git.chise.org Git - chise/xemacs-chise.git.1/blob - lisp/mule/japanese.el

   1 ;;; japanese.el --- Japanese support -*- coding: iso-2022-7bit; -*-
   2
   3 ;; Copyright (C) 1995 Electrotechnical Laboratory, JAPAN.
   4 ;; Licensed to the Free Software Foundation.
   5 ;; Copyright (C) 1997,1999,2002,2004,2005 MORIOKA Tomohiko
   6
   7 ;; Keywords: multilingual, Japanese
   8
   9 ;; This file is part of XEmacs.
  10
  11 ;; XEmacs is free software; you can redistribute it and/or modify it
  12 ;; under the terms of the GNU General Public License as published by
  13 ;; the Free Software Foundation; either version 2, or (at your option)
  14 ;; any later version.
  15
  16 ;; XEmacs is distributed in the hope that it will be useful, but
  17 ;; WITHOUT ANY WARRANTY; without even the implied warranty of
  18 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  19 ;; General Public License for more details.
  20
  21 ;; You should have received a copy of the GNU General Public License
  22 ;; along with XEmacs; see the file COPYING.  If not, write to the Free
  23 ;; Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA
  24 ;; 02111-1307, USA.
  25
  26 ;;; Commentary:
  27
  28 ;; For Japanese, character sets JISX0201, JISX0208, JISX0212 are
  29 ;; supported.
  30
  31 ;;; Code:
  32
  33 ;;; Syntax of Japanese characters.
  34 (modify-syntax-entry 'katakana-jisx0201 "w")
  35 (modify-syntax-entry 'japanese-jisx0212 "w")
  36
  37 (modify-syntax-entry 'japanese-jisx0208 "w")
  38 (loop for row in '(33 34 40)
  39       do (modify-syntax-entry `[japanese-jisx0208 ,row] "_"))
  40 (loop for char in '(?\e$B!<\e(B ?\e$B!+\e(B ?\e$B!,\e(B ?\e$B!3\e(B ?\e$B!4\e(B ?\e$B!5\e(B ?\e$B!6\e(B ?\e$B!7\e(B ?\e$B!8\e(B ?\e$B!9\e(B ?\e$B!:\e(B ?\e$B!;\e(B)
  41       do (modify-syntax-entry char "w"))
  42 (modify-syntax-entry ?\\e$B!J\e(B "(\e$B!K\e(B")
  43 (modify-syntax-entry ?\\e$B!N\e(B "(\e$B!O\e(B")
  44 (modify-syntax-entry ?\\e$B!P\e(B "(\e$B!Q\e(B")
  45 (modify-syntax-entry ?\\e$B!V\e(B "(\e$B!W\e(B")
  46 (modify-syntax-entry ?\\e$B!X\e(B "(\e$B!Y\e(B")
  47 (modify-syntax-entry ?\\e$B!K\e(B ")\e$B!J\e(B")
  48 (modify-syntax-entry ?\\e$B!O\e(B ")\e$B!N\e(B")
  49 (modify-syntax-entry ?\\e$B!Q\e(B ")\e$B!P\e(B")
  50 (modify-syntax-entry ?\\e$B!W\e(B ")\e$B!V\e(B")
  51 (modify-syntax-entry ?\\e$B!Y\e(B ")\e$B!X\e(B")
  52
  53 ;;; Character categories S, A, H, K, G, Y, and C
  54 (define-category ?S "Japanese 2-byte symbol character.")
  55 (modify-category-entry [japanese-jisx0208 33] ?S)
  56 (modify-category-entry [japanese-jisx0208 34] ?S)
  57 (modify-category-entry [japanese-jisx0208 40] ?S)
  58 (define-category ?A "Japanese 2-byte Alphanumeric character.")
  59 (modify-category-entry [japanese-jisx0208 35] ?A)
  60 (define-category ?H "Japanese 2-byte Hiragana character.")
  61 (modify-category-entry [japanese-jisx0208 36] ?H)
  62 (define-category ?K "Japanese 2-byte Katakana character.")
  63 (modify-category-entry [japanese-jisx0208 37] ?K)
  64 (define-category ?G "Japanese 2-byte Greek character.")
