Some Clojure functions to help read custom access log files into maps

gistfile1.clj

(defn comment? [s]
  (.startsWith s "#"))

(defn not-comment? [s]
  (not (comment? s)))

(defn remove-comments [line]
  (filter not-comment? line))

(defn nil-if-hyphen [s]
  (if (not= s "-") s))

(defn str->int 
  "Returns an int if the string parses as an int, otherwise returns input unaltered"
  [str]
  (if (re-matches (re-pattern "\\d+\\.?\\d+") str) 
    (read-string str)
    str))

;; #Fields: date time s-ip cs-method cs-uri-stem cs-uri-query s-port cs-username c-ip cs(User-Agent) sc-status sc-substatus sc-win32-status time-taken
(defn is-format-line? [s]
  (= (.substring s 0 8) "#Fields:"))

(defn find-first-format-line [lines]
  (first (filter is-format-line? lines)))

(defn read-format-into-keywords [s]
  (map keyword (filter not-comment? (clojure.string/split s #" "))))

(defn read-format-from-file [f]
  (let [file-contents (line-seq (clojure.java.io/reader (clojure.java.io/file f)))]
    (read-format-into-keywords (find-first-format-line file-contents) )))

(defn zipmap-line-data
  "Returns a map with the keywords mapped to data from a log line."
  [columns line]
   (let [line-data (map str->int (re-seq #"[^ ']+|'[^']*'" line))]
    (zipmap columns line-data)))

(defn read-data-from-file [file]
  (let [columns (read-format-from-file file)
        file-without-comments (remove-comments (line-seq (clojure.java.io/reader (clojure.java.io/file file))))]
    (map #(zipmap-line-data columns %1) file-without-comments)))

(def summarize (juxt :time :c-ip :time-taken :cs-uri))

;;; Example usage
(def data (read-data-from-file "path/to/file.log"))
(def longest200 (take 200 (reverse (sort-by :time-taken data))))
(pprint (sort (map summarize longest200)))