778
+ − 1 ;;; viet-util.el --- utilities for Vietnamese -*- coding: iso-2022-7bit; -*-
771
+ − 2
+ − 3 ;; Copyright (C) 1995 Electrotechnical Laboratory, JAPAN.
788
+ − 4 ;; Copyright (C) 2002 Ben Wing.
771
+ − 5 ;; Licensed to the Free Software Foundation.
+ − 6
+ − 7 ;; Keywords: mule, multilingual, Vietnamese
+ − 8
+ − 9 ;; This file is part of XEmacs.
+ − 10
+ − 11 ;; XEmacs is free software; you can redistribute it and/or modify it
+ − 12 ;; under the terms of the GNU General Public License as published by
+ − 13 ;; the Free Software Foundation; either version 2, or (at your option)
+ − 14 ;; any later version.
+ − 15
+ − 16 ;; XEmacs is distributed in the hope that it will be useful, but
+ − 17 ;; WITHOUT ANY WARRANTY; without even the implied warranty of
+ − 18 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ − 19 ;; General Public License for more details.
+ − 20
+ − 21 ;; You should have received a copy of the GNU General Public License
+ − 22 ;; along with XEmacs; see the file COPYING. If not, write to the Free
+ − 23 ;; Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA
+ − 24 ;; 02111-1307, USA.
+ − 25
778
+ − 26 ;;; Synched up with: Emacs 21.1 (language/viet-util.el).
771
+ − 27
+ − 28 ;;; Commentary:
+ − 29
+ − 30 ;; Vietnamese uses ASCII characters and additional 134 unique
+ − 31 ;; characters (these are Latin alphabets with various diacritical and
+ − 32 ;; tone marks). As far as I know, Vietnamese now has 4 different ways
+ − 33 ;; for representing these characters: VISCII, VSCII, VIQR, and
+ − 34 ;; Unicode. VISCII and VSCII are simple 1-byte code which assigns 134
+ − 35 ;; unique characters in control-code area (0x00..0x1F) and right half
+ − 36 ;; area (0x80..0xFF). VIQR is a menmonic encoding specification
+ − 37 ;; representing diacritical marks by following ASCII characters.
+ − 38
+ − 39 ;;; Code:
+ − 40
+ − 41 ;;;###autoload
+ − 42 (defun viet-encode-viscii-char (char)
+ − 43 "Return VISCII character code of CHAR if appropriate."
4072
+ − 44 (check-argument-type #'characterp char)
+ − 45 (if (eq char ?~)
+ − 46 char
+ − 47 (setq char (encode-coding-string char 'viscii))
+ − 48 (if (and (= 1 (length char))
+ − 49 (not (eq (aref char 0) ?~)))
+ − 50 (aref char 0))))
771
+ − 51
+ − 52 ;; VIQR is a menmonic encoding specification for Vietnamese.
+ − 53 ;; It represents diacritical marks by ASCII characters as follows:
+ − 54 ;; ------------+----------+--------
+ − 55 ;; mark | mnemonic | example
+ − 56 ;; ------------+----------+---------
+ − 57 ;; breve | ( | a( -> ,1e(B
+ − 58 ;; circumflex | ^ | a^ -> ,1b(B
