Mercurial > hg > xemacs-beta
annotate lisp/mule/viet-util.el @ 5938:699264ac20f1 cygwin
tls.h (why?) and skip-ascii.h added
author | Henry Thompson <ht@markup.co.uk> |
---|---|
date | Wed, 08 Dec 2021 18:43:18 +0000 |
parents | ac37a5f7e5be |
children |
rev | line source |
---|---|
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 | |
5402
308d34e9f07d
Changed bulk of GPLv2 or later files identified by script
Mats Lidell <matsl@xemacs.org>
parents:
4072
diff
changeset
|
11 ;; XEmacs is free software: you can redistribute it and/or modify it |
308d34e9f07d
Changed bulk of GPLv2 or later files identified by script
Mats Lidell <matsl@xemacs.org>
parents:
4072
diff
changeset
|
12 ;; under the terms of the GNU General Public License as published by the |
308d34e9f07d
Changed bulk of GPLv2 or later files identified by script
Mats Lidell <matsl@xemacs.org>
parents:
4072
diff
changeset
|
13 ;; Free Software Foundation, either version 3 of the License, or (at your |
308d34e9f07d
Changed bulk of GPLv2 or later files identified by script
Mats Lidell <matsl@xemacs.org>
parents:
4072
diff
changeset
|
14 ;; option) any later version. |
771 | 15 |
5402
308d34e9f07d
Changed bulk of GPLv2 or later files identified by script
Mats Lidell <matsl@xemacs.org>
parents:
4072
diff
changeset
|
16 ;; XEmacs is distributed in the hope that it will be useful, but WITHOUT |
308d34e9f07d
Changed bulk of GPLv2 or later files identified by script
Mats Lidell <matsl@xemacs.org>
parents:
4072
diff
changeset
|
17 ;; ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or |
308d34e9f07d
Changed bulk of GPLv2 or later files identified by script
Mats Lidell <matsl@xemacs.org>
parents:
4072
diff
changeset
|
18 ;; FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License |
308d34e9f07d
Changed bulk of GPLv2 or later files identified by script
Mats Lidell <matsl@xemacs.org>
parents:
4072
diff
changeset
|
19 ;; for more details. |
771 | 20 |
21 ;; You should have received a copy of the GNU General Public License | |
5402
308d34e9f07d
Changed bulk of GPLv2 or later files identified by script
Mats Lidell <matsl@xemacs.org>
parents:
4072
diff
changeset
|
22 ;; along with XEmacs. If not, see <http://www.gnu.org/licenses/>. |
771 | 23 |
778 | 24 ;;; Synched up with: Emacs 21.1 (language/viet-util.el). |
771 | 25 |
26 ;;; Commentary: | |
27 | |
28 ;; Vietnamese uses ASCII characters and additional 134 unique | |
29 ;; characters (these are Latin alphabets with various diacritical and | |
30 ;; tone marks). As far as I know, Vietnamese now has 4 different ways | |
31 ;; for representing these characters: VISCII, VSCII, VIQR, and | |
32 ;; Unicode. VISCII and VSCII are simple 1-byte code which assigns 134 | |
33 ;; unique characters in control-code area (0x00..0x1F) and right half | |
34 ;; area (0x80..0xFF). VIQR is a menmonic encoding specification | |
35 ;; representing diacritical marks by following ASCII characters. | |
36 | |
37 ;;; Code: | |
38 | |
39 ;;;###autoload | |
40 (defun viet-encode-viscii-char (char) | |
