google.clj

(def punc-regex #"\p{P}")

(defn wordify [s]
  (map #(clojure.contrib.string/replace-re punc-regex "" (clojure.contrib.string/lower-case %1)) (.split s " ")))

(defn collect [pairs]
  (loop [remPairs pairs
         dict {}]
    (if (empty? remPairs)
      (seq dict)
      (let [key (first (first remPairs))
            val (last (first remPairs))]
        (if (contains? dict key)
          (recur (rest remPairs)
                 (assoc dict key
                        (cons val (get dict key ))))
          (recur (rest remPairs) (assoc dict key [val])))))))

(defn map-reduce [coll mapper reducer]
  (map
   (fn [pair]
     [(first pair)
      (reduce reducer [] (last pair))
      ])
   (collect (map mapper coll))))


(defn reducer [rv cv]
  (if (some #{cv} rv)
    rv
    (cons cv rv)))

(defn mapper [word]
  [(apply str (sort word)) word])


(defn anagrams 
  "find all anagrams in sentence"
  [sentence]
  (filter #(> (count %1) 1) (map last (map-reduce (wordify sentence) mapper reducer))))