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 ?\  "This character counts as a space for indentation purposes.")
  84
  85 ;; Keep the followings for `kinsoku' processing.  See comments in
  86 ;; kinsoku.el.
  87 (define-category ?> "A character which can't be placed at beginning of line.")
  88 (define-category ?< "A character which can't be placed at end of line.")
  89
  90 \f
  91 ;;; Setting syntax and category.
  92
  93 ;; ASCII
  94
  95 (let ((ch 32))
  96   (while (< ch 127)                     ; All ASCII characters have
  97     (modify-category-entry ch ?a)       ; the category `a' (ASCII)
  98     (modify-category-entry ch ?l)       ; and `l' (Latin).
  99     (setq ch (1+ ch))))
 100
 101 ;; Arabic character set
 102
 103 (let ((charsets '(arabic-iso8859-6
 104                   arabic-digit
 105                   arabic-1-column
 106                   arabic-2-column)))
 107   (while charsets
 108     (modify-syntax-entry (make-char (car charsets)) "w")
 109     (modify-category-entry (make-char (car charsets)) ?b)
 110     (setq charsets (cdr charsets))))
 111
 112 ;; Chinese character set (GB2312)
 113
 114 (modify-syntax-entry (make-char 'chinese-gb2312) "w")
 115 (modify-syntax-entry (make-char 'chinese-gb2312 33) "_")
 116 (modify-syntax-entry (make-char 'chinese-gb2312 34) "_")
 117 (modify-syntax-entry (make-char 'chinese-gb2312 41) "_")
 118 (modify-syntax-entry ?\\e$A!2\e(B "(\e$A!3\e(B")
 119 (modify-syntax-entry ?\\e$A!4\e(B "(\e$A!5\e(B")
 120 (modify-syntax-entry ?\\e$A!6\e(B "(\e$A!7\e(B")
 121 (modify-syntax-entry ?\\e$A!8\e(B "(\e$A!9\e(B")
 122 (modify-syntax-entry ?\\e$A!:\e(B "(\e$A!;\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$A!3\e(B ")\e$A!2\e(B")
 126 (modify-syntax-entry ?\\e$A!5\e(B ")\e$A!4\e(B")
 127 (modify-syntax-entry ?\\e$A!7\e(B ")\e$A!6\e(B")
 128 (modify-syntax-entry ?\\e$A!9\e(B ")\e$A!8\e(B")
 129 (modify-syntax-entry ?\\e$A!;\e(B ")\e$A!:\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
 133 (modify-category-entry (make-char 'chinese-gb2312) ?c)
 134 (modify-category-entry (make-char 'chinese-gb2312) ?\|)
 135 (modify-category-entry (make-char 'chinese-gb2312 35) ?A)
 136 (modify-category-entry (make-char 'chinese-gb2312 36) ?H)
 137 (modify-category-entry (make-char 'chinese-gb2312 37) ?K)
 138 (modify-category-entry (make-char 'chinese-gb2312 38) ?G)
 139 (modify-category-entry (make-char 'chinese-gb2312 39) ?Y)
 140 (modify-category-entry (make-char 'chinese-gb2312 35) ?A)
 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\7f\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$(5!z\e(B"        "w"     ?6) ; digit
 246                   ))
 247       elm chars len syntax category to ch i)
 248   (while deflist
 249     (setq elm (car deflist))
 250     (setq chars (car elm)
 251           len (length chars)
 252           syntax (nth 1 elm)
 253           category (nth 2 elm)
 254           i 0)
 255     (while (< i len)
 256       (if (= (aref chars i) ?-)
 257           (setq i (1+ i)
 258                 to (sref chars i))
 259         (setq ch (sref chars i)
 260               to ch))
 261       (while (<= ch to)
 262         (modify-syntax-entry ch syntax)
 263         (modify-category-entry ch category)
 264         (setq ch (1+ ch)))
 265       (setq i (+ i (char-bytes to))))
 266     (setq deflist (cdr deflist))))
 267
 268 ;; Ethiopic character set
 269
 270 (modify-category-entry (make-char 'ethiopic) ?e)
 271 (modify-syntax-entry (make-char 'ethiopic) "w")
 272 (let ((chars '(?\e$(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
 285 (modify-category-entry (make-char 'latin-iso8859-1 160) ?\ )
 286 (modify-category-entry (make-char 'latin-iso8859-2 160) ?\ )
 287 (modify-category-entry (make-char 'latin-iso8859-3 160) ?\ )
 288 (modify-category-entry (make-char 'latin-iso8859-4 160) ?\ )
 289 (modify-category-entry (make-char 'latin-iso8859-9 160) ?\ )
 290
 291 ;; Greek character set (ISO-8859-7)
 292
 293 (modify-category-entry (make-char 'greek-iso8859-7) ?g)
 294
 295 (let ((c 182))
 296   (while (< c 255)
 297     (modify-syntax-entry (make-char 'greek-iso8859-7 c) "w")
 298     (setq c (1+ c))))
 299 (modify-syntax-entry (make-char 'greek-iso8859-7 160) "w") ; NBSP
 300 (modify-syntax-entry ?\e,F7\e(B ".")
