Commit | Line | Data |
---|---|---|
60370d40 | 1 | ;;; mule-cmds.el --- commands for mulitilingual environment |
a985cd2f | 2 | ;; Copyright (C) 1995, 2003 Electrotechnical Laboratory, JAPAN. |
fa526c4a | 3 | ;; Licensed to the Free Software Foundation. |
e76ef161 | 4 | ;; Copyright (C) 2000, 2001, 2002, 2003 Free Software Foundation, Inc. |
4ed46869 KH |
5 | |
6 | ;; Keywords: mule, multilingual | |
7 | ||
8 | ;; This file is part of GNU Emacs. | |
9 | ||
10 | ;; GNU Emacs is free software; you can redistribute it and/or modify | |
11 | ;; it under the terms of the GNU General Public License as published by | |
12 | ;; the Free Software Foundation; either version 2, or (at your option) | |
13 | ;; any later version. | |
14 | ||
15 | ;; GNU Emacs is distributed in the hope that it will be useful, | |
16 | ;; but WITHOUT ANY WARRANTY; without even the implied warranty of | |
17 | ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
18 | ;; GNU General Public License for more details. | |
19 | ||
20 | ;; You should have received a copy of the GNU General Public License | |
369314dc KH |
21 | ;; along with GNU Emacs; see the file COPYING. If not, write to the |
22 | ;; Free Software Foundation, Inc., 59 Temple Place - Suite 330, | |
23 | ;; Boston, MA 02111-1307, USA. | |
4ed46869 | 24 | |
60370d40 PJ |
25 | ;;; Commentary: |
26 | ||
4ed46869 KH |
27 | ;;; Code: |
28 | ||
cda74479 DL |
29 | (eval-when-compile (defvar dos-codepage)) |
30 | ||
4ed46869 KH |
31 | ;;; MULE related key bindings and menus. |
32 | ||
0709d285 | 33 | (defvar mule-keymap (make-sparse-keymap) |
33d17698 | 34 | "Keymap for Mule (Multilingual environment) specific commands.") |
4ed46869 | 35 | |
8f81f784 | 36 | ;; Keep "C-x C-m ..." for mule specific commands. |
0709d285 | 37 | (define-key ctl-x-map "\C-m" mule-keymap) |
ef8a8c8c | 38 | |
4ed46869 | 39 | (define-key mule-keymap "f" 'set-buffer-file-coding-system) |
d80dee05 | 40 | (define-key mule-keymap "r" 'revert-buffer-with-coding-system) |
2308d0d7 | 41 | (define-key mule-keymap "F" 'set-file-name-coding-system) |
4ed46869 | 42 | (define-key mule-keymap "t" 'set-terminal-coding-system) |
15b3e511 KH |
43 | (define-key mule-keymap "k" 'set-keyboard-coding-system) |
44 | (define-key mule-keymap "p" 'set-buffer-process-coding-system) | |
7624ebb9 KH |
45 | (define-key mule-keymap "x" 'set-selection-coding-system) |
46 | (define-key mule-keymap "X" 'set-next-selection-coding-system) | |
8b784951 | 47 | (define-key mule-keymap "\C-\\" 'set-input-method) |
15b3e511 | 48 | (define-key mule-keymap "c" 'universal-coding-system-argument) |
b4fba33f | 49 | (define-key mule-keymap "l" 'set-language-environment) |
4ed46869 | 50 | |
538d88fb EZ |
51 | (defvar mule-menu-keymap |
52 | (make-sparse-keymap "Mule (Multilingual Environment)") | |
33d17698 | 53 | "Keymap for Mule (Multilingual environment) menu specific commands.") |
15b3e511 | 54 | |
dcad02bc EZ |
55 | (defvar describe-language-environment-map |
56 | (make-sparse-keymap "Describe Language Environment")) | |
15b3e511 | 57 | |
dcad02bc EZ |
58 | (defvar setup-language-environment-map |
59 | (make-sparse-keymap "Set Language Environment")) | |
15b3e511 | 60 | |
dcad02bc EZ |
61 | (defvar set-coding-system-map |
62 | (make-sparse-keymap "Set Coding System")) | |
15b3e511 | 63 | |
15b3e511 | 64 | (define-key-after mule-menu-keymap [set-language-environment] |
cda74479 DL |
65 | (list 'menu-item "Set Language Environment" setup-language-environment-map |
66 | :help "Multilingual environment suitable for a specific language")) | |
a61f401d | 67 | (define-key-after mule-menu-keymap [mouse-set-font] |
538d88fb | 68 | '(menu-item "Set Font/Fontset" mouse-set-font |
cda74479 DL |
69 | :visible (fboundp 'generate-fontset-menu) |
70 | :help "Select a font from list of known fonts/fontsets")) | |
15b3e511 KH |
71 | (define-key-after mule-menu-keymap [separator-mule] |
72 | '("--") | |
73 | t) | |
74 | (define-key-after mule-menu-keymap [toggle-input-method] | |
538d88fb | 75 | '(menu-item "Toggle Input Method" toggle-input-method) |
15b3e511 | 76 | t) |
8b784951 | 77 | (define-key-after mule-menu-keymap [set-input-method] |
538d88fb | 78 | '(menu-item "Select Input Method..." set-input-method) |
15b3e511 | 79 | t) |
cda74479 DL |
80 | (define-key-after mule-menu-keymap [describe-input-method] |
81 | '(menu-item "Describe Input Method" describe-input-method)) | |
15b3e511 KH |
82 | (define-key-after mule-menu-keymap [separator-input-method] |
83 | '("--") | |
84 | t) | |
15b3e511 | 85 | (define-key-after mule-menu-keymap [set-various-coding-system] |
cda74479 | 86 | (list 'menu-item "Set Coding Systems" set-coding-system-map |
be9650bc | 87 | :enable 'default-enable-multibyte-characters)) |
538d88fb EZ |
88 | (define-key-after mule-menu-keymap [view-hello-file] |
89 | '(menu-item "Show Multi-lingual Text" view-hello-file | |
90 | :enable (file-readable-p | |
91 | (expand-file-name "HELLO" data-directory)) | |
92 | :help "Display file which says HELLO in many languages") | |
15b3e511 KH |
93 | t) |
94 | (define-key-after mule-menu-keymap [separator-coding-system] | |
95 | '("--") | |
96 | t) | |
538d88fb EZ |
97 | (define-key-after mule-menu-keymap [describe-language-environment] |
98 | (list 'menu-item "Describe Language Environment" | |
99 | describe-language-environment-map | |
cda74479 | 100 | :help "Show multilingual settings for a specific language") |
15b3e511 | 101 | t) |
538d88fb EZ |
102 | (define-key-after mule-menu-keymap [describe-input-method] |
103 | '(menu-item "Describe Input Method..." describe-input-method | |
cda74479 | 104 | :help "Keyboard layout for a specific input method") |
538d88fb EZ |
105 | t) |
106 | (define-key-after mule-menu-keymap [describe-coding-system] | |
107 | '(menu-item "Describe Coding System..." describe-coding-system) | |
108 | t) | |
cda74479 DL |
109 | (define-key-after mule-menu-keymap [list-character-sets] |
110 | '(menu-item "List Character Sets" list-character-sets | |
111 | :help "Show table of available character sets")) | |
538d88fb EZ |
112 | (define-key-after mule-menu-keymap [mule-diag] |
113 | '(menu-item "Show All of Mule Status" mule-diag | |
114 | :help "Display multilingual environment settings") | |
15b3e511 KH |
115 | t) |
116 | ||
3a151e98 | 117 | (define-key-after set-coding-system-map [universal-coding-system-argument] |
538d88fb EZ |
118 | '(menu-item "For Next Command" universal-coding-system-argument |
119 | :help "Coding system to be used by next command") | |
3a151e98 | 120 | t) |
d80dee05 RS |
121 | (define-key-after set-coding-system-map [separator-1] |
122 | '("--") | |
123 | t) | |
124 | (define-key-after set-coding-system-map [set-buffer-file-coding-system] | |
125 | '(menu-item "For Saving This Buffer" set-buffer-file-coding-system | |
126 | :help "How to encode this buffer when saved") | |
127 | t) | |
128 | (define-key-after set-coding-system-map [revert-buffer-with-coding-system] | |
129 | '(menu-item "For Reverting This File Now" revert-buffer-with-coding-system | |
130 | :enable buffer-file-name | |
131 | :help "Revisit this file immediately using specified coding system") | |
132 | t) | |
2308d0d7 KH |
133 | (define-key-after set-coding-system-map [set-file-name-coding-system] |
134 | '(menu-item "For File Name" set-file-name-coding-system | |
135 | :help "How to decode/encode file names") | |
136 | t) | |
d80dee05 RS |
137 | (define-key-after set-coding-system-map [separator-2] |
138 | '("--") | |
15b3e511 | 139 | t) |
d80dee05 | 140 | |
15b3e511 | 141 | (define-key-after set-coding-system-map [set-keyboard-coding-system] |
538d88fb EZ |
142 | '(menu-item "For Keyboard" set-keyboard-coding-system |
143 | :help "How to decode keyboard input") | |
15b3e511 | 144 | t) |
d80dee05 RS |
145 | (define-key-after set-coding-system-map [set-terminal-coding-system] |
146 | '(menu-item "For Terminal" set-terminal-coding-system | |
147 | :enable (null (memq window-system '(x w32 mac))) | |
148 | :help "How to encode terminal output") | |
149 | t) | |
150 | (define-key-after set-coding-system-map [separator-3] | |
151 | '("--") | |
15b3e511 | 152 | t) |
7624ebb9 | 153 | (define-key-after set-coding-system-map [set-selection-coding-system] |
538d88fb EZ |
154 | '(menu-item "For X Selections/Clipboard" set-selection-coding-system |
155 | :visible (display-selections-p) | |
156 | :help "How to en/decode data to/from selection/clipboard") | |
7624ebb9 KH |
157 | t) |
158 | (define-key-after set-coding-system-map [set-next-selection-coding-system] | |
538d88fb EZ |
159 | '(menu-item "For Next X Selection" set-next-selection-coding-system |
160 | :visible (display-selections-p) | |
161 | :help "How to en/decode next selection/clipboard operation") | |
7624ebb9 | 162 | t) |
d80dee05 RS |
163 | (define-key-after set-coding-system-map [set-buffer-process-coding-system] |
164 | '(menu-item "For I/O with Subprocess" set-buffer-process-coding-system | |
165 | :visible (fboundp 'start-process) | |
166 | :enable (get-buffer-process (current-buffer)) | |
167 | :help "How to en/decode I/O from/to subprocess connected to this buffer") | |
168 | t) | |
169 | ||
170 | ||
15b3e511 | 171 | (define-key setup-language-environment-map |
538d88fb | 172 | [Default] '(menu-item "Default" setup-specified-language-environment)) |
4ed46869 | 173 | |
cda74479 DL |
174 | (define-key describe-language-environment-map |
175 | [Default] '(menu-item "Default" describe-specified-language-support)) | |
176 | ||
4ed46869 KH |
177 | ;; This should be a single character key binding because users use it |
178 | ;; very frequently while editing multilingual text. Now we can use | |
179 | ;; only two such keys: "\C-\\" and "\C-^", but the latter is not | |
180 | ;; convenient because it requires shifting on most keyboards. An | |
181 | ;; alternative is "\C-\]" which is now bound to `abort-recursive-edit' | |
182 | ;; but it won't be used that frequently. | |
183 | (define-key global-map "\C-\\" 'toggle-input-method) | |
184 | ||
5f395df3 SM |
185 | ;; This is no good because people often type Shift-SPC |
186 | ;; meaning to type SPC. -- rms. | |
187 | ;; ;; Here's an alternative key binding for X users (Shift-SPACE). | |
188 | ;; (define-key global-map [?\S- ] 'toggle-input-method) | |
b4fba33f | 189 | |
464cc130 KH |
190 | ;;; Mule related hyperlinks. |
191 | (defconst help-xref-mule-regexp-template | |
192 | (purecopy (concat "\\(\\<\\(" | |
193 | "\\(coding system\\)\\|" | |
d0c40faa KH |
194 | "\\(input method\\)\\|" |
195 | "\\(character set\\)\\|" | |
196 | "\\(charset\\)" | |
464cc130 KH |
197 | "\\)\\s-+\\)?" |
198 | ;; Note starting with word-syntax character: | |
199 | "`\\(\\sw\\(\\sw\\|\\s_\\)+\\)'"))) | |
200 | ||
26d87040 EZ |
201 | (defun coding-system-change-eol-conversion (coding-system eol-type) |
202 | "Return a coding system which differs from CODING-SYSTEM in eol conversion. | |
203 | The returned coding system converts end-of-line by EOL-TYPE | |
204 | but text as the same way as CODING-SYSTEM. | |
205 | EOL-TYPE should be `unix', `dos', `mac', or nil. | |
206 | If EOL-TYPE is nil, the returned coding system detects | |
207 | how end-of-line is formatted automatically while decoding. | |
208 | ||
209 | EOL-TYPE can be specified by an integer 0, 1, or 2. | |
210 | They means `unix', `dos', and `mac' respectively." | |
211 | (if (symbolp eol-type) | |
212 | (setq eol-type (cond ((eq eol-type 'unix) 0) | |
213 | ((eq eol-type 'dos) 1) | |
214 | ((eq eol-type 'mac) 2) | |
215 | (t eol-type)))) | |
216 | (let ((orig-eol-type (coding-system-eol-type coding-system))) | |
217 | (if (vectorp orig-eol-type) | |
218 | (if (not eol-type) | |
219 | coding-system | |
220 | (aref orig-eol-type eol-type)) | |
221 | (let ((base (coding-system-base coding-system))) | |
222 | (if (not eol-type) | |
223 | base | |
224 | (if (= eol-type orig-eol-type) | |
225 | coding-system | |
226 | (setq orig-eol-type (coding-system-eol-type base)) | |
227 | (if (vectorp orig-eol-type) | |
228 | (aref orig-eol-type eol-type)))))))) | |
229 | ||
230 | (defun coding-system-change-text-conversion (coding-system coding) | |
231 | "Return a coding system which differs from CODING-SYSTEM in text conversion. | |
232 | The returned coding system converts text by CODING | |
233 | but end-of-line as the same way as CODING-SYSTEM. | |
234 | If CODING is nil, the returned coding system detects | |
235 | how text is formatted automatically while decoding." | |
1d77e15a JR |
236 | (let ((eol-type (coding-system-eol-type coding-system))) |
237 | (coding-system-change-eol-conversion | |
238 | (if coding coding 'undecided) | |
239 | (if (numberp eol-type) (aref [unix dos mac] eol-type))))) | |
26d87040 | 240 | |
4ed46869 | 241 | (defun toggle-enable-multibyte-characters (&optional arg) |
6998e1a1 RS |
242 | "Change whether this buffer uses multibyte characters. |
243 | With arg, use multibyte characters if the arg is positive. | |
244 | ||
245 | Note that this command does not convert the byte contents of | |
246 | the buffer; it only changes the way those bytes are interpreted. | |
247 | In general, therefore, this command *changes* the sequence of | |
248 | characters that the current buffer contains. | |
249 | ||
250 | We suggest you avoid using use this command unless you know what you | |
251 | are doing. If you use it by mistake, and the buffer is now displayed | |
252 | wrong, use this command again to toggle back to the right mode." | |
4ed46869 | 253 | (interactive "P") |
b7079457 RS |
254 | (let ((new-flag |
255 | (if (null arg) (null enable-multibyte-characters) | |
256 | (> (prefix-numeric-value arg) 0)))) | |
257 | (set-buffer-multibyte new-flag)) | |
4ed46869 KH |
258 | (force-mode-line-update)) |
259 | ||
260 | (defun view-hello-file () | |
261 | "Display the HELLO file which list up many languages and characters." | |
262 | (interactive) | |
8f81f784 KH |
263 | ;; We have to decode the file in any environment. |
264 | (let ((default-enable-multibyte-characters t) | |
95fa03b4 | 265 | (coding-system-for-read 'iso-2022-7bit)) |
544dd975 | 266 | (view-file (expand-file-name "HELLO" data-directory)))) |
4ed46869 | 267 | |
9e9a77a6 | 268 | (defun universal-coding-system-argument (coding-system) |
15b3e511 | 269 | "Execute an I/O command using the specified coding system." |
9e9a77a6 RS |
270 | (interactive |
271 | (let ((default (and buffer-file-coding-system | |
34104362 KH |
272 | (not (eq (coding-system-type buffer-file-coding-system) |
273 | t)) | |
9e9a77a6 RS |
274 | buffer-file-coding-system))) |
275 | (list (read-coding-system | |
276 | (if default | |
277 | (format "Coding system for following command (default, %s): " default) | |
278 | "Coding system for following command: ") | |
279 | default)))) | |
280 | (let* ((keyseq (read-key-sequence | |
e14a8f4c | 281 | (format "Command to execute with %s:" coding-system))) |
04363179 GM |
282 | (cmd (key-binding keyseq)) |
283 | prefix) | |
284 | ||
285 | (when (eq cmd 'universal-argument) | |
286 | (call-interactively cmd) | |
a1506d29 | 287 | |
04363179 GM |
288 | ;; Process keys bound in `universal-argument-map'. |
289 | (while (progn | |
290 | (setq keyseq (read-key-sequence nil t) | |
