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 ;; Copyright (C) 2001, 2002 Free Software Foundation, Inc.
   6
   7 ;; Keywords: multibyte character, character set, syntax, category
   8
   9 ;; This file is part of GNU Emacs.
  10
  11 ;; GNU Emacs is free software; you can redistribute it and/or modify
  12 ;; it 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 ;; GNU Emacs is distributed in the hope that it will be useful,
  17 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  18 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  19 ;; GNU General Public License for more details.
  20
  21 ;; You should have received a copy of the GNU General Public License
  22 ;; along with GNU Emacs; see the file COPYING.  If not, write to the
  23 ;; Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  24 ;; Boston, MA 02111-1307, USA.
  25
  26 ;;; Commentary:
  27
  28 ;; This file contains multibyte characters.  Save this file always in
  29 ;; the coding system `iso-2022-7bit'.
  30
  31 ;; This file does not define the syntax for Latin-N character sets;
  32 ;; those are defined by the files latin-N.el.
  33
  34 ;;; Code:
  35
  36 ;;; Predefined categories.
  37
  38 ;; For each character set.
  39
  40 (define-category ?a "ASCII")
  41 (define-category ?l "Latin")
  42 (define-category ?t "Thai")
  43 (define-category ?g "Greek")
  44 (define-category ?b "Arabic")
  45 (define-category ?w "Hebrew")
  46 (define-category ?y "Cyrillic")
  47 (define-category ?k "Japanese katakana")
  48 (define-category ?r "Japanese roman")
  49 (define-category ?c "Chinese")
  50 (define-category ?j "Japanese")
  51 (define-category ?h "Korean")
  52 (define-category ?e "Ethiopic (Ge'ez)")
  53 (define-category ?v "Vietnamese")
  54 (define-category ?i "Indian")
  55 (define-category ?o "Lao")
  56 (define-category ?q "Tibetan")
  57
  58 ;; For each group (row) of 2-byte character sets.
  59
  60 (define-category ?A "Alpha-numeric characters of 2-byte character sets")
  61 (define-category ?C "Chinese (Han) characters of 2-byte character sets")
  62 (define-category ?G "Greek characters of 2-byte character sets")
  63 (define-category ?H "Japanese Hiragana characters of 2-byte character sets")
  64 (define-category ?K "Japanese Katakana characters of 2-byte character sets")
  65 (define-category ?N "Korean Hangul characters of 2-byte character sets")
  66 (define-category ?Y "Cyrillic characters of 2-byte character sets")
  67 (define-category ?I "Indian Glyphs")
  68
  69 ;; For phonetic classifications.
  70
  71 (define-category ?0 "consonant")
  72 (define-category ?1 "base (independent) vowel")
  73 (define-category ?2 "upper diacritical mark (including upper vowel)")
  74 (define-category ?3 "lower diacritical mark (including lower vowel)")
  75 (define-category ?4 "tone mark")
  76 (define-category ?5 "symbol")
  77 (define-category ?6 "digit")
  78 (define-category ?7 "vowel-modifying diacritical mark")
  79 (define-category ?8 "vowel-signs")
  80 (define-category ?9 "semivowel lower")
  81
  82 ;; For filling.
  83 (define-category ?| "While filling, we can break a line at this character.")
  84
  85 ;; For indentation calculation.
  86 (define-category ?
  87   "This character counts as a space for indentation purposes.")
  88
  89 ;; Keep the following for `kinsoku' processing.  See comments in
  90 ;; kinsoku.el.
  91 (define-category ?> "A character which can't be placed at beginning of line.")
  92 (define-category ?< "A character which can't be placed at end of line.")
  93
  94 ;; Combining
  95 (define-category ?^ "Combining diacritic or mark")
  96 \f
  97 ;;; Setting syntax and category.
  98
  99 ;; ASCII
 100
 101 (let ((ch 32))
 102   (while (< ch 127)                     ; All ASCII characters have
 103     (modify-category-entry ch ?a)       ; the category `a' (ASCII)
 104     (modify-category-entry ch ?l)       ; and `l' (Latin).
 105     (setq ch (1+ ch))))
 106
 107 ;; Arabic character set
 108
 109 (let ((charsets '(arabic-iso8859-6
 110                   arabic-digit
 111                   arabic-1-column
 112                   arabic-2-column)))
 113   (while charsets
 114 ;;     (modify-syntax-entry (make-char (car charsets)) "w")
 115     (modify-category-entry (make-char (car charsets)) ?b)
 116     (setq charsets (cdr charsets))))
 117 (let ((ch #x600))
 118   (while (<= ch #x6ff)
 119     (modify-category-entry (decode-char 'ucs ch) ?b)
 120     (setq ch (1+ ch)))
 121   (setq ch #xfb50)
 122   (while (<= ch #xfdff)
 123     (modify-category-entry (decode-char 'ucs ch) ?b)
 124     (setq ch (1+ ch)))
 125   (setq ch #xfe70)
 126   (while (<= ch #xfefe)
 127     (modify-category-entry (decode-char 'ucs ch) ?b)
 128     (setq ch (1+ ch))))
 129
 130 ;; Chinese character set (GB2312)
 131
 132 ;; (modify-syntax-entry (make-char 'chinese-gb2312) "w")
 133 (modify-syntax-entry (make-char 'chinese-gb2312 33) "_")
 134 (modify-syntax-entry (make-char 'chinese-gb2312 34) "_")
 135 (modify-syntax-entry (make-char 'chinese-gb2312 41) "_")
 136 (modify-syntax-entry ?\\e$A!2\e(B "(\e$A!3\e(B")
 137 (modify-syntax-entry ?\\e$A!4\e(B "(\e$A!5\e(B")
 138 (modify-syntax-entry ?\\e$A!6\e(B "(\e$A!7\e(B")
 139 (modify-syntax-entry ?\\e$A!8\e(B "(\e$A!9\e(B")
 140 (modify-syntax-entry ?\\e$A!:\e(B "(\e$A!;\e(B")
 141 (modify-syntax-entry ?\\e$A!<\e(B "(\e$A!=\e(B")
 142 (modify-syntax-entry ?\\e$A!>\e(B "(\e$A!?\e(B")
 143 (modify-syntax-entry ?\\e$A!3\e(B ")\e$A!2\e(B")
 144 (modify-syntax-entry ?\\e$A!5\e(B ")\e$A!4\e(B")
 145 (modify-syntax-entry ?\\e$A!7\e(B ")\e$A!6\e(B")
 146 (modify-syntax-entry ?\\e$A!9\e(B ")\e$A!8\e(B")
 147 (modify-syntax-entry ?\\e$A!;\e(B ")\e$A!:\e(B")
 148 (modify-syntax-entry ?\\e$A!=\e(B ")\e$A!<\e(B")
 149 (modify-syntax-entry ?\\e$A!?\e(B ")\e$A!>\e(B")
 150 ;; Unicode equivalents of above
 151 (modify-syntax-entry ?\\e$,2=T\e(B "(\e$,2=U\e(B")
 152 (modify-syntax-entry ?\\e$,2=H\e(B "(\e$,2=I\e(B")
 153 (modify-syntax-entry ?\\e$,2=J\e(B "(\e$,2=K\e(B")
 154 (modify-syntax-entry ?\\e$,2=L\e(B "(\e$,2=M\e(B")
 155 (modify-syntax-entry ?\\e$,2=N\e(B "(\e$,2=O\e(B")
 156 (modify-syntax-entry ?\\e$,2=V\e(B "(\e$,2=W\e(B")
 157 (modify-syntax-entry ?\\e$,2=P\e(B "(\e$,2=Q\e(B")
 158 (modify-syntax-entry ?\\e$,2=U\e(B ")\e$,2=T\e(B")
 159 (modify-syntax-entry ?\\e$,2=I\e(B ")\e$,2=H\e(B")
 160 (modify-syntax-entry ?\\e$,2=K\e(B ")\e$,2=J\e(B")
 161 (modify-syntax-entry ?\\e$,2=M\e(B ")\e$,2=L\e(B")
 162 (modify-syntax-entry ?\\e$,2=O\e(B ")\e$,2=N\e(B")
 163 (modify-syntax-entry ?\\e$,2=W\e(B ")\e$,2=V\e(B")
 164 (modify-syntax-entry ?\\e$,2=Q\e(B ")\e$,2=P\e(B")
 165
 166 (modify-category-entry (make-char 'chinese-gb2312) ?c)
 167 (modify-category-entry (make-char 'chinese-gb2312) ?\|)
 168 (modify-category-entry (make-char 'chinese-gb2312 35) ?A)
 169 (modify-category-entry (make-char 'chinese-gb2312 36) ?H)
 170 (modify-category-entry (make-char 'chinese-gb2312 37) ?K)
 171 (modify-category-entry (make-char 'chinese-gb2312 38) ?G)
 172 (modify-category-entry (make-char 'chinese-gb2312 39) ?Y)
 173 (let ((row 48))
 174   (while (< row 127)
 175     (modify-category-entry (make-char 'chinese-gb2312 row) ?C)
 176     (setq row (1+ row))))
 177
 178 ;; Chinese character set (BIG5)
 179
 180 (let ((generic-big5-1-char (make-char 'chinese-big5-1))
 181       (generic-big5-2-char (make-char 'chinese-big5-2)))
 182 ;;   (modify-syntax-entry generic-big5-1-char "w")
 183 ;;   (modify-syntax-entry generic-big5-2-char "w")
 184
 185   (modify-category-entry generic-big5-1-char ?c)
 186   (modify-category-entry generic-big5-2-char ?c)
 187
 188   (modify-category-entry generic-big5-1-char ?C)
 189   (modify-category-entry generic-big5-2-char ?C)
 190
 191   (modify-category-entry generic-big5-1-char ?\|)
 192   (modify-category-entry generic-big5-2-char ?\|))
 193
 194
 195 ;; Chinese character set (CNS11643)
 196
 197 (let ((cns-list '(chinese-cns11643-1
 198                   chinese-cns11643-2
 199                   chinese-cns11643-3
 200                   chinese-cns11643-4
 201                   chinese-cns11643-5
 202                   chinese-cns11643-6
 203                   chinese-cns11643-7))
 204       generic-char)
 205   (while cns-list
 206     (setq generic-char (make-char (car cns-list)))
 207 ;;     (modify-syntax-entry generic-char "w")
 208     (modify-category-entry generic-char ?c)
 209     (modify-category-entry generic-char ?C)
 210     (modify-category-entry generic-char ?|)
 211     (setq cns-list (cdr cns-list))))
 212
 213 ;; Cyrillic character set (ISO-8859-5)
 214
 215 (modify-category-entry (make-char 'cyrillic-iso8859-5) ?y)
 216
 217 (modify-syntax-entry (make-char 'cyrillic-iso8859-5 160) " ")
 218 (modify-syntax-entry ?\e,L-\e(B ".")
