lisp/international/characters.el

   1 ;;; characters.el --- set syntax and category for multibyte characters
   2
   3 ;; Copyright (C) 1995, 1997 Electrotechnical Laboratory, JAPAN.
   4 ;; Licensed to the Free Software Foundation.
   5
   6 ;; Keywords: multibyte character, character set, syntax, category
   7
   8 ;; This file is part of GNU Emacs.
   9
  10 ;; GNU Emacs is free software; you can redistribute it and/or modify
  11 ;; it under the terms of the GNU General Public License as published by
  12 ;; the Free Software Foundation; either version 2, or (at your option)
  13 ;; any later version.
  14
  15 ;; GNU Emacs is distributed in the hope that it will be useful,
  16 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  17 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  18 ;; GNU General Public License for more details.
  19
  20 ;; You should have received a copy of the GNU General Public License
  21 ;; along with GNU Emacs; see the file COPYING.  If not, write to the
  22 ;; Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  23 ;; Boston, MA 02111-1307, USA.
  24
  25 ;;; Commentary:
  26
  27 ;; This file contains multibyte characters.  Save this file always in
  28 ;; the coding system `iso-2022-7bit'.
  29
  30 ;; This file does not define the syntax for Latin-N character sets;
  31 ;; those are defined by the files latin-N.el.
  32
  33 ;;; Predefined categories.
  34
  35 ;; For each character set.
  36
  37 (define-category ?a "ASCII")
  38 (define-category ?l "Latin")
  39 (define-category ?t "Thai")
  40 (define-category ?g "Greek")
  41 (define-category ?b "Arabic")
  42 (define-category ?w "Hebrew")
  43 (define-category ?y "Cyrillic")
  44 (define-category ?k "Japanese katakana")
  45 (define-category ?r "Japanese roman")
  46 (define-category ?c "Chinese")
  47 (define-category ?j "Japanese")
  48 (define-category ?h "Korean")
  49 (define-category ?e "Ethiopic (Ge'ez)")
  50 (define-category ?v "Vietnamese")
  51 (define-category ?i "Indian")
  52 (define-category ?o "Lao")
  53 (define-category ?q "Tibetan")
  54
  55 ;; For each group (row) of 2-byte character sets.
  56
  57 (define-category ?A "Alpha-numeric characters of 2-byte character sets")
  58 (define-category ?C "Chinese (Han) characters of 2-byte character sets")
  59 (define-category ?G "Greek characters of 2-byte character sets")
  60 (define-category ?H "Japanese Hiragana characters of 2-byte character sets")
  61 (define-category ?K "Japanese Katakana characters of 2-byte character sets")
  62 (define-category ?N "Korean Hangul characters of 2-byte character sets")
  63 (define-category ?Y "Cyrillic characters of 2-byte character sets")
  64 (define-category ?I "Indian Glyphs")
  65
  66 ;; For phonetic classifications.
  67
  68 (define-category ?0 "consonant")
  69 (define-category ?1 "base (independent) vowel")
  70 (define-category ?2 "upper diacritical mark (including upper vowel)")
  71 (define-category ?3 "lower diacritical mark (including lower vowel)")
  72 (define-category ?4 "tone mark")
  73 (define-category ?5 "symbol")
  74 (define-category ?6 "digit")
  75 (define-category ?7 "vowel-modifying diacritical mark")
  76 (define-category ?8 "vowel-signs")
  77 (define-category ?9 "semivowel lower")
  78
  79 ;; For filling.
  80 (define-category ?| "While filling, we can break a line at this character.")
  81
  82 ;; For indentation calculation.
  83 (define-category ?
  84   "This character counts as a space for indentation purposes.")
  85
  86 ;; Keep the following for `kinsoku' processing.  See comments in
  87 ;; kinsoku.el.
  88 (define-category ?> "A character which can't be placed at beginning of line.")
  89 (define-category ?< "A character which can't be placed at end of line.")
  90
  91 \f
  92 ;;; Setting syntax and category.
  93
  94 ;; ASCII
  95
  96 (let ((ch 32))
  97   (while (< ch 127)                     ; All ASCII characters have
  98     (modify-category-entry ch ?a)       ; the category `a' (ASCII)
  99     (modify-category-entry ch ?l)       ; and `l' (Latin).