291 | cmd (key-binding keyseq t)) | |
292 | (not (eq cmd 'universal-argument-other-key))) | |
293 | (let ((current-prefix-arg prefix-arg) | |
d37ef0f6 | 294 | ;; Have to bind `last-command-char' here so that |
04363179 GM |
295 | ;; `digit-argument', for isntance, can compute the |
296 | ;; prefix arg. | |
297 | (last-command-char (aref keyseq 0))) | |
298 | (call-interactively cmd))) | |
299 | ||
300 | ;; This is the final call to `univeral-argument-other-key', which | |
301 | ;; set's the final `prefix-arg. | |
302 | (let ((current-prefix-arg prefix-arg)) | |
303 | (call-interactively cmd)) | |
a1506d29 | 304 | |
04363179 GM |
305 | ;; Read the command to execute with the given prefix arg. |
306 | (setq prefix prefix-arg | |
307 | keyseq (read-key-sequence nil t) | |
308 | cmd (key-binding keyseq))) | |
309 | ||
15b3e511 | 310 | (let ((coding-system-for-read coding-system) |
04363179 | 311 | (coding-system-for-write coding-system) |
0e9ec609 | 312 | (coding-system-require-warning t) |
04363179 | 313 | (current-prefix-arg prefix)) |
15b3e511 KH |
314 | (message "") |
315 | (call-interactively cmd)))) | |
316 | ||
de94d711 | 317 | (defun set-default-coding-systems (coding-system) |
0c3154d2 | 318 | "Set default value of various coding systems to CODING-SYSTEM. |
387136f6 | 319 | This sets the following coding systems: |
0c3154d2 | 320 | o coding system of a newly created buffer |
8efc03e1 KH |
321 | o default coding system for subprocess I/O |
322 | This also sets the following values: | |
d37ef0f6 | 323 | o default value used as `file-name-coding-system' for converting file names. |
03c35c83 EZ |
324 | o default value for the command `set-terminal-coding-system' (not on MSDOS) |
325 | o default value for the command `set-keyboard-coding-system'." | |
de94d711 KH |
326 | (check-coding-system coding-system) |
327 | (setq-default buffer-file-coding-system coding-system) | |
716184d4 RS |
328 | (if default-enable-multibyte-characters |
329 | (setq default-file-name-coding-system coding-system)) | |
03c35c83 EZ |
330 | ;; If coding-system is nil, honor that on MS-DOS as well, so |
331 | ;; that they could reset the terminal coding system. | |
332 | (unless (and (eq window-system 'pc) coding-system) | |
333 | (setq default-terminal-coding-system coding-system)) | |
de94d711 | 334 | (setq default-keyboard-coding-system coding-system) |
1d77e15a JR |
335 | ;; Preserve eol-type from existing default-process-coding-systems. |
336 | ;; On non-unix-like systems in particular, these may have been set | |
337 | ;; carefully by the user, or by the startup code, to deal with the | |
338 | ;; users shell appropriately, so should not be altered by changing | |
339 | ;; language environment. | |
340 | (let ((output-coding | |
341 | (coding-system-change-text-conversion | |
342 | (car default-process-coding-system) coding-system)) | |
343 | (input-coding | |
344 | (coding-system-change-text-conversion | |
345 | (cdr default-process-coding-system) coding-system))) | |
346 | (setq default-process-coding-system | |
347 | (cons output-coding input-coding)))) | |
de94d711 | 348 | |
45d08cb2 | 349 | (defalias 'update-iso-coding-systems 'update-coding-systems-internal) |
2598a293 | 350 | (make-obsolete 'update-iso-coding-systems 'update-coding-systems-internal "20.3") |
45d08cb2 | 351 | |
0c3154d2 KH |
352 | (defun prefer-coding-system (coding-system) |
353 | "Add CODING-SYSTEM at the front of the priority list for automatic detection. | |
387136f6 | 354 | This also sets the following coding systems: |
0c3154d2 | 355 | o coding system of a newly created buffer |
8efc03e1 KH |
356 | o default coding system for subprocess I/O |
357 | This also sets the following values: | |
c654de1d | 358 | o default value used as `file-name-coding-system' for converting file names. |
03c35c83 EZ |
359 | o default value for the command `set-terminal-coding-system' (not on MSDOS) |
360 | o default value for the command `set-keyboard-coding-system' | |
361 | ||
bd3ac67e EZ |
362 | If CODING-SYSTEM specifies a certain type of EOL conversion, the coding |
363 | systems set by this function will use that type of EOL conversion. | |
364 | ||
03c35c83 EZ |
365 | This command does not change the default value of terminal coding system |
366 | for MS-DOS terminal, because DOS terminals only support a single coding | |
367 | system, and Emacs automatically sets the default to that coding system at | |
5beaa0d1 KH |
368 | startup. |
369 | ||
c654de1d DL |
370 | A coding system that requires automatic detection of text |
371 | encoding (e.g. undecided, unix) can't be preferred. | |
372 | ||
373 | See also `coding-category-list' and `coding-system-category'." | |
0c3154d2 KH |
374 | (interactive "zPrefer coding system: ") |
375 | (if (not (and coding-system (coding-system-p coding-system))) | |
376 | (error "Invalid coding system `%s'" coding-system)) | |
377 | (let ((coding-category (coding-system-category coding-system)) | |
bd3ac67e EZ |
378 | (base (coding-system-base coding-system)) |
379 | (eol-type (coding-system-eol-type coding-system))) | |
0c3154d2 KH |
380 | (if (not coding-category) |
381 | ;; CODING-SYSTEM is no-conversion or undecided. | |
382 | (error "Can't prefer the coding system `%s'" coding-system)) | |
8efc03e1 | 383 | (set coding-category (or base coding-system)) |
45d08cb2 | 384 | (update-coding-systems-internal) |
812cad80 | 385 | (or (eq coding-category (car coding-category-list)) |
0c3154d2 | 386 | ;; We must change the order. |
812cad80 | 387 | (set-coding-priority (list coding-category))) |
8efc03e1 KH |
388 | (if (and base (interactive-p)) |
389 | (message "Highest priority is set to %s (base of %s)" | |
390 | base coding-system)) | |
bd3ac67e | 391 | ;; If they asked for specific EOL conversion, honor that. |
6f9dc4fd | 392 | (if (memq eol-type '(0 1 2)) |
bd3ac67e EZ |
393 | (setq coding-system |
394 | (coding-system-change-eol-conversion base eol-type)) | |
395 | (setq coding-system base)) | |
396 | (set-default-coding-systems coding-system))) | |
0c3154d2 | 397 | |
b5edd1d1 KH |
398 | (defvar sort-coding-systems-predicate nil |
399 | "If non-nil, a predicate function to sort coding systems. | |
400 | ||
401 | It is called with two coding systems, and should return t if the first | |
402 | one is \"less\" than the second. | |
403 | ||
404 | The function `sort-coding-systems' use it.") | |
405 | ||
406 | (defun sort-coding-systems (codings) | |
407 | "Sort coding system list CODINGS by a priority of each coding system. | |
408 | ||
409 | If a coding system is most preferred, it has the highest priority. | |
410 | Otherwise, a coding system corresponds to some MIME charset has higher | |
411 | priorities. Among them, a coding system included in `coding-system' | |
412 | key of the current language environment has higher priorities. See | |
413 | also the documentation of `language-info-alist'. | |
414 | ||
415 | If the variable `sort-coding-systems-predicate' (which see) is | |
416 | non-nil, it is used to sort CODINGS in the different way than above." | |
417 | (if sort-coding-systems-predicate | |
418 | (sort codings sort-coding-systems-predicate) | |
da1ebad1 KH |
419 | (let* ((from-categories (mapcar #'(lambda (x) (symbol-value x)) |
420 | coding-category-list)) | |
421 | (most-preferred (car from-categories)) | |
b5edd1d1 KH |
422 | (lang-preferred (get-language-info current-language-environment |
423 | 'coding-system)) | |
424 | (func (function | |
425 | (lambda (x) | |
426 | (let ((base (coding-system-base x))) | |
da1ebad1 KH |
427 | ;; We calculate the priority number 0..255 by |
428 | ;; using the 8 bits PMMLCEII as this: | |
429 | ;; P: 1 iff most preferred. | |
430 | ;; MM: greater than 0 iff mime-charset. | |
431 | ;; L: 1 iff one of the current lang. env.'s codings. | |
432 | ;; C: 1 iff one of codings listed in the category list. | |
433 | ;; E: 1 iff not XXX-with-esc | |
434 | ;; II: if iso-2022 based, 0..3, else 1. | |
435 | (logior | |
436 | (lsh (if (eq base most-preferred) 1 0) 7) | |
437 | (lsh | |
438 | (let ((mime (coding-system-get base 'mime-charset))) | |
b439e72a DL |
439 | ;; Prefer coding systems corresponding to a |
440 | ;; MIME charset. | |
b5edd1d1 | 441 | (if mime |
b439e72a DL |
442 | ;; Lower utf-16 priority so that we |
443 | ;; normally prefer utf-8 to it, and put | |
444 | ;; x-ctext below that. | |
de814dd9 KH |
445 | (cond ((string-match "utf-16" |
446 | (symbol-name mime)) | |
da1ebad1 | 447 | 2) |
b439e72a | 448 | ((string-match "^x-" (symbol-name mime)) |
da1ebad1 KH |
449 | 1) |
450 | (t 3)) | |
b5edd1d1 | 451 | 0)) |
da1ebad1 KH |
452 | 5) |
453 | (lsh (if (memq base lang-preferred) 1 0) 4) | |
454 | (lsh (if (memq base from-categories) 1 0) 3) | |
455 | (lsh (if (string-match "-with-esc\\'" | |
456 | (symbol-name base)) | |
457 | 0 1) 2) | |
458 | (if (eq (coding-system-type base) 2) | |
459 | ;; For ISO based coding systems, prefer | |
460 | ;; one that doesn't use escape sequences. | |
461 | (let ((flags (coding-system-flags base))) | |
462 | (if (or (consp (aref flags 0)) | |
463 | (consp (aref flags 1)) | |
464 | (consp (aref flags 2)) | |
465 | (consp (aref flags 3))) | |
466 | (if (or (aref flags 8) (aref flags 9)) | |
467 | 0 | |
468 | 1) | |
469 | 2)) | |
470 | 1))))))) | |
b5edd1d1 KH |
471 | (sort codings (function (lambda (x y) |
472 | (> (funcall func x) (funcall func y)))))))) | |
54b226f7 | 473 | |
3fc7dfe5 | 474 | (defun find-coding-systems-region (from to) |
54b226f7 | 475 | "Return a list of proper coding systems to encode a text between FROM and TO. |
d37ef0f6 | 476 | If FROM is a string, find coding systems in that instead of the buffer. |
54b226f7 KH |
477 | All coding systems in the list can safely encode any multibyte characters |
478 | in the text. | |
479 | ||
e8dd0160 | 480 | If the text contains no multibyte characters, return a list of a single |
3fc7dfe5 | 481 | element `undecided'." |
b5edd1d1 KH |
482 | (let ((codings (find-coding-systems-region-internal from to))) |
483 | (if (eq codings t) | |
484 | ;; The text contains only ASCII characters. Any coding | |
485 | ;; systems are safe. | |
486 | '(undecided) | |
487 | ;; We need copy-sequence because sorting will alter the argument. | |
488 | (sort-coding-systems (copy-sequence codings))))) | |
54b226f7 | 489 | |
3fc7dfe5 KH |
490 | (defun find-coding-systems-string (string) |
491 | "Return a list of proper coding systems to encode STRING. | |
492 | All coding systems in the list can safely encode any multibyte characters | |
493 | in STRING. | |
494 | ||
e8dd0160 | 495 | If STRING contains no multibyte characters, return a list of a single |
3fc7dfe5 | 496 | element `undecided'." |
b5edd1d1 | 497 | (find-coding-systems-region string nil)) |
3fc7dfe5 KH |
498 | |
499 | (defun find-coding-systems-for-charsets (charsets) | |
500 | "Return a list of proper coding systems to encode characters of CHARSETS. | |
6053d86a KH |
501 | CHARSETS is a list of character sets. |
502 | It actually checks at most the first 96 characters of each charset. | |
503 | So, if a charset of dimension two is included in CHARSETS, the value may | |
504 | contain a coding system that can't encode all characters of the charset." | |
b5edd1d1 KH |
505 | (cond ((or (null charsets) |
506 | (and (= (length charsets) 1) | |
507 | (eq 'ascii (car charsets)))) | |
508 | '(undecided)) | |
509 | ((or (memq 'eight-bit-control charsets) | |
510 | (memq 'eight-bit-graphic charsets)) | |
511 | '(raw-text emacs-mule)) | |
512 | (t | |
513 | (let ((codings t) | |
6053d86a | 514 | charset l str) |
b5edd1d1 KH |
515 | (while (and codings charsets) |
516 | (setq charset (car charsets) charsets (cdr charsets)) | |
517 | (unless (eq charset 'ascii) | |
6053d86a KH |
518 | (setq str (make-string 96 32)) |
519 | (if (= (charset-dimension charset) 1) | |
520 | (if (= (charset-chars charset) 96) | |
521 | (dotimes (i 96) | |
522 | (aset str i (make-char charset (+ i 32)))) | |
523 | (dotimes (i 94) | |
524 | (aset str i (make-char charset (+ i 33))))) | |
525 | (if (= (charset-chars charset) 96) | |
526 | (dotimes (i 96) | |
527 | (aset str i (make-char charset 32 (+ i 32)))) | |
528 | (dotimes (i 94) | |
529 | (aset str i (make-char charset 33 (+ i 33)))))) | |
530 | (setq l (find-coding-systems-string str)) | |
b5edd1d1 KH |
531 | (if (eq codings t) |
532 | (setq codings l) | |
533 | (let ((ll nil)) | |
6053d86a KH |
534 | (dolist (elt codings) |
535 | (if (memq elt l) | |
536 | (setq ll (cons elt ll)))) | |
b5edd1d1 | 537 | (setq codings ll))))) |
6053d86a | 538 | codings)))) |
54b226f7 | 539 | |
51ed58ea KH |
540 | (defun find-multibyte-characters (from to &optional maxcount excludes) |
541 | "Find multibyte characters in the region specified by FROM and TO. | |
542 | If FROM is a string, find multibyte characters in the string. | |
543 | The return value is an alist of the following format: | |
544 | ((CHARSET COUNT CHAR ...) ...) | |
545 | where | |
546 | CHARSET is a character set, | |
547 | COUNT is a number of characters, | |
548 | CHARs are found characters of the character set. | |
549 | Optional 3rd arg MAXCOUNT limits how many CHARs are put in the above list. | |
251d4f4b KH |
550 | Optional 4th arg EXCLUDE is a list of character sets to be ignored. |
551 | ||
552 | For invalid characters, CHARs are actually strings." | |
51ed58ea KH |
553 | (let ((chars nil) |
554 | charset char) | |
555 | (if (stringp from) | |
556 | (let ((idx 0)) | |
557 | (while (setq idx (string-match "[^\000-\177]" from idx)) | |
558 | (setq char (aref from idx) | |
559 | charset (char-charset char)) | |
251d4f4b KH |
560 | (if (eq charset 'unknown) |
561 | (setq char (match-string 0))) | |
14333e31 KH |
562 | (if (or (memq charset '(unknown |
563 | eight-bit-control eight-bit-graphic)) | |
251d4f4b | 564 | (not (or (eq excludes t) (memq charset excludes)))) |
51ed58ea KH |
565 | (let ((slot (assq charset chars))) |
566 | (if slot | |
567 | (if (not (memq char (nthcdr 2 slot))) | |
568 | (let ((count (nth 1 slot))) | |
569 | (setcar (cdr slot) (1+ count)) | |
570 | (if (or (not maxcount) (< count maxcount)) | |
571 | (nconc slot (list char))))) | |
572 | (setq chars (cons (list charset 1 char) chars))))) | |
573 | (setq idx (1+ idx)))) | |
574 | (save-excursion | |
575 | (goto-char from) | |
576 | (while (re-search-forward "[^\000-\177]" to t) | |
577 | (setq char (preceding-char) | |
578 | charset (char-charset char)) | |
251d4f4b KH |
579 | (if (eq charset 'unknown) |
580 | (setq char (match-string 0))) | |
14333e31 | 581 | (if (or (memq charset '(unknown eight-bit-control eight-bit-graphic)) |
251d4f4b | 582 | (not (or (eq excludes t) (memq charset excludes)))) |
51ed58ea KH |
583 | (let ((slot (assq charset chars))) |
584 | (if slot | |
251d4f4b | 585 | (if (not (member char (nthcdr 2 slot))) |
51ed58ea KH |
586 | (let ((count (nth 1 slot))) |
587 | (setcar (cdr slot) (1+ count)) | |
588 | (if (or (not maxcount) (< count maxcount)) | |
589 | (nconc slot (list char))))) | |
590 | (setq chars (cons (list charset 1 char) chars)))))))) | |
591 | (nreverse chars))) | |
592 | ||
738746ba KH |
593 | |
594 | (defun search-unencodable-char (coding-system) | |
595 | "Search forward from point for a character that is not encodable. | |
