comment fix
[bpt/emacs.git] / lisp / international / mule-cmds.el
CommitLineData
4ed46869
KH
1;;; mule-cmds.el --- Commands for mulitilingual environment
2
4ed46869 3;; Copyright (C) 1995 Electrotechnical Laboratory, JAPAN.
fa526c4a 4;; Licensed to the Free Software Foundation.
cda74479 5;; Copyright (C) 2000 Free Software Foundation, Inc.
4ed46869
KH
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
369314dc
KH
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.
4ed46869
KH
25
26;;; Code:
27
cda74479
DL
28(eval-when-compile (defvar dos-codepage))
29
4ed46869
KH
30;;; MULE related key bindings and menus.
31
0709d285 32(defvar mule-keymap (make-sparse-keymap)
33d17698 33 "Keymap for Mule (Multilingual environment) specific commands.")
4ed46869 34
8f81f784 35;; Keep "C-x C-m ..." for mule specific commands.
0709d285 36(define-key ctl-x-map "\C-m" mule-keymap)
ef8a8c8c 37
4ed46869
KH
38(define-key mule-keymap "f" 'set-buffer-file-coding-system)
39(define-key mule-keymap "t" 'set-terminal-coding-system)
15b3e511
KH
40(define-key mule-keymap "k" 'set-keyboard-coding-system)
41(define-key mule-keymap "p" 'set-buffer-process-coding-system)
7624ebb9
KH
42(define-key mule-keymap "x" 'set-selection-coding-system)
43(define-key mule-keymap "X" 'set-next-selection-coding-system)
8b784951 44(define-key mule-keymap "\C-\\" 'set-input-method)
15b3e511 45(define-key mule-keymap "c" 'universal-coding-system-argument)
b4fba33f 46(define-key mule-keymap "l" 'set-language-environment)
4ed46869 47
281d03ec 48(define-key help-map "\C-L" 'describe-language-environment)
ac4a3a2d 49(define-key help-map "L" 'describe-language-environment)
4ed46869 50(define-key help-map "\C-\\" 'describe-input-method)
ac4a3a2d 51(define-key help-map "I" 'describe-input-method)
d0b9c3ab 52(define-key help-map "C" 'describe-coding-system)
4ed46869
KH
53(define-key help-map "h" 'view-hello-file)
54
538d88fb
EZ
55(defvar mule-menu-keymap
56 (make-sparse-keymap "Mule (Multilingual Environment)")
33d17698 57 "Keymap for Mule (Multilingual environment) menu specific commands.")
15b3e511 58
dcad02bc
EZ
59(defvar describe-language-environment-map
60 (make-sparse-keymap "Describe Language Environment"))
15b3e511 61
dcad02bc
EZ
62(defvar setup-language-environment-map
63 (make-sparse-keymap "Set Language Environment"))
15b3e511 64
dcad02bc
EZ
65(defvar set-coding-system-map
66 (make-sparse-keymap "Set Coding System"))
15b3e511 67
15b3e511 68(define-key-after mule-menu-keymap [set-language-environment]
cda74479
DL
69 (list 'menu-item "Set Language Environment" setup-language-environment-map
70 :help "Multilingual environment suitable for a specific language"))
a61f401d 71(define-key-after mule-menu-keymap [mouse-set-font]
538d88fb 72 '(menu-item "Set Font/Fontset" mouse-set-font
cda74479
DL
73 :visible (fboundp 'generate-fontset-menu)
74 :help "Select a font from list of known fonts/fontsets"))
15b3e511
KH
75(define-key-after mule-menu-keymap [separator-mule]
76 '("--")
77 t)
78(define-key-after mule-menu-keymap [toggle-input-method]
538d88fb 79 '(menu-item "Toggle Input Method" toggle-input-method)
15b3e511 80 t)
8b784951 81(define-key-after mule-menu-keymap [set-input-method]
538d88fb 82 '(menu-item "Select Input Method..." set-input-method)
15b3e511 83 t)
cda74479
DL
84(define-key-after mule-menu-keymap [describe-input-method]
85 '(menu-item "Describe Input Method" describe-input-method))
15b3e511
KH
86(define-key-after mule-menu-keymap [separator-input-method]
87 '("--")
88 t)
15b3e511 89(define-key-after mule-menu-keymap [set-various-coding-system]
cda74479
DL
90 (list 'menu-item "Set Coding Systems" set-coding-system-map
91 :enable 'enable-multibyte-characters))
538d88fb
EZ
92(define-key-after mule-menu-keymap [view-hello-file]
93 '(menu-item "Show Multi-lingual Text" view-hello-file
94 :enable (file-readable-p
95 (expand-file-name "HELLO" data-directory))
96 :help "Display file which says HELLO in many languages")
15b3e511
KH
97 t)
98(define-key-after mule-menu-keymap [separator-coding-system]
99 '("--")
100 t)
538d88fb
EZ
101(define-key-after mule-menu-keymap [describe-language-environment]
102 (list 'menu-item "Describe Language Environment"
103 describe-language-environment-map
cda74479 104 :help "Show multilingual settings for a specific language")
15b3e511 105 t)
538d88fb
EZ
106(define-key-after mule-menu-keymap [describe-input-method]
107 '(menu-item "Describe Input Method..." describe-input-method
cda74479 108 :help "Keyboard layout for a specific input method")
538d88fb
EZ
109 t)
110(define-key-after mule-menu-keymap [describe-coding-system]
111 '(menu-item "Describe Coding System..." describe-coding-system)
112 t)
cda74479
DL
113(define-key-after mule-menu-keymap [list-character-sets]
114 '(menu-item "List Character Sets" list-character-sets
115 :help "Show table of available character sets"))
538d88fb
EZ
116(define-key-after mule-menu-keymap [mule-diag]
117 '(menu-item "Show All of Mule Status" mule-diag
118 :help "Display multilingual environment settings")
15b3e511
KH
119 t)
120
121(define-key-after set-coding-system-map [set-buffer-file-coding-system]
538d88fb
EZ
122 '(menu-item "For Saving this Buffer" set-buffer-file-coding-system
123 :help "How to encode this buffer on disk")
15b3e511 124 t)
3a151e98 125(define-key-after set-coding-system-map [universal-coding-system-argument]
538d88fb
EZ
126 '(menu-item "For Next Command" universal-coding-system-argument
127 :help "Coding system to be used by next command")
3a151e98 128 t)
15b3e511 129(define-key-after set-coding-system-map [set-terminal-coding-system]
538d88fb
EZ
130 '(menu-item "For Terminal" set-terminal-coding-system
131 :enable (null (memq window-system '(x w32 mac)))
132 :help "How to encode terminal output")
15b3e511
KH
133 t)
134(define-key-after set-coding-system-map [set-keyboard-coding-system]
538d88fb
EZ
135 '(menu-item "For Keyboard" set-keyboard-coding-system
136 :help "How to decode keyboard input")
15b3e511
KH
137 t)
138(define-key-after set-coding-system-map [set-buffer-process-coding-system]
538d88fb
EZ
139 '(menu-item "For I/O with Subprocess" set-buffer-process-coding-system
140 :visible (fboundp 'start-process)
141 :enable (get-buffer-process (current-buffer))
142 :help "How to en/decode I/O from/to subprocess connected to this buffer")
15b3e511 143 t)
7624ebb9 144(define-key-after set-coding-system-map [set-selection-coding-system]
538d88fb
EZ
145 '(menu-item "For X Selections/Clipboard" set-selection-coding-system
146 :visible (display-selections-p)
147 :help "How to en/decode data to/from selection/clipboard")
7624ebb9
KH
148 t)
149(define-key-after set-coding-system-map [set-next-selection-coding-system]
538d88fb
EZ
150 '(menu-item "For Next X Selection" set-next-selection-coding-system
151 :visible (display-selections-p)
152 :help "How to en/decode next selection/clipboard operation")
7624ebb9 153 t)
15b3e511 154(define-key setup-language-environment-map
538d88fb 155 [Default] '(menu-item "Default" setup-specified-language-environment))
4ed46869 156
cda74479
DL
157(define-key describe-language-environment-map
158 [Default] '(menu-item "Default" describe-specified-language-support))
159
4ed46869
KH
160;; This should be a single character key binding because users use it
161;; very frequently while editing multilingual text. Now we can use
162;; only two such keys: "\C-\\" and "\C-^", but the latter is not
163;; convenient because it requires shifting on most keyboards. An
164;; alternative is "\C-\]" which is now bound to `abort-recursive-edit'
165;; but it won't be used that frequently.
166(define-key global-map "\C-\\" 'toggle-input-method)
167
a2ad45b9
RS
168;;; This is no good because people often type Shift-SPC
169;;; meaning to type SPC. -- rms.
170;;; ;; Here's an alternative key binding for X users (Shift-SPACE).
171;;; (define-key global-map [?\S- ] 'toggle-input-method)
b4fba33f 172
464cc130
KH
173;;; Mule related hyperlinks.
174(defconst help-xref-mule-regexp-template
175 (purecopy (concat "\\(\\<\\("
176 "\\(coding system\\)\\|"
d0c40faa
KH
177 "\\(input method\\)\\|"
178 "\\(character set\\)\\|"
179 "\\(charset\\)"
464cc130
KH
180 "\\)\\s-+\\)?"
181 ;; Note starting with word-syntax character:
182 "`\\(\\sw\\(\\sw\\|\\s_\\)+\\)'")))
183
26d87040
EZ
184(defun coding-system-change-eol-conversion (coding-system eol-type)
185 "Return a coding system which differs from CODING-SYSTEM in eol conversion.
186The returned coding system converts end-of-line by EOL-TYPE
187but text as the same way as CODING-SYSTEM.
188EOL-TYPE should be `unix', `dos', `mac', or nil.
189If EOL-TYPE is nil, the returned coding system detects
190how end-of-line is formatted automatically while decoding.
191
192EOL-TYPE can be specified by an integer 0, 1, or 2.
193They means `unix', `dos', and `mac' respectively."
194 (if (symbolp eol-type)
195 (setq eol-type (cond ((eq eol-type 'unix) 0)
196 ((eq eol-type 'dos) 1)
197 ((eq eol-type 'mac) 2)
198 (t eol-type))))
199 (let ((orig-eol-type (coding-system-eol-type coding-system)))
200 (if (vectorp orig-eol-type)
201 (if (not eol-type)
202 coding-system
203 (aref orig-eol-type eol-type))
204 (let ((base (coding-system-base coding-system)))
205 (if (not eol-type)
206 base
207 (if (= eol-type orig-eol-type)
208 coding-system
209 (setq orig-eol-type (coding-system-eol-type base))
210 (if (vectorp orig-eol-type)
211 (aref orig-eol-type eol-type))))))))
212
213(defun coding-system-change-text-conversion (coding-system coding)
214 "Return a coding system which differs from CODING-SYSTEM in text conversion.
215The returned coding system converts text by CODING
216but end-of-line as the same way as CODING-SYSTEM.
217If CODING is nil, the returned coding system detects
218how text is formatted automatically while decoding."
219 (if (not coding)
220 (coding-system-base coding-system)
221 (let ((eol-type (coding-system-eol-type coding-system)))
222 (coding-system-change-eol-conversion
223 coding
224 (if (numberp eol-type) (aref [unix dos mac] eol-type))))))
225
4ed46869 226(defun toggle-enable-multibyte-characters (&optional arg)
6998e1a1
RS
227 "Change whether this buffer uses multibyte characters.
228With arg, use multibyte characters if the arg is positive.
229
230Note that this command does not convert the byte contents of
231the buffer; it only changes the way those bytes are interpreted.
232In general, therefore, this command *changes* the sequence of
233characters that the current buffer contains.
234
235We suggest you avoid using use this command unless you know what you
236are doing. If you use it by mistake, and the buffer is now displayed
237wrong, use this command again to toggle back to the right mode."