  65 (modify-category-entry [japanese-jisx0208 38] ?G)
  66 (define-category ?Y "Japanese 2-byte Cyrillic character.")
  67 (modify-category-entry [japanese-jisx0208 39] ?Y)
  68 (define-category ?C "Japanese 2-byte Kanji characters.")
  69 (loop for row from 48 to 126
  70       do (modify-category-entry `[japanese-jisx0208 ,row] ?C))
  71 (loop for char in '(?\e$B!<\e(B ?\e$B!+\e(B ?\e$B!,\e(B)
  72       do (modify-category-entry char ?K)
  73          (modify-category-entry char ?H))
  74 (loop for char in '(?\e$B!3\e(B ?\e$B!4\e(B ?\e$B!5\e(B ?\e$B!6\e(B ?\e$B!7\e(B ?\e$B!8\e(B ?\e$B!9\e(B ?\e$B!:\e(B ?\e$B!;\e(B)
  75       do (modify-category-entry char ?C))
  76 (modify-category-entry 'japanese-jisx0212 ?C)
  77
  78 (defvar japanese-word-regexp
  79   "\\cA+\\cH*\\|\\cK+\\cH*\\|\\cC+\\cH*\\|\\cH+\\|\\ck+\\|\\sw+"
  80   "Regular expression used to match a Japanese word.")
  81
  82 (set-word-regexp japanese-word-regexp)
  83 (setq forward-word-regexp  "\\w\\>")
  84 (setq backward-word-regexp "\\<\\w")
  85
  86 ;;; Paragraph setting
  87 (setq sentence-end
  88       (concat
  89        "\\("
  90        "\\("
  91        "[.?!][]\"')}]*"
  92        "\\|"
  93        "[\e$B!%!)!*\e(B][\e$B!O!I!G!K!Q!M!S!U!W!Y\e(B]*"
  94        "\\)"
  95        "\\($\\|\t\\|  \\)"
  96        "\\|"
  97        "\e$B!#\e(B"
  98        "\\)"
  99        "[ \t\n]*"))
 100
 101 ;; allow paragraphs to start with a zenkaku space
 102 (setq paragraph-start    "[ \e$B!!\e(B\t\n\f]")
 103 (setq paragraph-separate "[ \e$B!!\e(B\t\f]*$")
 104
 105 ;; EGG specific setup
 106 (define-egg-environment 'japanese
 107   "Japanese settings for egg."
 108   (lambda ()
 109     (when (not (featurep 'egg-jpn))
 110       (load "its-hira")
 111       (load "its-kata")
 112       (load "its-hankaku")
 113       (load "its-zenkaku")
 114       (setq its:*standard-modes*
 115             (append
 116              (list (its:get-mode-map "roma-kana")
 117                    (its:get-mode-map "roma-kata")
 118                    (its:get-mode-map "downcase")
 119                    (its:get-mode-map "upcase")
 120                    (its:get-mode-map "zenkaku-downcase")
 121                    (its:get-mode-map "zenkaku-upcase"))
 122              its:*standard-modes*))
 123       (provide 'egg-jpn))
 124     (setq wnn-server-type 'jserver)
 125     ;; Can't do this here any more.  Must do it when selecting egg-wnn
 126     ;; or egg-sj3
 127     ;; (setq egg-default-startup-file "eggrc-wnn")
 128     (setq-default its:*current-map* (its:get-mode-map "roma-kana"))))
 129
 130 ;; stuff for providing grammatic processing of Japanese text
 131 ;; something like this should probably be created for all environments...
 132 ;; #### Arrgh.  This stuff should defvar'd in either fill.el or kinsoku.el.
 133 ;; Then the language environment should set these things, probably buffer-
 134 ;; locally.