596 | It asks which coding system to check. | |
597 | If such a character is found, set point after that character. | |
598 | Otherwise, don't move point. | |
599 | ||
600 | When called from a program, the value is a position of the found character, | |
601 | or nil if all characters are encodable." | |
602 | (interactive | |
603 | (list (let ((default (or buffer-file-coding-system 'us-ascii))) | |
604 | (read-coding-system | |
605 | (format "Coding-system (default, %s): " default) | |
606 | default)))) | |
607 | (let ((pos (unencodable-char-position (point) (point-max) coding-system))) | |
608 | (if pos | |
609 | (goto-char (1+ pos)) | |
610 | (message "All following characters are encodable by %s" coding-system)) | |
611 | pos)) | |
612 | ||
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 KH |
621 | (defvar select-safe-coding-system-accept-default-p nil |
622 | "If non-nil, a function to control the behaviour of coding system selection. | |
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 | ||
627 | (defun select-safe-coding-system (from to &optional default-coding-system | |
efa2c6d7 | 628 | accept-default-p file) |
d5266ddf KH |
629 | "Ask a user to select a safe coding system from candidates. |
630 | The candidates of coding systems which can safely encode a text | |
b5edd1d1 KH |
631 | between FROM and TO are shown in a popup window. Among them, the most |
632 | proper one is suggested as the default. | |
633 | ||
a985cd2f KH |
634 | The list of `buffer-file-coding-system' of the current buffer, |
635 | the `default-buffer-file-coding-system', and the | |
b5edd1d1 KH |
636 | most preferred coding system (if it corresponds to a MIME charset) is |
637 | treated as the default coding system list. Among them, the first one | |
d37ef0f6 DL |
638 | that safely encodes the text is normally selected silently and |
639 | returned without any user interaction. See also the command | |
640 | `prefer-coding-system'. | |
641 | ||
642 | However, the user is queried if the chosen coding system is | |
643 | inconsistent with what would be selected by `set-auto-coding' from | |
644 | coding cookies &c. if the contents of the region were read from a | |
645 | file. (That could lead to data corruption in a file subsequently | |
646 | re-visited and edited.) | |
54b226f7 | 647 | |
b5edd1d1 KH |
648 | Optional 3rd arg DEFAULT-CODING-SYSTEM specifies a coding system or a |
649 | list of coding systems to be prepended to the default coding system | |
0e9ec609 KH |
650 | list. However, if DEFAULT-CODING-SYSTEM is a list and the first |
651 | element is t, the cdr part is used as the defualt coding system list, | |
a985cd2f KH |
652 | i.e. `buffer-file-coding-system', `default-buffer-file-coding-system', |
653 | and the most preferred coding system are not used. | |
54b226f7 | 654 | |
b5edd1d1 KH |
655 | Optional 4th arg ACCEPT-DEFAULT-P, if non-nil, is a function to |
656 | determine the acceptability of the silently selected coding system. | |
657 | It is called with that coding system, and should return nil if it | |
658 | should not be silently selected and thus user interaction is required. | |
659 | ||
efa2c6d7 RS |
660 | Optional 5th arg FILE is the file name to use for this purpose. |
661 | That is different from `buffer-file-name' when handling `write-region' | |
662 | \(for example). | |
663 | ||
b5edd1d1 KH |
664 | The variable `select-safe-coding-system-accept-default-p', if |
665 | non-nil, overrides ACCEPT-DEFAULT-P. | |
54b226f7 KH |
666 | |
667 | Kludgy feature: if FROM is a string, the string is the target text, | |
668 | and TO is ignored." | |
b5edd1d1 KH |
669 | (if (and default-coding-system |
670 | (not (listp default-coding-system))) | |
671 | (setq default-coding-system (list default-coding-system))) | |
672 | ||
0e9ec609 KH |
673 | (let ((no-other-defaults nil)) |
674 | (if (eq (car default-coding-system) t) | |
675 | (setq no-other-defaults t | |
676 | default-coding-system (cdr default-coding-system))) | |
677 | ||
678 | ;; Change elements of the list to (coding . base-coding). | |
679 | (setq default-coding-system | |
680 | (mapcar (function (lambda (x) (cons x (coding-system-base x)))) | |
681 | default-coding-system)) | |
682 | ||
a985cd2f KH |
683 | ;; From now on, the list of defaults is reversed. |
684 | (setq default-coding-system (nreverse default-coding-system)) | |
685 | ||
0e9ec609 KH |
686 | (unless no-other-defaults |
687 | ;; If buffer-file-coding-system is not nil nor undecided, append it | |
688 | ;; to the defaults. | |
689 | (if buffer-file-coding-system | |
690 | (let ((base (coding-system-base buffer-file-coding-system))) | |
691 | (or (eq base 'undecided) | |
0e9ec609 | 692 | (rassq base default-coding-system) |
a985cd2f KH |
693 | (push (cons buffer-file-coding-system base) |
694 | default-coding-system)))) | |
695 | ||
696 | ;; If default-buffer-file-coding-system is not nil nor undecided, | |
697 | ;; append it to the defaults. | |
698 | (if default-buffer-file-coding-system | |
699 | (let ((base (coding-system-base default-buffer-file-coding-system))) | |
700 | (or (eq base 'undecided) | |
701 | (rassq base default-coding-system) | |
702 | (push (cons default-buffer-file-coding-system base) | |
703 | default-coding-system)))) | |
0e9ec609 KH |
704 | |
705 | ;; If the most preferred coding system has the property mime-charset, | |
706 | ;; append it to the defaults. | |
707 | (let ((tail coding-category-list) | |
708 | preferred base) | |
a985cd2f | 709 | (while (and tail (not (setq preferred (symbol-value (car tail))))) |
0e9ec609 KH |
710 | (setq tail (cdr tail))) |
711 | (and (coding-system-p preferred) | |
712 | (setq base (coding-system-base preferred)) | |
713 | (coding-system-get preferred 'mime-charset) | |
0e9ec609 | 714 | (not (rassq base default-coding-system)) |
a985cd2f KH |
715 | (push (cons preferred base) |
716 | default-coding-system))))) | |
b5edd1d1 KH |
717 | |
718 | (if select-safe-coding-system-accept-default-p | |
719 | (setq accept-default-p select-safe-coding-system-accept-default-p)) | |
720 | ||
721 | (let ((codings (find-coding-systems-region from to)) | |
722 | (coding-system nil) | |
45610c98 | 723 | (bufname (buffer-name)) |
c0d3ed97 | 724 | safe rejected unsafe) |
b5edd1d1 KH |
725 | (if (eq (car codings) 'undecided) |
726 | ;; Any coding system is ok. | |
727 | (setq coding-system t) | |
c0d3ed97 KH |
728 | ;; Classify the defaults into safe, rejected, and unsafe. |
729 | (dolist (elt default-coding-system) | |
730 | (if (memq (cdr elt) codings) | |
731 | (if (and (functionp accept-default-p) | |
732 | (not (funcall accept-default-p (cdr elt)))) | |
733 | (push (car elt) rejected) | |
734 | (push (car elt) safe)) | |
735 | (push (car elt) unsafe))) | |
736 | (if safe | |
a985cd2f | 737 | (setq coding-system (car safe)))) |
c0d3ed97 | 738 | |
b5edd1d1 | 739 | ;; If all the defaults failed, ask a user. |
c0d3ed97 KH |
740 | (when (not coding-system) |
741 | ;; At first, if some defaults are unsafe, record at most 11 | |
742 | ;; problematic characters and their positions for them by turning | |
743 | ;; (CODING ...) | |
744 | ;; into | |
745 | ;; ((CODING (POS . CHAR) (POS . CHAR) ...) ...) | |
746 | (if unsafe | |
747 | (if (stringp from) | |
748 | (setq unsafe | |
749 | (mapcar #'(lambda (coding) | |
750 | (cons coding | |
751 | (mapcar #'(lambda (pos) | |
752 | (cons pos (aref from pos))) | |
753 | (unencodable-char-position | |
754 | 0 (length from) coding | |
755 | 11 from)))) | |
756 | unsafe)) | |
a1506d29 | 757 | (setq unsafe |
c0d3ed97 KH |
758 | (mapcar #'(lambda (coding) |
759 | (cons coding | |
760 | (mapcar #'(lambda (pos) | |
761 | (cons pos (char-after pos))) | |
762 | (unencodable-char-position | |
763 | from to coding 11)))) | |
764 | unsafe)))) | |
738746ba KH |
765 | |
766 | ;; Change each safe coding system to the corresponding | |
b5edd1d1 KH |
767 | ;; mime-charset name if it is also a coding system. Such a name |
768 | ;; is more friendly to users. | |
769 | (let ((l codings) | |
34104362 KH |
770 | mime-charset) |
771 | (while l | |
772 | (setq mime-charset (coding-system-get (car l) 'mime-charset)) | |
773 | (if (and mime-charset (coding-system-p mime-charset)) | |
774 | (setcar l mime-charset)) | |
775 | (setq l (cdr l)))) | |
776 | ||
43afed8c RS |
777 | ;; Don't offer variations with locking shift, which you |
778 | ;; basically never want. | |
779 | (let (l) | |
780 | (dolist (elt codings (setq codings (nreverse l))) | |
781 | (unless (or (eq 'coding-category-iso-7-else | |
782 | (coding-system-category elt)) | |
783 | (eq 'coding-category-iso-8-else | |
784 | (coding-system-category elt))) | |
785 | (push elt l)))) | |
786 | ||
050afd64 SM |
787 | ;; Remove raw-text, emacs-mule and no-conversion unless nothing |
788 | ;; else is available. | |
789 | (setq codings | |
790 | (or (delq 'raw-text | |
791 | (delq 'emacs-mule | |
792 | (delq 'no-conversion codings))) | |
793 | '(raw-text emacs-mule no-conversion))) | |
794 | ||
738746ba KH |
795 | (let ((window-configuration (current-window-configuration))) |
796 | (save-excursion | |
c0d3ed97 KH |
797 | ;; If some defaults are unsafe, make sure the offending |
798 | ;; buffer is displayed. | |
799 | (when (and unsafe (not (stringp from))) | |
738746ba | 800 | (pop-to-buffer bufname) |
c0d3ed97 KH |
801 | (goto-char (apply 'min (mapcar #'(lambda (x) (car (cadr x))) |
802 | unsafe)))) | |
803 | ;; Then ask users to select one from CODINGS while showing | |
804 | ;; the reason why none of the defaults are not used. | |
738746ba | 805 | (with-output-to-temp-buffer "*Warning*" |
873a4be1 | 806 | (save-excursion |
738746ba KH |
807 | (set-buffer standard-output) |
808 | (if (not default-coding-system) | |
809 | (insert "No default coding systems to try for " | |
810 | (if (stringp from) | |
811 | (format "string \"%s\"." from) | |
812 | (format "buffer `%s'." bufname))) | |
813 | (insert | |
814 | "These default coding systems were tried to encode" | |
815 | (if (stringp from) | |
816 | (concat " \"" (if (> (length from) 10) | |
817 | (concat (substring from 0 10) "...\"") | |
818 | (concat from "\""))) | |
819 | (format " text\nin the buffer `%s'" bufname)) | |
820 | ":\n") | |
821 | (let ((pos (point)) | |
822 | (fill-prefix " ")) | |
c0d3ed97 KH |
823 | (mapc #'(lambda (x) (princ " ") (princ (car x))) |
824 | default-coding-system) | |
738746ba KH |
825 | (insert "\n") |
826 | (fill-region-as-paragraph pos (point))) | |
c0d3ed97 KH |
827 | (when rejected |
828 | (insert "These safely encodes the target text, | |
d37ef0f6 | 829 | but it is not recommended for encoding text in this context, |
c0d3ed97 KH |
830 | e.g., for sending an email message.\n ") |
831 | (mapc #'(lambda (x) (princ " ") (princ x)) rejected) | |
832 | (insert "\n")) | |
833 | (when unsafe | |
a1506d29 | 834 | (insert (if rejected "And the others" |
c0d3ed97 KH |
835 | "However, each of them") |
836 | " encountered these problematic characters:\n") | |
738746ba KH |
837 | (mapc |
838 | #'(lambda (coding) | |
839 | (insert (format " %s:" (car coding))) | |
c0d3ed97 KH |
840 | (let ((i 0) |
841 | (func1 | |
842 | #'(lambda (bufname pos) | |
843 | (when (buffer-live-p (get-buffer bufname)) | |
844 | (pop-to-buffer bufname) | |
845 | (goto-char pos)))) | |
846 | (func2 | |
738746ba KH |
847 | #'(lambda (bufname pos coding) |
848 | (when (buffer-live-p (get-buffer bufname)) | |
849 | (pop-to-buffer bufname) | |
850 | (if (< (point) pos) | |
851 | (goto-char pos) | |
852 | (forward-char 1) | |
853 | (search-unencodable-char coding) | |
c0d3ed97 KH |
854 | (forward-char -1)))))) |
855 | (dolist (elt (cdr coding)) | |
856 | (insert " ") | |
857 | (if (stringp from) | |
858 | (insert (if (< i 10) (cdr elt) "...")) | |
859 | (if (< i 10) | |
860 | (insert-text-button | |
861 | (cdr elt) | |
862 | :type 'help-xref | |
863 | 'help-echo | |
864 | "mouse-2, RET: jump to this character" | |
865 | 'help-function func1 | |
866 | 'help-args (list bufname (car elt))) | |
867 | (insert-text-button | |
868 | "..." | |
869 | :type 'help-xref | |
870 | 'help-echo | |
871 | "mouse-2, RET: next unencodable character" | |
872 | 'help-function func2 | |
873 | 'help-args (list bufname (car elt) | |
874 | (car coding))))) | |
875 | (setq i (1+ i)))) | |
738746ba | 876 | (insert "\n")) |
c0d3ed97 | 877 | unsafe) |
738746ba | 878 | (insert "\ |
d37ef0f6 | 879 | The first problematic character is at point in the displayed buffer,\n" |
738746ba | 880 | (substitute-command-keys "\ |
d37ef0f6 | 881 | and \\[universal-argument] \\[what-cursor-position] will give information about it.\n")))) |
c0d3ed97 | 882 | (insert (if safe |
738746ba KH |
883 | "\nSelect the above, or " |
884 | "\nSelect ") | |
885 | "\ | |
d37ef0f6 | 886 | one of the following safe coding systems, or edit the buffer:\n") |
738746ba KH |
887 | (let ((pos (point)) |
888 | (fill-prefix " ")) | |
889 | (mapcar (function (lambda (x) (princ " ") (princ x))) | |
890 | codings) | |
891 | (insert "\n") | |
0e9ec609 KH |
892 | (fill-region-as-paragraph pos (point))) |
893 | (insert "Or specify any other coding system | |
9ddf362e | 894 | on your risk of losing the problematic characters.\n"))) |
738746ba KH |
895 | |
896 | ;; Read a coding system. | |
0e9ec609 KH |
897 | (setq default-coding-system (or (car safe) (car codings))) |
898 | (setq coding-system | |
a1506d29 | 899 | (read-coding-system |
0e9ec609 KH |
900 | (format "Select coding system (default %s): " |
901 | default-coding-system) | |
902 | default-coding-system)) | |
903 | (setq last-coding-system-specified coding-system)) | |
904 | ||
738746ba KH |
905 | (kill-buffer "*Warning*") |
906 | (set-window-configuration window-configuration))) | |
b5edd1d1 KH |
907 | |
908 | (if (vectorp (coding-system-eol-type coding-system)) | |
909 | (let ((eol (coding-system-eol-type buffer-file-coding-system))) | |
910 | (if (numberp eol) | |
911 | (setq coding-system | |
912 | (coding-system-change-eol-conversion coding-system eol))))) | |
913 | ||
914 | (if (eq coding-system t) | |
915 | (setq coding-system buffer-file-coding-system)) | |
43afed8c RS |
916 | ;; Check we're not inconsistent with what `coding:' spec &c would |
917 | ;; give when file is re-read. | |
9e7638c3 RS |
918 | ;; But don't do this if we explicitly ignored the cookie |
919 | ;; by using `find-file-literally'. | |
920 | (unless (or (stringp from) find-file-literally) | |
efa2c6d7 RS |
921 | (let ((auto-cs (save-excursion |
922 | (save-restriction | |
923 | (widen) | |
924 | (narrow-to-region from to) | |
43afed8c | 925 | (goto-char (point-min)) |
efa2c6d7 | 926 | (set-auto-coding (or file buffer-file-name "") |
43afed8c | 927 | (buffer-size)))))) |
872c0ba8 | 928 | (if (and auto-cs coding-system |
d37ef0f6 DL |
929 | ;; Don't barf if writing a compressed file, say. |
930 | ;; This check perhaps isn't ideal, but is probably | |
931 | ;; the best thing to do. | |
932 | (not (auto-coding-alist-lookup (or file buffer-file-name ""))) | |
43afed8c RS |
933 | (not (coding-system-equal (coding-system-base coding-system) |
934 | (coding-system-base auto-cs)))) | |
935 | (unless (yes-or-no-p | |
936 | (format "Selected encoding %s disagrees with \ | |
