Commit | Line | Data |
---|---|---|
caff3c0a | 1 | ;;; mule-cmds.el --- commands for multilingual environment -*-coding: iso-2022-7bit -*- |
9ee5b744 | 2 | |
d4877ac1 | 3 | ;; Copyright (C) 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005, |
e3fe4da0 | 4 | ;; 2006, 2007, 2008 Free Software Foundation, Inc. |
7976eda0 | 5 | ;; Copyright (C) 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004, |
e3fe4da0 | 6 | ;; 2005, 2006, 2007, 2008 |
2fd125a3 KH |
7 | ;; National Institute of Advanced Industrial Science and Technology (AIST) |
8 | ;; Registration Number H14PRO021 | |
8f924df7 | 9 | ;; Copyright (C) 2003 |
97941b05 KH |
10 | ;; National Institute of Advanced Industrial Science and Technology (AIST) |
11 | ;; Registration Number H13PRO009 | |
4ed46869 | 12 | |
1f547b92 | 13 | ;; Keywords: mule, i18n |
4ed46869 KH |
14 | |
15 | ;; This file is part of GNU Emacs. | |
16 | ||
17 | ;; GNU Emacs is free software; you can redistribute it and/or modify | |
18 | ;; it under the terms of the GNU General Public License as published by | |
d7142f3e | 19 | ;; the Free Software Foundation; either version 3, or (at your option) |
4ed46869 KH |
20 | ;; any later version. |
21 | ||
22 | ;; GNU Emacs is distributed in the hope that it will be useful, | |
23 | ;; but WITHOUT ANY WARRANTY; without even the implied warranty of | |
24 | ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
25 | ;; GNU General Public License for more details. | |
26 | ||
27 | ;; You should have received a copy of the GNU General Public License | |
369314dc | 28 | ;; along with GNU Emacs; see the file COPYING. If not, write to the |
3a35cf56 LK |
29 | ;; Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor, |
30 | ;; Boston, MA 02110-1301, USA. | |
4ed46869 | 31 | |
60370d40 PJ |
32 | ;;; Commentary: |
33 | ||
4ed46869 KH |
34 | ;;; Code: |
35 | ||
2375e71a KH |
36 | (eval-when-compile |
37 | (defvar dos-codepage) | |
38 | (autoload 'widget-value "wid-edit")) | |
cda74479 | 39 | |
4a91d930 JB |
40 | (defvar mac-system-coding-system) |
41 | (defvar mac-system-locale) | |
42 | ||
4ed46869 KH |
43 | ;;; MULE related key bindings and menus. |
44 | ||
0709d285 | 45 | (defvar mule-keymap (make-sparse-keymap) |
33d17698 | 46 | "Keymap for Mule (Multilingual environment) specific commands.") |
4ed46869 | 47 | |
8f81f784 | 48 | ;; Keep "C-x C-m ..." for mule specific commands. |
0709d285 | 49 | (define-key ctl-x-map "\C-m" mule-keymap) |
ef8a8c8c | 50 | |
4ed46869 | 51 | (define-key mule-keymap "f" 'set-buffer-file-coding-system) |
d80dee05 | 52 | (define-key mule-keymap "r" 'revert-buffer-with-coding-system) |
2308d0d7 | 53 | (define-key mule-keymap "F" 'set-file-name-coding-system) |
4ed46869 | 54 | (define-key mule-keymap "t" 'set-terminal-coding-system) |
15b3e511 KH |
55 | (define-key mule-keymap "k" 'set-keyboard-coding-system) |
56 | (define-key mule-keymap "p" 'set-buffer-process-coding-system) | |
7624ebb9 KH |
57 | (define-key mule-keymap "x" 'set-selection-coding-system) |
58 | (define-key mule-keymap "X" 'set-next-selection-coding-system) | |
8b784951 | 59 | (define-key mule-keymap "\C-\\" 'set-input-method) |
15b3e511 | 60 | (define-key mule-keymap "c" 'universal-coding-system-argument) |
b4fba33f | 61 | (define-key mule-keymap "l" 'set-language-environment) |
4ed46869 | 62 | |
538d88fb EZ |
63 | (defvar mule-menu-keymap |
64 | (make-sparse-keymap "Mule (Multilingual Environment)") | |
33d17698 | 65 | "Keymap for Mule (Multilingual environment) menu specific commands.") |
15b3e511 | 66 | |
dcad02bc EZ |
67 | (defvar describe-language-environment-map |
68 | (make-sparse-keymap "Describe Language Environment")) | |
15b3e511 | 69 | |
dcad02bc EZ |
70 | (defvar setup-language-environment-map |
71 | (make-sparse-keymap "Set Language Environment")) | |
15b3e511 | 72 | |
dcad02bc EZ |
73 | (defvar set-coding-system-map |
74 | (make-sparse-keymap "Set Coding System")) | |
15b3e511 | 75 | |
15b3e511 | 76 | (define-key-after mule-menu-keymap [set-language-environment] |
ad9f6125 | 77 | (list 'menu-item "Set Language Environment" setup-language-environment-map)) |
15b3e511 KH |
78 | (define-key-after mule-menu-keymap [separator-mule] |
79 | '("--") | |
80 | t) | |
81 | (define-key-after mule-menu-keymap [toggle-input-method] | |
538d88fb | 82 | '(menu-item "Toggle Input Method" toggle-input-method) |
15b3e511 | 83 | t) |
8b784951 | 84 | (define-key-after mule-menu-keymap [set-input-method] |
538d88fb | 85 | '(menu-item "Select Input Method..." set-input-method) |
15b3e511 | 86 | t) |
cda74479 DL |
87 | (define-key-after mule-menu-keymap [describe-input-method] |
88 | '(menu-item "Describe Input Method" describe-input-method)) | |
15b3e511 KH |
89 | (define-key-after mule-menu-keymap [separator-input-method] |
90 | '("--") | |
91 | t) | |
15b3e511 | 92 | (define-key-after mule-menu-keymap [set-various-coding-system] |
cda74479 | 93 | (list 'menu-item "Set Coding Systems" set-coding-system-map |
be9650bc | 94 | :enable 'default-enable-multibyte-characters)) |
538d88fb EZ |
95 | (define-key-after mule-menu-keymap [view-hello-file] |
96 | '(menu-item "Show Multi-lingual Text" view-hello-file | |
97 | :enable (file-readable-p | |
98 | (expand-file-name "HELLO" data-directory)) | |
99 | :help "Display file which says HELLO in many languages") | |
15b3e511 KH |
100 | t) |
101 | (define-key-after mule-menu-keymap [separator-coding-system] | |
102 | '("--") | |
103 | t) | |
538d88fb EZ |
104 | (define-key-after mule-menu-keymap [describe-language-environment] |
105 | (list 'menu-item "Describe Language Environment" | |
106 | describe-language-environment-map | |
cda74479 | 107 | :help "Show multilingual settings for a specific language") |
15b3e511 | 108 | t) |
538d88fb EZ |
109 | (define-key-after mule-menu-keymap [describe-input-method] |
110 | '(menu-item "Describe Input Method..." describe-input-method | |
cda74479 | 111 | :help "Keyboard layout for a specific input method") |
538d88fb EZ |
112 | t) |
113 | (define-key-after mule-menu-keymap [describe-coding-system] | |
114 | '(menu-item "Describe Coding System..." describe-coding-system) | |
115 | t) | |
cda74479 DL |
116 | (define-key-after mule-menu-keymap [list-character-sets] |
117 | '(menu-item "List Character Sets" list-character-sets | |
118 | :help "Show table of available character sets")) | |
538d88fb EZ |
119 | (define-key-after mule-menu-keymap [mule-diag] |
120 | '(menu-item "Show All of Mule Status" mule-diag | |
121 | :help "Display multilingual environment settings") | |
15b3e511 KH |
122 | t) |
123 | ||
3a151e98 | 124 | (define-key-after set-coding-system-map [universal-coding-system-argument] |
538d88fb EZ |
125 | '(menu-item "For Next Command" universal-coding-system-argument |
126 | :help "Coding system to be used by next command") | |
3a151e98 | 127 | t) |
d80dee05 RS |
128 | (define-key-after set-coding-system-map [separator-1] |
129 | '("--") | |
130 | t) | |
131 | (define-key-after set-coding-system-map [set-buffer-file-coding-system] | |
132 | '(menu-item "For Saving This Buffer" set-buffer-file-coding-system | |
133 | :help "How to encode this buffer when saved") | |
134 | t) | |
135 | (define-key-after set-coding-system-map [revert-buffer-with-coding-system] | |
136 | '(menu-item "For Reverting This File Now" revert-buffer-with-coding-system | |
137 | :enable buffer-file-name | |
138 | :help "Revisit this file immediately using specified coding system") | |
139 | t) | |
2308d0d7 KH |
140 | (define-key-after set-coding-system-map [set-file-name-coding-system] |
141 | '(menu-item "For File Name" set-file-name-coding-system | |
142 | :help "How to decode/encode file names") | |
143 | t) | |
d80dee05 RS |
144 | (define-key-after set-coding-system-map [separator-2] |
145 | '("--") | |
15b3e511 | 146 | t) |
d80dee05 | 147 | |
15b3e511 | 148 | (define-key-after set-coding-system-map [set-keyboard-coding-system] |
538d88fb EZ |
149 | '(menu-item "For Keyboard" set-keyboard-coding-system |
150 | :help "How to decode keyboard input") | |
15b3e511 | 151 | t) |
d80dee05 RS |
152 | (define-key-after set-coding-system-map [set-terminal-coding-system] |
153 | '(menu-item "For Terminal" set-terminal-coding-system | |
dc2be2fa | 154 | :enable (null (memq initial-window-system '(x w32 mac))) |
d80dee05 RS |
155 | :help "How to encode terminal output") |
156 | t) | |
157 | (define-key-after set-coding-system-map [separator-3] | |
158 | '("--") | |
15b3e511 | 159 | t) |
7624ebb9 | 160 | (define-key-after set-coding-system-map [set-selection-coding-system] |
538d88fb EZ |
161 | '(menu-item "For X Selections/Clipboard" set-selection-coding-system |
162 | :visible (display-selections-p) | |
163 | :help "How to en/decode data to/from selection/clipboard") | |
7624ebb9 KH |
164 | t) |
165 | (define-key-after set-coding-system-map [set-next-selection-coding-system] | |
538d88fb EZ |
166 | '(menu-item "For Next X Selection" set-next-selection-coding-system |
167 | :visible (display-selections-p) | |
168 | :help "How to en/decode next selection/clipboard operation") | |
7624ebb9 | 169 | t) |
d80dee05 RS |
170 | (define-key-after set-coding-system-map [set-buffer-process-coding-system] |
171 | '(menu-item "For I/O with Subprocess" set-buffer-process-coding-system | |
172 | :visible (fboundp 'start-process) | |
173 | :enable (get-buffer-process (current-buffer)) | |
174 | :help "How to en/decode I/O from/to subprocess connected to this buffer") | |
175 | t) | |
176 | ||
177 | ||
15b3e511 | 178 | (define-key setup-language-environment-map |
538d88fb | 179 | [Default] '(menu-item "Default" setup-specified-language-environment)) |
4ed46869 | 180 | |
cda74479 DL |
181 | (define-key describe-language-environment-map |
182 | [Default] '(menu-item "Default" describe-specified-language-support)) | |
183 | ||
4ed46869 KH |
184 | ;; This should be a single character key binding because users use it |
185 | ;; very frequently while editing multilingual text. Now we can use | |
186 | ;; only two such keys: "\C-\\" and "\C-^", but the latter is not | |
187 | ;; convenient because it requires shifting on most keyboards. An | |
188 | ;; alternative is "\C-\]" which is now bound to `abort-recursive-edit' | |
189 | ;; but it won't be used that frequently. | |
190 | (define-key global-map "\C-\\" 'toggle-input-method) | |
191 | ||
5f395df3 SM |
192 | ;; This is no good because people often type Shift-SPC |
193 | ;; meaning to type SPC. -- rms. | |
194 | ;; ;; Here's an alternative key binding for X users (Shift-SPACE). | |
195 | ;; (define-key global-map [?\S- ] 'toggle-input-method) | |
b4fba33f | 196 | |
464cc130 KH |
197 | ;;; Mule related hyperlinks. |
198 | (defconst help-xref-mule-regexp-template | |
199 | (purecopy (concat "\\(\\<\\(" | |
200 | "\\(coding system\\)\\|" | |
d0c40faa KH |
201 | "\\(input method\\)\\|" |
202 | "\\(character set\\)\\|" | |
203 | "\\(charset\\)" | |
464cc130 KH |
204 | "\\)\\s-+\\)?" |
205 | ;; Note starting with word-syntax character: | |
206 | "`\\(\\sw\\(\\sw\\|\\s_\\)+\\)'"))) | |
207 | ||
26d87040 | 208 | (defun coding-system-change-eol-conversion (coding-system eol-type) |
caff3c0a | 209 | "Return a coding system which differs from CODING-SYSTEM in EOL conversion. |
26d87040 EZ |
210 | The returned coding system converts end-of-line by EOL-TYPE |
211 | but text as the same way as CODING-SYSTEM. | |
212 | EOL-TYPE should be `unix', `dos', `mac', or nil. | |
213 | If EOL-TYPE is nil, the returned coding system detects | |
214 | how end-of-line is formatted automatically while decoding. | |
215 | ||
216 | EOL-TYPE can be specified by an integer 0, 1, or 2. | |
217 | They means `unix', `dos', and `mac' respectively." | |
218 | (if (symbolp eol-type) | |
219 | (setq eol-type (cond ((eq eol-type 'unix) 0) | |
220 | ((eq eol-type 'dos) 1) | |
221 | ((eq eol-type 'mac) 2) | |
222 | (t eol-type)))) | |
1d5b0c66 CY |
223 | ;; We call `coding-system-base' before `coding-system-eol-type', |
224 | ;; because the coding-system may not be initialized until then. | |
225 | (let* ((base (coding-system-base coding-system)) | |
226 | (orig-eol-type (coding-system-eol-type coding-system))) | |
227 | (cond ((vectorp orig-eol-type) | |
228 | (if (not eol-type) | |
229 | coding-system | |
230 | (aref orig-eol-type eol-type))) | |
231 | ((not eol-type) | |
232 | base) | |
233 | ((= eol-type orig-eol-type) | |
234 | coding-system) | |
235 | ((progn (setq orig-eol-type (coding-system-eol-type base)) | |
236 | (vectorp orig-eol-type)) | |
237 | (aref orig-eol-type eol-type))))) | |
26d87040 EZ |
238 | |
239 | (defun coding-system-change-text-conversion (coding-system coding) | |
240 | "Return a coding system which differs from CODING-SYSTEM in text conversion. | |
241 | The returned coding system converts text by CODING | |
242 | but end-of-line as the same way as CODING-SYSTEM. | |
243 | If CODING is nil, the returned coding system detects | |
244 | how text is formatted automatically while decoding." | |
1d77e15a JR |
245 | (let ((eol-type (coding-system-eol-type coding-system))) |
246 | (coding-system-change-eol-conversion | |
247 | (if coding coding 'undecided) | |
248 | (if (numberp eol-type) (aref [unix dos mac] eol-type))))) | |
26d87040 | 249 | |
4ed46869 | 250 | (defun toggle-enable-multibyte-characters (&optional arg) |
6998e1a1 RS |
251 | "Change whether this buffer uses multibyte characters. |
252 | With arg, use multibyte characters if the arg is positive. | |
253 | ||
254 | Note that this command does not convert the byte contents of | |
255 | the buffer; it only changes the way those bytes are interpreted. | |
256 | In general, therefore, this command *changes* the sequence of | |
257 | characters that the current buffer contains. | |
258 | ||
caff3c0a JB |
259 | We suggest you avoid using this command unless you know what you are |
260 | doing. If you use it by mistake, and the buffer is now displayed | |
6998e1a1 | 261 | wrong, use this command again to toggle back to the right mode." |
4ed46869 | 262 | (interactive "P") |
b7079457 RS |
263 | (let ((new-flag |
264 | (if (null arg) (null enable-multibyte-characters) | |
265 | (> (prefix-numeric-value arg) 0)))) | |
266 | (set-buffer-multibyte new-flag)) | |
4ed46869 KH |
267 | (force-mode-line-update)) |
268 | ||
269 | (defun view-hello-file () | |
270 | "Display the HELLO file which list up many languages and characters." | |
271 | (interactive) | |
8f81f784 KH |
272 | ;; We have to decode the file in any environment. |
273 | (let ((default-enable-multibyte-characters t) | |
95fa03b4 | 274 | (coding-system-for-read 'iso-2022-7bit)) |
544dd975 | 275 | (view-file (expand-file-name "HELLO" data-directory)))) |
4ed46869 | 276 | |
9e9a77a6 | 277 | (defun universal-coding-system-argument (coding-system) |
15b3e511 | 278 | "Execute an I/O command using the specified coding system." |
9e9a77a6 RS |
279 | (interactive |
280 | (let ((default (and buffer-file-coding-system | |
34104362 | 281 | (not (eq (coding-system-type buffer-file-coding-system) |
a0d96cad | 282 | 'undecided)) |
9e9a77a6 RS |
283 | buffer-file-coding-system))) |
284 | (list (read-coding-system | |
285 | (if default | |
5b76833f | 286 | (format "Coding system for following command (default %s): " default) |
9e9a77a6 RS |
287 | "Coding system for following command: ") |
288 | default)))) | |
289 | (let* ((keyseq (read-key-sequence | |
e14a8f4c | 290 | (format "Command to execute with %s:" coding-system))) |
04363179 GM |
291 | (cmd (key-binding keyseq)) |
292 | prefix) | |
293 | ||
294 | (when (eq cmd 'universal-argument) | |
295 | (call-interactively cmd) | |
a1506d29 | 296 | |
04363179 GM |
297 | ;; Process keys bound in `universal-argument-map'. |
298 | (while (progn | |
299 | (setq keyseq (read-key-sequence nil t) | |
300 | cmd (key-binding keyseq t)) | |
301 | (not (eq cmd 'universal-argument-other-key))) | |
302 | (let ((current-prefix-arg prefix-arg) | |
d37ef0f6 | 303 | ;; Have to bind `last-command-char' here so that |
6b61353c | 304 | ;; `digit-argument', for instance, can compute the |
04363179 GM |
305 | ;; prefix arg. |
306 | (last-command-char (aref keyseq 0))) | |
307 | (call-interactively cmd))) | |
308 | ||
1f547b92 | 309 | ;; This is the final call to `universal-argument-other-key', which |
04363179 GM |
310 | ;; set's the final `prefix-arg. |
311 | (let ((current-prefix-arg prefix-arg)) | |
312 | (call-interactively cmd)) | |
a1506d29 | 313 | |
04363179 GM |
314 | ;; Read the command to execute with the given prefix arg. |
315 | (setq prefix prefix-arg | |
316 | keyseq (read-key-sequence nil t) | |
317 | cmd (key-binding keyseq))) | |
318 | ||
15b3e511 | 319 | (let ((coding-system-for-read coding-system) |
04363179 | 320 | (coding-system-for-write coding-system) |
0e9ec609 | 321 | (coding-system-require-warning t) |
04363179 | 322 | (current-prefix-arg prefix)) |
15b3e511 KH |
323 | (message "") |
324 | (call-interactively cmd)))) | |
325 | ||
de94d711 | 326 | (defun set-default-coding-systems (coding-system) |
0c3154d2 | 327 | "Set default value of various coding systems to CODING-SYSTEM. |
387136f6 | 328 | This sets the following coding systems: |
0c3154d2 | 329 | o coding system of a newly created buffer |
8efc03e1 KH |
330 | o default coding system for subprocess I/O |
331 | This also sets the following values: | |
d3e4babd | 332 | o default value used as `file-name-coding-system' for converting file names |
caff3c0a | 333 | if CODING-SYSTEM is ASCII-compatible |
03c35c83 | 334 | o default value for the command `set-terminal-coding-system' (not on MSDOS) |
d3e4babd | 335 | o default value for the command `set-keyboard-coding-system' |
caff3c0a | 336 | if CODING-SYSTEM is ASCII-compatible" |
de94d711 KH |
337 | (check-coding-system coding-system) |
338 | (setq-default buffer-file-coding-system coding-system) | |
6b61353c KH |
339 | (if (fboundp 'ucs-set-table-for-input) |
340 | (dolist (buffer (buffer-list)) | |
341 | (or (local-variable-p 'buffer-file-coding-system buffer) | |
342 | (ucs-set-table-for-input buffer)))) | |
343 | ||
db9aec47 | 344 | (if (eq system-type 'darwin) |
a41118cc | 345 | ;; The file-name coding system on Darwin systems is always utf-8. |
db9aec47 KH |