 301 (modify-syntax-entry ?\e,F;\e(B ".")
 302 (modify-syntax-entry ?\e,F=\e(B ".")
 303
 304 ;; Hebrew character set (ISO-8859-8)
 305
 306 (modify-category-entry (make-char 'hebrew-iso8859-8) ?w)
 307
 308 (let ((c 224))
 309   (while (< c 251)
 310     (modify-syntax-entry (make-char 'hebrew-iso8859-8 c) "w")
 311     (setq c (1+ c))))
 312 (modify-syntax-entry (make-char 'hebrew-iso8859-8 160) "w") ; NBSP
 313
 314 ;; Indian character set (IS 13194 and other Emacs original Indian charsets)
 315
 316 (modify-category-entry (make-char 'indian-is13194) ?i)
 317 (modify-category-entry (make-char 'indian-2-column) ?I)
 318 (modify-category-entry (make-char 'indian-1-column) ?I)
 319
 320 ;; Japanese character set (JISX0201-kana, JISX0201-roman, JISX0208, JISX0212)
 321
 322 (modify-category-entry (make-char 'katakana-jisx0201) ?k)
 323 (modify-category-entry (make-char 'latin-jisx0201) ?r)
 324 (modify-category-entry (make-char 'japanese-jisx0208) ?j)
 325 (modify-category-entry (make-char 'japanese-jisx0212) ?j)
 326 (modify-category-entry (make-char 'katakana-jisx0201) ?\|)
 327 (modify-category-entry (make-char 'japanese-jisx0208) ?\|)
 328 (modify-category-entry (make-char 'japanese-jisx0212) ?\|)
 329
 330 ;; JISX0208
 331 (modify-syntax-entry (make-char 'japanese-jisx0208) "w")
 332 (modify-syntax-entry (make-char 'japanese-jisx0208 33) "_")
 333 (modify-syntax-entry (make-char 'japanese-jisx0208 34) "_")
 334 (modify-syntax-entry (make-char 'japanese-jisx0208 40) "_")
 335 (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$B!;\e(B)))
 336   (while chars
 337     (modify-syntax-entry (car chars) "w")
 338     (setq chars (cdr chars))))
 339 (modify-syntax-entry ?\\e$B!J\e(B "(\e$B!K\e(B")
 340 (modify-syntax-entry ?\\e$B!N\e(B "(\e$B!O\e(B")
 341 (modify-syntax-entry ?\\e$B!P\e(B "(\e$B!Q\e(B")
 342 (modify-syntax-entry ?\\e$B!V\e(B "(\e$B!W\e(B")
 343 (modify-syntax-entry ?\\e$B!X\e(B "(\e$B!Y\e(B")
 344 (modify-syntax-entry ?\\e$B!K\e(B ")\e$B!J\e(B")
 345 (modify-syntax-entry ?\\e$B!O\e(B ")\e$B!N\e(B")
 346 (modify-syntax-entry ?\\e$B!Q\e(B ")\e$B!P\e(B")
 347 (modify-syntax-entry ?\\e$B!W\e(B ")\e$B!V\e(B")
 348 (modify-syntax-entry ?\\e$B!Y\e(B ")\e$B!X\e(B")
 349
 350 (modify-category-entry (make-char 'japanese-jisx0208 35) ?A)
 351 (modify-category-entry (make-char 'japanese-jisx0208 36) ?H)
 352 (modify-category-entry (make-char 'japanese-jisx0208 37) ?K)
 353 (modify-category-entry (make-char 'japanese-jisx0208 38) ?G)
 354 (modify-category-entry (make-char 'japanese-jisx0208 39) ?Y)
 355 (let ((row 48))
 356   (while (< row 127)
 357     (modify-category-entry (make-char 'japanese-jisx0208 row) ?C)
 358     (setq row (1+ row))))
 359 (modify-category-entry ?\e$B!<\e(B ?K)
 360 (let ((chars '(?\e$B!+\e(B ?\e$B!,\e(B)))
 361   (while chars
 362     (modify-category-entry (car chars) ?K)
 363     (modify-category-entry (car chars) ?H)
 364     (setq chars (cdr chars))))
 365 (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$B!;\e(B)))
 366   (while chars
 367     (modify-category-entry (car chars) ?C)
 368     (setq chars (cdr chars))))
 369
 370 ;; JISX0212
 371 (modify-syntax-entry (make-char 'japanese-jisx0212) "w")
 372 (modify-syntax-entry (make-char 'japanese-jisx0212 33) "_")
 373 (modify-syntax-entry (make-char 'japanese-jisx0212 34) "_")
 374 (modify-syntax-entry (make-char 'japanese-jisx0212 35) "_")
 375
 376 (modify-category-entry (make-char 'japanese-jisx0212 ) ?C)
 377
 378 ;; JISX0201-Kana
 379 (modify-syntax-entry (make-char 'katakana-jisx0201) "w")
 380 (let ((chars '(?\e(I!\e(B ?\e(I"\e(B ?\e(I#\e(B ?\e(I$\e(B ?\e(I%\e(B)))
 381   (while chars
 382     (modify-syntax-entry (car chars) ".")