 100     (setq ch (1+ ch))))
 101
 102 ;; Arabic character set
 103
 104 (let ((charsets '(arabic-iso8859-6
 105                   arabic-digit
 106                   arabic-1-column
 107                   arabic-2-column)))
 108   (while charsets
 109     (modify-syntax-entry (make-char (car charsets)) "w")
 110     (modify-category-entry (make-char (car charsets)) ?b)
 111     (setq charsets (cdr charsets))))
 112
 113 ;; Chinese character set (GB2312)
 114
 115 (modify-syntax-entry (make-char 'chinese-gb2312) "w")
 116 (modify-syntax-entry (make-char 'chinese-gb2312 33) "_")
 117 (modify-syntax-entry (make-char 'chinese-gb2312 34) "_")
 118 (modify-syntax-entry (make-char 'chinese-gb2312 41) "_")
 119 (modify-syntax-entry ?\\e$A!2\e(B "(\e$A!3\e(B")
 120 (modify-syntax-entry ?\\e$A!4\e(B "(\e$A!5\e(B")
 121 (modify-syntax-entry ?\\e$A!6\e(B "(\e$A!7\e(B")
 122 (modify-syntax-entry ?\\e$A!8\e(B "(\e$A!9\e(B")
 123 (modify-syntax-entry ?\\e$A!:\e(B "(\e$A!;\e(B")
 124 (modify-syntax-entry ?\\e$A!<\e(B "(\e$A!=\e(B")
 125 (modify-syntax-entry ?\\e$A!>\e(B "(\e$A!?\e(B")
 126 (modify-syntax-entry ?\\e$A!3\e(B ")\e$A!2\e(B")
 127 (modify-syntax-entry ?\\e$A!5\e(B ")\e$A!4\e(B")
 128 (modify-syntax-entry ?\\e$A!7\e(B ")\e$A!6\e(B")
 129 (modify-syntax-entry ?\\e$A!9\e(B ")\e$A!8\e(B")
 130 (modify-syntax-entry ?\\e$A!;\e(B ")\e$A!:\e(B")
 131 (modify-syntax-entry ?\\e$A!=\e(B ")\e$A!<\e(B")
 132 (modify-syntax-entry ?\\e$A!?\e(B ")\e$A!>\e(B")
 133
 134 (modify-category-entry (make-char 'chinese-gb2312) ?c)
 135 (modify-category-entry (make-char 'chinese-gb2312) ?\|)
 136 (modify-category-entry (make-char 'chinese-gb2312 35) ?A)
 137 (modify-category-entry (make-char 'chinese-gb2312 36) ?H)
 138 (modify-category-entry (make-char 'chinese-gb2312 37) ?K)
 139 (modify-category-entry (make-char 'chinese-gb2312 38) ?G)
 140 (modify-category-entry (make-char 'chinese-gb2312 39) ?Y)
 141 (let ((row 48))
 142   (while (< row 127)
 143     (modify-category-entry (make-char 'chinese-gb2312 row) ?C)
 144     (setq row (1+ row))))
 145
 146 ;; Chinese character set (BIG5)
 147
 148 (let ((generic-big5-1-char (make-char 'chinese-big5-1))
 149       (generic-big5-2-char (make-char 'chinese-big5-2)))
 150   (modify-syntax-entry generic-big5-1-char "w")
 151   (modify-syntax-entry generic-big5-2-char "w")
 152
 153   (modify-category-entry generic-big5-1-char ?c)
 154   (modify-category-entry generic-big5-2-char ?c)
 155
 156   (modify-category-entry generic-big5-1-char ?C)
 157   (modify-category-entry generic-big5-2-char ?C)
 158
 159   (modify-category-entry generic-big5-1-char ?\|)
 160   (modify-category-entry generic-big5-2-char ?\|))
 161
 162
 163 ;; Chinese character set (CNS11643)
 164
 165 (let ((cns-list '(chinese-cns11643-1
 166                   chinese-cns11643-2
 167                   chinese-cns11643-3
 168                   chinese-cns11643-4
 169                   chinese-cns11643-5
 170                   chinese-cns11643-6
 171                   chinese-cns11643-7))
 172       generic-char)
 173   (while cns-list
 174     (setq generic-char (make-char (car cns-list)))
 175     (modify-syntax-entry generic-char "w")
 176     (modify-category-entry generic-char ?c)
 177     (modify-category-entry generic-char ?C)
 178     (modify-category-entry generic-char ?|)
 179     (setq cns-list (cdr cns-list))))
 180
 181 ;; Cyrillic character set (ISO-8859-5)
 182
 183 (modify-category-entry (make-char 'cyrillic-iso8859-5) ?y)
 184
 185 (modify-syntax-entry (make-char 'cyrillic-iso8859-5 160) " ")
 186 (modify-syntax-entry ?\e,L-\e(B ".")
 187 (modify-syntax-entry ?\e,Lp\e(B ".")
 188 (modify-syntax-entry ?\e,L}\e(B ".")