937 | %s specified by file contents. Really save (else edit coding cookies \ | |
938 | and try again)? " coding-system auto-cs)) | |
41bfe57a RS |
939 | (error "Save aborted"))))) |
940 | coding-system)) | |
54b226f7 KH |
941 | |
942 | (setq select-safe-coding-system-function 'select-safe-coding-system) | |
943 | ||
46babb23 KH |
944 | (defun select-message-coding-system () |
945 | "Return a coding system to encode the outgoing message of the current buffer. | |
946 | It at first tries the first coding system found in these variables | |
947 | in this order: | |
948 | (1) local value of `buffer-file-coding-system' | |
949 | (2) value of `sendmail-coding-system' | |
b5edd1d1 KH |
950 | (3) value of `default-sendmail-coding-system' |
951 | (4) value of `default-buffer-file-coding-system' | |
46babb23 KH |
952 | If the found coding system can't encode the current buffer, |
953 | or none of them are bound to a coding system, | |
48e41165 | 954 | it asks the user to select a proper coding system." |
46babb23 | 955 | (let ((coding (or (and (local-variable-p 'buffer-file-coding-system) |
b5edd1d1 KH |
956 | buffer-file-coding-system) |
957 | sendmail-coding-system | |
958 | default-sendmail-coding-system | |
959 | default-buffer-file-coding-system))) | |
46babb23 | 960 | (if (eq coding 'no-conversion) |
d37ef0f6 | 961 | ;; We should never use no-conversion for outgoing mail. |
46babb23 KH |
962 | (setq coding nil)) |
963 | (if (fboundp select-safe-coding-system-function) | |
964 | (funcall select-safe-coding-system-function | |
b5edd1d1 KH |
965 | (point-min) (point-max) coding |
966 | (function (lambda (x) (coding-system-get x 'mime-charset)))) | |
46babb23 | 967 | coding))) |
4ed46869 | 968 | \f |
03c35c83 | 969 | ;;; Language support stuff. |
4ed46869 | 970 | |
4ed46869 | 971 | (defvar language-info-alist nil |
2c395d56 | 972 | "Alist of language environment definitions. |
4ed46869 KH |
973 | Each element looks like: |
974 | (LANGUAGE-NAME . ((KEY . INFO) ...)) | |
2c395d56 RS |
975 | where LANGUAGE-NAME is a string, the name of the language environment, |
976 | KEY is a symbol denoting the kind of information, and | |
977 | INFO is the data associated with KEY. | |
978 | Meaningful values for KEY include | |
979 | ||
980 | documentation value is documentation of what this language environment | |
981 | is meant for, and how to use it. | |
982 | charset value is a list of the character sets used by this | |
983 | language environment. | |
d37ef0f6 DL |
984 | sample-text value is an expression which is evalled to generate |
985 | a line of text written using characters appropriate | |
986 | for this language environment. | |
2c395d56 RS |
987 | setup-function value is a function to call to switch to this |
988 | language environment. | |
989 | exit-function value is a function to call to leave this | |
990 | language environment. | |
991 | coding-system value is a list of coding systems that are good | |
992 | for saving text written in this language environment. | |
993 | This list serves as suggestions to the user; | |
994 | in effect, as a kind of documentation. | |
995 | coding-priority value is a list of coding systems for this language | |
996 | environment, in order of decreasing priority. | |
997 | This is used to set up the coding system priority | |
45d08cb2 | 998 | list when you switch to this language environment. |
ddb5c041 | 999 | nonascii-translation |
7624ebb9 | 1000 | value is a translation table to be set in the |
45d08cb2 | 1001 | variable `nonascii-translation-table' in this |
7624ebb9 KH |
1002 | language environment, or a character set from |
1003 | which `nonascii-insert-offset' is calculated. | |
ddb5c041 KH |
1004 | input-method value is a default input method for this language |
1005 | environment. | |
7624ebb9 KH |
1006 | features value is a list of features requested in this |
1007 | language environment. | |
ddb5c041 KH |
1008 | |
1009 | The following keys take effect only when multibyte characters are | |
1010 | globally disabled, i.e. the value of `default-enable-multibyte-characters' | |
1011 | is nil. | |
1012 | ||
1013 | unibyte-syntax value is a library name to load to set | |
e8dd0160 | 1014 | unibyte 8-bit character syntaxes for this |
ddb5c041 KH |
1015 | language environment. |
1016 | ||
1017 | unibyte-display value is a coding system to encode characters | |
1018 | for the terminal. Characters in the range | |
1019 | of 160 to 255 display not as octal escapes, | |
1020 | but as non-ASCII characters in this language | |
1021 | environment.") | |
2c395d56 RS |
1022 | |
1023 | (defun get-language-info (lang-env key) | |
1024 | "Return information listed under KEY for language environment LANG-ENV. | |
1025 | KEY is a symbol denoting the kind of information. | |
1026 | For a list of useful values for KEY and their meanings, | |
1027 | see `language-info-alist'." | |
1028 | (if (symbolp lang-env) | |
1029 | (setq lang-env (symbol-name lang-env))) | |
1030 | (let ((lang-slot (assoc-ignore-case lang-env language-info-alist))) | |
4ed46869 KH |
1031 | (if lang-slot |
1032 | (cdr (assq key (cdr lang-slot)))))) | |
1033 | ||
f08adf27 | 1034 | (defun set-language-info (lang-env key info) |
2c395d56 RS |
1035 | "Modify part of the definition of language environment LANG-ENV. |
1036 | Specifically, this stores the information INFO under KEY | |
1037 | in the definition of this language environment. | |
4ed46869 | 1038 | KEY is a symbol denoting the kind of information. |
2c395d56 | 1039 | INFO is the value for that information. |
281d03ec | 1040 | |
2c395d56 | 1041 | For a list of useful values for KEY and their meanings, |
f08adf27 | 1042 | see `language-info-alist'." |
2c395d56 RS |
1043 | (if (symbolp lang-env) |
1044 | (setq lang-env (symbol-name lang-env))) | |
4ed46869 | 1045 | (let (lang-slot key-slot) |
2c395d56 | 1046 | (setq lang-slot (assoc lang-env language-info-alist)) |
4ed46869 | 1047 | (if (null lang-slot) ; If no slot for the language, add it. |
2c395d56 | 1048 | (setq lang-slot (list lang-env) |
4ed46869 KH |
1049 | language-info-alist (cons lang-slot language-info-alist))) |
1050 | (setq key-slot (assq key lang-slot)) | |
1051 | (if (null key-slot) ; If no slot for the key, add it. | |
1052 | (progn | |
1053 | (setq key-slot (list key)) | |
1054 | (setcdr lang-slot (cons key-slot (cdr lang-slot))))) | |
590dbcba | 1055 | (setcdr key-slot (purecopy info)) |
1169bd86 | 1056 | ;; Update the custom-type of `current-language-environment'. |
590dbcba | 1057 | (put 'current-language-environment 'custom-type |
1169bd86 MR |
1058 | (cons 'choice (mapcar |
1059 | (lambda (lang) | |
756e055f MR |
1060 | (list 'const lang)) |
1061 | (sort (mapcar 'car language-info-alist) 'string<)))))) | |
4ed46869 | 1062 | |
2c395d56 RS |
1063 | (defun set-language-info-alist (lang-env alist &optional parents) |
1064 | "Store ALIST as the definition of language environment LANG-ENV. | |
1065 | ALIST is an alist of KEY and INFO values. See the documentation of | |
98c6d6ed | 1066 | `language-info-alist' for the meanings of KEY and INFO. |
54b226f7 | 1067 | |
2c395d56 | 1068 | Optional arg PARENTS is a list of parent menu names; it specifies |
d37ef0f6 | 1069 | where to put this language environment in the |
2c395d56 RS |
1070 | Describe Language Environment and Set Language Environment menus. |
1071 | For example, (\"European\") means to put this language environment | |
1072 | in the European submenu in each of those two menus." | |
1073 | (if (symbolp lang-env) | |
1074 | (setq lang-env (symbol-name lang-env))) | |
54b226f7 KH |
1075 | (let ((describe-map describe-language-environment-map) |
1076 | (setup-map setup-language-environment-map)) | |
1077 | (if parents | |
1078 | (let ((l parents) | |
9deed82f | 1079 | map parent-symbol parent prompt) |
54b226f7 KH |
1080 | (while l |
1081 | (if (symbolp (setq parent-symbol (car l))) | |
1082 | (setq parent (symbol-name parent)) | |
1083 | (setq parent parent-symbol parent-symbol (intern parent))) | |
1084 | (setq map (lookup-key describe-map (vector parent-symbol))) | |
9deed82f EZ |
1085 | ;; This prompt string is for define-prefix-command, so |
1086 | ;; that the map it creates will be suitable for a menu. | |
1087 | (or map (setq prompt (format "%s Environment" parent))) | |
54b226f7 KH |
1088 | (if (not map) |
1089 | (progn | |
1090 | (setq map (intern (format "describe-%s-environment-map" | |
1091 | (downcase parent)))) | |
9deed82f | 1092 | (define-prefix-command map nil prompt) |
54b226f7 KH |
1093 | (define-key-after describe-map (vector parent-symbol) |
1094 | (cons parent map) t))) | |
1095 | (setq describe-map (symbol-value map)) | |
1096 | (setq map (lookup-key setup-map (vector parent-symbol))) | |
1097 | (if (not map) | |
1098 | (progn | |
1099 | (setq map (intern (format "setup-%s-environment-map" | |
1100 | (downcase parent)))) | |
9deed82f | 1101 | (define-prefix-command map nil prompt) |
54b226f7 KH |
1102 | (define-key-after setup-map (vector parent-symbol) |
1103 | (cons parent map) t))) | |
1104 | (setq setup-map (symbol-value map)) | |
1105 | (setq l (cdr l))))) | |
f08adf27 RS |
1106 | |
1107 | ;; Set up menu items for this language env. | |
7624ebb9 | 1108 | (let ((doc (assq 'documentation alist))) |
f08adf27 RS |
1109 | (when doc |
1110 | (define-key-after describe-map (vector (intern lang-env)) | |
7624ebb9 KH |
1111 | (cons lang-env 'describe-specified-language-support) t))) |
1112 | (define-key-after setup-map (vector (intern lang-env)) | |
1113 | (cons lang-env 'setup-specified-language-environment) t) | |
f08adf27 | 1114 | |
54b226f7 | 1115 | (while alist |
f08adf27 | 1116 | (set-language-info lang-env (car (car alist)) (cdr (car alist))) |
54b226f7 | 1117 | (setq alist (cdr alist))))) |
4ed46869 | 1118 | |
ae302641 | 1119 | (defun read-language-name (key prompt &optional default) |
2c395d56 | 1120 | "Read a language environment name which has information for KEY. |
ddb5c041 | 1121 | If KEY is nil, read any language environment. |
2c395d56 RS |
1122 | Prompt with PROMPT. DEFAULT is the default choice of language environment. |
1123 | This returns a language environment name as a string." | |
4ed46869 KH |
1124 | (let* ((completion-ignore-case t) |
1125 | (name (completing-read prompt | |
1126 | language-info-alist | |
ddb5c041 | 1127 | (and key |
ca429a25 | 1128 | (function (lambda (elm) (and (listp elm) (assq key elm))))) |
ae302641 | 1129 | t nil nil default))) |
13e82c04 | 1130 | (if (and (> (length name) 0) |
ddb5c041 KH |
1131 | (or (not key) |
1132 | (get-language-info name key))) | |
13e82c04 | 1133 | name))) |
4ed46869 KH |
1134 | \f |
1135 | ;;; Multilingual input methods. | |
d37ef0f6 | 1136 | (defgroup leim nil |
d0c40faa KH |
1137 | "LEIM: Libraries of Emacs Input Methods." |
1138 | :group 'mule) | |
4ed46869 | 1139 | |
d0b9c3ab KH |
1140 | (defconst leim-list-file-name "leim-list.el" |
1141 | "Name of LEIM list file. | |
1142 | This file contains a list of libraries of Emacs input methods (LEIM) | |
1143 | in the format of Lisp expression for registering each input method. | |
1144 | Emacs loads this file at startup time.") | |
1145 | ||
2e224638 | 1146 | (defvar leim-list-header (format |
d37ef0f6 | 1147 | ";;; %s -- list of LEIM (Library of Emacs Input Method) -*-coding: iso-2022-7bit;-*- |
d0b9c3ab KH |
1148 | ;; |
1149 | ;; This file contains a list of LEIM (Library of Emacs Input Method) | |
c654de1d DL |
1150 | ;; methods in the same directory as this file. Loading this file |
1151 | ;; registers all the input methods in Emacs. | |
d0b9c3ab | 1152 | ;; |
d33d5fbe | 1153 | ;; Each entry has the form: |
d0b9c3ab KH |
1154 | ;; (register-input-method |
1155 | ;; INPUT-METHOD LANGUAGE-NAME ACTIVATE-FUNC | |
1156 | ;; TITLE DESCRIPTION | |
1157 | ;; ARG ...) | |
c654de1d | 1158 | ;; See the function `register-input-method' for the meanings of the arguments. |
d0b9c3ab KH |
1159 | ;; |
1160 | ;; If this directory is included in load-path, Emacs automatically | |
1161 | ;; loads this file at startup time. | |
1162 | ||
1163 | " | |
1164 | leim-list-file-name) | |
1165 | "Header to be inserted in LEIM list file.") | |
1166 | ||
e55e92ee | 1167 | (defvar leim-list-entry-regexp "^(register-input-method" |
d0b9c3ab KH |
1168 | "Regexp matching head of each entry in LEIM list file. |
1169 | See also the variable `leim-list-header'") | |
1170 | ||
1171 | (defvar update-leim-list-functions | |
1172 | '(quail-update-leim-list-file) | |
1173 | "List of functions to call to update LEIM list file. | |
1174 | Each function is called with one arg, LEIM directory name.") | |
1175 | ||
a337fe7f RS |
1176 | (defun update-leim-list-file (&rest dirs) |
1177 | "Update LEIM list file in directories DIRS." | |
d0b9c3ab KH |
1178 | (let ((functions update-leim-list-functions)) |
1179 | (while functions | |
a337fe7f | 1180 | (apply (car functions) dirs) |
d0b9c3ab KH |
1181 | (setq functions (cdr functions))))) |
1182 | ||
4ed46869 KH |
1183 | (defvar current-input-method nil |
1184 | "The current input method for multilingual text. | |
96db204a | 1185 | If nil, that means no input method is activated now.") |
4ed46869 KH |
1186 | (make-variable-buffer-local 'current-input-method) |
1187 | (put 'current-input-method 'permanent-local t) | |
1188 | ||
1189 | (defvar current-input-method-title nil | |
d0b9c3ab | 1190 | "Title string of the current input method shown in mode line.") |
4ed46869 KH |
1191 | (make-variable-buffer-local 'current-input-method-title) |
1192 | (put 'current-input-method-title 'permanent-local t) | |
1193 | ||
b4fba33f | 1194 | (defcustom default-input-method nil |
8861c593 | 1195 | "*Default input method for multilingual text (a string). |
b4fba33f | 1196 | This is the input method activated automatically by the command |
9b10b5a3 | 1197 | `toggle-input-method' (\\[toggle-input-method])." |
8861c593 | 1198 | :group 'mule |
5806e8a6 GM |
1199 | :type '(choice (const nil) string) |
1200 | :set-after '(current-language-environment)) | |
b4fba33f | 1201 | |
0f835e87 KH |
1202 | (put 'input-method-function 'permanent-local t) |
1203 | ||
723a427a KH |
1204 | (defvar input-method-history nil |
1205 | "History list for some commands that read input methods.") | |
1206 | (make-variable-buffer-local 'input-method-history) | |
1207 | (put 'input-method-history 'permanent-local t) | |
4ed46869 KH |
1208 | |
1209 | (defvar inactivate-current-input-method-function nil | |
1210 | "Function to call for inactivating the current input method. | |
1211 | Every input method should set this to an appropriate value when activated. | |
f17ccaee KH |
1212 | This function is called with no argument. |
1213 | ||
1214 | This function should never change the value of `current-input-method'. | |
1215 | It is set to nil by the function `inactivate-input-method'.") | |
4ed46869 KH |
1216 | (make-variable-buffer-local 'inactivate-current-input-method-function) |
1217 | (put 'inactivate-current-input-method-function 'permanent-local t) | |
1218 | ||
1219 | (defvar describe-current-input-method-function nil | |
1220 | "Function to call for describing the current input method. | |
1221 | This function is called with no argument.") | |
