(mail-mode-fill-paragraph): Do not get the
[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)
2f1fa038 601 (insert "The following default coding systems were tried:\n")
b5edd1d1
KH
602 (let ((pos (point))
603 (fill-prefix " "))
604 (mapcar (function (lambda (x) (princ " ") (princ (car x))))
605 default-coding-system)
606 (insert "\n")
607 (fill-region-as-paragraph pos (point)))
2f1fa038
EZ
608 (insert
609 (if (consp coding-system)
610 (concat (format "%s safely encodes the target text,\n"
611 (car coding-system))
612 "but it is not recommended for encoding text in this context,\n"
613 "e.g., for sending an email message.\n")
614 "However, none of them safely encodes the target text.\n"))
b5edd1d1 615 (insert (if (consp coding-system)
2f1fa038
EZ
616 "\nSelect the above, or "
617 "\nSelect ")
618 "one of the following safe coding systems:\n")
b5edd1d1
KH
619 (let ((pos (point))
620 (fill-prefix " "))
621 (mapcar (function (lambda (x) (princ " ") (princ x)))
622 codings)
623 (insert "\n")
624 (fill-region-as-paragraph pos (point)))))
625
626 ;; Read a coding system.
627 (if (consp coding-system)
628 (setq codings (cons (car coding-system) codings)))
629 (let* ((safe-names (mapcar (lambda (x) (list (symbol-name x)))
630 codings))
631 (name (completing-read
632 (format "Select coding system (default %s): "
633 (car codings))
634 safe-names nil t nil nil
635 (car (car safe-names)))))
636 (setq last-coding-system-specified (intern name)
637 coding-system last-coding-system-specified)))
638 (kill-buffer "*Warning*")))
639
640 (if (vectorp (coding-system-eol-type coding-system))
641 (let ((eol (coding-system-eol-type buffer-file-coding-system)))
642 (if (numberp eol)
643 (setq coding-system
644 (coding-system-change-eol-conversion coding-system eol)))))
645
646 (if (eq coding-system t)
647 (setq coding-system buffer-file-coding-system))
e56d7900 648 coding-system))
54b226f7
KH
649
650(setq select-safe-coding-system-function 'select-safe-coding-system)
651
46babb23
KH
652(defun select-message-coding-system ()
653 "Return a coding system to encode the outgoing message of the current buffer.
654It at first tries the first coding system found in these variables
655in this order:
656 (1) local value of `buffer-file-coding-system'
657 (2) value of `sendmail-coding-system'
b5edd1d1
KH
658 (3) value of `default-sendmail-coding-system'
659 (4) value of `default-buffer-file-coding-system'
46babb23
KH
660If the found coding system can't encode the current buffer,
661or none of them are bound to a coding system,
48e41165 662it asks the user to select a proper coding system."
46babb23 663 (let ((coding (or (and (local-variable-p 'buffer-file-coding-system)
b5edd1d1
KH
664 buffer-file-coding-system)
665 sendmail-coding-system
666 default-sendmail-coding-system
667 default-buffer-file-coding-system)))
46babb23
KH
668 (if (eq coding 'no-conversion)
669 ;; We should never use no-conversion for outgoing mails.
670 (setq coding nil))
671 (if (fboundp select-safe-coding-system-function)
672 (funcall select-safe-coding-system-function
b5edd1d1
KH
673 (point-min) (point-max) coding
674 (function (lambda (x) (coding-system-get x 'mime-charset))))
46babb23 675 coding)))
4ed46869 676\f
03c35c83 677;;; Language support stuff.
4ed46869 678
4ed46869 679(defvar language-info-alist nil
2c395d56 680 "Alist of language environment definitions.
4ed46869
KH
681Each element looks like:
682 (LANGUAGE-NAME . ((KEY . INFO) ...))
2c395d56
RS
683where LANGUAGE-NAME is a string, the name of the language environment,
684KEY is a symbol denoting the kind of information, and
685INFO is the data associated with KEY.
686Meaningful values for KEY include
687
688 documentation value is documentation of what this language environment
689 is meant for, and how to use it.
690 charset value is a list of the character sets used by this
691 language environment.
692 sample-text value is one line of text,
693 written using those character sets,
694 appropriate for this language environment.
695 setup-function value is a function to call to switch to this
696 language environment.
697 exit-function value is a function to call to leave this
698 language environment.
699 coding-system value is a list of coding systems that are good
700 for saving text written in this language environment.
701 This list serves as suggestions to the user;
702 in effect, as a kind of documentation.
703 coding-priority value is a list of coding systems for this language
704 environment, in order of decreasing priority.
705 This is used to set up the coding system priority
45d08cb2 706 list when you switch to this language environment.
ddb5c041 707 nonascii-translation
7624ebb9 708 value is a translation table to be set in the
45d08cb2 709 variable `nonascii-translation-table' in this
7624ebb9
KH
710 language environment, or a character set from
711 which `nonascii-insert-offset' is calculated.
ddb5c041
KH
712 input-method value is a default input method for this language
713 environment.
7624ebb9
KH
714 features value is a list of features requested in this
715 language environment.
ddb5c041
KH
716
717The following keys take effect only when multibyte characters are
718globally disabled, i.e. the value of `default-enable-multibyte-characters'
719is nil.
720
721 unibyte-syntax value is a library name to load to set
e8dd0160 722 unibyte 8-bit character syntaxes for this
ddb5c041
KH
723 language environment.
724
725 unibyte-display value is a coding system to encode characters
726 for the terminal. Characters in the range
727 of 160 to 255 display not as octal escapes,
728 but as non-ASCII characters in this language
729 environment.")
2c395d56
RS
730
731(defun get-language-info (lang-env key)
732 "Return information listed under KEY for language environment LANG-ENV.
733KEY is a symbol denoting the kind of information.
734For a list of useful values for KEY and their meanings,
735see `language-info-alist'."
736 (if (symbolp lang-env)
737 (setq lang-env (symbol-name lang-env)))
738 (let ((lang-slot (assoc-ignore-case lang-env language-info-alist)))
4ed46869
KH
739 (if lang-slot
740 (cdr (assq key (cdr lang-slot))))))
741
f08adf27 742(defun set-language-info (lang-env key info)
2c395d56
RS
743 "Modify part of the definition of language environment LANG-ENV.
744Specifically, this stores the information INFO under KEY
745in the definition of this language environment.
4ed46869 746KEY is a symbol denoting the kind of information.
2c395d56 747INFO is the value for that information.
281d03ec 748
2c395d56 749For a list of useful values for KEY and their meanings,
f08adf27 750see `language-info-alist'."
2c395d56
RS
751 (if (symbolp lang-env)
752 (setq lang-env (symbol-name lang-env)))
4ed46869 753 (let (lang-slot key-slot)
2c395d56 754 (setq lang-slot (assoc lang-env language-info-alist))
4ed46869 755 (if (null lang-slot) ; If no slot for the language, add it.
2c395d56 756 (setq lang-slot (list lang-env)
4ed46869
KH
757 language-info-alist (cons lang-slot language-info-alist)))
758 (setq key-slot (assq key lang-slot))
759 (if (null key-slot) ; If no slot for the key, add it.
760 (progn
761 (setq key-slot (list key))
762 (setcdr lang-slot (cons key-slot (cdr lang-slot)))))
cda74479 763 (setcdr key-slot (purecopy info))))
4ed46869 764
2c395d56
RS
765(defun set-language-info-alist (lang-env alist &optional parents)
766 "Store ALIST as the definition of language environment LANG-ENV.
767ALIST is an alist of KEY and INFO values. See the documentation of
98c6d6ed 768`language-info-alist' for the meanings of KEY and INFO.
54b226f7 769
2c395d56
RS
770Optional arg PARENTS is a list of parent menu names; it specifies
771where to put this language environment in the
772Describe Language Environment and Set Language Environment menus.
773For example, (\"European\") means to put this language environment
774in the European submenu in each of those two menus."
