core.clj

(ns bucketlogs.core
  (:use clojure.java.io))

(defn re-groupmap
  "Return a hash-map of group names and matches generated by computing regex
matches for named groups."
  [ptn names s]
  (let [matcher (.matcher ptn sample)]
    (.find matcher)
    (let [groups (for [name names]
                   [(keyword name) (try (.group matcher name)
                                        (catch IllegalStateException e
                                          nil)
                                        (catch IllegalArgumentException e
                                          "No group found"))])]
      (into (hash-map) groups))))

(def groups ["ownerid" "bucketname" "timestamp" "remotehost" "requester"
             "requestid" "operation" "key" "requesturi" "status" "error"
             "bytessent" "objectsize" "totaltime" "turnaround" "referrer"
             "useragent"])

(def ptn (re-pattern
          (str "(?s)^(?<ownerid>[\\w\\d]+) "
               "(?<bucketname>[\\w\\d-]+) "
               "\\[(?<timestamp>\\d{2}/[A-Za-z]{3}/[\\d:]+) [\\-\\+]\\d{4}\\] "
               "(?<remotehost>[\\d\\., -]+) "
               "(?<requester>[\\w\\d\\:/]+) "
               "(?<requestid>\\w+) "
               "(?<operation>[A-Z\\.]+) "
               "(?<key>[\\w/\\.]+) "
               "\\\"(?<requesturi>[\\w/\\. ]+)\\\" "
               "(?<status>\\d{2,3}) "
               "(?<error>[\\w\\-]+) "
               "(?<bytessent>\\d+|\\-) "
               "(?<objectsize>\\d+) "
               "(?<totaltime>\\d+) "
               "(?<turnaround>\\d+) "
               "\\\"(?<referrer>[^\\\"\\s]+|\\-)\\\" "
               "\\\"(?<useragent>[^\\\"]+)\\\" *")))

(with-open [rdr (reader "resources/sample.log")]
  (doall (map (partial re-groupmap ptn groups) (line-seq rdr))))