 189 (let ((tbl (standard-case-table)))
 190   (set-case-syntax-pair ?\e,L!\e(B ?\e,Lq\e(B tbl)
 191   (set-case-syntax-pair ?\e,L"\e(B ?\e,Lr\e(B tbl)
 192   (set-case-syntax-pair ?\e,L#\e(B ?\e,Ls\e(B tbl)
 193   (set-case-syntax-pair ?\e,L$\e(B ?\e,Lt\e(B tbl)
 194   (set-case-syntax-pair ?\e,L%\e(B ?\e,Lu\e(B tbl)
 195   (set-case-syntax-pair ?\e,L&\e(B ?\e,Lv\e(B tbl)
 196   (set-case-syntax-pair ?\e,L'\e(B ?\e,Lw\e(B tbl)
 197   (set-case-syntax-pair ?\e,L(\e(B ?\e,Lx\e(B tbl)
 198   (set-case-syntax-pair ?\e,L)\e(B ?\e,Ly\e(B tbl)
 199   (set-case-syntax-pair ?\e,L*\e(B ?\e,Lz\e(B tbl)
 200   (set-case-syntax-pair ?\e,L+\e(B ?\e,L{\e(B tbl)
 201   (set-case-syntax-pair ?\e,L,\e(B ?\e,L|\e(B tbl)
 202   (set-case-syntax-pair ?\e,L.\e(B ?\e,L~\e(B tbl)
 203   (set-case-syntax-pair ?\e,L/\e(B ?\e,L\x7f\e(B tbl)
 204   (set-case-syntax-pair ?\e,L0\e(B ?\e,LP\e(B tbl)
 205   (set-case-syntax-pair ?\e,L1\e(B ?\e,LQ\e(B tbl)
 206   (set-case-syntax-pair ?\e,L2\e(B ?\e,LR\e(B tbl)
 207   (set-case-syntax-pair ?\e,L3\e(B ?\e,LS\e(B tbl)
 208   (set-case-syntax-pair ?\e,L4\e(B ?\e,LT\e(B tbl)
 209   (set-case-syntax-pair ?\e,L5\e(B ?\e,LU\e(B tbl)
 210   (set-case-syntax-pair ?\e,L6\e(B ?\e,LV\e(B tbl)
 211   (set-case-syntax-pair ?\e,L7\e(B ?\e,LW\e(B tbl)
 212   (set-case-syntax-pair ?\e,L8\e(B ?\e,LX\e(B tbl)
 213   (set-case-syntax-pair ?\e,L9\e(B ?\e,LY\e(B tbl)
 214   (set-case-syntax-pair ?\e,L:\e(B ?\e,LZ\e(B tbl)
 215   (set-case-syntax-pair ?\e,L;\e(B ?\e,L[\e(B tbl)
 216   (set-case-syntax-pair ?\e,L<\e(B ?\e,L\\e(B tbl)
 217   (set-case-syntax-pair ?\e,L=\e(B ?\e,L]\e(B tbl)
 218   (set-case-syntax-pair ?\e,L>\e(B ?\e,L^\e(B tbl)
 219   (set-case-syntax-pair ?\e,L?\e(B ?\e,L_\e(B tbl)
 220   (set-case-syntax-pair ?\e,L@\e(B ?\e,L`\e(B tbl)
 221   (set-case-syntax-pair ?\e,LA\e(B ?\e,La\e(B tbl)
 222   (set-case-syntax-pair ?\e,LB\e(B ?\e,Lb\e(B tbl)
 223   (set-case-syntax-pair ?\e,LC\e(B ?\e,Lc\e(B tbl)
 224   (set-case-syntax-pair ?\e,LD\e(B ?\e,Ld\e(B tbl)
 225   (set-case-syntax-pair ?\e,LE\e(B ?\e,Le\e(B tbl)
 226   (set-case-syntax-pair ?\e,LF\e(B ?\e,Lf\e(B tbl)
 227   (set-case-syntax-pair ?\e,LG\e(B ?\e,Lg\e(B tbl)
 228   (set-case-syntax-pair ?\e,LH\e(B ?\e,Lh\e(B tbl)
 229   (set-case-syntax-pair ?\e,LI\e(B ?\e,Li\e(B tbl)
 230   (set-case-syntax-pair ?\e,LJ\e(B ?\e,Lj\e(B tbl)
 231   (set-case-syntax-pair ?\e,LK\e(B ?\e,Lk\e(B tbl)
 232   (set-case-syntax-pair ?\e,LL\e(B ?\e,Ll\e(B tbl)
 233   (set-case-syntax-pair ?\e,LM\e(B ?\e,Lm\e(B tbl)
 234   (set-case-syntax-pair ?\e,LN\e(B ?\e,Ln\e(B tbl)
 235   (set-case-syntax-pair ?\e,LO\e(B ?\e,Lo\e(B tbl))
 236
 237 ;; Devanagari character set
 238
 239 (let ((deflist  '(;; chars      syntax  category
 240                   ("\e$(5!!!"!#\e(B"      "w"     ?7) ; vowel-modifying diacritical mark
 241                                             ; chandrabindu, anuswar, visarga
