get_app_names_by_id.clj

(require '[cheshire.core :as json]
         '[clj-http.client :as client]
         '[tech.ml.dataset :as ds]
         '[flatland.ordered.set :refer [ordered-set]]
         )

;; used a qubole query to get all app-id's 
(defonce ds (ds/->dataset "/tmp/full_result_407911382.csv"))

;; ordered by # of impressions. 
(def ordered-app-id-numbers 
    (->> (get (ds/sort-by-column "impressions"  > ds) "app_id")
        vec
        (apply ordered-set )
        (remove nil?)
        (filter #(clojure.core/re-matches #"^[0-9]+$" %))))

(defn lookup-app-name-by-id2
    [app-id-string] 
    (let [response (client/get (str "https://itunes.apple.com/lookup?id=" app-id-string))
          body (json/parse-string (:body response))
          app-name (get-in body ["results" 0 "trackName"])          ]
       app-name))

;; write to csv. st
;; such that if api blocks I have everything thus far in file. 
;; had to restart couple of times cause the 
;; api gave me error 503.

(defn strip-name [name]
    (clojure.string/replace name #"[^\w-\s\:\–\&\!\.\(\)\|]" ""))

(spit "app-id-name2.csv" "app_id, name, stripped_name \n")
(doseq [app-id-number  ordered-app-id-numbers]
  (let [app-name (lookup-app-name-by-id2 app-id-number)]
    (Thread/sleep 100)
    (spit "app-id-name2.csv" (str app-id-number 
                                  ", " 
                                  app-name
                                  ", "
                                  (strip-name app-name)
                                  "\n") 
          :append true)))