775 (if (symbolp lang-env)
776 (setq lang-env (symbol-name lang-env)))
54b226f7
KH
777 (let ((describe-map describe-language-environment-map)
778 (setup-map setup-language-environment-map))
779 (if parents
780 (let ((l parents)
9deed82f 781 map parent-symbol parent prompt)
54b226f7
KH
782 (while l
783 (if (symbolp (setq parent-symbol (car l)))
784 (setq parent (symbol-name parent))
785 (setq parent parent-symbol parent-symbol (intern parent)))
786 (setq map (lookup-key describe-map (vector parent-symbol)))
9deed82f
EZ
787 ;; This prompt string is for define-prefix-command, so
788 ;; that the map it creates will be suitable for a menu.
789 (or map (setq prompt (format "%s Environment" parent)))
54b226f7
KH
790 (if (not map)
791 (progn
792 (setq map (intern (format "describe-%s-environment-map"
793 (downcase parent))))
9deed82f 794 (define-prefix-command map nil prompt)
54b226f7
KH
795 (define-key-after describe-map (vector parent-symbol)
796 (cons parent map) t)))
797 (setq describe-map (symbol-value map))
798 (setq map (lookup-key setup-map (vector parent-symbol)))
799 (if (not map)
800 (progn
801 (setq map (intern (format "setup-%s-environment-map"
802 (downcase parent))))
9deed82f 803 (define-prefix-command map nil prompt)
54b226f7
KH
804 (define-key-after setup-map (vector parent-symbol)
805 (cons parent map) t)))
806 (setq setup-map (symbol-value map))
807 (setq l (cdr l)))))
f08adf27
RS
808
809 ;; Set up menu items for this language env.
7624ebb9 810 (let ((doc (assq 'documentation alist)))
f08adf27
RS
811 (when doc
812 (define-key-after describe-map (vector (intern lang-env))
7624ebb9
KH
813 (cons lang-env 'describe-specified-language-support) t)))
814 (define-key-after setup-map (vector (intern lang-env))
815 (cons lang-env 'setup-specified-language-environment) t)
f08adf27 816
54b226f7 817 (while alist
f08adf27 818 (set-language-info lang-env (car (car alist)) (cdr (car alist)))
54b226f7 819 (setq alist (cdr alist)))))
4ed46869 820
ae302641 821(defun read-language-name (key prompt &optional default)
2c395d56 822 "Read a language environment name which has information for KEY.
ddb5c041 823If KEY is nil, read any language environment.
2c395d56
RS
824Prompt with PROMPT. DEFAULT is the default choice of language environment.
825This returns a language environment name as a string."
4ed46869
KH
826 (let* ((completion-ignore-case t)
827 (name (completing-read prompt
828 language-info-alist
ddb5c041
KH
829 (and key
830 (function (lambda (elm) (assq key elm))))
ae302641 831 t nil nil default)))
13e82c04 832 (if (and (> (length name) 0)
ddb5c041
KH
833 (or (not key)
834 (get-language-info name key)))
13e82c04 835 name)))
4ed46869
KH
836\f
837;;; Multilingual input methods.
d0c40faa
KH
838(defgroup leim nil
839 "LEIM: Libraries of Emacs Input Methods."
840 :group 'mule)
4ed46869 841
d0b9c3ab
KH
842(defconst leim-list-file-name "leim-list.el"
843 "Name of LEIM list file.
844This file contains a list of libraries of Emacs input methods (LEIM)
845in the format of Lisp expression for registering each input method.
846Emacs loads this file at startup time.")
847
2e224638
SM
848(defvar leim-list-header (format
849";;; %s -- list of LEIM (Library of Emacs Input Method)
d0b9c3ab
KH
850;;
851;; This file contains a list of LEIM (Library of Emacs Input Method)
e8dd0160 852;; in the same directory as this file. Loading this file registers
d0b9c3ab
KH
853;; the whole input methods in Emacs.
854;;
d33d5fbe 855;; Each entry has the form:
d0b9c3ab
KH
856;; (register-input-method
857;; INPUT-METHOD LANGUAGE-NAME ACTIVATE-FUNC
858;; TITLE DESCRIPTION
859;; ARG ...)
860;; See the function `register-input-method' for the meanings of arguments.
861;;
862;; If this directory is included in load-path, Emacs automatically
863;; loads this file at startup time.
864
865"
866 leim-list-file-name)
867 "Header to be inserted in LEIM list file.")
868
e55e92ee 869(defvar leim-list-entry-regexp "^(register-input-method"
d0b9c3ab
KH
870 "Regexp matching head of each entry in LEIM list file.
871See also the variable `leim-list-header'")
872
873(defvar update-leim-list-functions
874 '(quail-update-leim-list-file)
875 "List of functions to call to update LEIM list file.
876Each function is called with one arg, LEIM directory name.")
877
a337fe7f
RS
878(defun update-leim-list-file (&rest dirs)
879 "Update LEIM list file in directories DIRS."
d0b9c3ab
KH
880 (let ((functions update-leim-list-functions))
881 (while functions
a337fe7f 882 (apply (car functions) dirs)
d0b9c3ab
KH
883 (setq functions (cdr functions)))))
884
4ed46869
KH
885(defvar current-input-method nil
886 "The current input method for multilingual text.
96db204a 887If nil, that means no input method is activated now.")
4ed46869
KH
888(make-variable-buffer-local 'current-input-method)
889(put 'current-input-method 'permanent-local t)
890
891(defvar current-input-method-title nil
d0b9c3ab 892 "Title string of the current input method shown in mode line.")
4ed46869
KH
893(make-variable-buffer-local 'current-input-method-title)
894(put 'current-input-method-title 'permanent-local t)
895
b4fba33f 896(defcustom default-input-method nil
8861c593 897 "*Default input method for multilingual text (a string).
b4fba33f 898This is the input method activated automatically by the command
9b10b5a3 899`toggle-input-method' (\\[toggle-input-method])."
8861c593 900 :group 'mule
5806e8a6
GM
901 :type '(choice (const nil) string)
902 :set-after '(current-language-environment))
b4fba33f 903
0f835e87
KH
904(put 'input-method-function 'permanent-local t)
905
723a427a
KH
906(defvar input-method-history nil
907 "History list for some commands that read input methods.")
908(make-variable-buffer-local 'input-method-history)
909(put 'input-method-history 'permanent-local t)
4ed46869
KH
910
911(defvar inactivate-current-input-method-function nil
912 "Function to call for inactivating the current input method.
913Every input method should set this to an appropriate value when activated.
f17ccaee
KH
914This function is called with no argument.
915
916This function should never change the value of `current-input-method'.
917It is set to nil by the function `inactivate-input-method'.")
4ed46869
KH
918(make-variable-buffer-local 'inactivate-current-input-method-function)
919(put 'inactivate-current-input-method-function 'permanent-local t)
920
921(defvar describe-current-input-method-function nil
922 "Function to call for describing the current input method.
923This function is called with no argument.")
924(make-variable-buffer-local 'describe-current-input-method-function)
925(put 'describe-current-input-method-function 'permanent-local t)
926
d0b9c3ab 927(defvar input-method-alist nil
2c395d56 928 "Alist of input method names vs how to use them.
d0b9c3ab 929Each element has the form:
2c395d56
RS
930 (INPUT-METHOD LANGUAGE-ENV ACTIVATE-FUNC TITLE DESCRIPTION ARGS...)
931See the function `register-input-method' for the meanings of the elements.")
932
f08adf27 933(defun register-input-method (input-method lang-env &rest args)
2c395d56 934 "Register INPUT-METHOD as an input method for language environment ENV.
f08adf27 935INPUT-METHOD and LANG-ENV are symbols or strings.
d0b9c3ab 936
d0b9c3ab 937The remaining arguments are:
2c395d56
RS
938 ACTIVATE-FUNC, TITLE, DESCRIPTION, and ARGS...
939ACTIVATE-FUNC is a function to call to activate this method.
940TITLE is a string to show in the mode line when this method is active.
941DESCRIPTION is a string describing this method and what it is good for.
942The ARGS, if any, are passed as arguments to ACTIVATE-FUNC.
205814ee
KH
943All told, the arguments to ACTIVATE-FUNC are INPUT-METHOD and the ARGS.
944
945This function is mainly used in the file \"leim-list.el\" which is
946created at building time of emacs, registering all quail input methods
947contained in the emacs distribution.
948
949In case you want to register a new quail input method by yourself, be
950careful to use the same input method title as given in the third
951parameter of `quail-define-package' (if the values are different, the
952string specified in this function takes precedence).