346 | (setq default-file-name-coding-system 'utf-8) |
347 | (if (and default-enable-multibyte-characters | |
348 | (or (not coding-system) | |
349 | (coding-system-get coding-system 'ascii-compatible-p))) | |
350 | (setq default-file-name-coding-system coding-system))) | |
03c35c83 EZ |
351 | ;; If coding-system is nil, honor that on MS-DOS as well, so |
352 | ;; that they could reset the terminal coding system. | |
353 | (unless (and (eq window-system 'pc) coding-system) | |
354 | (setq default-terminal-coding-system coding-system)) | |
de94d711 | 355 | (setq default-keyboard-coding-system coding-system) |
1d77e15a JR |
356 | ;; Preserve eol-type from existing default-process-coding-systems. |
357 | ;; On non-unix-like systems in particular, these may have been set | |
358 | ;; carefully by the user, or by the startup code, to deal with the | |
359 | ;; users shell appropriately, so should not be altered by changing | |
360 | ;; language environment. | |
361 | (let ((output-coding | |
362 | (coding-system-change-text-conversion | |
363 | (car default-process-coding-system) coding-system)) | |
364 | (input-coding | |
365 | (coding-system-change-text-conversion | |
366 | (cdr default-process-coding-system) coding-system))) | |
367 | (setq default-process-coding-system | |
368 | (cons output-coding input-coding)))) | |
de94d711 | 369 | |
0c3154d2 KH |
370 | (defun prefer-coding-system (coding-system) |
371 | "Add CODING-SYSTEM at the front of the priority list for automatic detection. | |
387136f6 | 372 | This also sets the following coding systems: |
0c3154d2 | 373 | o coding system of a newly created buffer |
8efc03e1 KH |
374 | o default coding system for subprocess I/O |
375 | This also sets the following values: | |
caff3c0a | 376 | o default value used as `file-name-coding-system' for converting file names |
03c35c83 EZ |
377 | o default value for the command `set-terminal-coding-system' (not on MSDOS) |
378 | o default value for the command `set-keyboard-coding-system' | |
379 | ||
bd3ac67e EZ |
380 | If CODING-SYSTEM specifies a certain type of EOL conversion, the coding |
381 | systems set by this function will use that type of EOL conversion. | |
382 | ||
03c35c83 EZ |
383 | This command does not change the default value of terminal coding system |
384 | for MS-DOS terminal, because DOS terminals only support a single coding | |
385 | system, and Emacs automatically sets the default to that coding system at | |
5beaa0d1 KH |
386 | startup. |
387 | ||
c654de1d | 388 | A coding system that requires automatic detection of text |
8f924df7 | 389 | +encoding (e.g. undecided, unix) can't be preferred.." |
0c3154d2 KH |
390 | (interactive "zPrefer coding system: ") |
391 | (if (not (and coding-system (coding-system-p coding-system))) | |
392 | (error "Invalid coding system `%s'" coding-system)) | |
97941b05 KH |
393 | (if (memq (coding-system-type coding-system) '(raw-text undecided)) |
394 | (error "Can't prefer the coding system `%s'" coding-system)) | |
395 | (let ((base (coding-system-base coding-system)) | |
bd3ac67e | 396 | (eol-type (coding-system-eol-type coding-system))) |
97941b05 KH |
397 | (set-coding-system-priority base) |
398 | (and (interactive-p) | |
399 | (or (eq base coding-system) | |
400 | (message "Highest priority is set to %s (base of %s)" | |
401 | base coding-system))) | |
bd3ac67e | 402 | ;; If they asked for specific EOL conversion, honor that. |
6f9dc4fd | 403 | (if (memq eol-type '(0 1 2)) |
97941b05 KH |
404 | (setq base |
405 | (coding-system-change-eol-conversion base eol-type))) | |
406 | (set-default-coding-systems base))) | |
0c3154d2 | 407 | |
b5edd1d1 KH |
408 | (defvar sort-coding-systems-predicate nil |
409 | "If non-nil, a predicate function to sort coding systems. | |
410 | ||
411 | It is called with two coding systems, and should return t if the first | |
412 | one is \"less\" than the second. | |
413 | ||
414 | The function `sort-coding-systems' use it.") | |
415 | ||
416 | (defun sort-coding-systems (codings) | |
417 | "Sort coding system list CODINGS by a priority of each coding system. | |
caff3c0a | 418 | Return the sorted list. CODINGS is modified by side effects. |
b5edd1d1 KH |
419 | |
420 | If a coding system is most preferred, it has the highest priority. | |
caff3c0a JB |
421 | Otherwise, coding systems that correspond to MIME charsets have |
422 | higher priorities. Among them, a coding system included in the | |
423 | `coding-system' key of the current language environment has higher | |
424 | priority. See also the documentation of `language-info-alist'. | |
b5edd1d1 KH |
425 | |
426 | If the variable `sort-coding-systems-predicate' (which see) is | |
caff3c0a | 427 | non-nil, it is used to sort CODINGS instead." |
b5edd1d1 KH |
428 | (if sort-coding-systems-predicate |
429 | (sort codings sort-coding-systems-predicate) | |
8f924df7 KH |
430 | (let* ((from-priority (coding-system-priority-list)) |
431 | (most-preferred (car from-priority)) | |
b5edd1d1 KH |
432 | (lang-preferred (get-language-info current-language-environment |
433 | 'coding-system)) | |
434 | (func (function | |
435 | (lambda (x) | |
436 | (let ((base (coding-system-base x))) | |
da1ebad1 KH |
437 | ;; We calculate the priority number 0..255 by |
438 | ;; using the 8 bits PMMLCEII as this: | |
3ecd3a56 GM |
439 | ;; P: 1 if most preferred. |
440 | ;; MM: greater than 0 if mime-charset. | |
441 | ;; L: 1 if one of the current lang. env.'s codings. | |
442 | ;; C: 1 if one of codings listed in the category list. | |
443 | ;; E: 1 if not XXX-with-esc | |
da1ebad1 KH |
444 | ;; II: if iso-2022 based, 0..3, else 1. |
445 | (logior | |
446 | (lsh (if (eq base most-preferred) 1 0) 7) | |
447 | (lsh | |
8f924df7 | 448 | (let ((mime (coding-system-get base :mime-charset))) |
b439e72a DL |
449 | ;; Prefer coding systems corresponding to a |
450 | ;; MIME charset. | |
b5edd1d1 | 451 | (if mime |
b439e72a DL |
452 | ;; Lower utf-16 priority so that we |
453 | ;; normally prefer utf-8 to it, and put | |
454 | ;; x-ctext below that. | |
de814dd9 KH |
455 | (cond ((string-match "utf-16" |
456 | (symbol-name mime)) | |
da1ebad1 | 457 | 2) |
b439e72a | 458 | ((string-match "^x-" (symbol-name mime)) |
da1ebad1 KH |
459 | 1) |
460 | (t 3)) | |
b5edd1d1 | 461 | 0)) |
da1ebad1 KH |
462 | 5) |
463 | (lsh (if (memq base lang-preferred) 1 0) 4) | |
8f924df7 | 464 | (lsh (if (memq base from-priority) 1 0) 3) |
da1ebad1 KH |
465 | (lsh (if (string-match "-with-esc\\'" |
466 | (symbol-name base)) | |
467 | 0 1) 2) | |
8f924df7 KH |
468 | (if (eq (coding-system-type base) 'iso-2022) |
469 | (let ((category (coding-system-category base))) | |
470 | ;; For ISO based coding systems, prefer | |
471 | ;; one that doesn't use designation nor | |
472 | ;; locking/single shifting. | |
a0d96cad KH |
473 | (cond |
474 | ((or (eq category 'coding-category-iso-8-1) | |
475 | (eq category 'coding-category-iso-8-2)) | |
476 | 2) | |
477 | ((or (eq category 'coding-category-iso-7-tight) | |
478 | (eq category 'coding-category-iso-7)) | |
479 | 1) | |
480 | (t | |
481 | 0))) | |
482 | 1) | |
a6dfc99b | 483 | )))))) |
b5edd1d1 KH |
484 | (sort codings (function (lambda (x y) |
485 | (> (funcall func x) (funcall func y)))))))) | |
54b226f7 | 486 | |
3fc7dfe5 | 487 | (defun find-coding-systems-region (from to) |
54b226f7 | 488 | "Return a list of proper coding systems to encode a text between FROM and TO. |
8f924df7 | 489 | |
d37ef0f6 | 490 | If FROM is a string, find coding systems in that instead of the buffer. |
54b226f7 KH |
491 | All coding systems in the list can safely encode any multibyte characters |
492 | in the text. | |
493 | ||
e8dd0160 | 494 | If the text contains no multibyte characters, return a list of a single |
3fc7dfe5 | 495 | element `undecided'." |
b5edd1d1 KH |
496 | (let ((codings (find-coding-systems-region-internal from to))) |
497 | (if (eq codings t) | |
498 | ;; The text contains only ASCII characters. Any coding | |
499 | ;; systems are safe. | |
500 | '(undecided) | |
501 | ;; We need copy-sequence because sorting will alter the argument. | |
502 | (sort-coding-systems (copy-sequence codings))))) | |
54b226f7 | 503 | |
3fc7dfe5 KH |
504 | (defun find-coding-systems-string (string) |
505 | "Return a list of proper coding systems to encode STRING. | |
506 | All coding systems in the list can safely encode any multibyte characters | |
507 | in STRING. | |
508 | ||
e8dd0160 | 509 | If STRING contains no multibyte characters, return a list of a single |
3fc7dfe5 | 510 | element `undecided'." |
b5edd1d1 | 511 | (find-coding-systems-region string nil)) |
3fc7dfe5 KH |
512 | |
513 | (defun find-coding-systems-for-charsets (charsets) | |
514 | "Return a list of proper coding systems to encode characters of CHARSETS. | |
6053d86a | 515 | CHARSETS is a list of character sets. |
1f547b92 DL |
516 | |
517 | This only finds coding systems of type `charset', whose | |
518 | `:charset-list' property includes all of CHARSETS (plus `ascii' for | |
519 | ascii-compatible coding systems). It was used in older versions of | |
520 | Emacs, but is unlikely to be what you really want now." | |
521 | ;; Deal with aliases. | |
522 | (setq charsets (mapcar (lambda (c) | |
523 | (get-charset-property c :name)) | |
524 | charsets)) | |
b5edd1d1 KH |
525 | (cond ((or (null charsets) |
526 | (and (= (length charsets) 1) | |
527 | (eq 'ascii (car charsets)))) | |
528 | '(undecided)) | |
529 | ((or (memq 'eight-bit-control charsets) | |
530 | (memq 'eight-bit-graphic charsets)) | |
1f547b92 | 531 | '(raw-text utf-8-emacs)) |
b5edd1d1 | 532 | (t |
1f547b92 DL |
533 | (let (codings) |
534 | (dolist (cs (coding-system-list t)) | |
c7b4b466 DL |
535 | (let ((cs-charsets (and (eq (coding-system-type cs) 'charset) |
536 | (coding-system-charset-list cs))) | |
1f547b92 DL |
537 | (charsets charsets)) |
538 | (if (coding-system-get cs :ascii-compatible-p) | |
539 | (add-to-list 'cs-charsets 'ascii)) | |
540 | (if (catch 'ok | |
541 | (when cs-charsets | |
542 | (while charsets | |
543 | (unless (memq (pop charsets) cs-charsets) | |
544 | (throw 'ok nil))) | |
545 | t)) | |
546 | (push cs codings)))) | |
547 | (nreverse codings))))) | |
54b226f7 | 548 | |
51ed58ea KH |
549 | (defun find-multibyte-characters (from to &optional maxcount excludes) |
550 | "Find multibyte characters in the region specified by FROM and TO. | |
551 | If FROM is a string, find multibyte characters in the string. | |
552 | The return value is an alist of the following format: | |
553 | ((CHARSET COUNT CHAR ...) ...) | |
554 | where | |
555 | CHARSET is a character set, | |
556 | COUNT is a number of characters, | |
1f547b92 | 557 | CHARs are the characters found from the character set. |
51ed58ea | 558 | Optional 3rd arg MAXCOUNT limits how many CHARs are put in the above list. |
caff3c0a | 559 | Optional 4th arg EXCLUDES is a list of character sets to be ignored." |
51ed58ea KH |
560 | (let ((chars nil) |
561 | charset char) | |
562 | (if (stringp from) | |
a0d96cad KH |
563 | (if (multibyte-string-p from) |
564 | (let ((idx 0)) | |
565 | (while (setq idx (string-match "[^\000-\177]" from idx)) | |
566 | (setq char (aref from idx) | |
567 | charset (char-charset char)) | |
568 | (unless (memq charset excludes) | |
569 | (let ((slot (assq charset chars))) | |
570 | (if slot | |
571 | (if (not (memq char (nthcdr 2 slot))) | |
572 | (let ((count (nth 1 slot))) | |
573 | (setcar (cdr slot) (1+ count)) | |
574 | (if (or (not maxcount) (< count maxcount)) | |
575 | (nconc slot (list char))))) | |
576 | (setq chars (cons (list charset 1 char) chars))))) | |
577 | (setq idx (1+ idx))))) | |
578 | (if enable-multibyte-characters | |
579 | (save-excursion | |
580 | (goto-char from) | |
581 | (while (re-search-forward "[^\000-\177]" to t) | |
582 | (setq char (preceding-char) | |
583 | charset (char-charset char)) | |
584 | (unless (memq charset excludes) | |
51ed58ea KH |
585 | (let ((slot (assq charset chars))) |
586 | (if slot | |
a0d96cad | 587 | (if (not (member char (nthcdr 2 slot))) |
51ed58ea KH |
588 | (let ((count (nth 1 slot))) |
589 | (setcar (cdr slot) (1+ count)) | |
590 | (if (or (not maxcount) (< count maxcount)) | |
591 | (nconc slot (list char))))) | |
a0d96cad | 592 | (setq chars (cons (list charset 1 char) chars))))))))) |
51ed58ea KH |
593 | (nreverse chars))) |
594 | ||
738746ba KH |
595 | (defun search-unencodable-char (coding-system) |
596 | "Search forward from point for a character that is not encodable. | |
597 | It asks which coding system to check. | |
598 | If such a character is found, set point after that character. | |
599 | Otherwise, don't move point. | |
600 | ||
caff3c0a JB |
601 | When called from a program, the value is the position of the unencodable |
602 | character found, or nil if all characters are encodable." | |
738746ba KH |
603 | (interactive |
604 | (list (let ((default (or buffer-file-coding-system 'us-ascii))) | |
605 | (read-coding-system | |
5b76833f | 606 | (format "Coding-system (default %s): " default) |
738746ba KH |
607 | default)))) |
608 | (let ((pos (unencodable-char-position (point) (point-max) coding-system))) | |
609 | (if pos | |
610 | (goto-char (1+ pos)) | |
611 | (message "All following characters are encodable by %s" coding-system)) | |
612 | pos)) | |
613 | ||
c83c4f60 RS |
614 | (defvar last-coding-system-specified nil |
615 | "Most recent coding system explicitly specified by the user when asked. | |
616 | This variable is set whenever Emacs asks the user which coding system | |
617 | to use in order to write a file. If you set it to nil explicitly, | |
618 | then call `write-region', then afterward this variable will be non-nil | |
619 | only if the user was explicitly asked and specified a coding system.") | |
620 | ||
b5edd1d1 | 621 | (defvar select-safe-coding-system-accept-default-p nil |
fea6b736 | 622 | "If non-nil, a function to control the behavior of coding system selection. |
b5edd1d1 KH |
623 | The meaning is the same as the argument ACCEPT-DEFAULT-P of the |
624 | function `select-safe-coding-system' (which see). This variable | |
625 | overrides that argument.") | |
626 | ||
9ee5b744 SM |
627 | (defun select-safe-coding-system-interactively (from to codings unsafe |
628 | &optional rejected default) | |
629 | "Select interactively a coding system for the region FROM ... TO. | |
630 | FROM can be a string, as in `write-region'. | |
631 | CODINGS is the list of base coding systems known to be safe for this region, | |
632 | typically obtained with `find-coding-systems-region'. | |
633 | UNSAFE is a list of coding systems known to be unsafe for this region. | |
634 | REJECTED is a list of coding systems which were safe but for some reason | |
635 | were not recommended in the particular context. | |
636 | DEFAULT is the coding system to use by default in the query." | |
637 | ;; At first, if some defaults are unsafe, record at most 11 | |
638 | ;; problematic characters and their positions for them by turning | |
639 | ;; (CODING ...) | |
640 | ;; into | |
641 | ;; ((CODING (POS . CHAR) (POS . CHAR) ...) ...) | |
642 | (if unsafe | |
643 | (setq unsafe | |
644 | (mapcar #'(lambda (coding) | |
645 | (cons coding | |
646 | (if (stringp from) | |
647 | (mapcar #'(lambda (pos) | |
648 | (cons pos (aref from pos))) | |
649 | (unencodable-char-position | |
650 | 0 (length from) coding | |
651 | 11 from)) | |
652 | (mapcar #'(lambda (pos) | |
653 | (cons pos (char-after pos))) | |
654 | (unencodable-char-position | |
655 | from to coding 11))))) | |
656 | unsafe))) | |
657 | ||
658 | ;; Change each safe coding system to the corresponding | |
659 | ;; mime-charset name if it is also a coding system. Such a name | |
660 | ;; is more friendly to users. | |
661 | (let ((l codings) | |
662 | mime-charset) | |
663 | (while l | |
a77c22c2 KH |
664 | (setq mime-charset (coding-system-get (car l) :mime-charset)) |
665 | (if (and mime-charset (coding-system-p mime-charset) | |
666 | (coding-system-equal (car l) mime-charset)) | |
9ee5b744 SM |
667 | (setcar l mime-charset)) |
668 | (setq l (cdr l)))) | |
669 | ||
670 | ;; Don't offer variations with locking shift, which you | |
671 | ;; basically never want. | |
672 | (let (l) | |
673 | (dolist (elt codings (setq codings (nreverse l))) | |
674 | (unless (or (eq 'coding-category-iso-7-else | |
675 | (coding-system-category elt)) | |
676 | (eq 'coding-category-iso-8-else | |
677 | (coding-system-category elt))) | |
678 | (push elt l)))) | |
679 | ||
680 | ;; Remove raw-text, emacs-mule and no-conversion unless nothing | |
681 | ;; else is available. | |
682 | (setq codings | |
683 | (or (delq 'raw-text | |
684 | (delq 'emacs-mule | |
685 | (delq 'no-conversion codings))) | |
686 | '(raw-text emacs-mule no-conversion))) | |
687 | ||
688 | (let ((window-configuration (current-window-configuration)) | |
689 | (bufname (buffer-name)) | |
690 | coding-system) | |
691 | (save-excursion | |
692 | ;; If some defaults are unsafe, make sure the offending | |
693 | ;; buffer is displayed. | |
694 | (when (and unsafe (not (stringp from))) | |
695 | (pop-to-buffer bufname) | |
696 | (goto-char (apply 'min (mapcar #'(lambda (x) (car (cadr x))) | |
697 | unsafe)))) | |
698 | ;; Then ask users to select one from CODINGS while showing | |
699 | ;; the reason why none of the defaults are not used. | |
700 | (with-output-to-temp-buffer "*Warning*" | |
701 | (with-current-buffer standard-output | |
702 | (if (and (null rejected) (null unsafe)) | |
703 | (insert "No default coding systems to try for " | |
704 | (if (stringp from) | |
705 | (format "string \"%s\"." from) | |
706 | (format "buffer `%s'." bufname))) | |
707 | (insert | |
708 | "These default coding systems were tried to encode" | |
709 | (if (stringp from) | |
710 | (concat " \"" (if (> (length from) 10) | |
711 | (concat (substring from 0 10) "...\"") | |
712 | (concat from "\""))) | |
713 | (format " text\nin the buffer `%s'" bufname)) | |
714 | ":\n") | |
715 | (let ((pos (point)) | |
716 | (fill-prefix " ")) | |
717 | (dolist (x (append rejected unsafe)) | |
46c04718 | 718 | (princ " ") (princ x)) |
9ee5b744 SM |
719 | (insert "\n") |
720 | (fill-region-as-paragraph pos (point))) | |
