]> code.delx.au - gnu-emacs/blob - lisp/international/mule-cmds.el
(locale-language-names): Use "French" for "fr".
[gnu-emacs] / lisp / international / mule-cmds.el
1 ;;; mule-cmds.el --- commands for mulitilingual environment
2
3 ;; Copyright (C) 1995 Electrotechnical Laboratory, JAPAN.
4 ;; Licensed to the Free Software Foundation.
5 ;; Copyright (C) 2000, 2001 Free Software Foundation, Inc.
6
7 ;; Keywords: mule, multilingual
8
9 ;; This file is part of GNU Emacs.
10
11 ;; GNU Emacs is free software; you can redistribute it and/or modify
12 ;; it under the terms of the GNU General Public License as published by
13 ;; the Free Software Foundation; either version 2, or (at your option)
14 ;; any later version.
15
16 ;; GNU Emacs is distributed in the hope that it will be useful,
17 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
18 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
19 ;; GNU General Public License for more details.
20
21 ;; You should have received a copy of the GNU General Public License
22 ;; along with GNU Emacs; see the file COPYING. If not, write to the
23 ;; Free Software Foundation, Inc., 59 Temple Place - Suite 330,
24 ;; Boston, MA 02111-1307, USA.
25
26 ;;; Commentary:
27
28 ;;; Code:
29
30 (eval-when-compile (defvar dos-codepage))
31
32 ;;; MULE related key bindings and menus.
33
34 (defvar mule-keymap (make-sparse-keymap)
35 "Keymap for Mule (Multilingual environment) specific commands.")
36
37 ;; Keep "C-x C-m ..." for mule specific commands.
38 (define-key ctl-x-map "\C-m" mule-keymap)
39
40 (define-key mule-keymap "f" 'set-buffer-file-coding-system)
41 (define-key mule-keymap "t" 'set-terminal-coding-system)
42 (define-key mule-keymap "k" 'set-keyboard-coding-system)
43 (define-key mule-keymap "p" 'set-buffer-process-coding-system)
44 (define-key mule-keymap "x" 'set-selection-coding-system)
45 (define-key mule-keymap "X" 'set-next-selection-coding-system)
46 (define-key mule-keymap "\C-\\" 'set-input-method)
47 (define-key mule-keymap "c" 'universal-coding-system-argument)
48 (define-key mule-keymap "l" 'set-language-environment)
49
50 (define-key help-map "\C-L" 'describe-language-environment)
51 (define-key help-map "L" 'describe-language-environment)
52 (define-key help-map "\C-\\" 'describe-input-method)
53 (define-key help-map "I" 'describe-input-method)
54 (define-key help-map "C" 'describe-coding-system)
55 (define-key help-map "h" 'view-hello-file)
56
57 (defvar mule-menu-keymap
58 (make-sparse-keymap "Mule (Multilingual Environment)")
59 "Keymap for Mule (Multilingual environment) menu specific commands.")
60
61 (defvar describe-language-environment-map
62 (make-sparse-keymap "Describe Language Environment"))
63
64 (defvar setup-language-environment-map
65 (make-sparse-keymap "Set Language Environment"))
66
67 (defvar set-coding-system-map
68 (make-sparse-keymap "Set Coding System"))
69
70 (define-key-after mule-menu-keymap [set-language-environment]
71 (list 'menu-item "Set Language Environment" setup-language-environment-map
72 :help "Multilingual environment suitable for a specific language"))
73 (define-key-after mule-menu-keymap [mouse-set-font]
74 '(menu-item "Set Font/Fontset" mouse-set-font
75 :visible (fboundp 'generate-fontset-menu)
76 :help "Select a font from list of known fonts/fontsets"))
77 (define-key-after mule-menu-keymap [separator-mule]
78 '("--")
79 t)
80 (define-key-after mule-menu-keymap [toggle-input-method]
81 '(menu-item "Toggle Input Method" toggle-input-method)
82 t)
83 (define-key-after mule-menu-keymap [set-input-method]
84 '(menu-item "Select Input Method..." set-input-method)
85 t)
86 (define-key-after mule-menu-keymap [describe-input-method]
87 '(menu-item "Describe Input Method" describe-input-method))
88 (define-key-after mule-menu-keymap [separator-input-method]
89 '("--")
90 t)
91 (define-key-after mule-menu-keymap [set-various-coding-system]
92 (list 'menu-item "Set Coding Systems" set-coding-system-map
93 :enable 'enable-multibyte-characters))
94 (define-key-after mule-menu-keymap [view-hello-file]
95 '(menu-item "Show Multi-lingual Text" view-hello-file
96 :enable (file-readable-p
97 (expand-file-name "HELLO" data-directory))
98 :help "Display file which says HELLO in many languages")
99 t)
100 (define-key-after mule-menu-keymap [separator-coding-system]
101 '("--")
102 t)
103 (define-key-after mule-menu-keymap [describe-language-environment]
104 (list 'menu-item "Describe Language Environment"
105 describe-language-environment-map
106 :help "Show multilingual settings for a specific language")
107 t)
108 (define-key-after mule-menu-keymap [describe-input-method]
109 '(menu-item "Describe Input Method..." describe-input-method
110 :help "Keyboard layout for a specific input method")
111 t)
112 (define-key-after mule-menu-keymap [describe-coding-system]
113 '(menu-item "Describe Coding System..." describe-coding-system)
114 t)
115 (define-key-after mule-menu-keymap [list-character-sets]
116 '(menu-item "List Character Sets" list-character-sets
117 :help "Show table of available character sets"))
118 (define-key-after mule-menu-keymap [mule-diag]
119 '(menu-item "Show All of Mule Status" mule-diag
120 :help "Display multilingual environment settings")
121 t)
122
123 (define-key-after set-coding-system-map [set-buffer-file-coding-system]
124 '(menu-item "For Saving this Buffer" set-buffer-file-coding-system
125 :help "How to encode this buffer on disk")
126 t)
127 (define-key-after set-coding-system-map [universal-coding-system-argument]
128 '(menu-item "For Next Command" universal-coding-system-argument
129 :help "Coding system to be used by next command")
130 t)
131 (define-key-after set-coding-system-map [set-terminal-coding-system]
132 '(menu-item "For Terminal" set-terminal-coding-system
133 :enable (null (memq window-system '(x w32 mac)))
134 :help "How to encode terminal output")
135 t)
136 (define-key-after set-coding-system-map [set-keyboard-coding-system]
137 '(menu-item "For Keyboard" set-keyboard-coding-system
138 :help "How to decode keyboard input")
139 t)
140 (define-key-after set-coding-system-map [set-buffer-process-coding-system]
141 '(menu-item "For I/O with Subprocess" set-buffer-process-coding-system
142 :visible (fboundp 'start-process)
143 :enable (get-buffer-process (current-buffer))
144 :help "How to en/decode I/O from/to subprocess connected to this buffer")
145 t)
146 (define-key-after set-coding-system-map [set-selection-coding-system]
147 '(menu-item "For X Selections/Clipboard" set-selection-coding-system
148 :visible (display-selections-p)
149 :help "How to en/decode data to/from selection/clipboard")
150 t)
151 (define-key-after set-coding-system-map [set-next-selection-coding-system]
152 '(menu-item "For Next X Selection" set-next-selection-coding-system
153 :visible (display-selections-p)
154 :help "How to en/decode next selection/clipboard operation")
155 t)
156 (define-key setup-language-environment-map
157 [Default] '(menu-item "Default" setup-specified-language-environment))
158
159 (define-key describe-language-environment-map
160 [Default] '(menu-item "Default" describe-specified-language-support))
161
162 ;; This should be a single character key binding because users use it
163 ;; very frequently while editing multilingual text. Now we can use
164 ;; only two such keys: "\C-\\" and "\C-^", but the latter is not
165 ;; convenient because it requires shifting on most keyboards. An
166 ;; alternative is "\C-\]" which is now bound to `abort-recursive-edit'
167 ;; but it won't be used that frequently.
168 (define-key global-map "\C-\\" 'toggle-input-method)
169
170 ;; This is no good because people often type Shift-SPC
171 ;; meaning to type SPC. -- rms.
172 ;; ;; Here's an alternative key binding for X users (Shift-SPACE).
173 ;; (define-key global-map [?\S- ] 'toggle-input-method)
174
175 ;;; Mule related hyperlinks.
176 (defconst help-xref-mule-regexp-template
177 (purecopy (concat "\\(\\<\\("
178 "\\(coding system\\)\\|"
179 "\\(input method\\)\\|"
180 "\\(character set\\)\\|"
181 "\\(charset\\)"
182 "\\)\\s-+\\)?"
183 ;; Note starting with word-syntax character:
184 "`\\(\\sw\\(\\sw\\|\\s_\\)+\\)'")))
185
186 (defun coding-system-change-eol-conversion (coding-system eol-type)
187 "Return a coding system which differs from CODING-SYSTEM in eol conversion.
188 The returned coding system converts end-of-line by EOL-TYPE
189 but text as the same way as CODING-SYSTEM.
190 EOL-TYPE should be `unix', `dos', `mac', or nil.
191 If EOL-TYPE is nil, the returned coding system detects
192 how end-of-line is formatted automatically while decoding.
193
194 EOL-TYPE can be specified by an integer 0, 1, or 2.
195 They means `unix', `dos', and `mac' respectively."
196 (if (symbolp eol-type)
197 (setq eol-type (cond ((eq eol-type 'unix) 0)
198 ((eq eol-type 'dos) 1)
199 ((eq eol-type 'mac) 2)
200 (t eol-type))))
201 (let ((orig-eol-type (coding-system-eol-type coding-system)))
202 (if (vectorp orig-eol-type)
203 (if (not eol-type)
204 coding-system
205 (aref orig-eol-type eol-type))
206 (let ((base (coding-system-base coding-system)))
207 (if (not eol-type)
208 base
209 (if (= eol-type orig-eol-type)
210 coding-system
211 (setq orig-eol-type (coding-system-eol-type base))
212 (if (vectorp orig-eol-type)
213 (aref orig-eol-type eol-type))))))))
214
215 (defun coding-system-change-text-conversion (coding-system coding)
216 "Return a coding system which differs from CODING-SYSTEM in text conversion.
217 The returned coding system converts text by CODING
218 but end-of-line as the same way as CODING-SYSTEM.
219 If CODING is nil, the returned coding system detects
220 how text is formatted automatically while decoding."
221 (if (not coding)
222 (coding-system-base coding-system)
223 (let ((eol-type (coding-system-eol-type coding-system)))
224 (coding-system-change-eol-conversion
225 coding
226 (if (numberp eol-type) (aref [unix dos mac] eol-type))))))
227
228 (defun toggle-enable-multibyte-characters (&optional arg)
229 "Change whether this buffer uses multibyte characters.
230 With arg, use multibyte characters if the arg is positive.
231
232 Note that this command does not convert the byte contents of
233 the buffer; it only changes the way those bytes are interpreted.
234 In general, therefore, this command *changes* the sequence of
235 characters that the current buffer contains.
236
237 We suggest you avoid using use this command unless you know what you
238 are doing. If you use it by mistake, and the buffer is now displayed
239 wrong, use this command again to toggle back to the right mode."
240 (interactive "P")
241 (let ((new-flag
242 (if (null arg) (null enable-multibyte-characters)
243 (> (prefix-numeric-value arg) 0))))
244 (set-buffer-multibyte new-flag))
245 (force-mode-line-update))
246
247 (defun view-hello-file ()
248 "Display the HELLO file which list up many languages and characters."
