code.delx.au - gnu-emacs/blob - 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 characters 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 character 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 followings 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 (modify-category-entry (make-char 'chinese-gb2312 35) ?A)
 142 (let ((row 48))
 143   (while (< row 127)
 144     (modify-category-entry (make-char 'chinese-gb2312 row) ?C)
 145     (setq row (1+ row))))
 146
 147 ;; Chinese character set (BIG5)
 148
 149 (let ((generic-big5-1-char (make-char 'chinese-big5-1))
 150       (generic-big5-2-char (make-char 'chinese-big5-2)))
 151   (modify-syntax-entry generic-big5-1-char "w")
 152   (modify-syntax-entry generic-big5-2-char "w")
 153
 154   (modify-category-entry generic-big5-1-char ?c)
 155   (modify-category-entry generic-big5-2-char ?c)
 156
 157   (modify-category-entry generic-big5-1-char ?C)
 158   (modify-category-entry generic-big5-2-char ?C)
 159
 160   (modify-category-entry generic-big5-1-char ?\|)
 161   (modify-category-entry generic-big5-2-char ?\|))
 162
 163
 164 ;; Chinese character set (CNS11643)
 165
 166 (let ((cns-list '(chinese-cns11643-1
 167                   chinese-cns11643-2
 168                   chinese-cns11643-3
 169                   chinese-cns11643-4
 170                   chinese-cns11643-5
 171                   chinese-cns11643-6
 172                   chinese-cns11643-7))
 173       generic-char)
 174   (while cns-list
 175     (setq generic-char (make-char (car cns-list)))
 176     (modify-syntax-entry generic-char "w")
 177     (modify-category-entry generic-char ?c)
 178     (modify-category-entry generic-char ?C)
 179     (modify-category-entry generic-char ?|)
 180     (setq cns-list (cdr cns-list))))
 181
 182 ;; Cyrillic character set (ISO-8859-5)
 183
 184 (modify-category-entry (make-char 'cyrillic-iso8859-5) ?y)
 185
 186 (modify-syntax-entry (make-char 'cyrillic-iso8859-5 160) " ")
 187 (modify-syntax-entry ?\e,L-\e(B ".")
 188 (modify-syntax-entry ?\e,Lp\e(B ".")
 189 (modify-syntax-entry ?\e,L}\e(B ".")
 190 (let ((tbl (standard-case-table)))
 191   (set-case-syntax-pair ?\e,L!\e(B ?\e,Lq\e(B tbl)
 192   (set-case-syntax-pair ?\e,L"\e(B ?\e,Lr\e(B tbl)
 193   (set-case-syntax-pair ?\e,L#\e(B ?\e,Ls\e(B tbl)
 194   (set-case-syntax-pair ?\e,L$\e(B ?\e,Lt\e(B tbl)
 195   (set-case-syntax-pair ?\e,L%\e(B ?\e,Lu\e(B tbl)
 196   (set-case-syntax-pair ?\e,L&\e(B ?\e,Lv\e(B tbl)
 197   (set-case-syntax-pair ?\e,L'\e(B ?\e,Lw\e(B tbl)
 198   (set-case-syntax-pair ?\e,L(\e(B ?\e,Lx\e(B tbl)
 199   (set-case-syntax-pair ?\e,L)\e(B ?\e,Ly\e(B tbl)
 200   (set-case-syntax-pair ?\e,L*\e(B ?\e,Lz\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~\e(B tbl)
 204   (set-case-syntax-pair ?\e,L/\e(B ?\e,L\7f\e(B tbl)
 205   (set-case-syntax-pair ?\e,L0\e(B ?\e,LP\e(B tbl)
 206   (set-case-syntax-pair ?\e,L1\e(B ?\e,LQ\e(B tbl)
 207   (set-case-syntax-pair ?\e,L2\e(B ?\e,LR\e(B tbl)