4ed46869 238 (interactive "P")
b7079457
RS
239 (let ((new-flag
240 (if (null arg) (null enable-multibyte-characters)
241 (> (prefix-numeric-value arg) 0))))
242 (set-buffer-multibyte new-flag))
4ed46869
KH
243 (force-mode-line-update))
244
245(defun view-hello-file ()
246 "Display the HELLO file which list up many languages and characters."
247 (interactive)
8f81f784
KH
248 ;; We have to decode the file in any environment.
249 (let ((default-enable-multibyte-characters t)
95fa03b4 250 (coding-system-for-read 'iso-2022-7bit))
8f81f784 251 (find-file-read-only (expand-file-name "HELLO" data-directory))))
4ed46869 252
15b3e511
KH
253(defun universal-coding-system-argument ()
254 "Execute an I/O command using the specified coding system."
255 (interactive)
34104362
KH
256 (let* ((default (and buffer-file-coding-system
257 (not (eq (coding-system-type buffer-file-coding-system)
258 t))
259 buffer-file-coding-system))
260 (coding-system (read-coding-system
261 (if default
262 (format "Coding system for following command (default, %s): " default)
263 "Coding system for following command: ")
264 default))
15b3e511 265 (keyseq (read-key-sequence
e14a8f4c 266 (format "Command to execute with %s:" coding-system)))
15b3e511
KH
267 (cmd (key-binding keyseq)))
268 (let ((coding-system-for-read coding-system)
269 (coding-system-for-write coding-system))
270 (message "")
271 (call-interactively cmd))))
272
de94d711 273(defun set-default-coding-systems (coding-system)
0c3154d2 274 "Set default value of various coding systems to CODING-SYSTEM.
387136f6 275This sets the following coding systems:
0c3154d2 276 o coding system of a newly created buffer
8efc03e1
KH
277 o default coding system for subprocess I/O
278This also sets the following values:
387136f6 279 o default value used as file-name-coding-system for converting file names.
03c35c83
EZ
280 o default value for the command `set-terminal-coding-system' (not on MSDOS)
281 o default value for the command `set-keyboard-coding-system'."
de94d711
KH
282 (check-coding-system coding-system)
283 (setq-default buffer-file-coding-system coding-system)
716184d4
RS
284 (if default-enable-multibyte-characters
285 (setq default-file-name-coding-system coding-system))
03c35c83
EZ
286 ;; If coding-system is nil, honor that on MS-DOS as well, so
287 ;; that they could reset the terminal coding system.
288 (unless (and (eq window-system 'pc) coding-system)
289 (setq default-terminal-coding-system coding-system))
de94d711
KH
290 (setq default-keyboard-coding-system coding-system)
291 (setq default-process-coding-system (cons coding-system coding-system)))
292
45d08cb2 293(defalias 'update-iso-coding-systems 'update-coding-systems-internal)
2598a293 294(make-obsolete 'update-iso-coding-systems 'update-coding-systems-internal "20.3")
45d08cb2 295
0c3154d2
KH
296(defun prefer-coding-system (coding-system)
297 "Add CODING-SYSTEM at the front of the priority list for automatic detection.
387136f6 298This also sets the following coding systems:
0c3154d2 299 o coding system of a newly created buffer
8efc03e1
KH
300 o default coding system for subprocess I/O
301This also sets the following values:
387136f6 302 o default value used as file-name-coding-system for converting file names.
03c35c83
EZ
303 o default value for the command `set-terminal-coding-system' (not on MSDOS)
304 o default value for the command `set-keyboard-coding-system'
305
bd3ac67e
EZ
306If CODING-SYSTEM specifies a certain type of EOL conversion, the coding
307systems set by this function will use that type of EOL conversion.
308
03c35c83
EZ
309This command does not change the default value of terminal coding system
310for MS-DOS terminal, because DOS terminals only support a single coding
311system, and Emacs automatically sets the default to that coding system at
312startup."
0c3154d2
KH
313 (interactive "zPrefer coding system: ")
314 (if (not (and coding-system (coding-system-p coding-system)))
315 (error "Invalid coding system `%s'" coding-system))
316 (let ((coding-category (coding-system-category coding-system))
bd3ac67e
EZ
317 (base (coding-system-base coding-system))
318 (eol-type (coding-system-eol-type coding-system)))
0c3154d2
KH
319 (if (not coding-category)
320 ;; CODING-SYSTEM is no-conversion or undecided.
321 (error "Can't prefer the coding system `%s'" coding-system))
8efc03e1 322 (set coding-category (or base coding-system))
45d08cb2 323 (update-coding-systems-internal)
812cad80 324 (or (eq coding-category (car coding-category-list))
0c3154d2 325 ;; We must change the order.
812cad80 326 (set-coding-priority (list coding-category)))
8efc03e1
KH
327 (if (and base (interactive-p))
328 (message "Highest priority is set to %s (base of %s)"
329 base coding-system))
bd3ac67e 330 ;; If they asked for specific EOL conversion, honor that.
6f9dc4fd 331 (if (memq eol-type '(0 1 2))
bd3ac67e
EZ
332 (setq coding-system
333 (coding-system-change-eol-conversion base eol-type))
334 (setq coding-system base))
335 (set-default-coding-systems coding-system)))
0c3154d2 336
b5edd1d1
KH
337(defvar sort-coding-systems-predicate nil
338 "If non-nil, a predicate function to sort coding systems.
339
340It is called with two coding systems, and should return t if the first
341one is \"less\" than the second.
342
343The function `sort-coding-systems' use it.")
344
345(defun sort-coding-systems (codings)
346 "Sort coding system list CODINGS by a priority of each coding system.
347
348If a coding system is most preferred, it has the highest priority.
349Otherwise, a coding system corresponds to some MIME charset has higher
350priorities. Among them, a coding system included in `coding-system'
351key of the current language environment has higher priorities. See
352also the documentation of `language-info-alist'.
353
354If the variable `sort-coding-systems-predicate' (which see) is
355non-nil, it is used to sort CODINGS in the different way than above."
356 (if sort-coding-systems-predicate
357 (sort codings sort-coding-systems-predicate)
358 (let* ((most-preferred (symbol-value (car coding-category-list)))
359 (lang-preferred (get-language-info current-language-environment
360 'coding-system))
361 (func (function
362 (lambda (x)
363 (let ((base (coding-system-base x)))
364 (+ (if (eq base most-preferred) 64 0)
365 (let ((mime (coding-system-get base 'mime-charset)))
366 (if mime
367 (if (string-match "^x-" (symbol-name mime))
368 16 32)
369 0))
370 (if (memq base lang-preferred) 8 0)
371 (if (string-match "-with-esc$" (symbol-name base))
372 0 4)
373 (if (eq (coding-system-type base) 2)
374 ;; For ISO based coding systems, prefer
375 ;; one that doesn't use escape sequences.
376 (let ((flags (coding-system-flags base)))
377 (if (or (consp (aref flags 0))
378 (consp (aref flags 1))
379 (consp (aref flags 2))
380 (consp (aref flags 3)))
381 (if (or (aref flags 8) (aref flags 9))
382 0
383 1)
384 2))
385 1)))))))
386 (sort codings (function (lambda (x y)
387 (> (funcall func x) (funcall func y))))))))
54b226f7 388
3fc7dfe5 389(defun find-coding-systems-region (from to)
54b226f7
KH
390 "Return a list of proper coding systems to encode a text between FROM and TO.
391All coding systems in the list can safely encode any multibyte characters
392in the text.
393
e8dd0160 394If the text contains no multibyte characters, return a list of a single
3fc7dfe5 395element `undecided'."
b5edd1d1
KH
396 (let ((codings (find-coding-systems-region-internal from to)))
397 (if (eq codings t)
398 ;; The text contains only ASCII characters. Any coding
399 ;; systems are safe.
400 '(undecided)
401 ;; We need copy-sequence because sorting will alter the argument.
402 (sort-coding-systems (copy-sequence codings)))))
54b226f7 403
3fc7dfe5
KH
404(defun find-coding-systems-string (string)
405 "Return a list of proper coding systems to encode STRING.
406All coding systems in the list can safely encode any multibyte characters
407in STRING.
408
e8dd0160 409If STRING contains no multibyte characters, return a list of a single
3fc7dfe5 410element `undecided'."
b5edd1d1 411 (find-coding-systems-region string nil))
3fc7dfe5
KH
412
413(defun find-coding-systems-for-charsets (charsets)
414 "Return a list of proper coding systems to encode characters of CHARSETS.
415CHARSETS is a list of character sets."
b5edd1d1
KH
416 (cond ((or (null charsets)
417 (and (= (length charsets) 1)
418 (eq 'ascii (car charsets))))
419 '(undecided))
420 ((or (memq 'eight-bit-control charsets)
421 (memq 'eight-bit-graphic charsets))
422 '(raw-text emacs-mule))
423 (t
424 (let ((codings t)
425 charset l ll)
426 (while (and codings charsets)
427 (setq charset (car charsets) charsets (cdr charsets))
428 (unless (eq charset 'ascii)
429 (setq l (aref char-coding-system-table (make-char charset)))
430 (if (eq codings t)
431 (setq codings l)
432 (let ((ll nil))
433 (while codings
434 (if (memq (car codings) l)
435 (setq ll (cons (car codings) ll)))
436 (setq codings (cdr codings)))
437 (setq codings ll)))))
438 (append codings
439 (char-table-extra-slot char-coding-system-table 0))))))
54b226f7 440
51ed58ea
KH
441(defun find-multibyte-characters (from to &optional maxcount excludes)
442 "Find multibyte characters in the region specified by FROM and TO.
443If FROM is a string, find multibyte characters in the string.
444The return value is an alist of the following format:
445 ((CHARSET COUNT CHAR ...) ...)
446where
447 CHARSET is a character set,
448 COUNT is a number of characters,
449 CHARs are found characters of the character set.
450Optional 3rd arg MAXCOUNT limits how many CHARs are put in the above list.
251d4f4b
KH
451Optional 4th arg EXCLUDE is a list of character sets to be ignored.
452
453For invalid characters, CHARs are actually strings."
51ed58ea
KH
454 (let ((chars nil)
455 charset char)
456 (if (stringp from)
457 (let ((idx 0))
458 (while (setq idx (string-match "[^\000-\177]" from idx))
459 (setq char (aref from idx)
460 charset (char-charset char))
251d4f4b
KH
461 (if (eq charset 'unknown)
462 (setq char (match-string 0)))
14333e31
KH
463 (if (or (memq charset '(unknown
464 eight-bit-control eight-bit-graphic))
251d4f4b 465 (not (or (eq excludes t) (memq charset excludes))))
51ed58ea
KH
466 (let ((slot (assq charset chars)))
467 (if slot
468 (if (not (memq char (nthcdr 2 slot)))
469 (let ((count (nth 1 slot)))
470 (setcar (cdr slot) (1+ count))
471 (if (or (not maxcount) (< count maxcount))
472 (nconc slot (list char)))))
473 (setq chars (cons (list charset 1 char) chars)))))
474 (setq idx (1+ idx))))
475 (save-excursion
476 (goto-char from)
477 (while (re-search-forward "[^\000-\177]" to t)
478 (setq char (preceding-char)
479 charset (char-charset char))
251d4f4b
KH
480 (if (eq charset 'unknown)
481 (setq char (match-string 0)))
14333e31 482 (if (or (memq charset '(unknown eight-bit-control eight-bit-graphic))
251d4f4b 483 (not (or (eq excludes t) (memq charset excludes))))
51ed58ea
KH
484 (let ((slot (assq charset chars)))
485 (if slot
251d4f4b 486 (if (not (member char (nthcdr 2 slot)))
51ed58ea
KH
487 (let ((count (nth 1 slot)))
488 (setcar (cdr slot) (1+ count))
489 (if (or (not maxcount) (< count maxcount))
490 (nconc slot (list char)))))
491 (setq chars (cons (list charset 1 char) chars))))))))
492 (nreverse chars)))
493
c83c4f60
RS
494(defvar last-coding-system-specified nil
495 "Most recent coding system explicitly specified by the user when asked.