 219 (modify-syntax-entry ?\e,Lp\e(B ".")
 220 (modify-syntax-entry ?\e,L}\e(B ".")
 221 (let ((tbl (standard-case-table)))
 222   (set-case-syntax-pair ?\e,L!\e(B ?\e,Lq\e(B tbl)
 223   (set-case-syntax-pair ?\e,L"\e(B ?\e,Lr\e(B tbl)
 224   (set-case-syntax-pair ?\e,L#\e(B ?\e,Ls\e(B tbl)
 225   (set-case-syntax-pair ?\e,L$\e(B ?\e,Lt\e(B tbl)
 226   (set-case-syntax-pair ?\e,L%\e(B ?\e,Lu\e(B tbl)
 227   (set-case-syntax-pair ?\e,L&\e(B ?\e,Lv\e(B tbl)
 228   (set-case-syntax-pair ?\e,L'\e(B ?\e,Lw\e(B tbl)
 229   (set-case-syntax-pair ?\e,L(\e(B ?\e,Lx\e(B tbl)
 230   (set-case-syntax-pair ?\e,L)\e(B ?\e,Ly\e(B tbl)
 231   (set-case-syntax-pair ?\e,L*\e(B ?\e,Lz\e(B tbl)
 232   (set-case-syntax-pair ?\e,L+\e(B ?\e,L{\e(B tbl)
 233   (set-case-syntax-pair ?\e,L,\e(B ?\e,L|\e(B tbl)
 234   (set-case-syntax-pair ?\e,L.\e(B ?\e,L~\e(B tbl)
 235   (set-case-syntax-pair ?\e,L/\e(B ?\e,L\7f\e(B tbl)
 236   (set-case-syntax-pair ?\e,L0\e(B ?\e,LP\e(B tbl)
 237   (set-case-syntax-pair ?\e,L1\e(B ?\e,LQ\e(B tbl)
 238   (set-case-syntax-pair ?\e,L2\e(B ?\e,LR\e(B tbl)
 239   (set-case-syntax-pair ?\e,L3\e(B ?\e,LS\e(B tbl)
 240   (set-case-syntax-pair ?\e,L4\e(B ?\e,LT\e(B tbl)
 241   (set-case-syntax-pair ?\e,L5\e(B ?\e,LU\e(B tbl)
 242   (set-case-syntax-pair ?\e,L6\e(B ?\e,LV\e(B tbl)
 243   (set-case-syntax-pair ?\e,L7\e(B ?\e,LW\e(B tbl)
 244   (set-case-syntax-pair ?\e,L8\e(B ?\e,LX\e(B tbl)
 245   (set-case-syntax-pair ?\e,L9\e(B ?\e,LY\e(B tbl)
 246   (set-case-syntax-pair ?\e,L:\e(B ?\e,LZ\e(B tbl)
 247   (set-case-syntax-pair ?\e,L;\e(B ?\e,L[\e(B tbl)
 248   (set-case-syntax-pair ?\e,L<\e(B ?\e,L\\e(B tbl)
 249   (set-case-syntax-pair ?\e,L=\e(B ?\e,L]\e(B tbl)
 250   (set-case-syntax-pair ?\e,L>\e(B ?\e,L^\e(B tbl)
 251   (set-case-syntax-pair ?\e,L?\e(B ?\e,L_\e(B tbl)
 252   (set-case-syntax-pair ?\e,L@\e(B ?\e,L`\e(B tbl)
 253   (set-case-syntax-pair ?\e,LA\e(B ?\e,La\e(B tbl)
 254   (set-case-syntax-pair ?\e,LB\e(B ?\e,Lb\e(B tbl)
 255   (set-case-syntax-pair ?\e,LC\e(B ?\e,Lc\e(B tbl)
 256   (set-case-syntax-pair ?\e,LD\e(B ?\e,Ld\e(B tbl)
 257   (set-case-syntax-pair ?\e,LE\e(B ?\e,Le\e(B tbl)
 258   (set-case-syntax-pair ?\e,LF\e(B ?\e,Lf\e(B tbl)
 259   (set-case-syntax-pair ?\e,LG\e(B ?\e,Lg\e(B tbl)
 260   (set-case-syntax-pair ?\e,LH\e(B ?\e,Lh\e(B tbl)
 261   (set-case-syntax-pair ?\e,LI\e(B ?\e,Li\e(B tbl)
 262   (set-case-syntax-pair ?\e,LJ\e(B ?\e,Lj\e(B tbl)
 263   (set-case-syntax-pair ?\e,LK\e(B ?\e,Lk\e(B tbl)
 264   (set-case-syntax-pair ?\e,LL\e(B ?\e,Ll\e(B tbl)
 265   (set-case-syntax-pair ?\e,LM\e(B ?\e,Lm\e(B tbl)
 266   (set-case-syntax-pair ?\e,LN\e(B ?\e,Ln\e(B tbl)
 267   (set-case-syntax-pair ?\e,LO\e(B ?\e,Lo\e(B tbl)
 268   (set-case-syntax-pair ?\e$,1(!\e(B ?\e$,1(q\e(B tbl)
 269   (set-case-syntax-pair ?\e$,1("\e(B ?\e$,1(r\e(B tbl)
 270   (set-case-syntax-pair ?\e$,1(#\e(B ?\e$,1(s\e(B tbl)
 271   (set-case-syntax-pair ?\e$,1($\e(B ?\e$,1(t\e(B tbl)
 272   (set-case-syntax-pair ?\e$,1(%\e(B ?\e$,1(u\e(B tbl)
 273   (set-case-syntax-pair ?\e$,1(&\e(B ?\e$,1(v\e(B tbl)
 274   (set-case-syntax-pair ?\e$,1('\e(B ?\e$,1(w\e(B tbl)
 275   (set-case-syntax-pair ?\e$,1((\e(B ?\e$,1(x\e(B tbl)
 276   (set-case-syntax-pair ?\e$,1()\e(B ?\e$,1(y\e(B tbl)
 277   (set-case-syntax-pair ?\e$,1(*\e(B ?\e$,1(z\e(B tbl)
 278   (set-case-syntax-pair ?\e$,1(+\e(B ?\e$,1({\e(B tbl)
 279   (set-case-syntax-pair ?\e$,1(,\e(B ?\e$,1(|\e(B tbl)
 280   (set-case-syntax-pair ?\e$,1(.\e(B ?\e$,1(~\e(B tbl)
 281   (set-case-syntax-pair ?\e$,1(/\e(B ?\e$,1(\7f\e(B tbl)
 282   (set-case-syntax-pair ?\e$,1(0\e(B ?\e$,1(P\e(B tbl)
 283   (set-case-syntax-pair ?\e$,1(1\e(B ?\e$,1(Q\e(B tbl)
 284   (set-case-syntax-pair ?\e$,1(2\e(B ?\e$,1(R\e(B tbl)
 285   (set-case-syntax-pair ?\e$,1(3\e(B ?\e$,1(S\e(B tbl)
 286   (set-case-syntax-pair ?\e$,1(4\e(B ?\e$,1(T\e(B tbl)
 287   (set-case-syntax-pair ?\e$,1(5\e(B ?\e$,1(U\e(B tbl)
 288   (set-case-syntax-pair ?\e$,1(6\e(B ?\e$,1(V\e(B tbl)
 289   (set-case-syntax-pair ?\e$,1(7\e(B ?\e$,1(W\e(B tbl)
 290   (set-case-syntax-pair ?\e$,1(8\e(B ?\e$,1(X\e(B tbl)
 291   (set-case-syntax-pair ?\e$,1(9\e(B ?\e$,1(Y\e(B tbl)
 292   (set-case-syntax-pair ?\e$,1(:\e(B ?\e$,1(Z\e(B tbl)
 293   (set-case-syntax-pair ?\e$,1(;\e(B ?\e$,1([\e(B tbl)
 294   (set-case-syntax-pair ?\e$,1(<\e(B ?\e$,1(\\e(B tbl)
 295   (set-case-syntax-pair ?\e$,1(=\e(B ?\e$,1(]\e(B tbl)
 296   (set-case-syntax-pair ?\e$,1(>\e(B ?\e$,1(^\e(B tbl)
 297   (set-case-syntax-pair ?\e$,1(?\e(B ?\e$,1(_\e(B tbl)
 298   (set-case-syntax-pair ?\e$,1(@\e(B ?\e$,1(`\e(B tbl)
 299   (set-case-syntax-pair ?\e$,1(A\e(B ?\e$,1(a\e(B tbl)
 300   (set-case-syntax-pair ?\e$,1(B\e(B ?\e$,1(b\e(B tbl)
 301   (set-case-syntax-pair ?\e$,1(C\e(B ?\e$,1(c\e(B tbl)
 302   (set-case-syntax-pair ?\e$,1(D\e(B ?\e$,1(d\e(B tbl)
 303   (set-case-syntax-pair ?\e$,1(E\e(B ?\e$,1(e\e(B tbl)
 304   (set-case-syntax-pair ?\e$,1(F\e(B ?\e$,1(f\e(B tbl)
 305   (set-case-syntax-pair ?\e$,1(G\e(B ?\e$,1(g\e(B tbl)
 306   (set-case-syntax-pair ?\e$,1(H\e(B ?\e$,1(h\e(B tbl)
 307   (set-case-syntax-pair ?\e$,1(I\e(B ?\e$,1(i\e(B tbl)
 308   (set-case-syntax-pair ?\e$,1(J\e(B ?\e$,1(j\e(B tbl)
 309   (set-case-syntax-pair ?\e$,1(K\e(B ?\e$,1(k\e(B tbl)
 310   (set-case-syntax-pair ?\e$,1(L\e(B ?\e$,1(l\e(B tbl)
 311   (set-case-syntax-pair ?\e$,1(M\e(B ?\e$,1(m\e(B tbl)
 312   (set-case-syntax-pair ?\e$,1(N\e(B ?\e$,1(n\e(B tbl)
 313   (set-case-syntax-pair ?\e$,1(O\e(B ?\e$,1(o\e(B tbl))
 314
 315 ;; Devanagari character set
 316
 317 ;;; Commented out since the categories appear not to be used anywhere
 318 ;;; and word syntax is the default.