 135
 136 (defvar aletter (concat "\\(" ascii-char "\\|" kanji-char "\\)"))
 137 (defvar kanji-space-insertable (concat
 138            "\e$B!"\e(B" aletter                   "\\|"
 139            "\e$B!#\e(B" aletter                   "\\|"
 140            aletter "\e$B!J\e(B"                   "\\|"
 141            "\e$B!K\e(B" aletter                   "\\|"
 142            ascii-alphanumeric  kanji-kanji-char "\\|"
 143            kanji-kanji-char    ascii-alphanumeric ))
 144
 145 ;; #### will be moved to fill.el
 146 (defvar space-insertable (concat " " aletter "\\|" kanji-space-insertable)
 147   "Regexp for finding points that can have spaces inserted into them for justification")
 148 \f
 149 ;; (make-coding-system
 150 ;;  'iso-2022-jp 2 ?J
 151 ;;  "ISO 2022 based 7bit encoding for Japanese (MIME:ISO-2022-JP)"
 152 ;;  '((ascii japanese-jisx0208-1978 japanese-jisx0208
 153 ;;           latin-jisx0201 japanese-jisx0212 katakana-jisx0201) nil nil nil
 154 ;;    short ascii-eol ascii-cntl seven)
 155 ;;  '((safe-charsets ascii japanese-jisx0208-1978 japanese-jisx0208
 156 ;;                   latin-jisx0201 japanese-jisx0212 katakana-jisx0201)
 157 ;;    (mime-charset . iso-2022-jp)))
 158
 159 (make-coding-system
 160  'iso-2022-jp 'iso2022
 161  "Coding-system used for communication with mail and news in Japan."
 162  (let ((conf
 163         '(charset-g0 ascii
 164           short t
 165           seven t
 166           mnemonic "MULE/7bit")))
 167    (if (featurep 'utf-2000)
 168        (list* 'ccs-priority-list
 169               '(ascii
 170                 =jis-x0208@1983 =jis-x0208@1978
 171                 latin-jisx0201)
 172               conf)
 173      (list* 'input-charset-conversion
 174             '((latin-jisx0201 ascii)
 175               (japanese-jisx0208-1978 japanese-jisx0208))
 176             conf))))
 177
 178 (define-coding-system-alias 'junet 'iso-2022-jp)
 179
 180 ;; (make-coding-system
 181 ;;  'iso-2022-jp-2 2 ?J
 182 ;;  "ISO 2022 based 7bit encoding for CJK, Latin-1, and Greek (MIME:ISO-2022-JP-2)"
 183 ;;  '((ascii japanese-jisx0208-1978 japanese-jisx0208
 184 ;;           latin-jisx0201 japanese-jisx0212 katakana-jisx0201
 185 ;;           chinese-gb2312 korean-ksc5601) nil
 186 ;;           (nil latin-iso8859-1 greek-iso8859-7) nil
 187 ;;  short ascii-eol ascii-cntl seven nil single-shift)
 188 ;;  '((safe-charsets ascii japanese-jisx0208-1978 japanese-jisx0208
 189 ;;                   latin-jisx0201 japanese-jisx0212 katakana-jisx0201
 190 ;;                   chinese-gb2312 korean-ksc5601
 191 ;;                   latin-iso8859-1 greek-iso8859-7)
 192 ;;    (mime-charset . iso-2022-jp-2)))
 193
 194 ;; (make-coding-system
 195 ;;  'japanese-shift-jis 1 ?S
 196 ;;  "Shift-JIS 8-bit encoding for Japanese (MIME:SHIFT_JIS)"
 197 ;;  nil
 198 ;;  '((safe-charsets ascii japanese-jisx0208 japanese-jisx0208-1978
 199 ;;                   latin-jisx0201 katakana-jisx0201)
 200 ;;    (mime-charset . shift_jis)
 201 ;;    (charset-origin-alist (japanese-jisx0208 "SJIS" encode-sjis-char)
 202 ;;                          (katakana-jisx0201 "SJIS" encode-sjis-char))))
 203
 204 (make-coding-system
 205  'shift_jis 'shift-jis
 206  "Coding-system of Shift-JIS used in Japan."
