;;; china-util.el --- utilities for Chinese
-;; Copyright (C) 1995 Free Software Foundation, Inc.
;; Copyright (C) 1995 Electrotechnical Laboratory, JAPAN.
+;; Licensed to the Free Software Foundation.
;; Keywords: mule, multilingual, Chinese
;; Free Software Foundation, Inc., 59 Temple Place - Suite 330,
;; Boston, MA 02111-1307, USA.
-;;; Code:
-
-;;;###autoload
-(defun setup-chinese-gb-environment ()
- "Setup multilingual environment (MULE) for Chinese GB2312 users."
- (interactive)
- (setup-english-environment)
-
- (setq primary-language "Chinese-GB")
-
- (setq coding-category-iso-8-2 'cn-gb-2312)
- (setq coding-category-iso-else 'iso-2022-cn)
- (setq coding-category-big5 'cn-big5)
-
- (set-coding-priority
- '(coding-category-iso-7
- coding-category-iso-else
- coding-category-iso-8-2
- coding-category-big5
- coding-category-iso-8-1
- coding-category-emacs-mule))
-
- (setq-default buffer-file-coding-system 'cn-gb-2312)
- (set-terminal-coding-system-internal 'cn-gb-2312)
- (set-keyboard-coding-system-internal 'cn-gb-2312)
- (setq sendmail-coding-system nil
- rmail-file-coding-system 'iso-2022-cn)
-
- (setq default-input-method '("Chinese-GB" . "quail-py")))
-
-;;;###autoload
-(defun setup-chinese-big5-environment ()
- "Setup multilingual environment (MULE) for Chinese Big5 users."
- (interactive)
- (setup-english-environment)
-
- (setq primary-language "Chinese-BIG5")
-
- (setq coding-category-iso-8-2 'cn-big5)
- (setq coding-category-iso-else 'iso-2022-cn)
- (setq coding-category-big5 'cn-big5)
-
- (set-coding-priority
- '(coding-category-iso-7
- coding-category-iso-else
- coding-category-big5
- coding-category-iso-8-2
- coding-category-emacs-mule))
-
- (setq-default buffer-file-coding-system 'cn-big5)
- (set-terminal-coding-system-internal 'cn-big5)
- (set-keyboard-coding-system-internal 'cn-big5)
- (setq sendmail-coding-system nil
- rmail-file-coding-system 'iso-2022-cn)
-
- (setq default-input-method '("Chinese-BIG5" . "quail-py-b5")))
+;;; Commentary:
-;;;###autoload
-(defun setup-chinese-cns-environment ()
- "Setup multilingual environment (MULE) for Chinese CNS11643 family users."
- (interactive)
- (setup-english-environment)
-
- (setq primary-language "Chinese-CNS")
-
- (setq coding-category-iso-else 'iso-2022-cn)
- (setq coding-category-big5 'cn-big5)
- (setq coding-category-iso-8-2 'cn-big5)
-
- (set-coding-priority
- '(coding-category-iso-7
- coding-category-iso-else
- coding-category-iso-8-2
- coding-category-big5))
-
- (setq-default buffer-file-coding-system 'iso-2022-cn)
- (set-terminal-coding-system-internal 'iso-2022-cn)
- (set-keyboard-coding-system-internal 'iso-2022-cn)
- (setq sendmail-coding-system nil
- rmail-file-coding-system 'iso-2022-cn)
-
- (setq default-input-method '("Chinese-CNS" . "quail-quick-cns")))
+;;; Code:
;; Hz/ZW encoding stuffs
;; Regexp of ZW sequence to start GB2312.
(defvar zw-start-gb "^zW")
;; Regexp for start of GB2312 in an encoding mixture of HZ and ZW.
-(defvar hz/zw-start-gb (concat hz-gb-designnation "\\|" zw-start-gb))
+(defvar hz/zw-start-gb
+ (concat hz-gb-designnation "\\|" zw-start-gb "\\|[^\0-\177]"))
(defvar decode-hz-line-continuation nil
"Flag to tell if we should care line continuation convention of Hz.")
+(defconst hz-set-msb-table
+ (let ((str (make-string 127 0))
+ (i 0))
+ (while (< i 33)
+ (aset str i i)
+ (setq i (1+ i)))
+ (while (< i 127)
+ (aset str i (+ i 128))
+ (setq i (1+ i)))
+ str))
+
;;;###autoload
(defun decode-hz-region (beg end)
"Decode HZ/ZW encoded text in the current region.