496This variable is set whenever Emacs asks the user which coding system
497to use in order to write a file. If you set it to nil explicitly,
498then call `write-region', then afterward this variable will be non-nil
499only if the user was explicitly asked and specified a coding system.")
500
b5edd1d1
KH
501(defvar select-safe-coding-system-accept-default-p nil
502 "If non-nil, a function to control the behaviour of coding system selection.
503The meaning is the same as the argument ACCEPT-DEFAULT-P of the
504function `select-safe-coding-system' (which see). This variable
505overrides that argument.")
506
507(defun select-safe-coding-system (from to &optional default-coding-system
508 accept-default-p)
d5266ddf
KH
509 "Ask a user to select a safe coding system from candidates.
510The candidates of coding systems which can safely encode a text
b5edd1d1
KH
511between FROM and TO are shown in a popup window. Among them, the most
512proper one is suggested as the default.
513
514The list of `buffer-file-coding-system' of the current buffer and the
515most preferred coding system (if it corresponds to a MIME charset) is
516treated as the default coding system list. Among them, the first one
517that safely encodes the text is silently selected and returned without
518any user interaction. See also the command `prefer-coding-system'.
54b226f7 519
b5edd1d1
KH
520Optional 3rd arg DEFAULT-CODING-SYSTEM specifies a coding system or a
521list of coding systems to be prepended to the default coding system
522list.
54b226f7 523
b5edd1d1
KH
524Optional 4th arg ACCEPT-DEFAULT-P, if non-nil, is a function to
525determine the acceptability of the silently selected coding system.
526It is called with that coding system, and should return nil if it
527should not be silently selected and thus user interaction is required.
528
529The variable `select-safe-coding-system-accept-default-p', if
530non-nil, overrides ACCEPT-DEFAULT-P.
54b226f7
KH
531
532Kludgy feature: if FROM is a string, the string is the target text,
533and TO is ignored."
b5edd1d1
KH
534 (if (and default-coding-system
535 (not (listp default-coding-system)))
536 (setq default-coding-system (list default-coding-system)))
537
538 ;; Change elements of the list to (coding . base-coding).
539 (setq default-coding-system
540 (mapcar (function (lambda (x) (cons x (coding-system-base x))))
541 default-coding-system))
542
543 ;; If buffer-file-coding-system is not nil nor undecided, append it
544 ;; to the defaults.
545 (if buffer-file-coding-system
546 (let ((base (coding-system-base buffer-file-coding-system)))
547 (or (eq base 'undecided)
548 (assq buffer-file-coding-system default-coding-system)
549 (rassq base default-coding-system)
e56d7900 550 (setq default-coding-system
b5edd1d1
KH
551 (append default-coding-system
552 (list (cons buffer-file-coding-system base)))))))
553
554 ;; If the most preferred coding system has the property mime-charset,
555 ;; append it to the defaults.
556 (let* ((preferred (symbol-value (car coding-category-list)))
557 (base (coding-system-base preferred)))
558 (and (coding-system-get preferred 'mime-charset)
559 (not (assq preferred default-coding-system))
560 (not (rassq base default-coding-system))
561 (setq default-coding-system
562 (append default-coding-system (list (cons preferred base))))))
563
564 (if select-safe-coding-system-accept-default-p
565 (setq accept-default-p select-safe-coding-system-accept-default-p))
566
567 (let ((codings (find-coding-systems-region from to))
568 (coding-system nil)
569 (l default-coding-system))
570 (if (eq (car codings) 'undecided)
571 ;; Any coding system is ok.
572 (setq coding-system t)
573 ;; Try the defaults.
574 (while (and l (not coding-system))
575 (if (memq (cdr (car l)) codings)
576 (setq coding-system (car (car l)))
577 (setq l (cdr l))))
578 (if (and coding-system accept-default-p)
579 (or (funcall accept-default-p coding-system)
580 (setq coding-system (list coding-system)))))
581
582 ;; If all the defaults failed, ask a user.
583 (when (or (not coding-system) (consp coding-system))
34104362 584 ;; At first, change each coding system to the corresponding
b5edd1d1
KH
585 ;; mime-charset name if it is also a coding system. Such a name
586 ;; is more friendly to users.
587 (let ((l codings)
34104362
KH
588 mime-charset)
589 (while l
590 (setq mime-charset (coding-system-get (car l) 'mime-charset))
591 (if (and mime-charset (coding-system-p mime-charset))
592 (setcar l mime-charset))
593 (setq l (cdr l))))
594
b5edd1d1
KH
595 ;; Then ask users to select one form CODINGS.
596 (unwind-protect
597 (save-window-excursion
598 (with-output-to-temp-buffer "*Warning*"
599 (save-excursion
600 (set-buffer standard-output)
038f27a7
KH
601 (insert "These default coding systems were tried")
602 (if (stringp from)
603 (insert " to encode \""
604 (if (> (length from) 10)
605 (substring from 0 10)
606 from)
607 "...\""))
608 (insert ":\n")
b5edd1d1
KH
609 (let ((pos (point))
610 (fill-prefix " "))
611 (mapcar (function (lambda (x) (princ " ") (princ (car x))))
612 default-coding-system)
613 (insert "\n")
614 (fill-region-as-paragraph pos (point)))
2f1fa038
EZ
615 (insert
616 (if (consp coding-system)
617 (concat (format "%s safely encodes the target text,\n"
618 (car coding-system))
619 "but it is not recommended for encoding text in this context,\n"
620 "e.g., for sending an email message.\n")
621 "However, none of them safely encodes the target text.\n"))
b5edd1d1 622 (insert (if (consp coding-system)
2f1fa038
EZ
623 "\nSelect the above, or "
624 "\nSelect ")
625 "one of the following safe coding systems:\n")
b5edd1d1
KH
626 (let ((pos (point))
627 (fill-prefix " "))
628 (mapcar (function (lambda (x) (princ " ") (princ x)))
629 codings)
630 (insert "\n")
631 (fill-region-as-paragraph pos (point)))))
632
633 ;; Read a coding system.
634 (if (consp coding-system)
635 (setq codings (cons (car coding-system) codings)))
636 (let* ((safe-names (mapcar (lambda (x) (list (symbol-name x)))
637 codings))
638 (name (completing-read
639 (format "Select coding system (default %s): "
640 (car codings))
641 safe-names nil t nil nil
642 (car (car safe-names)))))
643 (setq last-coding-system-specified (intern name)
644 coding-system last-coding-system-specified)))
645 (kill-buffer "*Warning*")))
646
647 (if (vectorp (coding-system-eol-type coding-system))
648 (let ((eol (coding-system-eol-type buffer-file-coding-system)))
649 (if (numberp eol)
650 (setq coding-system
651 (coding-system-change-eol-conversion coding-system eol)))))
652
653 (if (eq coding-system t)
654 (setq coding-system buffer-file-coding-system))
e56d7900 655 coding-system))
54b226f7
KH
656
657(setq select-safe-coding-system-function 'select-safe-coding-system)
658
46babb23
KH
659(defun select-message-coding-system ()
660 "Return a coding system to encode the outgoing message of the current buffer.
661It at first tries the first coding system found in these variables
662in this order:
663 (1) local value of `buffer-file-coding-system'
664 (2) value of `sendmail-coding-system'
b5edd1d1
KH
665 (3) value of `default-sendmail-coding-system'
666 (4) value of `default-buffer-file-coding-system'
46babb23
KH
667If the found coding system can't encode the current buffer,
668or none of them are bound to a coding system,
48e41165 669it asks the user to select a proper coding system."
46babb23 670 (let ((coding (or (and (local-variable-p 'buffer-file-coding-system)
b5edd1d1
KH
671 buffer-file-coding-system)
672 sendmail-coding-system
673 default-sendmail-coding-system
674 default-buffer-file-coding-system)))
46babb23
KH
675 (if (eq coding 'no-conversion)
676 ;; We should never use no-conversion for outgoing mails.
677 (setq coding nil))
678 (if (fboundp select-safe-coding-system-function)
679 (funcall select-safe-coding-system-function
b5edd1d1
KH
680 (point-min) (point-max) coding
681 (function (lambda (x) (coding-system-get x 'mime-charset))))
46babb23 682 coding)))
4ed46869 683\f
03c35c83 684;;; Language support stuff.
4ed46869 685
4ed46869 686(defvar language-info-alist nil
2c395d56 687 "Alist of language environment definitions.
4ed46869
KH
688Each element looks like:
689 (LANGUAGE-NAME . ((KEY . INFO) ...))
2c395d56
RS
690where LANGUAGE-NAME is a string, the name of the language environment,
691KEY is a symbol denoting the kind of information, and
692INFO is the data associated with KEY.
693Meaningful values for KEY include
694
695 documentation value is documentation of what this language environment
696 is meant for, and how to use it.
697 charset value is a list of the character sets used by this
698 language environment.
699 sample-text value is one line of text,
700 written using those character sets,
701 appropriate for this language environment.
702 setup-function value is a function to call to switch to this
703 language environment.
704 exit-function value is a function to call to leave this
705 language environment.
706 coding-system value is a list of coding systems that are good
707 for saving text written in this language environment.
708 This list serves as suggestions to the user;
709 in effect, as a kind of documentation.
710 coding-priority value is a list of coding systems for this language
711 environment, in order of decreasing priority.
712 This is used to set up the coding system priority
45d08cb2 713 list when you switch to this language environment.
ddb5c041 714 nonascii-translation
7624ebb9 715 value is a translation table to be set in the
45d08cb2 716 variable `nonascii-translation-table' in this
7624ebb9
KH
717 language environment, or a character set from
718 which `nonascii-insert-offset' is calculated.
ddb5c041
KH
719 input-method value is a default input method for this language
720 environment.
7624ebb9
KH
721 features value is a list of features requested in this
722 language environment.
ddb5c041
KH
723
724The following keys take effect only when multibyte characters are
725globally disabled, i.e. the value of `default-enable-multibyte-characters'
726is nil.
727
728 unibyte-syntax value is a library name to load to set
e8dd0160 729 unibyte 8-bit character syntaxes for this
ddb5c041
KH
730 language environment.
731
732 unibyte-display value is a coding system to encode characters
733 for the terminal. Characters in the range
734 of 160 to 255 display not as octal escapes,
735 but as non-ASCII characters in this language
736 environment.")
2c395d56
RS
737
738(defun get-language-info (lang-env key)
739 "Return information listed under KEY for language environment LANG-ENV.
740KEY is a symbol denoting the kind of information.
741For a list of useful values for KEY and their meanings,
742see `language-info-alist'."
743 (if (symbolp lang-env)
744 (setq lang-env (symbol-name lang-env)))
745 (let ((lang-slot (assoc-ignore-case lang-env language-info-alist)))
4ed46869
KH
746 (if lang-slot
747 (cdr (assq key (cdr lang-slot))))))
748
f08adf27 749(defun set-language-info (lang-env key info)
2c395d56
RS
750 "Modify part of the definition of language environment LANG-ENV.
751Specifically, this stores the information INFO under KEY
752in the definition of this language environment.
4ed46869 753KEY is a symbol denoting the kind of information.
2c395d56 754INFO is the value for that information.
281d03ec 755
2c395d56 756For a list of useful values for KEY and their meanings,
f08adf27 757see `language-info-alist'."
2c395d56
RS
758 (if (symbolp lang-env)
759 (setq lang-env (symbol-name lang-env)))
4ed46869 760 (let (lang-slot key-slot)
2c395d56 761 (setq lang-slot (assoc lang-env language-info-alist))
4ed46869 762 (if (null lang-slot) ; If no slot for the language, add it.