 207  '(mnemonic "Ja/SJIS"))
 208
 209 ;; (define-coding-system-alias 'shift_jis 'japanese-shift-jis)
 210 ;; (define-coding-system-alias 'sjis 'japanese-shift-jis)
 211
 212 ;; (make-coding-system
 213 ;;  'japanese-iso-7bit-1978-irv 2 ?j
 214 ;;  "ISO 2022 based 7-bit encoding for Japanese JISX0208-1978 and JISX0201-Roman"
 215 ;;  '((ascii japanese-jisx0208-1978 japanese-jisx0208
 216 ;;           latin-jisx0201 japanese-jisx0212 katakana-jisx0201 t) nil nil nil
 217 ;;    short ascii-eol ascii-cntl seven nil nil use-roman use-oldjis)
 218 ;;  '(ascii japanese-jisx0208-1978 japanese-jisx0208 latin-jisx0201))
 219
 220 (make-coding-system
 221  'iso-2022-jp-1978-irv 'iso2022
 222  "Coding-system used for old JIS terminal."
 223  '(charset-g0 ascii
 224    short t
 225    seven t
 226    output-charset-conversion ((ascii latin-jisx0201)
 227                               (japanese-jisx0208 japanese-jisx0208-1978))
 228    mnemonic "Ja-78/7bit"
 229    ))
 230
 231 ;; (define-coding-system-alias 'iso-2022-jp-1978-irv 'japanese-iso-7bit-1978-irv)
 232 ;; (define-coding-system-alias 'old-jis 'japanese-iso-7bit-1978-irv)
 233
 234 (define-coding-system-alias 'old-jis 'iso-2022-jp-1978-irv)
 235
 236 ;; (make-coding-system
 237 ;;  'japanese-iso-8bit 2 ?E
 238 ;;  "ISO 2022 based EUC encoding for Japanese (MIME:EUC-JP)"
 239 ;;  '(ascii japanese-jisx0208 katakana-jisx0201 japanese-jisx0212
 240 ;;    short ascii-eol ascii-cntl nil nil single-shift)
 241 ;;  '((safe-charsets ascii latin-jisx0201 japanese-jisx0208 japanese-jisx0208-1978
 242 ;;                  katakana-jisx0201 japanese-jisx0212)
 243 ;;    (mime-charset . euc-jp)))
 244
 245 (make-coding-system
 246  'euc-jp 'iso2022
 247  "Coding-system of Japanese EUC (Extended Unix Code)."
 248  '(charset-g0 ascii
 249    charset-g1 japanese-jisx0208
 250    charset-g2 katakana-jisx0201
 251    charset-g3 japanese-jisx0212
 252    short t
 253    mnemonic "Ja/EUC"
 254    ))
 255
 256 ;; [tomo] Moved to mule-conf.el.
 257 ;; (make-coding-system
 258 ;;  'euc-jisx0213 'iso2022
 259 ;;  "Coding-system of Japanese EUC based on JIS X 0213."
 260 ;;  '(charset-g0 ascii
 261 ;;    charset-g1 japanese-jisx0213-1
 262 ;;    charset-g2 katakana-jisx0201
 263 ;;    charset-g3 japanese-jisx0213-2
 264 ;;    short t
 265 ;;    mnemonic "Ja/EUC0213"
 266 ;;    ))
 267
 268 ;; (define-coding-system-alias 'euc-japan-1990 'japanese-iso-8bit)
 269 ;; (define-coding-system-alias 'euc-japan 'japanese-iso-8bit)
 270 ;; (define-coding-system-alias 'euc-jp 'japanese-iso-8bit)
 271
 272 (define-coding-system-alias 'euc-japan 'euc-jp) ; only for w3
 273 (define-coding-system-alias 'japanese-euc 'euc-jp)
 274
 275 (set-language-info-alist
 276  "Japanese" '((setup-function . setup-japanese-environment-internal)
 277               (exit-function . exit-japanese-environment)
 278               (tutorial . "TUTORIAL.ja")
 279               (charset japanese-jisx0208 japanese-jisx0208-1978
 280                        japanese-jisx0212 latin-jisx0201 katakana-jisx0201)
 281               (coding-system iso-2022-jp euc-jp
 282                              shift_jis iso-2022-jp-2)
 283               (coding-priority iso-2022-jp euc-jp
 284                                shift_jis iso-2022-jp-2)
 285 ;;            (input-method . "japanese")
 286               (features japan-util)
 287               (sample-text . "Japanese (\e$BF|K\8l\e(B)   \e$B$3$s$K$A$O\e(B, \e(I:]FAJ\e(B")
 288               (documentation . t)))
 289
 290 ;;; japanese.el ends here