gistfile1.clj

;; Totally untested, but should give an idea.

(ns mn
  (:use lstoll.utils) ; https://github.com/lstoll/clj-utils
  (:require [clj-http.client :as http])) ; https://github.com/dakrone/clj-http

(def test-concurrency 10)

(defn test-url
  "Tests a URL, returning it's status code"
  [url]
  (:status (http/get url {:throw-exceptions false}))) ; Does a HTTP get, getting the results status

(defn test-urls
  "Taks a seq of URLs. Returns a lazy seq of results {:url URL :status HTTP-STATUS}"
  [urls-to-test]
  (pmap2 test-concurrency ; call to pmap2. Set concurrency to the saved var
         (fn [u] {:url u :status (test-url u)}) ; Map function - call test-url, return with URL in map
         urls-to-test)) ; Run this on the param

(defn broken-urls
  "Prints broken URLS."
  [urls]
  ;; Force evaluation of a sequence. The binding vector is like let.
  (doseq [err-url (filter #(>= (:status %) 500) ; filter for all 500+ status codes
                          (test-urls urls))]  ; in the result of testing all the urls
    (println (str "URL " (:url err-url) " is returning a status of " (:status err-url)))))

(broken-urls ["http://google.com" "http://microsoft.com"])