2c395d56 763 (setq lang-slot (list lang-env)
4ed46869
KH
764 language-info-alist (cons lang-slot language-info-alist)))
765 (setq key-slot (assq key lang-slot))
766 (if (null key-slot) ; If no slot for the key, add it.
767 (progn
768 (setq key-slot (list key))
769 (setcdr lang-slot (cons key-slot (cdr lang-slot)))))
cda74479 770 (setcdr key-slot (purecopy info))))
4ed46869 771
2c395d56
RS
772(defun set-language-info-alist (lang-env alist &optional parents)
773 "Store ALIST as the definition of language environment LANG-ENV.
774ALIST is an alist of KEY and INFO values. See the documentation of
98c6d6ed 775`language-info-alist' for the meanings of KEY and INFO.
54b226f7 776
2c395d56
RS
777Optional arg PARENTS is a list of parent menu names; it specifies
778where to put this language environment in the
779Describe Language Environment and Set Language Environment menus.
780For example, (\"European\") means to put this language environment
781in the European submenu in each of those two menus."
782 (if (symbolp lang-env)
783 (setq lang-env (symbol-name lang-env)))
54b226f7
KH
784 (let ((describe-map describe-language-environment-map)
785 (setup-map setup-language-environment-map))
786 (if parents
787 (let ((l parents)
9deed82f 788 map parent-symbol parent prompt)
54b226f7
KH
789 (while l
790 (if (symbolp (setq parent-symbol (car l)))
791 (setq parent (symbol-name parent))
792 (setq parent parent-symbol parent-symbol (intern parent)))
793 (setq map (lookup-key describe-map (vector parent-symbol)))
9deed82f
EZ
794 ;; This prompt string is for define-prefix-command, so
795 ;; that the map it creates will be suitable for a menu.
796 (or map (setq prompt (format "%s Environment" parent)))
54b226f7
KH
797 (if (not map)
798 (progn
799 (setq map (intern (format "describe-%s-environment-map"
800 (downcase parent))))
9deed82f 801 (define-prefix-command map nil prompt)
54b226f7
KH
802 (define-key-after describe-map (vector parent-symbol)
803 (cons parent map) t)))
804 (setq describe-map (symbol-value map))
805 (setq map (lookup-key setup-map (vector parent-symbol)))
806 (if (not map)
807 (progn
808 (setq map (intern (format "setup-%s-environment-map"
809 (downcase parent))))
9deed82f 810 (define-prefix-command map nil prompt)
54b226f7
KH
811 (define-key-after setup-map (vector parent-symbol)
812 (cons parent map) t)))
813 (setq setup-map (symbol-value map))
814 (setq l (cdr l)))))
f08adf27
RS
815
816 ;; Set up menu items for this language env.
7624ebb9 817 (let ((doc (assq 'documentation alist)))
f08adf27
RS
818 (when doc
819 (define-key-after describe-map (vector (intern lang-env))
7624ebb9
KH
820 (cons lang-env 'describe-specified-language-support) t)))
821 (define-key-after setup-map (vector (intern lang-env))
822 (cons lang-env 'setup-specified-language-environment) t)
f08adf27 823
54b226f7 824 (while alist
f08adf27 825 (set-language-info lang-env (car (car alist)) (cdr (car alist)))
54b226f7 826 (setq alist (cdr alist)))))
4ed46869 827
ae302641 828(defun read-language-name (key prompt &optional default)
2c395d56 829 "Read a language environment name which has information for KEY.
ddb5c041 830If KEY is nil, read any language environment.
2c395d56
RS
831Prompt with PROMPT. DEFAULT is the default choice of language environment.
832This returns a language environment name as a string."
4ed46869
KH
833 (let* ((completion-ignore-case t)
834 (name (completing-read prompt
835 language-info-alist
ddb5c041
KH
836 (and key
837 (function (lambda (elm) (assq key elm))))
ae302641 838 t nil nil default)))
13e82c04 839 (if (and (> (length name) 0)
ddb5c041
KH
840 (or (not key)
841 (get-language-info name key)))
13e82c04 842 name)))
4ed46869
KH
843\f
844;;; Multilingual input methods.
d0c40faa
KH
845(defgroup leim nil
846 "LEIM: Libraries of Emacs Input Methods."
847 :group 'mule)
4ed46869 848
d0b9c3ab
KH
849(defconst leim-list-file-name "leim-list.el"
850 "Name of LEIM list file.
851This file contains a list of libraries of Emacs input methods (LEIM)
852in the format of Lisp expression for registering each input method.
853Emacs loads this file at startup time.")
854
2e224638
SM
855(defvar leim-list-header (format
856";;; %s -- list of LEIM (Library of Emacs Input Method)
d0b9c3ab
KH
857;;
858;; This file contains a list of LEIM (Library of Emacs Input Method)
e8dd0160 859;; in the same directory as this file. Loading this file registers
d0b9c3ab
KH
860;; the whole input methods in Emacs.
861;;
d33d5fbe 862;; Each entry has the form:
d0b9c3ab
KH
863;; (register-input-method
864;; INPUT-METHOD LANGUAGE-NAME ACTIVATE-FUNC
865;; TITLE DESCRIPTION
866;; ARG ...)
867;; See the function `register-input-method' for the meanings of arguments.
868;;
869;; If this directory is included in load-path, Emacs automatically
870;; loads this file at startup time.
871
872"
873 leim-list-file-name)
874 "Header to be inserted in LEIM list file.")
875
e55e92ee 876(defvar leim-list-entry-regexp "^(register-input-method"
d0b9c3ab
KH
877 "Regexp matching head of each entry in LEIM list file.
878See also the variable `leim-list-header'")
879
880(defvar update-leim-list-functions
881 '(quail-update-leim-list-file)
882 "List of functions to call to update LEIM list file.
883Each function is called with one arg, LEIM directory name.")
884
a337fe7f
RS
885(defun update-leim-list-file (&rest dirs)
886 "Update LEIM list file in directories DIRS."
d0b9c3ab
KH
887 (let ((functions update-leim-list-functions))
888 (while functions
a337fe7f 889 (apply (car functions) dirs)
d0b9c3ab
KH
890 (setq functions (cdr functions)))))
891
4ed46869
KH
892(defvar current-input-method nil
893 "The current input method for multilingual text.
96db204a 894If nil, that means no input method is activated now.")
4ed46869
KH
895(make-variable-buffer-local 'current-input-method)
896(put 'current-input-method 'permanent-local t)
897
898(defvar current-input-method-title nil
d0b9c3ab 899 "Title string of the current input method shown in mode line.")
4ed46869
KH
900(make-variable-buffer-local 'current-input-method-title)
901(put 'current-input-method-title 'permanent-local t)
902
b4fba33f 903(defcustom default-input-method nil
8861c593 904 "*Default input method for multilingual text (a string).
b4fba33f 905This is the input method activated automatically by the command
9b10b5a3 906`toggle-input-method' (\\[toggle-input-method])."
8861c593 907 :group 'mule
5806e8a6
GM
908 :type '(choice (const nil) string)
909 :set-after '(current-language-environment))
b4fba33f 910
0f835e87
KH
911(put 'input-method-function 'permanent-local t)
912
723a427a
KH
913(defvar input-method-history nil
914 "History list for some commands that read input methods.")
915(make-variable-buffer-local 'input-method-history)
916(put 'input-method-history 'permanent-local t)
4ed46869
KH
917
918(defvar inactivate-current-input-method-function nil
919 "Function to call for inactivating the current input method.
920Every input method should set this to an appropriate value when activated.
f17ccaee
KH
921This function is called with no argument.
922
923This function should never change the value of `current-input-method'.
924It is set to nil by the function `inactivate-input-method'.")
4ed46869
KH
925(make-variable-buffer-local 'inactivate-current-input-method-function)
926(put 'inactivate-current-input-method-function 'permanent-local t)
927
928(defvar describe-current-input-method-function nil
929 "Function to call for describing the current input method.
930This function is called with no argument.")
931(make-variable-buffer-local 'describe-current-input-method-function)
932(put 'describe-current-input-method-function 'permanent-local t)
933
d0b9c3ab 934(defvar input-method-alist nil
2c395d56 935 "Alist of input method names vs how to use them.
d0b9c3ab 936Each element has the form:
2c395d56
RS
937 (INPUT-METHOD LANGUAGE-ENV ACTIVATE-FUNC TITLE DESCRIPTION ARGS...)
938See the function `register-input-method' for the meanings of the elements.")
939
f08adf27 940(defun register-input-method (input-method lang-env &rest args)
2c395d56 941 "Register INPUT-METHOD as an input method for language environment ENV.
f08adf27 942INPUT-METHOD and LANG-ENV are symbols or strings.
d0b9c3ab 943
d0b9c3ab 944The remaining arguments are:
2c395d56
RS
945 ACTIVATE-FUNC, TITLE, DESCRIPTION, and ARGS...
946ACTIVATE-FUNC is a function to call to activate this method.
947TITLE is a string to show in the mode line when this method is active.
948DESCRIPTION is a string describing this method and what it is good for.
949The ARGS, if any, are passed as arguments to ACTIVATE-FUNC.
205814ee
KH
950All told, the arguments to ACTIVATE-FUNC are INPUT-METHOD and the ARGS.
951
952This function is mainly used in the file \"leim-list.el\" which is
953created at building time of emacs, registering all quail input methods
954contained in the emacs distribution.
955
956In case you want to register a new quail input method by yourself, be
957careful to use the same input method title as given in the third
958parameter of `quail-define-package' (if the values are different, the
959string specified in this function takes precedence).
960
961The commands `describe-input-method' and `list-input-methods' need
962this duplicated values to show some information about input methods
963without loading the affected quail packages."
f08adf27
RS
964 (if (symbolp lang-env)
965 (setq lang-env (symbol-name lang-env)))
4ef06f75
KH
966 (if (symbolp input-method)
967 (setq input-method (symbol-name input-method)))
f08adf27 968 (let ((info (cons lang-env args))
d0b9c3ab
KH
969 (slot (assoc input-method input-method-alist)))
970 (if slot
971 (setcdr slot info)
972 (setq slot (cons input-method info))
973 (setq input-method-alist (cons slot input-method-alist)))))
974
4d5ac029 975(defun read-input-method-name (prompt &optional default inhibit-null)
d0b9c3ab 976 "Read a name of input method from a minibuffer prompting with PROMPT.
4d5ac029
RS
977If DEFAULT is non-nil, use that as the default,
978 and substitute it into PROMPT at the first `%s'.
4ef06f75
KH
979If INHIBIT-NULL is non-nil, null input signals an error.
980
981The return value is a string."
4d5ac029
RS
982 (if default
983 (setq prompt (format prompt default)))
d0b9c3ab 984 (let* ((completion-ignore-case t)
723a427a
KH
985 ;; This binding is necessary because input-method-history is
986 ;; buffer local.
d0b9c3ab 987 (input-method (completing-read prompt input-method-alist
87505a98
RS
988 nil t nil 'input-method-history
989 default)))
bf294e6e
KH
990 (if (and input-method (symbolp input-method))
991 (setq input-method (symbol-name input-method)))
d0b9c3ab
KH
992 (if (> (length input-method) 0)
993 input-method
994 (if inhibit-null
43807b77 995 (error "No valid input method is specified")))))
d0b9c3ab 996
d0b9c3ab 997(defun activate-input-method (input-method)
2c395d56
RS
998 "Switch to input method INPUT-METHOD for the current buffer.
999If some other input method is already active, turn it off first.
1000If INPUT-METHOD is nil, deactivate any current input method."
