gistfile1

(ns cascalog-tutorial.core
  (:use cascalog.api
        cascalog.playground
       [cheshire.core :only [parse-string]]))

(def my-tap
  (hfs-tap (cascading.scheme.hadoop.TextLine.)
           (str (System/getProperty "user.dir")
                "/records.json")))

(defn count-data [json]
  (let [record (parse-string json true)]
      (count (:data record []))))

(defn -main []
  (?<- (stdout) ; destination
     [?c] ; projection
     (my-tap _ ?line)
     (count-data ?line :> ?c))) ; pattern