 208   (set-case-syntax-pair ?\e,L3\e(B ?\e,LS\e(B tbl)
 209   (set-case-syntax-pair ?\e,L4\e(B ?\e,LT\e(B tbl)
 210   (set-case-syntax-pair ?\e,L5\e(B ?\e,LU\e(B tbl)
 211   (set-case-syntax-pair ?\e,L6\e(B ?\e,LV\e(B tbl)
 212   (set-case-syntax-pair ?\e,L7\e(B ?\e,LW\e(B tbl)
 213   (set-case-syntax-pair ?\e,L8\e(B ?\e,LX\e(B tbl)
 214   (set-case-syntax-pair ?\e,L9\e(B ?\e,LY\e(B tbl)
 215   (set-case-syntax-pair ?\e,L:\e(B ?\e,LZ\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,L@\e(B ?\e,L`\e(B tbl)
 222   (set-case-syntax-pair ?\e,LA\e(B ?\e,La\e(B tbl)
 223   (set-case-syntax-pair ?\e,LB\e(B ?\e,Lb\e(B tbl)
 224   (set-case-syntax-pair ?\e,LC\e(B ?\e,Lc\e(B tbl)
 225   (set-case-syntax-pair ?\e,LD\e(B ?\e,Ld\e(B tbl)
 226   (set-case-syntax-pair ?\e,LE\e(B ?\e,Le\e(B tbl)
 227   (set-case-syntax-pair ?\e,LF\e(B ?\e,Lf\e(B tbl)
 228   (set-case-syntax-pair ?\e,LG\e(B ?\e,Lg\e(B tbl)
 229   (set-case-syntax-pair ?\e,LH\e(B ?\e,Lh\e(B tbl)
 230   (set-case-syntax-pair ?\e,LI\e(B ?\e,Li\e(B tbl)
 231   (set-case-syntax-pair ?\e,LJ\e(B ?\e,Lj\e(B tbl)
 232   (set-case-syntax-pair ?\e,LK\e(B ?\e,Lk\e(B tbl)
 233   (set-case-syntax-pair ?\e,LL\e(B ?\e,Ll\e(B tbl)
 234   (set-case-syntax-pair ?\e,LM\e(B ?\e,Lm\e(B tbl)
 235   (set-case-syntax-pair ?\e,LN\e(B ?\e,Ln\e(B tbl)
 236   (set-case-syntax-pair ?\e,LO\e(B ?\e,Lo\e(B tbl))
 237
 238 ;; Devanagari character set
 239
 240 (let ((deflist  '(;; chars      syntax  category
 241                   ("\e$(5!!!"!#\e(B"      "w"     ?7) ; vowel-modifying diacritical mark
 242                                             ; chandrabindu, anuswar, visarga
 243                   ("\e$(5!$\e(B-\e$(5!2\e(B"        "w"     ?1) ; independent vowel
 244                   ("\e$(5!3\e(B-\e$(5!X\e(B"        "w"     ?0) ; consonant
 245                   ("\e$(5!Z\e(B-\e$(5!g\e(B"        "w"     ?8) ; matra
 246                   ("\e$(5!q\e(B-\e$(5!z\e(B"        "w"     ?6) ; digit
 247                   ))
 248       elm chars len syntax category to ch i)
 249   (while deflist
 250     (setq elm (car deflist))
 251     (setq chars (car elm)
 252           len (length chars)
 253           syntax (nth 1 elm)
 254           category (nth 2 elm)
 255           i 0)
 256     (while (< i len)
 257       (if (= (aref chars i) ?-)
 258           (setq i (1+ i)
 259                 to (sref chars i))
 260         (setq ch (sref chars i)
 261               to ch))
 262       (while (<= ch to)
 263         (modify-syntax-entry ch syntax)
 264         (modify-category-entry ch category)
 265         (setq ch (1+ ch)))
 266       (setq i (+ i (char-bytes to))))
 267     (setq deflist (cdr deflist))))
 268
 269 ;; Ethiopic character set
 270
 271 (modify-category-entry (make-char 'ethiopic) ?e)
 272 (modify-syntax-entry (make-char 'ethiopic) "w")
 273 (let ((chars '(?\e$(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)))
 274   (while chars
 275     (modify-syntax-entry (car chars) ".")