305a3cb6 1001 (if (and input-method (symbolp input-method))
4ef06f75 1002 (setq input-method (symbol-name input-method)))
723a427a
KH
1003 (if (and current-input-method
1004 (not (string= current-input-method input-method)))
305a3cb6 1005 (inactivate-input-method))
2c395d56 1006 (unless (or current-input-method (null input-method))
d0b9c3ab
KH
1007 (let ((slot (assoc input-method input-method-alist)))
1008 (if (null slot)
723a427a 1009 (error "Can't activate input method `%s'" input-method))
8efc03e1
KH
1010 (let ((func (nth 2 slot)))
1011 (if (functionp func)
1012 (apply (nth 2 slot) input-method (nthcdr 5 slot))
1013 (if (and (consp func) (symbolp (car func)) (symbolp (cdr func)))
1014 (progn
1015 (require (cdr func))
1016 (apply (car func) input-method (nthcdr 5 slot)))
1017 (error "Can't activate input method `%s'" input-method))))
d0b9c3ab 1018 (setq current-input-method input-method)
723a427a 1019 (setq current-input-method-title (nth 3 slot))
28885c0e
KH
1020 (unwind-protect
1021 (run-hooks 'input-method-activate-hook)
1022 (force-mode-line-update)))))
15b3e511 1023
15b3e511 1024(defun inactivate-input-method ()
f17ccaee 1025 "Turn off the current input method."
723a427a
KH
1026 (when current-input-method
1027 (if input-method-history
1028 (unless (string= current-input-method (car input-method-history))
1029 (setq input-method-history
1030 (cons current-input-method
1031 (delete current-input-method input-method-history))))
1032 (setq input-method-history (list current-input-method)))
1033 (unwind-protect
1034 (funcall inactivate-current-input-method-function)
15b3e511 1035 (unwind-protect
723a427a
KH
1036 (run-hooks 'input-method-inactivate-hook)
1037 (setq current-input-method nil
28885c0e
KH
1038 current-input-method-title nil)
1039 (force-mode-line-update)))))
4ed46869 1040
8b784951 1041(defun set-input-method (input-method)
2c395d56
RS
1042 "Select and activate input method INPUT-METHOD for the current buffer.
1043This also sets the default input method to the one you specify."
d0b9c3ab 1044 (interactive
723a427a 1045 (let* ((default (or (car input-method-history) default-input-method)))
42395763 1046 (list (read-input-method-name
87505a98 1047 (if default "Select input method (default %s): " "Select input method: ")
42395763 1048 default t))))
d0b9c3ab 1049 (activate-input-method input-method)
42395763 1050 (setq default-input-method input-method))
4ed46869
KH
1051
1052(defun toggle-input-method (&optional arg)
15b3e511 1053 "Turn on or off a multilingual text input method for the current buffer.
723a427a 1054
f2979bdb
KH
1055With no prefix argument, if an input method is currently activated,
1056turn it off. Otherwise, activate an input method -- the one most
1057recently used, or the one specified in `default-input-method', or
1058the one read from the minibuffer.
723a427a 1059
f2979bdb
KH
1060With a prefix argument, read an input method from the minibuffer and
1061turn it on.
723a427a 1062
f2979bdb
KH
1063The default is to use the most recent input method specified
1064\(not including the currently active input method, if any)."
4ed46869 1065 (interactive "P")
7ddbb5bc
RS
1066 (if (and current-input-method (not arg))
1067 (inactivate-input-method)
1068 (let ((default (or (car input-method-history) default-input-method)))
1069 (if (and arg default (equal current-input-method default)
1070 (> (length input-method-history) 1))
1071 (setq default (nth 1 input-method-history)))
723a427a
KH
1072 (activate-input-method
1073 (if (or arg (not default))
7ddbb5bc
RS
1074 (progn
1075 (read-input-method-name
1076 (if default "Input method (default %s): " "Input method: " )
1077 default t))
723a427a
KH
1078 default))
1079 (or default-input-method
1080 (setq default-input-method current-input-method)))))
d0b9c3ab
KH
1081
1082(defun describe-input-method (input-method)
2c395d56 1083 "Describe input method INPUT-METHOD."
d0b9c3ab
KH
1084 (interactive
1085 (list (read-input-method-name
1086 "Describe input method (default, current choice): ")))
78754934 1087 (if (and input-method (symbolp input-method))
4ef06f75 1088 (setq input-method (symbol-name input-method)))
d0b9c3ab
KH
1089 (if (null input-method)
1090 (describe-current-input-method)
464cc130
KH
1091 (let ((current current-input-method))
1092 (condition-case nil
1093 (progn
1094 (save-excursion
1095 (activate-input-method input-method)
1096 (describe-current-input-method))
1097 (activate-input-method current))
1098 (error
1099 (activate-input-method current)
1100 (with-output-to-temp-buffer "*Help*"
1101 (let ((elt (assoc input-method input-method-alist)))
1102 (princ (format
1103 "Input method: %s (`%s' in mode line) for %s\n %s\n"
1104 input-method (nth 3 elt) (nth 1 elt) (nth 4 elt))))))))))
d0b9c3ab
KH
1105
1106(defun describe-current-input-method ()
96db204a 1107 "Describe the input method currently in use."
4ed46869
KH
1108 (if current-input-method
1109 (if (and (symbolp describe-current-input-method-function)
1110 (fboundp describe-current-input-method-function))
1111 (funcall describe-current-input-method-function)
1112 (message "No way to describe the current input method `%s'"
f2979bdb 1113 current-input-method)
4ed46869 1114 (ding))
d0b9c3ab 1115 (error "No input method is activated now")))
4ed46869 1116
d3459641 1117(defun read-multilingual-string (prompt &optional initial-input input-method)
4ed46869
KH
1118 "Read a multilingual string from minibuffer, prompting with string PROMPT.
1119The input method selected last time is activated in minibuffer.
15b3e511 1120If optional second arg INITIAL-INPUT is non-nil, insert it in the minibuffer
d0b9c3ab
KH
1121initially.
1122Optional 3rd argument INPUT-METHOD specifies the input method
4ef06f75
KH
1123to be activated instead of the one selected last time. It is a symbol
1124or a string."
88d559ec
KH
1125 (setq input-method
1126 (or input-method
d3459641 1127 current-input-method
88d559ec
KH
1128 default-input-method
1129 (read-input-method-name "Input method: " nil t)))
3df60841 1130 (if (and input-method (symbolp input-method))
4ef06f75 1131 (setq input-method (symbol-name input-method)))
305a3cb6
KH
1132 (let ((prev-input-method current-input-method))
1133 (unwind-protect
1134 (progn
1135 (activate-input-method input-method)
1136 (read-string prompt initial-input nil nil t))
1137 (activate-input-method prev-input-method))))
4ed46869
KH
1138
1139;; Variables to control behavior of input methods. All input methods
1140;; should react to these variables.
1141
8efc03e1
KH
1142(defcustom input-method-verbose-flag 'default
1143 "*A flag to control extra guidance given by input methods.
1144The value should be nil, t, `complex-only', or `default'.
4ed46869 1145
cb29dfb6 1146The extra guidance is done by showing list of available keys in echo
8efc03e1
KH
1147area. When you use the input method in the minibuffer, the guidance
1148is shown at the bottom short window (split from the existing window).
c27c4ed8 1149
8efc03e1
KH
1150If the value is t, extra guidance is always given, if the value is
1151nil, extra guidance is always suppressed.
1152
1153If the value is `complex-only', only complex input methods such as
1154`chinese-py' and `japanese' give extra guidance.
1155
1156If the value is `default', complex input methods always give extra
1157guidance, but simple input methods give it only when you are not in
1158the minibuffer.
1159
1160See also the variable `input-method-highlight-flag'."
1161 :type '(choice (const t) (const nil) (const complex-only) (const default))
42395763
RS
1162 :group 'mule)
1163
1164(defcustom input-method-highlight-flag t
1165 "*If this flag is non-nil, input methods highlight partially-entered text.
1166For instance, while you are in the middle of a Quail input method sequence,
1167the text inserted so far is temporarily underlined.
8efc03e1
KH
1168The underlining goes away when you finish or abort the input method sequence.
1169See also the variable `input-method-verbose-flag'."
42395763
RS
1170 :type 'boolean
1171 :group 'mule)
4ed46869
KH
1172
1173(defvar input-method-activate-hook nil
f17ccaee
KH
1174 "Normal hook run just after an input method is activated.
1175
1176The variable `current-input-method' keeps the input method name
1177just activated.")
4ed46869
KH
1178
1179(defvar input-method-inactivate-hook nil
f17ccaee
KH
1180 "Normal hook run just after an input method is inactivated.
1181
1182The variable `current-input-method' still keeps the input method name
4d0e6a11 1183just inactivated.")
4ed46869
KH
1184
1185(defvar input-method-after-insert-chunk-hook nil
1186 "Normal hook run just after an input method insert some chunk of text.")
1187
dccca980
KH
1188(defvar input-method-exit-on-first-char nil
1189 "This flag controls a timing when an input method returns.
1190Usually, the input method does not return while there's a possibility
1191that it may find a different translation if a user types another key.
39e643e2
RS
1192But, it this flag is non-nil, the input method returns as soon as
1193the current key sequence gets long enough to have some valid translation.")
dccca980
KH
1194
1195(defvar input-method-use-echo-area nil
1196 "This flag controls how an input method shows an intermediate key sequence.
39e643e2
RS
1197Usually, the input method inserts the intermediate key sequence,
1198or candidate translations corresponding to the sequence,
1199at point in the current buffer.
1200But, if this flag is non-nil, it displays them in echo area instead.")
dccca980 1201
723a427a
KH
1202(defvar input-method-exit-on-invalid-key nil
1203 "This flag controls the behaviour of an input method on invalid key input.
1204Usually, when a user types a key which doesn't start any character
1205handled by the input method, the key is handled by turning off the
e8dd0160 1206input method temporarily. After that key, the input method is re-enabled.
723a427a
KH
1207But, if this flag is non-nil, the input method is never back on.")
1208
4ed46869 1209\f
8efc03e1
KH
1210(defvar set-language-environment-hook nil
1211 "Normal hook run after some language environment is set.
1212
1213When you set some hook function here, that effect usually should not
1214be inherited to another language environment. So, you had better set
1215another function in `exit-language-environment-hook' (which see) to
1216cancel the effect.")
1217
1218(defvar exit-language-environment-hook nil
1219 "Normal hook run after exiting from some language environment.
1220When this hook is run, the variable `current-language-environment'
1221is still bound to the language environment being exited.
1222
e8dd0160 1223This hook is mainly used for canceling the effect of
8efc03e1
KH
1224`set-language-environment-hook' (which-see).")
1225
b0648a00
RS
1226(put 'setup-specified-language-environment 'apropos-inhibit t)
1227
15b3e511 1228(defun setup-specified-language-environment ()
f08adf27 1229 "Switch to a specified language environment."
15b3e511 1230 (interactive)
f850d782 1231 (let (language-name)
15b3e511
KH
1232 (if (and (symbolp last-command-event)
1233 (or (not (eq last-command-event 'Default))
1234 (setq last-command-event 'English))
f850d782
RS
1235 (setq language-name (symbol-name last-command-event)))
1236 (set-language-environment language-name)
15b3e511 1237 (error "Bogus calling sequence"))))
4ed46869 1238
8861c593 1239(defcustom current-language-environment "English"
94d04df6 1240 "The last language environment specified with `set-language-environment'.
ebef6d93
KH
1241This variable should be set only with \\[customize], which is equivalent
1242to using the function `set-language-environment'."