249 (interactive)
250 ;; We have to decode the file in any environment.
251 (let ((default-enable-multibyte-characters t)
252 (coding-system-for-read 'iso-2022-7bit))
253 (find-file-read-only (expand-file-name "HELLO" data-directory))))
254
255 (defun universal-coding-system-argument ()
256 "Execute an I/O command using the specified coding system."
257 (interactive)
258 (let* ((default (and buffer-file-coding-system
259 (not (eq (coding-system-type buffer-file-coding-system)
260 t))
261 buffer-file-coding-system))
262 (coding-system (read-coding-system
263 (if default
264 (format "Coding system for following command (default, %s): " default)
265 "Coding system for following command: ")
266 default))
267 (keyseq (read-key-sequence
268 (format "Command to execute with %s:" coding-system)))
269 (cmd (key-binding keyseq))
270 prefix)
271
272 (when (eq cmd 'universal-argument)
273 (call-interactively cmd)
274
275 ;; Process keys bound in `universal-argument-map'.
276 (while (progn
277 (setq keyseq (read-key-sequence nil t)
278 cmd (key-binding keyseq t))
279 (not (eq cmd 'universal-argument-other-key)))
280 (let ((current-prefix-arg prefix-arg)
281 ;; Have to bind `last-command-char' here so that
282 ;; `digit-argument', for isntance, can compute the
283 ;; prefix arg.
284 (last-command-char (aref keyseq 0)))
285 (call-interactively cmd)))
286
287 ;; This is the final call to `univeral-argument-other-key', which
288 ;; set's the final `prefix-arg.
289 (let ((current-prefix-arg prefix-arg))
290 (call-interactively cmd))
291
292 ;; Read the command to execute with the given prefix arg.
293 (setq prefix prefix-arg
294 keyseq (read-key-sequence nil t)
295 cmd (key-binding keyseq)))
296
297 (let ((coding-system-for-read coding-system)
298 (coding-system-for-write coding-system)
299 (current-prefix-arg prefix))
300 (message "")
301 (call-interactively cmd))))
302
303 (defun set-default-coding-systems (coding-system)
304 "Set default value of various coding systems to CODING-SYSTEM.
305 This sets the following coding systems:
306 o coding system of a newly created buffer
307 o default coding system for subprocess I/O
308 This also sets the following values:
309 o default value used as file-name-coding-system for converting file names.
310 o default value for the command `set-terminal-coding-system' (not on MSDOS)
311 o default value for the command `set-keyboard-coding-system'."
312 (check-coding-system coding-system)
313 (setq-default buffer-file-coding-system coding-system)
314 (if default-enable-multibyte-characters
315 (setq default-file-name-coding-system coding-system))
316 ;; If coding-system is nil, honor that on MS-DOS as well, so
317 ;; that they could reset the terminal coding system.
318 (unless (and (eq window-system 'pc) coding-system)
319 (setq default-terminal-coding-system coding-system))
320 (setq default-keyboard-coding-system coding-system)
321 (setq default-process-coding-system (cons coding-system coding-system)))
322
323 (defalias 'update-iso-coding-systems 'update-coding-systems-internal)
324 (make-obsolete 'update-iso-coding-systems 'update-coding-systems-internal "20.3")
325
326 (defun prefer-coding-system (coding-system)
327 "Add CODING-SYSTEM at the front of the priority list for automatic detection.
328 This also sets the following coding systems:
329 o coding system of a newly created buffer
330 o default coding system for subprocess I/O
331 This also sets the following values:
332 o default value used as `file-name-coding-system' for converting file names.
333 o default value for the command `set-terminal-coding-system' (not on MSDOS)
334 o default value for the command `set-keyboard-coding-system'
335
336 If CODING-SYSTEM specifies a certain type of EOL conversion, the coding
337 systems set by this function will use that type of EOL conversion.
338
339 This command does not change the default value of terminal coding system
340 for MS-DOS terminal, because DOS terminals only support a single coding
341 system, and Emacs automatically sets the default to that coding system at
342 startup.
343
344 A coding system that requires automatic detection of text
345 encoding (e.g. undecided, unix) can't be preferred.
346
347 See also `coding-category-list' and `coding-system-category'."
348 (interactive "zPrefer coding system: ")
349 (if (not (and coding-system (coding-system-p coding-system)))
350 (error "Invalid coding system `%s'" coding-system))
351 (let ((coding-category (coding-system-category coding-system))
352 (base (coding-system-base coding-system))
353 (eol-type (coding-system-eol-type coding-system)))
354 (if (not coding-category)
355 ;; CODING-SYSTEM is no-conversion or undecided.
356 (error "Can't prefer the coding system `%s'" coding-system))
357 (set coding-category (or base coding-system))
358 (update-coding-systems-internal)
359 (or (eq coding-category (car coding-category-list))
360 ;; We must change the order.
361 (set-coding-priority (list coding-category)))
362 (if (and base (interactive-p))
363 (message "Highest priority is set to %s (base of %s)"
364 base coding-system))
365 ;; If they asked for specific EOL conversion, honor that.
366 (if (memq eol-type '(0 1 2))
367 (setq coding-system
368 (coding-system-change-eol-conversion base eol-type))
369 (setq coding-system base))
370 (set-default-coding-systems coding-system)))
371
372 (defvar sort-coding-systems-predicate nil
373 "If non-nil, a predicate function to sort coding systems.
374
375 It is called with two coding systems, and should return t if the first
376 one is \"less\" than the second.
377
378 The function `sort-coding-systems' use it.")
379
380 (defun sort-coding-systems (codings)
381 "Sort coding system list CODINGS by a priority of each coding system.
382
383 If a coding system is most preferred, it has the highest priority.
384 Otherwise, a coding system corresponds to some MIME charset has higher
385 priorities. Among them, a coding system included in `coding-system'
386 key of the current language environment has higher priorities. See
387 also the documentation of `language-info-alist'.
388
389 If the variable `sort-coding-systems-predicate' (which see) is
390 non-nil, it is used to sort CODINGS in the different way than above."
391 (if sort-coding-systems-predicate
392 (sort codings sort-coding-systems-predicate)
393 (let* ((most-preferred (symbol-value (car coding-category-list)))
394 (lang-preferred (get-language-info current-language-environment
395 'coding-system))
396 (func (function
397 (lambda (x)
398 (let ((base (coding-system-base x)))
399 (+ (if (eq base most-preferred) 64 0)
400 (let ((mime (coding-system-get base 'mime-charset)))
401 (if mime
402 (if (string-match "^x-" (symbol-name mime))
403 16 32)
404 0))
405 (if (memq base lang-preferred) 8 0)
406 (if (string-match "-with-esc$" (symbol-name base))
407 0 4)
408 (if (eq (coding-system-type base) 2)
409 ;; For ISO based coding systems, prefer
410 ;; one that doesn't use escape sequences.
411 (let ((flags (coding-system-flags base)))
412 (if (or (consp (aref flags 0))
413 (consp (aref flags 1))
414 (consp (aref flags 2))
415 (consp (aref flags 3)))
416 (if (or (aref flags 8) (aref flags 9))
417 0
418 1)
419 2))
420 1)))))))
421 (sort codings (function (lambda (x y)
422 (> (funcall func x) (funcall func y))))))))
423
424 (defun find-coding-systems-region (from to)
425 "Return a list of proper coding systems to encode a text between FROM and TO.
426 All coding systems in the list can safely encode any multibyte characters
427 in the text.
428
429 If the text contains no multibyte characters, return a list of a single
430 element `undecided'."
431 (let ((codings (find-coding-systems-region-internal from to)))
432 (if (eq codings t)
433 ;; The text contains only ASCII characters. Any coding
434 ;; systems are safe.
435 '(undecided)
436 ;; We need copy-sequence because sorting will alter the argument.
437 (sort-coding-systems (copy-sequence codings)))))
438
439 (defun find-coding-systems-string (string)
440 "Return a list of proper coding systems to encode STRING.
441 All coding systems in the list can safely encode any multibyte characters
442 in STRING.
443
444 If STRING contains no multibyte characters, return a list of a single
445 element `undecided'."
446 (find-coding-systems-region string nil))
447
448 (defun find-coding-systems-for-charsets (charsets)
449 "Return a list of proper coding systems to encode characters of CHARSETS.
450 CHARSETS is a list of character sets."
451 (cond ((or (null charsets)
452 (and (= (length charsets) 1)
453 (eq 'ascii (car charsets))))
454 '(undecided))
455 ((or (memq 'eight-bit-control charsets)
456 (memq 'eight-bit-graphic charsets))
457 '(raw-text emacs-mule))
458 (t
459 (let ((codings t)
460 charset l ll)
461 (while (and codings charsets)
462 (setq charset (car charsets) charsets (cdr charsets))
463 (unless (eq charset 'ascii)
464 (setq l (aref char-coding-system-table (make-char charset)))
465 (if (eq codings t)
466 (setq codings l)
467 (let ((ll nil))
468 (while codings
469 (if (memq (car codings) l)
470 (setq ll (cons (car codings) ll)))
471 (setq codings (cdr codings)))
472 (setq codings ll)))))
473 (append codings
474 (char-table-extra-slot char-coding-system-table 0))))))
475
476 (defun find-multibyte-characters (from to &optional maxcount excludes)
477 "Find multibyte characters in the region specified by FROM and TO.
478 If FROM is a string, find multibyte characters in the string.
479 The return value is an alist of the following format:
480 ((CHARSET COUNT CHAR ...) ...)
481 where
482 CHARSET is a character set,
483 COUNT is a number of characters,
484 CHARs are found characters of the character set.
485 Optional 3rd arg MAXCOUNT limits how many CHARs are put in the above list.
486 Optional 4th arg EXCLUDE is a list of character sets to be ignored.
487
488 For invalid characters, CHARs are actually strings."
489 (let ((chars nil)
490 charset char)
491 (if (stringp from)
492 (let ((idx 0))
493 (while (setq idx (string-match "[^\000-\177]" from idx))
494 (setq char (aref from idx)
495 charset (char-charset char))
496 (if (eq charset 'unknown)
497 (setq char (match-string 0)))
498 (if (or (memq charset '(unknown
499 eight-bit-control eight-bit-graphic))
500 (not (or (eq excludes t) (memq charset excludes))))
501 (let ((slot (assq charset chars)))
502 (if slot
503 (if (not (memq char (nthcdr 2 slot)))
504 (let ((count (nth 1 slot)))
505 (setcar (cdr slot) (1+ count))
506 (if (or (not maxcount) (< count maxcount))
507 (nconc slot (list char)))))
508 (setq chars (cons (list charset 1 char) chars)))))
509 (setq idx (1+ idx))))
510 (save-excursion
511 (goto-char from)
512 (while (re-search-forward "[^\000-\177]" to t)
513 (setq char (preceding-char)
514 charset (char-charset char))
515 (if (eq charset 'unknown)
516 (setq char (match-string 0)))
517 (if (or (memq charset '(unknown eight-bit-control eight-bit-graphic))
518 (not (or (eq excludes t) (memq charset excludes))))
519 (let ((slot (assq charset chars)))
520 (if slot
521 (if (not (member char (nthcdr 2 slot)))
522 (let ((count (nth 1 slot)))
523 (setcar (cdr slot) (1+ count))
524 (if (or (not maxcount) (< count maxcount))
525 (nconc slot (list char)))))
526 (setq chars (cons (list charset 1 char) chars))))))))
527 (nreverse chars)))
528
529 (defvar last-coding-system-specified nil
530 "Most recent coding system explicitly specified by the user when asked.