721 | (when rejected | |
3fc0b26e EZ |
722 | (insert "These safely encode the text in the buffer, |
723 | but are not recommended for encoding text in this context, | |
9ee5b744 SM |
724 | e.g., for sending an email message.\n ") |
725 | (dolist (x rejected) | |
726 | (princ " ") (princ x)) | |
727 | (insert "\n")) | |
728 | (when unsafe | |
3fc0b26e | 729 | (insert (if rejected "The other coding systems" |
9ee5b744 | 730 | "However, each of them") |
3fc0b26e | 731 | " encountered characters it couldn't encode:\n") |
9ee5b744 | 732 | (dolist (coding unsafe) |
3fc0b26e | 733 | (insert (format " %s cannot encode these:" (car coding))) |
9ee5b744 SM |
734 | (let ((i 0) |
735 | (func1 | |
736 | #'(lambda (bufname pos) | |
737 | (when (buffer-live-p (get-buffer bufname)) | |
738 | (pop-to-buffer bufname) | |
739 | (goto-char pos)))) | |
740 | (func2 | |
741 | #'(lambda (bufname pos coding) | |
742 | (when (buffer-live-p (get-buffer bufname)) | |
743 | (pop-to-buffer bufname) | |
744 | (if (< (point) pos) | |
745 | (goto-char pos) | |
746 | (forward-char 1) | |
747 | (search-unencodable-char coding) | |
748 | (forward-char -1)))))) | |
749 | (dolist (elt (cdr coding)) | |
750 | (insert " ") | |
751 | (if (stringp from) | |
752 | (insert (if (< i 10) (cdr elt) "...")) | |
753 | (if (< i 10) | |
754 | (insert-text-button | |
755 | (cdr elt) | |
756 | :type 'help-xref | |
4e21e5aa | 757 | 'face 'link |
9ee5b744 SM |
758 | 'help-echo |
759 | "mouse-2, RET: jump to this character" | |
760 | 'help-function func1 | |
761 | 'help-args (list bufname (car elt))) | |
762 | (insert-text-button | |
763 | "..." | |
764 | :type 'help-xref | |
4e21e5aa | 765 | 'face 'link |
9ee5b744 SM |
766 | 'help-echo |
767 | "mouse-2, RET: next unencodable character" | |
768 | 'help-function func2 | |
769 | 'help-args (list bufname (car elt) | |
770 | (car coding))))) | |
771 | (setq i (1+ i)))) | |
772 | (insert "\n")) | |
333f3572 | 773 | (insert (substitute-command-keys "\ |
7d03c5b1 | 774 | |
333f3572 KH |
775 | Click on a character (or switch to this window by `\\[other-window]'\n\ |
776 | and select the characters by RET) to jump to the place it appears,\n\ | |
3fc0b26e | 777 | where `\\[universal-argument] \\[what-cursor-position]' will give information about it.\n")))) |
7d03c5b1 | 778 | (insert (substitute-command-keys "\nSelect \ |
3fc0b26e EZ |
779 | one of the safe coding systems listed below,\n\ |
780 | or cancel the writing with \\[keyboard-quit] and edit the buffer\n\ | |
781 | to remove or modify the problematic characters,\n\ | |
782 | or specify any other coding system (and risk losing\n\ | |
783 | the problematic characters).\n\n")) | |
9ee5b744 SM |
784 | (let ((pos (point)) |
785 | (fill-prefix " ")) | |
786 | (dolist (x codings) | |
787 | (princ " ") (princ x)) | |
788 | (insert "\n") | |
7d03c5b1 | 789 | (fill-region-as-paragraph pos (point))))) |
9ee5b744 SM |
790 | |
791 | ;; Read a coding system. | |
792 | (setq coding-system | |
793 | (read-coding-system | |
794 | (format "Select coding system (default %s): " default) | |
795 | default)) | |
796 | (setq last-coding-system-specified coding-system)) | |
797 | ||
798 | (kill-buffer "*Warning*") | |
799 | (set-window-configuration window-configuration) | |
800 | coding-system)) | |
801 | ||
b5edd1d1 | 802 | (defun select-safe-coding-system (from to &optional default-coding-system |
efa2c6d7 | 803 | accept-default-p file) |
d5266ddf KH |
804 | "Ask a user to select a safe coding system from candidates. |
805 | The candidates of coding systems which can safely encode a text | |
b5edd1d1 KH |
806 | between FROM and TO are shown in a popup window. Among them, the most |
807 | proper one is suggested as the default. | |
808 | ||
a985cd2f KH |
809 | The list of `buffer-file-coding-system' of the current buffer, |
810 | the `default-buffer-file-coding-system', and the | |
b5edd1d1 KH |
811 | most preferred coding system (if it corresponds to a MIME charset) is |
812 | treated as the default coding system list. Among them, the first one | |
d37ef0f6 DL |
813 | that safely encodes the text is normally selected silently and |
814 | returned without any user interaction. See also the command | |
815 | `prefer-coding-system'. | |
816 | ||
817 | However, the user is queried if the chosen coding system is | |
29d04c4f | 818 | inconsistent with what would be selected by `find-auto-coding' from |
d37ef0f6 DL |
819 | coding cookies &c. if the contents of the region were read from a |
820 | file. (That could lead to data corruption in a file subsequently | |
821 | re-visited and edited.) | |
54b226f7 | 822 | |
b5edd1d1 KH |
823 | Optional 3rd arg DEFAULT-CODING-SYSTEM specifies a coding system or a |
824 | list of coding systems to be prepended to the default coding system | |
0e9ec609 | 825 | list. However, if DEFAULT-CODING-SYSTEM is a list and the first |
ccb06340 | 826 | element is t, the cdr part is used as the default coding system list, |
a985cd2f KH |
827 | i.e. `buffer-file-coding-system', `default-buffer-file-coding-system', |
828 | and the most preferred coding system are not used. | |
54b226f7 | 829 | |
b5edd1d1 KH |
830 | Optional 4th arg ACCEPT-DEFAULT-P, if non-nil, is a function to |
831 | determine the acceptability of the silently selected coding system. | |
832 | It is called with that coding system, and should return nil if it | |
833 | should not be silently selected and thus user interaction is required. | |
834 | ||
efa2c6d7 RS |
835 | Optional 5th arg FILE is the file name to use for this purpose. |
836 | That is different from `buffer-file-name' when handling `write-region' | |
837 | \(for example). | |
838 | ||
b5edd1d1 KH |
839 | The variable `select-safe-coding-system-accept-default-p', if |
840 | non-nil, overrides ACCEPT-DEFAULT-P. | |
54b226f7 KH |
841 | |
842 | Kludgy feature: if FROM is a string, the string is the target text, | |
843 | and TO is ignored." | |
b5edd1d1 KH |
844 | (if (and default-coding-system |
845 | (not (listp default-coding-system))) | |
846 | (setq default-coding-system (list default-coding-system))) | |
847 | ||
29d04c4f KH |
848 | (let ((no-other-defaults nil) |
849 | auto-cs) | |
d35deeec | 850 | (unless (or (stringp from) find-file-literally) |
29d04c4f KH |
851 | ;; Find an auto-coding that is specified for the the current |
852 | ;; buffer and file from the region FROM and TO. | |
853 | (save-excursion | |
854 | (save-restriction | |
855 | (widen) | |
856 | (goto-char from) | |
857 | (setq auto-cs (find-auto-coding (or file buffer-file-name "") | |
858 | (- to from))) | |
859 | (if auto-cs | |
860 | (if (coding-system-p (car auto-cs)) | |
861 | (setq auto-cs (car auto-cs)) | |
862 | (display-warning | |
863 | :warning | |
864 | (format "\ | |
865 | Invalid coding system `%s' is specified | |
866 | for the current buffer/file by the %s. | |
867 | It is highly recommended to fix it before writing to a file." | |
868 | (car auto-cs) | |
869 | (if (eq (cdr auto-cs) :coding) ":coding tag" | |
870 | (format "variable `%s'" (cdr auto-cs))))) | |
871 | (or (yes-or-no-p "Really proceed with writing? ") | |
872 | (error "Save aborted")) | |
873 | (setq auto-cs nil)))))) | |
874 | ||
0e9ec609 KH |
875 | (if (eq (car default-coding-system) t) |
876 | (setq no-other-defaults t | |
877 | default-coding-system (cdr default-coding-system))) | |
878 | ||
879 | ;; Change elements of the list to (coding . base-coding). | |
880 | (setq default-coding-system | |
881 | (mapcar (function (lambda (x) (cons x (coding-system-base x)))) | |
882 | default-coding-system)) | |
883 | ||
29d04c4f KH |
884 | (if (and auto-cs (not no-other-defaults)) |
885 | ;; If the file has a coding cookie, try to use it before anything | |
886 | ;; else (i.e. before default-coding-system which will typically come | |
887 | ;; from file-coding-system-alist). | |
888 | (let ((base (coding-system-base auto-cs))) | |
889 | (or (memq base '(nil undecided)) | |
890 | (rassq base default-coding-system) | |
891 | (push (cons auto-cs base) default-coding-system)))) | |
892 | ||
0e9ec609 KH |
893 | (unless no-other-defaults |
894 | ;; If buffer-file-coding-system is not nil nor undecided, append it | |
895 | ;; to the defaults. | |
896 | (if buffer-file-coding-system | |
897 | (let ((base (coding-system-base buffer-file-coding-system))) | |
898 | (or (eq base 'undecided) | |
0e9ec609 | 899 | (rassq base default-coding-system) |
ccb06340 KH |
900 | (setq default-coding-system |
901 | (append default-coding-system | |
902 | (list (cons buffer-file-coding-system base))))))) | |
a985cd2f KH |
903 | |
904 | ;; If default-buffer-file-coding-system is not nil nor undecided, | |
905 | ;; append it to the defaults. | |
906 | (if default-buffer-file-coding-system | |
907 | (let ((base (coding-system-base default-buffer-file-coding-system))) | |
908 | (or (eq base 'undecided) | |
909 | (rassq base default-coding-system) | |
ccb06340 KH |
910 | (setq default-coding-system |
911 | (append default-coding-system | |
d35deeec | 912 | (list (cons default-buffer-file-coding-system |
ccb06340 | 913 | base))))))) |
0e9ec609 KH |
914 | |
915 | ;; If the most preferred coding system has the property mime-charset, | |
916 | ;; append it to the defaults. | |
8f924df7 KH |
917 | (let ((preferred (coding-system-priority-list t)) |
918 | base) | |
0e9ec609 KH |
919 | (and (coding-system-p preferred) |
920 | (setq base (coding-system-base preferred)) | |
8f924df7 | 921 | (coding-system-get preferred :mime-charset) |
0e9ec609 | 922 | (not (rassq base default-coding-system)) |
ccb06340 KH |
923 | (setq default-coding-system |
924 | (append default-coding-system | |
925 | (list (cons preferred base))))))) | |
29d04c4f KH |
926 | |
927 | (if select-safe-coding-system-accept-default-p | |
928 | (setq accept-default-p select-safe-coding-system-accept-default-p)) | |
929 | ||
bae7cd08 KH |
930 | ;; Decide the eol-type from the top of the default codings, |
931 | ;; buffer-file-coding-system, or | |
932 | ;; default-buffer-file-coding-system. | |
933 | (if default-coding-system | |
934 | (let ((default-eol-type (coding-system-eol-type | |
935 | (caar default-coding-system)))) | |
936 | (if (and (vectorp default-eol-type) buffer-file-coding-system) | |
d35deeec | 937 | (setq default-eol-type (coding-system-eol-type |
bae7cd08 KH |
938 | buffer-file-coding-system))) |
939 | (if (and (vectorp default-eol-type) default-buffer-file-coding-system) | |
d35deeec | 940 | (setq default-eol-type (coding-system-eol-type |
bae7cd08 KH |
941 | default-buffer-file-coding-system))) |
942 | (if (and default-eol-type (not (vectorp default-eol-type))) | |
943 | (dolist (elt default-coding-system) | |
944 | (setcar elt (coding-system-change-eol-conversion | |
945 | (car elt) default-eol-type)))))) | |
946 | ||
29d04c4f KH |
947 | (let ((codings (find-coding-systems-region from to)) |
948 | (coding-system nil) | |
86c3a9fb | 949 | (tick (if (not (stringp from)) (buffer-modified-tick))) |
29d04c4f | 950 | safe rejected unsafe) |
bae7cd08 KH |
951 | (if (eq (car codings) 'undecided) |
952 | ;; Any coding system is ok. | |
953 | (setq coding-system (caar default-coding-system)) | |
954 | ;; Reverse the list so that elements are accumulated in safe, | |
955 | ;; rejected, and unsafe in the correct order. | |
956 | (setq default-coding-system (nreverse default-coding-system)) | |
957 | ||
958 | ;; Classify the defaults into safe, rejected, and unsafe. | |
959 | (dolist (elt default-coding-system) | |
960 | (if (or (eq (car codings) 'undecided) | |
961 | (memq (cdr elt) codings)) | |
962 | (if (and (functionp accept-default-p) | |
963 | (not (funcall accept-default-p (cdr elt)))) | |
964 | (push (car elt) rejected) | |
965 | (push (car elt) safe)) | |
966 | (push (car elt) unsafe))) | |
967 | (if safe | |
968 | (setq coding-system (car safe)))) | |
29d04c4f KH |
969 | |
970 | ;; If all the defaults failed, ask a user. | |
bae7cd08 | 971 | (when (not coding-system) |
29d04c4f KH |
972 | (setq coding-system (select-safe-coding-system-interactively |
973 | from to codings unsafe rejected (car codings)))) | |
974 | ||
29d04c4f KH |
975 | ;; Check we're not inconsistent with what `coding:' spec &c would |
976 | ;; give when file is re-read. | |
977 | ;; But don't do this if we explicitly ignored the cookie | |
978 | ;; by using `find-file-literally'. | |
979 | (when (and auto-cs | |
980 | (not (and | |
981 | coding-system | |
982 | (memq (coding-system-type coding-system) '(0 5))))) | |
6b61353c KH |
983 | ;; Merge coding-system and auto-cs as far as possible. |
984 | (if (not coding-system) | |
985 | (setq coding-system auto-cs) | |
986 | (if (not auto-cs) | |
987 | (setq auto-cs coding-system) | |
988 | (let ((eol-type-1 (coding-system-eol-type coding-system)) | |
989 | (eol-type-2 (coding-system-eol-type auto-cs))) | |
990 | (if (eq (coding-system-base coding-system) 'undecided) | |
991 | (setq coding-system (coding-system-change-text-conversion | |
992 | coding-system auto-cs)) | |
993 | (if (eq (coding-system-base auto-cs) 'undecided) | |
994 | (setq auto-cs (coding-system-change-text-conversion | |
995 | auto-cs coding-system)))) | |
996 | (if (vectorp eol-type-1) | |
997 | (or (vectorp eol-type-2) | |
998 | (setq coding-system (coding-system-change-eol-conversion | |
999 | coding-system eol-type-2))) | |
1000 | (if (vectorp eol-type-2) | |
1001 | (setq auto-cs (coding-system-change-eol-conversion | |
1002 | auto-cs eol-type-1))))))) | |
1003 | ||
1004 | (if (and auto-cs | |
d37ef0f6 DL |
1005 | ;; Don't barf if writing a compressed file, say. |
1006 | ;; This check perhaps isn't ideal, but is probably | |
1007 | ;; the best thing to do. | |
1008 | (not (auto-coding-alist-lookup (or file buffer-file-name ""))) | |
6b61353c | 1009 | (not (coding-system-equal coding-system auto-cs))) |
43afed8c RS |
1010 | (unless (yes-or-no-p |
1011 | (format "Selected encoding %s disagrees with \ | |
1012 | %s specified by file contents. Really save (else edit coding cookies \ | |
1013 | and try again)? " coding-system auto-cs)) | |
29d04c4f | 1014 | (error "Save aborted")))) |
86c3a9fb KH |
1015 | (when (and tick (/= tick (buffer-modified-tick))) |
1016 | (error "Cancelled because the buffer was modified")) | |
29d04c4f | 1017 | coding-system))) |
54b226f7 KH |
1018 | |
1019 | (setq select-safe-coding-system-function 'select-safe-coding-system) | |
1020 | ||
46babb23 KH |
1021 | (defun select-message-coding-system () |
1022 | "Return a coding system to encode the outgoing message of the current buffer. | |
1023 | It at first tries the first coding system found in these variables | |
1024 | in this order: | |
1025 | (1) local value of `buffer-file-coding-system' | |
1026 | (2) value of `sendmail-coding-system' | |
b5edd1d1 KH |
1027 | (3) value of `default-sendmail-coding-system' |
1028 | (4) value of `default-buffer-file-coding-system' | |
46babb23 KH |
1029 | If the found coding system can't encode the current buffer, |
1030 | or none of them are bound to a coding system, | |
48e41165 | 1031 | it asks the user to select a proper coding system." |
46babb23 | 1032 | (let ((coding (or (and (local-variable-p 'buffer-file-coding-system) |
b5edd1d1 KH |
1033 | buffer-file-coding-system) |
1034 | sendmail-coding-system | |
1035 | default-sendmail-coding-system | |
1036 | default-buffer-file-coding-system))) | |
46babb23 | 1037 | (if (eq coding 'no-conversion) |
d37ef0f6 | 1038 | ;; We should never use no-conversion for outgoing mail. |
46babb23 KH |
1039 | (setq coding nil)) |
1040 | (if (fboundp select-safe-coding-system-function) | |
1041 | (funcall select-safe-coding-system-function | |
b5edd1d1 | 1042 | (point-min) (point-max) coding |
209c73b0 | 1043 | (function (lambda (x) (coding-system-get x :mime-charset)))) |
46babb23 | 1044 | coding))) |
4ed46869 | 1045 | \f |
03c35c83 | 1046 | ;;; Language support stuff. |
4ed46869 | 1047 | |
4ed46869 | 1048 | (defvar language-info-alist nil |
2c395d56 | 1049 | "Alist of language environment definitions. |
4ed46869 KH |
1050 | Each element looks like: |
1051 | (LANGUAGE-NAME . ((KEY . INFO) ...)) | |
2c395d56 RS |
1052 | where LANGUAGE-NAME is a string, the name of the language environment, |
1053 | KEY is a symbol denoting the kind of information, and | |
1054 | INFO is the data associated with KEY. | |
1055 | Meaningful values for KEY include | |
1056 | ||
1057 | documentation value is documentation of what this language environment | |
1058 | is meant for, and how to use it. | |
7dd42fb1 KH |
1059 | charset value is a list of the character sets mainly used |
1060 | by this language environment. | |
d37ef0f6 DL |
1061 | sample-text value is an expression which is evalled to generate |
1062 | a line of text written using characters appropriate | |
1063 | for this language environment. | |
2c395d56 RS |
1064 | setup-function value is a function to call to switch to this |
1065 | language environment. | |
1066 | exit-function value is a function to call to leave this | |
1067 | language environment. | |
1068 | coding-system value is a list of coding systems that are good | |
1069 | for saving text written in this language environment. | |
1070 | This list serves as suggestions to the user; | |
1071 | in effect, as a kind of documentation. | |
1072 | coding-priority value is a list of coding systems for this language | |
1073 | environment, in order of decreasing priority. | |
1074 | This is used to set up the coding system priority | |
45d08cb2 | 1075 | list when you switch to this language environment. |
ddb5c041 | 1076 | nonascii-translation |
7dd42fb1 KH |
1077 | value is a charset of dimension one to use for |
1078 | converting a unibyte character to multibyte | |
1079 | and vice versa. | |
ddb5c041 KH |
1080 | input-method value is a default input method for this language |
1081 | environment. | |
7624ebb9 KH |
1082 | features value is a list of features requested in this |
1083 | language environment. | |
6b61353c KH |
1084 | ctext-non-standard-encodings |
1085 | value is a list of non-standard encoding | |
1086 | names used in extended segments of CTEXT. | |
1087 | See the variable | |
1088 | `ctext-non-standard-encodings' for more | |
