:; #-*- mode: nendo; syntax: scheme -*-;;
;;;
;;; henkan.nnd - 変換エンジンのコア
;;;  
;;;   Copyright (c) 2010  Kiyoka Nishiyama  <kiyoka@sumibi.org>
;;;   
;;;   Redistribution and use in source and binary forms, with or without
;;;   modification, are permitted provided that the following conditions
;;;   are met:
;;;   
;;;   1. Redistributions of source code must retain the above copyright
;;;      notice, this list of conditions and the following disclaimer.
;;;  
;;;   2. Redistributions in binary form must reproduce the above copyright
;;;      notice, this list of conditions and the following disclaimer in the
;;;      documentation and/or other materials provided with the distribution.
;;;  
;;;   3. Neither the name of the authors nor the names of its contributors
;;;      may be used to endorse or promote products derived from this
;;;      software without specific prior written permission.
;;;  
;;;   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
;;;   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
;;;   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
;;;   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
;;;   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
;;;   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED
;;;   TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
;;;   PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
;;;   LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
;;;   NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
;;;   SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
;;;  
;;;  $Id: 
;;;
(use srfi-1)
(use rfc.json)
(use sekka.util)
(use sekka.convert-jisyo)
(use sekka.jisyo-db)
(use sekka.alphabet-lib)
(use sekka.sharp-number)
(use sekka.google-ime)
(require "sekka/approximatesearch")


;; 曖昧検索の評価値にヒューリスティックを加える
;;  "nn" のように n が重なるほど、"ん" の可能性が強くなる。
;;  "にゃ" "にゅ" "にょ" が出現した場合は、n がそちらの解釈に取られたものと解釈してその件数分引いておく。
(define (nn-heuristics keyword lst)
  (map
   (lambda (x)
     (let ((point  (first   x))
           (roman  (second  x))
           (target (third   x)))
       (let* ((nn-number (min
                          (vector-length (keyword.scan #/nn/))
                          (vector-length (roman.scan   #/nn/))
                          (- (vector-length (target.scan  #/ん/))
                             (vector-length (target.scan  #/[ゃゅょ]/)))))
              (nn-number (max 0 nn-number)))
         (list
          (if (= 0 nn-number)
              point
              (* point (* 1.1 nn-number)))
          roman
          target))))
   lst))


(define (flatten-vector vec)
  (vec.flatten))


;; 曖昧検索
;;  リストで返す。 (ソート済み検索結果)
(define (approximate-search userid kvs keyword type limit)
  (let ([jarow-threshold
         (if (= "h" type)
             0.975
             0.94)])

    (let* ((a-search (ApproximateSearch.new jarow-threshold))
           (lst      (to-list
                      (flatten-vector
                       (map
                        (lambda (x)
                          (let* ((val (vector-ref x 0))
                                 (k   (vector-ref x 1))
                                 (v   (dict-get userid kvs k ""))
                                 (v-pair (string-split-first-and-rest v))
                                 (v-first (car v-pair))
                                 (v-rest  (cdr v-pair)))
                            (map
                             (lambda (element)
                               (list val k (+ v-first element)))
                             (v-rest.split "/"))))
                        (a-search.search userid kvs keyword type)))))
           (nn-heuristics-result (nn-heuristics keyword lst))
           (sorted-lst  (sort-by nn-heuristics-result (lambda (item) (- 1.0 (car item))))))
      (if (= limit 0)
          sorted-lst
          (take* sorted-lst limit)))))


;; 厳密検索
(define (exact-search userid kvs keyword)
  (let1 v (dict-get userid kvs keyword #f)
    (if v
        `((1.0 ,keyword ,v))
        '())))


;; split "/a;annotation/b/c/" into `(("a" "annotation" ,src) ("b" #f ,src) ("c" #f ,src))  alist
(define (split-henkan-kouho str src . okuri)
  (define (append-j-type lst)
    (map
     (lambda (x)
       (append x (list 'j)))
     lst))
    
  (let* ((trimmed (cond
                   ((rxmatch #/^[\/](.+)$/ str)
                    => (lambda (m)
                         (rxmatch-substring m 1)))
                   (else
                    "")))
         (splitted (filter
                    (lambda (str) (< 0 str.size))
                    (to-list (trimmed.split "/"))))
         (okuri    (get-optional okuri "")))
    ;; pickup annotation
    (append-j-type
     (map
      (lambda (entry)
        (let1 vec (entry.split ";")
          (list (+ (vector-ref vec 0) okuri)
                (vector-ref vec 1 #f)
                src)))
      splitted))))