 319 ;; (let ((deflist       '(;; chars      syntax  category
 320 ;;                ("\e$(5!!!"!#\e(B"      "w"     ?7) ; vowel-modifying diacritical mark
 321 ;;                                          ; chandrabindu, anuswar, visarga
 322 ;;                ("\e$(5!$\e(B-\e$(5!2\e(B"        "w"     ?1) ; independent vowel
 323 ;;                ("\e$(5!3\e(B-\e$(5!X\e(B"        "w"     ?0) ; consonant
 324 ;;                ("\e$(5!Z\e(B-\e$(5!g\e(B"        "w"     ?8) ; matra
 325 ;;                ("\e$(5!q\e(B-\e$(5!z\e(B"        "w"     ?6) ; digit
 326 ;;                ;; Unicode equivalents
 327 ;;                ("\e$,15A5B5C\e(B"      "w"     ?7) ; vowel-modifying diacritical mark
 328 ;;                                          ; chandrabindu, anuswar, visarga
 329 ;;                ("\e$,15E\e(B-\e$,15M\e(B"        "w"     ?1) ; independent vowel
 330 ;;                ("\e$,15U\e(B-\e$,15y\e(B"        "w"     ?0) ; consonant
 331 ;;                ("\e$,15~\e(B-\e$,16)\e(B"        "w"     ?8) ; matra
 332 ;;                ("\e$,16F\e(B-\e$,16O\e(B"        "w"     ?6) ; digit
 333 ;;                ))
 334 ;;       elm chars len syntax category to ch i)
 335 ;;   (while deflist
 336 ;;     (setq elm (car deflist))
 337 ;;     (setq chars (car elm)
 338 ;;        len (length chars)
 339 ;;        syntax (nth 1 elm)
 340 ;;        category (nth 2 elm)
 341 ;;        i 0)
 342 ;;     (while (< i len)
 343 ;;       (if (= (aref chars i) ?-)
 344 ;;        (setq i (1+ i)
 345 ;;              to (aref chars i))
 346 ;;      (setq ch (aref chars i)
 347 ;;            to ch))
 348 ;;       (while (<= ch to)
 349 ;;      (modify-syntax-entry ch syntax)
 350 ;;      (modify-category-entry ch category)
 351 ;;      (setq ch (1+ ch)))
 352 ;;       (setq i (1+ i)))
 353 ;;     (setq deflist (cdr deflist))))
 354
 355 ;; Ethiopic character set
 356
 357 (modify-category-entry (make-char 'ethiopic) ?e)
 358 ;; (modify-syntax-entry (make-char 'ethiopic) "w")
 359 (dotimes (i (1+ (- #x137c #x1200)))
 360   (modify-category-entry (decode-char 'ucs (+ #x1200 i)) ?e))
 361 (let ((chars '(?\e$(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$(3%x\e(B
 362                ;; Unicode equivalents of the above:
 363                ?\e$,1Q!\e(B ?\e$,1Q"\e(B ?\e$,1Q#\e(B ?\e$,1Q$\e(B ?\e$,1Q%\e(B ?\e$,1Q&\e(B ?\e$,1Q'\e(B ?\e$,1Q(\e(B ?\e$,3op\e(B ?\e$,3o{\e(B ?\e$,3o|\e(B ?\e$,3o}\e(B ?\e$,3o~\e(B ?\e$,3o\7f\e(B)))
 364   (while chars
 365     (modify-syntax-entry (car chars) ".")
 366     (setq chars (cdr chars))))
 367
 368 ;; Greek character set (ISO-8859-7)
 369
 370 (modify-category-entry (make-char 'greek-iso8859-7) ?g)
 371 (let ((c #x370))
 372   (while (<= c #x3ff)
 373     (modify-category-entry (decode-char 'ucs c) ?g)
 374     (setq c (1+ c))))
 375
 376 ;; (let ((c 182))
 377 ;;   (while (< c 255)
 378 ;;     (modify-syntax-entry (make-char 'greek-iso8859-7 c) "w")
 379 ;;     (setq c (1+ c))))
 380 ;; (modify-syntax-entry (make-char 'greek-iso8859-7 160) "w") ; NBSP
 381 (modify-syntax-entry ?\e,F7\e(B ".")
 382 (modify-syntax-entry ?\e,F;\e(B ".")
 383 (modify-syntax-entry ?\e,F=\e(B ".")
 384 (let ((tbl (standard-case-table)))
 385   ;; Fixme: non-letter syntax copied from latin-1, but that's dubious
 386   ;; in several cases.
 387   (set-case-syntax ?\e,F!\e(B "." tbl)
 388   (set-case-syntax ?\e,F"\e(B "." tbl)
 389   (set-case-syntax ?\e,F&\e(B "." tbl)
 390   (set-case-syntax ?\e,F&\e(B "_" tbl)
 391   (set-case-syntax ?\e,F'\e(B "." tbl)
 392   (set-case-syntax ?\e,F)\e(B "_" tbl)
 393   (set-case-syntax ?\e,F+\e(B "." tbl)
 394   (set-case-syntax ?\e,F,\e(B "_" tbl)
 395   (set-case-syntax ?\e,F-\e(B "_" tbl)
 396   (set-case-syntax ?\e,F/\e(B "." tbl)
 397   (set-case-syntax ?\e,F0\e(B "_" tbl)
 398   (set-case-syntax ?\e,F1\e(B "_" tbl)
 399 ;;  (set-case-syntax ?\e,F7\e(B "_" tbl)
 400 ;;  (set-case-syntax ?\e,F=\e(B "_" tbl)
 401   (set-case-syntax-pair ?\e,FA\e(B ?\e,Fa\e(B tbl)
 402   (set-case-syntax-pair ?\e,FB\e(B ?\e,Fb\e(B tbl)
 403   (set-case-syntax-pair ?\e,FC\e(B ?\e,Fc\e(B tbl)
 404   (set-case-syntax-pair ?\e,FD\e(B ?\e,Fd\e(B tbl)
 405   (set-case-syntax-pair ?\e,FE\e(B ?\e,Fe\e(B tbl)
 406   (set-case-syntax-pair ?\e,FF\e(B ?\e,Ff\e(B tbl)
 407   (set-case-syntax-pair ?\e,FG\e(B ?\e,Fg\e(B tbl)
 408   (set-case-syntax-pair ?\e,FH\e(B ?\e,Fh\e(B tbl)
 409   (set-case-syntax-pair ?\e,FI\e(B ?\e,Fi\e(B tbl)
 410   (set-case-syntax-pair ?\e,FJ\e(B ?\e,Fj\e(B tbl)
 411   (set-case-syntax-pair ?\e,FK\e(B ?\e,Fk\e(B tbl)
 412   (set-case-syntax-pair ?\e,FL\e(B ?\e,Fl\e(B tbl)
 413   (set-case-syntax-pair ?\e,FM\e(B ?\e,Fm\e(B tbl)
 414   (set-case-syntax-pair ?\e,FN\e(B ?\e,Fn\e(B tbl)
 415   (set-case-syntax-pair ?\e,FO\e(B ?\e,Fo\e(B tbl)
 416   (set-case-syntax-pair ?\e,FP\e(B ?\e,Fp\e(B tbl)
 417   (set-case-syntax-pair ?\e,FQ\e(B ?\e,Fq\e(B tbl)
 418   (set-case-syntax-pair ?\e,FS\e(B ?\e,Fs\e(B tbl)
 419   (set-case-syntax-pair ?\e,FT\e(B ?\e,Ft\e(B tbl)
 420   (set-case-syntax-pair ?\e,FU\e(B ?\e,Fu\e(B tbl)
 421   (set-case-syntax-pair ?\e,FV\e(B ?\e,Fv\e(B tbl)
 422   (set-case-syntax-pair ?\e,FW\e(B ?\e,Fw\e(B tbl)
 423   (set-case-syntax-pair ?\e,FX\e(B ?\e,Fx\e(B tbl)
 424   (set-case-syntax-pair ?\e,FY\e(B ?\e,Fy\e(B tbl)
 425   (set-case-syntax-pair ?\e,FZ\e(B ?\e,Fz\e(B tbl)
 426   (set-case-syntax-pair ?\e,F[\e(B ?\e,F{\e(B tbl)
 427   (set-case-syntax-pair ?\e,F?\e(B ?\e,F~\e(B tbl)
 428   (set-case-syntax-pair ?\e,F>\e(B ?\e,F}\e(B tbl)
 429   (set-case-syntax-pair ?\e,F<\e(B ?\e,F|\e(B tbl)
 430   (set-case-syntax-pair ?\e,F6\e(B ?\e,F\\e(B tbl)
 431   (set-case-syntax-pair ?\e,F8\e(B ?\e,F]\e(B tbl)
 432   (set-case-syntax-pair ?\e,F9\e(B ?\e,F^\e(B tbl)
 433   (set-case-syntax-pair ?\e,F:\e(B ?\e,F_\e(B tbl)
 434   ;; Unicode equivalents
 435   (set-case-syntax-pair ?\e$,1&q\e(B ?\e$,1'1\e(B tbl)
 436   (set-case-syntax-pair ?\e$,1&r\e(B ?\e$,1'2\e(B tbl)
 437   (set-case-syntax-pair ?\e$,1&s\e(B ?\e$,1'3\e(B tbl)
 438   (set-case-syntax-pair ?\e$,1&t\e(B ?\e$,1'4\e(B tbl)
 439   (set-case-syntax-pair ?\e$,1&u\e(B ?\e$,1'5\e(B tbl)
 440   (set-case-syntax-pair ?\e$,1&v\e(B ?\e$,1'6\e(B tbl)
 441   (set-case-syntax-pair ?\e$,1&w\e(B ?\e$,1'7\e(B tbl)
 442   (set-case-syntax-pair ?\e$,1&x\e(B ?\e$,1'8\e(B tbl)
 443   (set-case-syntax-pair ?\e$,1&y\e(B ?\e$,1'9\e(B tbl)
 444   (set-case-syntax-pair ?\e$,1&z\e(B ?\e$,1':\e(B tbl)
 445   (set-case-syntax-pair ?\e$,1&{\e(B ?\e$,1';\e(B tbl)
 446   (set-case-syntax-pair ?\e$,1&|\e(B ?\e$,1'<\e(B tbl)
 447   (set-case-syntax-pair ?\e$,1&}\e(B ?\e$,1'=\e(B tbl)
 448   (set-case-syntax-pair ?\e$,1&~\e(B ?\e$,1'>\e(B tbl)
 449   (set-case-syntax-pair ?\e$,1&\7f\e(B ?\e$,1'?\e(B tbl)
 450   (set-case-syntax-pair ?\e$,1' \e(B ?\e$,1'@\e(B tbl)
 451   (set-case-syntax-pair ?\e$,1'!\e(B ?\e$,1'A\e(B tbl)
 452   (set-case-syntax-pair ?\e$,1'#\e(B ?\e$,1'C\e(B tbl)
 453   (set-case-syntax-pair ?\e$,1'$\e(B ?\e$,1'D\e(B tbl)
 454   (set-case-syntax-pair ?\e$,1'%\e(B ?\e$,1'E\e(B tbl)
 455   (set-case-syntax-pair ?\e$,1'&\e(B ?\e$,1'F\e(B tbl)
 456   (set-case-syntax-pair ?\e$,1''\e(B ?\e$,1'G\e(B tbl)
 457   (set-case-syntax-pair ?\e$,1'(\e(B ?\e$,1'H\e(B tbl)
 458   (set-case-syntax-pair ?\e$,1')\e(B ?\e$,1'I\e(B tbl)
 459   (set-case-syntax-pair ?\e$,1'*\e(B ?\e$,1'J\e(B tbl)
 460   (set-case-syntax-pair ?\e$,1'+\e(B ?\e$,1'K\e(B tbl)