531 This variable is set whenever Emacs asks the user which coding system
532 to use in order to write a file. If you set it to nil explicitly,
533 then call `write-region', then afterward this variable will be non-nil
534 only if the user was explicitly asked and specified a coding system.")
535
536 (defvar select-safe-coding-system-accept-default-p nil
537 "If non-nil, a function to control the behaviour of coding system selection.
538 The meaning is the same as the argument ACCEPT-DEFAULT-P of the
539 function `select-safe-coding-system' (which see). This variable
540 overrides that argument.")
541
542 (defun select-safe-coding-system (from to &optional default-coding-system
543 accept-default-p)
544 "Ask a user to select a safe coding system from candidates.
545 The candidates of coding systems which can safely encode a text
546 between FROM and TO are shown in a popup window. Among them, the most
547 proper one is suggested as the default.
548
549 The list of `buffer-file-coding-system' of the current buffer and the
550 most preferred coding system (if it corresponds to a MIME charset) is
551 treated as the default coding system list. Among them, the first one
552 that safely encodes the text is silently selected and returned without
553 any user interaction. See also the command `prefer-coding-system'.
554
555 Optional 3rd arg DEFAULT-CODING-SYSTEM specifies a coding system or a
556 list of coding systems to be prepended to the default coding system
557 list.
558
559 Optional 4th arg ACCEPT-DEFAULT-P, if non-nil, is a function to
560 determine the acceptability of the silently selected coding system.
561 It is called with that coding system, and should return nil if it
562 should not be silently selected and thus user interaction is required.
563
564 The variable `select-safe-coding-system-accept-default-p', if
565 non-nil, overrides ACCEPT-DEFAULT-P.
566
567 Kludgy feature: if FROM is a string, the string is the target text,
568 and TO is ignored."
569 (if (and default-coding-system
570 (not (listp default-coding-system)))
571 (setq default-coding-system (list default-coding-system)))
572
573 ;; Change elements of the list to (coding . base-coding).
574 (setq default-coding-system
575 (mapcar (function (lambda (x) (cons x (coding-system-base x))))
576 default-coding-system))
577
578 ;; If buffer-file-coding-system is not nil nor undecided, append it
579 ;; to the defaults.
580 (if buffer-file-coding-system
581 (let ((base (coding-system-base buffer-file-coding-system)))
582 (or (eq base 'undecided)
583 (assq buffer-file-coding-system default-coding-system)
584 (rassq base default-coding-system)
585 (setq default-coding-system
586 (append default-coding-system
587 (list (cons buffer-file-coding-system base)))))))
588
589 ;; If the most preferred coding system has the property mime-charset,
590 ;; append it to the defaults.
591 (let ((tail coding-category-list)
592 preferred base)
593 (while (and tail
594 (not (setq preferred (symbol-value (car tail)))))
595 (setq tail (cdr tail)))
596 (and (coding-system-p preferred)
597 (setq base (coding-system-base preferred))
598 (coding-system-get preferred 'mime-charset)
599 (not (assq preferred default-coding-system))
600 (not (rassq base default-coding-system))
601 (setq default-coding-system
602 (append default-coding-system (list (cons preferred base))))))
603
604 (if select-safe-coding-system-accept-default-p
605 (setq accept-default-p select-safe-coding-system-accept-default-p))
606
607 (let ((codings (find-coding-systems-region from to))
608 (coding-system nil)
609 (l default-coding-system))
610 (if (eq (car codings) 'undecided)
611 ;; Any coding system is ok.
612 (setq coding-system t)
613 ;; Try the defaults.
614 (while (and l (not coding-system))
615 (if (memq (cdr (car l)) codings)
616 (setq coding-system (car (car l)))
617 (setq l (cdr l))))
618 (if (and coding-system accept-default-p)
619 (or (funcall accept-default-p coding-system)
620 (setq coding-system (list coding-system)))))
621
622 ;; If all the defaults failed, ask a user.
623 (when (or (not coding-system) (consp coding-system))
624 ;; At first, change each coding system to the corresponding
625 ;; mime-charset name if it is also a coding system. Such a name
626 ;; is more friendly to users.
627 (let ((l codings)
628 mime-charset)
629 (while l
630 (setq mime-charset (coding-system-get (car l) 'mime-charset))
631 (if (and mime-charset (coding-system-p mime-charset))
632 (setcar l mime-charset))
633 (setq l (cdr l))))
634
635 ;; Then ask users to select one form CODINGS.
636 (unwind-protect
637 (save-window-excursion
638 (with-output-to-temp-buffer "*Warning*"
639 (save-excursion
640 (set-buffer standard-output)
641 (if (not default-coding-system)
642 (insert "No default coding systems to try.")
643 (insert "These default coding systems were tried")
644 (if (stringp from)
645 (insert " to encode \""
646 (if (> (length from) 10)
647 (substring from 0 10)
648 from)
649 "...\""))
650 (insert ":\n")
651 (let ((pos (point))
652 (fill-prefix " "))
653 (mapcar (function (lambda (x)
654 (princ " ") (princ (car x))))
655 default-coding-system)
656 (insert "\n")
657 (fill-region-as-paragraph pos (point)))
658 (insert
659 (if (consp coding-system)
660 (concat (format "%s safely encodes the target text,\n"
661 (car coding-system))
662 "but it is not recommended for encoding text in this context,\n"
663 "e.g., for sending an email message.\n")
664 "However, none of them safely encodes the target text.\n")))
665 (insert (if (consp coding-system)
666 "\nSelect the above, or "
667 "\nSelect ")
668 "one of the following safe coding systems:\n")
669 (let ((pos (point))
670 (fill-prefix " "))
671 (mapcar (function (lambda (x) (princ " ") (princ x)))
672 codings)
673 (insert "\n")
674 (fill-region-as-paragraph pos (point)))))
675
676 ;; Read a coding system.
677 (if (consp coding-system)
678 (setq codings (cons (car coding-system) codings)))
679 (let* ((safe-names (mapcar (lambda (x) (list (symbol-name x)))
680 codings))
681 (name (completing-read
682 (format "Select coding system (default %s): "
683 (car codings))
684 safe-names nil t nil nil
685 (car (car safe-names)))))
686 (setq last-coding-system-specified (intern name)
687 coding-system last-coding-system-specified)))
688 (kill-buffer "*Warning*")))
689
690 (if (vectorp (coding-system-eol-type coding-system))
691 (let ((eol (coding-system-eol-type buffer-file-coding-system)))
692 (if (numberp eol)
693 (setq coding-system
694 (coding-system-change-eol-conversion coding-system eol)))))
695
696 (if (eq coding-system t)
697 (setq coding-system buffer-file-coding-system))
698 coding-system))
699
700 (setq select-safe-coding-system-function 'select-safe-coding-system)
701
702 (defun select-message-coding-system ()
703 "Return a coding system to encode the outgoing message of the current buffer.
704 It at first tries the first coding system found in these variables
705 in this order:
706 (1) local value of `buffer-file-coding-system'
707 (2) value of `sendmail-coding-system'
708 (3) value of `default-sendmail-coding-system'
709 (4) value of `default-buffer-file-coding-system'
710 If the found coding system can't encode the current buffer,
711 or none of them are bound to a coding system,
712 it asks the user to select a proper coding system."
713 (let ((coding (or (and (local-variable-p 'buffer-file-coding-system)
714 buffer-file-coding-system)
715 sendmail-coding-system
716 default-sendmail-coding-system
717 default-buffer-file-coding-system)))
718 (if (eq coding 'no-conversion)
719 ;; We should never use no-conversion for outgoing mails.
720 (setq coding nil))
721 (if (fboundp select-safe-coding-system-function)
722 (funcall select-safe-coding-system-function
723 (point-min) (point-max) coding
724 (function (lambda (x) (coding-system-get x 'mime-charset))))
725 coding)))
726 \f
727 ;;; Language support stuff.
728
729 (defvar language-info-alist nil
730 "Alist of language environment definitions.
731 Each element looks like:
732 (LANGUAGE-NAME . ((KEY . INFO) ...))
733 where LANGUAGE-NAME is a string, the name of the language environment,
734 KEY is a symbol denoting the kind of information, and
735 INFO is the data associated with KEY.
736 Meaningful values for KEY include
737
738 documentation value is documentation of what this language environment
739 is meant for, and how to use it.
740 charset value is a list of the character sets used by this
741 language environment.
742 sample-text value is one line of text,
743 written using those character sets,
744 appropriate for this language environment.
745 setup-function value is a function to call to switch to this
746 language environment.
747 exit-function value is a function to call to leave this
748 language environment.
749 coding-system value is a list of coding systems that are good
750 for saving text written in this language environment.
751 This list serves as suggestions to the user;
752 in effect, as a kind of documentation.
753 coding-priority value is a list of coding systems for this language
754 environment, in order of decreasing priority.
755 This is used to set up the coding system priority
756 list when you switch to this language environment.
757 nonascii-translation
758 value is a translation table to be set in the
759 variable `nonascii-translation-table' in this
760 language environment, or a character set from
761 which `nonascii-insert-offset' is calculated.
762 input-method value is a default input method for this language
763 environment.
764 features value is a list of features requested in this
765 language environment.
766
767 The following keys take effect only when multibyte characters are
768 globally disabled, i.e. the value of `default-enable-multibyte-characters'
769 is nil.
770
771 unibyte-syntax value is a library name to load to set
772 unibyte 8-bit character syntaxes for this
773 language environment.
774
775 unibyte-display value is a coding system to encode characters
776 for the terminal. Characters in the range
777 of 160 to 255 display not as octal escapes,
778 but as non-ASCII characters in this language
779 environment.")
780
781 (defun get-language-info (lang-env key)
782 "Return information listed under KEY for language environment LANG-ENV.
783 KEY is a symbol denoting the kind of information.
784 For a list of useful values for KEY and their meanings,
785 see `language-info-alist'."
786 (if (symbolp lang-env)
787 (setq lang-env (symbol-name lang-env)))
788 (let ((lang-slot (assoc-ignore-case lang-env language-info-alist)))
789 (if lang-slot
790 (cdr (assq key (cdr lang-slot))))))
791
792 (defun set-language-info (lang-env key info)
793 "Modify part of the definition of language environment LANG-ENV.
794 Specifically, this stores the information INFO under KEY
795 in the definition of this language environment.
796 KEY is a symbol denoting the kind of information.
797 INFO is the value for that information.
798
799 For a list of useful values for KEY and their meanings,
800 see `language-info-alist'."
801 (if (symbolp lang-env)
802 (setq lang-env (symbol-name lang-env)))
803 (let (lang-slot key-slot)
804 (setq lang-slot (assoc lang-env language-info-alist))
805 (if (null lang-slot) ; If no slot for the language, add it.
806 (setq lang-slot (list lang-env)
807 language-info-alist (cons lang-slot language-info-alist)))
808 (setq key-slot (assq key lang-slot))
809 (if (null key-slot) ; If no slot for the key, add it.