 242                   ("\e$(5!$\e(B-\e$(5!2\e(B"        "w"     ?1) ; independent vowel
 243                   ("\e$(5!3\e(B-\e$(5!X\e(B"        "w"     ?0) ; consonant
 244                   ("\e$(5!Z\e(B-\e$(5!g\e(B"        "w"     ?8) ; matra
 245                   ("\e$(5!q\e(B-\e$(5!z\e(B"        "w"     ?6) ; digit
 246                   ))
 247       elm chars len syntax category to ch i)
 248   (while deflist
 249     (setq elm (car deflist))
 250     (setq chars (car elm)
 251           len (length chars)
 252           syntax (nth 1 elm)
 253           category (nth 2 elm)
 254           i 0)
 255     (while (< i len)
 256       (if (= (aref chars i) ?-)
 257           (setq i (1+ i)
 258                 to (aref chars i))
 259         (setq ch (aref chars i)
 260               to ch))
 261       (while (<= ch to)
 262         (modify-syntax-entry ch syntax)
 263         (modify-category-entry ch category)
 264         (setq ch (1+ ch)))
 265       (setq i (1+ i)))
 266     (setq deflist (cdr deflist))))
 267
 268 ;; Ethiopic character set
 269
 270 (modify-category-entry (make-char 'ethiopic) ?e)
 271 (modify-syntax-entry (make-char 'ethiopic) "w")
 272 (let ((chars '(?\e$(3$h\e(B ?\e$(3$i\e(B ?\e$(3$j\e(B ?\e$(3$k\e(B ?\e$(3$l\e(B ?\e$(3$m\e(B ?\e$(3$n\e(B ?\e$(3$o\e(B ?\e$(3%i\e(B ?\e$(3%t\e(B ?\e$(3%u\e(B ?\e$(3%v\e(B ?\e$(3%w\e(B ?\e$(3%x\e(B)))
 273   (while chars
 274     (modify-syntax-entry (car chars) ".")
 275     (setq chars (cdr chars))))
 276
 277 ;; European character set (Latin-1,2,3,4,5)
 278
 279 (modify-category-entry (make-char 'latin-iso8859-1) ?l)
 280 (modify-category-entry (make-char 'latin-iso8859-2) ?l)
 281 (modify-category-entry (make-char 'latin-iso8859-3) ?l)
 282 (modify-category-entry (make-char 'latin-iso8859-4) ?l)
 283 (modify-category-entry (make-char 'latin-iso8859-9) ?l)
 284 (modify-category-entry (make-char 'latin-iso8859-14) ?l)
 285 (modify-category-entry (make-char 'latin-iso8859-15) ?l)
 286
 287 (modify-category-entry (make-char 'latin-iso8859-1 160) ?\ )
 288 (modify-category-entry (make-char 'latin-iso8859-2 160) ?\ )
 289 (modify-category-entry (make-char 'latin-iso8859-3 160) ?\ )
 290 (modify-category-entry (make-char 'latin-iso8859-4 160) ?\ )
 291 (modify-category-entry (make-char 'latin-iso8859-9 160) ?\ )
 292 (modify-category-entry (make-char 'latin-iso8859-14 160) ?\ )
 293 (modify-category-entry (make-char 'latin-iso8859-15 160) ?\ )
 294
 295 ;; Greek character set (ISO-8859-7)
 296
 297 (modify-category-entry (make-char 'greek-iso8859-7) ?g)
 298
 299 (let ((c 182))
 300   (while (< c 255)
 301     (modify-syntax-entry (make-char 'greek-iso8859-7 c) "w")
 302     (setq c (1+ c))))
 303 (modify-syntax-entry (make-char 'greek-iso8859-7 160) "w") ; NBSP
 304 (modify-syntax-entry ?\e,F7\e(B ".")
 305 (modify-syntax-entry ?\e,F;\e(B ".")
 306 (modify-syntax-entry ?\e,F=\e(B ".")
