clojure.clj

(ns fuzzy-schema.core-test
  (:require [schema
             [core :as s]
             [utils :as utils]]
            [schema.spec.core :as spec]))

(def sample-schema
  {:zookeeper/configuration {:zookeeper/address s/Str
                             :zookeeper/port s/Num
                             }
   :onyx/onyx-id s/Str
   (s/optional-key :w) s/Num})

(defn levenshtein [str1 str2]
  "a Clojure levenshtein implementation using transient data structure"
  (let [n (count str1) m (count str2)]
    (cond
      (= 0 n) m
      (= 0 m) n
      :else
      (let [prev-col (transient (vec (range (inc m)))) col (transient [])] ; initialization for the first column.
        (dotimes [i n]
          (assoc! col 0 (inc i)) ; update col[0]
          (dotimes [j m]
            (assoc! col (inc j)  ; update col[1..m]
                    (min (inc (nth col j))
                         (inc (nth prev-col (inc j)))
                         (+ (get prev-col j) (if (= (nth str1 i) (nth str2 j)) 0 1)))))
          (dotimes [i (count prev-col)]
            (assoc! prev-col i (get col i)))) ;
        (last (persistent! col)))))) ; last element of last column

(defn get-distances [k possible-keys]
  (mapv (juxt #(levenshtein (str k) (str %)) identity) possible-keys))

(def sample-data {:zookeeper/configuration {:zookeeper/address "100"
                                            :zookeeper/ports 99}
                  :onyx/onyx-id "HI"})

(defn fuzzy-match-walk [schema]
  (spec/run-checker
   (fn [s params]
     (let [checker (spec/checker (s/spec s) params)]
       (fn [data]
         (if-let [err (utils/error-val (checker data))]
           (mapv (fn [[k v]]
                   (when (= v 'disallowed-key)
                     (println k " Is an invalid key."
                              "Did you mean: "
                              (mapv second (sort-by first
                                                    (get-distances k s))))))
                 err)
           data))))
   true
   schema))


                                        ;((fuzzy-match-walk sample-schema) sample-data)