core.clj

(ns pyppejson.core
  (:require [cheshire.core :refer :all])
  (:gen-class))

(def user-paths
  [[:user]
   [:retweeted_status :user]
   [:retweeted_status :quoted_status :user]
   [:quoted_status :user]])

(defn filter-users [tweet path fields]
  (update-in tweet path select-keys fields))

(defn filter-tweet [tweet paths]
  (reduce #(filter-users %1 %2 #{:id_str :name}) tweet paths))

(defn get-all-users [tweet paths]
  (let [all-users (map #(get-in tweet %) paths)]
    (zipmap (map :id_str all-users) all-users)))

(defn -main
  [& args]
  (let [tweet (parse-string (slurp "example-tweet.json") true)
        tweet-lite (filter-tweet tweet user-paths)]
    (println "EXTRACTED USERS")
    (println (get-all-users tweet-lite user-paths))
    (println "WHOLE TWEET WITH LITE USERS")
    (println (generate-string tweet-lite))))

example-output.txt

EXTRACTED USERS
{5 {:id_str 5, :name Quoter}, 1 {:id_str 1, :name Pyppe}, 2 {:id_str 2, :name Retweeter}, 0 {:id_str 0, :name Petra Sievinen}}
WHOLE TWEET WITH LITE USERS
{"filter_level":"low","timestamp_ms":"1433966461293","quoted_status":{"id_str":"4","user":{"id_str":"5","name":"Quoter"},"geo":null,"favorited":false,"retweeted":false,"possibly_sensitive":false,"filter_level":"low","lang":"fi"},"entities":{"hashtags":[{"text":"lovedungeon","indices":[57,70]}],"trends":[],"urls":[{"url":"https://t.co/abc","expanded_url":"https://twitter.com/diipadaapa/status/1234","display_url":"twitter.com/diipadaapa/sta…","indices":[72,95]}],"user_mentions":[{"screen_name":"foo","name":"Foo bar","id":102025211,"id_str":"102025211","indices":[0,13]}],"symbols":[]},"retweeted_status":{"hip":"hei","user":{"id_str":"2","name":"Retweeter"},"quoted_status":{"some":"property","user":{"id_str":"1","name":"Pyppe"}}},"lang":"fi","truncated":false,"retweeted":false,"id_str":"608725574924808192","favorited":false,"user":{"id_str":"0","name":"Petra Sievinen"},"retweet_count":0,"favorite_count":0,"created_at":"Wed Jun 10 20:01:01 +0000 2015","text":"Testing"}