953
954The commands `describe-input-method' and `list-input-methods' need
955this duplicated values to show some information about input methods
956without loading the affected quail packages."
f08adf27
RS
957 (if (symbolp lang-env)
958 (setq lang-env (symbol-name lang-env)))
4ef06f75
KH
959 (if (symbolp input-method)
960 (setq input-method (symbol-name input-method)))
f08adf27 961 (let ((info (cons lang-env args))
d0b9c3ab
KH
962 (slot (assoc input-method input-method-alist)))
963 (if slot
964 (setcdr slot info)
965 (setq slot (cons input-method info))
966 (setq input-method-alist (cons slot input-method-alist)))))
967
4d5ac029 968(defun read-input-method-name (prompt &optional default inhibit-null)
d0b9c3ab 969 "Read a name of input method from a minibuffer prompting with PROMPT.
4d5ac029
RS
970If DEFAULT is non-nil, use that as the default,
971 and substitute it into PROMPT at the first `%s'.
4ef06f75
KH
972If INHIBIT-NULL is non-nil, null input signals an error.
973
974The return value is a string."
4d5ac029
RS
975 (if default
976 (setq prompt (format prompt default)))
d0b9c3ab 977 (let* ((completion-ignore-case t)
723a427a
KH
978 ;; This binding is necessary because input-method-history is
979 ;; buffer local.
d0b9c3ab 980 (input-method (completing-read prompt input-method-alist
87505a98
RS
981 nil t nil 'input-method-history
982 default)))
bf294e6e
KH
983 (if (and input-method (symbolp input-method))
984 (setq input-method (symbol-name input-method)))
d0b9c3ab
KH
985 (if (> (length input-method) 0)
986 input-method
987 (if inhibit-null
43807b77 988 (error "No valid input method is specified")))))
d0b9c3ab 989
d0b9c3ab 990(defun activate-input-method (input-method)
2c395d56
RS
991 "Switch to input method INPUT-METHOD for the current buffer.
992If some other input method is already active, turn it off first.
993If INPUT-METHOD is nil, deactivate any current input method."
305a3cb6 994 (if (and input-method (symbolp input-method))
4ef06f75 995 (setq input-method (symbol-name input-method)))
723a427a
KH
996 (if (and current-input-method
997 (not (string= current-input-method input-method)))
305a3cb6 998 (inactivate-input-method))
2c395d56 999 (unless (or current-input-method (null input-method))
d0b9c3ab
KH
1000 (let ((slot (assoc input-method input-method-alist)))
1001 (if (null slot)
723a427a 1002 (error "Can't activate input method `%s'" input-method))
8efc03e1
KH
1003 (let ((func (nth 2 slot)))
1004 (if (functionp func)
1005 (apply (nth 2 slot) input-method (nthcdr 5 slot))
1006 (if (and (consp func) (symbolp (car func)) (symbolp (cdr func)))
1007 (progn
1008 (require (cdr func))
1009 (apply (car func) input-method (nthcdr 5 slot)))
1010 (error "Can't activate input method `%s'" input-method))))
d0b9c3ab 1011 (setq current-input-method input-method)
723a427a 1012 (setq current-input-method-title (nth 3 slot))
28885c0e
KH
1013 (unwind-protect
1014 (run-hooks 'input-method-activate-hook)
1015 (force-mode-line-update)))))
15b3e511 1016
15b3e511 1017(defun inactivate-input-method ()
f17ccaee 1018 "Turn off the current input method."
723a427a
KH
1019 (when current-input-method
1020 (if input-method-history
1021 (unless (string= current-input-method (car input-method-history))
1022 (setq input-method-history
1023 (cons current-input-method
1024 (delete current-input-method input-method-history))))
1025 (setq input-method-history (list current-input-method)))
1026 (unwind-protect
1027 (funcall inactivate-current-input-method-function)
15b3e511 1028 (unwind-protect
723a427a
KH
1029 (run-hooks 'input-method-inactivate-hook)
1030 (setq current-input-method nil
28885c0e
KH
1031 current-input-method-title nil)
1032 (force-mode-line-update)))))
4ed46869 1033
8b784951 1034(defun set-input-method (input-method)
2c395d56
RS
1035 "Select and activate input method INPUT-METHOD for the current buffer.
1036This also sets the default input method to the one you specify."
d0b9c3ab 1037 (interactive
723a427a 1038 (let* ((default (or (car input-method-history) default-input-method)))
42395763 1039 (list (read-input-method-name
87505a98 1040 (if default "Select input method (default %s): " "Select input method: ")
42395763 1041 default t))))
d0b9c3ab 1042 (activate-input-method input-method)
42395763 1043 (setq default-input-method input-method))
4ed46869
KH
1044
1045(defun toggle-input-method (&optional arg)
15b3e511 1046 "Turn on or off a multilingual text input method for the current buffer.
723a427a 1047
f2979bdb
KH
1048With no prefix argument, if an input method is currently activated,
1049turn it off. Otherwise, activate an input method -- the one most
1050recently used, or the one specified in `default-input-method', or
1051the one read from the minibuffer.
723a427a 1052
f2979bdb
KH
1053With a prefix argument, read an input method from the minibuffer and
1054turn it on.
723a427a 1055
f2979bdb
KH
1056The default is to use the most recent input method specified
1057\(not including the currently active input method, if any)."
4ed46869 1058 (interactive "P")
7ddbb5bc
RS
1059 (if (and current-input-method (not arg))
1060 (inactivate-input-method)
1061 (let ((default (or (car input-method-history) default-input-method)))
1062 (if (and arg default (equal current-input-method default)
1063 (> (length input-method-history) 1))
1064 (setq default (nth 1 input-method-history)))
723a427a
KH
1065 (activate-input-method
1066 (if (or arg (not default))
7ddbb5bc
RS
1067 (progn
1068 (read-input-method-name
1069 (if default "Input method (default %s): " "Input method: " )
1070 default t))
723a427a
KH
1071 default))
1072 (or default-input-method
1073 (setq default-input-method current-input-method)))))
d0b9c3ab
KH
1074
1075(defun describe-input-method (input-method)
2c395d56 1076 "Describe input method INPUT-METHOD."
d0b9c3ab
KH
1077 (interactive
1078 (list (read-input-method-name
1079 "Describe input method (default, current choice): ")))
78754934 1080 (if (and input-method (symbolp input-method))
4ef06f75 1081 (setq input-method (symbol-name input-method)))
d0b9c3ab
KH
1082 (if (null input-method)
1083 (describe-current-input-method)
464cc130
KH
1084 (let ((current current-input-method))
1085 (condition-case nil
1086 (progn
1087 (save-excursion
1088 (activate-input-method input-method)
1089 (describe-current-input-method))
1090 (activate-input-method current))
1091 (error
1092 (activate-input-method current)
1093 (with-output-to-temp-buffer "*Help*"
1094 (let ((elt (assoc input-method input-method-alist)))
1095 (princ (format
1096 "Input method: %s (`%s' in mode line) for %s\n %s\n"
1097 input-method (nth 3 elt) (nth 1 elt) (nth 4 elt))))))))))
d0b9c3ab
KH
1098
1099(defun describe-current-input-method ()
96db204a 1100 "Describe the input method currently in use."
4ed46869
KH
1101 (if current-input-method
1102 (if (and (symbolp describe-current-input-method-function)
1103 (fboundp describe-current-input-method-function))
1104 (funcall describe-current-input-method-function)
1105 (message "No way to describe the current input method `%s'"
f2979bdb 1106 current-input-method)
4ed46869 1107 (ding))
d0b9c3ab 1108 (error "No input method is activated now")))
4ed46869 1109
d3459641 1110(defun read-multilingual-string (prompt &optional initial-input input-method)
4ed46869
KH
1111 "Read a multilingual string from minibuffer, prompting with string PROMPT.
1112The input method selected last time is activated in minibuffer.
15b3e511 1113If optional second arg INITIAL-INPUT is non-nil, insert it in the minibuffer
d0b9c3ab
KH
1114initially.
1115Optional 3rd argument INPUT-METHOD specifies the input method
4ef06f75
KH
1116to be activated instead of the one selected last time. It is a symbol
1117or a string."