+ − 59 ;; horn | + | o+ -> ,1=(B
+ − 60 ;; ------------+----------+---------
+ − 61 ;; acute | ' | a' -> ,1a(B
+ − 62 ;; grave | ` | a` -> ,1`(B
+ − 63 ;; hook above | ? | a? -> ,1d(B
+ − 64 ;; tilde | ~ | a~ -> ,1c(B
+ − 65 ;; dot below | . | a. -> ,1U(B
+ − 66 ;; ------------+----------+---------
+ − 67 ;; d bar | dd | dd -> ,1p(B
+ − 68 ;; ------------+----------+---------
+ − 69
+ − 70 (defvar viet-viqr-alist
+ − 71 '(;; lowercase
+ − 72 (?,1!(B . "a('") ; 161
+ − 73 (?,1"(B . "a(`") ; 162
+ − 74 (?,1#(B . "a(.") ; 163
+ − 75 (?,1$(B . "a^'") ; 164
+ − 76 (?,1%(B . "a^`") ; 165
+ − 77 (?,1&(B . "a^?") ; 166
+ − 78 (?,1'(B . "a^.") ; 167
+ − 79 (?,1((B . "e~") ; 168
+ − 80 (?,1)(B . "e.") ; 169
+ − 81 (?,1*(B . "e^'") ; 170
+ − 82 (?,1+(B . "e^`") ; 171
+ − 83 (?,1,(B . "e^?") ; 172
+ − 84 (?,1-(B . "e^~") ; 173
+ − 85 (?,1.(B . "e^.") ; 174
+ − 86 (?,1/(B . "o^'") ; 175
+ − 87 (?,10(B . "o^`") ; 176
+ − 88 (?,11(B . "o^?") ; 177
+ − 89 (?,12(B . "o^~") ; 178
+ − 90 (?,15(B . "o^.") ; 181
+ − 91 (?,16(B . "o+`") ; 182
+ − 92 (?,17(B . "o+?") ; 183
+ − 93 (?,18(B . "i.") ; 184
+ − 94 (?,1=(B . "o+") ; 189
+ − 95 (?,1>(B . "o+'") ; 190
+ − 96 (?,1F(B . "a(?") ; 198
+ − 97 (?,1G(B . "a(~") ; 199
+ − 98 (?,1O(B . "y`") ; 207
+ − 99 (?,1Q(B . "u+'") ; 209
+ − 100 (?,1U(B . "a.") ; 213
+ − 101 (?,1V(B . "y?") ; 214
+ − 102 (?,1W(B . "u+`") ; 215
+ − 103 (?,1X(B . "u+?") ; 216
+ − 104 (?,1[(B . "y~") ; 219
+ − 105 (?,1\(B . "y.") ; 220
+ − 106 (?,1^(B . "o+~") ; 222
+ − 107 (?,1_(B . "u+") ; 223
+ − 108 (?,1`(B . "a`") ; 224
+ − 109 (?,1a(B . "a'") ; 225
+ − 110 (?,1b(B . "a^") ; 226
+ − 111 (?,1c(B . "a~") ; 227
+ − 112 (?,1d(B . "a?") ; 228
+ − 113 (?,1e(B . "a(") ; 229
+ − 114 (?,1f(B . "u+~") ; 230
+ − 115 (?,1g(B . "a^~") ; 231
+ − 116 (?,1h(B . "e`") ; 232
+ − 117 (?,1i(B . "e'") ; 233
+ − 118 (?,1j(B . "e^") ; 234
+ − 119 (?,1k(B . "e?") ; 235
+ − 120 (?,1l(B . "i`") ; 236
+ − 121 (?,1m(B . "i'") ; 237
+ − 122 (?,1n(B . "i~") ; 238
+ − 123 (?,1o(B . "i?") ; 239
+ − 124 (?,1p(B . "dd") ; 240
+ − 125 (?,1q(B . "u+.") ; 241
+ − 126 (?,1r(B . "o`") ; 242
+ − 127 (?,1s(B . "o'") ; 243
+ − 128 (?,1t(B . "o^") ; 244
+ − 129 (?,1u(B . "o~") ; 245
+ − 130 (?,1v(B . "o?") ; 246
+ − 131 (?,1w(B . "o.") ; 247
+ − 132 (?,1x(B . "u.") ; 248
+ − 133 (?,1y(B . "u`") ; 249
+ − 134 (?,1z(B . "u'") ; 250
+ − 135 (?,1{(B . "u~") ; 251
+ − 136 (?,1|(B . "u?") ; 252
+ − 137 (?,1}(B . "y'") ; 253
+ − 138 (?,1~(B . "o+.") ; 254
+ − 139
+ − 140 ;; upper case
+ − 141 (?,2!(B . "A('") ; 161
+ − 142 (?,2"(B . "A(`") ; 162