 383     (setq chars (cdr chars))))
 384
 385 ;; Korean character set (KSC5601)
 386
 387 (modify-syntax-entry (make-char 'korean-ksc5601) "w")
 388 (modify-syntax-entry (make-char 'korean-ksc5601 33) "_")
 389 (modify-syntax-entry (make-char 'korean-ksc5601 34) "_")
 390 (modify-syntax-entry (make-char 'korean-ksc5601 38) "_")
 391 (modify-syntax-entry (make-char 'korean-ksc5601 39) "_")
 392 (modify-syntax-entry (make-char 'korean-ksc5601 40) "_")
 393 (modify-syntax-entry (make-char 'korean-ksc5601 41) "_")
 394
 395 (modify-category-entry (make-char 'korean-ksc5601) ?h)
 396 (modify-category-entry (make-char 'korean-ksc5601 35) ?A)
 397 (modify-category-entry (make-char 'korean-ksc5601 37) ?G)
 398 (modify-category-entry (make-char 'korean-ksc5601 42) ?H)
 399 (modify-category-entry (make-char 'korean-ksc5601 43) ?K)
 400 (modify-category-entry (make-char 'korean-ksc5601 44) ?Y)
 401
 402 ;; Lao character set
 403
 404 (modify-category-entry (make-char 'lao) ?o)
 405
 406 (let ((deflist  '(;; chars      syntax  category
 407                   ("\e(1!\e(B-\e(1N\e(B"    "w"     ?0) ; consonant
 408                   ("\e(1PRS]`\e(B-\e(1d\e(B"        "w"     ?1) ; vowel base
 409                   ("\e(1QT\e(B-\e(1W[m\e(B" "w"     ?2) ; vowel upper
 410                   ("\e(1XY\e(B"           "w"     ?3) ; vowel lower
 411                   ("\e(1h\e(B-\e(1l\e(B"    "w"     ?4) ; tone mark
 412                   ("\e(1\\e(B"            "w"     ?9) ; semivowel lower
 413                   ("\e(1p\e(B-\e(1y\e(B"    "w"     ?6) ; digit
 414                   ("\e(1Of\e(B"           "_"     ?0) ; symbol
 415                   ))
 416       elm chars len syntax category to ch i)
 417   (while deflist
 418     (setq elm (car deflist))
 419     (setq chars (car elm)
 420           len (length chars)
 421           syntax (nth 1 elm)
 422           category (nth 2 elm)
 423           i 0)
 424     (while (< i len)
 425       (if (= (aref chars i) ?-)
 426           (setq i (1+ i)
 427                 to (sref chars i))
 428         (setq ch (sref chars i)
 429               to ch))
 430       (while (<= ch to)
 431         (modify-syntax-entry ch syntax)
 432         (modify-category-entry ch category)
 433         (setq ch (1+ ch)))
 434       (setq i (+ i (char-bytes to))))
 435     (setq deflist (cdr deflist))))
 436
 437 ;; Thai character set (TIS620)
 438
 439 (modify-category-entry (make-char 'thai-tis620) ?t)
 440
 441 (let ((deflist  '(;; chars      syntax  category
 442                   ("\e,T!\e(B-\e,TCEG\e(B-\e,TN\e(B"  "w"     ?0) ; consonant
 443                   ("\e,TDFPRS`\e(B-\e,Te\e(B"       "w"     ?1) ; vowel base
 444                   ("\e,TQT\e(B-\e,TWgn\e(B" "w"     ?2) ; vowel upper
 445                   ("\e,TX\e(B-\e,TZ\e(B"    "w"     ?3) ; vowel lower
 446                   ("\e,Th\e(B-\e,Tm\e(B"    "w"     ?4) ; tone mark
 447                   ("\e,Tp\e(B-\e,Ty\e(B"    "w"     ?6) ; digit
 448                   ("\e,TOf_oz{\e(B"       "_"     ?5) ; symbol
 449                   ))
 450       elm chars len syntax category to ch i)
 451   (while deflist
 452     (setq elm (car deflist))
 453     (setq chars (car elm)
 454           len (length chars)
 455           syntax (nth 1 elm)
 456           category (nth 2 elm)
 457           i 0)
 458     (while (< i len)
 459       (if (= (aref chars i) ?-)
 460           (setq i (1+ i)
 461                 to (sref chars i))
 462         (setq ch (sref chars i)
 463               to ch))
 464       (while (<= ch to)
 465         (modify-syntax-entry ch syntax)
 466         (modify-category-entry ch category)