94d04df6 1243 :link '(custom-manual "(emacs)Language Environments")
dff1aa24 1244 :set (lambda (symbol value) (set-language-environment value))
94d04df6
DL
1245 :get (lambda (x)
1246 (or (car-safe (assoc-ignore-case
1247 (if (symbolp current-language-environment)
1248 (symbol-name current-language-environment)
1249 current-language-environment)
1250 language-info-alist))
1251 "English"))
1252 :type (cons 'choice (mapcar (lambda (lang)
1253 (list 'const (car lang)))
1254 language-info-alist))
8861c593
RS
1255 :initialize 'custom-initialize-default
1256 :group 'mule
1257 :type 'string)
f850d782 1258
ddb5c041
KH
1259(defun reset-language-environment ()
1260 "Reset multilingual environment of Emacs to the default status.
1261
1262The default status is as follows:
1263
1264 The default value of buffer-file-coding-system is nil.
1265 The default coding system for process I/O is nil.
1266 The default value for the command `set-terminal-coding-system' is nil.
1267 The default value for the command `set-keyboard-coding-system' is nil.
1268
1269 The order of priorities of coding categories and the coding system
1270 bound to each category are as follows
1271 coding category coding system
1272 --------------------------------------------------
1273 coding-category-iso-8-2 iso-latin-1
1274 coding-category-iso-8-1 iso-latin-1
1275 coding-category-iso-7-tight iso-2022-jp
1276 coding-category-iso-7 iso-2022-7bit
1277 coding-category-iso-7-else iso-2022-7bit-lock
1278 coding-category-iso-8-else iso-2022-8bit-ss2
1279 coding-category-emacs-mule emacs-mule
1280 coding-category-raw-text raw-text
1281 coding-category-sjis japanese-shift-jis
1282 coding-category-big5 chinese-big5
1283 coding-category-ccl nil
e8dd0160 1284 coding-category-binary no-conversion
ddb5c041
KH
1285"
1286 (interactive)
1287 ;; This function formerly set default-enable-multibyte-characters to t,
1288 ;; but that is incorrect. It should not alter the unibyte/multibyte choice.
1289
1290 (setq coding-category-iso-7-tight 'iso-2022-jp
1291 coding-category-iso-7 'iso-2022-7bit
1292 coding-category-iso-8-1 'iso-latin-1
1293 coding-category-iso-8-2 'iso-latin-1
1294 coding-category-iso-7-else 'iso-2022-7bit-lock
1295 coding-category-iso-8-else 'iso-2022-8bit-ss2
1296 coding-category-emacs-mule 'emacs-mule
1297 coding-category-raw-text 'raw-text
1298 coding-category-sjis 'japanese-shift-jis
1299 coding-category-big5 'chinese-big5
9bfcd269
KH
1300 coding-category-utf-8 nil
1301 coding-category-utf-16-be nil
1302 coding-category-utf-16-le nil
ddb5c041
KH
1303 coding-category-ccl nil
1304 coding-category-binary 'no-conversion)
1305
1306 (set-coding-priority
1307 '(coding-category-iso-8-1
1308 coding-category-iso-8-2
1309 coding-category-iso-7-tight
1310 coding-category-iso-7
1311 coding-category-iso-7-else
1312 coding-category-iso-8-else
1313 coding-category-emacs-mule
1314 coding-category-raw-text
1315 coding-category-sjis
1316 coding-category-big5
1317 coding-category-ccl
9bfcd269
KH
1318 coding-category-binary
1319 coding-category-utf-16-be
1320 coding-category-utf-16-le
1321 coding-category-utf-8))
ddb5c041 1322
91693d18
KH
1323 (update-coding-systems-internal)
1324
ddb5c041 1325 (set-default-coding-systems nil)
b5edd1d1 1326 (setq default-sendmail-coding-system 'iso-latin-1)
0c47a7c8 1327 (setq default-process-coding-system '(undecided . iso-latin-1))
b5edd1d1 1328
ddb5c041
KH
1329 ;; Don't alter the terminal and keyboard coding systems here.
1330 ;; The terminal still supports the same coding system
1331 ;; that it supported a minute ago.
1332;;; (set-terminal-coding-system-internal nil)
1333;;; (set-keyboard-coding-system-internal nil)
1334
1335 (setq nonascii-translation-table nil
1336 nonascii-insert-offset 0))
1337
0c47a7c8
KH
1338(reset-language-environment)
1339
40c81f74
PE
1340(defun set-display-table-and-terminal-coding-system (language-name)
1341 "Set up the display table and terminal coding system for LANGUAGE-NAME."
1342 (let ((coding (get-language-info language-name 'unibyte-display)))
1343 (if coding
1344 (standard-display-european-internal)
1345 (standard-display-default (if (eq window-system 'pc) 128 160) 255)
1346 (aset standard-display-table 146 nil))
1347 (or (eq window-system 'pc)
1348 (set-terminal-coding-system coding))))
1349
166246f7 1350(defun set-language-environment (language-name)
6c05d680
RS
1351 "Set up multi-lingual environment for using LANGUAGE-NAME.
1352This sets the coding system priority and the default input method
8861c593
RS
1353and sometimes other things. LANGUAGE-NAME should be a string
1354which is the name of a language environment. For example, \"Latin-1\"
1355specifies the character set for the major languages of Western Europe."
8efc03e1 1356 (interactive (list (read-language-name
ddb5c041 1357 nil
8efc03e1 1358 "Set language environment (default, English): ")))
4ef06f75
KH
1359 (if language-name
1360 (if (symbolp language-name)
1361 (setq language-name (symbol-name language-name)))
1362 (setq language-name "English"))
ddb5c041 1363 (or (assoc-ignore-case language-name language-info-alist)
f850d782 1364 (error "Language environment not defined: %S" language-name))
8efc03e1
KH
1365 (if current-language-environment
1366 (let ((func (get-language-info current-language-environment
1367 'exit-function)))
e63645c2
KH
1368 (run-hooks 'exit-language-environment-hook)
1369 (if (fboundp func) (funcall func))))
03c35c83
EZ
1370 (let ((default-eol-type (coding-system-eol-type
1371 default-buffer-file-coding-system)))
1372 (reset-language-environment)
ddb5c041 1373
03c35c83
EZ
1374 (setq current-language-environment language-name)
1375 (set-language-environment-coding-systems language-name default-eol-type))
ddb5c041
KH
1376 (let ((input-method (get-language-info language-name 'input-method)))
1377 (when input-method
1378 (setq default-input-method input-method)
1379 (if input-method-history
1380 (setq input-method-history
1381 (cons input-method
1382 (delete input-method input-method-history))))))
ec241f58
EZ
1383 (let ((nonascii (get-language-info language-name 'nonascii-translation))
1384 (dos-table
4e2ac2d9
EZ
1385 (if (eq window-system 'pc)
1386 (intern
d9c0a50e 1387 (format "cp%d-nonascii-translation-table" dos-codepage)))))
03c35c83
EZ
1388 (cond
1389 ((char-table-p nonascii)
1390 (setq nonascii-translation-table nonascii))
ec241f58 1391 ((and (eq window-system 'pc) (boundp dos-table))
03c35c83
EZ
1392 ;; DOS terminals' default is to use a special non-ASCII translation
1393 ;; table as appropriate for the installed codepage.
ec241f58 1394 (setq nonascii-translation-table (symbol-value dos-table)))
03c35c83
EZ
1395 ((charsetp nonascii)
1396 (setq nonascii-insert-offset (- (make-char nonascii) 128)))))
ddb5c041 1397
63283a8f 1398 ;; Unibyte setups if necessary.
ddb5c041 1399 (unless default-enable-multibyte-characters
63283a8f 1400 ;; Syntax and case table.
ddb5c041
KH
1401 (let ((syntax (get-language-info language-name 'unibyte-syntax)))
1402 (if syntax
1403 (let ((set-case-syntax-set-multibyte nil))
63283a8f
KH
1404 (load syntax nil t))
1405 ;; No information for syntax and case. Reset to the defaults.
1406 (let ((syntax-table (standard-syntax-table))
1407 (case-table (standard-case-table))
03c35c83 1408 (ch (if (eq window-system 'pc) 128 160)))
63283a8f
KH
1409 (while (< ch 256)
1410 (modify-syntax-entry ch " " syntax-table)
1411 (aset case-table ch ch)
1412 (setq ch (1+ ch)))
1413 (set-char-table-extra-slot case-table 0 nil)
1414 (set-char-table-extra-slot case-table 1 nil)
1415 (set-char-table-extra-slot case-table 2 nil))
1416 (set-standard-case-table (standard-case-table))
1417 (let ((list (buffer-list)))
1418 (while list
1419 (with-current-buffer (car list)
1420 (set-case-table (standard-case-table)))
1421 (setq list (cdr list))))))
40c81f74 1422 (set-display-table-and-terminal-coding-system language-name))
ddb5c041
KH
1423
1424 (let ((required-features (get-language-info language-name 'features)))
1425 (while required-features
1426 (require (car required-features))
1427 (setq required-features (cdr required-features))))
1428 (let ((func (get-language-info language-name 'setup-function)))
1429 (if (fboundp func)
1430 (funcall func)))
8efc03e1 1431 (run-hooks 'set-language-environment-hook)
f850d782 1432 (force-mode-line-update t))
4ed46869 1433
51a8fc1d
RS
1434(defun standard-display-european-internal ()
1435 ;; Actually set up direct output of non-ASCII characters.
03c35c83
EZ
1436 (standard-display-8bit (if (eq window-system 'pc) 128 160) 255)
1437 ;; Unibyte Emacs on MS-DOS wants to display all 8-bit characters with
1438 ;; the native font, and codes 160 and 146 stand for something very
1439 ;; different there.
1440 (or (and (eq window-system 'pc) (not default-enable-multibyte-characters))
1441 (progn
1442 ;; Make non-line-break space display as a plain space.
1443 ;; Most X fonts do the wrong thing for code 160.
1444 (aset standard-display-table 160 [32])
1445 ;; Most Windows programs send out apostrophe's as \222. Most X fonts
1446 ;; don't contain a character at that position. Map it to the ASCII
1447 ;; apostrophe.
1448 (aset standard-display-table 146 [39]))))
1449
1450(defun set-language-environment-coding-systems (language-name
1451 &optional eol-type)
1452 "Do various coding system setups for language environment LANGUAGE-NAME.
1453
1454The optional arg EOL-TYPE specifies the eol-type of the default value
1455of buffer-file-coding-system set by this function."
54b226f7
KH
1456 (let* ((priority (get-language-info language-name 'coding-priority))
1457 (default-coding (car priority)))
1458 (if priority
1459 (let ((categories (mapcar 'coding-system-category priority)))
03c35c83
EZ
1460 (set-default-coding-systems
1461 (if (memq eol-type '(0 1 2 unix dos mac))
1462 (coding-system-change-eol-conversion default-coding eol-type)
1463 default-coding))
46babb23 1464 (setq default-sendmail-coding-system default-coding)
54b226f7
KH
1465 (set-coding-priority categories)
1466 (while priority
1467 (set (car categories) (car priority))
1468 (setq priority (cdr priority) categories (cdr categories)))
45d08cb2 1469 (update-coding-systems-internal)))))
54b226f7 1470
4ed46869
KH
1471;; Print all arguments with `princ', then print "\n".
1472(defsubst princ-list (&rest args)
1473 (while args (princ (car args)) (setq args (cdr args)))
1474 (princ "\n"))
1475
b0648a00
RS
1476(put 'describe-specified-language-support 'apropos-inhibit t)
1477
48082651 1478;; Print a language specific information such as input methods,
13e82c04 1479;; charsets, and coding systems. This function is intended to be
48082651 1480;; called from the menu:
281d03ec 1481;; [menu-bar mule describe-language-environment LANGUAGE]
48082651
KH
1482;; and should not run it by `M-x describe-current-input-method-function'.
1483(defun describe-specified-language-support ()
96db204a 1484 "Describe how Emacs supports the specified language environment."