41 "Return VISCII character code of CHAR if appropriate." | |
4072 | 42 (check-argument-type #'characterp char) |
43 (if (eq char ?~) | |
44 char | |
45 (setq char (encode-coding-string char 'viscii)) | |
5366
f00192e1cd49
Examining the result of #'length: `eql', not `=', it's better style & cheaper
Aidan Kehoe <kehoea@parhasard.net>
parents:
4072
diff
changeset
|
46 (if (and (eql 1 (length char)) |
f00192e1cd49
Examining the result of #'length: `eql', not `=', it's better style & cheaper
Aidan Kehoe <kehoea@parhasard.net>
parents:
4072
diff
changeset
|
47 (not (eql (aref char 0) ?~))) |
4072 | 48 (aref char 0)))) |
771 | 49 |
50 ;; VIQR is a menmonic encoding specification for Vietnamese. | |
51 ;; It represents diacritical marks by ASCII characters as follows: | |
52 ;; ------------+----------+-------- | |
53 ;; mark | mnemonic | example | |
54 ;; ------------+----------+--------- | |
55 ;; breve | ( | a( -> ,1e(B | |
56 ;; circumflex | ^ | a^ -> ,1b(B | |
57 ;; horn | + | o+ -> ,1=(B | |
58 ;; ------------+----------+--------- | |
59 ;; acute | ' | a' -> ,1a(B | |
60 ;; grave | ` | a` -> ,1`(B | |
61 ;; hook above | ? | a? -> ,1d(B | |
62 ;; tilde | ~ | a~ -> ,1c(B | |
63 ;; dot below | . | a. -> ,1U(B | |
64 ;; ------------+----------+--------- | |
65 ;; d bar | dd | dd -> ,1p(B | |
66 ;; ------------+----------+--------- | |
67 | |
68 (defvar viet-viqr-alist | |
69 '(;; lowercase | |
70 (?,1!(B . "a('") ; 161 | |
71 (?,1"(B . "a(`") ; 162 | |
72 (?,1#(B . "a(.") ; 163 | |
73 (?,1$(B . "a^'") ; 164 | |
74 (?,1%(B . "a^`") ; 165 | |
75 (?,1&(B . "a^?") ; 166 | |
76 (?,1'(B . "a^.") ; 167 | |
77 (?,1((B . "e~") ; 168 | |
78 (?,1)(B . "e.") ; 169 | |
79 (?,1*(B . "e^'") ; 170 | |
80 (?,1+(B . "e^`") ; 171 | |
81 (?,1,(B . "e^?") ; 172 | |
82 (?,1-(B . "e^~") ; 173 | |
83 (?,1.(B . "e^.") ; 174 | |
84 (?,1/(B . "o^'") ; 175 | |
85 (?,10(B . "o^`") ; 176 | |
86 (?,11(B . "o^?") ; 177 | |
87 (?,12(B . "o^~") ; 178 | |
88 (?,15(B . "o^.") ; 181 | |
89 (?,16(B . "o+`") ; 182 | |
90 (?,17(B . "o+?") ; 183 | |
91 (?,18(B . "i.") ; 184 | |
92 (?,1=(B . "o+") ; 189 | |
93 (?,1>(B . "o+'") ; 190 | |
94 (?,1F(B . "a(?") ; 198 | |
95 (?,1G(B . "a(~") ; 199 | |
96 (?,1O(B . "y`") ; 207 | |
97 (?,1Q(B . "u+'") ; 209 | |
98 (?,1U(B . "a.") ; 213 | |
99 (?,1V(B . "y?") ; 214 | |
100 (?,1W(B . "u+`") ; 215 | |
101 (?,1X(B . "u+?") ; 216 | |
102 (?,1[(B . "y~") ; 219 | |
103 (?,1\(B . "y.") ; 220 | |
104 (?,1^(B . "o+~") ; 222 | |
105 (?,1_(B . "u+") ; 223 | |
106 (?,1`(B . "a`") ; 224 | |
107 (?,1a(B . "a'") ; 225 | |
108 (?,1b(B . "a^") ; 226 | |
109 (?,1c(B . "a~") ; 227 | |
110 (?,1d(B . "a?") ; 228 | |
111 (?,1e(B . "a(") ; 229 | |
112 (?,1f(B . "u+~") ; 230 | |
113 (?,1g(B . "a^~") ; 231 | |
114 (?,1h(B . "e`") ; 232 | |
115 (?,1i(B . "e'") ; 233 | |
116 (?,1j(B . "e^") ; 234 | |
117 (?,1k(B . "e?") ; 235 | |
118 (?,1l(B . "i`") ; 236 | |
119 (?,1m(B . "i'") ; 237 | |
120 (?,1n(B . "i~") ; 238 | |
121 (?,1o(B . "i?") ; 239 | |
122 (?,1p(B . "dd") ; 240 | |