88d559ec
KH
1118 (setq input-method
1119 (or input-method
d3459641 1120 current-input-method
88d559ec
KH
1121 default-input-method
1122 (read-input-method-name "Input method: " nil t)))
3df60841 1123 (if (and input-method (symbolp input-method))
4ef06f75 1124 (setq input-method (symbol-name input-method)))
305a3cb6
KH
1125 (let ((prev-input-method current-input-method))
1126 (unwind-protect
1127 (progn
1128 (activate-input-method input-method)
1129 (read-string prompt initial-input nil nil t))
1130 (activate-input-method prev-input-method))))
4ed46869
KH
1131
1132;; Variables to control behavior of input methods. All input methods
1133;; should react to these variables.
1134
8efc03e1
KH
1135(defcustom input-method-verbose-flag 'default
1136 "*A flag to control extra guidance given by input methods.
1137The value should be nil, t, `complex-only', or `default'.
4ed46869 1138
cb29dfb6 1139The extra guidance is done by showing list of available keys in echo
8efc03e1
KH
1140area. When you use the input method in the minibuffer, the guidance
1141is shown at the bottom short window (split from the existing window).
c27c4ed8 1142
8efc03e1
KH
1143If the value is t, extra guidance is always given, if the value is
1144nil, extra guidance is always suppressed.
1145
1146If the value is `complex-only', only complex input methods such as
1147`chinese-py' and `japanese' give extra guidance.
1148
1149If the value is `default', complex input methods always give extra
1150guidance, but simple input methods give it only when you are not in
1151the minibuffer.
1152
1153See also the variable `input-method-highlight-flag'."
1154 :type '(choice (const t) (const nil) (const complex-only) (const default))
42395763
RS
1155 :group 'mule)
1156
1157(defcustom input-method-highlight-flag t
1158 "*If this flag is non-nil, input methods highlight partially-entered text.
1159For instance, while you are in the middle of a Quail input method sequence,
1160the text inserted so far is temporarily underlined.
8efc03e1
KH
1161The underlining goes away when you finish or abort the input method sequence.
1162See also the variable `input-method-verbose-flag'."
42395763
RS
1163 :type 'boolean
1164 :group 'mule)
4ed46869
KH
1165
1166(defvar input-method-activate-hook nil
f17ccaee
KH
1167 "Normal hook run just after an input method is activated.
1168
1169The variable `current-input-method' keeps the input method name
1170just activated.")
4ed46869
KH
1171
1172(defvar input-method-inactivate-hook nil
f17ccaee
KH
1173 "Normal hook run just after an input method is inactivated.
1174
1175The variable `current-input-method' still keeps the input method name
4d0e6a11 1176just inactivated.")
4ed46869
KH
1177
1178(defvar input-method-after-insert-chunk-hook nil
1179 "Normal hook run just after an input method insert some chunk of text.")
1180
dccca980
KH
1181(defvar input-method-exit-on-first-char nil
1182 "This flag controls a timing when an input method returns.
1183Usually, the input method does not return while there's a possibility
1184that it may find a different translation if a user types another key.
39e643e2
RS
1185But, it this flag is non-nil, the input method returns as soon as
1186the current key sequence gets long enough to have some valid translation.")
dccca980
KH
1187
1188(defvar input-method-use-echo-area nil
1189 "This flag controls how an input method shows an intermediate key sequence.
39e643e2
RS
1190Usually, the input method inserts the intermediate key sequence,
1191or candidate translations corresponding to the sequence,
1192at point in the current buffer.
1193But, if this flag is non-nil, it displays them in echo area instead.")
dccca980 1194
723a427a
KH
1195(defvar input-method-exit-on-invalid-key nil
1196 "This flag controls the behaviour of an input method on invalid key input.
1197Usually, when a user types a key which doesn't start any character
1198handled by the input method, the key is handled by turning off the
e8dd0160 1199input method temporarily. After that key, the input method is re-enabled.
723a427a
KH
1200But, if this flag is non-nil, the input method is never back on.")
1201
4ed46869 1202\f
8efc03e1
KH
1203(defvar set-language-environment-hook nil
1204 "Normal hook run after some language environment is set.
1205
1206When you set some hook function here, that effect usually should not
1207be inherited to another language environment. So, you had better set
1208another function in `exit-language-environment-hook' (which see) to
1209cancel the effect.")
1210
1211(defvar exit-language-environment-hook nil
1212 "Normal hook run after exiting from some language environment.
1213When this hook is run, the variable `current-language-environment'
1214is still bound to the language environment being exited.
1215
e8dd0160 1216This hook is mainly used for canceling the effect of
8efc03e1
KH
1217`set-language-environment-hook' (which-see).")
1218
b0648a00
RS
1219(put 'setup-specified-language-environment 'apropos-inhibit t)
1220
15b3e511 1221(defun setup-specified-language-environment ()
f08adf27 1222 "Switch to a specified language environment."
15b3e511 1223 (interactive)
f850d782 1224 (let (language-name)
15b3e511
KH
1225 (if (and (symbolp last-command-event)
1226 (or (not (eq last-command-event 'Default))
1227 (setq last-command-event 'English))
f850d782
RS
1228 (setq language-name (symbol-name last-command-event)))
1229 (set-language-environment language-name)
15b3e511 1230 (error "Bogus calling sequence"))))
4ed46869 1231
8861c593 1232(defcustom current-language-environment "English"
94d04df6 1233 "The last language environment specified with `set-language-environment'.
ebef6d93
KH
1234This variable should be set only with \\[customize], which is equivalent
1235to using the function `set-language-environment'."
94d04df6 1236 :link '(custom-manual "(emacs)Language Environments")
dff1aa24 1237 :set (lambda (symbol value) (set-language-environment value))
94d04df6
DL
1238 :get (lambda (x)
1239 (or (car-safe (assoc-ignore-case
1240 (if (symbolp current-language-environment)
1241 (symbol-name current-language-environment)
1242 current-language-environment)
1243 language-info-alist))
1244 "English"))
1245 :type (cons 'choice (mapcar (lambda (lang)
1246 (list 'const (car lang)))
1247 language-info-alist))
8861c593
RS
1248 :initialize 'custom-initialize-default
1249 :group 'mule
1250 :type 'string)
f850d782 1251
ddb5c041
KH
1252(defun reset-language-environment ()
1253 "Reset multilingual environment of Emacs to the default status.
1254
1255The default status is as follows:
1256
1257 The default value of buffer-file-coding-system is nil.
1258 The default coding system for process I/O is nil.
1259 The default value for the command `set-terminal-coding-system' is nil.
1260 The default value for the command `set-keyboard-coding-system' is nil.
1261
1262 The order of priorities of coding categories and the coding system
1263 bound to each category are as follows
1264 coding category coding system
1265 --------------------------------------------------
1266 coding-category-iso-8-2 iso-latin-1
1267 coding-category-iso-8-1 iso-latin-1
1268 coding-category-iso-7-tight iso-2022-jp
1269 coding-category-iso-7 iso-2022-7bit
1270 coding-category-iso-7-else iso-2022-7bit-lock
1271 coding-category-iso-8-else iso-2022-8bit-ss2
1272 coding-category-emacs-mule emacs-mule
1273 coding-category-raw-text raw-text
1274 coding-category-sjis japanese-shift-jis
1275 coding-category-big5 chinese-big5
1276 coding-category-ccl nil
e8dd0160 1277 coding-category-binary no-conversion
ddb5c041
KH
1278"
1279 (interactive)
1280 ;; This function formerly set default-enable-multibyte-characters to t,
1281 ;; but that is incorrect. It should not alter the unibyte/multibyte choice.
