gengo03.clj

(ns gengo03
  (:require [cheshire.core :as cheshire]
            [clojure.java.io :as io]))

;; https://nlp100.github.io/ja/ch03.html

;; 20
(def r (io/reader (io/resource "jawiki-country.json")))
(def jsons (doall
            (map (fn [line]
                   (cheshire/decode line))
                 (line-seq r))))
(def uk (first (filter (fn [data] (= (get data "title") "イギリス")) jsons)))
(def uk-text (get uk "text"))


;; 21
(doseq [[category-line] (re-seq #"\[\[Category\:(.*?)\|*.*\]\]" uk-text)]
  (println category-line))

;; 22
(doseq [[_ category] (re-seq #"\[\[Category\:(.*)\]\]" uk-text)]
  (println (clojure.string/replace category #"\|.*" "")))

;; 23
(doseq [[m section] (re-seq #"==+(.*?)==+" uk-text)]
  (let [e (re-find #"=+" m)
        c (dec (count e))]
    (println (str (clojure.string/trim section) " " c))))

;; 24
(doseq [[_ category] (re-seq #"\[\[ファイル\:(.*)\]\]" uk-text)]
  (println (clojure.string/replace category #"\|.*" "")))

;; 25
(import '(java.io BufferedReader StringReader))
;; TODO reduceじゃなくて、 take-while, drop-whileをつかっていけそう
;; (def raw-lines (-> (reduce (fn [{:keys [lines inner-section?]} line]
;;                             (if inner-section?
;;                               (if (= line "}}")
;;                                 {:lines (conj lines line) :inner-section? false}
;;                                 {:lines (conj lines line) :inner-section? inner-section?})
;;                               (if (= line "{{基礎情報 国")
;;                                 {:lines lines :inner-section? true}
;;                                 {:lines lines :inner-section? inner-section?})))
;;                           {:lines []
;;                            :inner-section? false}
;;                           (line-seq (BufferedReader. (StringReader. uk-text))))
;;                    :lines))
(def raw-lines (->> (line-seq (BufferedReader. (StringReader. uk-text)))
                   (drop-while (fn [line] (not= line "{{基礎情報 国")))
                   (drop 1) ;; drop line of "{{基礎情報 国"
                   (take-while (fn [line] (not= line "}}")))))
(def templates (->> raw-lines
                    (map (fn [raw-line]
                           (let [[_ k v] (re-find #"\|(.*) = (.*)" raw-line)]
                             [k v])))
                    (filter (fn [[k _]] (not (nil? k))))
                    (into {})))

;; 26
(def templates2 (->> templates
                          (map (fn [[k v]]
                                 [k (clojure.string/replace v #"'''''(.*?)'''''" "$1")]))
                          (map (fn [[k v]]
                                 [k (clojure.string/replace v #"'''(.*?)'''" "$1")]))
                          (map (fn [[k v]]
                                 [k (clojure.string/replace v #"''(.*?)''" "$1")]))
                          (into {})))

;; 27

(def templates3 (->> templates2
                     (map (fn [[k v]]
                            [k (clojure.string/replace v #"\[\[.*?\|(.*)\]\]" "$1")]))
                     (map (fn [[k v]]
                           [k (clojure.string/replace v #"\[\[(.*?)\]\]" "$1")]))
                     (into {})))

;; 28
;; いまいち
(def templates4 (->> templates3
                     (map (fn [[k v]]
                            [k (clojure.string/replace v #"\<ref\>.*\<\/ref\>" "")]))
                     (map (fn [[k v]]
                            [k (clojure.string/replace v #"\<ref .*?\>" "")]))
                     (map (fn [[k v]]
                            [k (clojure.string/replace v #"\<ref\>" "")]))
                     (map (fn [[k v]]
                            [k (clojure.string/replace v #"\<br \/\>" "")]))
                     (map (fn [[k v]]
                            [k (clojure.string/replace v #"\{\{(.*?)\}\}" "")]))
                     (into {})))
(doseq [[k v] templates4]
  (println (str k " -> " v)))

;; 29
(def wikidata (let [filename (get templates4 "国旗画像")
                    url (str "https://en.wikipedia.org/w/api.php?action=query&prop=imageinfo&iiprop=url&titles=File%3A"
                             (clojure.string/replace filename #" " "+")
                             "&format=json")]
                (cheshire/decode (slurp url))))
(get-in wikidata ["query" "pages" "23473560" "imageinfo" 0 "url"])