 307
 308 ;; Hebrew character set (ISO-8859-8)
 309
 310 (modify-category-entry (make-char 'hebrew-iso8859-8) ?w)
 311
 312 (let ((c 224))
 313   (while (< c 251)
 314     (modify-syntax-entry (make-char 'hebrew-iso8859-8 c) "w")
 315     (setq c (1+ c))))
 316 (modify-syntax-entry (make-char 'hebrew-iso8859-8 160) "w") ; NBSP
 317
 318 ;; Indian character set (IS 13194 and other Emacs original Indian charsets)
 319
 320 (modify-category-entry (make-char 'indian-is13194) ?i)
 321 (modify-category-entry (make-char 'indian-2-column) ?I)
 322 (modify-category-entry (make-char 'indian-1-column) ?I)
 323
 324 (let ((deflist
 325         '(;; chars      syntax  category
 326           ("\e(5!"#\e(B"  "w"     ?7) ; vowel-modifying diacritical mark
 327                                     ; chandrabindu, anuswar, visarga
 328           ("\e(5$\e(B-\e(52\e(B"    "w"     ?1) ; base (independent) vowel
 329           ("\e(53\e(B-\e(5X\e(B"    "w"     ?0) ; consonant
 330           ("\e(5Z\e(B-\e(5g\e(B"    "w"     ?8) ; matra
 331           ("\e(5q\e(B-\e(5z\e(B"    "w"     ?6) ; digit
 332           ))
 333       elm chars len syntax category to ch i)
 334   (while deflist
 335     (setq elm (car deflist))
 336     (setq chars (car elm)
 337           len (length chars)
 338           syntax (nth 1 elm)
 339           category (nth 2 elm)
 340           i 0)
 341     (while (< i len)
 342       (if (= (aref chars i) ?-)
 343           (setq i (1+ i)
 344                 to (aref chars i))
 345         (setq ch (aref chars i)
 346               to ch))
 347       (while (<= ch to)
 348         (modify-syntax-entry ch syntax)
 349         (modify-category-entry ch category)
 350         (setq ch (1+ ch)))
 351       (setq i (1+ i)))
 352     (setq deflist (cdr deflist))))
 353
 354
 355 ;; Japanese character set (JISX0201-kana, JISX0201-roman, JISX0208, JISX0212)
 356
 357 (modify-category-entry (make-char 'katakana-jisx0201) ?k)
 358 (modify-category-entry (make-char 'latin-jisx0201) ?r)
 359 (modify-category-entry (make-char 'japanese-jisx0208) ?j)
 360 (modify-category-entry (make-char 'japanese-jisx0212) ?j)
 361 (modify-category-entry (make-char 'katakana-jisx0201) ?\|)
 362 (modify-category-entry (make-char 'japanese-jisx0208) ?\|)
 363 (modify-category-entry (make-char 'japanese-jisx0212) ?\|)
 364
 365 ;; JISX0208
 366 (modify-syntax-entry (make-char 'japanese-jisx0208) "w")
 367 (modify-syntax-entry (make-char 'japanese-jisx0208 33) "_")
 368 (modify-syntax-entry (make-char 'japanese-jisx0208 34) "_")
 369 (modify-syntax-entry (make-char 'japanese-jisx0208 40) "_")
 370 (let ((chars '(?\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)))
 371   (while chars
 372     (modify-syntax-entry (car chars) "w")
 373     (setq chars (cdr chars))))
 374 (modify-syntax-entry ?\\e$B!J\e(B "(\e$B!K\e(B")
 375 (modify-syntax-entry ?\\e$B!N\e(B "(\e$B!O\e(B")
 376 (modify-syntax-entry ?\\e$B!P\e(B "(\e$B!Q\e(B")
 377 (modify-syntax-entry ?\\e$B!V\e(B "(\e$B!W\e(B")
 378 (modify-syntax-entry ?\\e$B!X\e(B "(\e$B!Y\e(B")
 379 (modify-syntax-entry ?\\e$B!K\e(B ")\e$B!J\e(B")
 380 (modify-syntax-entry ?\\e$B!O\e(B ")\e$B!N\e(B")
 381 (modify-syntax-entry ?\\e$B!Q\e(B ")\e$B!P\e(B")
 382 (modify-syntax-entry ?\\e$B!W\e(B ")\e$B!V\e(B")
 383 (modify-syntax-entry ?\\e$B!Y\e(B ")\e$B!X\e(B")
 384
 385 (modify-category-entry (make-char 'japanese-jisx0208 35) ?A)
 386 (modify-category-entry (make-char 'japanese-jisx0208 36) ?H)
 387 (modify-category-entry (make-char 'japanese-jisx0208 37) ?K)
 388 (modify-category-entry (make-char 'japanese-jisx0208 38) ?G)
 389 (modify-category-entry (make-char 'japanese-jisx0208 39) ?Y)
 390 (let ((row 48))
 391   (while (< row 127)
 392     (modify-category-entry (make-char 'japanese-jisx0208 row) ?C)
 393     (setq row (1+ row))))
 394 (modify-category-entry ?\e$B!<\e(B ?K)
 395 (let ((chars '(?\e$B!+\e(B ?\e$B!,\e(B)))
 396   (while chars
 397     (modify-category-entry (car chars) ?K)
 398     (modify-category-entry (car chars) ?H)
 399     (setq chars (cdr chars))))
 400 (let ((chars '(?\e$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)))
 401   (while chars
 402     (modify-category-entry (car chars) ?C)
 403     (setq chars (cdr chars))))
 404
 405 ;; JISX0212
 406 (modify-syntax-entry (make-char 'japanese-jisx0212) "w")
 407 (modify-syntax-entry (make-char 'japanese-jisx0212 33) "_")
 408 (modify-syntax-entry (make-char 'japanese-jisx0212 34) "_")
 409 (modify-syntax-entry (make-char 'japanese-jisx0212 35) "_")
 410
 411 (modify-category-entry (make-char 'japanese-jisx0212 ) ?C)
 412
 413 ;; JISX0201-Kana
 414 (modify-syntax-entry (make-char 'katakana-jisx0201) "w")
 415 (let ((chars '(?\e(I!\e(B ?\e(I"\e(B ?\e(I#\e(B ?\e(I$\e(B ?\e(I%\e(B)))
 416   (while chars
 417     (modify-syntax-entry (car chars) ".")
