ext_en_th.clj

(require '[clojure.data.xml :as xml]
         '[clojure.string :as str]
         '[clojure.java.io :as io])

(defn ext-segs [segs]
  (->> segs
       (mapcat :content)
       (str/join " ")))

(defn ext-tuv [tuv]
  [(-> tuv
       :attrs
       :xml/lang
       keyword)
   (->> tuv
       :content
       (filter #(= (:tag %) :seg))
       ext-segs)])

(defn ext-tu [tu]
  (->> tu
       :content
       (filter #(= (:tag %) :tuv))
       (mapcat ext-tuv)
       (apply array-map)))

(defn extract [path]
  (let [doc (xml/parse path)
        body (-> doc :content (nth 1) :content)
        tu-list (->> body
                      (filter #(= (:tag %) :tu)))]
    (map ext-tu tu-list)))

(defn valid-pair? [p]
  (and (> (count (:en p)) 0)
       (> (count (:th p)) 0)
       (not (re-matches #"^\s+$" (:en p)))
       (not (re-matches #"^\s+$" (:th p)))))

(with-open [w-en (io/writer "open.en")
            w-th (io/writer "open.th")]
  (doseq [path ["opensub/en-th.tmx"
                "gnome/en-th.tmx"]]
    (with-open [r (io/reader path)]
      (->> (extract r)
           (filter valid-pair?)
           (map #(do
                   (prn %)
                   (.write w-en (:en %))
                   (.write w-en "\n")
                   (.write w-th (:th %))
                   (.write w-th "\n")))
           doall))))