(interactive "r")
(save-excursion
(save-restriction
- (narrow-to-region beg end)
-
- ;; We, at first, convert HZ/ZW to `iso-2022-7',
- ;; then decode it.
-
- ;; "~\n" -> "\n"
- (goto-char (point-min))
- (while (search-forward "~" nil t)
- (if (= (following-char) ?\n) (delete-char -1))
- (if (not (eobp)) (forward-char 1)))
-
- ;; "^zW...\n" -> Chinese GB2312
- ;; "~{...~}" -> Chinese GB2312
- (goto-char (point-min))
- (let ((chinese-found nil))
+ (let (pos ch)
+ (narrow-to-region beg end)
+
+ ;; We, at first, convert HZ/ZW to `euc-china',
+ ;; then decode it.
+
+ ;; "~\n" -> "\n", "~~" -> "~"
+ (goto-char (point-min))
+ (while (search-forward "~" nil t)
+ (setq ch (following-char))
+ (if (or (= ch ?\n) (= ch ?~)) (delete-char -1)))
+
+ ;; "^zW...\n" -> Chinese GB2312
+ ;; "~{...~}" -> Chinese GB2312
+ (goto-char (point-min))
+ (setq beg nil)
(while (re-search-forward hz/zw-start-gb nil t)
- (if (= (char-after (match-beginning 0)) ?z)
- ;; ZW -> iso-20227-7
- (progn
- (delete-char -2)
- (insert iso2022-gb-designation)
- (end-of-line)
- (insert iso2022-ascii-designation))
- ;; HZ -> iso-20227-7
- (delete-char -2)
- (insert iso2022-gb-designation)
- (let ((pos (save-excursion (end-of-line) (point))))
- (if (search-forward hz-ascii-designnation pos t)
- (replace-match iso2022-ascii-designation)
- (if (not decode-hz-line-continuation)
- (insert iso2022-ascii-designation)))))
- (setq chinese-found t))
- (if (or chinese-found
- (let ((enable-multibyte-characters nil))
- ;; Here we check if the text contains EUC (China) codes.
- ;; If any, we had better decode them also.
- (goto-char (point-min))
- (re-search-forward "[\240-\377]" nil t)))
- (decode-coding-region (point-min) (point-max) 'euc-china)))
-
- ;; "~~" -> "~"
- (goto-char (point-min))
- (while (search-forward "~~" nil t) (delete-char -1))
+ (setq pos (match-beginning 0)
+ ch (char-after pos))
+ ;; Record the first position to start conversion.
+ (or beg (setq beg pos))
+ (end-of-line)
+ (setq end (point))
+ (if (>= ch 128) ; 8bit GB2312
+ nil
+ (goto-char pos)
+ (delete-char 2)
+ (setq end (- end 2))
+ (if (= ch ?z) ; ZW -> euc-china
+ (progn
+ (translate-region (point) end hz-set-msb-table)
+ (goto-char end))
+ (if (search-forward hz-ascii-designnation
+ (if decode-hz-line-continuation nil end)
+ t)
+ (delete-char -2))
+ (setq end (point))
+ (translate-region pos (point) hz-set-msb-table))))
+ (if beg
+ (decode-coding-region beg end 'euc-china)))
(- (point-max) (point-min)))))
;;;###autoload
;; Chinese GB2312 -> "~{...~}"
(goto-char (point-min))
(if (re-search-forward "\\cc" nil t)
- (let ((enable-multibyte-characters nil)
- pos)
+ (let (pos)
(goto-char (setq pos (match-beginning 0)))
- (encode-coding-region pos (point-max) 'iso-2022-7)
+ (encode-coding-region pos (point-max) 'iso-2022-7bit)
(goto-char pos)
(while (search-forward iso2022-gb-designation nil t)
(delete-char -3)
(encode-hz-region (point-min) (point-max)))
;;
-(provide 'language/china-util)
+(provide 'china-util)
-;;; Local Variables:
-;;; generated-autoload-file: "../loaddefs.el"
-;;; End:
;;; china-util.el ends here
HCoop / bpt / emacs.git / blobdiff