 418     (setq chars (cdr chars))))
 419
 420 ;; Korean character set (KSC5601)
 421
 422 (modify-syntax-entry (make-char 'korean-ksc5601) "w")
 423 (modify-syntax-entry (make-char 'korean-ksc5601 33) "_")
 424 (modify-syntax-entry (make-char 'korean-ksc5601 34) "_")
 425 (modify-syntax-entry (make-char 'korean-ksc5601 38) "_")
 426 (modify-syntax-entry (make-char 'korean-ksc5601 39) "_")
 427 (modify-syntax-entry (make-char 'korean-ksc5601 40) "_")
 428 (modify-syntax-entry (make-char 'korean-ksc5601 41) "_")
 429
 430 (modify-category-entry (make-char 'korean-ksc5601) ?h)
 431 (modify-category-entry (make-char 'korean-ksc5601 35) ?A)
 432 (modify-category-entry (make-char 'korean-ksc5601 37) ?G)
 433 (modify-category-entry (make-char 'korean-ksc5601 42) ?H)
 434 (modify-category-entry (make-char 'korean-ksc5601 43) ?K)
 435 (modify-category-entry (make-char 'korean-ksc5601 44) ?Y)
 436
 437 ;; Lao character set
 438
 439 (modify-category-entry (make-char 'lao) ?o)
 440
 441 (let ((deflist  '(;; chars      syntax  category
 442                   ("\e(1!\e(B-\e(1N\e(B"    "w"     ?0) ; consonant
 443                   ("\e(1PRS]`\e(B-\e(1d\e(B"        "w"     ?1) ; vowel base
 444                   ("\e(1QT\e(B-\e(1W[m\e(B" "w"     ?2) ; vowel upper
 445                   ("\e(1XY\e(B"           "w"     ?3) ; vowel lower
 446                   ("\e(1h\e(B-\e(1l\e(B"    "w"     ?4) ; tone mark
 447                   ("\e(1\\e(B"            "w"     ?9) ; semivowel lower
 448                   ("\e(1p\e(B-\e(1y\e(B"    "w"     ?6) ; digit
 449                   ("\e(1Of\e(B"           "_"     ?5) ; symbol
 450                   ))
 451       elm chars len syntax category to ch i)
 452   (while deflist
 453     (setq elm (car deflist))
 454     (setq chars (car elm)
 455           len (length chars)
 456           syntax (nth 1 elm)
 457           category (nth 2 elm)
 458           i 0)
 459     (while (< i len)
 460       (if (= (aref chars i) ?-)
 461           (setq i (1+ i)
 462                 to (aref chars i))
 463         (setq ch (aref chars i)
 464               to ch))
 465       (while (<= ch to)
 466         (modify-syntax-entry ch syntax)
 467         (modify-category-entry ch category)
 468         (setq ch (1+ ch)))
 469       (setq i (1+ i)))
 470     (setq deflist (cdr deflist))))
 471
 472 ;; Thai character set (TIS620)
 473
 474 (modify-category-entry (make-char 'thai-tis620) ?t)
 475
 476 (let ((deflist  '(;; chars      syntax  category
 477                   ("\e,T!\e(B-\e,TCEG\e(B-\e,TN\e(B"  "w"     ?0) ; consonant
 478                   ("\e,TDFPRS`\e(B-\e,Te\e(B"       "w"     ?1) ; vowel base
 479                   ("\e,TQT\e(B-\e,TWgn\e(B" "w"     ?2) ; vowel upper
 480                   ("\e,TX\e(B-\e,TZ\e(B"    "w"     ?3) ; vowel lower
 481                   ("\e,Th\e(B-\e,Tm\e(B"    "w"     ?4) ; tone mark
 482                   ("\e,Tp\e(B-\e,Ty\e(B"    "w"     ?6) ; digit
 483                   ("\e,TOf_oz{\e(B"       "_"     ?5) ; symbol
 484                   ))
 485       elm chars len syntax category to ch i)
 486   (while deflist
 487     (setq elm (car deflist))