1222 | (make-variable-buffer-local 'describe-current-input-method-function) | |
1223 | (put 'describe-current-input-method-function 'permanent-local t) | |
1224 | ||
d0b9c3ab | 1225 | (defvar input-method-alist nil |
2c395d56 | 1226 | "Alist of input method names vs how to use them. |
d0b9c3ab | 1227 | Each element has the form: |
2c395d56 RS |
1228 | (INPUT-METHOD LANGUAGE-ENV ACTIVATE-FUNC TITLE DESCRIPTION ARGS...) |
1229 | See the function `register-input-method' for the meanings of the elements.") | |
1230 | ||
f08adf27 | 1231 | (defun register-input-method (input-method lang-env &rest args) |
bf42aa15 | 1232 | "Register INPUT-METHOD as an input method for language environment LANG-ENV. |
f08adf27 | 1233 | INPUT-METHOD and LANG-ENV are symbols or strings. |
d0b9c3ab | 1234 | |
d0b9c3ab | 1235 | The remaining arguments are: |
2c395d56 RS |
1236 | ACTIVATE-FUNC, TITLE, DESCRIPTION, and ARGS... |
1237 | ACTIVATE-FUNC is a function to call to activate this method. | |
1238 | TITLE is a string to show in the mode line when this method is active. | |
1239 | DESCRIPTION is a string describing this method and what it is good for. | |
1240 | The ARGS, if any, are passed as arguments to ACTIVATE-FUNC. | |
205814ee KH |
1241 | All told, the arguments to ACTIVATE-FUNC are INPUT-METHOD and the ARGS. |
1242 | ||
c654de1d DL |
1243 | This function is mainly used in the file \"leim-list.el\" which is |
1244 | created at Emacs build time, registering all Quail input methods | |
0b6cadff | 1245 | contained in the Emacs distribution. |
205814ee | 1246 | |
0b6cadff | 1247 | In case you want to register a new Quail input method by yourself, be |
205814ee | 1248 | careful to use the same input method title as given in the third |
0b6cadff DL |
1249 | parameter of `quail-define-package'. (If the values are different, the |
1250 | string specified in this function takes precedence.) | |
205814ee KH |
1251 | |
1252 | The commands `describe-input-method' and `list-input-methods' need | |
0b6cadff DL |
1253 | these duplicated values to show some information about input methods |
1254 | without loading the relevant Quail packages." | |
f08adf27 RS |
1255 | (if (symbolp lang-env) |
1256 | (setq lang-env (symbol-name lang-env))) | |
4ef06f75 KH |
1257 | (if (symbolp input-method) |
1258 | (setq input-method (symbol-name input-method))) | |
f08adf27 | 1259 | (let ((info (cons lang-env args)) |
d0b9c3ab KH |
1260 | (slot (assoc input-method input-method-alist))) |
1261 | (if slot | |
1262 | (setcdr slot info) | |
1263 | (setq slot (cons input-method info)) | |
1264 | (setq input-method-alist (cons slot input-method-alist))))) | |
1265 | ||
4d5ac029 | 1266 | (defun read-input-method-name (prompt &optional default inhibit-null) |
d0b9c3ab | 1267 | "Read a name of input method from a minibuffer prompting with PROMPT. |
4d5ac029 | 1268 | If DEFAULT is non-nil, use that as the default, |
0b6cadff | 1269 | and substitute it into PROMPT at the first `%s'. |
4ef06f75 KH |
1270 | If INHIBIT-NULL is non-nil, null input signals an error. |
1271 | ||
1272 | The return value is a string." | |
4d5ac029 RS |
1273 | (if default |
1274 | (setq prompt (format prompt default))) | |
d0b9c3ab | 1275 | (let* ((completion-ignore-case t) |
c54044ff KH |
1276 | ;; As it is quite normal to change input method in the |
1277 | ;; minibuffer, we must enable it even if | |
1278 | ;; enable-recursive-minibuffers is currently nil. | |
1279 | (enable-recursive-minibuffers t) | |
723a427a KH |
1280 | ;; This binding is necessary because input-method-history is |
1281 | ;; buffer local. | |
d0b9c3ab | 1282 | (input-method (completing-read prompt input-method-alist |
87505a98 RS |
1283 | nil t nil 'input-method-history |
1284 | default))) | |
bf294e6e KH |
1285 | (if (and input-method (symbolp input-method)) |
1286 | (setq input-method (symbol-name input-method))) | |
d0b9c3ab KH |
1287 | (if (> (length input-method) 0) |
1288 | input-method | |
1289 | (if inhibit-null | |
43807b77 | 1290 | (error "No valid input method is specified"))))) |
d0b9c3ab | 1291 | |
d0b9c3ab | 1292 | (defun activate-input-method (input-method) |
2c395d56 RS |
1293 | "Switch to input method INPUT-METHOD for the current buffer. |
1294 | If some other input method is already active, turn it off first. | |
1295 | If INPUT-METHOD is nil, deactivate any current input method." | |
305a3cb6 | 1296 | (if (and input-method (symbolp input-method)) |
4ef06f75 | 1297 | (setq input-method (symbol-name input-method))) |
723a427a KH |
1298 | (if (and current-input-method |
1299 | (not (string= current-input-method input-method))) | |
305a3cb6 | 1300 | (inactivate-input-method)) |
2c395d56 | 1301 | (unless (or current-input-method (null input-method)) |
d0b9c3ab KH |
1302 | (let ((slot (assoc input-method input-method-alist))) |
1303 | (if (null slot) | |
723a427a | 1304 | (error "Can't activate input method `%s'" input-method)) |
278dd6ac | 1305 | (setq current-input-method-title nil) |
8efc03e1 KH |
1306 | (let ((func (nth 2 slot))) |
1307 | (if (functionp func) | |
1308 | (apply (nth 2 slot) input-method (nthcdr 5 slot)) | |
1309 | (if (and (consp func) (symbolp (car func)) (symbolp (cdr func))) | |
1310 | (progn | |
1311 | (require (cdr func)) | |
1312 | (apply (car func) input-method (nthcdr 5 slot))) | |
1313 | (error "Can't activate input method `%s'" input-method)))) | |
d0b9c3ab | 1314 | (setq current-input-method input-method) |
278dd6ac KH |
1315 | (or (stringp current-input-method-title) |
1316 | (setq current-input-method-title (nth 3 slot))) | |
28885c0e KH |
1317 | (unwind-protect |
1318 | (run-hooks 'input-method-activate-hook) | |
1319 | (force-mode-line-update))))) | |
15b3e511 | 1320 | |
15b3e511 | 1321 | (defun inactivate-input-method () |
f17ccaee | 1322 | "Turn off the current input method." |
723a427a KH |
1323 | (when current-input-method |
1324 | (if input-method-history | |
1325 | (unless (string= current-input-method (car input-method-history)) | |
1326 | (setq input-method-history | |
1327 | (cons current-input-method | |
1328 | (delete current-input-method input-method-history)))) | |
1329 | (setq input-method-history (list current-input-method))) | |
1330 | (unwind-protect | |
1331 | (funcall inactivate-current-input-method-function) | |
15b3e511 | 1332 | (unwind-protect |
723a427a KH |
1333 | (run-hooks 'input-method-inactivate-hook) |
1334 | (setq current-input-method nil | |
2b1ec973 | 1335 | input-method-function nil |
28885c0e KH |
1336 | current-input-method-title nil) |
1337 | (force-mode-line-update))))) | |
4ed46869 | 1338 | |
8b784951 | 1339 | (defun set-input-method (input-method) |
2c395d56 | 1340 | "Select and activate input method INPUT-METHOD for the current buffer. |
bc406911 | 1341 | This also sets the default input method to the one you specify. |
402dbbd1 EZ |
1342 | If INPUT-METHOD is nil, this function turns off the input method, and |
1343 | also causes you to be prompted for a name of an input method the next | |
1344 | time you invoke \\[toggle-input-method]. | |
1345 | ||
bc406911 EZ |
1346 | To deactivate the input method interactively, use \\[toggle-input-method]. |
1347 | To deactivate it programmatically, use \\[inactivate-input-method]." | |
d0b9c3ab | 1348 | (interactive |
723a427a | 1349 | (let* ((default (or (car input-method-history) default-input-method))) |
42395763 | 1350 | (list (read-input-method-name |
87505a98 | 1351 | (if default "Select input method (default %s): " "Select input method: ") |
42395763 | 1352 | default t)))) |
d0b9c3ab | 1353 | (activate-input-method input-method) |
f4990970 PA |
1354 | (setq default-input-method input-method) |
1355 | (when (interactive-p) | |
1356 | (customize-mark-as-set 'default-input-method)) | |
1357 | default-input-method) | |
4ed46869 KH |
1358 | |
1359 | (defun toggle-input-method (&optional arg) | |
f8ec20be RS |
1360 | "Enable or disable multilingual text input method for the current buffer. |
1361 | Only one input method can be enabled at any time in a given buffer. | |
1362 | ||
1363 | The normal action is to enable an input method if none was | |
1364 | enabled, and disable the current one otherwise. Which input method | |
1365 | to enable can be determined in various ways--either the one most | |
1366 | recently used, or the one specified by `default-input-method', or | |
1367 | as a last resort by reading the name of an input method in the | |
1368 | minibuffer. | |
1369 | ||
1370 | With a prefix argument, read an input method name with the minibuffer | |
1371 | and enable that one. The default is the most recent input method specified | |
f2979bdb | 1372 | \(not including the currently active input method, if any)." |
f8ec20be | 1373 | |
4ed46869 | 1374 | (interactive "P") |
7ddbb5bc RS |
1375 | (if (and current-input-method (not arg)) |
1376 | (inactivate-input-method) | |
1377 | (let ((default (or (car input-method-history) default-input-method))) | |
1378 | (if (and arg default (equal current-input-method default) | |
1379 | (> (length input-method-history) 1)) | |
1380 | (setq default (nth 1 input-method-history))) | |
723a427a KH |
1381 | (activate-input-method |
1382 | (if (or arg (not default)) | |
7ddbb5bc RS |
1383 | (progn |
1384 | (read-input-method-name | |
1385 | (if default "Input method (default %s): " "Input method: " ) | |
1386 | default t)) | |
723a427a | 1387 | default)) |
f4990970 | 1388 | (unless default-input-method |
d37ef0f6 | 1389 | (prog1 |
f4990970 PA |
1390 | (setq default-input-method current-input-method) |
1391 | (when (interactive-p) | |
1392 | (customize-mark-as-set 'default-input-method))))))) | |
d0b9c3ab KH |
1393 | |
1394 | (defun describe-input-method (input-method) | |
2c395d56 | 1395 | "Describe input method INPUT-METHOD." |
d0b9c3ab KH |
1396 | (interactive |
1397 | (list (read-input-method-name | |
1398 | "Describe input method (default, current choice): "))) | |
78754934 | 1399 | (if (and input-method (symbolp input-method)) |
4ef06f75 | 1400 | (setq input-method (symbol-name input-method))) |
43125c28 RS |
1401 | (help-setup-xref (list #'describe-input-method |
1402 | (or input-method current-input-method)) | |
f80e2142 RS |
1403 | (interactive-p)) |
1404 | ||
d0b9c3ab KH |
1405 | (if (null input-method) |
1406 | (describe-current-input-method) | |
464cc130 KH |
1407 | (let ((current current-input-method)) |
1408 | (condition-case nil | |
1409 | (progn | |
1410 | (save-excursion | |
1411 | (activate-input-method input-method) | |
1412 | (describe-current-input-method)) | |
1413 | (activate-input-method current)) | |
d37ef0f6 | 1414 | (error |
464cc130 | 1415 | (activate-input-method current) |
5f395df3 SM |
1416 | (help-setup-xref (list #'describe-input-method input-method) |
1417 | (interactive-p)) | |
1418 | (with-output-to-temp-buffer (help-buffer) | |
464cc130 KH |
1419 | (let ((elt (assoc input-method input-method-alist))) |
1420 | (princ (format | |
1421 | "Input method: %s (`%s' in mode line) for %s\n %s\n" | |
1422 | input-method (nth 3 elt) (nth 1 elt) (nth 4 elt)))))))))) | |
d0b9c3ab KH |
1423 | |
1424 | (defun describe-current-input-method () | |
f80e2142 RS |
1425 | "Describe the input method currently in use. |
1426 | This is a subroutine for `describe-input-method'." | |
4ed46869 KH |
1427 | (if current-input-method |
1428 | (if (and (symbolp describe-current-input-method-function) | |
1429 | (fboundp describe-current-input-method-function)) | |
1430 | (funcall describe-current-input-method-function) | |
1431 | (message "No way to describe the current input method `%s'" | |
f2979bdb | 1432 | current-input-method) |
4ed46869 | 1433 | (ding)) |
d0b9c3ab | 1434 | (error "No input method is activated now"))) |
4ed46869 | 1435 | |
d3459641 | 1436 | (defun read-multilingual-string (prompt &optional initial-input input-method) |
4ed46869 KH |
1437 | "Read a multilingual string from minibuffer, prompting with string PROMPT. |
1438 | The input method selected last time is activated in minibuffer. | |
15b3e511 | 1439 | If optional second arg INITIAL-INPUT is non-nil, insert it in the minibuffer |
d0b9c3ab KH |
1440 | initially. |
1441 | Optional 3rd argument INPUT-METHOD specifies the input method | |
4ef06f75 KH |
1442 | to be activated instead of the one selected last time. It is a symbol |
1443 | or a string." | |
88d559ec KH |
1444 | (setq input-method |
1445 | (or input-method | |
d3459641 | 1446 | current-input-method |
88d559ec KH |
1447 | default-input-method |
1448 | (read-input-method-name "Input method: " nil t))) | |
3df60841 | 1449 | (if (and input-method (symbolp input-method)) |
4ef06f75 | 1450 | (setq input-method (symbol-name input-method))) |
305a3cb6 KH |
1451 | (let ((prev-input-method current-input-method)) |
1452 | (unwind-protect | |
1453 | (progn | |
1454 | (activate-input-method input-method) | |
1455 | (read-string prompt initial-input nil nil t)) | |
1456 | (activate-input-method prev-input-method)))) | |
4ed46869 KH |
1457 | |
1458 | ;; Variables to control behavior of input methods. All input methods | |
1459 | ;; should react to these variables. | |
1460 | ||
8efc03e1 KH |
1461 | (defcustom input-method-verbose-flag 'default |
1462 | "*A flag to control extra guidance given by input methods. | |
1463 | The value should be nil, t, `complex-only', or `default'. | |
4ed46869 | 1464 | |
cb29dfb6 | 1465 | The extra guidance is done by showing list of available keys in echo |
8efc03e1 KH |
1466 | area. When you use the input method in the minibuffer, the guidance |
1467 | is shown at the bottom short window (split from the existing window). | |
c27c4ed8 | 1468 | |
8efc03e1 KH |
1469 | If the value is t, extra guidance is always given, if the value is |
1470 | nil, extra guidance is always suppressed. | |
1471 | ||
1472 | If the value is `complex-only', only complex input methods such as | |
1473 | `chinese-py' and `japanese' give extra guidance. | |
1474 | ||
1475 | If the value is `default', complex input methods always give extra | |
1476 | guidance, but simple input methods give it only when you are not in | |
1477 | the minibuffer. | |
1478 | ||
1479 | See also the variable `input-method-highlight-flag'." | |
d37ef0f6 DL |
1480 | :type '(choice (const :tag "Always" t) (const :tag "Never" nil) |
1481 | (const complex-only) (const default)) | |
42395763 RS |
1482 | :group 'mule) |
1483 | ||
1484 | (defcustom input-method-highlight-flag t | |
1485 | "*If this flag is non-nil, input methods highlight partially-entered text. | |
1486 | For instance, while you are in the middle of a Quail input method sequence, | |
1487 | the text inserted so far is temporarily underlined. | |
8efc03e1 KH |
1488 | The underlining goes away when you finish or abort the input method sequence. |
1489 | See also the variable `input-method-verbose-flag'." | |
42395763 RS |
1490 | :type 'boolean |
1491 | :group 'mule) | |
4ed46869 KH |
1492 | |
1493 | (defvar input-method-activate-hook nil | |
f17ccaee KH |
1494 | "Normal hook run just after an input method is activated. |
1495 | ||
1496 | The variable `current-input-method' keeps the input method name | |
1497 | just activated.") | |
4ed46869 KH |
1498 | |
1499 | (defvar input-method-inactivate-hook nil | |
f17ccaee KH |