1089 | detail. | |
ddb5c041 KH |
1090 | |
1091 | The following keys take effect only when multibyte characters are | |
1092 | globally disabled, i.e. the value of `default-enable-multibyte-characters' | |
1093 | is nil. | |
1094 | ||
ddb5c041 KH |
1095 | unibyte-display value is a coding system to encode characters |
1096 | for the terminal. Characters in the range | |
1097 | of 160 to 255 display not as octal escapes, | |
1098 | but as non-ASCII characters in this language | |
1099 | environment.") | |
2c395d56 RS |
1100 | |
1101 | (defun get-language-info (lang-env key) | |
1102 | "Return information listed under KEY for language environment LANG-ENV. | |
1103 | KEY is a symbol denoting the kind of information. | |
1104 | For a list of useful values for KEY and their meanings, | |
1105 | see `language-info-alist'." | |
1106 | (if (symbolp lang-env) | |
1107 | (setq lang-env (symbol-name lang-env))) | |
f15466c5 | 1108 | (let ((lang-slot (assoc-string lang-env language-info-alist t))) |
4ed46869 KH |
1109 | (if lang-slot |
1110 | (cdr (assq key (cdr lang-slot)))))) | |
1111 | ||
f08adf27 | 1112 | (defun set-language-info (lang-env key info) |
2c395d56 RS |
1113 | "Modify part of the definition of language environment LANG-ENV. |
1114 | Specifically, this stores the information INFO under KEY | |
1115 | in the definition of this language environment. | |
4ed46869 | 1116 | KEY is a symbol denoting the kind of information. |
2c395d56 | 1117 | INFO is the value for that information. |
281d03ec | 1118 | |
2c395d56 | 1119 | For a list of useful values for KEY and their meanings, |
f08adf27 | 1120 | see `language-info-alist'." |
2c395d56 RS |
1121 | (if (symbolp lang-env) |
1122 | (setq lang-env (symbol-name lang-env))) | |
eec5c8f9 KH |
1123 | (set-language-info-internal lang-env key info) |
1124 | (if (equal lang-env current-language-environment) | |
d042f8b4 | 1125 | (cond ((eq key 'coding-priority) |
e9a8ed3c KH |
1126 | (set-language-environment-coding-systems lang-env) |
1127 | (set-language-environment-charset lang-env)) | |
d042f8b4 KH |
1128 | ((eq key 'input-method) |
1129 | (set-language-environment-input-method lang-env)) | |
1130 | ((eq key 'nonascii-translation) | |
1131 | (set-language-environment-nonascii-translation lang-env)) | |
1132 | ((eq key 'charset) | |
1133 | (set-language-environment-charset lang-env)) | |
d042f8b4 KH |
1134 | ((and (not default-enable-multibyte-characters) |
1135 | (or (eq key 'unibyte-syntax) (eq key 'unibyte-display))) | |
1136 | (set-language-environment-unibyte lang-env))))) | |
eec5c8f9 KH |
1137 | |
1138 | (defun set-language-info-internal (lang-env key info) | |
1139 | "Internal use only. | |
1140 | Arguments are the same as `set-language-info'." | |
4ed46869 | 1141 | (let (lang-slot key-slot) |
2c395d56 | 1142 | (setq lang-slot (assoc lang-env language-info-alist)) |
4ed46869 | 1143 | (if (null lang-slot) ; If no slot for the language, add it. |
2c395d56 | 1144 | (setq lang-slot (list lang-env) |
4ed46869 KH |
1145 | language-info-alist (cons lang-slot language-info-alist))) |
1146 | (setq key-slot (assq key lang-slot)) | |
1147 | (if (null key-slot) ; If no slot for the key, add it. | |
1148 | (progn | |
1149 | (setq key-slot (list key)) | |
1150 | (setcdr lang-slot (cons key-slot (cdr lang-slot))))) | |
590dbcba | 1151 | (setcdr key-slot (purecopy info)) |
1169bd86 | 1152 | ;; Update the custom-type of `current-language-environment'. |
590dbcba | 1153 | (put 'current-language-environment 'custom-type |
1169bd86 MR |
1154 | (cons 'choice (mapcar |
1155 | (lambda (lang) | |
756e055f MR |
1156 | (list 'const lang)) |
1157 | (sort (mapcar 'car language-info-alist) 'string<)))))) | |
4ed46869 | 1158 | |
2c395d56 RS |
1159 | (defun set-language-info-alist (lang-env alist &optional parents) |
1160 | "Store ALIST as the definition of language environment LANG-ENV. | |
1161 | ALIST is an alist of KEY and INFO values. See the documentation of | |
98c6d6ed | 1162 | `language-info-alist' for the meanings of KEY and INFO. |
54b226f7 | 1163 | |
2c395d56 | 1164 | Optional arg PARENTS is a list of parent menu names; it specifies |
d37ef0f6 | 1165 | where to put this language environment in the |
2c395d56 RS |
1166 | Describe Language Environment and Set Language Environment menus. |
1167 | For example, (\"European\") means to put this language environment | |
1168 | in the European submenu in each of those two menus." | |
1169 | (if (symbolp lang-env) | |
1170 | (setq lang-env (symbol-name lang-env))) | |
54b226f7 KH |
1171 | (let ((describe-map describe-language-environment-map) |
1172 | (setup-map setup-language-environment-map)) | |
1173 | (if parents | |
1174 | (let ((l parents) | |
9deed82f | 1175 | map parent-symbol parent prompt) |
54b226f7 KH |
1176 | (while l |
1177 | (if (symbolp (setq parent-symbol (car l))) | |
1178 | (setq parent (symbol-name parent)) | |
1179 | (setq parent parent-symbol parent-symbol (intern parent))) | |
1180 | (setq map (lookup-key describe-map (vector parent-symbol))) | |
9deed82f EZ |
1181 | ;; This prompt string is for define-prefix-command, so |
1182 | ;; that the map it creates will be suitable for a menu. | |
1183 | (or map (setq prompt (format "%s Environment" parent))) | |
54b226f7 KH |
1184 | (if (not map) |
1185 | (progn | |
1186 | (setq map (intern (format "describe-%s-environment-map" | |
1187 | (downcase parent)))) | |
9deed82f | 1188 | (define-prefix-command map nil prompt) |
54b226f7 KH |
1189 | (define-key-after describe-map (vector parent-symbol) |
1190 | (cons parent map) t))) | |
1191 | (setq describe-map (symbol-value map)) | |
1192 | (setq map (lookup-key setup-map (vector parent-symbol))) | |
1193 | (if (not map) | |
1194 | (progn | |
1195 | (setq map (intern (format "setup-%s-environment-map" | |
1196 | (downcase parent)))) | |
9deed82f | 1197 | (define-prefix-command map nil prompt) |
54b226f7 KH |
1198 | (define-key-after setup-map (vector parent-symbol) |
1199 | (cons parent map) t))) | |
1200 | (setq setup-map (symbol-value map)) | |
1201 | (setq l (cdr l))))) | |
f08adf27 RS |
1202 | |
1203 | ;; Set up menu items for this language env. | |
7624ebb9 | 1204 | (let ((doc (assq 'documentation alist))) |
f08adf27 RS |
1205 | (when doc |
1206 | (define-key-after describe-map (vector (intern lang-env)) | |
7624ebb9 KH |
1207 | (cons lang-env 'describe-specified-language-support) t))) |
1208 | (define-key-after setup-map (vector (intern lang-env)) | |
1209 | (cons lang-env 'setup-specified-language-environment) t) | |
f08adf27 | 1210 | |
eec5c8f9 KH |
1211 | (dolist (elt alist) |
1212 | (set-language-info-internal lang-env (car elt) (cdr elt))) | |
d35deeec | 1213 | |
eec5c8f9 KH |
1214 | (if (equal lang-env current-language-environment) |
1215 | (set-language-environment lang-env)))) | |
4ed46869 | 1216 | |
ae302641 | 1217 | (defun read-language-name (key prompt &optional default) |
2c395d56 | 1218 | "Read a language environment name which has information for KEY. |
ddb5c041 | 1219 | If KEY is nil, read any language environment. |
2c395d56 RS |
1220 | Prompt with PROMPT. DEFAULT is the default choice of language environment. |
1221 | This returns a language environment name as a string." | |
4ed46869 KH |
1222 | (let* ((completion-ignore-case t) |
1223 | (name (completing-read prompt | |
1224 | language-info-alist | |
ddb5c041 | 1225 | (and key |
ca429a25 | 1226 | (function (lambda (elm) (and (listp elm) (assq key elm))))) |
ae302641 | 1227 | t nil nil default))) |
13e82c04 | 1228 | (if (and (> (length name) 0) |
ddb5c041 KH |
1229 | (or (not key) |
1230 | (get-language-info name key))) | |
13e82c04 | 1231 | name))) |
4ed46869 KH |
1232 | \f |
1233 | ;;; Multilingual input methods. | |
d37ef0f6 | 1234 | (defgroup leim nil |
d0c40faa KH |
1235 | "LEIM: Libraries of Emacs Input Methods." |
1236 | :group 'mule) | |
4ed46869 | 1237 | |
d0b9c3ab KH |
1238 | (defconst leim-list-file-name "leim-list.el" |
1239 | "Name of LEIM list file. | |
1240 | This file contains a list of libraries of Emacs input methods (LEIM) | |
1241 | in the format of Lisp expression for registering each input method. | |
1242 | Emacs loads this file at startup time.") | |
1243 | ||
2e224638 | 1244 | (defvar leim-list-header (format |
d37ef0f6 | 1245 | ";;; %s -- list of LEIM (Library of Emacs Input Method) -*-coding: iso-2022-7bit;-*- |
d0b9c3ab KH |
1246 | ;; |
1247 | ;; This file contains a list of LEIM (Library of Emacs Input Method) | |
c654de1d DL |
1248 | ;; methods in the same directory as this file. Loading this file |
1249 | ;; registers all the input methods in Emacs. | |
d0b9c3ab | 1250 | ;; |
d33d5fbe | 1251 | ;; Each entry has the form: |
d0b9c3ab KH |
1252 | ;; (register-input-method |
1253 | ;; INPUT-METHOD LANGUAGE-NAME ACTIVATE-FUNC | |
1254 | ;; TITLE DESCRIPTION | |
1255 | ;; ARG ...) | |
c654de1d | 1256 | ;; See the function `register-input-method' for the meanings of the arguments. |
d0b9c3ab KH |
1257 | ;; |
1258 | ;; If this directory is included in load-path, Emacs automatically | |
1259 | ;; loads this file at startup time. | |
1260 | ||
1261 | " | |
1262 | leim-list-file-name) | |
1263 | "Header to be inserted in LEIM list file.") | |
1264 | ||
e55e92ee | 1265 | (defvar leim-list-entry-regexp "^(register-input-method" |
d0b9c3ab | 1266 | "Regexp matching head of each entry in LEIM list file. |
caff3c0a | 1267 | See also the variable `leim-list-header'.") |
d0b9c3ab KH |
1268 | |
1269 | (defvar update-leim-list-functions | |
1270 | '(quail-update-leim-list-file) | |
1271 | "List of functions to call to update LEIM list file. | |
1272 | Each function is called with one arg, LEIM directory name.") | |
1273 | ||
a337fe7f RS |
1274 | (defun update-leim-list-file (&rest dirs) |
1275 | "Update LEIM list file in directories DIRS." | |
d0b9c3ab KH |
1276 | (let ((functions update-leim-list-functions)) |
1277 | (while functions | |
a337fe7f | 1278 | (apply (car functions) dirs) |
d0b9c3ab KH |
1279 | (setq functions (cdr functions))))) |
1280 | ||
4ed46869 KH |
1281 | (defvar current-input-method nil |
1282 | "The current input method for multilingual text. | |
96db204a | 1283 | If nil, that means no input method is activated now.") |
4ed46869 KH |
1284 | (make-variable-buffer-local 'current-input-method) |
1285 | (put 'current-input-method 'permanent-local t) | |
1286 | ||
1287 | (defvar current-input-method-title nil | |
d0b9c3ab | 1288 | "Title string of the current input method shown in mode line.") |
4ed46869 KH |
1289 | (make-variable-buffer-local 'current-input-method-title) |
1290 | (put 'current-input-method-title 'permanent-local t) | |
1291 | ||
b4fba33f | 1292 | (defcustom default-input-method nil |
caff3c0a | 1293 | "Default input method for multilingual text (a string). |
b4fba33f | 1294 | This is the input method activated automatically by the command |
9b10b5a3 | 1295 | `toggle-input-method' (\\[toggle-input-method])." |
1f547b92 | 1296 | :link '(custom-manual "(emacs)Input Methods") |
8861c593 | 1297 | :group 'mule |
d398dba6 DL |
1298 | :type '(choice (const nil) (string |
1299 | :completion-ignore-case t | |
1300 | :complete-function widget-string-complete | |
1301 | :completion-alist input-method-alist | |
1302 | :prompt-history input-method-history)) | |
5806e8a6 | 1303 | :set-after '(current-language-environment)) |
b4fba33f | 1304 | |
0f835e87 KH |
1305 | (put 'input-method-function 'permanent-local t) |
1306 | ||
723a427a | 1307 | (defvar input-method-history nil |
6f5d2452 EZ |
1308 | "History list of input methods read from the minibuffer. |
1309 | ||
1310 | Maximum length of the history list is determined by the value | |
1311 | of `history-length', which see.") | |
723a427a KH |
1312 | (make-variable-buffer-local 'input-method-history) |
1313 | (put 'input-method-history 'permanent-local t) | |
4ed46869 KH |
1314 | |
1315 | (defvar inactivate-current-input-method-function nil | |
1316 | "Function to call for inactivating the current input method. | |
1317 | Every input method should set this to an appropriate value when activated. | |
f17ccaee KH |
1318 | This function is called with no argument. |
1319 | ||
1320 | This function should never change the value of `current-input-method'. | |
1321 | It is set to nil by the function `inactivate-input-method'.") | |
4ed46869 KH |
1322 | (make-variable-buffer-local 'inactivate-current-input-method-function) |
1323 | (put 'inactivate-current-input-method-function 'permanent-local t) | |
1324 | ||
1325 | (defvar describe-current-input-method-function nil | |
1326 | "Function to call for describing the current input method. | |
1327 | This function is called with no argument.") | |
1328 | (make-variable-buffer-local 'describe-current-input-method-function) | |
1329 | (put 'describe-current-input-method-function 'permanent-local t) | |
1330 | ||
d0b9c3ab | 1331 | (defvar input-method-alist nil |
2c395d56 | 1332 | "Alist of input method names vs how to use them. |
d0b9c3ab | 1333 | Each element has the form: |
2c395d56 RS |
1334 | (INPUT-METHOD LANGUAGE-ENV ACTIVATE-FUNC TITLE DESCRIPTION ARGS...) |
1335 | See the function `register-input-method' for the meanings of the elements.") | |
1336 | ||
f08adf27 | 1337 | (defun register-input-method (input-method lang-env &rest args) |
bf42aa15 | 1338 | "Register INPUT-METHOD as an input method for language environment LANG-ENV. |
d0b9c3ab | 1339 | |
d35deeec | 1340 | INPUT-METHOD and LANG-ENV are symbols or strings. |
2c395d56 RS |
1341 | ACTIVATE-FUNC is a function to call to activate this method. |
1342 | TITLE is a string to show in the mode line when this method is active. | |
1343 | DESCRIPTION is a string describing this method and what it is good for. | |
1344 | The ARGS, if any, are passed as arguments to ACTIVATE-FUNC. | |
205814ee KH |
1345 | All told, the arguments to ACTIVATE-FUNC are INPUT-METHOD and the ARGS. |
1346 | ||
c654de1d DL |
1347 | This function is mainly used in the file \"leim-list.el\" which is |
1348 | created at Emacs build time, registering all Quail input methods | |
0b6cadff | 1349 | contained in the Emacs distribution. |
205814ee | 1350 | |
0b6cadff | 1351 | In case you want to register a new Quail input method by yourself, be |
205814ee | 1352 | careful to use the same input method title as given in the third |
0b6cadff DL |
1353 | parameter of `quail-define-package'. (If the values are different, the |
1354 | string specified in this function takes precedence.) | |
205814ee KH |
1355 | |
1356 | The commands `describe-input-method' and `list-input-methods' need | |
0b6cadff | 1357 | these duplicated values to show some information about input methods |
d35deeec JB |
1358 | without loading the relevant Quail packages. |
1359 | \n(fn INPUT-METHOD LANG-ENV ACTIVATE-FUNC TITLE DESCRIPTION &rest ARGS)" | |
f08adf27 RS |
1360 | (if (symbolp lang-env) |
1361 | (setq lang-env (symbol-name lang-env))) | |
4ef06f75 KH |
1362 | (if (symbolp input-method) |
1363 | (setq input-method (symbol-name input-method))) | |
f08adf27 | 1364 | (let ((info (cons lang-env args)) |
d0b9c3ab KH |
1365 | (slot (assoc input-method input-method-alist))) |
1366 | (if slot | |
1367 | (setcdr slot info) | |
1368 | (setq slot (cons input-method info)) | |
1369 | (setq input-method-alist (cons slot input-method-alist))))) | |
1370 | ||
4d5ac029 | 1371 | (defun read-input-method-name (prompt &optional default inhibit-null) |
d0b9c3ab | 1372 | "Read a name of input method from a minibuffer prompting with PROMPT. |
4d5ac029 | 1373 | If DEFAULT is non-nil, use that as the default, |
0b6cadff | 1374 | and substitute it into PROMPT at the first `%s'. |
4ef06f75 KH |
1375 | If INHIBIT-NULL is non-nil, null input signals an error. |
1376 | ||
1377 | The return value is a string." | |
4d5ac029 RS |
1378 | (if default |
1379 | (setq prompt (format prompt default))) | |
d0b9c3ab | 1380 | (let* ((completion-ignore-case t) |
c54044ff KH |
1381 | ;; As it is quite normal to change input method in the |
1382 | ;; minibuffer, we must enable it even if | |
1383 | ;; enable-recursive-minibuffers is currently nil. | |
1384 | (enable-recursive-minibuffers t) | |
723a427a KH |
1385 | ;; This binding is necessary because input-method-history is |
1386 | ;; buffer local. | |
d0b9c3ab | 1387 | (input-method (completing-read prompt input-method-alist |
87505a98 RS |
1388 | nil t nil 'input-method-history |
1389 | default))) | |
bf294e6e KH |
1390 | (if (and input-method (symbolp input-method)) |
1391 | (setq input-method (symbol-name input-method))) | |
d0b9c3ab KH |
1392 | (if (> (length input-method) 0) |
1393 | input-method | |
1394 | (if inhibit-null | |
43807b77 | 1395 | (error "No valid input method is specified"))))) |
d0b9c3ab | 1396 | |
d0b9c3ab | 1397 | (defun activate-input-method (input-method) |
2c395d56 RS |
1398 | "Switch to input method INPUT-METHOD for the current buffer. |
1399 | If some other input method is already active, turn it off first. | |
1400 | If INPUT-METHOD is nil, deactivate any current input method." | |
305a3cb6 | 1401 | (if (and input-method (symbolp input-method)) |
4ef06f75 | 1402 | (setq input-method (symbol-name input-method))) |
723a427a KH |
1403 | (if (and current-input-method |
1404 | (not (string= current-input-method input-method))) | |
305a3cb6 | 1405 | (inactivate-input-method)) |
2c395d56 | 1406 | (unless (or current-input-method (null input-method)) |
d0b9c3ab KH |
1407 | (let ((slot (assoc input-method input-method-alist))) |
1408 | (if (null slot) | |
723a427a | 1409 | (error "Can't activate input method `%s'" input-method)) |
278dd6ac | 1410 | (setq current-input-method-title nil) |
8efc03e1 KH |
1411 | (let ((func (nth 2 slot))) |
1412 | (if (functionp func) | |
1413 | (apply (nth 2 slot) input-method (nthcdr 5 slot)) | |
1414 | (if (and (consp func) (symbolp (car func)) (symbolp (cdr func))) | |
1415 | (progn | |
1416 | (require (cdr func)) | |
1417 | (apply (car func) input-method (nthcdr 5 slot))) | |
1418 | (error "Can't activate input method `%s'" input-method)))) | |
d0b9c3ab | 1419 | (setq current-input-method input-method) |
278dd6ac KH |
1420 | (or (stringp current-input-method-title) |
1421 | (setq current-input-method-title (nth 3 slot))) | |