 276     (setq chars (cdr chars))))
 277
 278 ;; European character set (Latin-1,2,3,4,5)
 279
 280 (modify-category-entry (make-char 'latin-iso8859-1) ?l)
 281 (modify-category-entry (make-char 'latin-iso8859-2) ?l)
 282 (modify-category-entry (make-char 'latin-iso8859-3) ?l)
 283 (modify-category-entry (make-char 'latin-iso8859-4) ?l)
 284 (modify-category-entry (make-char 'latin-iso8859-9) ?l)
 285
 286 (modify-category-entry (make-char 'latin-iso8859-1 160) ?\ )
 287 (modify-category-entry (make-char 'latin-iso8859-2 160) ?\ )
 288 (modify-category-entry (make-char 'latin-iso8859-3 160) ?\ )
 289 (modify-category-entry (make-char 'latin-iso8859-4 160) ?\ )
 290 (modify-category-entry (make-char 'latin-iso8859-9 160) ?\ )
 291
 292 ;; Greek character set (ISO-8859-7)
 293
 294 (modify-category-entry (make-char 'greek-iso8859-7) ?g)
 295
 296 (let ((c 182))
 297   (while (< c 255)
 298     (modify-syntax-entry (make-char 'greek-iso8859-7 c) "w")
 299     (setq c (1+ c))))
 300 (modify-syntax-entry (make-char 'greek-iso8859-7 160) "w") ; NBSP
 301 (modify-syntax-entry ?\e,F7\e(B ".")
 302 (modify-syntax-entry ?\e,F;\e(B ".")
 303 (modify-syntax-entry ?\e,F=\e(B ".")
 304
 305 ;; Hebrew character set (ISO-8859-8)
 306
 307 (modify-category-entry (make-char 'hebrew-iso8859-8) ?w)
 308
 309 (let ((c 224))
 310   (while (< c 251)
 311     (modify-syntax-entry (make-char 'hebrew-iso8859-8 c) "w")
 312     (setq c (1+ c))))
 313 (modify-syntax-entry (make-char 'hebrew-iso8859-8 160) "w") ; NBSP
 314
 315 ;; Indian character set (IS 13194 and other Emacs original Indian charsets)
 316
 317 (modify-category-entry (make-char 'indian-is13194) ?i)
 318 (modify-category-entry (make-char 'indian-2-column) ?I)
 319 (modify-category-entry (make-char 'indian-1-column) ?I)
 320
 321 ;; Japanese character set (JISX0201-kana, JISX0201-roman, JISX0208, JISX0212)
 322
 323 (modify-category-entry (make-char 'katakana-jisx0201) ?k)
 324 (modify-category-entry (make-char 'latin-jisx0201) ?r)
 325 (modify-category-entry (make-char 'japanese-jisx0208) ?j)
 326 (modify-category-entry (make-char 'japanese-jisx0212) ?j)
 327 (modify-category-entry (make-char 'katakana-jisx0201) ?\|)
 328 (modify-category-entry (make-char 'japanese-jisx0208) ?\|)
 329 (modify-category-entry (make-char 'japanese-jisx0212) ?\|)
 330
 331 ;; JISX0208
 332 (modify-syntax-entry (make-char 'japanese-jisx0208) "w")
 333 (modify-syntax-entry (make-char 'japanese-jisx0208 33) "_")
 334 (modify-syntax-entry (make-char 'japanese-jisx0208 34) "_")
 335 (modify-syntax-entry (make-char 'japanese-jisx0208 40) "_")
 336 (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$B!;\e(B)))
 337   (while chars
 338     (modify-syntax-entry (car chars) "w")
 339     (setq chars (cdr chars))))
 340 (modify-syntax-entry ?\\e$B!J\e(B "(\e$B!K\e(B")
 341 (modify-syntax-entry ?\\e$B!N\e(B "(\e$B!O\e(B")
 342 (modify-syntax-entry ?\\e$B!P\e(B "(\e$B!Q\e(B")
 343 (modify-syntax-entry ?\\e$B!V\e(B "(\e$B!W\e(B")
 344 (modify-syntax-entry ?\\e$B!X\e(B "(\e$B!Y\e(B")
 345 (modify-syntax-entry ?\\e$B!K\e(B ")\e$B!J\e(B")
 346 (modify-syntax-entry ?\\e$B!O\e(B ")\e$B!N\e(B")
 347 (modify-syntax-entry ?\\e$B!Q\e(B ")\e$B!P\e(B")
 348 (modify-syntax-entry ?\\e$B!W\e(B ")\e$B!V\e(B")
 349 (modify-syntax-entry ?\\e$B!Y\e(B ")\e$B!X\e(B")
 350
 351 (modify-category-entry (make-char 'japanese-jisx0208 35) ?A)
 352 (modify-category-entry (make-char 'japanese-jisx0208 36) ?H)
 353 (modify-category-entry (make-char 'japanese-jisx0208 37) ?K)
 354 (modify-category-entry (make-char 'japanese-jisx0208 38) ?G)