1500 | "Normal hook run just after an input method is inactivated. |
1501 | ||
1502 | The variable `current-input-method' still keeps the input method name | |
4d0e6a11 | 1503 | just inactivated.") |
4ed46869 KH |
1504 | |
1505 | (defvar input-method-after-insert-chunk-hook nil | |
1506 | "Normal hook run just after an input method insert some chunk of text.") | |
1507 | ||
dccca980 | 1508 | (defvar input-method-exit-on-first-char nil |
0b6cadff | 1509 | "This flag controls when an input method returns. |
dccca980 KH |
1510 | Usually, the input method does not return while there's a possibility |
1511 | that it may find a different translation if a user types another key. | |
39e643e2 RS |
1512 | But, it this flag is non-nil, the input method returns as soon as |
1513 | the current key sequence gets long enough to have some valid translation.") | |
dccca980 KH |
1514 | |
1515 | (defvar input-method-use-echo-area nil | |
1516 | "This flag controls how an input method shows an intermediate key sequence. | |
39e643e2 RS |
1517 | Usually, the input method inserts the intermediate key sequence, |
1518 | or candidate translations corresponding to the sequence, | |
1519 | at point in the current buffer. | |
1520 | But, if this flag is non-nil, it displays them in echo area instead.") | |
dccca980 | 1521 | |
723a427a KH |
1522 | (defvar input-method-exit-on-invalid-key nil |
1523 | "This flag controls the behaviour of an input method on invalid key input. | |
1524 | Usually, when a user types a key which doesn't start any character | |
1525 | handled by the input method, the key is handled by turning off the | |
e8dd0160 | 1526 | input method temporarily. After that key, the input method is re-enabled. |
723a427a KH |
1527 | But, if this flag is non-nil, the input method is never back on.") |
1528 | ||
4ed46869 | 1529 | \f |
8efc03e1 KH |
1530 | (defvar set-language-environment-hook nil |
1531 | "Normal hook run after some language environment is set. | |
1532 | ||
1533 | When you set some hook function here, that effect usually should not | |
1534 | be inherited to another language environment. So, you had better set | |
1535 | another function in `exit-language-environment-hook' (which see) to | |
1536 | cancel the effect.") | |
1537 | ||
1538 | (defvar exit-language-environment-hook nil | |
1539 | "Normal hook run after exiting from some language environment. | |
1540 | When this hook is run, the variable `current-language-environment' | |
1541 | is still bound to the language environment being exited. | |
1542 | ||
e8dd0160 | 1543 | This hook is mainly used for canceling the effect of |
8efc03e1 KH |
1544 | `set-language-environment-hook' (which-see).") |
1545 | ||
b0648a00 RS |
1546 | (put 'setup-specified-language-environment 'apropos-inhibit t) |
1547 | ||
15b3e511 | 1548 | (defun setup-specified-language-environment () |
f08adf27 | 1549 | "Switch to a specified language environment." |
15b3e511 | 1550 | (interactive) |
f850d782 | 1551 | (let (language-name) |
15b3e511 KH |
1552 | (if (and (symbolp last-command-event) |
1553 | (or (not (eq last-command-event 'Default)) | |
1554 | (setq last-command-event 'English)) | |
f850d782 | 1555 | (setq language-name (symbol-name last-command-event))) |
f4990970 PA |
1556 | (prog1 |
1557 | (set-language-environment language-name) | |
1558 | (customize-mark-as-set 'current-language-environment)) | |
15b3e511 | 1559 | (error "Bogus calling sequence")))) |
4ed46869 | 1560 | |
8861c593 | 1561 | (defcustom current-language-environment "English" |
94d04df6 | 1562 | "The last language environment specified with `set-language-environment'. |
ebef6d93 KH |
1563 | This variable should be set only with \\[customize], which is equivalent |
1564 | to using the function `set-language-environment'." | |
94d04df6 | 1565 | :link '(custom-manual "(emacs)Language Environments") |
dff1aa24 | 1566 | :set (lambda (symbol value) (set-language-environment value)) |
94d04df6 DL |
1567 | :get (lambda (x) |
1568 | (or (car-safe (assoc-ignore-case | |
1569 | (if (symbolp current-language-environment) | |
1570 | (symbol-name current-language-environment) | |
1571 | current-language-environment) | |
1572 | language-info-alist)) | |
1573 | "English")) | |
990a4108 MR |
1574 | ;; custom type will be updated with `set-language-info'. |
1575 | :type (if language-info-alist | |
1576 | (cons 'choice (mapcar | |
1577 | (lambda (lang) | |
756e055f MR |
1578 | (list 'const lang)) |
1579 | (sort (mapcar 'car language-info-alist) 'string<))) | |
990a4108 | 1580 | 'string) |
8861c593 | 1581 | :initialize 'custom-initialize-default |
eb9fc9e6 | 1582 | :group 'mule) |
f850d782 | 1583 | |
ddb5c041 KH |
1584 | (defun reset-language-environment () |
1585 | "Reset multilingual environment of Emacs to the default status. | |
1586 | ||
1587 | The default status is as follows: | |
1588 | ||
d37ef0f6 | 1589 | The default value of `buffer-file-coding-system' is nil. |
ddb5c041 KH |
1590 | The default coding system for process I/O is nil. |
1591 | The default value for the command `set-terminal-coding-system' is nil. | |
1592 | The default value for the command `set-keyboard-coding-system' is nil. | |
1593 | ||
1594 | The order of priorities of coding categories and the coding system | |
1595 | bound to each category are as follows | |
1596 | coding category coding system | |
1597 | -------------------------------------------------- | |
ddb5c041 | 1598 | coding-category-iso-8-1 iso-latin-1 |
370bd9cf KH |
1599 | coding-category-iso-8-2 iso-latin-1 |
1600 | coding-category-utf-8 mule-utf-8 | |
1601 | coding-category-utf-16-be mule-utf-16-be-with-signature | |
1602 | coding-category-utf-16-le mule-utf-16-le-with-signature | |
ddb5c041 KH |
1603 | coding-category-iso-7-tight iso-2022-jp |
1604 | coding-category-iso-7 iso-2022-7bit | |
1605 | coding-category-iso-7-else iso-2022-7bit-lock | |
1606 | coding-category-iso-8-else iso-2022-8bit-ss2 | |
1607 | coding-category-emacs-mule emacs-mule | |
1608 | coding-category-raw-text raw-text | |
1609 | coding-category-sjis japanese-shift-jis | |
1610 | coding-category-big5 chinese-big5 | |
1611 | coding-category-ccl nil | |
370bd9cf | 1612 | coding-category-binary no-conversion" |
ddb5c041 KH |
1613 | (interactive) |
1614 | ;; This function formerly set default-enable-multibyte-characters to t, | |
1615 | ;; but that is incorrect. It should not alter the unibyte/multibyte choice. | |
1616 | ||
1617 | (setq coding-category-iso-7-tight 'iso-2022-jp | |
1618 | coding-category-iso-7 'iso-2022-7bit | |
1619 | coding-category-iso-8-1 'iso-latin-1 | |
1620 | coding-category-iso-8-2 'iso-latin-1 | |
1621 | coding-category-iso-7-else 'iso-2022-7bit-lock | |
1622 | coding-category-iso-8-else 'iso-2022-8bit-ss2 | |
1623 | coding-category-emacs-mule 'emacs-mule | |
1624 | coding-category-raw-text 'raw-text | |
1625 | coding-category-sjis 'japanese-shift-jis | |
1626 | coding-category-big5 'chinese-big5 | |
370bd9cf KH |
1627 | coding-category-utf-16-be 'mule-utf-16-be-with-signature |
1628 | coding-category-utf-16-le 'mule-utf-16-le-with-signature | |
b909cfc0 KH |
1629 | coding-category-utf-8 'mule-utf-8 |
1630 | coding-category-ccl nil | |
1631 | coding-category-binary 'no-conversion) | |
ddb5c041 KH |
1632 | |
1633 | (set-coding-priority | |
1634 | '(coding-category-iso-8-1 | |
1635 | coding-category-iso-8-2 | |
370bd9cf KH |
1636 | coding-category-utf-8 |
1637 | coding-category-utf-16-be | |
1638 | coding-category-utf-16-le | |
ddb5c041 KH |
1639 | coding-category-iso-7-tight |
1640 | coding-category-iso-7 | |
1641 | coding-category-iso-7-else | |
1642 | coding-category-iso-8-else | |
d37ef0f6 | 1643 | coding-category-emacs-mule |
ddb5c041 KH |
1644 | coding-category-raw-text |
1645 | coding-category-sjis | |
1646 | coding-category-big5 | |
1647 | coding-category-ccl | |
370bd9cf | 1648 | coding-category-binary)) |
ddb5c041 | 1649 | |
91693d18 KH |
1650 | (update-coding-systems-internal) |
1651 | ||
ddb5c041 | 1652 | (set-default-coding-systems nil) |
b5edd1d1 | 1653 | (setq default-sendmail-coding-system 'iso-latin-1) |
787caf99 | 1654 | (setq default-file-name-coding-system 'iso-latin-1) |
1d77e15a JR |
1655 | ;; Preserve eol-type from existing default-process-coding-systems. |
1656 | ;; On non-unix-like systems in particular, these may have been set | |
1657 | ;; carefully by the user, or by the startup code, to deal with the | |
1658 | ;; users shell appropriately, so should not be altered by changing | |
1659 | ;; language environment. | |
1660 | (let ((output-coding | |
a099a2ff JR |
1661 | ;; When bootstrapping, coding-systems are not defined yet, so |
1662 | ;; we need to catch the error from check-coding-system. | |
d37ef0f6 | 1663 | (condition-case nil |
a099a2ff JR |
1664 | (coding-system-change-text-conversion |
1665 | (car default-process-coding-system) 'undecided) | |
1666 | (coding-system-error 'undecided))) | |
1d77e15a | 1667 | (input-coding |
a099a2ff JR |
1668 | (condition-case nil |
1669 | (coding-system-change-text-conversion | |
1670 | (cdr default-process-coding-system) 'iso-latin-1) | |
1671 | (coding-system-error 'iso-latin-1)))) | |
1d77e15a JR |
1672 | (setq default-process-coding-system |
1673 | (cons output-coding input-coding))) | |
b5edd1d1 | 1674 | |
ddb5c041 KH |
1675 | ;; Don't alter the terminal and keyboard coding systems here. |
1676 | ;; The terminal still supports the same coding system | |
1677 | ;; that it supported a minute ago. | |
1b8dc791 SM |
1678 | ;; (set-terminal-coding-system-internal nil) |
1679 | ;; (set-keyboard-coding-system-internal nil) | |
ddb5c041 KH |
1680 | |
1681 | (setq nonascii-translation-table nil | |
1682 | nonascii-insert-offset 0)) | |
1683 | ||
0c47a7c8 KH |
1684 | (reset-language-environment) |
1685 | ||
40c81f74 PE |
1686 | (defun set-display-table-and-terminal-coding-system (language-name) |
1687 | "Set up the display table and terminal coding system for LANGUAGE-NAME." | |
1688 | (let ((coding (get-language-info language-name 'unibyte-display))) | |
1689 | (if coding | |
1690 | (standard-display-european-internal) | |
1691 | (standard-display-default (if (eq window-system 'pc) 128 160) 255) | |
1692 | (aset standard-display-table 146 nil)) | |
1693 | (or (eq window-system 'pc) | |
1694 | (set-terminal-coding-system coding)))) | |
1695 | ||
166246f7 | 1696 | (defun set-language-environment (language-name) |
6c05d680 RS |
1697 | "Set up multi-lingual environment for using LANGUAGE-NAME. |
1698 | This sets the coding system priority and the default input method | |
8861c593 RS |
1699 | and sometimes other things. LANGUAGE-NAME should be a string |
1700 | which is the name of a language environment. For example, \"Latin-1\" | |
1701 | specifies the character set for the major languages of Western Europe." | |
8efc03e1 | 1702 | (interactive (list (read-language-name |
ddb5c041 | 1703 | nil |
8efc03e1 | 1704 | "Set language environment (default, English): "))) |
4ef06f75 KH |
1705 | (if language-name |
1706 | (if (symbolp language-name) | |
1707 | (setq language-name (symbol-name language-name))) | |
1708 | (setq language-name "English")) | |
ddb5c041 | 1709 | (or (assoc-ignore-case language-name language-info-alist) |
f850d782 | 1710 | (error "Language environment not defined: %S" language-name)) |
8efc03e1 KH |
1711 | (if current-language-environment |
1712 | (let ((func (get-language-info current-language-environment | |
1713 | 'exit-function))) | |
e63645c2 | 1714 | (run-hooks 'exit-language-environment-hook) |
5f395df3 | 1715 | (if (functionp func) (funcall func)))) |
03c35c83 EZ |
1716 | (let ((default-eol-type (coding-system-eol-type |
1717 | default-buffer-file-coding-system))) | |
1718 | (reset-language-environment) | |
ddb5c041 | 1719 | |
e522f07b | 1720 | ;; The features might set up coding systems. |
9c20a8d5 DL |
1721 | (let ((required-features (get-language-info language-name 'features))) |
1722 | (while required-features | |
1723 | (require (car required-features)) | |
1724 | (setq required-features (cdr required-features)))) | |
1725 | ||
03c35c83 EZ |
1726 | (setq current-language-environment language-name) |
1727 | (set-language-environment-coding-systems language-name default-eol-type)) | |
ddb5c041 KH |
1728 | (let ((input-method (get-language-info language-name 'input-method))) |
1729 | (when input-method | |
1730 | (setq default-input-method input-method) | |
1731 | (if input-method-history | |
1732 | (setq input-method-history | |
1733 | (cons input-method | |
1734 | (delete input-method input-method-history)))))) | |
ec241f58 EZ |
1735 | (let ((nonascii (get-language-info language-name 'nonascii-translation)) |
1736 | (dos-table | |
4e2ac2d9 EZ |
1737 | (if (eq window-system 'pc) |
1738 | (intern | |
d9c0a50e | 1739 | (format "cp%d-nonascii-translation-table" dos-codepage))))) |
03c35c83 EZ |
1740 | (cond |
1741 | ((char-table-p nonascii) | |
1742 | (setq nonascii-translation-table nonascii)) | |
ec241f58 | 1743 | ((and (eq window-system 'pc) (boundp dos-table)) |
03c35c83 EZ |
1744 | ;; DOS terminals' default is to use a special non-ASCII translation |
1745 | ;; table as appropriate for the installed codepage. | |
ec241f58 | 1746 | (setq nonascii-translation-table (symbol-value dos-table))) |
03c35c83 EZ |
1747 | ((charsetp nonascii) |
1748 | (setq nonascii-insert-offset (- (make-char nonascii) 128))))) | |
ddb5c041 | 1749 | |
63283a8f | 1750 | ;; Unibyte setups if necessary. |
ddb5c041 | 1751 | (unless default-enable-multibyte-characters |
63283a8f | 1752 | ;; Syntax and case table. |
ddb5c041 KH |
1753 | (let ((syntax (get-language-info language-name 'unibyte-syntax))) |
1754 | (if syntax | |
1755 | (let ((set-case-syntax-set-multibyte nil)) | |
63283a8f KH |
1756 | (load syntax nil t)) |
1757 | ;; No information for syntax and case. Reset to the defaults. | |
1758 | (let ((syntax-table (standard-syntax-table)) | |
1759 | (case-table (standard-case-table)) | |
03c35c83 | 1760 | (ch (if (eq window-system 'pc) 128 160))) |
63283a8f KH |
1761 | (while (< ch 256) |
1762 | (modify-syntax-entry ch " " syntax-table) | |
1763 | (aset case-table ch ch) | |
1764 | (setq ch (1+ ch))) | |
1765 | (set-char-table-extra-slot case-table 0 nil) | |
1766 | (set-char-table-extra-slot case-table 1 nil) | |
1767 | (set-char-table-extra-slot case-table 2 nil)) | |
1768 | (set-standard-case-table (standard-case-table)) | |
1769 | (let ((list (buffer-list))) | |
1770 | (while list | |
1771 | (with-current-buffer (car list) | |
1772 | (set-case-table (standard-case-table))) | |
1773 | (setq list (cdr list)))))) | |
40c81f74 | 1774 | (set-display-table-and-terminal-coding-system language-name)) |
ddb5c041 KH |
1775 | |
1776 | (let ((required-features (get-language-info language-name 'features))) | |
1777 | (while required-features | |
1778 | (require (car required-features)) | |
1779 | (setq required-features (cdr required-features)))) | |
1780 | (let ((func (get-language-info language-name 'setup-function))) | |
5f395df3 | 1781 | (if (functionp func) |
ddb5c041 | 1782 | (funcall func))) |
8efc03e1 | 1783 | (run-hooks 'set-language-environment-hook) |
f850d782 | 1784 | (force-mode-line-update t)) |
4ed46869 | 1785 | |
51a8fc1d RS |