28885c0e KH |
1422 | (unwind-protect |
1423 | (run-hooks 'input-method-activate-hook) | |
1424 | (force-mode-line-update))))) | |
15b3e511 | 1425 | |
15b3e511 | 1426 | (defun inactivate-input-method () |
f17ccaee | 1427 | "Turn off the current input method." |
723a427a KH |
1428 | (when current-input-method |
1429 | (if input-method-history | |
1430 | (unless (string= current-input-method (car input-method-history)) | |
1431 | (setq input-method-history | |
1432 | (cons current-input-method | |
1433 | (delete current-input-method input-method-history)))) | |
1434 | (setq input-method-history (list current-input-method))) | |
1435 | (unwind-protect | |
36f8618c KH |
1436 | (progn |
1437 | (setq input-method-function nil | |
1438 | current-input-method-title nil) | |
1439 | (funcall inactivate-current-input-method-function)) | |
15b3e511 | 1440 | (unwind-protect |
723a427a | 1441 | (run-hooks 'input-method-inactivate-hook) |
36f8618c | 1442 | (setq current-input-method nil) |
28885c0e | 1443 | (force-mode-line-update))))) |
4ed46869 | 1444 | |
e893eae2 | 1445 | (defun set-input-method (input-method &optional interactive) |
2c395d56 | 1446 | "Select and activate input method INPUT-METHOD for the current buffer. |
bc406911 | 1447 | This also sets the default input method to the one you specify. |
402dbbd1 EZ |
1448 | If INPUT-METHOD is nil, this function turns off the input method, and |
1449 | also causes you to be prompted for a name of an input method the next | |
1450 | time you invoke \\[toggle-input-method]. | |
e893eae2 RS |
1451 | When called interactively, the optional arg INTERACTIVE is non-nil, |
1452 | which marks the variable `default-input-method' as set for Custom buffers. | |
402dbbd1 | 1453 | |
bc406911 | 1454 | To deactivate the input method interactively, use \\[toggle-input-method]. |
caff3c0a | 1455 | To deactivate it programmatically, use `inactivate-input-method'." |
d0b9c3ab | 1456 | (interactive |
723a427a | 1457 | (let* ((default (or (car input-method-history) default-input-method))) |
42395763 | 1458 | (list (read-input-method-name |
87505a98 | 1459 | (if default "Select input method (default %s): " "Select input method: ") |
e893eae2 RS |
1460 | default t) |
1461 | t))) | |
d0b9c3ab | 1462 | (activate-input-method input-method) |
f4990970 | 1463 | (setq default-input-method input-method) |
e893eae2 | 1464 | (when interactive |
f4990970 PA |
1465 | (customize-mark-as-set 'default-input-method)) |
1466 | default-input-method) | |
4ed46869 | 1467 | |
023df4cf RS |
1468 | (defvar toggle-input-method-active nil |
1469 | "Non-nil inside `toggle-input-method'.") | |
1470 | ||
e893eae2 | 1471 | (defun toggle-input-method (&optional arg interactive) |
f8ec20be RS |
1472 | "Enable or disable multilingual text input method for the current buffer. |
1473 | Only one input method can be enabled at any time in a given buffer. | |
1474 | ||
1475 | The normal action is to enable an input method if none was | |
1476 | enabled, and disable the current one otherwise. Which input method | |
1477 | to enable can be determined in various ways--either the one most | |
1478 | recently used, or the one specified by `default-input-method', or | |
1479 | as a last resort by reading the name of an input method in the | |
1480 | minibuffer. | |
1481 | ||
1482 | With a prefix argument, read an input method name with the minibuffer | |
1483 | and enable that one. The default is the most recent input method specified | |
e893eae2 | 1484 | \(not including the currently active input method, if any). |
f8ec20be | 1485 | |
e893eae2 RS |
1486 | When called interactively, the optional arg INTERACTIVE is non-nil, |
1487 | which marks the variable `default-input-method' as set for Custom buffers." | |
1488 | ||
1489 | (interactive "P\np") | |
023df4cf RS |
1490 | (if toggle-input-method-active |
1491 | (error "Recursive use of `toggle-input-method'")) | |
7ddbb5bc RS |
1492 | (if (and current-input-method (not arg)) |
1493 | (inactivate-input-method) | |
023df4cf RS |
1494 | (let ((toggle-input-method-active t) |
1495 | (default (or (car input-method-history) default-input-method))) | |
7ddbb5bc RS |
1496 | (if (and arg default (equal current-input-method default) |
1497 | (> (length input-method-history) 1)) | |
1498 | (setq default (nth 1 input-method-history))) | |
723a427a KH |
1499 | (activate-input-method |
1500 | (if (or arg (not default)) | |
7ddbb5bc RS |
1501 | (progn |
1502 | (read-input-method-name | |
1503 | (if default "Input method (default %s): " "Input method: " ) | |
1504 | default t)) | |
723a427a | 1505 | default)) |
f4990970 | 1506 | (unless default-input-method |
d37ef0f6 | 1507 | (prog1 |
f4990970 | 1508 | (setq default-input-method current-input-method) |
e893eae2 | 1509 | (when interactive |
f4990970 | 1510 | (customize-mark-as-set 'default-input-method))))))) |
d0b9c3ab | 1511 | |
0855c6cd DL |
1512 | (eval-when-compile (autoload 'help-buffer "help-mode")) |
1513 | ||
d0b9c3ab | 1514 | (defun describe-input-method (input-method) |
2c395d56 | 1515 | "Describe input method INPUT-METHOD." |
d0b9c3ab KH |
1516 | (interactive |
1517 | (list (read-input-method-name | |
5b76833f | 1518 | "Describe input method (default current choice): "))) |
78754934 | 1519 | (if (and input-method (symbolp input-method)) |
4ef06f75 | 1520 | (setq input-method (symbol-name input-method))) |
43125c28 RS |
1521 | (help-setup-xref (list #'describe-input-method |
1522 | (or input-method current-input-method)) | |
f80e2142 RS |
1523 | (interactive-p)) |
1524 | ||
d0b9c3ab KH |
1525 | (if (null input-method) |
1526 | (describe-current-input-method) | |
464cc130 KH |
1527 | (let ((current current-input-method)) |
1528 | (condition-case nil | |
1529 | (progn | |
1530 | (save-excursion | |
1531 | (activate-input-method input-method) | |
1532 | (describe-current-input-method)) | |
1533 | (activate-input-method current)) | |
d37ef0f6 | 1534 | (error |
464cc130 | 1535 | (activate-input-method current) |
5f395df3 SM |
1536 | (help-setup-xref (list #'describe-input-method input-method) |
1537 | (interactive-p)) | |
1538 | (with-output-to-temp-buffer (help-buffer) | |
464cc130 KH |
1539 | (let ((elt (assoc input-method input-method-alist))) |
1540 | (princ (format | |
1541 | "Input method: %s (`%s' in mode line) for %s\n %s\n" | |
1542 | input-method (nth 3 elt) (nth 1 elt) (nth 4 elt)))))))))) | |
d0b9c3ab KH |
1543 | |
1544 | (defun describe-current-input-method () | |
f80e2142 RS |
1545 | "Describe the input method currently in use. |
1546 | This is a subroutine for `describe-input-method'." | |
4ed46869 KH |
1547 | (if current-input-method |
1548 | (if (and (symbolp describe-current-input-method-function) | |
1549 | (fboundp describe-current-input-method-function)) | |
1550 | (funcall describe-current-input-method-function) | |
1551 | (message "No way to describe the current input method `%s'" | |
f2979bdb | 1552 | current-input-method) |
4ed46869 | 1553 | (ding)) |
d0b9c3ab | 1554 | (error "No input method is activated now"))) |
4ed46869 | 1555 | |
d3459641 | 1556 | (defun read-multilingual-string (prompt &optional initial-input input-method) |
4ed46869 KH |
1557 | "Read a multilingual string from minibuffer, prompting with string PROMPT. |
1558 | The input method selected last time is activated in minibuffer. | |
15b3e511 | 1559 | If optional second arg INITIAL-INPUT is non-nil, insert it in the minibuffer |
d0b9c3ab KH |
1560 | initially. |
1561 | Optional 3rd argument INPUT-METHOD specifies the input method | |
4ef06f75 KH |
1562 | to be activated instead of the one selected last time. It is a symbol |
1563 | or a string." | |
88d559ec KH |
1564 | (setq input-method |
1565 | (or input-method | |
d3459641 | 1566 | current-input-method |
88d559ec KH |
1567 | default-input-method |
1568 | (read-input-method-name "Input method: " nil t))) | |
3df60841 | 1569 | (if (and input-method (symbolp input-method)) |
4ef06f75 | 1570 | (setq input-method (symbol-name input-method))) |
305a3cb6 KH |
1571 | (let ((prev-input-method current-input-method)) |
1572 | (unwind-protect | |
1573 | (progn | |
1574 | (activate-input-method input-method) | |
1575 | (read-string prompt initial-input nil nil t)) | |
1576 | (activate-input-method prev-input-method)))) | |
4ed46869 KH |
1577 | |
1578 | ;; Variables to control behavior of input methods. All input methods | |
1579 | ;; should react to these variables. | |
1580 | ||
8efc03e1 | 1581 | (defcustom input-method-verbose-flag 'default |
caff3c0a | 1582 | "A flag to control extra guidance given by input methods. |
8efc03e1 | 1583 | The value should be nil, t, `complex-only', or `default'. |
4ed46869 | 1584 | |
cb29dfb6 | 1585 | The extra guidance is done by showing list of available keys in echo |
8efc03e1 KH |
1586 | area. When you use the input method in the minibuffer, the guidance |
1587 | is shown at the bottom short window (split from the existing window). | |
c27c4ed8 | 1588 | |
8efc03e1 KH |
1589 | If the value is t, extra guidance is always given, if the value is |
1590 | nil, extra guidance is always suppressed. | |
1591 | ||
1592 | If the value is `complex-only', only complex input methods such as | |
1593 | `chinese-py' and `japanese' give extra guidance. | |
1594 | ||
1595 | If the value is `default', complex input methods always give extra | |
1596 | guidance, but simple input methods give it only when you are not in | |
1597 | the minibuffer. | |
1598 | ||
1599 | See also the variable `input-method-highlight-flag'." | |
d37ef0f6 DL |
1600 | :type '(choice (const :tag "Always" t) (const :tag "Never" nil) |
1601 | (const complex-only) (const default)) | |
42395763 RS |
1602 | :group 'mule) |
1603 | ||
1604 | (defcustom input-method-highlight-flag t | |
caff3c0a | 1605 | "If this flag is non-nil, input methods highlight partially-entered text. |
42395763 RS |
1606 | For instance, while you are in the middle of a Quail input method sequence, |
1607 | the text inserted so far is temporarily underlined. | |
8efc03e1 KH |
1608 | The underlining goes away when you finish or abort the input method sequence. |
1609 | See also the variable `input-method-verbose-flag'." | |
42395763 RS |
1610 | :type 'boolean |
1611 | :group 'mule) | |
4ed46869 | 1612 | |
1f547b92 | 1613 | (defcustom input-method-activate-hook nil |
f17ccaee KH |
1614 | "Normal hook run just after an input method is activated. |
1615 | ||
1616 | The variable `current-input-method' keeps the input method name | |
1f547b92 DL |
1617 | just activated." |
1618 | :type 'hook | |
1619 | :group 'mule) | |
4ed46869 | 1620 | |
1f547b92 | 1621 | (defcustom input-method-inactivate-hook nil |
f17ccaee KH |
1622 | "Normal hook run just after an input method is inactivated. |
1623 | ||
1624 | The variable `current-input-method' still keeps the input method name | |
1f547b92 DL |
1625 | just inactivated." |
1626 | :type 'hook | |
1627 | :group 'mule) | |
4ed46869 | 1628 | |
1f547b92 | 1629 | (defcustom input-method-after-insert-chunk-hook nil |
8f924df7 | 1630 | "Normal hook run just after an input method insert some chunk of text." |
1f547b92 DL |
1631 | :type 'hook |
1632 | :group 'mule) | |
4ed46869 | 1633 | |
dccca980 | 1634 | (defvar input-method-exit-on-first-char nil |
0b6cadff | 1635 | "This flag controls when an input method returns. |
dccca980 KH |
1636 | Usually, the input method does not return while there's a possibility |
1637 | that it may find a different translation if a user types another key. | |
caff3c0a | 1638 | But, if this flag is non-nil, the input method returns as soon as |
39e643e2 | 1639 | the current key sequence gets long enough to have some valid translation.") |
dccca980 | 1640 | |
1f547b92 | 1641 | (defcustom input-method-use-echo-area nil |
dccca980 | 1642 | "This flag controls how an input method shows an intermediate key sequence. |
39e643e2 RS |
1643 | Usually, the input method inserts the intermediate key sequence, |
1644 | or candidate translations corresponding to the sequence, | |
1645 | at point in the current buffer. | |
1f547b92 DL |
1646 | But, if this flag is non-nil, it displays them in echo area instead." |
1647 | :type 'hook | |
1648 | :group 'mule) | |
dccca980 | 1649 | |
723a427a | 1650 | (defvar input-method-exit-on-invalid-key nil |
fea6b736 | 1651 | "This flag controls the behavior of an input method on invalid key input. |
723a427a KH |
1652 | Usually, when a user types a key which doesn't start any character |
1653 | handled by the input method, the key is handled by turning off the | |
e8dd0160 | 1654 | input method temporarily. After that key, the input method is re-enabled. |
723a427a KH |
1655 | But, if this flag is non-nil, the input method is never back on.") |
1656 | ||
4ed46869 | 1657 | \f |
1f547b92 | 1658 | (defcustom set-language-environment-hook nil |
8efc03e1 KH |
1659 | "Normal hook run after some language environment is set. |
1660 | ||
1661 | When you set some hook function here, that effect usually should not | |
1662 | be inherited to another language environment. So, you had better set | |
1663 | another function in `exit-language-environment-hook' (which see) to | |
1f547b92 DL |
1664 | cancel the effect." |
1665 | :type 'hook | |
1666 | :group 'mule) | |
8efc03e1 | 1667 | |
1f547b92 | 1668 | (defcustom exit-language-environment-hook nil |
8efc03e1 KH |
1669 | "Normal hook run after exiting from some language environment. |
1670 | When this hook is run, the variable `current-language-environment' | |
1671 | is still bound to the language environment being exited. | |
1672 | ||
e8dd0160 | 1673 | This hook is mainly used for canceling the effect of |
caff3c0a | 1674 | `set-language-environment-hook' (which see)." |
1f547b92 DL |
1675 | :type 'hook |
1676 | :group 'mule) | |
8efc03e1 | 1677 | |
b0648a00 RS |
1678 | (put 'setup-specified-language-environment 'apropos-inhibit t) |
1679 | ||
15b3e511 | 1680 | (defun setup-specified-language-environment () |
f08adf27 | 1681 | "Switch to a specified language environment." |
15b3e511 | 1682 | (interactive) |
f850d782 | 1683 | (let (language-name) |
15b3e511 KH |
1684 | (if (and (symbolp last-command-event) |
1685 | (or (not (eq last-command-event 'Default)) | |
1686 | (setq last-command-event 'English)) | |
f850d782 | 1687 | (setq language-name (symbol-name last-command-event))) |
f4990970 PA |
1688 | (prog1 |
1689 | (set-language-environment language-name) | |
1690 | (customize-mark-as-set 'current-language-environment)) | |
15b3e511 | 1691 | (error "Bogus calling sequence")))) |
4ed46869 | 1692 | |
8861c593 | 1693 | (defcustom current-language-environment "English" |
94d04df6 | 1694 | "The last language environment specified with `set-language-environment'. |
ebef6d93 KH |
1695 | This variable should be set only with \\[customize], which is equivalent |
1696 | to using the function `set-language-environment'." | |
94d04df6 | 1697 | :link '(custom-manual "(emacs)Language Environments") |
dff1aa24 | 1698 | :set (lambda (symbol value) (set-language-environment value)) |
94d04df6 | 1699 | :get (lambda (x) |
f15466c5 | 1700 | (or (car-safe (assoc-string |
94d04df6 DL |
1701 | (if (symbolp current-language-environment) |
1702 | (symbol-name current-language-environment) | |
1703 | current-language-environment) | |
f15466c5 | 1704 | language-info-alist t)) |
94d04df6 | 1705 | "English")) |
990a4108 MR |
1706 | ;; custom type will be updated with `set-language-info'. |
1707 | :type (if language-info-alist | |
1708 | (cons 'choice (mapcar | |
1709 | (lambda (lang) | |
756e055f MR |
1710 | (list 'const lang)) |
1711 | (sort (mapcar 'car language-info-alist) 'string<))) | |
990a4108 | 1712 | 'string) |
8861c593 | 1713 | :initialize 'custom-initialize-default |
eb9fc9e6 | 1714 | :group 'mule) |
f850d782 | 1715 | |
ddb5c041 KH |
1716 | (defun reset-language-environment () |
1717 | "Reset multilingual environment of Emacs to the default status. | |
1718 | ||
1719 | The default status is as follows: | |
1720 | ||
d37ef0f6 | 1721 | The default value of `buffer-file-coding-system' is nil. |
ddb5c041 KH |
1722 | The default coding system for process I/O is nil. |
1723 | The default value for the command `set-terminal-coding-system' is nil. | |
1724 | The default value for the command `set-keyboard-coding-system' is nil. | |
1725 | ||
97941b05 KH |
1726 | The order of priorities of coding systems are as follows: |
1727 | utf-8 | |
1728 | iso-2022-7bit | |
1729 | iso-latin-1 | |
1730 | iso-2022-7bit-lock | |
1731 | iso-2022-8bit-ss2 | |
1732 | emacs-mule | |
1733 | raw-text" | |
ddb5c041 KH |
1734 | (interactive) |
1735 | ;; This function formerly set default-enable-multibyte-characters to t, | |
1736 | ;; but that is incorrect. It should not alter the unibyte/multibyte choice. | |
1737 | ||
97941b05 KH |
1738 | (set-coding-system-priority |
1739 | 'utf-8 | |
1740 | 'iso-2022-7bit | |
1741 | 'iso-latin-1 | |
1742 | 'iso-2022-7bit-lock | |
1743 | 'iso-2022-8bit-ss2 | |
1744 | 'emacs-mule | |
1745 | 'raw-text) | |
91693d18 | 1746 | |
ddb5c041 | 1747 | (set-default-coding-systems nil) |
b5edd1d1 | 1748 | (setq default-sendmail-coding-system 'iso-latin-1) |
a41118cc SM |
1749 | ;; On Darwin systems, this should be utf-8, but when this file is loaded |
1750 | ;; utf-8 is not yet defined, so we set it in set-locale-environment instead. | |
787caf99 | 1751 | (setq default-file-name-coding-system 'iso-latin-1) |
1d77e15a JR |
1752 | ;; Preserve eol-type from existing default-process-coding-systems. |
1753 | ;; On non-unix-like systems in particular, these may have been set | |
1754 | ;; carefully by the user, or by the startup code, to deal with the | |
1755 | ;; users shell appropriately, so should not be altered by changing | |
1756 | ;; language environment. | |
1757 | (let ((output-coding | |
a099a2ff JR |
1758 | ;; When bootstrapping, coding-systems are not defined yet, so |
1759 | ;; we need to catch the error from check-coding-system. | |
d37ef0f6 | 1760 | (condition-case nil |
a099a2ff JR |
1761 | (coding-system-change-text-conversion |
1762 | (car default-process-coding-system) 'undecided) | |
1763 | (coding-system-error 'undecided))) | |
1d77e15a | 1764 | (input-coding |
a099a2ff JR |