 355 (modify-category-entry (make-char 'japanese-jisx0208 39) ?Y)
 356 (let ((row 48))
 357   (while (< row 127)
 358     (modify-category-entry (make-char 'japanese-jisx0208 row) ?C)
 359     (setq row (1+ row))))
 360 (modify-category-entry ?\e$B!<\e(B ?K)
 361 (let ((chars '(?\e$B!+\e(B ?\e$B!,\e(B)))
 362   (while chars
 363     (modify-category-entry (car chars) ?K)
 364     (modify-category-entry (car chars) ?H)
 365     (setq chars (cdr chars))))
 366 (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$B!;\e(B)))
 367   (while chars
 368     (modify-category-entry (car chars) ?C)
 369     (setq chars (cdr chars))))
 370
 371 ;; JISX0212
 372 (modify-syntax-entry (make-char 'japanese-jisx0212) "w")
 373 (modify-syntax-entry (make-char 'japanese-jisx0212 33) "_")
 374 (modify-syntax-entry (make-char 'japanese-jisx0212 34) "_")
 375 (modify-syntax-entry (make-char 'japanese-jisx0212 35) "_")
 376
 377 (modify-category-entry (make-char 'japanese-jisx0212 ) ?C)
 378
 379 ;; JISX0201-Kana
 380 (modify-syntax-entry (make-char 'katakana-jisx0201) "w")
 381 (let ((chars '(?\e(I!\e(B ?\e(I"\e(B ?\e(I#\e(B ?\e(I$\e(B ?\e(I%\e(B)))
 382   (while chars
 383     (modify-syntax-entry (car chars) ".")
 384     (setq chars (cdr chars))))
 385
 386 ;; Korean character set (KSC5601)
 387
 388 (modify-syntax-entry (make-char 'korean-ksc5601) "w")
 389 (modify-syntax-entry (make-char 'korean-ksc5601 33) "_")
 390 (modify-syntax-entry (make-char 'korean-ksc5601 34) "_")
 391 (modify-syntax-entry (make-char 'korean-ksc5601 38) "_")
 392 (modify-syntax-entry (make-char 'korean-ksc5601 39) "_")
 393 (modify-syntax-entry (make-char 'korean-ksc5601 40) "_")
 394 (modify-syntax-entry (make-char 'korean-ksc5601 41) "_")
 395
 396 (modify-category-entry (make-char 'korean-ksc5601) ?h)
 397 (modify-category-entry (make-char 'korean-ksc5601 35) ?A)
 398 (modify-category-entry (make-char 'korean-ksc5601 37) ?G)
 399 (modify-category-entry (make-char 'korean-ksc5601 42) ?H)
 400 (modify-category-entry (make-char 'korean-ksc5601 43) ?K)
 401 (modify-category-entry (make-char 'korean-ksc5601 44) ?Y)
 402
 403 ;; Lao character set
 404
 405 (modify-category-entry (make-char 'lao) ?o)
 406
 407 (let ((deflist  '(;; chars      syntax  category
 408                   ("\e(1!\e(B-\e(1N\e(B"    "w"     ?0) ; consonant
 409                   ("\e(1PRS]`\e(B-\e(1d\e(B"        "w"     ?1) ; vowel base
 410                   ("\e(1QT\e(B-\e(1W[m\e(B" "w"     ?2) ; vowel upper
 411                   ("\e(1XY\e(B"           "w"     ?3) ; vowel lower
 412                   ("\e(1h\e(B-\e(1l\e(B"    "w"     ?4) ; tone mark
 413                   ("\e(1\\e(B"            "w"     ?9) ; semivowel lower
 414                   ("\e(1p\e(B-\e(1y\e(B"    "w"     ?6) ; digit
 415                   ("\e(1Of\e(B"           "_"     ?0) ; symbol
 416                   ))
 417       elm chars len syntax category to ch i)
 418   (while deflist
 419     (setq elm (car deflist))
 420     (setq chars (car elm)
 421           len (length chars)
 422           syntax (nth 1 elm)
 423           category (nth 2 elm)
 424           i 0)
 425     (while (< i len)
 426       (if (= (aref chars i) ?-)
 427           (setq i (1+ i)
 428                 to (sref chars i))
 429         (setq ch (sref chars i)
 430               to ch))
 431       (while (<= ch to)
 432         (modify-syntax-entry ch syntax)
 433         (modify-category-entry ch category)
 434         (setq ch (1+ ch)))
 435       (setq i (+ i (char-bytes to))))