;; 送り仮名なしの変換
(define (henkan-okuri-nashi userid kvs keyword limit)
  (let* ((result (approximate-search userid kvs keyword "k" limit))
         (kouho  (uniq (map
                        (lambda (x)
                          (third x))
                        result))))
    (let1 lst
        (append-map
         (lambda (value)
           (if-let1 m (rxmatch #/^C(.+)$/ value)
             (split-henkan-kouho
              (dict-get userid kvs (string-drop value 1))
              (rxmatch-substring m 1)
              ) ;; continue to fetch
             (split-henkan-kouho
              value
              keyword)))
         kouho)
      (if (= limit 0)
          lst
          (take* lst limit)))))
          
  
  
;; 送り仮名なし、かつ、数字で始まるキーワードの変換
(define (henkan-okuri-nashi-and-number userid kvs keyword limit)
  ;; 数字部分を抜きだして '#' に変換
  (let ([num-list (to-list (keyword.scan #/[0-9]+/))]
        [replaced (keyword.gsub #/[0-9]+/ "#")])
    ;; 辞書引きした結果の # 部分を 再度数値に戻す。
    (let* ([result1 (henkan-okuri-nashi userid kvs replaced limit)]
           [result2 (map
                     (lambda (entry)
                       (let* ([type-list (to-list (. (car entry) scan #/[#][0-9]/))]
                              [converted-list 
                               (map
                                (lambda  (_type _num)
                                  (sekka-henkan-sharp-number _type _num))
                                type-list
                                num-list)]
                              [format-str (. (car entry) gsub #/[#][0-9]/ "%s")])
                         ;; ユーザ入力クエリ中の数字のリストと辞書の中の # の個数がマッチしないといけない
                         (if (= (length num-list)
                                (length type-list))
                             (cons
                              (apply sprintf
                                     (cons format-str
                                           converted-list))
                              (append (drop-right (cdr entry) 1)
                                      (list 'n)))
                             #f)))
                     result1)])
      (filter
       (lambda (x) x)
       result2))))
           


;; 送り仮名ありの変換
(define (henkan-okuri-ari userid kvs keyword limit roman-method)
  (let* ((keyword  (let1 _pair (string-split-first-and-rest keyword)
                     (+ (sekka-downcase (car _pair))
                        (cdr _pair))))
         (m        (rxmatch #/^([a-z])([a-z\-^]*)([A-Z`+])([a-zA-Z]*)$/ keyword)))
    (if (not m)
        '() ;; 変換候補無し
        (let* ((result (approximate-search userid kvs
                                           (+ (rxmatch-substring m 1)
                                              (rxmatch-substring m 2)
                                              (rxmatch-substring m 3))
                                           "K"
                                           limit))
               (okurigana-lst  (gen-roman->hiragana (sekka-downcase
                                                     (+
                                                      (rxmatch-substring m 3)
                                                      (rxmatch-substring m 4)))
                                                    roman-method))
               (okurigana-lst  (if (null? okurigana-lst)
                                   '("")
                                   okurigana-lst))
               (kouho      (uniq (map
                                  (lambda (x)
                                    (third x))
                                  result))))
          (let1 lst
              (append-map
               (lambda (value)
                 (append-map
                  (lambda (x)
                    (if-let1 m (rxmatch #/^C(.+)$/ value)
                      (split-henkan-kouho
                       (dict-get userid kvs (string-drop value 1))
                       (rxmatch-substring m 1)
                       x) ;; continue to fetch
                      (split-henkan-kouho value keyword x)))
                  okurigana-lst))
               kouho)
            (if (= limit 0)
                lst
                (take* lst limit)))))))


;; 平仮名フレーズ辞書を使った曖昧検索
(define (henkan-hiragana-phrase userid kvs keyword limit roman-method)
  (let* ([result (approximate-search userid kvs
                                     (+ "=" (sekka-downcase keyword))
                                     "h"
                                     limit)]
         [uniq-result
          (delete-duplicates
           (map
            (lambda (x) (third x))
            result))]
          
         [lst (map
               (lambda (x)
                 (list x #f keyword 'h))
               uniq-result)])

    (if (= limit 0)
        lst
        (take* lst limit))))

  
  
;; 平仮名の変換
(define (henkan-hiragana userid kvs keyword roman-method)
  (let* ([phrase-limit  3]
         [str  (sekka-downcase keyword)]
         [hira (gen-roman->hiragana str roman-method)]
         [kata (gen-roman->katakana str roman-method)])
    (append
     (henkan-hiragana-phrase userid kvs keyword phrase-limit roman-method)
     (if (null? hira)
         `((,keyword #f ,keyword j))
         (append-map (lambda (h k)
                       `(
                         (,h #f ,keyword h)
                         (,k #f ,keyword k)))
                     hira
                     kata)))))