48082651 1485 (interactive)
281d03ec 1486 (let (language-name)
48082651 1487 (if (not (and (symbolp last-command-event)
cda74479
DL
1488 (or (not (eq last-command-event 'Default))
1489 (setq last-command-event 'English))
281d03ec 1490 (setq language-name (symbol-name last-command-event))))
48082651 1491 (error "Bogus calling sequence"))
281d03ec
RS
1492 (describe-language-environment language-name)))
1493
1494(defun describe-language-environment (language-name)
1495 "Describe how Emacs supports language environment LANGUAGE-NAME."
78754934
KH
1496 (interactive
1497 (list (read-language-name
1498 'documentation
8adfa8be 1499 "Describe language environment (default, current choice): ")))
f850d782
RS
1500 (if (null language-name)
1501 (setq language-name current-language-environment))
281d03ec
RS
1502 (if (or (null language-name)
1503 (null (get-language-info language-name 'documentation)))
1504 (error "No documentation for the specified language"))
4ef06f75
KH
1505 (if (symbolp language-name)
1506 (setq language-name (symbol-name language-name)))
464cc130
KH
1507 (let ((doc (get-language-info language-name 'documentation))
1508 pos)
48082651 1509 (with-output-to-temp-buffer "*Help*"
464cc130
KH
1510 (save-excursion
1511 (set-buffer standard-output)
1512 (insert language-name " language environment\n\n")
1513 (if (stringp doc)
1514 (insert doc "\n\n"))
e036b0a6
KH
1515 (condition-case nil
1516 (let ((str (eval (get-language-info language-name 'sample-text))))
1517 (if (stringp str)
1518 (insert "Sample text:\n " str "\n\n")))
1519 (error nil))
464cc130
KH
1520 (let ((input-method (get-language-info language-name 'input-method))
1521 (l (copy-sequence input-method-alist)))
1522 (insert "Input methods")
1523 (when input-method
1524 (insert " (default, " input-method ")")
1525 (setq input-method (assoc input-method input-method-alist))
1526 (setq l (cons input-method (delete input-method l))))
1527 (insert ":\n")
48082651 1528 (while l
464cc130
KH
1529 (when (string= language-name (nth 1 (car l)))
1530 (insert " " (car (car l)))
1531 (search-backward (car (car l)))
1532 (help-xref-button 0 #'describe-input-method (car (car l))
1533 "mouse-2, RET: describe this input method")
1534 (goto-char (point-max))
1535 (insert " (\"" (nth 3 (car l)) "\" in mode line)\n"))
1536 (setq l (cdr l)))
1537 (insert "\n"))
1538 (insert "Character sets:\n")
1539 (let ((l (get-language-info language-name 'charset)))
1540 (if (null l)
1541 (insert " nothing specific to " language-name "\n")
1542 (while l
1543 (insert " " (symbol-name (car l)))
1544 (search-backward (symbol-name (car l)))
1545 (help-xref-button 0 #'describe-character-set (car l)
1546 "mouse-2, RET: describe this character set")
1547 (goto-char (point-max))
1548 (insert ": " (charset-description (car l)) "\n")
1549 (setq l (cdr l)))))
1550 (insert "\n")
1551 (insert "Coding systems:\n")
1552 (let ((l (get-language-info language-name 'coding-system)))
1553 (if (null l)
1554 (insert " nothing specific to " language-name "\n")
1555 (while l
1556 (insert " " (symbol-name (car l)))
1557 (search-backward (symbol-name (car l)))
1558 (help-xref-button 0 #'describe-coding-system (car l)
1559 "mouse-2, RET: describe this coding system")
1560 (goto-char (point-max))
1561 (insert " (`"
1562 (coding-system-mnemonic (car l))
1563 "' in mode line):\n\t"
1564 (coding-system-doc-string (car l))
1565 "\n")
1566 (let ((aliases (coding-system-get (car l)
1567 'alias-coding-systems)))
1568 (when aliases
1569 (insert "\t(alias:")
1570 (while aliases
1571 (insert " " (symbol-name (car aliases)))
1572 (setq aliases (cdr aliases)))
1573 (insert ")\n")))
1574 (setq l (cdr l)))))
1575 (help-setup-xref (list #'describe-language-environment language-name)
1576 (interactive-p))))))
4ed46869 1577\f
40c81f74
PE
1578;;; Locales.
1579
0d7c5bb9
DL
1580(defvar locale-translation-file-name nil
1581 "File name for the system's file of locale-name aliases, or nil if none.")
40c81f74
PE
1582
1583(defvar locale-language-names
1584 '(
1585 ;; UTF-8 is not yet implemented.
1586 ;; Put this first, so that e.g. "ko.UTF-8" does not match "ko" below.
1587 (".*[._]utf" . nil)
1588
1589 ;; Locale names of the form LANGUAGE[_TERRITORY][.CODESET][@MODIFIER]
1590 ;; as specified in the Single Unix Spec, Version 2.
1591 ;; LANGUAGE is a language code taken from ISO 639:1988 (E/F)
1592 ;; with additions from ISO 639/RA Newsletter No.1/1989;
1593 ;; see Internet RFC 2165 (1997-06).
1594 ;; TERRITORY is a country code taken from ISO 3166.
1595 ;; CODESET and MODIFIER are implementation-dependent.
1596 ;;
1597 ; aa Afar
1598 ; ab Abkhazian
6ececc4d 1599 ("af" . "Latin-1") ; Afrikaans
40c81f74
PE
1600 ("am" . "Ethiopic") ; Amharic
1601 ; ar Arabic
1602 ; as Assamese
1603 ; ay Aymara
1604 ; az Azerbaijani
1605 ; ba Bashkir
6ececc4d
PE
1606 ("be" . "Latin-5") ; Byelorussian
1607 ("bg" . "Latin-5") ; Bulgarian
40c81f74
PE
1608 ; bh Bihari
1609 ; bi Bislama
1610 ; bn Bengali, Bangla
1611 ("bo" . "Tibetan")
1612 ("br" . "Latin-1") ; Breton
1613 ("ca" . "Latin-1") ; Catalan
1614 ; co Corsican
1615 ("cs" . "Czech")
6ececc4d 1616 ("cy" . "Latin-8") ; Welsh
40c81f74
PE
1617 ("da" . "Latin-1") ; Danish
1618 ("de" . "German")
1619 ; dz Bhutani
1620 ("el" . "Greek")
6ececc4d
PE
1621 ;; Users who specify "en" explicitly typically want Latin-1, not ASCII.
1622 ("en" . "Latin-1") ; English
40c81f74
PE
1623 ("eo" . "Latin-3") ; Esperanto
1624 ("es" . "Latin-1") ; Spanish
1625 ("et" . "Latin-4") ; Estonian
1626 ("eu" . "Latin-1") ; Basque
1627 ; fa Persian
1628 ("fi" . "Latin-1") ; Finnish
1629 ; fj Fiji
1630 ("fo" . "Latin-1") ; Faroese
1631 ("fr" . "Latin-1") ; French
1632 ("fy" . "Latin-1") ; Frisian
6ececc4d
PE
1633 ("ga" . "Latin-1") ; Irish Gaelic (new orthography)
1634 ("gd" . "Latin-1") ; Scots Gaelic
1635 ("gl" . "Latin-1") ; Galician
40c81f74
PE
1636 ; gn Guarani
1637 ; gu Gujarati
1638 ; ha Hausa
1639 ("he" . "Hebrew")
1640 ("hi" . "Devanagari") ; Hindi
1641 ("hr" . "Latin-2") ; Croatian
1642 ("hu" . "Latin-2") ; Hungarian
1643 ; hy Armenian
1644 ; ia Interlingua
1645 ("id" . "Latin-1") ; Indonesian
1646 ; ie Interlingue
1647 ; ik Inupiak
1648 ("is" . "Latin-1") ; Icelandic
1649 ("it" . "Latin-1") ; Italian
1650 ; iu Inuktitut
1651 ("ja" . "Japanese")
1652 ; jw Javanese
1653 ; ka Georgian
1654 ; kk Kazakh
6ececc4d 1655 ("kl" . "Latin-1") ; Greenlandic
40c81f74
PE
1656 ; km Cambodian
1657 ; kn Kannada
1658 ("ko" . "Korean")
1659 ; ks Kashmiri
1660 ; ku Kurdish
1661 ; ky Kirghiz
1662 ("la" . "Latin-1") ; Latin
1663 ; ln Lingala
1664 ("lo" . "Lao") ; Laothian
1665 ("lt" . "Latin-4") ; Lithuanian
1666 ("lv" . "Latin-4") ; Latvian, Lettish
1667 ; mg Malagasy
1668 ; mi Maori
6ececc4d 1669 ("mk" . "Latin-5") ; Macedonian
40c81f74
PE
1670 ; ml Malayalam
1671 ; mn Mongolian
1672 ; mo Moldavian
1673 ("mr" . "Devanagari") ; Marathi
1674 ; ms Malay
1675 ("mt" . "Latin-3") ; Maltese
1676 ; my Burmese
1677 ; na Nauru
1678 ("ne" . "Devanagari") ; Nepali
1679 ("nl" . "Latin-1") ; Dutch
1680 ("no" . "Latin-1") ; Norwegian
1681 ; oc Occitan
1682 ; om (Afan) Oromo
1683 ; or Oriya
1684 ; pa Punjabi
1685 ("pl" . "Latin-2") ; Polish
1686 ; ps Pashto, Pushto
1687 ("pt" . "Latin-1") ; Portuguese
1688 ; qu Quechua
6ececc4d 1689 ("rm" . "Latin-1") ; Rhaeto-Romanic
40c81f74
PE
1690 ; rn Kirundi
1691 ("ro" . "Romanian")
1692 ("ru.*[_.]koi8" . "Cyrillic-KOI8") ; Russian
6ececc4d 1693 ("ru" . "Latin-5") ; Russian
40c81f74
PE
1694 ; rw Kinyarwanda
1695 ("sa" . "Devanagari") ; Sanskrit
1696 ; sd Sindhi
1697 ; sg Sangho
1698 ("sh" . "Latin-2") ; Serbo-Croatian
1699 ; si Sinhalese
1700 ("sk" . "Slovak")
1701 ("sl" . "Slovenian")
1702 ; sm Samoan
1703 ; sn Shona
1704 ; so Somali
6ececc4d 1705 ("sq" . "Latin-1") ; Albanian
40c81f74
PE
1706 ("sr" . "Latin-2") ; Serbian (Latin alphabet)
1707 ; ss Siswati
1708 ; st Sesotho
1709 ; su Sundanese
1710 ("sv" . "Latin-1") ; Swedish
1711 ("sw" . "Latin-1") ; Swahili
1712 ; ta Tamil
1713 ; te Telugu
1714 ; tg Tajik
1715 ("th" . "Thai")
1716 ; ti Tigrinya
1717 ; tk Turkmen
6ececc4d 1718 ("tl" . "Latin-1") ; Tagalog
40c81f74
PE
1719 ; tn Setswana
1720 ; to Tonga
1721 ("tr" . "Latin-5") ; Turkish
1722 ; ts Tsonga
1723 ; tt Tatar
1724 ; tw Twi
1725 ; ug Uighur
6ececc4d 1726 ("uk" . "Latin-5") ; Ukrainian
40c81f74
PE
1727 ; ur Urdu
1728 ; uz Uzbek
1729 ("vi" . "Vietnamese")
1730 ; vo Volapuk
1731 ; wo Wolof
1732 ; xh Xhosa
1733 ; yi Yiddish
1734 ; yo Yoruba
1735 ; za Zhuang
1736 ("zh.*[._]big5" . "Chinese-BIG5")
1737 ("zh.*[._]gbk" . nil) ; Solaris 2.7; has gbk-0 as well as GB 2312.1980-0
1738 ("zh_tw" . "Chinese-CNS")
1739 ("zh" . "Chinese-GB")
1740 ; zu Zulu
1741
1742 ;; ISO standard locales
1743 ("c$" . "ASCII")
1744 ("posix$" . "ASCII")
1745
40c81f74
PE
1746 ;; The "IPA" Emacs language environment does not correspond
1747 ;; to any ISO 639 code, so let it stand for itself.
