search.clj

(ns batcher.search
  (:require [clojure.string :as string]))

(defn tokenise [text]
  (string/split text #" "))

(defn index-document [{:keys [id text] :as document}]
  (let [tokens (tokenise text)]
    (map hash-map tokens (repeat (set [id])))))

(defn index [documents]
  (apply merge-with into
         (apply concat (map index-document documents))))

(defn search [index query]
  (let [query-tokens (tokenise query)]
    (reduce into  #{} (map #(get index % #{}) query-tokens))))

(defn run []
  (let [d1 {:id :foo :text "hello there"}
        d2 {:id :bar :text "hello world"}
        idx (index [d1 d2])]
    [(search idx "hello there")
     (search idx "world")]))