markov.clj

(ns alice-clj.core)

(def text (slurp "resources/19033.txt"))
(def words (clojure.string/split text #"[^A-Za-z]+"))

(def pairs (partition 2 1 (map clojure.string/lower-case words)))

(defn add-pair [m [w1 w2]]
  (update-in m [w1 w2] (fn [count]
                         (if count (inc count) 1))))


(def freqs (reduce add-pair {} pairs))

(def sorted-freqs (into {} (map (fn [[w1 m]]
                                  [w1 (map first (sort-by second > m))]) freqs)))

(defn next-word [sorted-freqs w]
  (let [words (get sorted-freqs w)
        word (rand-nth (take 3 words))]
    word))

(defn upcase-word [s]
  (clojure.string/join (conj (rest s) (Character/toUpperCase (first s)))))

(defn sentence [n]
  (let [start (rand-nth words)
        [first-word & next-words] (take n (iterate (partial next-word sorted-freqs) start))]
    (clojure.string/join " " (conj next-words (upcase-word first-word)))))


(sentence 6)
Will be a little thing i
alice-clj.core> (sentence 8)
Said to be off with a low hall
alice-clj.core> (sentence 23)
Go and the project gutenberg tm work is to alice s the king said alice s the king said to alice in the
alice-clj.core>