810 (progn
811 (setq key-slot (list key))
812 (setcdr lang-slot (cons key-slot (cdr lang-slot)))))
813 (setcdr key-slot (purecopy info))))
814
815 (defun set-language-info-alist (lang-env alist &optional parents)
816 "Store ALIST as the definition of language environment LANG-ENV.
817 ALIST is an alist of KEY and INFO values. See the documentation of
818 `language-info-alist' for the meanings of KEY and INFO.
819
820 Optional arg PARENTS is a list of parent menu names; it specifies
821 where to put this language environment in the
822 Describe Language Environment and Set Language Environment menus.
823 For example, (\"European\") means to put this language environment
824 in the European submenu in each of those two menus."
825 (if (symbolp lang-env)
826 (setq lang-env (symbol-name lang-env)))
827 (let ((describe-map describe-language-environment-map)
828 (setup-map setup-language-environment-map))
829 (if parents
830 (let ((l parents)
831 map parent-symbol parent prompt)
832 (while l
833 (if (symbolp (setq parent-symbol (car l)))
834 (setq parent (symbol-name parent))
835 (setq parent parent-symbol parent-symbol (intern parent)))
836 (setq map (lookup-key describe-map (vector parent-symbol)))
837 ;; This prompt string is for define-prefix-command, so
838 ;; that the map it creates will be suitable for a menu.
839 (or map (setq prompt (format "%s Environment" parent)))
840 (if (not map)
841 (progn
842 (setq map (intern (format "describe-%s-environment-map"
843 (downcase parent))))
844 (define-prefix-command map nil prompt)
845 (define-key-after describe-map (vector parent-symbol)
846 (cons parent map) t)))
847 (setq describe-map (symbol-value map))
848 (setq map (lookup-key setup-map (vector parent-symbol)))
849 (if (not map)
850 (progn
851 (setq map (intern (format "setup-%s-environment-map"
852 (downcase parent))))
853 (define-prefix-command map nil prompt)
854 (define-key-after setup-map (vector parent-symbol)
855 (cons parent map) t)))
856 (setq setup-map (symbol-value map))
857 (setq l (cdr l)))))
858
859 ;; Set up menu items for this language env.
860 (let ((doc (assq 'documentation alist)))
861 (when doc
862 (define-key-after describe-map (vector (intern lang-env))
863 (cons lang-env 'describe-specified-language-support) t)))
864 (define-key-after setup-map (vector (intern lang-env))
865 (cons lang-env 'setup-specified-language-environment) t)
866
867 (while alist
868 (set-language-info lang-env (car (car alist)) (cdr (car alist)))
869 (setq alist (cdr alist)))))
870
871 (defun read-language-name (key prompt &optional default)
872 "Read a language environment name which has information for KEY.
873 If KEY is nil, read any language environment.
874 Prompt with PROMPT. DEFAULT is the default choice of language environment.
875 This returns a language environment name as a string."
876 (let* ((completion-ignore-case t)
877 (name (completing-read prompt
878 language-info-alist
879 (and key
880 (function (lambda (elm) (assq key elm))))
881 t nil nil default)))
882 (if (and (> (length name) 0)
883 (or (not key)
884 (get-language-info name key)))
885 name)))
886 \f
887 ;;; Multilingual input methods.
888 (defgroup leim nil
889 "LEIM: Libraries of Emacs Input Methods."
890 :group 'mule)
891
892 (defconst leim-list-file-name "leim-list.el"
893 "Name of LEIM list file.
894 This file contains a list of libraries of Emacs input methods (LEIM)
895 in the format of Lisp expression for registering each input method.
896 Emacs loads this file at startup time.")
897
898 (defvar leim-list-header (format
899 ";;; %s -- list of LEIM (Library of Emacs Input Method)
900 ;;
901 ;; This file contains a list of LEIM (Library of Emacs Input Method)
902 ;; methods in the same directory as this file. Loading this file
903 ;; registers all the input methods in Emacs.
904 ;;
905 ;; Each entry has the form:
906 ;; (register-input-method
907 ;; INPUT-METHOD LANGUAGE-NAME ACTIVATE-FUNC
908 ;; TITLE DESCRIPTION
909 ;; ARG ...)
910 ;; See the function `register-input-method' for the meanings of the arguments.
911 ;;
912 ;; If this directory is included in load-path, Emacs automatically
913 ;; loads this file at startup time.
914
915 "
916 leim-list-file-name)
917 "Header to be inserted in LEIM list file.")
918
919 (defvar leim-list-entry-regexp "^(register-input-method"
920 "Regexp matching head of each entry in LEIM list file.
921 See also the variable `leim-list-header'")
922
923 (defvar update-leim-list-functions
924 '(quail-update-leim-list-file)
925 "List of functions to call to update LEIM list file.
926 Each function is called with one arg, LEIM directory name.")
927
928 (defun update-leim-list-file (&rest dirs)
929 "Update LEIM list file in directories DIRS."
930 (let ((functions update-leim-list-functions))
931 (while functions
932 (apply (car functions) dirs)
933 (setq functions (cdr functions)))))
934
935 (defvar current-input-method nil
936 "The current input method for multilingual text.
937 If nil, that means no input method is activated now.")
938 (make-variable-buffer-local 'current-input-method)
939 (put 'current-input-method 'permanent-local t)
940
941 (defvar current-input-method-title nil
942 "Title string of the current input method shown in mode line.")
943 (make-variable-buffer-local 'current-input-method-title)
944 (put 'current-input-method-title 'permanent-local t)
945
946 (defcustom default-input-method nil
947 "*Default input method for multilingual text (a string).
948 This is the input method activated automatically by the command
949 `toggle-input-method' (\\[toggle-input-method])."
950 :group 'mule
951 :type '(choice (const nil) string)
952 :set-after '(current-language-environment))
953
954 (put 'input-method-function 'permanent-local t)
955
956 (defvar input-method-history nil
957 "History list for some commands that read input methods.")
958 (make-variable-buffer-local 'input-method-history)
959 (put 'input-method-history 'permanent-local t)
960
961 (defvar inactivate-current-input-method-function nil
962 "Function to call for inactivating the current input method.
963 Every input method should set this to an appropriate value when activated.
964 This function is called with no argument.
965
966 This function should never change the value of `current-input-method'.
967 It is set to nil by the function `inactivate-input-method'.")
968 (make-variable-buffer-local 'inactivate-current-input-method-function)
969 (put 'inactivate-current-input-method-function 'permanent-local t)
970
971 (defvar describe-current-input-method-function nil
972 "Function to call for describing the current input method.
973 This function is called with no argument.")
974 (make-variable-buffer-local 'describe-current-input-method-function)
975 (put 'describe-current-input-method-function 'permanent-local t)
976
977 (defvar input-method-alist nil
978 "Alist of input method names vs how to use them.
979 Each element has the form:
980 (INPUT-METHOD LANGUAGE-ENV ACTIVATE-FUNC TITLE DESCRIPTION ARGS...)
981 See the function `register-input-method' for the meanings of the elements.")
982
983 (defun register-input-method (input-method lang-env &rest args)
984 "Register INPUT-METHOD as an input method for language environment LANG-ENV.
985 INPUT-METHOD and LANG-ENV are symbols or strings.
986
987 The remaining arguments are:
988 ACTIVATE-FUNC, TITLE, DESCRIPTION, and ARGS...
989 ACTIVATE-FUNC is a function to call to activate this method.
990 TITLE is a string to show in the mode line when this method is active.
991 DESCRIPTION is a string describing this method and what it is good for.
992 The ARGS, if any, are passed as arguments to ACTIVATE-FUNC.
993 All told, the arguments to ACTIVATE-FUNC are INPUT-METHOD and the ARGS.
994
995 This function is mainly used in the file \"leim-list.el\" which is
996 created at Emacs build time, registering all Quail input methods
997 contained in the Emacs distribution.
998
999 In case you want to register a new Quail input method by yourself, be
1000 careful to use the same input method title as given in the third
1001 parameter of `quail-define-package'. (If the values are different, the
1002 string specified in this function takes precedence.)
1003
1004 The commands `describe-input-method' and `list-input-methods' need
1005 these duplicated values to show some information about input methods
1006 without loading the relevant Quail packages."
1007 (if (symbolp lang-env)
1008 (setq lang-env (symbol-name lang-env)))
1009 (if (symbolp input-method)
1010 (setq input-method (symbol-name input-method)))
1011 (let ((info (cons lang-env args))
1012 (slot (assoc input-method input-method-alist)))
1013 (if slot
1014 (setcdr slot info)
1015 (setq slot (cons input-method info))
1016 (setq input-method-alist (cons slot input-method-alist)))))
1017
1018 (defun read-input-method-name (prompt &optional default inhibit-null)
1019 "Read a name of input method from a minibuffer prompting with PROMPT.
1020 If DEFAULT is non-nil, use that as the default,
1021 and substitute it into PROMPT at the first `%s'.
1022 If INHIBIT-NULL is non-nil, null input signals an error.
1023
1024 The return value is a string."
1025 (if default
1026 (setq prompt (format prompt default)))
1027 (let* ((completion-ignore-case t)
1028 ;; As it is quite normal to change input method in the
1029 ;; minibuffer, we must enable it even if
1030 ;; enable-recursive-minibuffers is currently nil.
1031 (enable-recursive-minibuffers t)
1032 ;; This binding is necessary because input-method-history is
1033 ;; buffer local.
1034 (input-method (completing-read prompt input-method-alist
1035 nil t nil 'input-method-history
1036 default)))
1037 (if (and input-method (symbolp input-method))
1038 (setq input-method (symbol-name input-method)))
1039 (if (> (length input-method) 0)
1040 input-method
1041 (if inhibit-null
1042 (error "No valid input method is specified")))))
1043
1044 (defun activate-input-method (input-method)
1045 "Switch to input method INPUT-METHOD for the current buffer.
1046 If some other input method is already active, turn it off first.
1047 If INPUT-METHOD is nil, deactivate any current input method."
1048 (if (and input-method (symbolp input-method))
1049 (setq input-method (symbol-name input-method)))
1050 (if (and current-input-method
1051 (not (string= current-input-method input-method)))
1052 (inactivate-input-method))
1053 (unless (or current-input-method (null input-method))
1054 (let ((slot (assoc input-method input-method-alist)))
1055 (if (null slot)
1056 (error "Can't activate input method `%s'" input-method))
1057 (setq current-input-method-title nil)
1058 (let ((func (nth 2 slot)))
1059 (if (functionp func)
1060 (apply (nth 2 slot) input-method (nthcdr 5 slot))
1061 (if (and (consp func) (symbolp (car func)) (symbolp (cdr func)))
1062 (progn
1063 (require (cdr func))
1064 (apply (car func) input-method (nthcdr 5 slot)))
1065 (error "Can't activate input method `%s'" input-method))))
1066 (setq current-input-method input-method)
1067 (or (stringp current-input-method-title)
1068 (setq current-input-method-title (nth 3 slot)))
1069 (unwind-protect
1070 (run-hooks 'input-method-activate-hook)
1071 (force-mode-line-update)))))
1072
1073 (defun inactivate-input-method ()
1074 "Turn off the current input method."