;; アルファベットの単純変換
(define (henkan-alphabet kvs keyword)
  (let* ((zen  (gen-alphabet-han->zen keyword))
         (han  (gen-alphabet-zen->han keyword)))
    `(
      (,zen #f ,keyword z)
      (,han #f ,keyword l))))


;; 記号を含むキーワードの変換
(define (henkan-non-kanji userid kvs keyword)
  (let* ((result (exact-search userid kvs keyword))
         (kouho  (map
                  (lambda (x)
                    (third x))
                  result)))
    (append-map
     (lambda (value)
       (split-henkan-kouho value keyword))
     kouho)))


;; 数字だけで構成されるキーワードの変換
(define (henkan-number kvs keyword)
  ;; 数字部分を抜きだして '#' に変換
  `(
    (,(sekka-henkan-sharp-number "#1" keyword) nil ,keyword n)
    (,(sekka-henkan-sharp-number "#0" keyword) nil ,keyword n)
    (,(sekka-henkan-sharp-number "#2" keyword) nil ,keyword n)
    (,(sekka-henkan-sharp-number "#3" keyword) nil ,keyword n)))


;; keyword には ローマ字のみを受け付ける
(define (sekka-henkan userid kvs keyword limit roman-method)
  ;; 変換候補にindex番号を付加する
  (define (append-index-number kouho-list)
    (let1 count 0
      (map
       (lambda (x)
         (begin0
           (append x (list count))
           (set! count (+ count 1))))
       kouho-list)))
  
  ;; キーワードの種別で処理を分割する
  (append-index-number
   (cond
    ((rxmatch #/[A-Z`+]/ keyword)
     (let1 k (string-downcase-first keyword)
       (cond
        ((rxmatch #/[a-z][A-Z`+]/ k)
         ;; 送りあり
         (append
          (append
           (henkan-okuri-ari userid kvs k limit roman-method)
           (if (null? (gen-roman->hiragana (sekka-downcase k) roman-method))
               '()
              (henkan-hiragana userid kvs (sekka-downcase k) roman-method)))
          (henkan-alphabet kvs keyword)))
        (else
         ;; 送りなし
         (append
          (append
           (henkan-okuri-nashi userid kvs k limit)
           (if (null? (gen-roman->hiragana (sekka-downcase k) roman-method))
               '()
               (henkan-hiragana userid kvs (sekka-downcase k) roman-method)))
          (henkan-alphabet kvs keyword))))))
    ;; 10進数数値のみで構成されるキーワード
    ((rxmatch #/^[0-9]+$/ keyword)
     (henkan-number kvs keyword))
    ;; 10進数数値で始まるキーワード
    ((rxmatch #/^[0-9]+[0-9a-zA-Z@-]+$/ keyword)
     (append
      (henkan-okuri-nashi-and-number userid kvs keyword limit)
      (henkan-alphabet kvs keyword)))
    ((not (null? (gen-roman->hiragana keyword roman-method)))
     (append
      (append
       (henkan-hiragana userid kvs keyword roman-method)
       (henkan-alphabet kvs keyword))
      (henkan-okuri-nashi userid kvs keyword limit)))
    (else
     (append
      (henkan-non-kanji userid kvs keyword)
      (henkan-alphabet kvs keyword))))))



;; conversion #f => nil for EmacsLisp
(define (sekkaHenkan userid kvs cachesv keyword limit roman-method)
  (define cache-exp-second (* 60 60))
  (let* ((keyword (keyword.strip))
         (sekka-keyword (+ "sekka:" roman-method ":" (limit.to_s) ":" keyword)))
    (if-let1 fetched (and cachesv
                          (cachesv.get sekka-keyword #f))
      (begin
        keyword
        (read-from-string fetched))
      (let1 henkan-result (map
                           (lambda (x)
                             (map (lambda (val) (if val val nil)) x))
                           (sekka-henkan userid kvs keyword limit (make-keyword roman-method)))
        (and cachesv
             (not (null? henkan-result))
             (let1 fetched2 (cachesv.get  "sekka:(keys)" #f)
               (cachesv.put! sekka-keyword (write-to-string henkan-result) cache-exp-second)
               (cachesv.put! "sekka:(keys)" (if fetched2
                                                 (+ fetched2 " " sekka-keyword)
                                                 sekka-keyword)
                             cache-exp-second)
               (if-let1 v (cachesv.get  "sekka:(keys)" #f)
                 (begin
                   #?=(v.size)
                   #?=v)
                 #f)))
        henkan-result))))
;; Export to Ruby world
(export-to-ruby sekkaHenkan)