+ − 143 (?,2#(B . "A(.") ; 163
+ − 144 (?,2$(B . "A^'") ; 164
+ − 145 (?,2%(B . "A^`") ; 165
+ − 146 (?,2&(B . "A^?") ; 166
+ − 147 (?,2'(B . "A^.") ; 167
+ − 148 (?,2((B . "E~") ; 168
+ − 149 (?,2)(B . "E.") ; 169
+ − 150 (?,2*(B . "E^'") ; 170
+ − 151 (?,2+(B . "E^`") ; 171
+ − 152 (?,2,(B . "E^?") ; 172
+ − 153 (?,2-(B . "E^~") ; 173
+ − 154 (?,2.(B . "E^.") ; 174
+ − 155 (?,2/(B . "O^'") ; 175
+ − 156 (?,20(B . "O^`") ; 176
+ − 157 (?,21(B . "O^?") ; 177
+ − 158 (?,22(B . "O^~") ; 178
+ − 159 (?,25(B . "O^.") ; 181
+ − 160 (?,26(B . "O+`") ; 182
+ − 161 (?,27(B . "O+?") ; 183
+ − 162 (?,28(B . "I.") ; 184
+ − 163 (?,2=(B . "O+") ; 189
+ − 164 (?,2>(B . "O+'") ; 190
+ − 165 (?,2F(B . "A(?") ; 198
+ − 166 (?,2G(B . "A(~") ; 199
+ − 167 (?,2O(B . "Y`") ; 207
+ − 168 (?,2Q(B . "U+'") ; 209
+ − 169 (?,2U(B . "A.") ; 213
+ − 170 (?,2V(B . "Y?") ; 214
+ − 171 (?,2W(B . "U+`") ; 215
+ − 172 (?,2X(B . "U+?") ; 216
+ − 173 (?,2[(B . "Y~") ; 219
+ − 174 (?,2\(B . "Y.") ; 220
+ − 175 (?,2^(B . "O+~") ; 222
+ − 176 (?,2_(B . "U+") ; 223
+ − 177 (?,2`(B . "A`") ; 224
+ − 178 (?,2a(B . "A'") ; 225
+ − 179 (?,2b(B . "A^") ; 226
+ − 180 (?,2c(B . "A~") ; 227
+ − 181 (?,2d(B . "A?") ; 228
+ − 182 (?,2e(B . "A(") ; 229
+ − 183 (?,2f(B . "U+~") ; 230
+ − 184 (?,2g(B . "A^~") ; 231
+ − 185 (?,2h(B . "E`") ; 232
+ − 186 (?,2i(B . "E'") ; 233
+ − 187 (?,2j(B . "E^") ; 234
+ − 188 (?,2k(B . "E?") ; 235
+ − 189 (?,2l(B . "I`") ; 236
+ − 190 (?,2m(B . "I'") ; 237
+ − 191 (?,2n(B . "I~") ; 238
+ − 192 (?,2o(B . "I?") ; 239
+ − 193 (?,2p(B . "DD") ; 240
+ − 194 (?,2p(B . "dD") ; 240
+ − 195 (?,2p(B . "Dd") ; 240
+ − 196 (?,2q(B . "U+.") ; 241
+ − 197 (?,2r(B . "O`") ; 242
+ − 198 (?,2s(B . "O'") ; 243
+ − 199 (?,2t(B . "O^") ; 244
+ − 200 (?,2u(B . "O~") ; 245
+ − 201 (?,2v(B . "O?") ; 246
+ − 202 (?,2w(B . "O.") ; 247
+ − 203 (?,2x(B . "U.") ; 248
+ − 204 (?,2y(B . "U`") ; 249
+ − 205 (?,2z(B . "U'") ; 250
+ − 206 (?,2{(B . "U~") ; 251
+ − 207 (?,2|(B . "U?") ; 252
+ − 208 (?,2}(B . "Y'") ; 253
+ − 209 (?,2~(B . "O+.") ; 254
+ − 210
+ − 211 ;; escape from composition
+ − 212 (?\( . "\\(") ; breve (left parenthesis)
+ − 213 (?^ . "\\^") ; circumflex (caret)
+ − 214 (?+ . "\\+") ; horn (plus sign)
+ − 215 (?' . "\\'") ; acute (apostrophe)
+ − 216 (?` . "\\`") ; grave (backquote)
+ − 217 (?? . "\\?") ; hook above (question mark)
+ − 218 (?~ . "\\~") ; tilde (tilde)
+ − 219 (?. . "\\.") ; dot below (period)
+ − 220 (?d . "\\d") ; d-bar (d)
+ − 221 (?\\ . "\\\\") ; literal backslash
+ − 222 )
+ − 223 "Alist of Vietnamese characters vs corresponding `VIQR' string.")