1075 (when current-input-method
1076 (if input-method-history
1077 (unless (string= current-input-method (car input-method-history))
1078 (setq input-method-history
1079 (cons current-input-method
1080 (delete current-input-method input-method-history))))
1081 (setq input-method-history (list current-input-method)))
1082 (unwind-protect
1083 (funcall inactivate-current-input-method-function)
1084 (unwind-protect
1085 (run-hooks 'input-method-inactivate-hook)
1086 (setq current-input-method nil
1087 input-method-function nil
1088 current-input-method-title nil)
1089 (force-mode-line-update)))))
1090
1091 (defun set-input-method (input-method)
1092 "Select and activate input method INPUT-METHOD for the current buffer.
1093 This also sets the default input method to the one you specify.
1094 If INPUT-METHOD is nil, this function turns off the input method, and
1095 also causes you to be prompted for a name of an input method the next
1096 time you invoke \\[toggle-input-method].
1097
1098 To deactivate the input method interactively, use \\[toggle-input-method].
1099 To deactivate it programmatically, use \\[inactivate-input-method]."
1100 (interactive
1101 (let* ((default (or (car input-method-history) default-input-method)))
1102 (list (read-input-method-name
1103 (if default "Select input method (default %s): " "Select input method: ")
1104 default t))))
1105 (activate-input-method input-method)
1106 (setq default-input-method input-method))
1107
1108 (defun toggle-input-method (&optional arg)
1109 "Enable or disable multilingual text input method for the current buffer.
1110 Only one input method can be enabled at any time in a given buffer.
1111
1112 The normal action is to enable an input method if none was
1113 enabled, and disable the current one otherwise. Which input method
1114 to enable can be determined in various ways--either the one most
1115 recently used, or the one specified by `default-input-method', or
1116 as a last resort by reading the name of an input method in the
1117 minibuffer.
1118
1119 With a prefix argument, read an input method name with the minibuffer
1120 and enable that one. The default is the most recent input method specified
1121 \(not including the currently active input method, if any)."
1122
1123 (interactive "P")
1124 (if (and current-input-method (not arg))
1125 (inactivate-input-method)
1126 (let ((default (or (car input-method-history) default-input-method)))
1127 (if (and arg default (equal current-input-method default)
1128 (> (length input-method-history) 1))
1129 (setq default (nth 1 input-method-history)))
1130 (activate-input-method
1131 (if (or arg (not default))
1132 (progn
1133 (read-input-method-name
1134 (if default "Input method (default %s): " "Input method: " )
1135 default t))
1136 default))
1137 (or default-input-method
1138 (setq default-input-method current-input-method)))))
1139
1140 (defun describe-input-method (input-method)
1141 "Describe input method INPUT-METHOD."
1142 (interactive
1143 (list (read-input-method-name
1144 "Describe input method (default, current choice): ")))
1145 (if (and input-method (symbolp input-method))
1146 (setq input-method (symbol-name input-method)))
1147 (if (null input-method)
1148 (describe-current-input-method)
1149 (let ((current current-input-method))
1150 (condition-case nil
1151 (progn
1152 (save-excursion
1153 (activate-input-method input-method)
1154 (describe-current-input-method))
1155 (activate-input-method current))
1156 (error
1157 (activate-input-method current)
1158 (help-setup-xref (list #'describe-input-method input-method)
1159 (interactive-p))
1160 (with-output-to-temp-buffer (help-buffer)
1161 (let ((elt (assoc input-method input-method-alist)))
1162 (princ (format
1163 "Input method: %s (`%s' in mode line) for %s\n %s\n"
1164 input-method (nth 3 elt) (nth 1 elt) (nth 4 elt))))))))))
1165
1166 (defun describe-current-input-method ()
1167 "Describe the input method currently in use."
1168 (if current-input-method
1169 (if (and (symbolp describe-current-input-method-function)
1170 (fboundp describe-current-input-method-function))
1171 (funcall describe-current-input-method-function)
1172 (message "No way to describe the current input method `%s'"
1173 current-input-method)
1174 (ding))
1175 (error "No input method is activated now")))
1176
1177 (defun read-multilingual-string (prompt &optional initial-input input-method)
1178 "Read a multilingual string from minibuffer, prompting with string PROMPT.
1179 The input method selected last time is activated in minibuffer.
1180 If optional second arg INITIAL-INPUT is non-nil, insert it in the minibuffer
1181 initially.
1182 Optional 3rd argument INPUT-METHOD specifies the input method
1183 to be activated instead of the one selected last time. It is a symbol
1184 or a string."
1185 (setq input-method
1186 (or input-method
1187 current-input-method
1188 default-input-method
1189 (read-input-method-name "Input method: " nil t)))
1190 (if (and input-method (symbolp input-method))
1191 (setq input-method (symbol-name input-method)))
1192 (let ((prev-input-method current-input-method))
1193 (unwind-protect
1194 (progn
1195 (activate-input-method input-method)
1196 (read-string prompt initial-input nil nil t))
1197 (activate-input-method prev-input-method))))
1198
1199 ;; Variables to control behavior of input methods. All input methods
1200 ;; should react to these variables.
1201
1202 (defcustom input-method-verbose-flag 'default
1203 "*A flag to control extra guidance given by input methods.
1204 The value should be nil, t, `complex-only', or `default'.
1205
1206 The extra guidance is done by showing list of available keys in echo
1207 area. When you use the input method in the minibuffer, the guidance
1208 is shown at the bottom short window (split from the existing window).
1209
1210 If the value is t, extra guidance is always given, if the value is
1211 nil, extra guidance is always suppressed.
1212
1213 If the value is `complex-only', only complex input methods such as
1214 `chinese-py' and `japanese' give extra guidance.
1215
1216 If the value is `default', complex input methods always give extra
1217 guidance, but simple input methods give it only when you are not in
1218 the minibuffer.
1219
1220 See also the variable `input-method-highlight-flag'."
1221 :type '(choice (const t) (const nil) (const complex-only) (const default))
1222 :group 'mule)
1223
1224 (defcustom input-method-highlight-flag t
1225 "*If this flag is non-nil, input methods highlight partially-entered text.
1226 For instance, while you are in the middle of a Quail input method sequence,
1227 the text inserted so far is temporarily underlined.
1228 The underlining goes away when you finish or abort the input method sequence.
1229 See also the variable `input-method-verbose-flag'."
1230 :type 'boolean
1231 :group 'mule)
1232
1233 (defvar input-method-activate-hook nil
1234 "Normal hook run just after an input method is activated.
1235
1236 The variable `current-input-method' keeps the input method name
1237 just activated.")
1238
1239 (defvar input-method-inactivate-hook nil
1240 "Normal hook run just after an input method is inactivated.
1241
1242 The variable `current-input-method' still keeps the input method name
1243 just inactivated.")
1244
1245 (defvar input-method-after-insert-chunk-hook nil
1246 "Normal hook run just after an input method insert some chunk of text.")
1247
1248 (defvar input-method-exit-on-first-char nil
1249 "This flag controls when an input method returns.
1250 Usually, the input method does not return while there's a possibility
1251 that it may find a different translation if a user types another key.
1252 But, it this flag is non-nil, the input method returns as soon as
1253 the current key sequence gets long enough to have some valid translation.")
1254
1255 (defvar input-method-use-echo-area nil
1256 "This flag controls how an input method shows an intermediate key sequence.
1257 Usually, the input method inserts the intermediate key sequence,
1258 or candidate translations corresponding to the sequence,
1259 at point in the current buffer.
1260 But, if this flag is non-nil, it displays them in echo area instead.")
1261
1262 (defvar input-method-exit-on-invalid-key nil
1263 "This flag controls the behaviour of an input method on invalid key input.
1264 Usually, when a user types a key which doesn't start any character
1265 handled by the input method, the key is handled by turning off the
1266 input method temporarily. After that key, the input method is re-enabled.
1267 But, if this flag is non-nil, the input method is never back on.")
1268
1269 \f
1270 (defvar set-language-environment-hook nil
1271 "Normal hook run after some language environment is set.
1272
1273 When you set some hook function here, that effect usually should not
1274 be inherited to another language environment. So, you had better set
1275 another function in `exit-language-environment-hook' (which see) to
1276 cancel the effect.")
1277
1278 (defvar exit-language-environment-hook nil
1279 "Normal hook run after exiting from some language environment.
1280 When this hook is run, the variable `current-language-environment'
1281 is still bound to the language environment being exited.
1282
1283 This hook is mainly used for canceling the effect of
1284 `set-language-environment-hook' (which-see).")
1285
1286 (put 'setup-specified-language-environment 'apropos-inhibit t)
1287
1288 (defun setup-specified-language-environment ()
1289 "Switch to a specified language environment."
1290 (interactive)
1291 (let (language-name)
1292 (if (and (symbolp last-command-event)
1293 (or (not (eq last-command-event 'Default))
1294 (setq last-command-event 'English))
1295 (setq language-name (symbol-name last-command-event)))
1296 (set-language-environment language-name)
1297 (error "Bogus calling sequence"))))
1298
1299 (defcustom current-language-environment "English"
1300 "The last language environment specified with `set-language-environment'.
1301 This variable should be set only with \\[customize], which is equivalent
1302 to using the function `set-language-environment'."
1303 :link '(custom-manual "(emacs)Language Environments")
1304 :set (lambda (symbol value) (set-language-environment value))
1305 :get (lambda (x)
1306 (or (car-safe (assoc-ignore-case
1307 (if (symbolp current-language-environment)
1308 (symbol-name current-language-environment)
1309 current-language-environment)
1310 language-info-alist))
1311 "English"))
1312 :type (cons 'choice (mapcar (lambda (lang)
1313 (list 'const (car lang)))
1314 language-info-alist))
1315 :initialize 'custom-initialize-default
1316 :group 'mule
1317 :type 'string)
1318
1319 (defun reset-language-environment ()
1320 "Reset multilingual environment of Emacs to the default status.
1321
1322 The default status is as follows:
1323
1324 The default value of buffer-file-coding-system is nil.
1325 The default coding system for process I/O is nil.
1326 The default value for the command `set-terminal-coding-system' is nil.
1327 The default value for the command `set-keyboard-coding-system' is nil.
1328
1329 The order of priorities of coding categories and the coding system
1330 bound to each category are as follows
1331 coding category coding system
1332 --------------------------------------------------
1333 coding-category-iso-8-2 iso-latin-1
1334 coding-category-iso-8-1 iso-latin-1
1335 coding-category-iso-7-tight iso-2022-jp
1336 coding-category-iso-7 iso-2022-7bit
1337 coding-category-iso-7-else iso-2022-7bit-lock
1338 coding-category-iso-8-else iso-2022-8bit-ss2
1339 coding-category-emacs-mule emacs-mule
1340 coding-category-raw-text raw-text
1341 coding-category-sjis japanese-shift-jis
1342 coding-category-big5 chinese-big5
1343 coding-category-ccl nil
1344 coding-category-binary no-conversion
1345 coding-category-utf-16-be nil
1346 coding-category-utf-16-le nil
1347 coding-category-utf-8 mule-utf-8"
1348 (interactive)
1349 ;; This function formerly set default-enable-multibyte-characters to t,
1350 ;; but that is incorrect. It should not alter the unibyte/multibyte choice.