 467         (setq ch (1+ ch)))
 468       (setq i (+ i (char-bytes to))))
 469     (setq deflist (cdr deflist))))
 470
 471 ;; Tibetan character set
 472
 473 (let ((row 33))
 474   (while (< row 38)
 475     (modify-category-entry (make-char 'tibetan row) ?q)
 476     (setq row (1+ row))))
 477
 478 (modify-category-entry (make-char 'tibetan-1-column 33) ?q)
 479
 480 (let ((deflist  '(;; chars             syntax category
 481                   ("\e$(7"!\e(B-\e$(7"J\e(B"                "w"     ?0) ; consonant
 482                   ("\e$(7#!\e(B-\e$(7#J#P#Q\e(B"          "w"     ?0) ;
 483                   ("\e$(7$!\e(B-\e$(7$e\e(B"              "w"     ?0) ;
 484                   ("\e$(7%!\e(B-\e$(7%u\e(B"              "w"     ?0) ;
 485                   ("\e$(7"S"["\"]"^"a\e(B"       "w"      ?2) ; upper vowel
 486                   ("\e$(7"_"c"d"g"h"i"j"k"l\e(B" "w"      ?2) ; upper modifier
 487                   ("\e$(7!I"Q"U"e!e!g\e(B"       "w"      ?3) ; lowel vowel/modifier
 488                   ("\e$(7!P\e(B-\e$(7!Y!Z\e(B-\e$(7!c\e(B"            "w"     ?6) ; digit
 489                   ("\e$(7!;!=\e(B-\e$(7!B!D"`\e(B"        "."     ?|) ; line-break char
 490                   ("\e$(8!;!=!?!@!A!D"`\e(B"            "."     ?|) ;
 491                   ("\e$(7!8!;!=\e(B-\e$(7!B!D"`!m!d\e(B"  "."     ?>) ; prohibition
 492                   ("\e$(8!;!=!?!@!A!D"`\e(B"            "."     ?>) ;
 493                   ("\e$(7!0\e(B-\e$(7!:!l#R#S"f\e(B"      "."     ?<) ; prohibition
 494                   ("\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
 495                   ))
 496       elm chars len syntax category to ch i)
 497   (while deflist
 498     (setq elm (car deflist))
 499     (setq chars (car elm)
 500           len (length chars)
 501           syntax (nth 1 elm)
 502           category (nth 2 elm)
 503           i 0)
 504     (while (< i len)
 505       (if (= (aref chars i) ?-)
 506           (setq i (1+ i)
 507                 to (sref chars i))
 508         (setq ch (sref chars i)
 509               to ch))
 510       (while (<= ch to)
 511         (modify-syntax-entry ch syntax)
 512         (modify-category-entry ch category)
 513         (setq ch (1+ ch)))
 514       (setq i (+ i (char-bytes to))))
 515     (setq deflist (cdr deflist))))
 516
 517 ;; Vietnamese character set
 518
 519 (let ((lower (make-char 'vietnamese-viscii-lower))
 520       (upper (make-char 'vietnamese-viscii-upper)))
 521   (modify-syntax-entry lower "w")
 522   (modify-syntax-entry upper "w")
 523   (modify-category-entry lower ?v)
 524   (modify-category-entry upper ?v)
 525   (modify-category-entry lower ?l)      ; To make a word with
 526   (modify-category-entry upper ?l)      ; latin characters.
 527   )
 528
 529 \f
 530 ;;; Setting word boundary.
 531
 532 (setq word-combining-categories
 533       '((?l . ?l)))
 534
 535 (setq word-separating-categories        ;  (2-byte character sets)
 536       '((?A . ?K)                       ; Alpha numeric - Katakana
 537         (?A . ?C)                       ; Alpha numeric - Chinese
 538         (?H . ?A)                       ; Hiragana - Alpha numeric
 539         (?H . ?K)                       ; Hiragana - Katakana
 540         (?H . ?C)                       ; Hiragana - Chinese
 541         (?K . ?A)                       ; Katakana - Alpha numeric
 542         (?K . ?C)                       ; Katakana - Chinese
 543         (?C . ?A)                       ; Chinese - Alpha numeric
 544         (?C . ?K)                       ; Chinese - Katakana
 545         ))