 461   (set-case-syntax-pair ?\e$,1&o\e(B ?\e$,1'N\e(B tbl)
 462   (set-case-syntax-pair ?\e$,1&n\e(B ?\e$,1'M\e(B tbl)
 463   (set-case-syntax-pair ?\e$,1&l\e(B ?\e$,1'L\e(B tbl)
 464   (set-case-syntax-pair ?\e$,1&f\e(B ?\e$,1',\e(B tbl)
 465   (set-case-syntax-pair ?\e$,1&h\e(B ?\e$,1'-\e(B tbl)
 466   (set-case-syntax-pair ?\e$,1&i\e(B ?\e$,1'.\e(B tbl)
 467   (set-case-syntax-pair ?\e$,1&j\e(B ?\e$,1'/\e(B tbl))
 468
 469 ;; Hebrew character set (ISO-8859-8)
 470
 471 (modify-category-entry (make-char 'hebrew-iso8859-8) ?w)
 472 (let ((c #x591))
 473   (while (<= c #x5f4)
 474     (modify-category-entry (decode-char 'ucs c) ?w)
 475     (setq c (1+ c))))
 476
 477 (modify-syntax-entry (make-char 'hebrew-iso8859-8 208) ".") ; PASEQ
 478 (modify-syntax-entry (make-char 'hebrew-iso8859-8 211) ".") ; SOF PASUQ
 479 (modify-syntax-entry (decode-char 'ucs #x5be) ".") ; MAQAF
 480 (modify-syntax-entry (decode-char 'ucs #x5c0) ".") ; PASEQ
 481 (modify-syntax-entry (decode-char 'ucs #x5c3) ".") ; SOF PASUQ
 482 (modify-syntax-entry (decode-char 'ucs #x5f3) ".") ; GERESH
 483 (modify-syntax-entry (decode-char 'ucs #x5f4) ".") ; GERSHAYIM
 484
 485 ;; (let ((c 224))
 486 ;;   (while (< c 251)
 487 ;;     (modify-syntax-entry (make-char 'hebrew-iso8859-8 c) "w")
 488 ;;     (setq c (1+ c))))
 489 ;; (modify-syntax-entry (make-char 'hebrew-iso8859-8 160) "w") ; NBSP
 490
 491 ;; Indian character set (IS 13194 and other Emacs original Indian charsets)
 492
 493 (modify-category-entry (make-char 'indian-is13194) ?i)
 494 (modify-category-entry (make-char 'indian-2-column) ?I)
 495 (modify-category-entry (make-char 'indian-glyph) ?I)
 496 ;; Unicode Devanagari block
 497 (let ((c #x901))
 498   (while (<= c #x970)
 499     (modify-category-entry (decode-char 'ucs c) ?i)
 500     (setq c (1+ c))))
 501
 502 ;;; Commented out since the categories appear not to be used anywhere
 503 ;;; and word syntax is the default.
 504 ;; (let ((deflist                               ;
 505 ;;      '(;; chars      syntax  category
 506 ;;        ("\e(5!"#\e(B"  "w"     ?7) ; vowel-modifying diacritical mark
 507 ;;                                  ; chandrabindu, anuswar, visarga
 508 ;;        ("\e(5$\e(B-\e(52\e(B"    "w"     ?1) ; base (independent) vowel
 509 ;;        ("\e(53\e(B-\e(5X\e(B"    "w"     ?0) ; consonant
 510 ;;        ("\e(5Z\e(B-\e(5g\e(B"    "w"     ?8) ; matra
 511 ;;        ("\e(5q\e(B-\e(5z\e(B"    "w"     ?6) ; digit
 512 ;;        ))
 513 ;;       elm chars len syntax category to ch i)
 514 ;;   (while deflist
 515 ;;     (setq elm (car deflist))
 516 ;;     (setq chars (car elm)
 517 ;;        len (length chars)
 518 ;;        syntax (nth 1 elm)
 519 ;;        category (nth 2 elm)
 520 ;;        i 0)
 521 ;;     (while (< i len)
 522 ;;       (if (= (aref chars i) ?-)
 523 ;;        (setq i (1+ i)
 524 ;;              to (aref chars i))
 525 ;;      (setq ch (aref chars i)
 526 ;;            to ch))
 527 ;;       (while (<= ch to)
 528 ;;      (modify-syntax-entry ch syntax)
 529 ;;      (modify-category-entry ch category)
 530 ;;      (setq ch (1+ ch)))
 531 ;;       (setq i (1+ i)))
 532 ;;     (setq deflist (cdr deflist))))
 533
 534
 535 ;; Japanese character set (JISX0201-kana, JISX0201-roman, JISX0208, JISX0212)
 536
 537 (modify-category-entry (make-char 'katakana-jisx0201) ?k)
 538 (modify-category-entry (make-char 'katakana-jisx0201) ?j)
 539 (modify-category-entry (make-char 'latin-jisx0201) ?r)
 540 (modify-category-entry (make-char 'japanese-jisx0208) ?j)
 541 (modify-category-entry (make-char 'japanese-jisx0212) ?j)
 542 (modify-category-entry (make-char 'katakana-jisx0201) ?\|)
 543 (modify-category-entry (make-char 'japanese-jisx0208) ?\|)
 544 (modify-category-entry (make-char 'japanese-jisx0212) ?\|)
 545
 546 ;; Unicode equivalents of JISX0201-kana
 547 (let ((c #xff61))
 548   (while (<= c #xff9f)
 549     (modify-category-entry (decode-char 'ucs c) ?k)
 550     (modify-category-entry (decode-char 'ucs c) ?j)
 551     (modify-category-entry (decode-char 'ucs c) ?\|)
 552     (setq c (1+ c))))
 553
 554 ;; Katakana block
 555 (let ((c #x30a0))
 556   (while (<= c #x30ff)
 557     ;; ?K is double width, ?k isn't specified
 558     (modify-category-entry (decode-char 'ucs c) ?k)
 559     (modify-category-entry (decode-char 'ucs c) ?j)
 560     (modify-category-entry (decode-char 'ucs c) ?\|)
 561     (setq c (1+ c))))
 562
 563 ;; Hiragana block
 564 (let ((c #x3040))
 565   (while (<= c #x309f)
 566     ;; ?H is actually defined to be double width
 567     (modify-category-entry (decode-char 'ucs c) ?H)
 568     ;;(modify-category-entry (decode-char 'ucs c) ?j)
 569     (modify-category-entry (decode-char 'ucs c) ?\|)
 570     (setq c (1+ c))))
 571
 572 ;; JISX0208
 573 ;; (modify-syntax-entry (make-char 'japanese-jisx0208) "w")
 574 (modify-syntax-entry (make-char 'japanese-jisx0208 33) "_")
 575 (modify-syntax-entry (make-char 'japanese-jisx0208 34) "_")
 576 (modify-syntax-entry (make-char 'japanese-jisx0208 40) "_")
 577 (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)))
 578   (while chars
 579     (modify-syntax-entry (car chars) "w")
 580     (setq chars (cdr chars))))
 581 (modify-syntax-entry ?\\e$B!J\e(B "(\e$B!K\e(B")
 582 (modify-syntax-entry ?\\e$B!N\e(B "(\e$B!O\e(B")
 583 (modify-syntax-entry ?\\e$B!P\e(B "(\e$B!Q\e(B")
 584 (modify-syntax-entry ?\\e$B!V\e(B "(\e$B!W\e(B")
 585 (modify-syntax-entry ?\\e$B!X\e(B "(\e$B!Y\e(B")
 586 (modify-syntax-entry ?\\e$B!K\e(B ")\e$B!J\e(B")
 587 (modify-syntax-entry ?\\e$B!O\e(B ")\e$B!N\e(B")
 588 (modify-syntax-entry ?\\e$B!Q\e(B ")\e$B!P\e(B")
 589 (modify-syntax-entry ?\\e$B!W\e(B ")\e$B!V\e(B")
 590 (modify-syntax-entry ?\\e$B!Y\e(B ")\e$B!X\e(B")
 591
 592 (modify-category-entry (make-char 'japanese-jisx0208 35) ?A)
 593 (modify-category-entry (make-char 'japanese-jisx0208 36) ?H)
 594 (modify-category-entry (make-char 'japanese-jisx0208 37) ?K)
 595 (modify-category-entry (make-char 'japanese-jisx0208 38) ?G)
 596 (modify-category-entry (make-char 'japanese-jisx0208 39) ?Y)
 597 (let ((row 48))
 598   (while (< row 127)
 599     (modify-category-entry (make-char 'japanese-jisx0208 row) ?C)
 600     (setq row (1+ row))))
 601 (modify-category-entry ?\e$B!<\e(B ?K)
 602 (let ((chars '(?\e$B!+\e(B ?\e$B!,\e(B)))
 603   (while chars
 604     (modify-category-entry (car chars) ?K)
 605     (modify-category-entry (car chars) ?H)
 606     (setq chars (cdr chars))))
 607 (let ((chars '(?\e$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)))
 608   (while chars
 609     (modify-category-entry (car chars) ?C)
 610     (setq chars (cdr chars))))
 611
 612 ;; JISX0212
 613 ;; (modify-syntax-entry (make-char 'japanese-jisx0212) "w")
 614 (modify-syntax-entry (make-char 'japanese-jisx0212 33) "_")
 615 (modify-syntax-entry (make-char 'japanese-jisx0212 34) "_")
 616 (modify-syntax-entry (make-char 'japanese-jisx0212 35) "_")
 617
 618 (modify-category-entry (make-char 'japanese-jisx0212 ) ?C)
 619
 620 ;; JISX0201-Kana
 621 ;; (modify-syntax-entry (make-char 'katakana-jisx0201) "w")
 622 (let ((chars '(?\e(I!\e(B ?\e(I$\e(B ?\e(I%\e(B
 623                ;; Unicode:
 624                ?\e$,3sa\e(B ?\e$,3sd\e(B ?\e$,3se\e(B)))
 625   (while chars
 626     (modify-syntax-entry (car chars) ".")