1765 | (condition-case nil |
1766 | (coding-system-change-text-conversion | |
1767 | (cdr default-process-coding-system) 'iso-latin-1) | |
1768 | (coding-system-error 'iso-latin-1)))) | |
1d77e15a JR |
1769 | (setq default-process-coding-system |
1770 | (cons output-coding input-coding))) | |
b5edd1d1 | 1771 | |
ddb5c041 KH |
1772 | ;; Don't alter the terminal and keyboard coding systems here. |
1773 | ;; The terminal still supports the same coding system | |
1774 | ;; that it supported a minute ago. | |
1b8dc791 SM |
1775 | ;; (set-terminal-coding-system-internal nil) |
1776 | ;; (set-keyboard-coding-system-internal nil) | |
ddb5c041 | 1777 | |
7dd42fb1 | 1778 | (set-unibyte-charset 'iso-8859-1)) |
ddb5c041 | 1779 | |
0c47a7c8 KH |
1780 | (reset-language-environment) |
1781 | ||
97c57fb2 | 1782 | (defun set-display-table-and-terminal-coding-system (language-name &optional coding-system display) |
40c81f74 PE |
1783 | "Set up the display table and terminal coding system for LANGUAGE-NAME." |
1784 | (let ((coding (get-language-info language-name 'unibyte-display))) | |
735b7c87 KH |
1785 | (if (and coding |
1786 | (or (not coding-system) | |
1787 | (coding-system-equal coding coding-system))) | |
40c81f74 | 1788 | (standard-display-european-internal) |
a67ae60e EZ |
1789 | ;; The following 2 lines undo the 8-bit display that we set up |
1790 | ;; in standard-display-european-internal, which see. This is in | |
1791 | ;; case the user has used standard-display-european earlier in | |
1792 | ;; this session. (The MS-DOS port doesn't use that setup, so it | |
1793 | ;; doesn't need to undo it.) | |
d73a7bb8 JPW |
1794 | (when standard-display-table |
1795 | (dotimes (i 128) | |
b221615b | 1796 | (aset standard-display-table (+ i 128) nil)))) |
40c81f74 | 1797 | (or (eq window-system 'pc) |
97c57fb2 | 1798 | (set-terminal-coding-system (or coding-system coding) display)))) |
40c81f74 | 1799 | |
166246f7 | 1800 | (defun set-language-environment (language-name) |
6c05d680 RS |
1801 | "Set up multi-lingual environment for using LANGUAGE-NAME. |
1802 | This sets the coding system priority and the default input method | |
8861c593 RS |
1803 | and sometimes other things. LANGUAGE-NAME should be a string |
1804 | which is the name of a language environment. For example, \"Latin-1\" | |
1805 | specifies the character set for the major languages of Western Europe." | |
8efc03e1 | 1806 | (interactive (list (read-language-name |
ddb5c041 | 1807 | nil |
5b76833f | 1808 | "Set language environment (default English): "))) |
4ef06f75 KH |
1809 | (if language-name |
1810 | (if (symbolp language-name) | |
1811 | (setq language-name (symbol-name language-name))) | |
1812 | (setq language-name "English")) | |
f15466c5 | 1813 | (let ((slot (assoc-string language-name language-info-alist t))) |
95498fd0 | 1814 | (unless slot |
f850d782 | 1815 | (error "Language environment not defined: %S" language-name)) |
95498fd0 | 1816 | (setq language-name (car slot))) |
8efc03e1 KH |
1817 | (if current-language-environment |
1818 | (let ((func (get-language-info current-language-environment | |
1819 | 'exit-function))) | |
e63645c2 | 1820 | (run-hooks 'exit-language-environment-hook) |
5f395df3 | 1821 | (if (functionp func) (funcall func)))) |
ddb5c041 | 1822 | |
e9a8ed3c KH |
1823 | (reset-language-environment) |
1824 | ;; The features might set up coding systems. | |
ddb5c041 KH |
1825 | (let ((required-features (get-language-info language-name 'features))) |
1826 | (while required-features | |
1827 | (require (car required-features)) | |
1828 | (setq required-features (cdr required-features)))) | |
6b61353c | 1829 | |
e9a8ed3c KH |
1830 | (setq current-language-environment language-name) |
1831 | ||
1832 | (set-language-environment-coding-systems language-name) | |
1833 | (set-language-environment-input-method language-name) | |
1834 | (set-language-environment-nonascii-translation language-name) | |
1835 | (set-language-environment-charset language-name) | |
1836 | ;; Unibyte setups if necessary. | |
1837 | (unless default-enable-multibyte-characters | |
1838 | (set-language-environment-unibyte language-name)) | |
1839 | ||
ddb5c041 | 1840 | (let ((func (get-language-info language-name 'setup-function))) |
5f395df3 | 1841 | (if (functionp func) |
ddb5c041 | 1842 | (funcall func))) |
e9a8ed3c | 1843 | |
8efc03e1 | 1844 | (run-hooks 'set-language-environment-hook) |
f850d782 | 1845 | (force-mode-line-update t)) |
4ed46869 | 1846 | |
c3869589 | 1847 | (define-widget 'charset 'symbol |
1644d5b9 DL |
1848 | "An Emacs charset." |
1849 | :tag "Charset" | |
c3869589 DL |
1850 | :complete-function (lambda () |
1851 | (interactive) | |
1852 | (lisp-complete-symbol 'charsetp)) | |
1853 | :completion-ignore-case t | |
1854 | :value 'ascii | |
1855 | :validate (lambda (widget) | |
1856 | (unless (charsetp (widget-value widget)) | |
1857 | (widget-put widget :error (format "Invalid charset: %S" | |
1858 | (widget-value widget))) | |
1859 | widget)) | |
1860 | :prompt-history 'charset-history) | |
1861 | ||
1862 | (defcustom language-info-custom-alist nil | |
1863 | "Customizations of language environment parameters. | |
1864 | Value is an alist with elements like those of `language-info-alist'. | |
1865 | These are used to set values in `language-info-alist' which replace | |
1866 | the defaults. A typical use is replacing the default input method for | |
1867 | the environment. Use \\[describe-language-environment] to find the environment's settings. | |
1868 | ||
1869 | This option is intended for use at startup. Removing items doesn't | |
1870 | remove them from the language info until you next restart Emacs. | |
1871 | ||
1872 | Setting this variable directly does not take effect. See | |
1873 | `set-language-info-alist' for use in programs." | |
1874 | :group 'mule | |
8589dc17 | 1875 | :version "23.1" |
c3869589 DL |
1876 | :set (lambda (s v) |
1877 | (custom-set-default s v) | |
1878 | ;; Can't do this before language environments are set up. | |
1879 | (when v | |
1880 | ;; modify language-info-alist | |
1881 | (dolist (elt v) | |
1882 | (set-language-info-alist (car elt) (cdr elt))) | |
1883 | ;; re-set the environment in case its parameters changed | |
1884 | (set-language-environment current-language-environment))) | |
1885 | :type `(alist | |
1886 | :key-type (string :tag "Language environment" | |
1887 | :completion-ignore-case t | |
1888 | :complete-function widget-string-complete | |
1889 | :completion-alist language-info-alist) | |
1890 | :value-type | |
1891 | (alist :key-type symbol | |
1892 | :options ((documentation string) | |
1893 | (charset (repeat charset)) | |
1894 | (sample-text string) | |
1895 | (setup-function function) | |
1896 | (exit-function function) | |
1897 | (coding-system (repeat coding-system)) | |
1898 | (coding-priority (repeat coding-system)) | |
1899 | (nonascii-translation charset) | |
1900 | (input-method | |
1901 | (string | |
1902 | :completion-ignore-case t | |
1903 | :complete-function widget-string-complete | |
1904 | :completion-alist input-method-alist | |
1905 | :prompt-history input-method-history)) | |
1906 | (features (repeat symbol)) | |
1907 | (unibyte-display coding-system))))) | |
1908 | ||
51a8fc1d RS |
1909 | (defun standard-display-european-internal () |
1910 | ;; Actually set up direct output of non-ASCII characters. | |
03c35c83 EZ |
1911 | (standard-display-8bit (if (eq window-system 'pc) 128 160) 255) |
1912 | ;; Unibyte Emacs on MS-DOS wants to display all 8-bit characters with | |
1913 | ;; the native font, and codes 160 and 146 stand for something very | |
1914 | ;; different there. | |
1915 | (or (and (eq window-system 'pc) (not default-enable-multibyte-characters)) | |
1916 | (progn | |
6b626913 SM |
1917 | ;; Most X fonts used to do the wrong thing for latin-1 code 160. |
1918 | (unless (and (eq window-system 'x) | |
1919 | ;; XFree86 4 has fixed the fonts. | |
1920 | (string= "The XFree86 Project, Inc" (x-server-vendor)) | |
1921 | (> (aref (number-to-string (nth 2 (x-server-version))) 0) | |
1922 | ?3)) | |
1923 | ;; Make non-line-break space display as a plain space. | |
1924 | (aset standard-display-table 160 [32])) | |
1492f7ac | 1925 | ;; Most Windows programs send out apostrophes as \222. Most X fonts |
03c35c83 | 1926 | ;; don't contain a character at that position. Map it to the ASCII |
5f395df3 SM |
1927 | ;; apostrophe. [This is actually RIGHT SINGLE QUOTATION MARK, |
1928 | ;; U+2019, normally from the windows-1252 character set. XFree 4 | |
1929 | ;; fonts probably have the appropriate glyph at this position, | |
1930 | ;; so they could use standard-display-8bit. It's better to use a | |
1931 | ;; proper windows-1252 coding system. --fx] | |
6b626913 | 1932 | (aset standard-display-table 146 [39])))) |
03c35c83 | 1933 | |
e9a8ed3c KH |
1934 | (defun set-language-environment-coding-systems (language-name) |
1935 | "Do various coding system setups for language environment LANGUAGE-NAME." | |
54b226f7 | 1936 | (let* ((priority (get-language-info language-name 'coding-priority)) |
e9a8ed3c KH |
1937 | (default-coding (car priority)) |
1938 | (eol-type (coding-system-eol-type default-buffer-file-coding-system))) | |
97941b05 KH |
1939 | (when priority |
1940 | (set-default-coding-systems | |
1941 | (if (memq eol-type '(0 1 2 unix dos mac)) | |
1942 | (coding-system-change-eol-conversion default-coding eol-type) | |
1943 | default-coding)) | |
1944 | (setq default-sendmail-coding-system default-coding) | |
1945 | (apply 'set-coding-system-priority priority)))) | |
54b226f7 | 1946 | |
d042f8b4 KH |
1947 | (defun set-language-environment-input-method (language-name) |
1948 | "Do various input method setups for language environment LANGUAGE-NAME." | |
1949 | (let ((input-method (get-language-info language-name 'input-method))) | |
1950 | (when input-method | |
1951 | (setq default-input-method input-method) | |
1952 | (if input-method-history | |
1953 | (setq input-method-history | |
1954 | (cons input-method | |
1955 | (delete input-method input-method-history))))))) | |
1956 | ||
1957 | (defun set-language-environment-nonascii-translation (language-name) | |
1958 | "Do unibyte/multibyte translation setup for language environment LANGUAGE-NAME." | |
e9a8ed3c KH |
1959 | ;; Note: For DOS, we assumed that the charset cpXXX is already |
1960 | ;; defined. | |
1961 | (let ((nonascii (get-language-info language-name 'nonascii-translation))) | |
1962 | (if (eq window-system 'pc) | |
1963 | (setq nonascii (intern "cp%d" dos-codepage))) | |
1964 | (or (and (charsetp nonascii) | |
1965 | (get-charset-property nonascii :ascii-compatible-p)) | |
1966 | (setq nonascii 'iso-8859-1)) | |
1967 | (set-unibyte-charset nonascii))) | |
d042f8b4 KH |
1968 | |
1969 | (defun set-language-environment-charset (language-name) | |
1970 | "Do various charset setups for language environment LANGUAGE-NAME." | |
e9a8ed3c KH |
1971 | ;; Put higher priorities to such charsets that are supported by the |
1972 | ;; coding systems of higher priorities in this environment. | |
1973 | (let ((charsets (get-language-info language-name 'charset))) | |
1974 | (dolist (coding (get-language-info language-name 'coding-priority)) | |
1975 | (setq charsets (append charsets (coding-system-charset-list coding)))) | |
1976 | (if charsets | |
1977 | (apply 'set-charset-priority charsets)))) | |
d042f8b4 KH |
1978 | |
1979 | (defun set-language-environment-unibyte (language-name) | |
1980 | "Do various unibyte-mode setups for language environment LANGUAGE-NAME." | |
d042f8b4 KH |
1981 | (set-display-table-and-terminal-coding-system language-name)) |
1982 | ||
4ed46869 | 1983 | (defsubst princ-list (&rest args) |
caff3c0a | 1984 | "Print all arguments with `princ', then print \"\\n\"." |
4ed46869 KH |
1985 | (while args (princ (car args)) (setq args (cdr args))) |
1986 | (princ "\n")) | |
1987 | ||
b0648a00 RS |
1988 | (put 'describe-specified-language-support 'apropos-inhibit t) |
1989 | ||
6b61353c | 1990 | ;; Print language-specific information such as input methods, |
13e82c04 | 1991 | ;; charsets, and coding systems. This function is intended to be |
48082651 | 1992 | ;; called from the menu: |
281d03ec | 1993 | ;; [menu-bar mule describe-language-environment LANGUAGE] |
48082651 KH |
1994 | ;; and should not run it by `M-x describe-current-input-method-function'. |
1995 | (defun describe-specified-language-support () | |
96db204a | 1996 | "Describe how Emacs supports the specified language environment." |
48082651 | 1997 | (interactive) |
281d03ec | 1998 | (let (language-name) |
48082651 | 1999 | (if (not (and (symbolp last-command-event) |
cda74479 DL |
2000 | (or (not (eq last-command-event 'Default)) |
2001 | (setq last-command-event 'English)) | |
281d03ec | 2002 | (setq language-name (symbol-name last-command-event)))) |
48082651 | 2003 | (error "Bogus calling sequence")) |
281d03ec RS |
2004 | (describe-language-environment language-name))) |
2005 | ||
2006 | (defun describe-language-environment (language-name) | |
2007 | "Describe how Emacs supports language environment LANGUAGE-NAME." | |
78754934 KH |
2008 | (interactive |
2009 | (list (read-language-name | |
2010 | 'documentation | |
5b76833f | 2011 | "Describe language environment (default current choice): "))) |
f850d782 RS |
2012 | (if (null language-name) |
2013 | (setq language-name current-language-environment)) | |
281d03ec RS |
2014 | (if (or (null language-name) |
2015 | (null (get-language-info language-name 'documentation))) | |
2016 | (error "No documentation for the specified language")) | |
4ef06f75 KH |
2017 | (if (symbolp language-name) |
2018 | (setq language-name (symbol-name language-name))) | |
ef5a4730 KH |
2019 | (dolist (feature (get-language-info language-name 'features)) |
2020 | (require feature)) | |
6b626913 | 2021 | (let ((doc (get-language-info language-name 'documentation))) |
c3034e84 SM |
2022 | (help-setup-xref (list #'describe-language-environment language-name) |
2023 | (interactive-p)) | |
2024 | (with-output-to-temp-buffer (help-buffer) | |
464cc130 KH |
2025 | (save-excursion |
2026 | (set-buffer standard-output) | |
2027 | (insert language-name " language environment\n\n") | |
2028 | (if (stringp doc) | |
2029 | (insert doc "\n\n")) | |
e036b0a6 KH |
2030 | (condition-case nil |
2031 | (let ((str (eval (get-language-info language-name 'sample-text)))) | |
2032 | (if (stringp str) | |
2033 | (insert "Sample text:\n " str "\n\n"))) | |
2034 | (error nil)) | |
464cc130 | 2035 | (let ((input-method (get-language-info language-name 'input-method)) |
acd1b9bd KH |
2036 | (l (copy-sequence input-method-alist)) |
2037 | (first t)) | |
2038 | (when (and input-method | |
2039 | (setq input-method (assoc input-method l))) | |
2040 | (insert "Input methods (default " (car input-method) ")\n") | |
2041 | (setq l (cons input-method (delete input-method l)) | |
2042 | first nil)) | |
2043 | (dolist (elt l) | |
2044 | (when (or (eq input-method elt) | |
2045 | (eq t (compare-strings language-name nil nil | |
2046 | (nth 1 elt) nil nil t))) | |
2047 | (when first | |
2048 | (insert "Input methods:\n") | |
2049 | (setq first nil)) | |
2050 | (insert " " (car elt)) | |
2051 | (search-backward (car elt)) | |
2052 | (help-xref-button 0 'help-input-method (car elt)) | |
464cc130 | 2053 | (goto-char (point-max)) |
2fa7e202 | 2054 | (insert " (\"" |
acd1b9bd KH |
2055 | (if (stringp (nth 3 elt)) (nth 3 elt) (car (nth 3 elt))) |
2056 | "\" in mode line)\n"))) | |
2057 | (or first | |
2058 | (insert "\n"))) | |
464cc130 KH |
2059 | (insert "Character sets:\n") |
2060 | (let ((l (get-language-info language-name 'charset))) | |
2061 | (if (null l) | |
2062 | (insert " nothing specific to " language-name "\n") | |
2063 | (while l | |
2064 | (insert " " (symbol-name (car l))) | |
2065 | (search-backward (symbol-name (car l))) | |
467412aa | 2066 | (help-xref-button 0 'help-character-set (car l)) |
464cc130 KH |
2067 | (goto-char (point-max)) |
2068 | (insert ": " (charset-description (car l)) "\n") | |
2069 | (setq l (cdr l))))) | |
2070 | (insert "\n") | |
2071 | (insert "Coding systems:\n") | |
2072 | (let ((l (get-language-info language-name 'coding-system))) | |
2073 | (if (null l) | |
2074 | (insert " nothing specific to " language-name "\n") | |
2075 | (while l | |
2076 | (insert " " (symbol-name (car l))) | |
2077 | (search-backward (symbol-name (car l))) | |
467412aa | 2078 | (help-xref-button 0 'help-coding-system (car l)) |
464cc130 KH |
2079 | (goto-char (point-max)) |
2080 | (insert " (`" | |
2081 | (coding-system-mnemonic (car l)) | |
2082 | "' in mode line):\n\t" | |
2083 | (coding-system-doc-string (car l)) | |
2084 | "\n") | |
0855c6cd | 2085 | (let ((aliases (coding-system-aliases (car l)))) |
464cc130 KH |
2086 | (when aliases |
2087 | (insert "\t(alias:") | |
2088 | (while aliases | |
2089 | (insert " " (symbol-name (car aliases))) | |
2090 | (setq aliases (cdr aliases))) | |
2091 | (insert ")\n"))) | |
c3034e84 | 2092 | (setq l (cdr l))))))))) |
4ed46869 | 2093 | \f |
40c81f74 PE |
2094 | ;;; Locales. |
2095 | ||
0d7c5bb9 DL |
2096 | (defvar locale-translation-file-name nil |
2097 | "File name for the system's file of locale-name aliases, or nil if none.") | |
40c81f74 | 2098 | |
5f395df3 SM |
2099 | ;; The following definitions might as well be marked as constants and |
2100 | ;; purecopied, since they're normally used on startup, and probably | |
2101 | ;; should reflect the facilities of the base Emacs. | |
2102 | (defconst locale-language-names | |
2103 | (purecopy | |
2104 | '( | |
40c81f74 PE |
2105 | ;; Locale names of the form LANGUAGE[_TERRITORY][.CODESET][@MODIFIER] |
2106 | ;; as specified in the Single Unix Spec, Version 2. | |
2107 | ;; LANGUAGE is a language code taken from ISO 639:1988 (E/F) | |
2108 | ;; with additions from ISO 639/RA Newsletter No.1/1989; | |
5f395df3 SM |
2109 | ;; see Internet RFC 2165 (1997-06) and |
2110 | ;; http://www.evertype.com/standards/iso639/iso639-en.html | |
2111 | ;; TERRITORY is a country code taken from ISO 3166 | |
2112 | ;; http://www.din.de/gremien/nas/nabd/iso3166ma/codlstp1/en_listp1.html. | |
