #(apply str %) (partition-all part-size text))]

    (println "File path:" file-path)
    (println "Number of threads:" number-of-threads "-b" benchmark)
    (println "-----------------------------------------")

    (if (> benchmark 0)
      (let [start (. System (nanoTime))]
        (println "Starting benchmark")
        (println "Running example" benchmark "times.")
        (dotimes [n benchmark]
                    (time (apply merge-with + (mapv deref (mapv #(future (frequencies %)) parts)))))
        (println "Average runtime was" (/ (double (- (. System (nanoTime)) start)) (* 1000000.0 benchmark)) "ms"))

      (pprint/print-table
       (map (fn [[key value]] {"character" key "frequency" value})
            (do-sort-map (time (apply merge-with + (mapv deref (mapv #(future (frequencies %)) parts)))))))))

  (shutdown-agents))