123 (?,1q(B . "u+.") ; 241 | |
124 (?,1r(B . "o`") ; 242 | |
125 (?,1s(B . "o'") ; 243 | |
126 (?,1t(B . "o^") ; 244 | |
127 (?,1u(B . "o~") ; 245 | |
128 (?,1v(B . "o?") ; 246 | |
129 (?,1w(B . "o.") ; 247 | |
130 (?,1x(B . "u.") ; 248 | |
131 (?,1y(B . "u`") ; 249 | |
132 (?,1z(B . "u'") ; 250 | |
133 (?,1{(B . "u~") ; 251 | |
134 (?,1|(B . "u?") ; 252 | |
135 (?,1}(B . "y'") ; 253 | |
136 (?,1~(B . "o+.") ; 254 | |
137 | |
138 ;; upper case | |
139 (?,2!(B . "A('") ; 161 | |
140 (?,2"(B . "A(`") ; 162 | |
141 (?,2#(B . "A(.") ; 163 | |
142 (?,2$(B . "A^'") ; 164 | |
143 (?,2%(B . "A^`") ; 165 | |
144 (?,2&(B . "A^?") ; 166 | |
145 (?,2'(B . "A^.") ; 167 | |
146 (?,2((B . "E~") ; 168 | |
147 (?,2)(B . "E.") ; 169 | |
148 (?,2*(B . "E^'") ; 170 | |
149 (?,2+(B . "E^`") ; 171 | |
150 (?,2,(B . "E^?") ; 172 | |
151 (?,2-(B . "E^~") ; 173 | |
152 (?,2.(B . "E^.") ; 174 | |
153 (?,2/(B . "O^'") ; 175 | |
154 (?,20(B . "O^`") ; 176 | |
155 (?,21(B . "O^?") ; 177 | |
156 (?,22(B . "O^~") ; 178 | |
157 (?,25(B . "O^.") ; 181 | |
158 (?,26(B . "O+`") ; 182 | |
159 (?,27(B . "O+?") ; 183 | |
160 (?,28(B . "I.") ; 184 | |
161 (?,2=(B . "O+") ; 189 | |
162 (?,2>(B . "O+'") ; 190 | |
163 (?,2F(B . "A(?") ; 198 | |
164 (?,2G(B . "A(~") ; 199 | |
165 (?,2O(B . "Y`") ; 207 | |
166 (?,2Q(B . "U+'") ; 209 | |
167 (?,2U(B . "A.") ; 213 | |
168 (?,2V(B . "Y?") ; 214 | |
169 (?,2W(B . "U+`") ; 215 | |
170 (?,2X(B . "U+?") ; 216 | |
171 (?,2[(B . "Y~") ; 219 | |
172 (?,2\(B . "Y.") ; 220 | |
173 (?,2^(B . "O+~") ; 222 | |
174 (?,2_(B . "U+") ; 223 | |
175 (?,2`(B . "A`") ; 224 | |
176 (?,2a(B . "A'") ; 225 | |
177 (?,2b(B . "A^") ; 226 | |
178 (?,2c(B . "A~") ; 227 | |
179 (?,2d(B . "A?") ; 228 | |
180 (?,2e(B . "A(") ; 229 | |
181 (?,2f(B . "U+~") ; 230 | |
182 (?,2g(B . "A^~") ; 231 | |
183 (?,2h(B . "E`") ; 232 | |
184 (?,2i(B . "E'") ; 233 | |
185 (?,2j(B . "E^") ; 234 | |
186 (?,2k(B . "E?") ; 235 | |
187 (?,2l(B . "I`") ; 236 | |
188 (?,2m(B . "I'") ; 237 | |
189 (?,2n(B . "I~") ; 238 | |
190 (?,2o(B . "I?") ; 239 | |
191 (?,2p(B . "DD") ; 240 | |
192 (?,2p(B . "dD") ; 240 | |
193 (?,2p(B . "Dd") ; 240 | |
194 (?,2q(B . "U+.") ; 241 | |
195 (?,2r(B . "O`") ; 242 | |
196 (?,2s(B . "O'") ; 243 | |
197 (?,2t(B . "O^") ; 244 | |
198 (?,2u(B . "O~") ; 245 | |
199 (?,2v(B . "O?") ; 246 | |
200 (?,2w(B . "O.") ; 247 | |
201 (?,2x(B . "U.") ; 248 | |
202 (?,2y(B . "U`") ; 249 | |
203 (?,2z(B . "U'") ; 250 | |
204 (?,2{(B . "U~") ; 251 | |
205 (?,2|(B . "U?") ; 252 | |
206 (?,2}(B . "Y'") ; 253 | |
207 (?,2~(B . "O+.") ; 254 | |
208 | |
209 ;; escape from composition | |
210 (?\( . "\\(") ; breve (left parenthesis) | |
211 (?^ . "\\^") ; circumflex (caret) | |
212 (?+ . "\\+") ; horn (plus sign) | |
213 (?' . "\\'") ; acute (apostrophe) | |
214 (?` . "\\`") ; grave (backquote) | |
215 (?? . "\\?") ; hook above (question mark) | |
216 (?~ . "\\~") ; tilde (tilde) | |
217 (?. . "\\.") ; dot below (period) | |
218 (?d . "\\d") ; d-bar (d) | |
219 (?\\ . "\\\\") ; literal backslash | |