 488     (setq chars (car elm)
 489           len (length chars)
 490           syntax (nth 1 elm)
 491           category (nth 2 elm)
 492           i 0)
 493     (while (< i len)
 494       (if (= (aref chars i) ?-)
 495           (setq i (1+ i)
 496                 to (aref chars i))
 497         (setq ch (aref chars i)
 498               to ch))
 499       (while (<= ch to)
 500         (modify-syntax-entry ch syntax)
 501         (modify-category-entry ch category)
 502         (setq ch (1+ ch)))
 503       (setq i (1+ i)))
 504     (setq deflist (cdr deflist))))
 505
 506 ;; Tibetan character set
 507
 508 (modify-category-entry (make-char 'tibetan) ?q)
 509 (modify-category-entry (make-char 'tibetan-1-column) ?q)
 510
 511 (let ((deflist  '(;; chars             syntax category
 512                   ("\e$(7"!\e(B-\e$(7"J\e(B"                "w"     ?0) ; consonant
 513                   ("\e$(7#!\e(B-\e$(7#J#P#Q\e(B"          "w"     ?0) ;
 514                   ("\e$(7$!\e(B-\e$(7$e\e(B"              "w"     ?0) ;
 515                   ("\e$(7%!\e(B-\e$(7%u\e(B"              "w"     ?0) ;
 516                   ("\e$(7"S"["\"]"^"a\e(B"       "w"      ?2) ; upper vowel
 517                   ("\e$(7"_"c"d"g"h"i"j"k"l\e(B" "w"      ?2) ; upper modifier
 518                   ("\e$(7!I"Q"U"e!e!g\e(B"       "w"      ?3) ; lowel vowel/modifier
 519                   ("\e$(7!P\e(B-\e$(7!Y!Z\e(B-\e$(7!c\e(B"            "w"     ?6) ; digit
 520                   ("\e$(7!;!=\e(B-\e$(7!B!D"`\e(B"        "."     ?|) ; line-break char
 521                   ("\e$(8!;!=!?!@!A!D"`\e(B"            "."     ?|) ;
 522                   ("\e$(7!8!;!=\e(B-\e$(7!B!D"`!m!d\e(B"  "."     ?>) ; prohibition
 523                   ("\e$(8!;!=!?!@!A!D"`\e(B"            "."     ?>) ;
 524                   ("\e$(7!0\e(B-\e$(7!:!l#R#S"f\e(B"      "."     ?<) ; prohibition
 525                   ("\e$(7!C!E\e(B-\e$(7!H!J\e(B-\e$(7!O!f!h\e(B-\e$(7!k!n!o\e(B" "." ?q) ; others
 526                   ))
 527       elm chars len syntax category to ch i)
 528   (while deflist
 529     (setq elm (car deflist))
 530     (setq chars (car elm)
 531           len (length chars)
 532           syntax (nth 1 elm)
 533           category (nth 2 elm)
 534           i 0)
 535     (while (< i len)
 536       (if (= (aref chars i) ?-)
 537           (setq i (1+ i)
 538                 to (aref chars i))
 539         (setq ch (aref chars i)
 540               to ch))
 541       (while (<= ch to)
 542         (modify-syntax-entry ch syntax)
 543         (modify-category-entry ch category)
 544         (setq ch (1+ ch)))
 545       (setq i (1+ i)))
 546     (setq deflist (cdr deflist))))
 547
 548 ;; Vietnamese character set
 549
 550 (let ((lower (make-char 'vietnamese-viscii-lower))
 551       (upper (make-char 'vietnamese-viscii-upper)))
 552   (modify-syntax-entry lower "w")
 553   (modify-syntax-entry upper "w")
 554   (modify-category-entry lower ?v)
 555   (modify-category-entry upper ?v)
 556   (modify-category-entry lower ?l)      ; To make a word with
 557   (modify-category-entry upper ?l)      ; latin characters.