1748 ("ipa$" . "IPA")
1749
1750 ;; Nonstandard or obsolete language codes
1751 ("cz" . "Czech") ; e.g. Solaris 2.6
1752 ("ee" . "Latin-4") ; Estonian, e.g. X11R6.4
1753 ("iw" . "Hebrew") ; e.g. X11R6.4
6ececc4d 1754 ("sp" . "Latin-5") ; Serbian (Cyrillic alphabet), e.g. X11R6.4
40c81f74
PE
1755 ("su" . "Latin-1") ; Finnish, e.g. Solaris 2.6
1756 )
1757 "List of pairs of locale regexps and language names.
6ececc4d
PE
1758The first element whose locale regexp matches the start of a downcased locale
1759specifies the language name corresponding to that locale.
40c81f74
PE
1760If the language name is nil, there is no corresponding language environment.")
1761
6ececc4d
PE
1762(defvar locale-charset-language-names
1763 '((".*8859[-_]?1\\>" . "Latin-1")
1764 (".*8859[-_]?2\\>" . "Latin-2")
1765 (".*8859[-_]?3\\>" . "Latin-3")
1766 (".*8859[-_]?4\\>" . "Latin-4")
1767 (".*8859[-_]?9\\>" . "Latin-5")
1768 (".*8859[-_]?14\\>" . "Latin-8")
1769 (".*8859[-_]?15\\>" . "Latin-9")
1770 )
1771 "List of pairs of locale regexps and charset language names.
1772The first element whose locale regexp matches the start of a downcased locale
1773specifies the language name whose charsets corresponds to that locale.
1774This language name is used if its charsets disagree with the charsets of
1775the language name that would otherwise be used for this locale.")
1776
40c81f74
PE
1777(defvar locale-preferred-coding-systems
1778 '(("ja.*[._]euc" . japanese-iso-8bit)
1779 ("ja.*[._]jis7" . iso-2022-jp)
1780 ("ja.*[._]pck" . japanese-shift-jis)
1781 ("ja.*[._]sjis" . japanese-shift-jis)
40c81f74 1782 )
6ececc4d
PE
1783 "List of pairs of locale regexps and preferred coding systems.
1784The first element whose locale regexp matches the start of a downcased locale
1785specifies the coding system to prefer when using that locale.")
40c81f74
PE
1786
1787(defun locale-name-match (key alist)
1788 "Search for KEY in ALIST, which should be a list of regexp-value pairs.
1789Return the value corresponding to the first regexp that matches the
1790start of KEY, or nil if there is no match."
1791 (let (element)
1792 (while (and alist (not element))
1793 (if (string-match (concat "^\\(" (car (car alist)) "\\)") key)
1794 (setq element (car alist)))
1795 (setq alist (cdr alist)))
1796 (cdr element)))
1797
1798(defun set-locale-environment (locale-name)
1799 "Set up multi-lingual environment for using LOCALE-NAME.
1800This sets the coding system priority and the default input method
1801and sometimes other things. LOCALE-NAME should be a string
1802which is the name of a locale supported by the system;
1803often it is of the form xx_XX.CODE, where xx is a language,
1804XX is a country, and CODE specifies a character set and coding system.
1805For example, the locale name \"ja_JP.EUC\" might name a locale
1806for Japanese in Japan using the `japanese-iso-8bit' coding-system.
1807
1808If LOCALE-NAME is nil, its value is taken from the environment.
1809
1810The locale names supported by your system can typically be found in a
0812c1e8
DL
1811directory named `/usr/share/locale' or `/usr/lib/locale'. LOCALE-NAME
1812may be translated according to the table specified by
1813`locale-translation-file-name'.
1814
1815See also `locale-charset-language-names', `locale-language-names',
1816`locale-preferred-coding-systems' and `locale-coding-system'."
40c81f74 1817
0d7c5bb9
DL
1818 ;; Do this at runtime for the sake of binaries possibly transported
1819 ;; to a system without X.
1820 (setq locale-translation-file-name
1821 (let ((files
1822 '("/usr/lib/X11/locale/locale.alias" ; e.g. X11R6.4
1823 "/usr/X11R6/lib/X11/locale/locale.alias" ; e.g. RedHat 4.2
1824 "/usr/openwin/lib/locale/locale.alias" ; e.g. Solaris 2.6
1825 ;;
1826 ;; The following name appears after the X-related names above,
1827 ;; since the X-related names are what X actually uses.
1828 "/usr/share/locale/locale.alias" ; GNU/Linux sans X
1829 )))
1830 (while (and files (not (file-exists-p (car files))))
1831 (setq files (cdr files)))
1832 (car files)))
1833
40c81f74
PE
1834 (unless locale-name
1835 ;; Use the first of these three environment variables
1836 ;; that has a nonempty value.
1837 (let ((vars '("LC_ALL" "LC_CTYPE" "LANG")))
1838 (while (and vars (not (setq locale-name (getenv (car vars)))))
1839 (setq vars (cdr vars)))))
1840
1841 (when locale-name
1842
1843 ;; Translate "swedish" into "sv_SE.ISO8859-1", and so on,
1844 ;; using the translation file that many systems have.
1845 (when locale-translation-file-name
1846 (with-temp-buffer
1847 (insert-file-contents locale-translation-file-name)
1848 (when (re-search-forward
1849 (concat "^" (regexp-quote locale-name) ":?[ \t]+") nil t)
1850 (setq locale-name (buffer-substring (point) (line-end-position))))))
1851
1852 (setq locale-name (downcase locale-name))
1853
6ececc4d
PE
1854 (let ((language-name
1855 (locale-name-match locale-name locale-language-names))
1856 (charset-language-name
1857 (locale-name-match locale-name locale-charset-language-names))
1858 (coding-system
1859 (locale-name-match locale-name locale-preferred-coding-systems)))
1860
1861 (if (and charset-language-name
1862 (not
1863 (equal (get-language-info language-name 'charset)
1864 (get-language-info charset-language-name 'charset))))
1865 (setq language-name charset-language-name))
1866
40c81f74
PE
1867 (when language-name
1868
1869 ;; Set up for this character set. This is now the right way
1870 ;; to do it for both unibyte and multibyte modes.
1871 (set-language-environment language-name)
1872
1873 ;; If default-enable-multibyte-characters is nil,
1874 ;; we are using single-byte characters,
1875 ;; so the display table and terminal coding system are irrelevant.
1876 (when default-enable-multibyte-characters
1877 (set-display-table-and-terminal-coding-system language-name))
1878
1879 (setq locale-coding-system
1880 (car (get-language-info language-name 'coding-priority))))
1881
1882 (when coding-system
1883 (prefer-coding-system coding-system)
1884 (setq locale-coding-system coding-system)))))
1885\f
4ed46869
KH
1886;;; Charset property
1887
251d4f4b 1888(defun get-charset-property (charset propname)
4ed46869
KH
1889 "Return the value of CHARSET's PROPNAME property.
1890This is the last value stored with
96db204a 1891 (put-charset-property CHARSET PROPNAME VALUE)."
251d4f4b
KH
1892 (and (not (eq charset 'composition))
1893 (plist-get (charset-plist charset) propname)))
4ed46869 1894
251d4f4b 1895(defun put-charset-property (charset propname value)
4ed46869
KH
1896 "Store CHARSETS's PROPNAME property with value VALUE.
1897It can be retrieved with `(get-charset-property CHARSET PROPNAME)'."
1300d43f
KH
1898 (or (eq charset 'composition)
1899 (set-charset-plist charset
1900 (plist-put (charset-plist charset) propname value))))
4ed46869
KH
1901
1902;;; Character code property
1903(put 'char-code-property-table 'char-table-extra-slots 0)
1904
1905(defvar char-code-property-table
1906 (make-char-table 'char-code-property-table)
1907 "Char-table containing a property list of each character code.
1908
1909See also the documentation of `get-char-code-property' and
96db204a 1910`put-char-code-property'.")
4ed46869
KH
1911
1912(defun get-char-code-property (char propname)
1913 "Return the value of CHAR's PROPNAME property in `char-code-property-table'."
1914 (let ((plist (aref char-code-property-table char)))
1915 (if (listp plist)
1916 (car (cdr (memq propname plist))))))
1917
1918(defun put-char-code-property (char propname value)
1919 "Store CHAR's PROPNAME property with VALUE in `char-code-property-table'.
1920It can be retrieved with `(get-char-code-property CHAR PROPNAME)'."
1921 (let ((plist (aref char-code-property-table char)))
1922 (if plist
1923 (let ((slot (memq propname plist)))
1924 (if slot
1925 (setcar (cdr slot) value)
1926 (nconc plist (list propname value))))
1927 (aset char-code-property-table char (list propname value)))))
1928
a127b764
KH
1929\f
1930;; Pretty description of encoded string
1931
1932;; Alist of ISO 2022 control code vs the corresponding mnemonic string.
1933(defvar iso-2022-control-alist
1934 '((?\x1b . "ESC")
1935 (?\x0e . "SO")
1936 (?\x0f . "SI")
1937 (?\x8e . "SS2")
1938 (?\x8f . "SS3")
1939 (?\x9b . "CSI")))
1940
1941(defun encoded-string-description (str coding-system)
1942 "Return a pretty description of STR that is encoded by CODING-SYSTEM."
1943 (setq str (string-as-unibyte str))
993b2a7d 1944 (mapconcat
8c9d55a9 1945 (if (and coding-system (eq (coding-system-type coding-system) 2))
993b2a7d
KH
1946 ;; Try to get a pretty description for ISO 2022 escape sequences.
1947 (function (lambda (x) (or (cdr (assq x iso-2022-control-alist))
1948 (format "%02X" x))))
5de75f53 1949 (function (lambda (x) (format "0x%02X" x))))
993b2a7d 1950 str " "))
a127b764
KH
1951
1952(defun encode-coding-char (char coding-system)
1953 "Encode CHAR by CODING-SYSTEM and return the resulting string.
1954If CODING-SYSTEM can't safely encode CHAR, return nil."
bd953173
KH
1955 (let ((str1 (string-as-multibyte (char-to-string char)))
1956 (str2 (string-as-multibyte (make-string 2 char)))
a89f541b
KH
1957 (safe-chars (and coding-system
1958 (coding-system-get coding-system 'safe-chars)))
8c9d55a9 1959 (charset (char-charset char))
a127b764 1960 enc1 enc2 i1 i2)
a89f541b 1961 (when (or (eq safe-chars t)
8c9d55a9 1962 (eq charset 'ascii)
a89f541b 1963 (and safe-chars (aref safe-chars char)))
a127b764
KH
1964 ;; We must find the encoded string of CHAR. But, just encoding
1965 ;; CHAR will put extra control sequences (usually to designate
1966 ;; ASCII charaset) at the tail if type of CODING is ISO 2022.
1967 ;; To exclude such tailing bytes, we at first encode one-char
1968 ;; string and two-char string, then check how many bytes at the
1969 ;; tail of both encoded strings are the same.
1970
bd953173 1971 (setq enc1 (encode-coding-string str1 coding-system)
a127b764 1972 i1 (length enc1)
bd953173 1973 enc2 (encode-coding-string str2 coding-system)
a127b764
KH
1974 i2 (length enc2))
1975 (while (and (> i1 0) (= (aref enc1 (1- i1)) (aref enc2 (1- i2))))
1976 (setq i1 (1- i1) i2 (1- i2)))
1977
1978 ;; Now (substring enc1 i1) and (substring enc2 i2) are the same,
1979 ;; and they are the extra control sequences at the tail to
1980 ;; exclude.
1981 (substring enc2 0 i2))))
1982
1983
4ed46869 1984;;; mule-cmds.el ends here