 627     (setq chars (cdr chars))))
 628
 629 (modify-syntax-entry ?\\e(I"\e(B "(\e(I#\e(B")
 630 (modify-syntax-entry ?\\e(I#\e(B "(\e(I"\e(B")
 631
 632 ;; Korean character set (KSC5601)
 633
 634 ;; (modify-syntax-entry (make-char 'korean-ksc5601) "w")
 635 (modify-syntax-entry (make-char 'korean-ksc5601 33) "_")
 636 (modify-syntax-entry (make-char 'korean-ksc5601 34) "_")
 637 (modify-syntax-entry (make-char 'korean-ksc5601 38) "_")
 638 (modify-syntax-entry (make-char 'korean-ksc5601 39) "_")
 639 (modify-syntax-entry (make-char 'korean-ksc5601 40) "_")
 640 (modify-syntax-entry (make-char 'korean-ksc5601 41) "_")
 641
 642 (modify-category-entry (make-char 'korean-ksc5601) ?h)
 643 (modify-category-entry (make-char 'korean-ksc5601 35) ?A)
 644 (modify-category-entry (make-char 'korean-ksc5601 37) ?G)
 645 (modify-category-entry (make-char 'korean-ksc5601 42) ?H)
 646 (modify-category-entry (make-char 'korean-ksc5601 43) ?K)
 647 (modify-category-entry (make-char 'korean-ksc5601 44) ?Y)
 648
 649 ;; Latin character set (latin-1,2,3,4,5,8,9)
 650
 651 (modify-category-entry (make-char 'latin-iso8859-1) ?l)
 652 (modify-category-entry (make-char 'latin-iso8859-2) ?l)
 653 (modify-category-entry (make-char 'latin-iso8859-3) ?l)
 654 (modify-category-entry (make-char 'latin-iso8859-4) ?l)
 655 (modify-category-entry (make-char 'latin-iso8859-9) ?l)
 656 (modify-category-entry (make-char 'latin-iso8859-14) ?l)
 657 (modify-category-entry (make-char 'latin-iso8859-15) ?l)
 658
 659 (modify-category-entry (make-char 'latin-iso8859-1 160) ?\ )
 660 (modify-category-entry (make-char 'latin-iso8859-2 160) ?\ )
 661 (modify-category-entry (make-char 'latin-iso8859-3 160) ?\ )
 662 (modify-category-entry (make-char 'latin-iso8859-4 160) ?\ )
 663 (modify-category-entry (make-char 'latin-iso8859-9 160) ?\ )
 664 (modify-category-entry (make-char 'latin-iso8859-14 160) ?\ )
 665 (modify-category-entry (make-char 'latin-iso8859-15 160) ?\ )
 666
 667 ;; Lao character set
 668
 669 (modify-category-entry (make-char 'lao) ?o)
 670 (dotimes (i (1+ (- #xeff #xe80)))
 671   (modify-category-entry (decode-char 'ucs (+ i #xe80)) ?o))
 672
 673 (let ((deflist  '(;; chars      syntax  category
 674                   ("\e(1!\e(B-\e(1N\e(B"    "w"     ?0) ; consonant
 675                   ("\e(1PRS]`\e(B-\e(1d\e(B"        "w"     ?1) ; vowel base
 676                   ("\e(1QT\e(B-\e(1W[m\e(B" "w"     ?2) ; vowel upper
 677                   ("\e(1XY\e(B"           "w"     ?3) ; vowel lower
 678                   ("\e(1h\e(B-\e(1l\e(B"    "w"     ?4) ; tone mark
 679                   ("\e(1\\e(B"            "w"     ?9) ; semivowel lower
 680                   ("\e(1p\e(B-\e(1y\e(B"    "w"     ?6) ; digit
 681                   ("\e(1Of\e(B"           "_"     ?5) ; symbol
 682                   ;; Unicode equivalents
 683                   ("\e$,1D!\e(B-\e$,1DN\e(B"        "w"     ?0) ; consonant
 684                   ("\e$,1DPDRDSD]D`\e(B-\e$,1Dd\e(B"        "w"     ?1) ; vowel base
 685                   ("\e$,1DQDT\e(B-\e$,1DWD[Dm\e(B"  "w"     ?2) ; vowel upper
 686                   ("\e$,1DXDY\e(B"        "w"     ?3) ; vowel lower
 687                   ("\e$,1Dh\e(B-\e$,1Dk\e(B"        "w"     ?4) ; tone mark
 688                   ("\e$,1D\D]\e(B"        "w"     ?9) ; semivowel lower
 689                   ("\e$,1Dp\e(B-\e$,1Dy\e(B"        "w"     ?6) ; digit
 690                   ("\e$,1DODf\e(B"        "_"     ?5) ; symbol
 691                   ))
 692       elm chars len syntax category to ch i)
 693   (while deflist
 694     (setq elm (car deflist))
 695     (setq chars (car elm)
 696           len (length chars)
 697           syntax (nth 1 elm)
 698           category (nth 2 elm)
 699           i 0)
 700     (while (< i len)
 701       (if (= (aref chars i) ?-)
 702           (setq i (1+ i)
 703                 to (aref chars i))
 704         (setq ch (aref chars i)
 705               to ch))
 706       (while (<= ch to)
 707         (unless (string-equal syntax "w")
 708           (modify-syntax-entry ch syntax))
 709         (modify-category-entry ch category)
 710         (setq ch (1+ ch)))
 711       (setq i (1+ i)))
 712     (setq deflist (cdr deflist))))
 713
 714 ;; Thai character set (TIS620)
 715
 716 (modify-category-entry (make-char 'thai-tis620) ?t)
 717 (dotimes (i (1+ (- #xe7f #xe00)))
 718   (modify-category-entry (decode-char 'ucs (+ i #xe00)) ?t))
 719
 720 (let ((deflist  '(;; chars      syntax  category
 721                   ("\e,T!\e(B-\e,TCEG\e(B-\e,TN\e(B"  "w"     ?0) ; consonant
 722                   ("\e,TDFPRS`\e(B-\e,Te\e(B"       "w"     ?1) ; vowel base
 723                   ("\e,TQT\e(B-\e,TWgn\e(B" "w"     ?2) ; vowel upper
 724                   ("\e,TX\e(B-\e,TZ\e(B"    "w"     ?3) ; vowel lower
 725                   ("\e,Th\e(B-\e,Tm\e(B"    "w"     ?4) ; tone mark
 726                   ("\e,Tp\e(B-\e,Ty\e(B"    "w"     ?6) ; digit
 727                   ("\e,TOf_oz{\e(B"       "_"     ?5) ; symbol
 728                   ;; Unicode equivalents
 729                   ("\e$,1Ba\e(B-\e$,1C#C%C'\e(B-\e$,1C.\e(B"  "w"     ?0) ; consonant
 730                   ("\e$,1C$C&C0C2C3C@\e(B-\e$,1CE\e(B"      "w"     ?1) ; vowel base
 731                   ("\e$,1C1C4\e(B-\e$,1C7CGCN\e(B"  "w"     ?2) ; vowel upper
 732                   ("\e$,1C8\e(B-\e$,1C:\e(B"        "w"     ?3) ; vowel lower
 733                   ("\e$,1CH\e(B-\e$,1CM\e(B"        "w"     ?4) ; tone mark
 734                   ("\e$,1CP\e(B-\e$,1CY\e(B"        "w"     ?6) ; digit
 735                   ("\e$,1C/CFC?COCZC[\e(B"        "_"     ?5) ; symbol
 736                   ))
 737       elm chars len syntax category to ch i)
 738   (while deflist
 739     (setq elm (car deflist))
 740     (setq chars (car elm)
 741           len (length chars)
 742           syntax (nth 1 elm)
 743           category (nth 2 elm)
 744           i 0)
 745     (while (< i len)
 746       (if (= (aref chars i) ?-)
 747           (setq i (1+ i)
 748                 to (aref chars i))
 749         (setq ch (aref chars i)
 750               to ch))
 751       (while (<= ch to)
 752         (unless (string-equal syntax "w")
 753           (modify-syntax-entry ch syntax))
 754         (modify-category-entry ch category)
 755         (setq ch (1+ ch)))
 756       (setq i (1+ i)))
 757     (setq deflist (cdr deflist))))
 758
 759 ;; Tibetan character set
 760
 761 (modify-category-entry (make-char 'tibetan) ?q)
 762 (modify-category-entry (make-char 'tibetan-1-column) ?q)
 763 (dotimes (i (1+ (- #xfff #xf00)))
 764   (modify-category-entry (decode-char 'ucs (+ i #xf00)) ?q))
 765
 766 (let ((deflist  '(;; chars             syntax category
 767                   ("\e4\e$(7"!\e0"!\e1\e(B-\e4\e$(7"J\e0"J\e1\e4"K\e0"K\e1\e(B"              "w"     ?0) ; consonant
 768                   ("\e$(7#!\e(B-\e$(7#J#K#L#M!"!#\e(B"       "w"     ?0) ;
 769                   ("\e$(7$!\e(B-\e$(7$e\e(B"              "w"     ?0) ;