1786 | (defun standard-display-european-internal () |
1787 | ;; Actually set up direct output of non-ASCII characters. | |
03c35c83 EZ |
1788 | (standard-display-8bit (if (eq window-system 'pc) 128 160) 255) |
1789 | ;; Unibyte Emacs on MS-DOS wants to display all 8-bit characters with | |
1790 | ;; the native font, and codes 160 and 146 stand for something very | |
1791 | ;; different there. | |
1792 | (or (and (eq window-system 'pc) (not default-enable-multibyte-characters)) | |
1793 | (progn | |
1794 | ;; Make non-line-break space display as a plain space. | |
1795 | ;; Most X fonts do the wrong thing for code 160. | |
1796 | (aset standard-display-table 160 [32]) | |
1492f7ac DL |
1797 | ;; With luck, non-Latin-1 fonts are more recent and so don't |
1798 | ;; have this bug. | |
1799 | (aset standard-display-table 2208 [32]) ; Latin-1 NBSP | |
1800 | ;; Most Windows programs send out apostrophes as \222. Most X fonts | |
03c35c83 | 1801 | ;; don't contain a character at that position. Map it to the ASCII |
5f395df3 SM |
1802 | ;; apostrophe. [This is actually RIGHT SINGLE QUOTATION MARK, |
1803 | ;; U+2019, normally from the windows-1252 character set. XFree 4 | |
1804 | ;; fonts probably have the appropriate glyph at this position, | |
1805 | ;; so they could use standard-display-8bit. It's better to use a | |
1806 | ;; proper windows-1252 coding system. --fx] | |
03c35c83 EZ |
1807 | (aset standard-display-table 146 [39])))) |
1808 | ||
1809 | (defun set-language-environment-coding-systems (language-name | |
1810 | &optional eol-type) | |
1811 | "Do various coding system setups for language environment LANGUAGE-NAME. | |
1812 | ||
1813 | The optional arg EOL-TYPE specifies the eol-type of the default value | |
d37ef0f6 | 1814 | of `buffer-file-coding-system' set by this function." |
54b226f7 KH |
1815 | (let* ((priority (get-language-info language-name 'coding-priority)) |
1816 | (default-coding (car priority))) | |
1817 | (if priority | |
1818 | (let ((categories (mapcar 'coding-system-category priority))) | |
03c35c83 EZ |
1819 | (set-default-coding-systems |
1820 | (if (memq eol-type '(0 1 2 unix dos mac)) | |
1821 | (coding-system-change-eol-conversion default-coding eol-type) | |
1822 | default-coding)) | |
46babb23 | 1823 | (setq default-sendmail-coding-system default-coding) |
54b226f7 KH |
1824 | (set-coding-priority categories) |
1825 | (while priority | |
1826 | (set (car categories) (car priority)) | |
1827 | (setq priority (cdr priority) categories (cdr categories))) | |
45d08cb2 | 1828 | (update-coding-systems-internal))))) |
54b226f7 | 1829 | |
4ed46869 | 1830 | (defsubst princ-list (&rest args) |
d37ef0f6 | 1831 | "Print all arguments with `princ', then print \"\n\"." |
4ed46869 KH |
1832 | (while args (princ (car args)) (setq args (cdr args))) |
1833 | (princ "\n")) | |
1834 | ||
b0648a00 RS |
1835 | (put 'describe-specified-language-support 'apropos-inhibit t) |
1836 | ||
48082651 | 1837 | ;; Print a language specific information such as input methods, |
13e82c04 | 1838 | ;; charsets, and coding systems. This function is intended to be |
48082651 | 1839 | ;; called from the menu: |
281d03ec | 1840 | ;; [menu-bar mule describe-language-environment LANGUAGE] |
48082651 KH |
1841 | ;; and should not run it by `M-x describe-current-input-method-function'. |
1842 | (defun describe-specified-language-support () | |
96db204a | 1843 | "Describe how Emacs supports the specified language environment." |
48082651 | 1844 | (interactive) |
281d03ec | 1845 | (let (language-name) |
48082651 | 1846 | (if (not (and (symbolp last-command-event) |
cda74479 DL |
1847 | (or (not (eq last-command-event 'Default)) |
1848 | (setq last-command-event 'English)) | |
281d03ec | 1849 | (setq language-name (symbol-name last-command-event)))) |
48082651 | 1850 | (error "Bogus calling sequence")) |
281d03ec RS |
1851 | (describe-language-environment language-name))) |
1852 | ||
1853 | (defun describe-language-environment (language-name) | |
1854 | "Describe how Emacs supports language environment LANGUAGE-NAME." | |
78754934 KH |
1855 | (interactive |
1856 | (list (read-language-name | |
1857 | 'documentation | |
8adfa8be | 1858 | "Describe language environment (default, current choice): "))) |
f850d782 RS |
1859 | (if (null language-name) |
1860 | (setq language-name current-language-environment)) | |
281d03ec RS |
1861 | (if (or (null language-name) |
1862 | (null (get-language-info language-name 'documentation))) | |
1863 | (error "No documentation for the specified language")) | |
4ef06f75 KH |
1864 | (if (symbolp language-name) |
1865 | (setq language-name (symbol-name language-name))) | |
ef5a4730 KH |
1866 | (dolist (feature (get-language-info language-name 'features)) |
1867 | (require feature)) | |
464cc130 KH |
1868 | (let ((doc (get-language-info language-name 'documentation)) |
1869 | pos) | |
c3034e84 SM |
1870 | (help-setup-xref (list #'describe-language-environment language-name) |
1871 | (interactive-p)) | |
1872 | (with-output-to-temp-buffer (help-buffer) | |
464cc130 KH |
1873 | (save-excursion |
1874 | (set-buffer standard-output) | |
1875 | (insert language-name " language environment\n\n") | |
1876 | (if (stringp doc) | |
1877 | (insert doc "\n\n")) | |
e036b0a6 KH |
1878 | (condition-case nil |
1879 | (let ((str (eval (get-language-info language-name 'sample-text)))) | |
1880 | (if (stringp str) | |
1881 | (insert "Sample text:\n " str "\n\n"))) | |
1882 | (error nil)) | |
464cc130 KH |
1883 | (let ((input-method (get-language-info language-name 'input-method)) |
1884 | (l (copy-sequence input-method-alist))) | |
1885 | (insert "Input methods") | |
1886 | (when input-method | |
1887 | (insert " (default, " input-method ")") | |
1888 | (setq input-method (assoc input-method input-method-alist)) | |
1889 | (setq l (cons input-method (delete input-method l)))) | |
1890 | (insert ":\n") | |
48082651 | 1891 | (while l |
464cc130 KH |
1892 | (when (string= language-name (nth 1 (car l))) |
1893 | (insert " " (car (car l))) | |
1894 | (search-backward (car (car l))) | |
467412aa | 1895 | (help-xref-button 0 'help-input-method (car (car l))) |
464cc130 | 1896 | (goto-char (point-max)) |
2fa7e202 KH |
1897 | (insert " (\"" |
1898 | (if (stringp (nth 3 (car l))) | |
1899 | (nth 3 (car l)) | |
1900 | (car (nth 3 (car l)))) | |
1901 | "\" in mode line)\n")) | |
464cc130 KH |
1902 | (setq l (cdr l))) |
1903 | (insert "\n")) | |
1904 | (insert "Character sets:\n") | |
1905 | (let ((l (get-language-info language-name 'charset))) | |
1906 | (if (null l) | |
1907 | (insert " nothing specific to " language-name "\n") | |
1908 | (while l | |
1909 | (insert " " (symbol-name (car l))) | |
1910 | (search-backward (symbol-name (car l))) | |
467412aa | 1911 | (help-xref-button 0 'help-character-set (car l)) |
464cc130 KH |
1912 | (goto-char (point-max)) |
1913 | (insert ": " (charset-description (car l)) "\n") | |
1914 | (setq l (cdr l))))) | |
1915 | (insert "\n") | |
1916 | (insert "Coding systems:\n") | |
1917 | (let ((l (get-language-info language-name 'coding-system))) | |
1918 | (if (null l) | |
1919 | (insert " nothing specific to " language-name "\n") | |
1920 | (while l | |
1921 | (insert " " (symbol-name (car l))) | |
1922 | (search-backward (symbol-name (car l))) | |
467412aa | 1923 | (help-xref-button 0 'help-coding-system (car l)) |
464cc130 KH |
1924 | (goto-char (point-max)) |
1925 | (insert " (`" | |
1926 | (coding-system-mnemonic (car l)) | |
1927 | "' in mode line):\n\t" | |
1928 | (coding-system-doc-string (car l)) | |
1929 | "\n") | |
1930 | (let ((aliases (coding-system-get (car l) | |
1931 | 'alias-coding-systems))) | |
1932 | (when aliases | |
1933 | (insert "\t(alias:") | |
1934 | (while aliases | |
1935 | (insert " " (symbol-name (car aliases))) | |
1936 | (setq aliases (cdr aliases))) | |
1937 | (insert ")\n"))) | |
c3034e84 | 1938 | (setq l (cdr l))))))))) |
4ed46869 | 1939 | \f |
40c81f74 PE |
1940 | ;;; Locales. |
1941 | ||
0d7c5bb9 DL |
1942 | (defvar locale-translation-file-name nil |
1943 | "File name for the system's file of locale-name aliases, or nil if none.") | |
40c81f74 | 1944 | |
5f395df3 SM |
1945 | ;; The following definitions might as well be marked as constants and |
1946 | ;; purecopied, since they're normally used on startup, and probably | |
1947 | ;; should reflect the facilities of the base Emacs. | |
1948 | (defconst locale-language-names | |
1949 | (purecopy | |
1950 | '( | |
40c81f74 PE |
1951 | ;; Locale names of the form LANGUAGE[_TERRITORY][.CODESET][@MODIFIER] |
1952 | ;; as specified in the Single Unix Spec, Version 2. | |
1953 | ;; LANGUAGE is a language code taken from ISO 639:1988 (E/F) | |
1954 | ;; with additions from ISO 639/RA Newsletter No.1/1989; | |
5f395df3 SM |
1955 | ;; see Internet RFC 2165 (1997-06) and |
1956 | ;; http://www.evertype.com/standards/iso639/iso639-en.html | |
1957 | ;; TERRITORY is a country code taken from ISO 3166 | |
1958 | ;; http://www.din.de/gremien/nas/nabd/iso3166ma/codlstp1/en_listp1.html. | |
40c81f74 | 1959 | ;; CODESET and MODIFIER are implementation-dependent. |
5f395df3 | 1960 | |
40c81f74 PE |
1961 | ; aa Afar |
1962 | ; ab Abkhazian | |
6ececc4d | 1963 | ("af" . "Latin-1") ; Afrikaans |
40c81f74 | 1964 | ("am" . "Ethiopic") ; Amharic |
5f395df3 | 1965 | ; ar Arabic glibc uses 8859-6 |
40c81f74 PE |
1966 | ; as Assamese |
1967 | ; ay Aymara | |
1968 | ; az Azerbaijani | |
1969 | ; ba Bashkir | |
e522f07b | 1970 | ("be" . "Belarusian") ; Belarusian [Byelorussian until early 1990s] |
9c20a8d5 | 1971 | ("bg" . "Bulgarian") ; Bulgarian |
40c81f74 PE |
1972 | ; bh Bihari |
1973 | ; bi Bislama | |
1974 | ; bn Bengali, Bangla | |
1975 | ("bo" . "Tibetan") | |
1976 | ("br" . "Latin-1") ; Breton | |
d37ef0f6 | 1977 | ("bs" . "Latin-2") ; Bosnian |
40c81f74 PE |
1978 | ("ca" . "Latin-1") ; Catalan |
1979 | ; co Corsican | |
1980 | ("cs" . "Czech") | |
e76ef161 | 1981 | ("cy" . "Welsh") ; Welsh [glibc uses Latin-8. Did this change?] |
40c81f74 PE |
1982 | ("da" . "Latin-1") ; Danish |
1983 | ("de" . "German") | |
1984 | ; dz Bhutani | |
1985 | ("el" . "Greek") | |
6ececc4d | 1986 | ;; Users who specify "en" explicitly typically want Latin-1, not ASCII. |
e11cf111 DL |
1987 | ;; That's actually what the GNU locales define, modulo things like |
1988 | ;; en_IN -- fx. | |
6ececc4d | 1989 | ("en" . "Latin-1") ; English |
40c81f74 | 1990 | ("eo" . "Latin-3") ; Esperanto |
8c4b6822 | 1991 | ("es" . "Spanish") |
40c81f74 PE |
1992 | ("et" . "Latin-4") ; Estonian |
1993 | ("eu" . "Latin-1") ; Basque | |
d37ef0f6 | 1994 | ; fa Persian glibc uses utf-8 |
40c81f74 PE |
1995 | ("fi" . "Latin-1") ; Finnish |
1996 | ; fj Fiji | |
1997 | ("fo" . "Latin-1") ; Faroese | |
8d738cfd | 1998 | ("fr" . "French") ; French |
40c81f74 | 1999 | ("fy" . "Latin-1") ; Frisian |
6ececc4d PE |
2000 | ("ga" . "Latin-1") ; Irish Gaelic (new orthography) |
2001 | ("gd" . "Latin-1") ; Scots Gaelic | |
2002 | ("gl" . "Latin-1") ; Galician | |
40c81f74 PE |
2003 | ; gn Guarani |
2004 | ; gu Gujarati | |
d37ef0f6 | 2005 | ("gv" . "Latin-8") ; Manx Gaelic glibc uses 8859-1 |
40c81f74 PE |
2006 | ; ha Hausa |
2007 | ("he" . "Hebrew") | |
5f395df3 | 2008 | ("hi" . "Devanagari") ; Hindi glibc uses utf-8 |
03cb0324 | 2009 | ("hr" . "Croatian") ; Croatian |
40c81f74 PE |
2010 | ("hu" . "Latin-2") ; Hungarian |
2011 | ; hy Armenian | |
2012 | ; ia Interlingua | |
2013 | ("id" . "Latin-1") ; Indonesian | |
2014 | ; ie Interlingue | |
2015 | ; ik Inupiak | |
2016 | ("is" . "Latin-1") ; Icelandic | |
e76ef161 | 2017 | ("it" . "Italian") ; Italian |
40c81f74 PE |
2018 | ; iu Inuktitut |
2019 | ("ja" . "Japanese") | |
2020 | ; jw Javanese | |
9c20a8d5 | 2021 | ("ka" . "Georgian") ; Georgian |
40c81f74 | 2022 | ; kk Kazakh |
6ececc4d | 2023 | ("kl" . "Latin-1") ; Greenlandic |
40c81f74 PE |
2024 | ; km Cambodian |
2025 | ; kn Kannada | |
2026 | ("ko" . "Korean") | |
2027 | ; ks Kashmiri | |
2028 | ; ku Kurdish | |
5f395df3 | 2029 | ("kw" . "Latin-1") ; Cornish |
40c81f74 PE |
2030 | ; ky Kirghiz |
2031 | ("la" . "Latin-1") ; Latin | |
5f395df3 | 2032 | ("lb" . "Latin-1") ; Luxemburgish |
40c81f74 PE |
2033 | ; ln Lingala |
2034 | ("lo" . "Lao") ; Laothian | |
9c20a8d5 DL |
2035 | ("lt" . "Lithuanian") |
2036 | ("lv" . "Latvian") ; Latvian, Lettish | |
40c81f74 | 2037 | ; mg Malagasy |
5f395df3 | 2038 | ("mi" . "Latin-7") ; Maori |
f1282c7f | 2039 | ("mk" . "Cyrillic-ISO") ; Macedonian |
40c81f74 PE |
2040 | ; ml Malayalam |
2041 | ; mn Mongolian | |
2042 | ; mo Moldavian | |
5f395df3 SM |
2043 | ("mr" . "Devanagari") ; Marathi glibc uses utf-8 |
2044 | ("ms" . "Latin-1") ; Malay | |
40c81f74 PE |
2045 | ("mt" . "Latin-3") ; Maltese |
2046 | ; my Burmese | |
2047 | ; na Nauru | |
2048 | ("ne" . "Devanagari") ; Nepali | |
8c4b6822 | 2049 | ("nl" . "Dutch") |
40c81f74 | 2050 | ("no" . "Latin-1") ; Norwegian |
5f395df3 | 2051 | ("oc" . "Latin-1") ; Occitan |
40c81f74 PE |
2052 | ; om (Afan) Oromo |
2053 | ; or Oriya | |
2054 | ; pa Punjabi | |
2055 | ("pl" . "Latin-2") ; Polish | |
2056 | ; ps Pashto, Pushto | |
2057 | ("pt" . "Latin-1") ; Portuguese | |
2058 | ; qu Quechua | |
6ececc4d | 2059 | ("rm" . "Latin-1") ; Rhaeto-Romanic |
40c81f74 PE |
2060 | ; rn Kirundi |
2061 | ("ro" . "Romanian") | |
e76ef161 | 2062 | ("ru.*[_.]koi8" . "Russian") |
f1282c7f | 2063 | ("ru" . "Cyrillic-ISO") ; Russian |
40c81f74 PE |
2064 | ; rw Kinyarwanda |
2065 | ("sa" . "Devanagari") ; Sanskrit | |
2066 | ; sd Sindhi | |
5f395df3 | 2067 | ; se Northern Sami |
40c81f74 PE |
2068 | ; sg Sangho |
2069 | ("sh" . "Latin-2") ; Serbo-Croatian | |
2070 | ; si Sinhalese | |
2071 | ("sk" . "Slovak") | |
2072 | ("sl" . "Slovenian") | |
2073 | ; sm Samoan | |
2074 | ; sn Shona | |
2075 | ; so Somali | |
6ececc4d | 2076 | ("sq" . "Latin-1") ; Albanian |
40c81f74 | 2077 | ("sr" . "Latin-2") ; Serbian (Latin alphabet) |
e522f07b | 2078 | ("sr_YU@cyrillic" . "Cyrillic-ISO") ; per glibc |
40c81f74 PE |
2079 | ; ss Siswati |
2080 | ; st Sesotho | |
2081 | ; su Sundanese | |
03cb0324 | 2082 | ("sv" . "Swedish") ; Swedish |
40c81f74 | 2083 | ("sw" . "Latin-1") ; Swahili |
5f395df3 SM |
2084 | ; ta Tamil glibc uses utf-8 |
2085 | ; te Telugu glibc uses utf-8 | |
e522f07b | 2086 | ("tg" . "Tajik") |
40c81f74 PE |
2087 | ("th" . "Thai") |
2088 | ; ti Tigrinya | |
2089 | ; tk Turkmen | |
6ececc4d | 2090 | ("tl" . "Latin-1") ; Tagalog |
40c81f74 PE |
2091 | ; tn Setswana |
2092 | ; to Tonga | |
e76ef161 | 2093 | ("tr" . "Turkish") |
40c81f74 PE |
2094 | ; ts Tsonga |
2095 | ; tt Tatar | |
2096 | ; tw Twi | |
2097 | ; ug Uighur | |
9c20a8d5 | 2098 | ("uk" . "Ukrainian") ; Ukrainian |
5f395df3 SM |
2099 | ; ur Urdu glibc uses utf-8 |
2100 | ("uz" . "Latin-1") ; Uzbek | |
2101 | ("vi" . "Vietnamese") ; glibc uses utf-8 | |
40c81f74 | 2102 | ; vo Volapuk |
d37ef0f6 | 2103 | ("wa" . "Latin-1") ; Walloon |
40c81f74 PE |
2104 | ; wo Wolof |
2105 | ; xh Xhosa | |
9c20a8d5 | 2106 | ("yi" . "Windows-1255") ; Yiddish |
40c81f74 PE |
2107 | ; yo Yoruba |
2108 | ; za Zhuang | |
5f395df3 SM |
2109 | |
2110 | ; glibc: | |
2111 | ; zh_CN.GB18030/GB18030 \ | |
2112 | ; zh_CN.GBK/GBK \ | |
2113 | ; zh_HK/BIG5-HKSCS \ | |
5f395df3 | 2114 | |
40c81f74 PE |