1282
1283 (setq coding-category-iso-7-tight 'iso-2022-jp
1284 coding-category-iso-7 'iso-2022-7bit
1285 coding-category-iso-8-1 'iso-latin-1
1286 coding-category-iso-8-2 'iso-latin-1
1287 coding-category-iso-7-else 'iso-2022-7bit-lock
1288 coding-category-iso-8-else 'iso-2022-8bit-ss2
1289 coding-category-emacs-mule 'emacs-mule
1290 coding-category-raw-text 'raw-text
1291 coding-category-sjis 'japanese-shift-jis
1292 coding-category-big5 'chinese-big5
9bfcd269
KH
1293 coding-category-utf-8 nil
1294 coding-category-utf-16-be nil
1295 coding-category-utf-16-le nil
ddb5c041
KH
1296 coding-category-ccl nil
1297 coding-category-binary 'no-conversion)
1298
1299 (set-coding-priority
1300 '(coding-category-iso-8-1
1301 coding-category-iso-8-2
1302 coding-category-iso-7-tight
1303 coding-category-iso-7
1304 coding-category-iso-7-else
1305 coding-category-iso-8-else
1306 coding-category-emacs-mule
1307 coding-category-raw-text
1308 coding-category-sjis
1309 coding-category-big5
1310 coding-category-ccl
9bfcd269
KH
1311 coding-category-binary
1312 coding-category-utf-16-be
1313 coding-category-utf-16-le
1314 coding-category-utf-8))
ddb5c041 1315
91693d18
KH
1316 (update-coding-systems-internal)
1317
ddb5c041 1318 (set-default-coding-systems nil)
b5edd1d1 1319 (setq default-sendmail-coding-system 'iso-latin-1)
0c47a7c8 1320 (setq default-process-coding-system '(undecided . iso-latin-1))
b5edd1d1 1321
ddb5c041
KH
1322 ;; Don't alter the terminal and keyboard coding systems here.
1323 ;; The terminal still supports the same coding system
1324 ;; that it supported a minute ago.
1325;;; (set-terminal-coding-system-internal nil)
1326;;; (set-keyboard-coding-system-internal nil)
1327
1328 (setq nonascii-translation-table nil
1329 nonascii-insert-offset 0))
1330
0c47a7c8
KH
1331(reset-language-environment)
1332
40c81f74
PE
1333(defun set-display-table-and-terminal-coding-system (language-name)
1334 "Set up the display table and terminal coding system for LANGUAGE-NAME."
1335 (let ((coding (get-language-info language-name 'unibyte-display)))
1336 (if coding
1337 (standard-display-european-internal)
1338 (standard-display-default (if (eq window-system 'pc) 128 160) 255)
1339 (aset standard-display-table 146 nil))
1340 (or (eq window-system 'pc)
1341 (set-terminal-coding-system coding))))
1342
166246f7 1343(defun set-language-environment (language-name)
6c05d680
RS
1344 "Set up multi-lingual environment for using LANGUAGE-NAME.
1345This sets the coding system priority and the default input method
8861c593
RS
1346and sometimes other things. LANGUAGE-NAME should be a string
1347which is the name of a language environment. For example, \"Latin-1\"
1348specifies the character set for the major languages of Western Europe."
8efc03e1 1349 (interactive (list (read-language-name
ddb5c041 1350 nil
8efc03e1 1351 "Set language environment (default, English): ")))
4ef06f75
KH
1352 (if language-name
1353 (if (symbolp language-name)
1354 (setq language-name (symbol-name language-name)))
1355 (setq language-name "English"))
ddb5c041 1356 (or (assoc-ignore-case language-name language-info-alist)
f850d782 1357 (error "Language environment not defined: %S" language-name))
8efc03e1
KH
1358 (if current-language-environment
1359 (let ((func (get-language-info current-language-environment
1360 'exit-function)))
e63645c2
KH
1361 (run-hooks 'exit-language-environment-hook)
1362 (if (fboundp func) (funcall func))))
03c35c83
EZ
1363 (let ((default-eol-type (coding-system-eol-type
1364 default-buffer-file-coding-system)))
1365 (reset-language-environment)
ddb5c041 1366
03c35c83
EZ
1367 (setq current-language-environment language-name)
1368 (set-language-environment-coding-systems language-name default-eol-type))
ddb5c041
KH
1369 (let ((input-method (get-language-info language-name 'input-method)))
1370 (when input-method
1371 (setq default-input-method input-method)
1372 (if input-method-history
1373 (setq input-method-history
1374 (cons input-method
1375 (delete input-method input-method-history))))))
ec241f58
EZ
1376 (let ((nonascii (get-language-info language-name 'nonascii-translation))
1377 (dos-table
4e2ac2d9
EZ
1378 (if (eq window-system 'pc)
1379 (intern
d9c0a50e 1380 (format "cp%d-nonascii-translation-table" dos-codepage)))))
03c35c83
EZ
1381 (cond
1382 ((char-table-p nonascii)
1383 (setq nonascii-translation-table nonascii))
ec241f58 1384 ((and (eq window-system 'pc) (boundp dos-table))
03c35c83
EZ
1385 ;; DOS terminals' default is to use a special non-ASCII translation
1386 ;; table as appropriate for the installed codepage.
ec241f58 1387 (setq nonascii-translation-table (symbol-value dos-table)))
03c35c83
EZ
1388 ((charsetp nonascii)
1389 (setq nonascii-insert-offset (- (make-char nonascii) 128)))))
ddb5c041 1390
63283a8f 1391 ;; Unibyte setups if necessary.
ddb5c041 1392 (unless default-enable-multibyte-characters
63283a8f 1393 ;; Syntax and case table.
ddb5c041
KH
1394 (let ((syntax (get-language-info language-name 'unibyte-syntax)))
1395 (if syntax
1396 (let ((set-case-syntax-set-multibyte nil))
63283a8f
KH
1397 (load syntax nil t))
1398 ;; No information for syntax and case. Reset to the defaults.
1399 (let ((syntax-table (standard-syntax-table))
1400 (case-table (standard-case-table))
03c35c83 1401 (ch (if (eq window-system 'pc) 128 160)))
63283a8f
KH
1402 (while (< ch 256)
1403 (modify-syntax-entry ch " " syntax-table)
1404 (aset case-table ch ch)
1405 (setq ch (1+ ch)))
1406 (set-char-table-extra-slot case-table 0 nil)
1407 (set-char-table-extra-slot case-table 1 nil)
1408 (set-char-table-extra-slot case-table 2 nil))
1409 (set-standard-case-table (standard-case-table))
1410 (let ((list (buffer-list)))
1411 (while list
1412 (with-current-buffer (car list)
1413 (set-case-table (standard-case-table)))
1414 (setq list (cdr list))))))
40c81f74 1415 (set-display-table-and-terminal-coding-system language-name))
ddb5c041
KH
1416
1417 (let ((required-features (get-language-info language-name 'features)))
1418 (while required-features
1419 (require (car required-features))
1420 (setq required-features (cdr required-features))))
1421 (let ((func (get-language-info language-name 'setup-function)))
1422 (if (fboundp func)
1423 (funcall func)))
8efc03e1 1424 (run-hooks 'set-language-environment-hook)
f850d782 1425 (force-mode-line-update t))
4ed46869 1426
51a8fc1d
RS
1427(defun standard-display-european-internal ()
1428 ;; Actually set up direct output of non-ASCII characters.
03c35c83
EZ
1429 (standard-display-8bit (if (eq window-system 'pc) 128 160) 255)
1430 ;; Unibyte Emacs on MS-DOS wants to display all 8-bit characters with
1431 ;; the native font, and codes 160 and 146 stand for something very
1432 ;; different there.
1433 (or (and (eq window-system 'pc) (not default-enable-multibyte-characters))
1434 (progn
1435 ;; Make non-line-break space display as a plain space.
1436 ;; Most X fonts do the wrong thing for code 160.
1437 (aset standard-display-table 160 [32])
1438 ;; Most Windows programs send out apostrophe's as \222. Most X fonts
1439 ;; don't contain a character at that position. Map it to the ASCII
1440 ;; apostrophe.
1441 (aset standard-display-table 146 [39]))))
1442
1443(defun set-language-environment-coding-systems (language-name
1444 &optional eol-type)
1445 "Do various coding system setups for language environment LANGUAGE-NAME.
1446
1447The optional arg EOL-TYPE specifies the eol-type of the default value
1448of buffer-file-coding-system set by this function."
54b226f7
KH
1449 (let* ((priority (get-language-info language-name 'coding-priority))
1450 (default-coding (car priority)))
1451 (if priority
1452 (let ((categories (mapcar 'coding-system-category priority)))
03c35c83
EZ
1453 (set-default-coding-systems
1454 (if (memq eol-type '(0 1 2 unix dos mac))
1455 (coding-system-change-eol-conversion default-coding eol-type)
1456 default-coding))
46babb23 1457 (setq default-sendmail-coding-system default-coding)
54b226f7
KH
1458 (set-coding-priority categories)
1459 (while priority
1460 (set (car categories) (car priority))
1461 (setq priority (cdr priority) categories (cdr categories)))
45d08cb2 1462 (update-coding-systems-internal)))))
54b226f7 1463
4ed46869
KH
1464;; Print all arguments with `princ', then print "\n".