1351
1352 (setq coding-category-iso-7-tight 'iso-2022-jp
1353 coding-category-iso-7 'iso-2022-7bit
1354 coding-category-iso-8-1 'iso-latin-1
1355 coding-category-iso-8-2 'iso-latin-1
1356 coding-category-iso-7-else 'iso-2022-7bit-lock
1357 coding-category-iso-8-else 'iso-2022-8bit-ss2
1358 coding-category-emacs-mule 'emacs-mule
1359 coding-category-raw-text 'raw-text
1360 coding-category-sjis 'japanese-shift-jis
1361 coding-category-big5 'chinese-big5
1362 coding-category-utf-16-be nil
1363 coding-category-utf-16-le nil
1364 coding-category-utf-8 'mule-utf-8
1365 coding-category-ccl nil
1366 coding-category-binary 'no-conversion)
1367
1368 (set-coding-priority
1369 '(coding-category-iso-8-1
1370 coding-category-iso-8-2
1371 coding-category-iso-7-tight
1372 coding-category-iso-7
1373 coding-category-iso-7-else
1374 coding-category-iso-8-else
1375 coding-category-emacs-mule
1376 coding-category-raw-text
1377 coding-category-sjis
1378 coding-category-big5
1379 coding-category-ccl
1380 coding-category-binary
1381 coding-category-utf-16-be
1382 coding-category-utf-16-le
1383 coding-category-utf-8))
1384
1385 (update-coding-systems-internal)
1386
1387 (set-default-coding-systems nil)
1388 (setq default-sendmail-coding-system 'iso-latin-1)
1389 (setq default-process-coding-system '(undecided . iso-latin-1))
1390
1391 ;; Don't alter the terminal and keyboard coding systems here.
1392 ;; The terminal still supports the same coding system
1393 ;; that it supported a minute ago.
1394 ;;; (set-terminal-coding-system-internal nil)
1395 ;;; (set-keyboard-coding-system-internal nil)
1396
1397 (setq nonascii-translation-table nil
1398 nonascii-insert-offset 0))
1399
1400 (reset-language-environment)
1401
1402 (defun set-display-table-and-terminal-coding-system (language-name)
1403 "Set up the display table and terminal coding system for LANGUAGE-NAME."
1404 (let ((coding (get-language-info language-name 'unibyte-display)))
1405 (if coding
1406 (standard-display-european-internal)
1407 (standard-display-default (if (eq window-system 'pc) 128 160) 255)
1408 (aset standard-display-table 146 nil))
1409 (or (eq window-system 'pc)
1410 (set-terminal-coding-system coding))))
1411
1412 (defun set-language-environment (language-name)
1413 "Set up multi-lingual environment for using LANGUAGE-NAME.
1414 This sets the coding system priority and the default input method
1415 and sometimes other things. LANGUAGE-NAME should be a string
1416 which is the name of a language environment. For example, \"Latin-1\"
1417 specifies the character set for the major languages of Western Europe."
1418 (interactive (list (read-language-name
1419 nil
1420 "Set language environment (default, English): ")))
1421 (if language-name
1422 (if (symbolp language-name)
1423 (setq language-name (symbol-name language-name)))
1424 (setq language-name "English"))
1425 (or (assoc-ignore-case language-name language-info-alist)
1426 (error "Language environment not defined: %S" language-name))
1427 (if current-language-environment
1428 (let ((func (get-language-info current-language-environment
1429 'exit-function)))
1430 (run-hooks 'exit-language-environment-hook)
1431 (if (functionp func) (funcall func))))
1432 (let ((default-eol-type (coding-system-eol-type
1433 default-buffer-file-coding-system)))
1434 (reset-language-environment)
1435
1436 (setq current-language-environment language-name)
1437 (set-language-environment-coding-systems language-name default-eol-type))
1438 (let ((input-method (get-language-info language-name 'input-method)))
1439 (when input-method
1440 (setq default-input-method input-method)
1441 (if input-method-history
1442 (setq input-method-history
1443 (cons input-method
1444 (delete input-method input-method-history))))))
1445 (let ((nonascii (get-language-info language-name 'nonascii-translation))
1446 (dos-table
1447 (if (eq window-system 'pc)
1448 (intern
1449 (format "cp%d-nonascii-translation-table" dos-codepage)))))
1450 (cond
1451 ((char-table-p nonascii)
1452 (setq nonascii-translation-table nonascii))
1453 ((and (eq window-system 'pc) (boundp dos-table))
1454 ;; DOS terminals' default is to use a special non-ASCII translation
1455 ;; table as appropriate for the installed codepage.
1456 (setq nonascii-translation-table (symbol-value dos-table)))
1457 ((charsetp nonascii)
1458 (setq nonascii-insert-offset (- (make-char nonascii) 128)))))
1459
1460 ;; Unibyte setups if necessary.
1461 (unless default-enable-multibyte-characters
1462 ;; Syntax and case table.
1463 (let ((syntax (get-language-info language-name 'unibyte-syntax)))
1464 (if syntax
1465 (let ((set-case-syntax-set-multibyte nil))
1466 (load syntax nil t))
1467 ;; No information for syntax and case. Reset to the defaults.
1468 (let ((syntax-table (standard-syntax-table))
1469 (case-table (standard-case-table))
1470 (ch (if (eq window-system 'pc) 128 160)))
1471 (while (< ch 256)
1472 (modify-syntax-entry ch " " syntax-table)
1473 (aset case-table ch ch)
1474 (setq ch (1+ ch)))
1475 (set-char-table-extra-slot case-table 0 nil)
1476 (set-char-table-extra-slot case-table 1 nil)
1477 (set-char-table-extra-slot case-table 2 nil))
1478 (set-standard-case-table (standard-case-table))
1479 (let ((list (buffer-list)))
1480 (while list
1481 (with-current-buffer (car list)
1482 (set-case-table (standard-case-table)))
1483 (setq list (cdr list))))))
1484 (set-display-table-and-terminal-coding-system language-name))
1485
1486 (let ((required-features (get-language-info language-name 'features)))
1487 (while required-features
1488 (require (car required-features))
1489 (setq required-features (cdr required-features))))
1490 (let ((func (get-language-info language-name 'setup-function)))
1491 (if (functionp func)
1492 (funcall func)))
1493 (run-hooks 'set-language-environment-hook)
1494 (force-mode-line-update t))
1495
1496 (defun standard-display-european-internal ()
1497 ;; Actually set up direct output of non-ASCII characters.
1498 (standard-display-8bit (if (eq window-system 'pc) 128 160) 255)
1499 ;; Unibyte Emacs on MS-DOS wants to display all 8-bit characters with
1500 ;; the native font, and codes 160 and 146 stand for something very
1501 ;; different there.
1502 (or (and (eq window-system 'pc) (not default-enable-multibyte-characters))
1503 (progn
1504 ;; Make non-line-break space display as a plain space.
1505 ;; Most X fonts do the wrong thing for code 160.
1506 (aset standard-display-table 160 [32])
1507 ;; With luck, non-Latin-1 fonts are more recent and so don't
1508 ;; have this bug.
1509 (aset standard-display-table 2208 [32]) ; Latin-1 NBSP
1510 ;; Most Windows programs send out apostrophes as \222. Most X fonts
1511 ;; don't contain a character at that position. Map it to the ASCII
1512 ;; apostrophe. [This is actually RIGHT SINGLE QUOTATION MARK,
1513 ;; U+2019, normally from the windows-1252 character set. XFree 4
1514 ;; fonts probably have the appropriate glyph at this position,
1515 ;; so they could use standard-display-8bit. It's better to use a
1516 ;; proper windows-1252 coding system. --fx]
1517 (aset standard-display-table 146 [39]))))
1518
1519 (defun set-language-environment-coding-systems (language-name
1520 &optional eol-type)
1521 "Do various coding system setups for language environment LANGUAGE-NAME.
1522
1523 The optional arg EOL-TYPE specifies the eol-type of the default value
1524 of buffer-file-coding-system set by this function."
1525 (let* ((priority (get-language-info language-name 'coding-priority))
1526 (default-coding (car priority)))
1527 (if priority
1528 (let ((categories (mapcar 'coding-system-category priority)))
1529 (set-default-coding-systems
1530 (if (memq eol-type '(0 1 2 unix dos mac))
1531 (coding-system-change-eol-conversion default-coding eol-type)
1532 default-coding))
1533 (setq default-sendmail-coding-system default-coding)
1534 (set-coding-priority categories)
1535 (while priority
1536 (set (car categories) (car priority))
1537 (setq priority (cdr priority) categories (cdr categories)))
1538 (update-coding-systems-internal)))))
1539
1540 ;; Print all arguments with `princ', then print "\n".
1541 (defsubst princ-list (&rest args)
1542 (while args (princ (car args)) (setq args (cdr args)))
1543 (princ "\n"))
1544
1545 (put 'describe-specified-language-support 'apropos-inhibit t)
1546
1547 ;; Print a language specific information such as input methods,
1548 ;; charsets, and coding systems. This function is intended to be
1549 ;; called from the menu:
1550 ;; [menu-bar mule describe-language-environment LANGUAGE]
1551 ;; and should not run it by `M-x describe-current-input-method-function'.
1552 (defun describe-specified-language-support ()
1553 "Describe how Emacs supports the specified language environment."
1554 (interactive)
1555 (let (language-name)
1556 (if (not (and (symbolp last-command-event)
1557 (or (not (eq last-command-event 'Default))
1558 (setq last-command-event 'English))
1559 (setq language-name (symbol-name last-command-event))))
1560 (error "Bogus calling sequence"))
1561 (describe-language-environment language-name)))
1562
1563 (defun describe-language-environment (language-name)
1564 "Describe how Emacs supports language environment LANGUAGE-NAME."
1565 (interactive
1566 (list (read-language-name
1567 'documentation
1568 "Describe language environment (default, current choice): ")))
1569 (if (null language-name)
1570 (setq language-name current-language-environment))
1571 (if (or (null language-name)
1572 (null (get-language-info language-name 'documentation)))
1573 (error "No documentation for the specified language"))
1574 (if (symbolp language-name)
1575 (setq language-name (symbol-name language-name)))
1576 (let ((doc (get-language-info language-name 'documentation))
1577 pos)
1578 (help-setup-xref (list #'describe-language-environment language-name)
1579 (interactive-p))
1580 (with-output-to-temp-buffer (help-buffer)
1581 (save-excursion
1582 (set-buffer standard-output)
1583 (insert language-name " language environment\n\n")
1584 (if (stringp doc)
1585 (insert doc "\n\n"))
1586 (condition-case nil
1587 (let ((str (eval (get-language-info language-name 'sample-text))))
1588 (if (stringp str)
1589 (insert "Sample text:\n " str "\n\n")))
1590 (error nil))
1591 (let ((input-method (get-language-info language-name 'input-method))
1592 (l (copy-sequence input-method-alist)))
1593 (insert "Input methods")
1594 (when input-method
1595 (insert " (default, " input-method ")")
1596 (setq input-method (assoc input-method input-method-alist))
1597 (setq l (cons input-method (delete input-method l))))
1598 (insert ":\n")
1599 (while l
1600 (when (string= language-name (nth 1 (car l)))
1601 (insert " " (car (car l)))
1602 (search-backward (car (car l)))
1603 (help-xref-button 0 #'describe-input-method (car (car l))
1604 "mouse-2, RET: describe this input method")
1605 (goto-char (point-max))
1606 (insert " (\""
1607 (if (stringp (nth 3 (car l)))
1608 (nth 3 (car l))
1609 (car (nth 3 (car l))))
1610 "\" in mode line)\n"))
1611 (setq l (cdr l)))
1612 (insert "\n"))
1613 (insert "Character sets:\n")
1614 (let ((l (get-language-info language-name 'charset)))
1615 (if (null l)
1616 (insert " nothing specific to " language-name "\n")
1617 (while l
1618 (insert " " (symbol-name (car l)))
1619 (search-backward (symbol-name (car l)))
1620 (help-xref-button 0 #'describe-character-set (car l)
1621 "mouse-2, RET: describe this character set")
1622 (goto-char (point-max))
1623 (insert ": " (charset-description (car l)) "\n")
1624 (setq l (cdr l)))))
1625 (insert "\n")
1626 (insert "Coding systems:\n")
1627 (let ((l (get-language-info language-name 'coding-system)))
1628 (if (null l)
1629 (insert " nothing specific to " language-name "\n")
1630 (while l
1631 (insert " " (symbol-name (car l)))
1632 (search-backward (symbol-name (car l)))
1633 (help-xref-button 0 #'describe-coding-system (car l)
1634 "mouse-2, RET: describe this coding system")
1635 (goto-char (point-max))
1636 (insert " (`"
1637 (coding-system-mnemonic (car l))
1638 "' in mode line):\n\t"
1639 (coding-system-doc-string (car l))
1640 "\n")
1641 (let ((aliases (coding-system-get (car l)
1642 'alias-coding-systems)))
1643 (when aliases
1644 (insert "\t(alias:")
1645 (while aliases
1646 (insert " " (symbol-name (car aliases)))
1647 (setq aliases (cdr aliases)))
1648 (insert ")\n")))
1649 (setq l (cdr l)))))))))
1650 \f
1651 ;;; Locales.