 558   )
 559
 560 (let ((tbl (standard-case-table))
 561       (i 32))
 562   (while (< i 128)
 563     (set-case-syntax-pair (make-char 'vietnamese-viscii-upper i)
 564                           (make-char 'vietnamese-viscii-lower i)
 565                           tbl)
 566     (setq i (1+ i))))
 567
 568 \f
 569 ;;; Setting word boundary.
 570
 571 (setq word-combining-categories
 572       '((?l . ?l)))
 573
 574 (setq word-separating-categories        ;  (2-byte character sets)
 575       '((?A . ?K)                       ; Alpha numeric - Katakana
 576         (?A . ?C)                       ; Alpha numeric - Chinese
 577         (?H . ?A)                       ; Hiragana - Alpha numeric
 578         (?H . ?K)                       ; Hiragana - Katakana
 579         (?H . ?C)                       ; Hiragana - Chinese
 580         (?K . ?A)                       ; Katakana - Alpha numeric
 581         (?K . ?C)                       ; Katakana - Chinese
 582         (?C . ?A)                       ; Chinese - Alpha numeric
 583         (?C . ?K)                       ; Chinese - Katakana
 584         ))
 585
 586 \f
 587 ;; For each character set, put the information of the most proper
 588 ;; coding system to encode it by `preferred-coding-system' property.
 589
 590 (let ((l '((latin-iso8859-1     . iso-latin-1)
 591            (latin-iso8859-2     . iso-latin-2)
 592            (latin-iso8859-3     . iso-latin-3)
 593            (latin-iso8859-4     . iso-latin-4)
 594            (thai-tis620         . thai-tis620)
 595            (greek-iso8859-7     . greek-iso-8bit)
 596            (arabic-iso8859-6    . iso-2022-7bit)
 597            (hebrew-iso8859-8    . hebrew-iso-8bit)
 598            (katakana-jisx0201   . japanese-shift-jis)
 599            (latin-jisx0201      . japanese-shift-jis)
 600            (cyrillic-iso8859-5  . cyrillic-iso-8bit)
 601            (latin-iso8859-9     . iso-latin-5)
 602            (japanese-jisx0208-1978 . iso-2022-jp)
 603            (chinese-gb2312      . cn-gb-2312)
 604            (japanese-jisx0208   . iso-2022-jp)
 605            (korean-ksc5601      . iso-2022-kr)
 606            (japanese-jisx0212   . iso-2022-jp)
 607            (chinese-cns11643-1  . iso-2022-cn)
 608            (chinese-cns11643-2  . iso-2022-cn)
 609            (chinese-big5-1      . chinese-big5)
 610            (chinese-big5-2      . chinese-big5)
 611            (chinese-sisheng     . iso-2022-7bit)
 612            (ipa                 . iso-2022-7bit)
 613            (vietnamese-viscii-lower . vietnamese-viscii)
 614            (vietnamese-viscii-upper . vietnamese-viscii)
 615            (arabic-digit        . iso-2022-7bit)
 616            (arabic-1-column     . iso-2022-7bit)
 617            (ascii-right-to-left . iso-2022-7bit)
 618            (lao                 . lao)
 619            (arabic-2-column     . iso-2022-7bit)
 620            (indian-is13194      . devanagari)
 621            (indian-1-column     . devanagari)
 622            (tibetan-1-column    . tibetan)
 623            (ethiopic            . iso-2022-7bit)
 624            (chinese-cns11643-3  . iso-2022-cn)
 625            (chinese-cns11643-4  . iso-2022-cn)
 626            (chinese-cns11643-5  . iso-2022-cn)
 627            (chinese-cns11643-6  . iso-2022-cn)
 628            (chinese-cns11643-7  . iso-2022-cn)
 629            (indian-2-column     . devanagari)
 630            (tibetan             . tibetan)
 631            (latin-iso8859-14    . iso-latin-8)
 632            (latin-iso8859-15    . iso-latin-9))))
 633   (while l
 634     (put-charset-property (car (car l)) 'preferred-coding-system (cdr (car l)))
 635     (setq l (cdr l))))
 636
 637 \f
 638 ;; Setup auto-fill-chars for charsets that should invoke auto-filling.
 639 ;; SPACE and NEWLIE are already set.  Also put `nospace-between-words'
 640 ;; property to the charsets.
 641 (let ((l '(katakana-jisx0201
 642            japanese-jisx0208 japanese-jisx0212
 643            chinese-gb2312 chinese-big5-1 chinese-big5-2)))
 644   (while l
 645     (aset auto-fill-chars (make-char (car l)) t)
 646     (put-charset-property (car l) 'nospace-between-words t)
 647     (setq l (cdr l))))
 648
 649 ;;; Local Variables:
 650 ;;; coding: iso-2022-7bit
 651 ;;; End:
 652
 653 ;;; end of characters.el