1465(defsubst princ-list (&rest args)
1466 (while args (princ (car args)) (setq args (cdr args)))
1467 (princ "\n"))
1468
b0648a00
RS
1469(put 'describe-specified-language-support 'apropos-inhibit t)
1470
48082651 1471;; Print a language specific information such as input methods,
13e82c04 1472;; charsets, and coding systems. This function is intended to be
48082651 1473;; called from the menu:
281d03ec 1474;; [menu-bar mule describe-language-environment LANGUAGE]
48082651
KH
1475;; and should not run it by `M-x describe-current-input-method-function'.
1476(defun describe-specified-language-support ()
96db204a 1477 "Describe how Emacs supports the specified language environment."
48082651 1478 (interactive)
281d03ec 1479 (let (language-name)
48082651 1480 (if (not (and (symbolp last-command-event)
cda74479
DL
1481 (or (not (eq last-command-event 'Default))
1482 (setq last-command-event 'English))
281d03ec 1483 (setq language-name (symbol-name last-command-event))))
48082651 1484 (error "Bogus calling sequence"))
281d03ec
RS
1485 (describe-language-environment language-name)))
1486
1487(defun describe-language-environment (language-name)
1488 "Describe how Emacs supports language environment LANGUAGE-NAME."
78754934
KH
1489 (interactive
1490 (list (read-language-name
1491 'documentation
8adfa8be 1492 "Describe language environment (default, current choice): ")))
f850d782
RS
1493 (if (null language-name)
1494 (setq language-name current-language-environment))
281d03ec
RS
1495 (if (or (null language-name)
1496 (null (get-language-info language-name 'documentation)))
1497 (error "No documentation for the specified language"))
4ef06f75
KH
1498 (if (symbolp language-name)
1499 (setq language-name (symbol-name language-name)))
464cc130
KH
1500 (let ((doc (get-language-info language-name 'documentation))
1501 pos)
48082651 1502 (with-output-to-temp-buffer "*Help*"
464cc130
KH
1503 (save-excursion
1504 (set-buffer standard-output)
1505 (insert language-name " language environment\n\n")
1506 (if (stringp doc)
1507 (insert doc "\n\n"))
1508 (let ((str (get-language-info language-name 'sample-text)))
1509 (if (stringp str)
1510 (insert "Sample text:\n " str "\n\n")))
1511 (let ((input-method (get-language-info language-name 'input-method))
1512 (l (copy-sequence input-method-alist)))
1513 (insert "Input methods")
1514 (when input-method
1515 (insert " (default, " input-method ")")
1516 (setq input-method (assoc input-method input-method-alist))
1517 (setq l (cons input-method (delete input-method l))))
1518 (insert ":\n")
48082651 1519 (while l
464cc130
KH
1520 (when (string= language-name (nth 1 (car l)))
1521 (insert " " (car (car l)))
1522 (search-backward (car (car l)))
1523 (help-xref-button 0 #'describe-input-method (car (car l))
1524 "mouse-2, RET: describe this input method")
1525 (goto-char (point-max))
1526 (insert " (\"" (nth 3 (car l)) "\" in mode line)\n"))
1527 (setq l (cdr l)))
1528 (insert "\n"))
1529 (insert "Character sets:\n")
1530 (let ((l (get-language-info language-name 'charset)))
1531 (if (null l)
1532 (insert " nothing specific to " language-name "\n")
1533 (while l
1534 (insert " " (symbol-name (car l)))
1535 (search-backward (symbol-name (car l)))
1536 (help-xref-button 0 #'describe-character-set (car l)
1537 "mouse-2, RET: describe this character set")
1538 (goto-char (point-max))
1539 (insert ": " (charset-description (car l)) "\n")
1540 (setq l (cdr l)))))
1541 (insert "\n")
1542 (insert "Coding systems:\n")
1543 (let ((l (get-language-info language-name 'coding-system)))
1544 (if (null l)
1545 (insert " nothing specific to " language-name "\n")
1546 (while l
1547 (insert " " (symbol-name (car l)))
1548 (search-backward (symbol-name (car l)))
1549 (help-xref-button 0 #'describe-coding-system (car l)
1550 "mouse-2, RET: describe this coding system")
1551 (goto-char (point-max))
1552 (insert " (`"
1553 (coding-system-mnemonic (car l))
1554 "' in mode line):\n\t"
1555 (coding-system-doc-string (car l))
1556 "\n")
1557 (let ((aliases (coding-system-get (car l)
1558 'alias-coding-systems)))
1559 (when aliases
1560 (insert "\t(alias:")
1561 (while aliases
1562 (insert " " (symbol-name (car aliases)))
1563 (setq aliases (cdr aliases)))
1564 (insert ")\n")))
1565 (setq l (cdr l)))))
1566 (help-setup-xref (list #'describe-language-environment language-name)
1567 (interactive-p))))))
4ed46869 1568\f
40c81f74
PE
1569;;; Locales.
1570
0d7c5bb9
DL
1571(defvar locale-translation-file-name nil
1572 "File name for the system's file of locale-name aliases, or nil if none.")
40c81f74
PE
1573
1574(defvar locale-language-names
1575 '(
1576 ;; UTF-8 is not yet implemented.
1577 ;; Put this first, so that e.g. "ko.UTF-8" does not match "ko" below.
1578 (".*[._]utf" . nil)
1579
1580 ;; Locale names of the form LANGUAGE[_TERRITORY][.CODESET][@MODIFIER]
1581 ;; as specified in the Single Unix Spec, Version 2.
1582 ;; LANGUAGE is a language code taken from ISO 639:1988 (E/F)
1583 ;; with additions from ISO 639/RA Newsletter No.1/1989;
1584 ;; see Internet RFC 2165 (1997-06).
1585 ;; TERRITORY is a country code taken from ISO 3166.
1586 ;; CODESET and MODIFIER are implementation-dependent.
1587 ;;
1588 ; aa Afar
1589 ; ab Abkhazian
6ececc4d 1590 ("af" . "Latin-1") ; Afrikaans
40c81f74
PE
1591 ("am" . "Ethiopic") ; Amharic
1592 ; ar Arabic
1593 ; as Assamese
1594 ; ay Aymara
1595 ; az Azerbaijani
1596 ; ba Bashkir
6ececc4d
PE
1597 ("be" . "Latin-5") ; Byelorussian
1598 ("bg" . "Latin-5") ; Bulgarian
40c81f74
PE
1599 ; bh Bihari
1600 ; bi Bislama
1601 ; bn Bengali, Bangla
1602 ("bo" . "Tibetan")
1603 ("br" . "Latin-1") ; Breton
1604 ("ca" . "Latin-1") ; Catalan
1605 ; co Corsican
1606 ("cs" . "Czech")
6ececc4d 1607 ("cy" . "Latin-8") ; Welsh
40c81f74
PE
1608 ("da" . "Latin-1") ; Danish
1609 ("de" . "German")
1610 ; dz Bhutani
1611 ("el" . "Greek")
6ececc4d
PE
1612 ;; Users who specify "en" explicitly typically want Latin-1, not ASCII.