40c81f74 | 2113 | ;; CODESET and MODIFIER are implementation-dependent. |
5f395df3 | 2114 | |
6b61353c KH |
2115 | ;; jasonr comments: MS Windows uses three letter codes for |
2116 | ;; languages instead of the two letter ISO codes that POSIX | |
2117 | ;; uses. In most cases the first two letters are the same, so | |
2118 | ;; most of the regexps in locale-language-names work. Japanese | |
2119 | ;; and Chinese are exceptions, which are listed in the | |
2120 | ;; non-standard section at the bottom of locale-language-names. | |
2121 | ||
8dedddd5 KH |
2122 | ("aa_DJ" . "Latin-1") ; Afar |
2123 | ("aa" . "UTF-8") | |
2124 | ;; ab Abkhazian | |
6ececc4d | 2125 | ("af" . "Latin-1") ; Afrikaans |
8dedddd5 KH |
2126 | ("am" "Ethiopic" utf-8) ; Amharic |
2127 | ("an" . "Latin-9") ; Aragonese | |
5f395df3 | 2128 | ; ar Arabic glibc uses 8859-6 |
40c81f74 PE |
2129 | ; as Assamese |
2130 | ; ay Aymara | |
8dedddd5 | 2131 | ("az" . "UTF-8") ; Azerbaijani |
40c81f74 | 2132 | ; ba Bashkir |
8dedddd5 KH |
2133 | ("be" "Belarusian" cp1251) ; Belarusian [Byelorussian until early 1990s] |
2134 | ("bg" "Bulgarian" cp1251) ; Bulgarian | |
40c81f74 PE |
2135 | ; bh Bihari |
2136 | ; bi Bislama | |
8dedddd5 | 2137 | ("bn" . "UTF-8") ; Bengali, Bangla |
40c81f74 PE |
2138 | ("bo" . "Tibetan") |
2139 | ("br" . "Latin-1") ; Breton | |
d37ef0f6 | 2140 | ("bs" . "Latin-2") ; Bosnian |
8dedddd5 | 2141 | ("byn" . "UTF-8") ; Bilin; Blin |
40c81f74 PE |
2142 | ("ca" . "Latin-1") ; Catalan |
2143 | ; co Corsican | |
8dedddd5 KH |
2144 | ("cs" "Czech" iso-8859-2) |
2145 | ("cy" "Welsh" iso-8859-14) | |
40c81f74 | 2146 | ("da" . "Latin-1") ; Danish |
8dedddd5 | 2147 | ("de" "German" iso-8859-1) |
569a6374 | 2148 | ; dv Divehi |
40c81f74 | 2149 | ; dz Bhutani |
8dedddd5 | 2150 | ("el" "Greek" iso-8859-7) |
6ececc4d | 2151 | ;; Users who specify "en" explicitly typically want Latin-1, not ASCII. |
e11cf111 DL |
2152 | ;; That's actually what the GNU locales define, modulo things like |
2153 | ;; en_IN -- fx. | |
8dedddd5 | 2154 | ("en_IN" "English" utf-8) ; glibc uses utf-8 for English in India |
eec5c8f9 | 2155 | ("en" "English" iso-8859-1) ; English |
58802cd7 | 2156 | ("eo" . "Esperanto") ; Esperanto |
8dedddd5 KH |
2157 | ("es" "Spanish" iso-8859-1) |
2158 | ("et" . "Latin-1") ; Estonian | |
40c81f74 | 2159 | ("eu" . "Latin-1") ; Basque |
8dedddd5 | 2160 | ("fa" . "UTF-8") ; Persian |
40c81f74 | 2161 | ("fi" . "Latin-1") ; Finnish |
8dedddd5 | 2162 | ("fj" . "Latin-1") ; Fiji |
40c81f74 | 2163 | ("fo" . "Latin-1") ; Faroese |
8dedddd5 | 2164 | ("fr" "French" iso-8859-1) ; French |
40c81f74 | 2165 | ("fy" . "Latin-1") ; Frisian |
6ececc4d | 2166 | ("ga" . "Latin-1") ; Irish Gaelic (new orthography) |
8dedddd5 KH |
2167 | ("gd" . "Latin-9") ; Scots Gaelic |
2168 | ("gez" "Ethiopic" utf-8) ; Geez | |
2169 | ("gl" . "Latin-1") ; Gallegan; Galician | |
40c81f74 | 2170 | ; gn Guarani |
8dedddd5 KH |
2171 | ("gu" . "UTF-8") ; Gujarati |
2172 | ("gv" . "Latin-1") ; Manx Gaelic | |
40c81f74 | 2173 | ; ha Hausa |
8dedddd5 KH |
2174 | ("he" "Hebrew" iso-8859-8) |
2175 | ("hi" "Devanagari" utf-8) ; Hindi | |
2176 | ("hr" "Croatian" iso-8859-2) ; Croatian | |
40c81f74 PE |
2177 | ("hu" . "Latin-2") ; Hungarian |
2178 | ; hy Armenian | |
2179 | ; ia Interlingua | |
2180 | ("id" . "Latin-1") ; Indonesian | |
2181 | ; ie Interlingue | |
2182 | ; ik Inupiak | |
2183 | ("is" . "Latin-1") ; Icelandic | |
8dedddd5 | 2184 | ("it" "Italian" iso-8859-1) ; Italian |
40c81f74 | 2185 | ; iu Inuktitut |
8dedddd5 KH |
2186 | ("iw" "Hebrew" iso-8859-8) |
2187 | ("ja" "Japanese" euc-jp) | |
40c81f74 | 2188 | ; jw Javanese |
8dedddd5 | 2189 | ("ka" "Georgian" georgian-ps) ; Georgian |
40c81f74 | 2190 | ; kk Kazakh |
6ececc4d | 2191 | ("kl" . "Latin-1") ; Greenlandic |
40c81f74 | 2192 | ; km Cambodian |
8dedddd5 KH |
2193 | ("kn" "Kannada" utf-8) |
2194 | ("ko" "Korean" euc-kr) | |
40c81f74 PE |
2195 | ; ks Kashmiri |
2196 | ; ku Kurdish | |
5f395df3 | 2197 | ("kw" . "Latin-1") ; Cornish |
40c81f74 PE |
2198 | ; ky Kirghiz |
2199 | ("la" . "Latin-1") ; Latin | |
5f395df3 | 2200 | ("lb" . "Latin-1") ; Luxemburgish |
8dedddd5 | 2201 | ("lg" . "Laint-6") ; Ganda |
40c81f74 | 2202 | ; ln Lingala |
8dedddd5 KH |
2203 | ("lo" "Lao" utf-8) ; Laothian |
2204 | ("lt" "Lithuanian" iso-8859-13) | |
9c20a8d5 | 2205 | ("lv" . "Latvian") ; Latvian, Lettish |
40c81f74 | 2206 | ; mg Malagasy |
5f395df3 | 2207 | ("mi" . "Latin-7") ; Maori |
8dedddd5 KH |
2208 | ("mk" "Cyrillic-ISO" iso-8859-5) ; Macedonian |
2209 | ("ml" "Malayalam" utf-8) | |
2210 | ("mn" . "UTF-8") ; Mongolian | |
40c81f74 | 2211 | ; mo Moldavian |
8dedddd5 | 2212 | ("mr" "Devanagari" utf-8) ; Marathi |
5f395df3 | 2213 | ("ms" . "Latin-1") ; Malay |
40c81f74 PE |
2214 | ("mt" . "Latin-3") ; Maltese |
2215 | ; my Burmese | |
2216 | ; na Nauru | |
8dedddd5 KH |
2217 | ("nb" . "Latin-1") ; Norwegian |
2218 | ("ne" "Devanagari" utf-8) ; Nepali | |
2219 | ("nl" "Dutch" iso-8859-1) | |
40c81f74 | 2220 | ("no" . "Latin-1") ; Norwegian |
5f395df3 | 2221 | ("oc" . "Latin-1") ; Occitan |
8dedddd5 KH |
2222 | ("om_ET" . "UTF-8") ; (Afan) Oromo |
2223 | ("om" . "Latin-1") ; (Afan) Oromo | |
40c81f74 | 2224 | ; or Oriya |
8dedddd5 | 2225 | ("pa" . "UTF-8") ; Punjabi |
40c81f74 PE |
2226 | ("pl" . "Latin-2") ; Polish |
2227 | ; ps Pashto, Pushto | |
2228 | ("pt" . "Latin-1") ; Portuguese | |
2229 | ; qu Quechua | |
6ececc4d | 2230 | ("rm" . "Latin-1") ; Rhaeto-Romanic |
40c81f74 | 2231 | ; rn Kirundi |
8dedddd5 KH |
2232 | ("ro" "Romanian" iso-8859-2) |
2233 | ("ru_RU" "Russian" iso-8859-5) | |
2234 | ("ru_UA" "Russian" koi8-u) | |
40c81f74 PE |
2235 | ; rw Kinyarwanda |
2236 | ("sa" . "Devanagari") ; Sanskrit | |
2237 | ; sd Sindhi | |
8dedddd5 | 2238 | ("se" . "UTF-8") ; Northern Sami |
40c81f74 PE |
2239 | ; sg Sangho |
2240 | ("sh" . "Latin-2") ; Serbo-Croatian | |
2241 | ; si Sinhalese | |
8dedddd5 KH |
2242 | ("sid" . "UTF-8") ; Sidamo |
2243 | ("sk" "Slovak" iso-8859-2) | |
2244 | ("sl" "Slovenian" iso-8859-2) | |
40c81f74 PE |
2245 | ; sm Samoan |
2246 | ; sn Shona | |
8dedddd5 KH |
2247 | ("so_ET" "UTF-8") ; Somali |
2248 | ("so" "Latin-1") ; Somali | |
6ececc4d | 2249 | ("sq" . "Latin-1") ; Albanian |
40c81f74 PE |
2250 | ("sr" . "Latin-2") ; Serbian (Latin alphabet) |
2251 | ; ss Siswati | |
8dedddd5 | 2252 | ("st" . "Latin-1") ; Sesotho |
40c81f74 | 2253 | ; su Sundanese |
8dedddd5 | 2254 | ("sv" "Swedish" iso-8859-1) ; Swedish |
40c81f74 | 2255 | ("sw" . "Latin-1") ; Swahili |
8dedddd5 KH |
2256 | ("ta" "Tamil" utf-8) |
2257 | ("te" . "UTF-8") ; Telugu | |
2258 | ("tg" "Tajik" koi8-t) | |
2259 | ("th" "Thai" tis-620) | |
2260 | ("ti" "Ethiopic" utf-8) ; Tigrinya | |
2261 | ("tig_ER" . "UTF-8") ; Tigre | |
40c81f74 | 2262 | ; tk Turkmen |
6ececc4d | 2263 | ("tl" . "Latin-1") ; Tagalog |
40c81f74 PE |
2264 | ; tn Setswana |
2265 | ; to Tonga | |
8dedddd5 | 2266 | ("tr" "Turkish" iso-8859-9) |
40c81f74 | 2267 | ; ts Tsonga |
8dedddd5 | 2268 | ("tt" . "UTF-8") ; Tatar |
40c81f74 PE |
2269 | ; tw Twi |
2270 | ; ug Uighur | |
8dedddd5 KH |
2271 | ("uk" "Ukrainian" koi8-u) |
2272 | ("ur" . "UTF-8") ; Urdu | |
2273 | ("uz_UZ@cyrillic" . "UTF-8"); Uzbek | |
5f395df3 | 2274 | ("uz" . "Latin-1") ; Uzbek |
8dedddd5 | 2275 | ("vi" "Vietnamese" utf-8) |
40c81f74 | 2276 | ; vo Volapuk |
d37ef0f6 | 2277 | ("wa" . "Latin-1") ; Walloon |
40c81f74 | 2278 | ; wo Wolof |
8dedddd5 | 2279 | ("xh" . "Latin-1") ; Xhosa |
9c20a8d5 | 2280 | ("yi" . "Windows-1255") ; Yiddish |
40c81f74 PE |
2281 | ; yo Yoruba |
2282 | ; za Zhuang | |
8dedddd5 | 2283 | ("zh_HK" . "Chinese-Big5") |
05b81b42 | 2284 | ; zh_HK/BIG5-HKSCS \ |
8dedddd5 | 2285 | ("zh_TW" . "Chinese-Big5") |
05b81b42 KH |
2286 | ("zh_CN.GB2312" "Chinese-GB") |
2287 | ("zh_CN.GBK" "Chinese-GBK") | |
2288 | ("zh_CN.GB18030" "Chinese-GB18030") | |
f21605fa | 2289 | ("zh_CN.UTF-8" . "Chinese-GBK") |
8dedddd5 | 2290 | ("zh_CN" . "Chinese-GB") |
40c81f74 | 2291 | ("zh" . "Chinese-GB") |
8dedddd5 | 2292 | ("zu" . "Latin-1") ; Zulu |
40c81f74 PE |
2293 | |
2294 | ;; ISO standard locales | |
2295 | ("c$" . "ASCII") | |
2296 | ("posix$" . "ASCII") | |
2297 | ||
40c81f74 PE |
2298 | ;; The "IPA" Emacs language environment does not correspond |
2299 | ;; to any ISO 639 code, so let it stand for itself. | |
2300 | ("ipa$" . "IPA") | |
2301 | ||
2302 | ;; Nonstandard or obsolete language codes | |
2303 | ("cz" . "Czech") ; e.g. Solaris 2.6 | |
2304 | ("ee" . "Latin-4") ; Estonian, e.g. X11R6.4 | |
2305 | ("iw" . "Hebrew") ; e.g. X11R6.4 | |
f1282c7f | 2306 | ("sp" . "Cyrillic-ISO") ; Serbian (Cyrillic alphabet), e.g. X11R6.4 |
40c81f74 | 2307 | ("su" . "Latin-1") ; Finnish, e.g. Solaris 2.6 |
2e86ceaa | 2308 | ("jp" . "Japanese") ; e.g. MS Windows |
f21605fa | 2309 | ("chs" . "Chinese-GBK") ; MS Windows Chinese Simplified |
86d9e628 | 2310 | ("cht" . "Chinese-BIG5") ; MS Windows Chinese Traditional |
569a6374 EZ |
2311 | ("gbz" . "UTF-8") ; MS Windows Dari Persian |
2312 | ("div" . "UTF-8") ; MS Windows Divehi (Maldives) | |
2313 | ("wee" . "Latin-2") ; MS Windows Lower Sorbian | |
2314 | ("wen" . "Latin-2") ; MS Windows Upper Sorbian | |
367ca50f | 2315 | )) |
8dedddd5 | 2316 | "Alist of locale regexps vs the corresponding languages and coding systems. |
caff3c0a | 2317 | Each element has this form: |
8dedddd5 KH |
2318 | \(LOCALE-REGEXP LANG-ENV CODING-SYSTEM) |
2319 | The first element whose LOCALE-REGEXP matches the start of a | |
caff3c0a | 2320 | downcased locale specifies the LANG-ENV \(language environment) |
8dedddd5 KH |
2321 | and CODING-SYSTEM corresponding to that locale. If there is no |
2322 | appropriate language environment, the element may have this form: | |
2323 | \(LOCALE-REGEXP . LANG-ENV) | |
2324 | In this case, LANG-ENV is one of generic language environments for an | |
2325 | specific encoding such as \"Latin-1\" and \"UTF-8\".") | |
40c81f74 | 2326 | |
5f395df3 SM |
2327 | (defconst locale-charset-language-names |
2328 | (purecopy | |
2329 | '((".*8859[-_]?1\\>" . "Latin-1") | |
2330 | (".*8859[-_]?2\\>" . "Latin-2") | |
2331 | (".*8859[-_]?3\\>" . "Latin-3") | |
2332 | (".*8859[-_]?4\\>" . "Latin-4") | |
2333 | (".*8859[-_]?9\\>" . "Latin-5") | |
2334 | (".*8859[-_]?14\\>" . "Latin-8") | |
2335 | (".*8859[-_]?15\\>" . "Latin-9") | |
6b61353c KH |
2336 | (".*utf\\(?:-?8\\)?\\>" . "UTF-8") |
2337 | ;; utf-8@euro exists, so put this last. (@euro really specifies | |
2338 | ;; the currency, rather than the charset.) | |
e522f07b | 2339 | (".*@euro\\>" . "Latin-9"))) |
6ececc4d PE |
2340 | "List of pairs of locale regexps and charset language names. |
2341 | The first element whose locale regexp matches the start of a downcased locale | |
6b61353c | 2342 | specifies the language name whose charset corresponds to that locale. |
8dedddd5 | 2343 | This language name is used if the locale is not listed in |
caff3c0a | 2344 | `locale-language-names'.") |
6ececc4d | 2345 | |
5f395df3 SM |
2346 | (defconst locale-preferred-coding-systems |
2347 | (purecopy | |
8dedddd5 KH |
2348 | '((".*8859[-_]?1\\>" . iso-8859-1) |
2349 | (".*8859[-_]?2\\>" . iso-8859-2) | |
2350 | (".*8859[-_]?3\\>" . iso-8859-3) | |
2351 | (".*8859[-_]?4\\>" . iso-8859-4) | |
2352 | (".*8859[-_]?9\\>" . iso-8859-9) | |
2353 | (".*8859[-_]?14\\>" . iso-8859-14) | |
2354 | (".*8859[-_]?15\\>" . iso-8859-15) | |
2355 | (".*utf\\(?:-?8\\)?" . utf-8) | |
2356 | ;; utf-8@euro exists, so put this after utf-8. (@euro really | |
2357 | ;; specifies the currency, rather than the charset.) | |
2358 | (".*@euro" . iso-8859-15) | |
2359 | ("koi8-?r" . koi8-r) | |
2360 | ("koi8-?u" . koi8-u) | |
2361 | ("tcvn" . tcvn) | |
38c05d07 | 2362 | ("big5[-_]?hkscs" . big5-hkscs) |
8dedddd5 KH |
2363 | ("big5" . big5) |
2364 | ("euc-?tw" . euc-tw) | |
38c05d07 KH |
2365 | ("euc-?cn" .euc-cn) |
2366 | ("gb2312" . gb2312) | |
2367 | ("gbk" . gbk) | |
2368 | ("gb18030" . gb18030) | |
8dedddd5 | 2369 | ("ja.*[._]euc" . japanese-iso-8bit) |
5f395df3 SM |
2370 | ("ja.*[._]jis7" . iso-2022-jp) |
2371 | ("ja.*[._]pck" . japanese-shift-jis) | |
2372 | ("ja.*[._]sjis" . japanese-shift-jis) | |
69210880 | 2373 | ("jpn" . japanese-shift-jis) ; MS-Windows uses this. |
8dedddd5 | 2374 | )) |
6ececc4d PE |
2375 | "List of pairs of locale regexps and preferred coding systems. |
2376 | The first element whose locale regexp matches the start of a downcased locale | |
8dedddd5 KH |
2377 | specifies the coding system to prefer when using that locale. |
2378 | This coding system is used if the locale specifies a specific charset.") | |
40c81f74 PE |
2379 | |
2380 | (defun locale-name-match (key alist) | |
2381 | "Search for KEY in ALIST, which should be a list of regexp-value pairs. | |
2382 | Return the value corresponding to the first regexp that matches the | |
2383 | start of KEY, or nil if there is no match." | |
2384 | (let (element) | |
2385 | (while (and alist (not element)) | |
ddf6b836 | 2386 | (if (string-match (concat "\\`\\(?:" (car (car alist)) "\\)") key) |
40c81f74 PE |
2387 | (setq element (car alist))) |
2388 | (setq alist (cdr alist))) | |
2389 | (cdr element))) | |
2390 | ||
58b78d5b | 2391 | (defun locale-charset-match-p (charset1 charset2) |
91ed0599 | 2392 | "Whether charset names (strings) CHARSET1 and CHARSET2 are equivalent. |
58b78d5b DL |
2393 | Matching is done ignoring case and any hyphens and underscores in the |
2394 | names. E.g. `ISO_8859-1' and `iso88591' both match `iso-8859-1'." | |
2395 | (setq charset1 (replace-regexp-in-string "[-_]" "" charset1)) | |
2396 | (setq charset2 (replace-regexp-in-string "[-_]" "" charset2)) | |
2397 | (eq t (compare-strings charset1 nil nil charset2 nil nil t))) | |
2398 | ||
6b61353c KH |
2399 | (defvar locale-charset-alist nil |
2400 | "Coding system alist keyed on locale-style charset name. | |
2401 | Used by `locale-charset-to-coding-system'.") | |
2402 | ||
2403 | (defun locale-charset-to-coding-system (charset) | |
2404 | "Find coding system corresponding to CHARSET. | |
2405 | CHARSET is any sort of non-Emacs charset name, such as might be used | |
2406 | in a locale codeset, or elsewhere. It is matched to a coding system | |
2407 | first by case-insensitive lookup in `locale-charset-alist'. Then | |
2408 | matches are looked for in the coding system list, treating case and | |
2409 | the characters `-' and `_' as insignificant. The coding system base | |
2410 | is returned. Thus, for instance, if charset \"ISO8859-2\", | |
2411 | `iso-latin-2' is returned." | |
f15466c5 | 2412 | (or (car (assoc-string charset locale-charset-alist t)) |
6b61353c KH |
2413 | (let ((cs coding-system-alist) |
2414 | c) | |
2415 | (while (and (not c) cs) | |
2416 | (if (locale-charset-match-p charset (caar cs)) | |
2417 | (setq c (intern (caar cs))) | |
2418 | (pop cs))) | |
2419 | (if c (coding-system-base c))))) | |
2420 | ||
2421 | ;; Fixme: This ought to deal with the territory part of the locale | |
2422 | ;; too, for setting things such as calendar holidays, ps-print paper | |
2423 | ;; size, spelling dictionary. | |
2424 | ||
36ab8612 | 2425 | (defun set-locale-environment (&optional locale-name frame) |
40c81f74 | 2426 | "Set up multi-lingual environment for using LOCALE-NAME. |
758f07de RS |
2427 | This sets the language environment, the coding system priority, |
2428 | the default input method and sometimes other things. | |
2429 | ||
58b78d5b | 2430 | LOCALE-NAME should be a string which is the name of a locale supported |
6b61353c | 2431 | by the system. Often it is of the form xx_XX.CODE, where xx is a |
58b78d5b DL |
2432 | language, XX is a country, and CODE specifies a character set and |
2433 | coding system. For example, the locale name \"ja_JP.EUC\" might name | |
2434 | a locale for Japanese in Japan using the `japanese-iso-8bit' | |
2435 | coding-system. The name may also have a modifier suffix, e.g. `@euro' | |
2436 | or `@cyrillic'. | |
40c81f74 | 2437 | |
758f07de | 2438 | If LOCALE-NAME is nil, its value is taken from the environment |
d37ef0f6 | 2439 | variables LC_ALL, LC_CTYPE and LANG (the first one that is set). |
40c81f74 PE |
2440 | |
2441 | The locale names supported by your system can typically be found in a | |
0812c1e8 | 2442 | directory named `/usr/share/locale' or `/usr/lib/locale'. LOCALE-NAME |
758f07de | 2443 | will be translated according to the table specified by |
0812c1e8 DL |
2444 | `locale-translation-file-name'. |
2445 | ||
36ab8612 MB |
2446 | If FRAME is non-nil, only set the keyboard coding system and the |
2447 | terminal coding system for the terminal of that frame, and don't | |
2448 | touch session-global parameters like the language environment. | |
97c57fb2 | 2449 | |
0812c1e8 DL |
2450 | See also `locale-charset-language-names', `locale-language-names', |
2451 | `locale-preferred-coding-systems' and `locale-coding-system'." | |
758f07de | 2452 | (interactive "sSet environment for locale: ") |
a1506d29 | 2453 | |
0d7c5bb9 DL |
2454 | ;; Do this at runtime for the sake of binaries possibly transported |
2455 | ;; to a system without X. | |
2456 | (setq locale-translation-file-name | |
2457 | (let ((files | |
6bba8c70 KH |
2458 | '("/usr/share/X11/locale/locale.alias" ; e.g. X11R7 |
2459 | "/usr/lib/X11/locale/locale.alias" ; e.g. X11R6.4 | |
6b61353c | 2460 | "/usr/X11R6/lib/X11/locale/locale.alias" ; XFree86, e.g. RedHat 4.2 |
0d7c5bb9 DL |
2461 | "/usr/openwin/lib/locale/locale.alias" ; e.g. Solaris 2.6 |
2462 | ;; | |
2463 | ;; The following name appears after the X-related names above, | |
2464 | ;; since the X-related names are what X actually uses. | |
2465 | "/usr/share/locale/locale.alias" ; GNU/Linux sans X | |
2466 | ))) | |
2467 | (while (and files (not (file-exists-p (car files)))) | |
2468 | (setq files (cdr files))) | |
2469 | (car files))) | |
2470 | ||
7008ccac GM |
2471 | (let ((locale locale-name)) |
2472 | ||
2473 | (unless locale | |
2474 | ;; Use the first of these three environment variables | |
2475 | ;; that has a nonempty value. | |
2476 | (let ((vars '("LC_ALL" "LC_CTYPE" "LANG"))) | |
db6bd804 DL |
2477 | (while (and vars |
2478 | (= 0 (length locale))) ; nil or empty string | |
36ab8612 | 2479 | (setq locale (getenv (pop vars) frame))))) |
7008ccac | 2480 | |
02de72e9 SM |