 436     (setq deflist (cdr deflist))))
 437
 438 ;; Thai character set (TIS620)
 439
 440 (modify-category-entry (make-char 'thai-tis620) ?t)
 441
 442 (let ((deflist  '(;; chars      syntax  category
 443                   ("\e,T!\e(B-\e,TCEG\e(B-\e,TN\e(B"  "w"     ?0) ; consonant
 444                   ("\e,TDFPRS`\e(B-\e,Te\e(B"       "w"     ?1) ; vowel base
 445                   ("\e,TQT\e(B-\e,TWgn\e(B" "w"     ?2) ; vowel upper
 446                   ("\e,TX\e(B-\e,TZ\e(B"    "w"     ?3) ; vowel lower
 447                   ("\e,Th\e(B-\e,Tm\e(B"    "w"     ?4) ; tone mark
 448                   ("\e,Tp\e(B-\e,Ty\e(B"    "w"     ?6) ; digit
 449                   ("\e,TOf_oz{\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 (sref chars i))
 463         (setq ch (sref 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 (+ i (char-bytes to))))
 470     (setq deflist (cdr deflist))))
 471
 472 ;; Tibetan character set
 473
 474 (let ((row 33))
 475   (while (< row 38)
 476     (modify-category-entry (make-char 'tibetan row) ?q)
 477     (setq row (1+ row))))
 478
 479 (modify-category-entry (make-char 'tibetan-1-column 33) ?q)
 480
 481 (let ((deflist  '(;; chars             syntax category
 482                   ("\e$(7"!\e(B-\e$(7"J\e(B"                "w"     ?0) ; consonant
 483                   ("\e$(7#!\e(B-\e$(7#J#P#Q\e(B"          "w"     ?0) ;
 484                   ("\e$(7$!\e(B-\e$(7$e\e(B"              "w"     ?0) ;
 485                   ("\e$(7%!\e(B-\e$(7%u\e(B"              "w"     ?0) ;
 486                   ("\e$(7"S"["\"]"^"a\e(B"       "w"      ?2) ; upper vowel
 487                   ("\e$(7"_"c"d"g"h"i"j"k"l\e(B" "w"      ?2) ; upper modifier
 488                   ("\e$(7!I"Q"U"e!e!g\e(B"       "w"      ?3) ; lowel vowel/modifier
 489                   ("\e$(7!P\e(B-\e$(7!Y!Z\e(B-\e$(7!c\e(B"            "w"     ?6) ; digit
 490                   ("\e$(7!;!=\e(B-\e$(7!B!D"`\e(B"        "."     ?|) ; line-break char
 491                   ("\e$(8!;!=!?!@!A!D"`\e(B"            "."     ?|) ;
 492                   ("\e$(7!8!;!=\e(B-\e$(7!B!D"`!m!d\e(B"  "."     ?>) ; prohibition
 493                   ("\e$(8!;!=!?!@!A!D"`\e(B"            "."     ?>) ;
 494                   ("\e$(7!0\e(B-\e$(7!:!l#R#S"f\e(B"      "."     ?<) ; prohibition
 495                   ("\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
 496                   ))
 497       elm chars len syntax category to ch i)
 498   (while deflist
 499     (setq elm (car deflist))
 500     (setq chars (car elm)
 501           len (length chars)
 502           syntax (nth 1 elm)
 503           category (nth 2 elm)
 504           i 0)
 505     (while (< i len)
 506       (if (= (aref chars i) ?-)
 507           (setq i (1+ i)
 508                 to (sref chars i))
 509         (setq ch (sref chars i)
 510               to ch))
 511       (while (<= ch to)
 512         (modify-syntax-entry ch syntax)
 513         (modify-category-entry ch category)
 514         (setq ch (1+ ch)))
 515       (setq i (+ i (char-bytes to))))
 516     (setq deflist (cdr deflist))))
 517
 518 ;; Vietnamese character set
 519
 520 (let ((lower (make-char 'vietnamese-viscii-lower))
 521       (upper (make-char 'vietnamese-viscii-upper)))
 522   (modify-syntax-entry lower "w")
 523   (modify-syntax-entry upper "w")
 524   (modify-category-entry lower ?v)
 525   (modify-category-entry upper ?v)
 526   (modify-category-entry lower ?l)      ; To make a word with
 527   (modify-category-entry upper ?l)      ; latin characters.
