Emacs Lisp to count Japanese zenkaku/hankaku character mixed string.

count-japanese-chars.el

(defun asciip (char-code)
  (and (<= 32 char-code)
       (<= char-code 126)))

;; (asciip ?a)                             ; => t
;; (asciip ?あ)                            ; => nil


;; see also: japan-util.el
;; char-code のリスト
(defconst japanese-hankaku-chars (string-to-list
           "ｱｲｳｴｵｶｷｸｹｺｻｼｽｾｿﾀﾁﾂﾃﾄﾅﾆﾇﾈﾉﾊﾋﾌﾍﾎﾏﾐﾑﾒﾓﾔﾕﾖﾗﾘﾙﾚﾛﾜｦﾝｧｨｩｪｫｯｬｭｮ､｡･ﾞﾟ`ｰ｢｣"))

;; その他に半角として扱う文字があればここに入れる
;; char-code のリスト
(defvar misc-hankaku-chars (string-to-list
                              ""))

(defun japanese-hankaku-p (char-code)
  (memq char-code
        japanese-hankaku-chars))

(defun misc-hankaku-p (char-code)
  (memq char-code
        misc-hankaku-chars))

;; (japanese-hankaku-p ?ョ)                ; => nil
;; (japanese-hankaku-p ?｣)                 ; => (65379)


(defun zenkakup (char-code)
  (and (not (asciip char-code))
       (not (japanese-hankaku-p char-code))
       (not (misc-hankaku-p char-code))))

;; (zenkakup ?a)                           ; => nil
;; (zenkakup ?あ)                          ; => t


(defun count-chars-as-zenkaku-in-string (str)
  (let ((count 0)
        (zenkaku-count 0)
        char-code)
    (while (< count (length str))
      (setq char-code (aref str count))
    
      (setq zenkaku-count
            (+ zenkaku-count 
               (if (zenkakup char-code) 1 0.5)))
    
      (setq count (1+ count)))
    ;; zenkaku-count
    (ceiling zenkaku-count) ; 切り上げ
    ))

;; (count-chars-as-zenkaku-in-string "")             ; => 0 (#o0, #x0)
;; (count-chars-as-zenkaku-in-string "a")            ; => 1 (#o1, #x1)
;; (count-chars-as-zenkaku-in-string "ab")           ; => 1 (#o1, #x1)
;; (count-chars-as-zenkaku-in-string "abc")          ; => 2 (#o2, #x2)
;; (count-chars-as-zenkaku-in-string "123")          ; => 2 (#o2, #x2)
;; (count-chars-as-zenkaku-in-string "a b c ")       ; => 3 (#o3, #x3)
;; (count-chars-as-zenkaku-in-string "abあ")         ; => 2 (#o2, #x2)
;; (count-chars-as-zenkaku-in-string "abcあ")        ; => 3 (#o3, #x3)
;; (count-chars-as-zenkaku-in-string "ab(あ[い]う)") ; => 6 (#o6, #x6)
;; (count-chars-as-zenkaku-in-string "ｼﾞｬﾝﾊﾟｰ")      ; => 4 (#o4, #x4)


(defun count-chars-as-zenkaku (begin end)
  (let ((pos begin)
        (count 0)
        (zenkaku-count 0))
    (while (< pos end)
      (setq zenkaku-count
            (+ zenkaku-count 
               (cond
                ((eq (char-after pos) ?\n)
                 0)
                ((zenkakup (char-after pos))
                 1)
                (t
                 0.5)
                )
               ))
      
      (setq pos (1+ pos)))

    ;; zenkaku-count
    (ceiling zenkaku-count) ; 切り上げ
    ))

(defun count-japanese-chars (begin end)
  (interactive "r")
  (message "%d" (count-chars-as-zenkaku begin end)))

;; (global-set-key (kbd "C-M-=") 'count-japanese-chars)

;; (defun count-lines-and-chars ()
;;   (if mark-active
;;       (format "%d jchars"
;;               (count- (region-beginning) (region-end))
;;               )
;;     ""))

;; (add-to-list 'default-mode-line-format
;;              '(:eval (count-lines-and-chars)))