2481 | (unless locale |
2482 | ;; The two tests are kept separate so the byte-compiler sees | |
2483 | ;; that mac-get-preference is only called after checking its existence. | |
2484 | (when (fboundp 'mac-get-preference) | |
2485 | (setq locale (mac-get-preference "AppleLocale")) | |
2486 | (unless locale | |
2487 | (let ((languages (mac-get-preference "AppleLanguages"))) | |
2488 | (unless (= (length languages) 0) ; nil or empty vector | |
2489 | (setq locale (aref languages 0))))))) | |
cfd67504 YM |
2490 | (unless (or locale (not (boundp 'mac-system-locale))) |
2491 | (setq locale mac-system-locale)) | |
2492 | ||
7008ccac GM |
2493 | (when locale |
2494 | ||
2495 | ;; Translate "swedish" into "sv_SE.ISO8859-1", and so on, | |
2496 | ;; using the translation file that many systems have. | |
2497 | (when locale-translation-file-name | |
2498 | (with-temp-buffer | |
8e9e2c18 | 2499 | (set-buffer-multibyte nil) |
7008ccac GM |
2500 | (insert-file-contents locale-translation-file-name) |
2501 | (when (re-search-forward | |
2502 | (concat "^" (regexp-quote locale) ":?[ \t]+") nil t) | |
2503 | (setq locale (buffer-substring (point) (line-end-position)))))) | |
2504 | ||
2505 | ;; Leave the system locales alone if the caller did not specify | |
2506 | ;; an explicit locale name, as their defaults are set from | |
2507 | ;; LC_MESSAGES and LC_TIME, not LC_CTYPE, and the user might not | |
2508 | ;; want to set them to the same value as LC_CTYPE. | |
2509 | (when locale-name | |
2510 | (setq system-messages-locale locale) | |
2511 | (setq system-time-locale locale)) | |
2512 | ||
2513 | (setq locale (downcase locale)) | |
2514 | ||
2515 | (let ((language-name | |
2516 | (locale-name-match locale locale-language-names)) | |
2517 | (charset-language-name | |
2518 | (locale-name-match locale locale-charset-language-names)) | |
cdb1af30 EZ |
2519 | (default-eol-type (coding-system-eol-type |
2520 | default-buffer-file-coding-system)) | |
7008ccac | 2521 | (coding-system |
6b61353c KH |
2522 | (or (locale-name-match locale locale-preferred-coding-systems) |
2523 | (when locale | |
2524 | (if (string-match "\\.\\([^@]+\\)" locale) | |
2525 | (locale-charset-to-coding-system | |
cfd67504 YM |
2526 | (match-string 1 locale)))) |
2527 | (and (eq system-type 'macos) mac-system-coding-system)))) | |
7008ccac | 2528 | |
8dedddd5 KH |
2529 | (if (consp language-name) |
2530 | ;; locale-language-names specify both lang-env and coding. | |
2531 | ;; But, what specified in locale-preferred-coding-systems | |
2532 | ;; has higher priority. | |
2533 | (setq coding-system (or coding-system | |
2534 | (nth 1 language-name)) | |
2535 | language-name (car language-name)) | |
2536 | ;; Otherwise, if locale is not listed in locale-language-names, | |
2537 | ;; use what listed in locale-charset-language-names. | |
2538 | (if (not language-name) | |
2539 | (setq language-name charset-language-name))) | |
7008ccac | 2540 | |
cdb1af30 EZ |
2541 | ;; If a specific EOL conversion was specified in the default |
2542 | ;; buffer-file-coding-system, preserve it in the coding system | |
2543 | ;; we will be using from now on. | |
8bca692e EZ |
2544 | (if (and (memq default-eol-type '(0 1 2 unix dos mac)) |
2545 | coding-system | |
2546 | (coding-system-p coding-system)) | |
cdb1af30 EZ |
2547 | (setq coding-system (coding-system-change-eol-conversion |
2548 | coding-system default-eol-type))) | |
2549 | ||
7008ccac GM |
2550 | (when language-name |
2551 | ||
2552 | ;; Set up for this character set. This is now the right way | |
2553 | ;; to do it for both unibyte and multibyte modes. | |
36ab8612 | 2554 | (unless frame |
97c57fb2 | 2555 | (set-language-environment language-name)) |
7008ccac GM |
2556 | |
2557 | ;; If default-enable-multibyte-characters is nil, | |
2558 | ;; we are using single-byte characters, | |
2559 | ;; so the display table and terminal coding system are irrelevant. | |
2560 | (when default-enable-multibyte-characters | |
ff76e074 | 2561 | (set-display-table-and-terminal-coding-system |
36ab8612 | 2562 | language-name coding-system frame)) |
7008ccac | 2563 | |
166ce29f DL |
2564 | ;; Set the `keyboard-coding-system' if appropriate (tty |
2565 | ;; only). At least X and MS Windows can generate | |
2566 | ;; multilingual input. | |
68bba4e4 KL |
2567 | ;; XXX This was disabled unless `window-system', but that |
2568 | ;; leads to buggy behaviour when a tty frame is opened | |
2569 | ;; later. Setting the keyboard coding system has no adverse | |
2570 | ;; effect on X, so let's do it anyway. -- Lorentey | |
2571 | (let ((kcs (or coding-system | |
2572 | (car (get-language-info language-name | |
2573 | 'coding-system))))) | |
36ab8612 | 2574 | (if kcs (set-keyboard-coding-system kcs frame))) |
dc2be2fa | 2575 | |
36ab8612 | 2576 | (unless frame |
97c57fb2 KL |
2577 | (setq locale-coding-system |
2578 | (car (get-language-info language-name 'coding-priority))))) | |
7008ccac | 2579 | |
36ab8612 | 2580 | (when (and (not frame) |
97c57fb2 | 2581 | coding-system |
8dedddd5 KH |
2582 | (not (coding-system-equal coding-system |
2583 | locale-coding-system))) | |
7008ccac | 2584 | (prefer-coding-system coding-system) |
218e7ce3 KH |
2585 | ;; Fixme: perhaps prefer-coding-system should set this too. |
2586 | ;; But it's not the time to do such a fundamental change. | |
2587 | (setq default-sendmail-coding-system coding-system) | |
58b78d5b | 2588 | (setq locale-coding-system coding-system)) |
4bc5d766 | 2589 | |
c7b4b466 | 2590 | (when (get-language-info current-language-environment 'coding-priority) |
8f924df7 KH |
2591 | (let ((codeset (locale-info 'codeset)) |
2592 | (coding-system (car (coding-system-priority-list)))) | |
2593 | (when codeset | |
2594 | (let ((cs (coding-system-aliases coding-system)) | |
2595 | result) | |
2596 | (while (and cs (not result)) | |
2597 | (setq result | |
2598 | (locale-charset-match-p (symbol-name (pop cs)) | |
2599 | (locale-info 'codeset)))) | |
2600 | (unless result | |
c7b4b466 | 2601 | (message "Warning: Default coding system `%s' disagrees with |
8f924df7 | 2602 | system codeset `%s' for this locale." coding-system codeset)))))))) |
e76ef161 | 2603 | |
52c7f9ee | 2604 | ;; On Windows, override locale-coding-system, |
26245233 JR |
2605 | ;; default-file-name-coding-system, keyboard-coding-system, |
2606 | ;; terminal-coding-system with system codepage. | |
5eb94383 | 2607 | (when (boundp 'w32-ansi-code-page) |
893b49bb JR |
2608 | (let ((code-page-coding (intern (format "cp%d" w32-ansi-code-page)))) |
2609 | (when (coding-system-p code-page-coding) | |
36ab8612 MB |
2610 | (unless frame (setq locale-coding-system code-page-coding)) |
2611 | (set-keyboard-coding-system code-page-coding frame) | |
d4aa48db | 2612 | (set-terminal-coding-system code-page-coding frame) |
59938af3 EZ |
2613 | ;; Set default-file-name-coding-system last, so that Emacs |
2614 | ;; doesn't try to use cpNNNN when it defines keyboard and | |
2615 | ;; terminal encoding. That's because the above two lines | |
2616 | ;; will want to load code-pages.el, where cpNNNN are | |
2617 | ;; defined; if default-file-name-coding-system were set to | |
2618 | ;; cpNNNN while these two lines run, Emacs will want to use | |
2619 | ;; it for encoding the file name it wants to load. And that | |
2620 | ;; will fail, since cpNNNN is not yet usable until | |
2621 | ;; code-pages.el finishes loading. | |
2622 | (setq default-file-name-coding-system code-page-coding)))) | |
893b49bb | 2623 | |
a41118cc | 2624 | (when (eq system-type 'darwin) |
ff76e074 SM |
2625 | ;; On Darwin, file names are always encoded in utf-8, no matter |
2626 | ;; the locale. | |
2627 | (setq default-file-name-coding-system 'utf-8) | |
2628 | ;; Mac OS X's Terminal.app by default uses utf-8 regardless of | |
2629 | ;; the locale. | |
2630 | (when (and (null window-system) | |
36ab8612 | 2631 | (equal (getenv "TERM_PROGRAM" frame) "Apple_Terminal")) |
ff76e074 SM |
2632 | (set-terminal-coding-system 'utf-8) |
2633 | (set-keyboard-coding-system 'utf-8))) | |
a41118cc | 2634 | |
e76ef161 | 2635 | ;; Default to A4 paper if we're not in a C, POSIX or US locale. |
3479c806 | 2636 | ;; (See comments in Flocale_info.) |
36ab8612 | 2637 | (unless frame |
97c57fb2 KL |
2638 | (let ((locale locale) |
2639 | (paper (locale-info 'paper))) | |
2640 | (if paper | |
2641 | ;; This will always be null at the time of writing. | |
2642 | (cond | |
2643 | ((equal paper '(216 279)) | |
2644 | (setq ps-paper-type 'letter)) | |
2645 | ((equal paper '(210 297)) | |
2646 | (setq ps-paper-type 'a4))) | |
2647 | (let ((vars '("LC_ALL" "LC_PAPER" "LANG"))) | |
2648 | (while (and vars (= 0 (length locale))) | |
36ab8612 | 2649 | (setq locale (getenv (pop vars) frame)))) |
97c57fb2 KL |
2650 | (when locale |
2651 | ;; As of glibc 2.2.5, these are the only US Letter locales, | |
2652 | ;; and the rest are A4. | |
2653 | (setq ps-paper-type | |
2654 | (or (locale-name-match locale '(("c$" . letter) | |
2655 | ("posix$" . letter) | |
2656 | (".._us" . letter) | |
2657 | (".._pr" . letter) | |
2658 | (".._ca" . letter) | |
2659 | ("enu$" . letter) ; Windows | |
2660 | ("esu$" . letter) | |
2661 | ("enc$" . letter) | |
2662 | ("frc$" . letter))) | |
2663 | 'a4))))))) | |
e76ef161 | 2664 | nil) |
40c81f74 | 2665 | \f |
16400c32 KH |
2666 | ;;; Character property |
2667 | ||
2668 | ;; Each element has the form (PROP . TABLE). | |
2669 | ;; PROP is a symbol representing a character property. | |
2670 | ;; TABLE is a char-table containing the property value for each character. | |
2671 | ;; TABLE may be a name of file to load to build a char-table. | |
2672 | ;; Don't modify this variable directly but use `define-char-code-property'. | |
2673 | ||
2674 | (defvar char-code-property-alist nil | |
2675 | "Alist of character property name vs char-table containing property values. | |
2676 | Internal use only.") | |
2677 | ||
2678 | (put 'char-code-property-table 'char-table-extra-slots 5) | |
2679 | ||
2680 | (defun define-char-code-property (name table &optional docstring) | |
2681 | "Define NAME as a character code property given by TABLE. | |
2682 | TABLE is a char-table of purpose `char-code-property-table' with | |
2683 | these extra slots: | |
2684 | 1st: NAME. | |
2685 | 2nd: Function to call to get a property value of a character. | |
caff3c0a | 2686 | It is called with three arguments CHAR, VAL, and TABLE, where |
16400c32 KH |
2687 | CHAR is a character, VAL is the value of (aref TABLE CHAR). |
2688 | 3rd: Function to call to put a property value of a character. | |
2689 | It is called with the same arguments as above. | |
2690 | 4th: Function to call to get a description string of a property value. | |
2691 | It is called with one argument VALUE, a property value. | |
2692 | 5th: Data used by the above functions. | |
2693 | ||
2694 | TABLE may be a name of file to load to build a char-table. The | |
2695 | file should contain a call of `define-char-code-property' with a | |
2696 | char-table of the above format as the argument TABLE. | |
2697 | ||
2698 | TABLE may also be nil, in which case no property value is pre-assigned. | |
2699 | ||
2700 | Optional 3rd argment DOCSTRING is a documentation string of the property. | |
2701 | ||
2702 | See also the documentation of `get-char-code-property' and | |
2703 | `put-char-code-property'." | |
2704 | (or (symbolp name) | |
2705 | (error "Not a symbol: %s" name)) | |
2706 | (if (char-table-p table) | |
2707 | (or (and (eq (char-table-subtype table) 'char-code-property-table) | |
2708 | (eq (char-table-extra-slot table 0) name)) | |
2709 | (error "Invalid char-table: %s" table)) | |
2710 | (or (stringp table) | |
2711 | (error "Not a char-table nor a file name: %s" table))) | |
2712 | (let ((slot (assq name char-code-property-alist))) | |
2713 | (if slot | |
2714 | (setcdr slot table) | |
2715 | (setq char-code-property-alist | |
2716 | (cons (cons name table) char-code-property-alist)))) | |
2717 | (put name 'char-code-property-documentation docstring)) | |
4ed46869 KH |
2718 | |
2719 | (defvar char-code-property-table | |
2720 | (make-char-table 'char-code-property-table) | |
2721 | "Char-table containing a property list of each character code. | |
16400c32 | 2722 | This table is used for properties not listed in `char-code-property-alist'. |
4ed46869 | 2723 | See also the documentation of `get-char-code-property' and |
96db204a | 2724 | `put-char-code-property'.") |
4ed46869 KH |
2725 | |
2726 | (defun get-char-code-property (char propname) | |
16400c32 KH |
2727 | "Return the value of CHAR's PROPNAME property." |
2728 | (let ((slot (assq propname char-code-property-alist))) | |
2729 | (if slot | |
2730 | (let (table value func) | |
2731 | (if (stringp (cdr slot)) | |
2732 | (load (cdr slot))) | |
2733 | (setq table (cdr slot) | |
2734 | value (aref table char) | |
2735 | func (char-table-extra-slot table 1)) | |
2736 | (if (functionp func) | |
2737 | (setq value (funcall func char value table))) | |
2738 | value) | |
2739 | (plist-get (aref char-code-property-table char) propname)))) | |
4ed46869 KH |
2740 | |
2741 | (defun put-char-code-property (char propname value) | |
16400c32 | 2742 | "Store CHAR's PROPNAME property with VALUE. |
4ed46869 | 2743 | It can be retrieved with `(get-char-code-property CHAR PROPNAME)'." |
16400c32 KH |
2744 | (let ((slot (assq propname char-code-property-alist))) |
2745 | (if slot | |
2746 | (let (table func) | |
2747 | (if (stringp (cdr slot)) | |
2748 | (load (cdr slot))) | |
2749 | (setq table (cdr slot) | |
2750 | func (char-table-extra-slot table 2)) | |
2751 | (if (functionp func) | |
2752 | (funcall func char value table) | |
2753 | (aset table char value))) | |
2754 | (let* ((plist (aref char-code-property-table char)) | |
2755 | (x (plist-put plist propname value))) | |
2756 | (or (eq x plist) | |
2757 | (aset char-code-property-table char x)))) | |
2758 | value)) | |
2759 | ||
2760 | (defun char-code-property-description (prop value) | |
2761 | "Return a description string of character property PROP's value VALUE. | |
2762 | If there's no description string for VALUE, return nil." | |
2763 | (let ((slot (assq prop char-code-property-alist))) | |
2764 | (if slot | |
2765 | (let (table func) | |
2766 | (if (stringp (cdr slot)) | |
2767 | (load (cdr slot))) | |
2768 | (setq table (cdr slot) | |
2769 | func (char-table-extra-slot table 3)) | |
2770 | (if (functionp func) | |
2771 | (funcall func value)))))) | |
4ed46869 | 2772 | |
a127b764 KH |
2773 | \f |
2774 | ;; Pretty description of encoded string | |
2775 | ||
2776 | ;; Alist of ISO 2022 control code vs the corresponding mnemonic string. | |
2777 | (defvar iso-2022-control-alist | |
2778 | '((?\x1b . "ESC") | |
2779 | (?\x0e . "SO") | |
2780 | (?\x0f . "SI") | |
2781 | (?\x8e . "SS2") | |
2782 | (?\x8f . "SS3") | |
2783 | (?\x9b . "CSI"))) | |
2784 | ||
2785 | (defun encoded-string-description (str coding-system) | |
2786 | "Return a pretty description of STR that is encoded by CODING-SYSTEM." | |
2787 | (setq str (string-as-unibyte str)) | |
993b2a7d | 2788 | (mapconcat |
a6dfc99b | 2789 | (if (and coding-system (eq (coding-system-type coding-system) 'iso-2022)) |
993b2a7d KH |
2790 | ;; Try to get a pretty description for ISO 2022 escape sequences. |
2791 | (function (lambda (x) (or (cdr (assq x iso-2022-control-alist)) | |
695effcc JL |
2792 | (format "#x%02X" x)))) |
2793 | (function (lambda (x) (format "#x%02X" x)))) | |
993b2a7d | 2794 | str " ")) |
a127b764 KH |
2795 | |
2796 | (defun encode-coding-char (char coding-system) | |
2797 | "Encode CHAR by CODING-SYSTEM and return the resulting string. | |
2798 | If CODING-SYSTEM can't safely encode CHAR, return nil." | |
bc304a20 DL |
2799 | (let ((str1 (string-as-multibyte (string char))) |
2800 | (str2 (string-as-multibyte (string char char))) | |
a127b764 | 2801 | enc1 enc2 i1 i2) |
bc304a20 DL |
2802 | (when (memq (coding-system-base coding-system) |
2803 | (find-coding-systems-string str1)) | |
a127b764 KH |
2804 | ;; We must find the encoded string of CHAR. But, just encoding |
2805 | ;; CHAR will put extra control sequences (usually to designate | |
bc304a20 | 2806 | ;; ASCII charset) at the tail if type of CODING is ISO 2022. |
a127b764 KH |
2807 | ;; To exclude such tailing bytes, we at first encode one-char |
2808 | ;; string and two-char string, then check how many bytes at the | |
2809 | ;; tail of both encoded strings are the same. | |
2810 | ||
bd953173 | 2811 | (setq enc1 (encode-coding-string str1 coding-system) |
a127b764 | 2812 | i1 (length enc1) |
bd953173 | 2813 | enc2 (encode-coding-string str2 coding-system) |
a127b764 KH |
2814 | i2 (length enc2)) |
2815 | (while (and (> i1 0) (= (aref enc1 (1- i1)) (aref enc2 (1- i2)))) | |
2816 | (setq i1 (1- i1) i2 (1- i2))) | |
2817 | ||
2818 | ;; Now (substring enc1 i1) and (substring enc2 i2) are the same, | |
2819 | ;; and they are the extra control sequences at the tail to | |
2820 | ;; exclude. | |
2821 | (substring enc2 0 i2)))) | |
2822 | ||
d2f613a0 DL |
2823 | ;; Backwards compatibility. These might be better with :init-value t, |
2824 | ;; but that breaks loadup. | |
2825 | (define-minor-mode unify-8859-on-encoding-mode | |
2826 | "Obsolete." | |
2827 | :group 'mule | |
2828 | :global t) | |
2829 | (define-minor-mode unify-8859-on-decoding-mode | |
2830 | "Obsolete." | |
2831 | :group 'mule | |
2832 | :global t) | |
a127b764 | 2833 | |
7dd42fb1 KH |
2834 | (defvar nonascii-insert-offset 0 "This variable is obsolete.") |
2835 | (defvar nonascii-translation-table nil "This variable is obsolete.") | |
2836 | ||
cb9e47dd KH |
2837 | (defun ucs-insert (arg) |
2838 | "Insert a character of the given Unicode code point. | |
2839 | Interactively, prompts for a hex string giving the code." | |
2840 | (interactive "sUnicode (hex): ") | |
2841 | (or (integerp arg) | |
2842 | (setq arg (string-to-number arg 16))) | |
2843 | (if (or (< arg 0) (> arg #x10FFFF)) | |
2844 | (error "Not a Unicode character code: 0x%X" arg)) | |
2845 | (insert arg)) | |
2846 | ||
a127b764 | 2847 | |
9ee5b744 | 2848 | ;; arch-tag: b382c432-4b36-460e-bf4c-05efd0bb18dc |
4ed46869 | 2849 | ;;; mule-cmds.el ends here |