1613 ("en" . "Latin-1") ; English
40c81f74
PE
1614 ("eo" . "Latin-3") ; Esperanto
1615 ("es" . "Latin-1") ; Spanish
1616 ("et" . "Latin-4") ; Estonian
1617 ("eu" . "Latin-1") ; Basque
1618 ; fa Persian
1619 ("fi" . "Latin-1") ; Finnish
1620 ; fj Fiji
1621 ("fo" . "Latin-1") ; Faroese
1622 ("fr" . "Latin-1") ; French
1623 ("fy" . "Latin-1") ; Frisian
6ececc4d
PE
1624 ("ga" . "Latin-1") ; Irish Gaelic (new orthography)
1625 ("gd" . "Latin-1") ; Scots Gaelic
1626 ("gl" . "Latin-1") ; Galician
40c81f74
PE
1627 ; gn Guarani
1628 ; gu Gujarati
1629 ; ha Hausa
1630 ("he" . "Hebrew")
1631 ("hi" . "Devanagari") ; Hindi
1632 ("hr" . "Latin-2") ; Croatian
1633 ("hu" . "Latin-2") ; Hungarian
1634 ; hy Armenian
1635 ; ia Interlingua
1636 ("id" . "Latin-1") ; Indonesian
1637 ; ie Interlingue
1638 ; ik Inupiak
1639 ("is" . "Latin-1") ; Icelandic
1640 ("it" . "Latin-1") ; Italian
1641 ; iu Inuktitut
1642 ("ja" . "Japanese")
1643 ; jw Javanese
1644 ; ka Georgian
1645 ; kk Kazakh
6ececc4d 1646 ("kl" . "Latin-1") ; Greenlandic
40c81f74
PE
1647 ; km Cambodian
1648 ; kn Kannada
1649 ("ko" . "Korean")
1650 ; ks Kashmiri
1651 ; ku Kurdish
1652 ; ky Kirghiz
1653 ("la" . "Latin-1") ; Latin
1654 ; ln Lingala
1655 ("lo" . "Lao") ; Laothian
1656 ("lt" . "Latin-4") ; Lithuanian
1657 ("lv" . "Latin-4") ; Latvian, Lettish
1658 ; mg Malagasy
1659 ; mi Maori
6ececc4d 1660 ("mk" . "Latin-5") ; Macedonian
40c81f74
PE
1661 ; ml Malayalam
1662 ; mn Mongolian
1663 ; mo Moldavian
1664 ("mr" . "Devanagari") ; Marathi
1665 ; ms Malay
1666 ("mt" . "Latin-3") ; Maltese
1667 ; my Burmese
1668 ; na Nauru
1669 ("ne" . "Devanagari") ; Nepali
1670 ("nl" . "Latin-1") ; Dutch
1671 ("no" . "Latin-1") ; Norwegian
1672 ; oc Occitan
1673 ; om (Afan) Oromo
1674 ; or Oriya
1675 ; pa Punjabi
1676 ("pl" . "Latin-2") ; Polish
1677 ; ps Pashto, Pushto
1678 ("pt" . "Latin-1") ; Portuguese
1679 ; qu Quechua
6ececc4d 1680 ("rm" . "Latin-1") ; Rhaeto-Romanic
40c81f74
PE
1681 ; rn Kirundi
1682 ("ro" . "Romanian")
1683 ("ru.*[_.]koi8" . "Cyrillic-KOI8") ; Russian
6ececc4d 1684 ("ru" . "Latin-5") ; Russian
40c81f74
PE
1685 ; rw Kinyarwanda
1686 ("sa" . "Devanagari") ; Sanskrit
1687 ; sd Sindhi
1688 ; sg Sangho
1689 ("sh" . "Latin-2") ; Serbo-Croatian
1690 ; si Sinhalese
1691 ("sk" . "Slovak")
1692 ("sl" . "Slovenian")
1693 ; sm Samoan
1694 ; sn Shona
1695 ; so Somali
6ececc4d 1696 ("sq" . "Latin-1") ; Albanian
40c81f74
PE
1697 ("sr" . "Latin-2") ; Serbian (Latin alphabet)
1698 ; ss Siswati
1699 ; st Sesotho
1700 ; su Sundanese
1701 ("sv" . "Latin-1") ; Swedish
1702 ("sw" . "Latin-1") ; Swahili
1703 ; ta Tamil
1704 ; te Telugu
1705 ; tg Tajik
1706 ("th" . "Thai")
1707 ; ti Tigrinya
1708 ; tk Turkmen
6ececc4d 1709 ("tl" . "Latin-1") ; Tagalog
40c81f74
PE
1710 ; tn Setswana
1711 ; to Tonga
1712 ("tr" . "Latin-5") ; Turkish
1713 ; ts Tsonga
1714 ; tt Tatar
1715 ; tw Twi
1716 ; ug Uighur
6ececc4d 1717 ("uk" . "Latin-5") ; Ukrainian
40c81f74
PE
1718 ; ur Urdu
1719 ; uz Uzbek
1720 ("vi" . "Vietnamese")
1721 ; vo Volapuk
1722 ; wo Wolof
1723 ; xh Xhosa
1724 ; yi Yiddish
1725 ; yo Yoruba
1726 ; za Zhuang
1727 ("zh.*[._]big5" . "Chinese-BIG5")
1728 ("zh.*[._]gbk" . nil) ; Solaris 2.7; has gbk-0 as well as GB 2312.1980-0
1729 ("zh_tw" . "Chinese-CNS")
1730 ("zh" . "Chinese-GB")
1731 ; zu Zulu
1732
1733 ;; ISO standard locales
1734 ("c$" . "ASCII")
1735 ("posix$" . "ASCII")
1736
40c81f74
PE
1737 ;; The "IPA" Emacs language environment does not correspond
1738 ;; to any ISO 639 code, so let it stand for itself.
1739 ("ipa$" . "IPA")
1740
1741 ;; Nonstandard or obsolete language codes
1742 ("cz" . "Czech") ; e.g. Solaris 2.6
1743 ("ee" . "Latin-4") ; Estonian, e.g. X11R6.4
1744 ("iw" . "Hebrew") ; e.g. X11R6.4
6ececc4d 1745 ("sp" . "Latin-5") ; Serbian (Cyrillic alphabet), e.g. X11R6.4
40c81f74
PE
1746 ("su" . "Latin-1") ; Finnish, e.g. Solaris 2.6
1747 )
1748 "List of pairs of locale regexps and language names.
6ececc4d
PE
1749The first element whose locale regexp matches the start of a downcased locale
1750specifies the language name corresponding to that locale.
40c81f74
PE
1751If the language name is nil, there is no corresponding language environment.")
1752
6ececc4d
PE
1753(defvar locale-charset-language-names
1754 '((".*8859[-_]?1\\>" . "Latin-1")
1755 (".*8859[-_]?2\\>" . "Latin-2")
1756 (".*8859[-_]?3\\>" . "Latin-3")
1757 (".*8859[-_]?4\\>" . "Latin-4")
1758 (".*8859[-_]?9\\>" . "Latin-5")
1759 (".*8859[-_]?14\\>" . "Latin-8")
1760 (".*8859[-_]?15\\>" . "Latin-9")
1761 )
1762 "List of pairs of locale regexps and charset language names.
1763The first element whose locale regexp matches the start of a downcased locale
1764specifies the language name whose charsets corresponds to that locale.
1765This language name is used if its charsets disagree with the charsets of
1766the language name that would otherwise be used for this locale.")
1767
40c81f74
PE
1768(defvar locale-preferred-coding-systems
1769 '(("ja.*[._]euc" . japanese-iso-8bit)
1770 ("ja.*[._]jis7" . iso-2022-jp)
1771 ("ja.*[._]pck" . japanese-shift-jis)
1772 ("ja.*[._]sjis" . japanese-shift-jis)
40c81f74 1773 )
6ececc4d
PE
1774 "List of pairs of locale regexps and preferred coding systems.
1775The first element whose locale regexp matches the start of a downcased locale
1776specifies the coding system to prefer when using that locale.")
40c81f74
PE
1777
1778(defun locale-name-match (key alist)
1779 "Search for KEY in ALIST, which should be a list of regexp-value pairs.
1780Return the value corresponding to the first regexp that matches the
1781start of KEY, or nil if there is no match."
1782 (let (element)
1783 (while (and alist (not element))
1784 (if (string-match (concat "^\\(" (car (car alist)) "\\)") key)
1785 (setq element (car alist)))
1786 (setq alist (cdr alist)))
1787 (cdr element)))
1788
1789(defun set-locale-environment (locale-name)
1790 "Set up multi-lingual environment for using LOCALE-NAME.
1791This sets the coding system priority and the default input method
1792and sometimes other things. LOCALE-NAME should be a string
1793which is the name of a locale supported by the system;
1794often it is of the form xx_XX.CODE, where xx is a language,
1795XX is a country, and CODE specifies a character set and coding system.
1796For example, the locale name \"ja_JP.EUC\" might name a locale
1797for Japanese in Japan using the `japanese-iso-8bit' coding-system.
