error classifier

error-classifer.clj

(ns error-classifier
  (:require [clojure.test :refer :all]))

(def known-errors ["Anonymous access not allowed"
                   "Invalid trace, not on the current datasource"
                   "sync placeholders were requested with a limit of"
                   "constraint \"company_campaign_dismisses_pkey\""
                   "constraint \"company_placeholder_customers_reached_pkey"
                   "constraint \"company_campaign_banner_customers_reached_pkey\""
                   "constraint \"company_channel_customers_reached_pkey\""
                   "constraint \"company_channel_conversions_pkey\""
                   "value too long for type character varying"
                   "Trail already finished on sync page"
                   "Invalid tracking data for trail"
                   "Trail is gone"
                   "Invalid trail identifier"
                   "Trail action without triggering action found"
                   "Trail has already completed for another campaign"
                   "campaign-priorities.invoke" ; import errors (should be warnings)
                   ])

(defn classify [file]
  (let [counters (->> (reduce #(assoc % %2 {:count 0}) {} known-errors)
                      (merge {"other" {:count 0}}))]
    (with-open [rdr (clojure.java.io/reader file)]
      (loop [lines (line-seq rdr)
             counters counters]
        (if (seq lines)
          (let [line (first lines)
                s (some #(re-find (re-pattern %) line) known-errors)]
            (if s
              (recur (rest lines) (update-in counters [s :count] inc))
              (do
                (println "Other error" line)
                (recur (rest lines) (update-in counters ["other" :count] inc)))))
          (->> counters
               (map (fn [[key val]] (merge val {:key key})))
               (sort-by :count)
               reverse
               (every? #(do (println (format "%1$6d: %2$s"  (:count %) (:key %)))
                            true))))))))