1652
1653 (defvar locale-translation-file-name nil
1654 "File name for the system's file of locale-name aliases, or nil if none.")
1655
1656 ;; The following definitions might as well be marked as constants and
1657 ;; purecopied, since they're normally used on startup, and probably
1658 ;; should reflect the facilities of the base Emacs.
1659 (defconst locale-language-names
1660 (purecopy
1661 '(
1662 ;; UTF-8 is not yet implemented.
1663 ;; Put this first, so that e.g. "ko.UTF-8" does not match "ko" below.
1664 (".*[._]utf" . nil)
1665
1666 ;; Locale names of the form LANGUAGE[_TERRITORY][.CODESET][@MODIFIER]
1667 ;; as specified in the Single Unix Spec, Version 2.
1668 ;; LANGUAGE is a language code taken from ISO 639:1988 (E/F)
1669 ;; with additions from ISO 639/RA Newsletter No.1/1989;
1670 ;; see Internet RFC 2165 (1997-06) and
1671 ;; http://www.evertype.com/standards/iso639/iso639-en.html
1672 ;; TERRITORY is a country code taken from ISO 3166
1673 ;; http://www.din.de/gremien/nas/nabd/iso3166ma/codlstp1/en_listp1.html.
1674 ;; CODESET and MODIFIER are implementation-dependent.
1675
1676 ; aa Afar
1677 ; ab Abkhazian
1678 ("af" . "Latin-1") ; Afrikaans
1679 ("am" . "Ethiopic") ; Amharic
1680 ; ar Arabic glibc uses 8859-6
1681 ; as Assamese
1682 ; ay Aymara
1683 ; az Azerbaijani
1684 ; ba Bashkir
1685 ("be" . "Latin-5") ; Byelorussian
1686 ("bg" . "Latin-5") ; Bulgarian
1687 ; bh Bihari
1688 ; bi Bislama
1689 ; bn Bengali, Bangla
1690 ("bo" . "Tibetan")
1691 ("br" . "Latin-1") ; Breton
1692 ("ca" . "Latin-1") ; Catalan
1693 ; co Corsican
1694 ("cs" . "Czech")
1695 ("cy" . "Latin-8") ; Welsh
1696 ("da" . "Latin-1") ; Danish
1697 ("de" . "German")
1698 ; dz Bhutani
1699 ("el" . "Greek")
1700 ;; Users who specify "en" explicitly typically want Latin-1, not ASCII.
1701 ("en" . "Latin-1") ; English
1702 ("eo" . "Latin-3") ; Esperanto
1703 ("es" . "Spanish")
1704 ("et" . "Latin-4") ; Estonian
1705 ("eu" . "Latin-1") ; Basque
1706 ; fa Persian
1707 ("fi" . "Latin-1") ; Finnish
1708 ; fj Fiji
1709 ("fo" . "Latin-1") ; Faroese
1710 ("fr" . "French") ; French
1711 ("fy" . "Latin-1") ; Frisian
1712 ("ga" . "Latin-1") ; Irish Gaelic (new orthography)
1713 ("gd" . "Latin-1") ; Scots Gaelic
1714 ("gl" . "Latin-1") ; Galician
1715 ; gn Guarani
1716 ; gu Gujarati
1717 ("gv" . "Latin-8") ; Manx Gaelic
1718 ; ha Hausa
1719 ("he" . "Hebrew")
1720 ("hi" . "Devanagari") ; Hindi glibc uses utf-8
1721 ("hr" . "Latin-2") ; Croatian
1722 ("hu" . "Latin-2") ; Hungarian
1723 ; hy Armenian
1724 ; ia Interlingua
1725 ("id" . "Latin-1") ; Indonesian
1726 ; ie Interlingue
1727 ; ik Inupiak
1728 ("is" . "Latin-1") ; Icelandic
1729 ("it" . "Latin-1") ; Italian
1730 ; iu Inuktitut
1731 ("ja" . "Japanese")
1732 ; jw Javanese
1733 ; ka Georgian
1734 ; kk Kazakh
1735 ("kl" . "Latin-1") ; Greenlandic
1736 ; km Cambodian
1737 ; kn Kannada
1738 ("ko" . "Korean")
1739 ; ks Kashmiri
1740 ; ku Kurdish
1741 ("kw" . "Latin-1") ; Cornish
1742 ; ky Kirghiz
1743 ("la" . "Latin-1") ; Latin
1744 ("lb" . "Latin-1") ; Luxemburgish
1745 ; ln Lingala
1746 ("lo" . "Lao") ; Laothian
1747 ("lt" . "Latin-4") ; Lithuanian
1748 ("lv" . "Latin-4") ; Latvian, Lettish
1749 ; mg Malagasy
1750 ("mi" . "Latin-7") ; Maori
1751 ("mk" . "Latin-5") ; Macedonian
1752 ; ml Malayalam
1753 ; mn Mongolian
1754 ; mo Moldavian
1755 ("mr" . "Devanagari") ; Marathi glibc uses utf-8
1756 ("ms" . "Latin-1") ; Malay
1757 ("mt" . "Latin-3") ; Maltese
1758 ; my Burmese
1759 ; na Nauru
1760 ("ne" . "Devanagari") ; Nepali
1761 ("nl" . "Dutch")
1762 ("no" . "Latin-1") ; Norwegian
1763 ("oc" . "Latin-1") ; Occitan
1764 ; om (Afan) Oromo
1765 ; or Oriya
1766 ; pa Punjabi
1767 ("pl" . "Latin-2") ; Polish
1768 ; ps Pashto, Pushto
1769 ("pt" . "Latin-1") ; Portuguese
1770 ; qu Quechua
1771 ("rm" . "Latin-1") ; Rhaeto-Romanic
1772 ; rn Kirundi
1773 ("ro" . "Romanian")
1774 ("ru.*[_.]koi8" . "Cyrillic-KOI8") ; Russian
1775 ("ru" . "Latin-5") ; Russian
1776 ; rw Kinyarwanda
1777 ("sa" . "Devanagari") ; Sanskrit
1778 ; sd Sindhi
1779 ; se Northern Sami
1780 ; sg Sangho
1781 ("sh" . "Latin-2") ; Serbo-Croatian
1782 ; si Sinhalese
1783 ("sk" . "Slovak")
1784 ("sl" . "Slovenian")
1785 ; sm Samoan
1786 ; sn Shona
1787 ; so Somali
1788 ("sq" . "Latin-1") ; Albanian
1789 ("sr" . "Latin-2") ; Serbian (Latin alphabet)
1790 ; ss Siswati
1791 ; st Sesotho
1792 ; su Sundanese
1793 ("sv" . "Latin-1") ; Swedish
1794 ("sw" . "Latin-1") ; Swahili
1795 ; ta Tamil glibc uses utf-8
1796 ; te Telugu glibc uses utf-8
1797 ; tg Tajik "Cyrillic-KOI8-T"
1798 ("th" . "Thai")
1799 ; ti Tigrinya
1800 ; tk Turkmen
1801 ("tl" . "Latin-1") ; Tagalog
1802 ; tn Setswana
1803 ; to Tonga
1804 ("tr" . "Latin-5") ; Turkish
1805 ; ts Tsonga
1806 ; tt Tatar
1807 ; tw Twi
1808 ; ug Uighur
1809 ("uk" . "Latin-5") ; Ukrainian
1810 ; ur Urdu glibc uses utf-8
1811 ("uz" . "Latin-1") ; Uzbek
1812 ("vi" . "Vietnamese") ; glibc uses utf-8
1813 ; vo Volapuk
1814 ; wo Wolof
1815 ; xh Xhosa
1816 ; yi Yiddish
1817 ; yo Yoruba
1818 ; za Zhuang
1819
1820 ; glibc:
1821 ; zh_CN.GB18030/GB18030 \
1822 ; zh_CN.GBK/GBK \
1823 ; zh_HK/BIG5-HKSCS \
1824 ; zh_TW/BIG5 \
1825 ; zh_TW.EUC-TW/EUC-TW \
1826
1827 ("zh.*[._]big5" . "Chinese-BIG5")
1828 ("zh.*[._]gbk" . nil) ; Solaris 2.7; has gbk-0 as well as GB 2312.1980-0
1829 ("zh_tw" . "Chinese-CNS")
1830 ("zh" . "Chinese-GB")
1831 ; zu Zulu
1832
1833 ;; ISO standard locales
1834 ("c$" . "ASCII")
1835 ("posix$" . "ASCII")
1836
1837 ;; The "IPA" Emacs language environment does not correspond
1838 ;; to any ISO 639 code, so let it stand for itself.
1839 ("ipa$" . "IPA")
1840
1841 ;; Nonstandard or obsolete language codes
1842 ("cz" . "Czech") ; e.g. Solaris 2.6
1843 ("ee" . "Latin-4") ; Estonian, e.g. X11R6.4
1844 ("iw" . "Hebrew") ; e.g. X11R6.4
1845 ("sp" . "Latin-5") ; Serbian (Cyrillic alphabet), e.g. X11R6.4
1846 ("su" . "Latin-1") ; Finnish, e.g. Solaris 2.6
1847 ("jp" . "Japanese") ; e.g. MS Windows
1848 ("chs" . "Chinese-GB") ; MS Windows Chinese Simplified
1849 ("cht" . "Chinese-BIG5") ; MS Windows Chinese Traditional
1850 ))
1851 "List of pairs of locale regexps and language names.
1852 The first element whose locale regexp matches the start of a downcased locale
1853 specifies the language name corresponding to that locale.