2115 | ("zh.*[._]big5" . "Chinese-BIG5") |
2116 | ("zh.*[._]gbk" . nil) ; Solaris 2.7; has gbk-0 as well as GB 2312.1980-0 | |
e76ef161 DL |
2117 | ("zh_tw" . "Chinese-CNS") ; glibc uses big5 |
2118 | ("zh_tw[._]euc-tw" . "Chinese-EUC-TW") | |
40c81f74 PE |
2119 | ("zh" . "Chinese-GB") |
2120 | ; zu Zulu | |
2121 | ||
2122 | ;; ISO standard locales | |
2123 | ("c$" . "ASCII") | |
2124 | ("posix$" . "ASCII") | |
2125 | ||
40c81f74 PE |
2126 | ;; The "IPA" Emacs language environment does not correspond |
2127 | ;; to any ISO 639 code, so let it stand for itself. | |
2128 | ("ipa$" . "IPA") | |
2129 | ||
2130 | ;; Nonstandard or obsolete language codes | |
2131 | ("cz" . "Czech") ; e.g. Solaris 2.6 | |
2132 | ("ee" . "Latin-4") ; Estonian, e.g. X11R6.4 | |
2133 | ("iw" . "Hebrew") ; e.g. X11R6.4 | |
f1282c7f | 2134 | ("sp" . "Cyrillic-ISO") ; Serbian (Cyrillic alphabet), e.g. X11R6.4 |
40c81f74 | 2135 | ("su" . "Latin-1") ; Finnish, e.g. Solaris 2.6 |
2e86ceaa | 2136 | ("jp" . "Japanese") ; e.g. MS Windows |
86d9e628 JR |
2137 | ("chs" . "Chinese-GB") ; MS Windows Chinese Simplified |
2138 | ("cht" . "Chinese-BIG5") ; MS Windows Chinese Traditional | |
367ca50f | 2139 | )) |
40c81f74 | 2140 | "List of pairs of locale regexps and language names. |
6ececc4d PE |
2141 | The first element whose locale regexp matches the start of a downcased locale |
2142 | specifies the language name corresponding to that locale. | |
40c81f74 PE |
2143 | If the language name is nil, there is no corresponding language environment.") |
2144 | ||
5f395df3 SM |
2145 | (defconst locale-charset-language-names |
2146 | (purecopy | |
2147 | '((".*8859[-_]?1\\>" . "Latin-1") | |
2148 | (".*8859[-_]?2\\>" . "Latin-2") | |
2149 | (".*8859[-_]?3\\>" . "Latin-3") | |
2150 | (".*8859[-_]?4\\>" . "Latin-4") | |
2151 | (".*8859[-_]?9\\>" . "Latin-5") | |
2152 | (".*8859[-_]?14\\>" . "Latin-8") | |
2153 | (".*8859[-_]?15\\>" . "Latin-9") | |
e522f07b DL |
2154 | (".*utf\\(-?8\\)\\>" . "UTF-8") |
2155 | ;; utf-8@euro exists, so put this last. (@euro really specifies | |
2156 | ;; the currency, rather than the charset.) | |
2157 | (".*@euro\\>" . "Latin-9"))) | |
6ececc4d PE |
2158 | "List of pairs of locale regexps and charset language names. |
2159 | The first element whose locale regexp matches the start of a downcased locale | |
2160 | specifies the language name whose charsets corresponds to that locale. | |
2161 | This language name is used if its charsets disagree with the charsets of | |
2162 | the language name that would otherwise be used for this locale.") | |
2163 | ||
5f395df3 SM |
2164 | (defconst locale-preferred-coding-systems |
2165 | (purecopy | |
2166 | '(("ja.*[._]euc" . japanese-iso-8bit) | |
2167 | ("ja.*[._]jis7" . iso-2022-jp) | |
2168 | ("ja.*[._]pck" . japanese-shift-jis) | |
2169 | ("ja.*[._]sjis" . japanese-shift-jis) | |
69210880 | 2170 | ("jpn" . japanese-shift-jis) ; MS-Windows uses this. |
9c20a8d5 | 2171 | (".*[._]utf" . utf-8))) |
6ececc4d PE |
2172 | "List of pairs of locale regexps and preferred coding systems. |
2173 | The first element whose locale regexp matches the start of a downcased locale | |
2174 | specifies the coding system to prefer when using that locale.") | |
40c81f74 PE |
2175 | |
2176 | (defun locale-name-match (key alist) | |
2177 | "Search for KEY in ALIST, which should be a list of regexp-value pairs. | |
2178 | Return the value corresponding to the first regexp that matches the | |
2179 | start of KEY, or nil if there is no match." | |
2180 | (let (element) | |
2181 | (while (and alist (not element)) | |
ddf6b836 | 2182 | (if (string-match (concat "\\`\\(?:" (car (car alist)) "\\)") key) |
40c81f74 PE |
2183 | (setq element (car alist))) |
2184 | (setq alist (cdr alist))) | |
2185 | (cdr element))) | |
2186 | ||
758f07de | 2187 | (defun set-locale-environment (&optional locale-name) |
40c81f74 | 2188 | "Set up multi-lingual environment for using LOCALE-NAME. |
758f07de RS |
2189 | This sets the language environment, the coding system priority, |
2190 | the default input method and sometimes other things. | |
2191 | ||
2192 | LOCALE-NAME should be a string | |
40c81f74 PE |
2193 | which is the name of a locale supported by the system; |
2194 | often it is of the form xx_XX.CODE, where xx is a language, | |
2195 | XX is a country, and CODE specifies a character set and coding system. | |
2196 | For example, the locale name \"ja_JP.EUC\" might name a locale | |
2197 | for Japanese in Japan using the `japanese-iso-8bit' coding-system. | |
2198 | ||
758f07de | 2199 | If LOCALE-NAME is nil, its value is taken from the environment |
d37ef0f6 | 2200 | variables LC_ALL, LC_CTYPE and LANG (the first one that is set). |
40c81f74 PE |
2201 | |
2202 | The locale names supported by your system can typically be found in a | |
0812c1e8 | 2203 | directory named `/usr/share/locale' or `/usr/lib/locale'. LOCALE-NAME |
758f07de | 2204 | will be translated according to the table specified by |
0812c1e8 DL |
2205 | `locale-translation-file-name'. |
2206 | ||
2207 | See also `locale-charset-language-names', `locale-language-names', | |
2208 | `locale-preferred-coding-systems' and `locale-coding-system'." | |
758f07de | 2209 | (interactive "sSet environment for locale: ") |
a1506d29 | 2210 | |
0d7c5bb9 DL |
2211 | ;; Do this at runtime for the sake of binaries possibly transported |
2212 | ;; to a system without X. | |
2213 | (setq locale-translation-file-name | |
2214 | (let ((files | |
2215 | '("/usr/lib/X11/locale/locale.alias" ; e.g. X11R6.4 | |
2216 | "/usr/X11R6/lib/X11/locale/locale.alias" ; e.g. RedHat 4.2 | |
2217 | "/usr/openwin/lib/locale/locale.alias" ; e.g. Solaris 2.6 | |
2218 | ;; | |
2219 | ;; The following name appears after the X-related names above, | |
2220 | ;; since the X-related names are what X actually uses. | |
2221 | "/usr/share/locale/locale.alias" ; GNU/Linux sans X | |
2222 | ))) | |
2223 | (while (and files (not (file-exists-p (car files)))) | |
2224 | (setq files (cdr files))) | |
2225 | (car files))) | |
2226 | ||
7008ccac GM |
2227 | (let ((locale locale-name)) |
2228 | ||
2229 | (unless locale | |
2230 | ;; Use the first of these three environment variables | |
2231 | ;; that has a nonempty value. | |
2232 | (let ((vars '("LC_ALL" "LC_CTYPE" "LANG"))) | |
db6bd804 DL |
2233 | (while (and vars |
2234 | (= 0 (length locale))) ; nil or empty string | |
2235 | (setq locale (getenv (pop vars)))))) | |
7008ccac GM |
2236 | |
2237 | (when locale | |
2238 | ||
2239 | ;; Translate "swedish" into "sv_SE.ISO8859-1", and so on, | |
2240 | ;; using the translation file that many systems have. | |
2241 | (when locale-translation-file-name | |
2242 | (with-temp-buffer | |
2243 | (insert-file-contents locale-translation-file-name) | |
2244 | (when (re-search-forward | |
2245 | (concat "^" (regexp-quote locale) ":?[ \t]+") nil t) | |
2246 | (setq locale (buffer-substring (point) (line-end-position)))))) | |
2247 | ||
2248 | ;; Leave the system locales alone if the caller did not specify | |
2249 | ;; an explicit locale name, as their defaults are set from | |
2250 | ;; LC_MESSAGES and LC_TIME, not LC_CTYPE, and the user might not | |
2251 | ;; want to set them to the same value as LC_CTYPE. | |
2252 | (when locale-name | |
2253 | (setq system-messages-locale locale) | |
2254 | (setq system-time-locale locale)) | |
2255 | ||
2256 | (setq locale (downcase locale)) | |
2257 | ||
2258 | (let ((language-name | |
2259 | (locale-name-match locale locale-language-names)) | |
2260 | (charset-language-name | |
2261 | (locale-name-match locale locale-charset-language-names)) | |
2262 | (coding-system | |
2263 | (locale-name-match locale locale-preferred-coding-systems))) | |
2264 | ||
5f395df3 | 2265 | ;; Give preference to charset-language-name over language-name. |
7008ccac GM |
2266 | (if (and charset-language-name |
2267 | (not | |
2268 | (equal (get-language-info language-name 'charset) | |
2269 | (get-language-info charset-language-name 'charset)))) | |
2270 | (setq language-name charset-language-name)) | |
2271 | ||
2272 | (when language-name | |
2273 | ||
2274 | ;; Set up for this character set. This is now the right way | |
2275 | ;; to do it for both unibyte and multibyte modes. | |
2276 | (set-language-environment language-name) | |
2277 | ||
2278 | ;; If default-enable-multibyte-characters is nil, | |
2279 | ;; we are using single-byte characters, | |
2280 | ;; so the display table and terminal coding system are irrelevant. | |
2281 | (when default-enable-multibyte-characters | |
2282 | (set-display-table-and-terminal-coding-system language-name)) | |
2283 | ||
166ce29f DL |
2284 | ;; Set the `keyboard-coding-system' if appropriate (tty |
2285 | ;; only). At least X and MS Windows can generate | |
2286 | ;; multilingual input. | |
2287 | (unless window-system | |
2288 | (let ((kcs (or coding-system | |
2289 | (car (get-language-info language-name | |
2290 | 'coding-system))))) | |
2291 | (if kcs (set-keyboard-coding-system kcs)))) | |
ddf6b836 | 2292 | |
7008ccac GM |
2293 | (setq locale-coding-system |
2294 | (car (get-language-info language-name 'coding-priority)))) | |
2295 | ||
2296 | (when coding-system | |
2297 | (prefer-coding-system coding-system) | |
e76ef161 DL |
2298 | (setq locale-coding-system coding-system)))) |
2299 | ||
2300 | ;; Default to A4 paper if we're not in a C, POSIX or US locale. | |
2301 | ;; (See comments in Flanginfo.) | |
2302 | (let ((locale locale) | |
2303 | (paper (langinfo 'paper))) | |
2304 | (if paper | |
2305 | ;; This will always be null at the time of writing. | |
2306 | (cond | |
2307 | ((equal paper '(216 279)) | |
2308 | (setq ps-paper-type 'letter)) | |
2309 | ((equal paper '(210 297)) | |
2310 | (setq ps-paper-type 'a4))) | |
2311 | (let ((vars '("LC_ALL" "LC_PAPER" "LANG"))) | |
2312 | (while (and vars (= 0 (length locale))) | |
2313 | (setq locale (getenv (pop vars))))) | |
2314 | (when locale | |
2315 | ;; As of glibc 2.2.5, these are the only US Letter locales, | |
2316 | ;; and the rest are A4. | |
2317 | (setq ps-paper-type | |
2318 | (or (locale-name-match locale '(("c$" . letter) | |
2319 | ("posix$" . letter) | |
2320 | (".._us" . letter) | |
2321 | (".._pr" . letter) | |
2322 | (".._ca" . letter))) | |
2323 | 'a4)))))) | |
2324 | nil) | |
40c81f74 | 2325 | \f |
4ed46869 KH |
2326 | ;;; Charset property |
2327 | ||
251d4f4b | 2328 | (defun get-charset-property (charset propname) |
4ed46869 KH |
2329 | "Return the value of CHARSET's PROPNAME property. |
2330 | This is the last value stored with | |
96db204a | 2331 | (put-charset-property CHARSET PROPNAME VALUE)." |
251d4f4b KH |
2332 | (and (not (eq charset 'composition)) |
2333 | (plist-get (charset-plist charset) propname))) | |
4ed46869 | 2334 | |
251d4f4b | 2335 | (defun put-charset-property (charset propname value) |
4ed46869 KH |
2336 | "Store CHARSETS's PROPNAME property with value VALUE. |
2337 | It can be retrieved with `(get-charset-property CHARSET PROPNAME)'." | |
1300d43f KH |
2338 | (or (eq charset 'composition) |
2339 | (set-charset-plist charset | |
2340 | (plist-put (charset-plist charset) propname value)))) | |
4ed46869 KH |
2341 | |
2342 | ;;; Character code property | |
2343 | (put 'char-code-property-table 'char-table-extra-slots 0) | |
2344 | ||
2345 | (defvar char-code-property-table | |
2346 | (make-char-table 'char-code-property-table) | |
2347 | "Char-table containing a property list of each character code. | |
2348 | ||
2349 | See also the documentation of `get-char-code-property' and | |
96db204a | 2350 | `put-char-code-property'.") |
4ed46869 KH |
2351 | |
2352 | (defun get-char-code-property (char propname) | |
2353 | "Return the value of CHAR's PROPNAME property in `char-code-property-table'." | |
2354 | (let ((plist (aref char-code-property-table char))) | |
2355 | (if (listp plist) | |
2356 | (car (cdr (memq propname plist)))))) | |
2357 | ||
2358 | (defun put-char-code-property (char propname value) | |
2359 | "Store CHAR's PROPNAME property with VALUE in `char-code-property-table'. | |
2360 | It can be retrieved with `(get-char-code-property CHAR PROPNAME)'." | |
2361 | (let ((plist (aref char-code-property-table char))) | |
2362 | (if plist | |
2363 | (let ((slot (memq propname plist))) | |
2364 | (if slot | |
2365 | (setcar (cdr slot) value) | |
2366 | (nconc plist (list propname value)))) | |
2367 | (aset char-code-property-table char (list propname value))))) | |
2368 | ||
a127b764 KH |
2369 | \f |
2370 | ;; Pretty description of encoded string | |
2371 | ||
2372 | ;; Alist of ISO 2022 control code vs the corresponding mnemonic string. | |
2373 | (defvar iso-2022-control-alist | |
2374 | '((?\x1b . "ESC") | |
2375 | (?\x0e . "SO") | |
2376 | (?\x0f . "SI") | |
2377 | (?\x8e . "SS2") | |
2378 | (?\x8f . "SS3") | |
2379 | (?\x9b . "CSI"))) | |
2380 | ||
2381 | (defun encoded-string-description (str coding-system) | |
2382 | "Return a pretty description of STR that is encoded by CODING-SYSTEM." | |
2383 | (setq str (string-as-unibyte str)) | |
993b2a7d | 2384 | (mapconcat |
8c9d55a9 | 2385 | (if (and coding-system (eq (coding-system-type coding-system) 2)) |
993b2a7d KH |
2386 | ;; Try to get a pretty description for ISO 2022 escape sequences. |
2387 | (function (lambda (x) (or (cdr (assq x iso-2022-control-alist)) | |
2388 | (format "%02X" x)))) | |
5de75f53 | 2389 | (function (lambda (x) (format "0x%02X" x)))) |
993b2a7d | 2390 | str " ")) |
a127b764 KH |
2391 | |
2392 | (defun encode-coding-char (char coding-system) | |
2393 | "Encode CHAR by CODING-SYSTEM and return the resulting string. | |
2394 | If CODING-SYSTEM can't safely encode CHAR, return nil." | |
bd953173 KH |
2395 | (let ((str1 (string-as-multibyte (char-to-string char))) |
2396 | (str2 (string-as-multibyte (make-string 2 char))) | |
a89f541b KH |
2397 | (safe-chars (and coding-system |
2398 | (coding-system-get coding-system 'safe-chars))) | |
8c9d55a9 | 2399 | (charset (char-charset char)) |
a127b764 | 2400 | enc1 enc2 i1 i2) |
a89f541b | 2401 | (when (or (eq safe-chars t) |
8c9d55a9 | 2402 | (eq charset 'ascii) |
a89f541b | 2403 | (and safe-chars (aref safe-chars char))) |
a127b764 KH |
2404 | ;; We must find the encoded string of CHAR. But, just encoding |
2405 | ;; CHAR will put extra control sequences (usually to designate | |
2406 | ;; ASCII charaset) at the tail if type of CODING is ISO 2022. | |
2407 | ;; To exclude such tailing bytes, we at first encode one-char | |
2408 | ;; string and two-char string, then check how many bytes at the | |
2409 | ;; tail of both encoded strings are the same. | |
2410 | ||
bd953173 | 2411 | (setq enc1 (encode-coding-string str1 coding-system) |
a127b764 | 2412 | i1 (length enc1) |
bd953173 | 2413 | enc2 (encode-coding-string str2 coding-system) |
a127b764 KH |
2414 | i2 (length enc2)) |
2415 | (while (and (> i1 0) (= (aref enc1 (1- i1)) (aref enc2 (1- i2)))) | |
2416 | (setq i1 (1- i1) i2 (1- i2))) | |
2417 | ||
2418 | ;; Now (substring enc1 i1) and (substring enc2 i2) are the same, | |
2419 | ;; and they are the extra control sequences at the tail to | |
2420 | ;; exclude. | |
2421 | (substring enc2 0 i2)))) | |
2422 | ||
2423 | ||
4ed46869 | 2424 | ;;; mule-cmds.el ends here |