core.clj

(ns riaktest.core
  (:require [clojure.string :as string]
            [clj-time.core :as time-core]
            [clj-time.format :as time-format]
            [clojurewerkz.welle.core :as wc]
            [clojurewerkz.welle.mr :as mr]
            [clojurewerkz.welle.kv :as wkv]
            [clojurewerkz.welle.buckets :as wbuckets]))

(wc/connect!)

;;(load "riaktest/core")(require '[riaktest.core :as rt])
;;(require '[clojurewerkz.welle.kv :as wkv])
;;(require '[clojurewerkz.welle.buckets :as wbuckets])

;; create the bucket like this:
;;(wbuckets/update "pages")

;; add a dummy page with 5 different retrieval states:
(comment
  (dotimes [n 5]
    (let [datestamp (rfc3339-datestamp!)]
      (rt/store-page! "pages"
                      "http://example.com/"
                      datestamp
                      {:uri "http://example.com/"
                       :datestamp datestamp
                       :value (str n)})
      (Thread/sleep 2000))))

;; utility fn to get rid of test data:
(comment
  (wkv/delete-all "pages" (wbuckets/keys-in "pages")))

(defn rfc3339-datestamp!
  "Returns the RFC3339 current time (e.g. 2012-07-09T09:31:01.579Z)."
  []
  (time-format/unparse
    (time-format/formatters :date-time)
    (time-core/now)))

(defn store-page!
  "Stores the provided value in given bucket, using a key which starts
   with the uri argument, which is followed the ^ character and
   finally an RFC3339 datestamp."
  [bucket uri datestamp value]
  ;; application/clojure is another possible :content-type,
  ;; but pr-str with text/plain gives much cleaner results
  ;; (i.e. a string value of {:a :b} instead of
  ;; "#=(clojure.lang.PersistentArrayMap/create {:a :b})"
  (wkv/store bucket
             (str uri "^" datestamp)
             (pr-str value)
             :content-type "text/plain"))

(def kv-js-fn "function(obj) { return [[obj.key, obj.values[0].data]];}")

(defn get-page-states
  "Accepts a riak bucket and the uri for the requested page. Returns a
   sequence of pairs, with the first element in each pair being the
   datestamp for the individual state and the second element its
   value."
  [bucket uri]
  (map (fn [[k v]]
         [(second (string/split k #"\^"))
          (read-string v)])
       (mr/map-reduce
        {:inputs {:bucket bucket
                  :key_filters [["starts_with" (str uri "^")]]}
         :query [{:map {:language "javascript"
                        :source kv-js-fn
                        :keep false}}
                 {:reduce {:language "erlang"
                           :module "riak_kv_mapreduce"
                           :function "reduce_sort"}}]})))

;; example values returned by the get-page-states fn:
(comment
  [["2013-03-17T12:23:01.036Z"
    {:uri "http://example.com/",
     :datestamp "2013-03-17T12:23:01.036Z",
     :value "0"}]
   ["2013-03-17T12:23:03.041Z"
    {:uri "http://example.com/",
     :datestamp "2013-03-17T12:23:03.041Z",
     :value "1"}]
   ["2013-03-17T12:23:05.049Z"
    {:uri "http://example.com/",
     :datestamp "2013-03-17T12:23:05.049Z",
     :value "2"}]
   ["2013-03-17T12:23:07.056Z"
    {:uri "http://example.com/",
     :datestamp "2013-03-17T12:23:07.056Z",
     :value "3"}]
   ["2013-03-17T12:23:09.064Z"
    {:uri "http://example.com/",
     :datestamp "2013-03-17T12:23:09.064Z",
     :value "4"}]])