 770                   ("\e$(7%!\e(B-\e$(7%u\e(B"              "w"     ?0) ;
 771                   ("\e$(7"S"["\"]"^"a\e(B"       "w"      ?2) ; upper vowel
 772                   ("\e$(7"_"c"d"g"h"i"j"k"l\e(B" "w"      ?2) ; upper modifier
 773                   ("\e$(7!I"Q"R"U"e!e!g\e(B"       "w"    ?3) ; lowel vowel/modifier
 774                   ("\e$(7!P\e(B-\e$(7!Y!Z\e(B-\e$(7!c\e(B"            "w"     ?6) ; digit
 775                   ("\e$(7!;!=\e(B-\e$(7!B!D"`\e(B"        "."     ?|) ; line-break char
 776                   ("\e$(8!;!=!?!@!A!D"`\e(B"            "."     ?|) ;
 777                   ("\e$(7!8!;!=\e(B-\e$(7!B!D"`!m!d\e(B"  "."     ?>) ; prohibition
 778                   ("\e$(8!;!=!?!@!A!D"`\e(B"            "."     ?>) ;
 779                   ("\e$(7!0\e(B-\e$(7!:!l#R#S"f\e(B"      "."     ?<) ; prohibition
 780                   ("\e$(7!C!E\e(B-\e$(7!H!J\e(B-\e$(7!O!f!h\e(B-\e$(7!k!n!o#O#P\e(B-\e$(7#`\e(B" "." ?q) ; others
 781
 782                   ;; Unicode version (not complete)
 783                   ("\e$,1F \e(B-\e$,1FIFJ\e(B"              "w"     ?0) ; consonant
 784                   ("\e$,1Fp\e(B-\e$,1G9G:G;G<\e(B"       "w"     ?0) ;
 785                   ("\e$,1FRFZF[F\F]F`\e(B"       "w"      ?2) ; upper vowel
 786                   ("\e$,1F^FbFcFfFgFhFiFjFk\e(B" "w"      ?2) ; upper modifier
 787                   ("\e$,1EYFPFQFTFdEuEw\e(B"       "w"    ?3) ; lowel vowel/modifier
 788                   ("\e$,1E`\e(B-\e$,1EiEj\e(B-\e$,1Es\e(B"            "w"     ?6) ; digit
 789                   ("\e$,1EKEM\e(B-\e$,1ERETF_\e(B"        "."     ?|) ; line-break char
 790                   ("\e$,1EHEKEM\e(B-\e$,1ERETF_E}Et\e(B"  "."     ?>) ; prohibition
 791                   ("\e$,1E@\e(B-\e$,1EJE|GAGBFe\e(B"      "."     ?<) ; prohibition
 792                   ("\e$,1ESEU\e(B-\e$,1EXEZ\e(B-\e$,1E_EvEx\e(B-\e$,1E{E~E\7fG>G?\e(B-\e$,1GO\e(B" "." ?q) ; others
 793                   ))
 794       elm chars len syntax category to ch i)
 795   (while deflist
 796     (setq elm (car deflist))
 797     (setq chars (car elm)
 798           len (length chars)
 799           syntax (nth 1 elm)
 800           category (nth 2 elm)
 801           i 0)
 802     (while (< i len)
 803       (if (= (aref chars i) ?-)
 804           (setq i (1+ i)
 805                 to (aref chars i))
 806         (setq ch (aref chars i)
 807               to ch))
 808       (while (<= ch to)
 809         (unless (string-equal syntax "w")
 810           (modify-syntax-entry ch syntax))
 811         (modify-category-entry ch category)
 812         (setq ch (1+ ch)))
 813       (setq i (1+ i)))
 814     (setq deflist (cdr deflist))))
 815
 816 ;; Vietnamese character set
 817
 818 (let ((lower (make-char 'vietnamese-viscii-lower))
 819       (upper (make-char 'vietnamese-viscii-upper)))
 820 ;;   (modify-syntax-entry lower "w")
 821 ;;   (modify-syntax-entry upper "w")
 822   (modify-category-entry lower ?v)
 823   (modify-category-entry upper ?v)
 824   (modify-category-entry lower ?l)      ; To make a word with
 825   (modify-category-entry upper ?l)      ; latin characters.
 826   )
 827
 828 (let ((tbl (standard-case-table))
 829       (i 32))
 830   (while (< i 128)
 831     (set-case-syntax-pair (make-char 'vietnamese-viscii-upper i)
 832                           (make-char 'vietnamese-viscii-lower i)
 833                           tbl)
 834     (setq i (1+ i))))
 835
 836 ;; Unicode (mule-unicode-0100-24ff)
 837
 838 (let ((tbl (standard-case-table)) c)
 839
 840 ;; In some languages, U+0049 LATIN CAPITAL LETTER I and U+0131 LATIN
 841 ;; SMALL LETTER DOTLESS I make a case pair, and so do U+0130 LATIN
 842 ;; CAPITAL LETTER I WITH DOT ABOVE and U+0069 LATIN SMALL LETTER I.
 843 ;; Thus we have to check language-environment to handle casing
 844 ;; correctly.  Currently only I<->i is available.
 845
 846   ;; Latin Extended-A, Latin Extended-B
 847   (setq c #x0100)
 848   (while (<= c #x0233)
 849     (modify-category-entry (decode-char 'ucs c) ?l)
 850     (and (or (<= c #x012e)
 851              (and (>= c #x014a) (<= c #x0177)))
 852          (zerop (% c 2))
 853          (set-case-syntax-pair
 854           (decode-char 'ucs c) (decode-char 'ucs (1+ c)) tbl))
 855     (and (>= c #x013a)
 856          (<= c #x0148)
 857          (zerop (% c 2))
 858          (set-case-syntax-pair
 859           (decode-char 'ucs (1- c)) (decode-char 'ucs c) tbl))
 860     (setq c (1+ c)))
 861   (set-case-syntax-pair ?\e$,1 R\e(B ?\e$,1 S\e(B tbl)
 862   (set-case-syntax-pair ?\e$,1 T\e(B ?\e$,1 U\e(B tbl)
 863   (set-case-syntax-pair ?\e$,1 V\e(B ?\e$,1 W\e(B tbl)
 864 ;;;  (set-case-syntax-pair ?\e$,1!8\e(B ?\e,A\7f\e(B tbl)     ; these two have different length!
 865   (set-case-syntax-pair ?\e$,1!9\e(B ?\e$,1!:\e(B tbl)
 866   (set-case-syntax-pair ?\e$,1!;\e(B ?\e$,1!<\e(B tbl)
 867   (set-case-syntax-pair ?\e$,1!=\e(B ?\e$,1!>\e(B tbl)
 868
 869   ;; Latin Extended-B
 870   (set-case-syntax-pair ?\e$,1!A\e(B ?\e$,1#S\e(B tbl)
 871   (set-case-syntax-pair ?\e$,1!B\e(B ?\e$,1!C\e(B tbl)
 872   (set-case-syntax-pair ?\e$,1!D\e(B ?\e$,1!E\e(B tbl)
 873   (set-case-syntax-pair ?\e$,1!F\e(B ?\e$,1#T\e(B tbl)
 874   (set-case-syntax-pair ?\e$,1!G\e(B ?\e$,1!H\e(B tbl)
 875   (set-case-syntax-pair ?\e$,1!I\e(B ?\e$,1#V\e(B tbl)
 876   (set-case-syntax-pair ?\e$,1!J\e(B ?\e$,1#W\e(B tbl)
 877   (set-case-syntax-pair ?\e$,1!K\e(B ?\e$,1!L\e(B tbl)
 878   (set-case-syntax-pair ?\e$,1!N\e(B ?\e$,1"=\e(B tbl)
 879   (set-case-syntax-pair ?\e$,1!O\e(B ?\e$,1#Y\e(B tbl)
 880   (set-case-syntax-pair ?\e$,1!P\e(B ?\e$,1#[\e(B tbl)
 881   (set-case-syntax-pair ?\e$,1!Q\e(B ?\e$,1!R\e(B tbl)
 882   (set-case-syntax-pair ?\e$,1!S\e(B ?\e$,1#`\e(B tbl)
 883   (set-case-syntax-pair ?\e$,1!T\e(B ?\e$,1#c\e(B tbl)
 884   (set-case-syntax-pair ?\e$,1!V\e(B ?\e$,1#i\e(B tbl)
 885   (set-case-syntax-pair ?\e$,1!W\e(B ?\e$,1#h\e(B tbl)
 886   (set-case-syntax-pair ?\e$,1!X\e(B ?\e$,1!Y\e(B tbl)
 887   (set-case-syntax-pair ?\e$,1!\\e(B ?\e$,1#o\e(B tbl)
 888   (set-case-syntax-pair ?\e$,1!]\e(B ?\e$,1#r\e(B tbl)
 889   (set-case-syntax-pair ?\e$,1!_\e(B ?\e$,1#u\e(B tbl)
 890   (set-case-syntax-pair ?\e$,1!`\e(B ?\e$,1!a\e(B tbl)
 891   (set-case-syntax-pair ?\e$,1!b\e(B ?\e$,1!c\e(B tbl)
 892   (set-case-syntax-pair ?\e$,1!d\e(B ?\e$,1!e\e(B tbl)
 893   (set-case-syntax-pair ?\e$,1!f\e(B ?\e$,1$ \e(B tbl)
 894   (set-case-syntax-pair ?\e$,1!g\e(B ?\e$,1!h\e(B tbl)
 895   (set-case-syntax-pair ?\e$,1!i\e(B ?\e$,1$#\e(B tbl)
 896   (set-case-syntax-pair ?\e$,1!l\e(B ?\e$,1!m\e(B tbl)
 897   (set-case-syntax-pair ?\e$,1!n\e(B ?\e$,1$(\e(B tbl)
 898   (set-case-syntax-pair ?\e$,1!o\e(B ?\e$,1!p\e(B tbl)
 899   (set-case-syntax-pair ?\e$,1!q\e(B ?\e$,1$*\e(B tbl)
 900   (set-case-syntax-pair ?\e$,1!r\e(B ?\e$,1$+\e(B tbl)
 901   (set-case-syntax-pair ?\e$,1!s\e(B ?\e$,1!t\e(B tbl)
 902   (set-case-syntax-pair ?\e$,1!u\e(B ?\e$,1!v\e(B tbl)
 903   (set-case-syntax-pair ?\e$,1!w\e(B ?\e$,1$2\e(B tbl)