1798
1799If LOCALE-NAME is nil, its value is taken from the environment.
1800
1801The locale names supported by your system can typically be found in a
1802directory named `/usr/share/locale' or `/usr/lib/locale'."
1803
0d7c5bb9
DL
1804 ;; Do this at runtime for the sake of binaries possibly transported
1805 ;; to a system without X.
1806 (setq locale-translation-file-name
1807 (let ((files
1808 '("/usr/lib/X11/locale/locale.alias" ; e.g. X11R6.4
1809 "/usr/X11R6/lib/X11/locale/locale.alias" ; e.g. RedHat 4.2
1810 "/usr/openwin/lib/locale/locale.alias" ; e.g. Solaris 2.6
1811 ;;
1812 ;; The following name appears after the X-related names above,
1813 ;; since the X-related names are what X actually uses.
1814 "/usr/share/locale/locale.alias" ; GNU/Linux sans X
1815 )))
1816 (while (and files (not (file-exists-p (car files))))
1817 (setq files (cdr files)))
1818 (car files)))
1819
40c81f74
PE
1820 (unless locale-name
1821 ;; Use the first of these three environment variables
1822 ;; that has a nonempty value.
1823 (let ((vars '("LC_ALL" "LC_CTYPE" "LANG")))
1824 (while (and vars (not (setq locale-name (getenv (car vars)))))
1825 (setq vars (cdr vars)))))
1826
1827 (when locale-name
1828
1829 ;; Translate "swedish" into "sv_SE.ISO8859-1", and so on,
1830 ;; using the translation file that many systems have.
1831 (when locale-translation-file-name
1832 (with-temp-buffer
1833 (insert-file-contents locale-translation-file-name)
1834 (when (re-search-forward
1835 (concat "^" (regexp-quote locale-name) ":?[ \t]+") nil t)
1836 (setq locale-name (buffer-substring (point) (line-end-position))))))
1837
1838 (setq locale-name (downcase locale-name))
1839
6ececc4d
PE
1840 (let ((language-name
1841 (locale-name-match locale-name locale-language-names))
1842 (charset-language-name
1843 (locale-name-match locale-name locale-charset-language-names))
1844 (coding-system
1845 (locale-name-match locale-name locale-preferred-coding-systems)))
1846
1847 (if (and charset-language-name
1848 (not
1849 (equal (get-language-info language-name 'charset)
1850 (get-language-info charset-language-name 'charset))))
1851 (setq language-name charset-language-name))
1852
40c81f74
PE
1853 (when language-name
1854
1855 ;; Set up for this character set. This is now the right way
1856 ;; to do it for both unibyte and multibyte modes.
1857 (set-language-environment language-name)
1858
1859 ;; If default-enable-multibyte-characters is nil,
1860 ;; we are using single-byte characters,
1861 ;; so the display table and terminal coding system are irrelevant.
1862 (when default-enable-multibyte-characters
1863 (set-display-table-and-terminal-coding-system language-name))
1864
1865 (setq locale-coding-system
1866 (car (get-language-info language-name 'coding-priority))))
1867
1868 (when coding-system
1869 (prefer-coding-system coding-system)
1870 (setq locale-coding-system coding-system)))))
1871\f
4ed46869
KH
1872;;; Charset property
1873
251d4f4b 1874(defun get-charset-property (charset propname)
4ed46869
KH
1875 "Return the value of CHARSET's PROPNAME property.
1876This is the last value stored with
96db204a 1877 (put-charset-property CHARSET PROPNAME VALUE)."
251d4f4b
KH
1878 (and (not (eq charset 'composition))
1879 (plist-get (charset-plist charset) propname)))
4ed46869 1880
251d4f4b 1881(defun put-charset-property (charset propname value)
4ed46869
KH
1882 "Store CHARSETS's PROPNAME property with value VALUE.
1883It can be retrieved with `(get-charset-property CHARSET PROPNAME)'."
1300d43f
KH
1884 (or (eq charset 'composition)
1885 (set-charset-plist charset
1886 (plist-put (charset-plist charset) propname value))))
4ed46869
KH
1887
1888;;; Character code property
1889(put 'char-code-property-table 'char-table-extra-slots 0)
1890
1891(defvar char-code-property-table
1892 (make-char-table 'char-code-property-table)
1893 "Char-table containing a property list of each character code.
1894
1895See also the documentation of `get-char-code-property' and
96db204a 1896`put-char-code-property'.")
4ed46869
KH
1897
1898(defun get-char-code-property (char propname)
1899 "Return the value of CHAR's PROPNAME property in `char-code-property-table'."
1900 (let ((plist (aref char-code-property-table char)))
1901 (if (listp plist)
1902 (car (cdr (memq propname plist))))))
1903
1904(defun put-char-code-property (char propname value)
1905 "Store CHAR's PROPNAME property with VALUE in `char-code-property-table'.
1906It can be retrieved with `(get-char-code-property CHAR PROPNAME)'."
1907 (let ((plist (aref char-code-property-table char)))
1908 (if plist
1909 (let ((slot (memq propname plist)))
1910 (if slot
1911 (setcar (cdr slot) value)
1912 (nconc plist (list propname value))))
1913 (aset char-code-property-table char (list propname value)))))
1914
a127b764
KH
1915\f
1916;; Pretty description of encoded string
1917
1918;; Alist of ISO 2022 control code vs the corresponding mnemonic string.
1919(defvar iso-2022-control-alist
1920 '((?\x1b . "ESC")
1921 (?\x0e . "SO")
1922 (?\x0f . "SI")
1923 (?\x8e . "SS2")
1924 (?\x8f . "SS3")
1925 (?\x9b . "CSI")))
1926
1927(defun encoded-string-description (str coding-system)
1928 "Return a pretty description of STR that is encoded by CODING-SYSTEM."
1929 (setq str (string-as-unibyte str))
993b2a7d 1930 (mapconcat
8c9d55a9 1931 (if (and coding-system (eq (coding-system-type coding-system) 2))
993b2a7d
KH
1932 ;; Try to get a pretty description for ISO 2022 escape sequences.
1933 (function (lambda (x) (or (cdr (assq x iso-2022-control-alist))
1934 (format "%02X" x))))
5de75f53 1935 (function (lambda (x) (format "0x%02X" x))))
993b2a7d 1936 str " "))
a127b764
KH
1937
1938(defun encode-coding-char (char coding-system)
1939 "Encode CHAR by CODING-SYSTEM and return the resulting string.
1940If CODING-SYSTEM can't safely encode CHAR, return nil."
bd953173
KH
1941 (let ((str1 (string-as-multibyte (char-to-string char)))
1942 (str2 (string-as-multibyte (make-string 2 char)))
a127b764
KH
1943 (safe-charsets (and coding-system
1944 (coding-system-get coding-system 'safe-charsets)))
8c9d55a9 1945 (charset (char-charset char))
a127b764
KH
1946 enc1 enc2 i1 i2)
1947 (when (or (eq safe-charsets t)
8c9d55a9
KH
1948 (eq charset 'ascii)
1949 (memq charset safe-charsets))
a127b764
KH
1950 ;; We must find the encoded string of CHAR. But, just encoding
1951 ;; CHAR will put extra control sequences (usually to designate
1952 ;; ASCII charaset) at the tail if type of CODING is ISO 2022.
1953 ;; To exclude such tailing bytes, we at first encode one-char
1954 ;; string and two-char string, then check how many bytes at the
1955 ;; tail of both encoded strings are the same.
1956
bd953173 1957 (setq enc1 (encode-coding-string str1 coding-system)
a127b764 1958 i1 (length enc1)
bd953173 1959 enc2 (encode-coding-string str2 coding-system)
a127b764
KH
1960 i2 (length enc2))
1961 (while (and (> i1 0) (= (aref enc1 (1- i1)) (aref enc2 (1- i2))))
1962 (setq i1 (1- i1) i2 (1- i2)))
1963
1964 ;; Now (substring enc1 i1) and (substring enc2 i2) are the same,
1965 ;; and they are the extra control sequences at the tail to
1966 ;; exclude.
1967 (substring enc2 0 i2))))
1968
1969
4ed46869 1970;;; mule-cmds.el ends here