1854 If the language name is nil, there is no corresponding language environment.")
1855
1856 (defconst locale-charset-language-names
1857 (purecopy
1858 '((".*8859[-_]?1\\>" . "Latin-1")
1859 (".*8859[-_]?2\\>" . "Latin-2")
1860 (".*8859[-_]?3\\>" . "Latin-3")
1861 (".*8859[-_]?4\\>" . "Latin-4")
1862 (".*8859[-_]?9\\>" . "Latin-5")
1863 (".*8859[-_]?14\\>" . "Latin-8")
1864 (".*8859[-_]?15\\>" . "Latin-9")
1865 (".*@euro\\>" . "Latin-9")
1866 ))
1867 "List of pairs of locale regexps and charset language names.
1868 The first element whose locale regexp matches the start of a downcased locale
1869 specifies the language name whose charsets corresponds to that locale.
1870 This language name is used if its charsets disagree with the charsets of
1871 the language name that would otherwise be used for this locale.")
1872
1873 (defconst locale-preferred-coding-systems
1874 (purecopy
1875 '(("ja.*[._]euc" . japanese-iso-8bit)
1876 ("ja.*[._]jis7" . iso-2022-jp)
1877 ("ja.*[._]pck" . japanese-shift-jis)
1878 ("ja.*[._]sjis" . japanese-shift-jis)
1879 ))
1880 "List of pairs of locale regexps and preferred coding systems.
1881 The first element whose locale regexp matches the start of a downcased locale
1882 specifies the coding system to prefer when using that locale.")
1883
1884 (defconst standard-keyboard-coding-systems
1885 (purecopy
1886 '(iso-latin-1 iso-latin-2 iso-latin-3 iso-latin-4 iso-latin-5
1887 iso-latin-6 iso-latin-7 iso-latin-8 iso-latin-9))
1888 "Coding systems that are commonly used for keyboards.
1889 `set-locale-environment' will set the `keyboard-coding-system' if the
1890 coding-system specified by the locale setting is a member of this list.")
1891
1892 (defun locale-name-match (key alist)
1893 "Search for KEY in ALIST, which should be a list of regexp-value pairs.
1894 Return the value corresponding to the first regexp that matches the
1895 start of KEY, or nil if there is no match."
1896 (let (element)
1897 (while (and alist (not element))
1898 (if (string-match (concat "\\`\\(?:" (car (car alist)) "\\)") key)
1899 (setq element (car alist)))
1900 (setq alist (cdr alist)))
1901 (cdr element)))
1902
1903 (defun set-locale-environment (&optional locale-name)
1904 "Set up multi-lingual environment for using LOCALE-NAME.
1905 This sets the language environment, the coding system priority,
1906 the default input method and sometimes other things.
1907
1908 LOCALE-NAME should be a string
1909 which is the name of a locale supported by the system;
1910 often it is of the form xx_XX.CODE, where xx is a language,
1911 XX is a country, and CODE specifies a character set and coding system.
1912 For example, the locale name \"ja_JP.EUC\" might name a locale
1913 for Japanese in Japan using the `japanese-iso-8bit' coding-system.
1914
1915 If LOCALE-NAME is nil, its value is taken from the environment
1916 variables LC_ALL, LC_CTYLE and LANG (the first one that is set).
1917
1918 The locale names supported by your system can typically be found in a
1919 directory named `/usr/share/locale' or `/usr/lib/locale'. LOCALE-NAME
1920 will be translated according to the table specified by
1921 `locale-translation-file-name'.
1922
1923 See also `locale-charset-language-names', `locale-language-names',
1924 `locale-preferred-coding-systems' and `locale-coding-system'."
1925 (interactive "sSet environment for locale: ")
1926
1927 ;; Do this at runtime for the sake of binaries possibly transported
1928 ;; to a system without X.
1929 (setq locale-translation-file-name
1930 (let ((files
1931 '("/usr/lib/X11/locale/locale.alias" ; e.g. X11R6.4
1932 "/usr/X11R6/lib/X11/locale/locale.alias" ; e.g. RedHat 4.2
1933 "/usr/openwin/lib/locale/locale.alias" ; e.g. Solaris 2.6
1934 ;;
1935 ;; The following name appears after the X-related names above,
1936 ;; since the X-related names are what X actually uses.
1937 "/usr/share/locale/locale.alias" ; GNU/Linux sans X
1938 )))
1939 (while (and files (not (file-exists-p (car files))))
1940 (setq files (cdr files)))
1941 (car files)))
1942
1943 (let ((locale locale-name))
1944
1945 (unless locale
1946 ;; Use the first of these three environment variables
1947 ;; that has a nonempty value.
1948 (let ((vars '("LC_ALL" "LC_CTYPE" "LANG")))
1949 (while (and vars (not (setq locale (getenv (car vars)))))
1950 (setq vars (cdr vars)))))
1951
1952 (when locale
1953
1954 ;; Translate "swedish" into "sv_SE.ISO8859-1", and so on,
1955 ;; using the translation file that many systems have.
1956 (when locale-translation-file-name
1957 (with-temp-buffer
1958 (insert-file-contents locale-translation-file-name)
1959 (when (re-search-forward
1960 (concat "^" (regexp-quote locale) ":?[ \t]+") nil t)
1961 (setq locale (buffer-substring (point) (line-end-position))))))
1962
1963 ;; Leave the system locales alone if the caller did not specify
1964 ;; an explicit locale name, as their defaults are set from
1965 ;; LC_MESSAGES and LC_TIME, not LC_CTYPE, and the user might not
1966 ;; want to set them to the same value as LC_CTYPE.
1967 (when locale-name
1968 (setq system-messages-locale locale)
1969 (setq system-time-locale locale))
1970
1971 (setq locale (downcase locale))
1972
1973 (let ((language-name
1974 (locale-name-match locale locale-language-names))
1975 (charset-language-name
1976 (locale-name-match locale locale-charset-language-names))
1977 (coding-system
1978 (locale-name-match locale locale-preferred-coding-systems)))
1979
1980 ;; Give preference to charset-language-name over language-name.
1981 (if (and charset-language-name
1982 (not
1983 (equal (get-language-info language-name 'charset)
1984 (get-language-info charset-language-name 'charset))))
1985 (setq language-name charset-language-name))
1986
1987 (when language-name
1988
1989 ;; Set up for this character set. This is now the right way
1990 ;; to do it for both unibyte and multibyte modes.
1991 (set-language-environment language-name)
1992
1993 ;; If default-enable-multibyte-characters is nil,
1994 ;; we are using single-byte characters,
1995 ;; so the display table and terminal coding system are irrelevant.
1996 (when default-enable-multibyte-characters
1997 (set-display-table-and-terminal-coding-system language-name))
1998
1999 ;; Set the `keyboard-coding-system' if appropriate.
2000 (let ((kcs (or coding-system
2001 (car (get-language-info language-name
2002 'coding-system)))))
2003 (if (memq kcs standard-keyboard-coding-systems)
2004 (set-keyboard-coding-system kcs)))
2005
2006 (setq locale-coding-system
2007 (car (get-language-info language-name 'coding-priority))))
2008
2009 (when coding-system
2010 (prefer-coding-system coding-system)
2011 (setq locale-coding-system coding-system))))))
2012 \f
2013 ;;; Charset property
2014
2015 (defun get-charset-property (charset propname)
2016 "Return the value of CHARSET's PROPNAME property.
2017 This is the last value stored with
2018 (put-charset-property CHARSET PROPNAME VALUE)."
2019 (and (not (eq charset 'composition))
2020 (plist-get (charset-plist charset) propname)))
2021
2022 (defun put-charset-property (charset propname value)
2023 "Store CHARSETS's PROPNAME property with value VALUE.
2024 It can be retrieved with `(get-charset-property CHARSET PROPNAME)'."
2025 (or (eq charset 'composition)
2026 (set-charset-plist charset
2027 (plist-put (charset-plist charset) propname value))))
2028
2029 ;;; Character code property
2030 (put 'char-code-property-table 'char-table-extra-slots 0)
2031
2032 (defvar char-code-property-table
2033 (make-char-table 'char-code-property-table)
2034 "Char-table containing a property list of each character code.
2035
2036 See also the documentation of `get-char-code-property' and
2037 `put-char-code-property'.")
2038
2039 (defun get-char-code-property (char propname)
2040 "Return the value of CHAR's PROPNAME property in `char-code-property-table'."
2041 (let ((plist (aref char-code-property-table char)))
2042 (if (listp plist)
2043 (car (cdr (memq propname plist))))))
2044
2045 (defun put-char-code-property (char propname value)
2046 "Store CHAR's PROPNAME property with VALUE in `char-code-property-table'.
2047 It can be retrieved with `(get-char-code-property CHAR PROPNAME)'."
2048 (let ((plist (aref char-code-property-table char)))
2049 (if plist
2050 (let ((slot (memq propname plist)))
2051 (if slot
2052 (setcar (cdr slot) value)
2053 (nconc plist (list propname value))))
2054 (aset char-code-property-table char (list propname value)))))
2055
2056 \f
2057 ;; Pretty description of encoded string
2058
2059 ;; Alist of ISO 2022 control code vs the corresponding mnemonic string.
2060 (defvar iso-2022-control-alist
2061 '((?\x1b . "ESC")
2062 (?\x0e . "SO")
2063 (?\x0f . "SI")
2064 (?\x8e . "SS2")
2065 (?\x8f . "SS3")
2066 (?\x9b . "CSI")))
2067
2068 (defun encoded-string-description (str coding-system)
2069 "Return a pretty description of STR that is encoded by CODING-SYSTEM."
2070 (setq str (string-as-unibyte str))
2071 (mapconcat
2072 (if (and coding-system (eq (coding-system-type coding-system) 2))
2073 ;; Try to get a pretty description for ISO 2022 escape sequences.
2074 (function (lambda (x) (or (cdr (assq x iso-2022-control-alist))
2075 (format "%02X" x))))
2076 (function (lambda (x) (format "0x%02X" x))))
2077 str " "))
2078
2079 (defun encode-coding-char (char coding-system)
2080 "Encode CHAR by CODING-SYSTEM and return the resulting string.
2081 If CODING-SYSTEM can't safely encode CHAR, return nil."
2082 (let ((str1 (string-as-multibyte (char-to-string char)))
2083 (str2 (string-as-multibyte (make-string 2 char)))
2084 (safe-chars (and coding-system
2085 (coding-system-get coding-system 'safe-chars)))
2086 (charset (char-charset char))
2087 enc1 enc2 i1 i2)
2088 (when (or (eq safe-chars t)
2089 (eq charset 'ascii)
2090 (and safe-chars (aref safe-chars char)))
2091 ;; We must find the encoded string of CHAR. But, just encoding
2092 ;; CHAR will put extra control sequences (usually to designate
2093 ;; ASCII charaset) at the tail if type of CODING is ISO 2022.
2094 ;; To exclude such tailing bytes, we at first encode one-char
2095 ;; string and two-char string, then check how many bytes at the
2096 ;; tail of both encoded strings are the same.
2097
2098 (setq enc1 (encode-coding-string str1 coding-system)
2099 i1 (length enc1)
2100 enc2 (encode-coding-string str2 coding-system)
2101 i2 (length enc2))
2102 (while (and (> i1 0) (= (aref enc1 (1- i1)) (aref enc2 (1- i2))))
2103 (setq i1 (1- i1) i2 (1- i2)))
2104
2105 ;; Now (substring enc1 i1) and (substring enc2 i2) are the same,
2106 ;; and they are the extra control sequences at the tail to
2107 ;; exclude.
2108 (substring enc2 0 i2))))
2109
2110
2111 ;;; mule-cmds.el ends here