220 ) | |
221 "Alist of Vietnamese characters vs corresponding `VIQR' string.") | |
222 | |
223 ;; Regular expression matching single Vietnamese character represented | |
224 ;; by VIQR. | |
225 (defconst viqr-regexp | |
226 "[aeiouyAEIOUY]\\([(^+]?['`?~.]\\|[(^+]\\)\\|[Dd][Dd]") | |
227 | |
228 ;;;###autoload | |
229 (defun viet-decode-viqr-region (from to) | |
1196 | 230 "Convert `VIQR' mnemonics of the current region to Vietnamese characters. |
771 | 231 When called from a program, expects two arguments, |
232 positions (integers or markers) specifying the stretch of the region." | |
233 (interactive "r") | |
234 (save-restriction | |
235 (narrow-to-region from to) | |
236 (goto-char (point-min)) | |
237 (while (re-search-forward viqr-regexp nil t) | |
238 (let* ((viqr (buffer-substring (match-beginning 0) (match-end 0))) | |
239 (ch (car (rassoc viqr viet-viqr-alist)))) | |
240 (if ch | |
241 (progn | |
242 (delete-region (match-beginning 0) (match-end 0)) | |
243 (insert ch))))))) | |
244 | |
245 ;;;###autoload | |
246 (defun viet-decode-viqr-buffer () | |
1196 | 247 "Convert `VIQR' mnemonics of the current buffer to Vietnamese characters." |
771 | 248 (interactive) |
249 (viet-decode-viqr-region (point-min) (point-max))) | |
250 | |
251 ;;;###autoload | |
252 (defun viet-encode-viqr-region (from to) | |
1196 | 253 "Convert Vietnamese characters of the current region to `VIQR' mnemonics. |
771 | 254 When called from a program, expects two arguments, |
255 positions (integers or markers) specifying the stretch of the region." | |
256 (interactive "r") | |
257 (save-restriction | |
258 (narrow-to-region from to) | |
259 (goto-char (point-min)) | |
260 (while (re-search-forward "\\cv" nil t) | |
261 (let* ((ch (preceding-char)) | |
262 (viqr (cdr (assq ch viet-viqr-alist)))) | |
263 (if viqr | |
264 (progn | |
265 (delete-char -1) | |
266 (insert viqr))))))) | |
267 | |
268 ;;;###autoload | |
269 (defun viet-encode-viqr-buffer () | |
1196 | 270 "Convert Vietnamese characters of the current buffer to `VIQR' mnemonics." |
771 | 271 (interactive) |
272 (viet-encode-viqr-region (point-min) (point-max))) | |
273 | |
274 ;;;###autoload | |
275 (defun viqr-post-read-conversion (len) | |
276 (save-excursion | |
277 (save-restriction | |
278 (narrow-to-region (point) (+ (point) len)) | |
279 (let ((buffer-modified-p (buffer-modified-p))) | |
280 (viet-decode-viqr-region (point-min) (point-max)) | |
281 (set-buffer-modified-p buffer-modified-p) | |
282 (- (point-max) (point-min)))))) | |
283 | |
284 ;;;###autoload | |
285 (defun viqr-pre-write-conversion (from to) | |
286 (let ((old-buf (current-buffer))) | |
287 (set-buffer (generate-new-buffer " *temp*")) | |
288 (if (stringp from) | |
289 (insert from) | |
290 (insert-buffer-substring old-buf from to)) | |
291 (viet-encode-viqr-region (point-min) (point-max)) | |
292 ;; Should return nil as annotations. | |
293 nil)) | |
294 | |
295 ;;; | |
296 (provide 'viet-util) | |
297 | |
298 ;;; viet-util.el ends here |