;; Flush henkan-result cache data on cachesv(memcached)
(define (flushCacheServer cachesv)
  #?="--- do (flushCacheServer) ---"
  (if-let1 fetched (cachesv.get  "sekka:(keys)" #f)
    (begin
      (for-each
       (lambda (x)
         (cachesv.delete  #?=x))
       (to-list (fetched.split #/[ ]+/)))
      (cachesv.delete "sekka:(keys)"))
    #f))
;; Export to Ruby world
(export-to-ruby flushCacheServer)


;; 確定処理: 最終確定語を変換候補の先頭に持ってくる。
;;  key   ... "developer" や "へんかん" など、変換候補レコードのキーとなる文字列
;;  tango ... "変換" など、変換候補から最終確定した、変換候補の文字列
;;  登録失敗したら nil を返す
(define (sekkaKakutei userid kvs cachesv key tango)
  (define (fetch userid kvs key)
    (dict-get userid kvs key #f))

  (define (write-user-entry userid kvs key value)
    (kvs.put! (+ userid ":" key) value))

  (define (join-henkan-kouho lst)
    (+ "/"
       (string-join
        (map
         (lambda (x)
           (if (second x)
               (+ (first x) ";" (second x))
               (first x)))
         lst)
        "/")))

  (let1 tango (if (is-hiragana-and-okuri key)
                  (drop-okuri tango)
                  tango)
    (if-let1 kouho-str (fetch userid kvs key)
      (if (rxmatch #/^\// kouho-str)
          (let* ((kouho-lst (split-henkan-kouho kouho-str key))
                 (no1
                  (filter (lambda (x)
                               (eq? (car x) tango))
                             kouho-lst))
                 (other
                  (filter (lambda (x)
                               (not (eq? (car x) tango)))
                             kouho-lst))
                 (new-kouho-str
                  (join-henkan-kouho (append no1 other))))
            (if (not (= kouho-str new-kouho-str))
                (begin
                  (write-user-entry userid kvs key new-kouho-str)
                  (and cachesv
                       (flushCacheServer cachesv))
                  tango)
                nil))
          nil)
      nil)))
;; Export to Ruby world
(export-to-ruby sekkaKakutei)


;; ユーザー辞書の全削除 (useridで指定したユーザーの辞書)
;; 削除した件数を返す。
(define (flushUserJisyo userid kvs)
  (let1 lst (to-list (kvs.forward_match_keys (+ userid ":")))
    (for-each
     (lambda (k)
       (kvs.delete k))
     lst)
    (length lst)))
;; Export to Ruby world
(export-to-ruby flushUserJisyo)


;; ユーザー定義語彙の登録処理
(define (registerUserJisyo userid kvs dict-line)
  (define user-keylist '())
  (define (insert-to-db sekka-jisyo-data)
    (for-each
     (lambda (entry)
       (let* ((kv (to-list (entry.split #/[ ]+/)))
              (k (first  kv))
              (v (second kv)))
         (set! user-keylist (cons k user-keylist))
         (append-entry userid kvs k v)))
     sekka-jisyo-data))

  ;; "ユーザー語彙を"(stored)"にpush!する"
  (define (kvs-push! userid kvs entry-str)
    (let* ((orig (kvs.get (+ userid ":(stored)") "()"))
           (orig (read-from-string orig)))
      (kvs.put! (+ userid ":(stored)") (write-to-string (append orig (list entry-str))))
      entry-str))

  (cond
   ((memv dict-line (read-from-string (kvs.get (+ userid ":(stored)") "()")))
    #?=(+ "already stored userid=[" userid "] tango=[" dict-line "]")
    #f)
   (else
    (let1 lst (convert-skk-jisyo-f (StringIO.new (+ dict-line "\n")))
      (if (null? lst)
            (begin
              #?=(+ "Error user dict format error userid=[" userid "] tango=[" dict-line "]")
              #f)
            (begin
              (insert-to-db lst)
              (setup-trie-index userid kvs user-keylist #f)
              (kvs-push! userid kvs dict-line)
              #?=(+ "user dict stored userid=[" userid "] tango=[" dict-line "]")
              #t))))))
;; Export to Ruby world
(export-to-ruby registerUserJisyo)


;; ユーザー定義語彙の登録処理
(define (googleIme keyword proxy-host proxy-port)
  (request-google-ime keyword #f proxy-host proxy-port))
;; Export to Ruby world
(export-to-ruby googleIme)