+ − 224
+ − 225 ;; Regular expression matching single Vietnamese character represented
+ − 226 ;; by VIQR.
+ − 227 (defconst viqr-regexp
+ − 228 "[aeiouyAEIOUY]\\([(^+]?['`?~.]\\|[(^+]\\)\\|[Dd][Dd]")
+ − 229
+ − 230 ;;;###autoload
+ − 231 (defun viet-decode-viqr-region (from to)
1196
+ − 232 "Convert `VIQR' mnemonics of the current region to Vietnamese characters.
771
+ − 233 When called from a program, expects two arguments,
+ − 234 positions (integers or markers) specifying the stretch of the region."
+ − 235 (interactive "r")
+ − 236 (save-restriction
+ − 237 (narrow-to-region from to)
+ − 238 (goto-char (point-min))
+ − 239 (while (re-search-forward viqr-regexp nil t)
+ − 240 (let* ((viqr (buffer-substring (match-beginning 0) (match-end 0)))
+ − 241 (ch (car (rassoc viqr viet-viqr-alist))))
+ − 242 (if ch
+ − 243 (progn
+ − 244 (delete-region (match-beginning 0) (match-end 0))
+ − 245 (insert ch)))))))
+ − 246
+ − 247 ;;;###autoload
+ − 248 (defun viet-decode-viqr-buffer ()
1196
+ − 249 "Convert `VIQR' mnemonics of the current buffer to Vietnamese characters."
771
+ − 250 (interactive)
+ − 251 (viet-decode-viqr-region (point-min) (point-max)))
+ − 252
+ − 253 ;;;###autoload
+ − 254 (defun viet-encode-viqr-region (from to)
1196
+ − 255 "Convert Vietnamese characters of the current region to `VIQR' mnemonics.
771
+ − 256 When called from a program, expects two arguments,
+ − 257 positions (integers or markers) specifying the stretch of the region."
+ − 258 (interactive "r")
+ − 259 (save-restriction
+ − 260 (narrow-to-region from to)
+ − 261 (goto-char (point-min))
+ − 262 (while (re-search-forward "\\cv" nil t)
+ − 263 (let* ((ch (preceding-char))
+ − 264 (viqr (cdr (assq ch viet-viqr-alist))))
+ − 265 (if viqr
+ − 266 (progn
+ − 267 (delete-char -1)
+ − 268 (insert viqr)))))))
+ − 269
+ − 270 ;;;###autoload
+ − 271 (defun viet-encode-viqr-buffer ()
1196
+ − 272 "Convert Vietnamese characters of the current buffer to `VIQR' mnemonics."
771
+ − 273 (interactive)
+ − 274 (viet-encode-viqr-region (point-min) (point-max)))
+ − 275
+ − 276 ;;;###autoload
+ − 277 (defun viqr-post-read-conversion (len)
+ − 278 (save-excursion
+ − 279 (save-restriction
+ − 280 (narrow-to-region (point) (+ (point) len))
+ − 281 (let ((buffer-modified-p (buffer-modified-p)))
+ − 282 (viet-decode-viqr-region (point-min) (point-max))
+ − 283 (set-buffer-modified-p buffer-modified-p)
+ − 284 (- (point-max) (point-min))))))
+ − 285
+ − 286 ;;;###autoload
+ − 287 (defun viqr-pre-write-conversion (from to)
+ − 288 (let ((old-buf (current-buffer)))
+ − 289 (set-buffer (generate-new-buffer " *temp*"))
+ − 290 (if (stringp from)
+ − 291 (insert from)
+ − 292 (insert-buffer-substring old-buf from to))
+ − 293 (viet-encode-viqr-region (point-min) (point-max))
+ − 294 ;; Should return nil as annotations.
+ − 295 nil))
+ − 296
+ − 297 ;;;
+ − 298 (provide 'viet-util)
+ − 299
+ − 300 ;;; viet-util.el ends here
