Lucenalog: Datalog interface to Lucene in 10 lines

logicrels-lucene.clj

(ns lucenalog.core
    "Lucenalog = Datalog interface to Lucene in 10 lines.

Simple but powerful.

Use

  (db/add (index) {:a \"foo\" :b \"bar\"})

to index a map with Lucene.  Then you can use the relation
'lucenalog-rel' from core.logic.  That relation works on maps.  Bound map
values produce a Lucene query, which returns matching maps.  See
'lucenalog-test' for an example."


  (:require [clucy.core :as db]
            [clojure.core.logic :as logic]))

;; (defproject lucenalog "0.0.1-SNAPSHOT"
;;   :description "Lucenalog = Datalog interface to Lucene in 10 lines"
;;   :dependencies [[org.clojure/clojure "1.2.1"]
;;                  [clucy "0.3.0"]
;;                  [org.clojure/core.logic "0.6.6"]
;;                  ]
;;   :main lucenalog.core)


(set! *warn-on-reflection* true)

;; Accessory functions.  Not important.

(let [m (ref {:max-query-results 1024
              :verbose true})]
  (defn config
    "Lucenalog configuration.  Call with no args to see the current
configuration.  Get a configuration property's value by passing the
property to this function.  Change the config using the two-argument
dispatch."
    ([] @m)
    ([k] (@m k))
    ([k v] (dosync (alter m assoc k v)))))


(defonce index ;; Get the default Lucene index.
  (let [i (db/memory-index)]
    (fn [] i)))

(defn- note [& args]
  "Println the args if (config :verbose).  Return the last arg."
  (when (config :verbose)
    (apply println :note args))
  (last args))


;; Generate the Lucene query string.

(defn- lucene-query
  "Query Lucene based on the given object and Substitutions.  The
query Q should be a map that has some values bound by substitutions A.
The generated Lucene query string looks like 'p1:v1 AND p2:v2', where
Q contains :p1 lv1 and :p2 lv2 and the substitutions take lv1 to v1 and
lv2 to v2.  That query string is then used in the Lucene query."
  ([q a]
     (db/search (index)
                (let [walked
                      (logic/walk* a q)
                      query
                      (reduce str
                              (interpose " AND "
                                         (map (fn [[k v]]
                                                (str (name k) ":" v))
                                              (remove (comp logic/lvar? last)
                                                      walked))))]
                  (note :lucene-search walked q a :query query))
                (config :max-query-results))))

;; The core.logic relation.

(defn lucenalog-rel [q]
  "A clojure.core.logic relation backed by Lucene.  Lucene query
generated by lucene-query based on the given map."
  (fn [a]
    (logic/to-stream
     (map #(logic/unify a % q)
          (lucene-query q a)))))


;; Tests and examples.

(defn lucenalog-test
  "Simple check that Lucenalog is working.  Returns true if things
look okay."
  ([]
     ;; Let's use our own in-memory Lucene index.
     (binding [index (let [i (db/memory-index)]
                       (fn [] i))]
       (with-open [i ^org.apache.lucene.store.Directory (index)]
         ;; Add a little data.
         (doseq [m [{:a "a1" :b "z1"}
                    {:a "z1" :b "b1"}
                    {:a "a1" :b "z2"}
                    {:a "z1" :b "b2"}]]
           (db/add i m))
         ;; Check to see if we can chase a1 to b1 and b2 via z1.
         (let [expect #{"b1" "b2"}
               got (set
                    (logic/run* [q]
                                (logic/fresh [a b c]
                                             (logic/== a "a1")
                                             (lucenalog-rel {:a a :b b})
                                             (lucenalog-rel {:a b :b c})
                                             (logic/== q c))))
               passed? (= expect got)]
           (println :test passed? :got got :expected expect)
           passed?)))))

(defn -main ([& args] (lucenalog-test)))

Thanks! Updated.