 904   (set-case-syntax-pair ?\e$,1!x\e(B ?\e$,1!y\e(B tbl)
 905   (set-case-syntax-pair ?\e$,1!|\e(B ?\e$,1!}\e(B tbl)
 906   (set-case-syntax-pair ?\e$,1"$\e(B ?\e$,1"&\e(B tbl)
 907   (set-case-syntax-pair ?\e$,1"%\e(B ?\e$,1"&\e(B tbl)
 908   (set-case-syntax-pair ?\e$,1"'\e(B ?\e$,1")\e(B tbl)
 909   (set-case-syntax-pair ?\e$,1"(\e(B ?\e$,1")\e(B tbl)
 910   (set-case-syntax-pair ?\e$,1"*\e(B ?\e$,1",\e(B tbl)
 911   (set-case-syntax-pair ?\e$,1"+\e(B ?\e$,1",\e(B tbl)
 912   (set-case-syntax-pair ?\e$,1"-\e(B ?\e$,1".\e(B tbl)
 913   (set-case-syntax-pair ?\e$,1"/\e(B ?\e$,1"0\e(B tbl)
 914   (set-case-syntax-pair ?\e$,1"1\e(B ?\e$,1"2\e(B tbl)
 915   (set-case-syntax-pair ?\e$,1"3\e(B ?\e$,1"4\e(B tbl)
 916   (set-case-syntax-pair ?\e$,1"5\e(B ?\e$,1"6\e(B tbl)
 917   (set-case-syntax-pair ?\e$,1"7\e(B ?\e$,1"8\e(B tbl)
 918   (set-case-syntax-pair ?\e$,1"9\e(B ?\e$,1":\e(B tbl)
 919   (set-case-syntax-pair ?\e$,1";\e(B ?\e$,1"<\e(B tbl)
 920   (set-case-syntax-pair ?\e$,1">\e(B ?\e$,1"?\e(B tbl)
 921   (set-case-syntax-pair ?\e$,1"@\e(B ?\e$,1"A\e(B tbl)
 922   (set-case-syntax-pair ?\e$,1"B\e(B ?\e$,1"C\e(B tbl)
 923   (set-case-syntax-pair ?\e$,1"D\e(B ?\e$,1"E\e(B tbl)
 924   (set-case-syntax-pair ?\e$,1"F\e(B ?\e$,1"G\e(B tbl)
 925   (set-case-syntax-pair ?\e$,1"H\e(B ?\e$,1"I\e(B tbl)
 926   (set-case-syntax-pair ?\e$,1"J\e(B ?\e$,1"K\e(B tbl)
 927   (set-case-syntax-pair ?\e$,1"L\e(B ?\e$,1"M\e(B tbl)
 928   (set-case-syntax-pair ?\e$,1"N\e(B ?\e$,1"O\e(B tbl)
 929   ;; 01F0; F; 006A 030C; # LATIN SMALL LETTER J WITH CARON
 930   (set-case-syntax-pair ?\e$,1"Q\e(B ?\e$,1"S\e(B tbl)
 931   (set-case-syntax-pair ?\e$,1"R\e(B ?\e$,1"S\e(B tbl)
 932   (set-case-syntax-pair ?\e$,1"T\e(B ?\e$,1"U\e(B tbl)
 933   (set-case-syntax-pair ?\e$,1"V\e(B ?\e$,1!U\e(B tbl)
 934   (set-case-syntax-pair ?\e$,1"W\e(B ?\e$,1!\7f\e(B tbl)
 935   (set-case-syntax-pair ?\e$,1"X\e(B ?\e$,1"Y\e(B tbl)
 936   (set-case-syntax-pair ?\e$,1"Z\e(B ?\e$,1"[\e(B tbl)
 937   (set-case-syntax-pair ?\e$,1"\\e(B ?\e$,1"]\e(B tbl)
 938   (set-case-syntax-pair ?\e$,1"^\e(B ?\e$,1"_\e(B tbl)
 939   (set-case-syntax-pair ?\e$,1"`\e(B ?\e$,1"a\e(B tbl)
 940   (set-case-syntax-pair ?\e$,1"b\e(B ?\e$,1"c\e(B tbl)
 941   (set-case-syntax-pair ?\e$,1"d\e(B ?\e$,1"e\e(B tbl)
 942   (set-case-syntax-pair ?\e$,1"f\e(B ?\e$,1"g\e(B tbl)
 943   (set-case-syntax-pair ?\e$,1"h\e(B ?\e$,1"i\e(B tbl)
 944   (set-case-syntax-pair ?\e$,1"j\e(B ?\e$,1"k\e(B tbl)
 945   (set-case-syntax-pair ?\e$,1"l\e(B ?\e$,1"m\e(B tbl)
 946   (set-case-syntax-pair ?\e$,1"n\e(B ?\e$,1"o\e(B tbl)
 947   (set-case-syntax-pair ?\e$,1"p\e(B ?\e$,1"q\e(B tbl)
 948   (set-case-syntax-pair ?\e$,1"r\e(B ?\e$,1"s\e(B tbl)
 949   (set-case-syntax-pair ?\e$,1"t\e(B ?\e$,1"u\e(B tbl)
 950   (set-case-syntax-pair ?\e$,1"v\e(B ?\e$,1"w\e(B tbl)
 951   (set-case-syntax-pair ?\e$,1"x\e(B ?\e$,1"y\e(B tbl)
 952   (set-case-syntax-pair ?\e$,1"z\e(B ?\e$,1"{\e(B tbl)
 953   (set-case-syntax-pair ?\e$,1"|\e(B ?\e$,1"}\e(B tbl)
 954   (set-case-syntax-pair ?\e$,1"~\e(B ?\e$,1"\7f\e(B tbl)
 955   (set-case-syntax-pair ?\e$,1#"\e(B ?\e$,1##\e(B tbl)
 956   (set-case-syntax-pair ?\e$,1#$\e(B ?\e$,1#%\e(B tbl)
 957   (set-case-syntax-pair ?\e$,1#&\e(B ?\e$,1#'\e(B tbl)
 958   (set-case-syntax-pair ?\e$,1#(\e(B ?\e$,1#)\e(B tbl)
 959   (set-case-syntax-pair ?\e$,1#*\e(B ?\e$,1#+\e(B tbl)
 960   (set-case-syntax-pair ?\e$,1#,\e(B ?\e$,1#-\e(B tbl)
 961   (set-case-syntax-pair ?\e$,1#.\e(B ?\e$,1#/\e(B tbl)
 962   (set-case-syntax-pair ?\e$,1#0\e(B ?\e$,1#1\e(B tbl)
 963   (set-case-syntax-pair ?\e$,1#2\e(B ?\e$,1#3\e(B tbl)
 964
 965   ;; Latin Extended Additional
 966   (setq c #x1e00)
 967   (while (<= c #x1ef9)
 968     (modify-category-entry (decode-char 'ucs c) ?l)
 969     (and (zerop (% c 2))
 970          (or (<= c #x1e94) (>= c #x1ea0))
 971          (set-case-syntax-pair
 972           (decode-char 'ucs c) (decode-char 'ucs (1+ c)) tbl))
 973     (setq c (1+ c)))
 974
 975   ;; Greek
 976   (setq c #x0370)
 977   (while (<= c #x03ff)
 978     (modify-category-entry (decode-char 'ucs c) ?g)
 979     (if (or (and (>= c #x0391) (<= c #x03a1))
 980             (and (>= c #x03a3) (<= c #x03ab)))
 981         (set-case-syntax-pair
 982          (decode-char 'ucs c) (decode-char 'ucs (+ c 32)) tbl))
 983     (and (>= c #x03da)
 984          (<= c #x03ee)
 985          (zerop (% c 2))
 986          (set-case-syntax-pair
 987           (decode-char 'ucs c) (decode-char 'ucs (1+ c)) tbl))
 988     (setq c (1+ c)))
 989   (set-case-syntax-pair ?\e$,1&f\e(B ?\e$,1',\e(B tbl)
 990   (set-case-syntax-pair ?\e$,1&h\e(B ?\e$,1'-\e(B tbl)
 991   (set-case-syntax-pair ?\e$,1&i\e(B ?\e$,1'.\e(B tbl)
 992   (set-case-syntax-pair ?\e$,1&j\e(B ?\e$,1'/\e(B tbl)
 993   (set-case-syntax-pair ?\e$,1&l\e(B ?\e$,1'L\e(B tbl)
 994   (set-case-syntax-pair ?\e$,1&n\e(B ?\e$,1'M\e(B tbl)
 995   (set-case-syntax-pair ?\e$,1&o\e(B ?\e$,1'N\e(B tbl)
 996
 997   ;; Armenian
 998   (setq c #x531)
 999   (while (<= c #x556)
1000     (set-case-syntax-pair (decode-char 'ucs c)
1001                           (decode-char 'ucs (+ c #x30)) tbl)
1002     (setq c (1+ c)))
1003
1004   ;; Greek Extended
1005   (setq c #x1f00)
1006   (while (<= c #x1fff)
1007     (modify-category-entry (decode-char 'ucs c) ?g)
1008     (and (<= (logand c #x000f) 7)
1009          (<= c #x1fa7)
1010          (not (memq c '(#x1f50 #x1f52 #x1f54 #x1f56)))
1011          (/= (logand c #x00f0) 7)
1012          (set-case-syntax-pair
1013           (decode-char 'ucs (+ c 8)) (decode-char 'ucs c) tbl))
1014     (setq c (1+ c)))
1015   (set-case-syntax-pair ?\e$,1qx\e(B ?\e$,1qp\e(B tbl)
1016   (set-case-syntax-pair ?\e$,1qy\e(B ?\e$,1qq\e(B tbl)
1017   (set-case-syntax-pair ?\e$,1qz\e(B ?\e$,1q0\e(B tbl)
1018   (set-case-syntax-pair ?\e$,1q{\e(B ?\e$,1q1\e(B tbl)
1019   (set-case-syntax-pair ?\e$,1q|\e(B ?\e$,1qs\e(B tbl)
1020   (set-case-syntax-pair ?\e$,1r(\e(B ?\e$,1q2\e(B tbl)
1021   (set-case-syntax-pair ?\e$,1r)\e(B ?\e$,1q3\e(B tbl)
1022   (set-case-syntax-pair ?\e$,1r*\e(B ?\e$,1q4\e(B tbl)
1023   (set-case-syntax-pair ?\e$,1r+\e(B ?\e$,1q5\e(B tbl)
1024   (set-case-syntax-pair ?\e$,1r,\e(B ?\e$,1r#\e(B tbl)
1025   (set-case-syntax-pair ?\e$,1r8\e(B ?\e$,1r0\e(B tbl)
1026   (set-case-syntax-pair ?\e$,1r9\e(B ?\e$,1r1\e(B tbl)
1027   (set-case-syntax-pair ?\e$,1r:\e(B ?\e$,1q6\e(B tbl)
1028   (set-case-syntax-pair ?\e$,1r;\e(B ?\e$,1q7\e(B tbl)
1029   (set-case-syntax-pair ?\e$,1rH\e(B ?\e$,1r@\e(B tbl)
1030   (set-case-syntax-pair ?\e$,1rI\e(B ?\e$,1rA\e(B tbl)
1031   (set-case-syntax-pair ?\e$,1rJ\e(B ?\e$,1q:\e(B tbl)
1032   (set-case-syntax-pair ?\e$,1rK\e(B ?\e$,1q;\e(B tbl)
1033   (set-case-syntax-pair ?\e$,1rL\e(B ?\e$,1rE\e(B tbl)
1034   (set-case-syntax-pair ?\e$,1rX\e(B ?\e$,1q8\e(B tbl)
1035   (set-case-syntax-pair ?\e$,1rY\e(B ?\e$,1q9\e(B tbl)
1036   (set-case-syntax-pair ?\e$,1rZ\e(B ?\e$,1q<\e(B tbl)
1037   (set-case-syntax-pair ?\e$,1r[\e(B ?\e$,1q=\e(B tbl)