 528   )
 529
 530 \f
 531 ;;; Setting word boundary.
 532
 533 (setq word-combining-categories
 534       '((?l . ?l)))
 535
 536 (setq word-separating-categories        ;  (2-byte character sets)
 537       '((?A . ?K)                       ; Alpha numeric - Katakana
 538         (?A . ?C)                       ; Alpha numeric - Chinese
 539         (?H . ?A)                       ; Hiragana - Alpha numeric
 540         (?H . ?K)                       ; Hiragana - Katakana
 541         (?H . ?C)                       ; Hiragana - Chinese
 542         (?K . ?A)                       ; Katakana - Alpha numeric
 543         (?K . ?C)                       ; Katakana - Chinese
 544         (?C . ?A)                       ; Chinese - Alpha numeric
 545         (?C . ?K)                       ; Chinese - Katakana
 546         ))
 547
 548 \f
 549 ;; For each character set, put the information of the most proper
 550 ;; coding system to encode it by `prefered-coding-system' property.
 551
 552 (let ((l '((latin-iso8859-1     . iso-latin-1)
 553            (latin-iso8859-2     . iso-latin-2)
 554            (latin-iso8859-3     . iso-latin-3)
 555            (latin-iso8859-4     . iso-latin-4)
 556            (thai-tis620         . thai-tis620)
 557            (greek-iso8859-7     . greek-iso-8bit)
 558            (arabic-iso8859-6    . iso-2022-7bit)
 559            (hebrew-iso8859-8    . hebrew-iso-8bit)
 560            (katakana-jisx0201   . japanese-shift-jis)
 561            (latin-jisx0201      . japanese-shift-jis)
 562            (cyrillic-iso8859-5  . cyrillic-iso-8bit)
 563            (latin-iso8859-9     . iso-latin-5)
 564            (japanese-jisx0208-1978 . iso-2022-jp)
 565            (chinese-gb2312      . cn-gb-2312)
 566            (japanese-jisx0208   . iso-2022-jp)
 567            (korean-ksc5601      . iso-2022-kr)
 568            (japanese-jisx0212   . iso-2022-jp)
 569            (chinese-cns11643-1  . iso-2022-cn)
 570            (chinese-cns11643-2  . iso-2022-cn)
 571            (chinese-big5-1      . chinese-big5)
 572            (chinese-big5-2      . chinese-big5)
 573            (chinese-sisheng     . iso-2022-7bit)
 574            (ipa                 . iso-2022-7bit)
 575            (vietnamese-viscii-lower . vietnamese-viscii)
 576            (vietnamese-viscii-upper . vietnamese-viscii)
 577            (arabic-digit        . iso-2022-7bit)
 578            (arabic-1-column     . iso-2022-7bit)
 579            (ascii-right-to-left . iso-2022-7bit)
 580            (lao                 . lao)
 581            (arabic-2-column     . iso-2022-7bit)
 582            (indian-is13194      . devanagari)
 583            (indian-1-column     . devanagari)
 584            (tibetan-1-column    . tibetan)
 585            (ethiopic            . iso-2022-jp)
 586            (chinese-cns11643-3  . iso-2022-cn)
 587            (chinese-cns11643-4  . iso-2022-cn)
 588            (chinese-cns11643-5  . iso-2022-cn)
 589            (chinese-cns11643-6  . iso-2022-cn)
 590            (chinese-cns11643-7  . iso-2022-cn)
 591            (indian-2-column     . devanagari)
 592            (tibetan             . tibetan))))
 593   (while l
 594     (put-charset-property (car (car l)) 'prefered-coding-system (cdr (car l)))
 595     (setq l (cdr l))))
 596
 597
 598 ;;; Local Variables:
 599 ;;; coding: iso-2022-7bit
 600 ;;; End:
 601
 602 ;;; end of characters.el