core.clj

(ns domlink-neo4j-import.core
  (:require [clojure.string :as s])
  (:gen-class))

(defn get-name-id
 [name-str {:keys [names next-id] :as r :or {names {} next-id 1}}]
 (let [id (names name-str)]
   (if id
     [id r]
     (let [id next-id
           next-id (inc next-id)]
       [id {:names (assoc names name-str id) :next-id next-id}]))))

(defn process-data
  [inf outf]
  (binding [*flush-on-newline* false
            *out* outf]
    (->> (line-seq inf)
         (map #(s/split % #"\s+"))
         (reduce
          (fn [names [from to link-count]]
            (let [[from-id names] (get-name-id from names)
                  [to-id names] (get-name-id to names)]
              (println from-id to-id link-count)
              names))
          {})
         ;; we have the get-name-id map now, get the names from it
         :names
         (map #(println (s/join "\t" %)))
         ;; map is lazy, so make sure we process all operations
         doall)
    nil))

(defn -main
  "Reads an input file with lines containing
        from-name to-name num-count
and replaces names with unique ids.

XXX output id->name mapping in the end"
  ([inname]
     (with-open [inf (clojure.java.io/reader inname)
                 outf (clojure.java.io/writer java.lang.System/out)]
       (process-data inf outf)))
  ([inname outname]
     (with-open [inf (clojure.java.io/reader inname)
                 outf (clojure.java.io/writer outname)]
       (process-data inf outf))))