1038   (set-case-syntax-pair ?\e$,1r\\e(B ?\e$,1rS\e(B tbl)
1039
1040   ;; cyrillic
1041   (setq c #x0400)
1042   (while (<= c #x04ff)
1043     (modify-category-entry (decode-char 'ucs c) ?y)
1044     (and (>= c #x0400)
1045          (<= c #x040f)
1046          (set-case-syntax-pair
1047           (decode-char 'ucs c) (decode-char 'ucs (+ c 80)) tbl))
1048     (and (>= c #x0410)
1049          (<= c #x042f)
1050          (set-case-syntax-pair
1051           (decode-char 'ucs c) (decode-char 'ucs (+ c 32)) tbl))
1052     (and (zerop (% c 2))
1053          (or (and (>= c #x0460) (<= c #x0480))
1054              (and (>= c #x048c) (<= c #x04be))
1055              (and (>= c #x04d0) (<= c #x04f4)))
1056          (set-case-syntax-pair
1057           (decode-char 'ucs c) (decode-char 'ucs (1+ c)) tbl))
1058     (setq c (1+ c)))
1059   (set-case-syntax-pair ?\e$,1*!\e(B ?\e$,1*"\e(B tbl)
1060   (set-case-syntax-pair ?\e$,1*#\e(B ?\e$,1*$\e(B tbl)
1061   (set-case-syntax-pair ?\e$,1*'\e(B ?\e$,1*(\e(B tbl)
1062   (set-case-syntax-pair ?\e$,1*+\e(B ?\e$,1*,\e(B tbl)
1063   (set-case-syntax-pair ?\e$,1*X\e(B ?\e$,1*Y\e(B tbl)
1064
1065   ;; general punctuation
1066   (setq c #x2000)
1067   (while (<= c #x200b)
1068     (set-case-syntax c " " tbl)
1069     (setq c (1+ c)))
1070   (setq c #x2010)
1071   (while (<= c #x2027)
1072     (set-case-syntax c "_" tbl)
1073     (setq c (1+ c)))
1074
1075   ;; Roman numerals
1076   (setq c #x2160)
1077   (while (<= c #x216f)
1078     (set-case-syntax-pair (decode-char 'ucs c)
1079                           (decode-char 'ucs (+ c #x10)) tbl)
1080     (setq c (1+ c)))
1081
1082   ;; Circled Latin
1083   (setq c #x24b6)
1084   (while (<= c #x24cf)
1085     (set-case-syntax-pair (decode-char 'ucs c)
1086                           (decode-char 'ucs (+ c 26)) tbl)
1087     (modify-category-entry (decode-char 'ucs c) ?l)
1088     (modify-category-entry (decode-char 'ucs (+ c 26)) ?l)
1089     (setq c (1+ c)))
1090
1091   ;; Fullwidth Latin
1092   (setq c #xff21)
1093   (while (<= c #xff3a)
1094     (set-case-syntax-pair (decode-char 'ucs c)
1095                           (decode-char 'ucs (+ c #x20)) tbl)
1096     (modify-category-entry (decode-char 'ucs c) ?l)
1097     (modify-category-entry (decode-char 'ucs (+ c #x20)) ?l)
1098     (setq c (1+ c)))
1099
1100   ;; Ohm, Kelvin, Angstrom
1101   (set-case-syntax-pair ?\e$,1uf\e(B ?\e$,1'I\e(B tbl)
1102 ;;;  These mess up the case conversion of k and \e,Ae\e(B.
1103 ;;;  (set-case-syntax-pair ?\e$,1uj\e(B ?k tbl)
1104 ;;;  (set-case-syntax-pair ?\e$,1uk\e(B ?\e,Ae\e(B tbl)
1105
1106   ;; Combining diacritics
1107   (setq c #x300)
1108   (while (<= c #x362)
1109     (modify-category-entry (decode-char 'ucs c) ?^)
1110     (setq c (1+ c)))
1111
1112   ;; Combining marks
1113   (setq c #x20d0)
1114   (while (<= c #x20e3)
1115     (modify-category-entry (decode-char 'ucs c) ?^)
1116     (setq c (1+ c)))
1117
1118   ;; Fixme: syntax for symbols &c
1119   )
1120 \f
1121 ;;; Setting word boundary.
1122
1123 (setq word-combining-categories
1124       '((?l . ?l)))
1125
1126 (setq word-separating-categories        ;  (2-byte character sets)
1127       '((?A . ?K)                       ; Alpha numeric - Katakana
1128         (?A . ?C)                       ; Alpha numeric - Chinese
1129         (?H . ?A)                       ; Hiragana - Alpha numeric
1130         (?H . ?K)                       ; Hiragana - Katakana
1131         (?H . ?C)                       ; Hiragana - Chinese
1132         (?K . ?A)                       ; Katakana - Alpha numeric
1133         (?K . ?C)                       ; Katakana - Chinese
1134         (?C . ?A)                       ; Chinese - Alpha numeric
1135         (?C . ?K)                       ; Chinese - Katakana
1136         ))
1137
1138 \f
1139 ;; For each character set, put the information of the most proper
1140 ;; coding system to encode it by `preferred-coding-system' property.
1141
1142 (let ((l '((latin-iso8859-1     . iso-latin-1)
1143            (latin-iso8859-2     . iso-latin-2)
1144            (latin-iso8859-3     . iso-latin-3)
1145            (latin-iso8859-4     . iso-latin-4)
1146            (thai-tis620         . thai-tis620)
1147            (greek-iso8859-7     . greek-iso-8bit)
1148            (arabic-iso8859-6    . iso-2022-7bit)
1149            (hebrew-iso8859-8    . hebrew-iso-8bit)
1150            (katakana-jisx0201   . japanese-shift-jis)
1151            (latin-jisx0201      . japanese-shift-jis)
1152            (cyrillic-iso8859-5  . cyrillic-iso-8bit)
1153            (latin-iso8859-9     . iso-latin-5)
1154            (japanese-jisx0208-1978 . iso-2022-jp)
1155            (chinese-gb2312      . cn-gb-2312)
1156            (japanese-jisx0208   . iso-2022-jp)
1157            (korean-ksc5601      . iso-2022-kr)
1158            (japanese-jisx0212   . iso-2022-jp)
1159            (chinese-cns11643-1  . iso-2022-cn)
1160            (chinese-cns11643-2  . iso-2022-cn)
1161            (chinese-big5-1      . chinese-big5)
1162            (chinese-big5-2      . chinese-big5)
1163            (chinese-sisheng     . iso-2022-7bit)
1164            (ipa                 . iso-2022-7bit)
1165            (vietnamese-viscii-lower . vietnamese-viscii)
1166            (vietnamese-viscii-upper . vietnamese-viscii)
1167            (arabic-digit        . iso-2022-7bit)
1168            (arabic-1-column     . iso-2022-7bit)
1169            (ascii-right-to-left . iso-2022-7bit)
1170            (lao                 . lao)
1171            (arabic-2-column     . iso-2022-7bit)
1172            (indian-is13194      . devanagari)
1173            (indian-glyph        . devanagari)
1174            (tibetan-1-column    . tibetan)
1175            (ethiopic            . iso-2022-7bit)
1176            (chinese-cns11643-3  . iso-2022-cn)
1177            (chinese-cns11643-4  . iso-2022-cn)
1178            (chinese-cns11643-5  . iso-2022-cn)
1179            (chinese-cns11643-6  . iso-2022-cn)
1180            (chinese-cns11643-7  . iso-2022-cn)
1181            (indian-2-column     . devanagari)
1182            (tibetan             . tibetan)
1183            (latin-iso8859-14    . iso-latin-8)
1184            (latin-iso8859-15    . iso-latin-9))))
1185   (while l
1186     (put-charset-property (car (car l)) 'preferred-coding-system (cdr (car l)))
1187     (setq l (cdr l))))
1188
1189 \f
1190 ;; Setup auto-fill-chars for charsets that should invoke auto-filling.
1191 ;; SPACE and NEWLINE are already set.  Also put `nospace-between-words'
1192 ;; property on the charsets.
1193 (let ((l '(katakana-jisx0201
1194            japanese-jisx0208 japanese-jisx0212
1195            chinese-gb2312 chinese-big5-1 chinese-big5-2)))
1196   (while l
1197     (aset auto-fill-chars (make-char (car l)) t)
1198     (put-charset-property (car l) 'nospace-between-words t)
1199     (setq l (cdr l))))
1200
1201 ;;; Local Variables:
1202 ;;; coding: iso-2022-7bit
1203 ;;; End:
1204
1205 ;;; characters.el ends here