(lispy_mouse_names): Add additional mouse names.
[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.
4ed46869
KH
5
6;; Keywords: mule, multilingual
7
8;; This file is part of GNU Emacs.
9
10;; GNU Emacs is free software; you can redistribute it and/or modify
11;; it under the terms of the GNU General Public License as published by
12;; the Free Software Foundation; either version 2, or (at your option)
13;; any later version.
14
15;; GNU Emacs is distributed in the hope that it will be useful,
16;; but WITHOUT ANY WARRANTY; without even the implied warranty of
17;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
18;; GNU General Public License for more details.
19
20;; You should have received a copy of the GNU General Public License
369314dc
KH
21;; along with GNU Emacs; see the file COPYING. If not, write to the
22;; Free Software Foundation, Inc., 59 Temple Place - Suite 330,
23;; Boston, MA 02111-1307, USA.
4ed46869
KH
24
25;;; Code:
26
27;;; MULE related key bindings and menus.
28
0709d285 29(defvar mule-keymap (make-sparse-keymap)
33d17698 30 "Keymap for Mule (Multilingual environment) specific commands.")
4ed46869 31
8f81f784 32;; Keep "C-x C-m ..." for mule specific commands.
0709d285 33(define-key ctl-x-map "\C-m" mule-keymap)
ef8a8c8c 34
4ed46869
KH
35(define-key mule-keymap "f" 'set-buffer-file-coding-system)
36(define-key mule-keymap "t" 'set-terminal-coding-system)
15b3e511
KH
37(define-key mule-keymap "k" 'set-keyboard-coding-system)
38(define-key mule-keymap "p" 'set-buffer-process-coding-system)
7624ebb9
KH
39(define-key mule-keymap "x" 'set-selection-coding-system)
40(define-key mule-keymap "X" 'set-next-selection-coding-system)
8b784951 41(define-key mule-keymap "\C-\\" 'set-input-method)
15b3e511 42(define-key mule-keymap "c" 'universal-coding-system-argument)
b4fba33f 43(define-key mule-keymap "l" 'set-language-environment)
4ed46869 44
281d03ec 45(define-key help-map "\C-L" 'describe-language-environment)
ac4a3a2d 46(define-key help-map "L" 'describe-language-environment)
4ed46869 47(define-key help-map "\C-\\" 'describe-input-method)
ac4a3a2d 48(define-key help-map "I" 'describe-input-method)
d0b9c3ab 49(define-key help-map "C" 'describe-coding-system)
4ed46869
KH
50(define-key help-map "h" 'view-hello-file)
51
538d88fb
EZ
52(defvar mule-menu-keymap
53 (make-sparse-keymap "Mule (Multilingual Environment)")
33d17698 54 "Keymap for Mule (Multilingual environment) menu specific commands.")
15b3e511 55
dcad02bc
EZ
56(defvar describe-language-environment-map
57 (make-sparse-keymap "Describe Language Environment"))
15b3e511 58
dcad02bc
EZ
59(defvar setup-language-environment-map
60 (make-sparse-keymap "Set Language Environment"))
15b3e511 61
dcad02bc
EZ
62(defvar set-coding-system-map
63 (make-sparse-keymap "Set Coding System"))
15b3e511 64
15b3e511 65(define-key-after mule-menu-keymap [set-language-environment]
538d88fb
EZ
66 (list 'menu-item "Set Language Environment" setup-language-environment-map
67 :help "Multilingual environment suitable for specific language")
15b3e511 68 t)
a61f401d 69(define-key-after mule-menu-keymap [mouse-set-font]
538d88fb
EZ
70 '(menu-item "Set Font/Fontset" mouse-set-font
71 :visible (fboundp 'generate-fontset-menu)
72 :help "Select a font from list of known fonts/fontsets")
a61f401d 73 t)
15b3e511
KH
74(define-key-after mule-menu-keymap [separator-mule]
75 '("--")
76 t)
77(define-key-after mule-menu-keymap [toggle-input-method]
538d88fb 78 '(menu-item "Toggle Input Method" toggle-input-method)
15b3e511 79 t)
8b784951 80(define-key-after mule-menu-keymap [set-input-method]
538d88fb 81 '(menu-item "Select Input Method..." set-input-method)
15b3e511
KH
82 t)
83(define-key-after mule-menu-keymap [separator-input-method]
84 '("--")
85 t)
15b3e511 86(define-key-after mule-menu-keymap [set-various-coding-system]
538d88fb
EZ
87 (list 'menu-item "Set Coding Systems" set-coding-system-map)
88 t)
89(define-key-after mule-menu-keymap [view-hello-file]
90 '(menu-item "Show Multi-lingual Text" view-hello-file
91 :enable (file-readable-p
92 (expand-file-name "HELLO" data-directory))
93 :help "Display file which says HELLO in many languages")
15b3e511
KH
94 t)
95(define-key-after mule-menu-keymap [separator-coding-system]
96 '("--")
97 t)
538d88fb
EZ
98(define-key-after mule-menu-keymap [describe-language-environment]
99 (list 'menu-item "Describe Language Environment"
100 describe-language-environment-map
101 :help "Show multilingual settings for specific language")
15b3e511 102 t)
538d88fb
EZ
103(define-key-after mule-menu-keymap [describe-input-method]
104 '(menu-item "Describe Input Method..." describe-input-method
105 :help "Keyboard layout for specific input method")
106 t)
107(define-key-after mule-menu-keymap [describe-coding-system]
108 '(menu-item "Describe Coding System..." describe-coding-system)
109 t)
110(define-key-after mule-menu-keymap [mule-diag]
111 '(menu-item "Show All of Mule Status" mule-diag
112 :help "Display multilingual environment settings")
15b3e511
KH
113 t)
114
115(define-key-after set-coding-system-map [set-buffer-file-coding-system]
538d88fb
EZ
116 '(menu-item "For Saving this Buffer" set-buffer-file-coding-system
117 :help "How to encode this buffer on disk")
15b3e511 118 t)
3a151e98 119(define-key-after set-coding-system-map [universal-coding-system-argument]
538d88fb
EZ
120 '(menu-item "For Next Command" universal-coding-system-argument
121 :help "Coding system to be used by next command")
3a151e98 122 t)
15b3e511 123(define-key-after set-coding-system-map [set-terminal-coding-system]
538d88fb
EZ
124 '(menu-item "For Terminal" set-terminal-coding-system
125 :enable (null (memq window-system '(x w32 mac)))
126 :help "How to encode terminal output")
15b3e511
KH
127 t)
128(define-key-after set-coding-system-map [set-keyboard-coding-system]
538d88fb
EZ
129 '(menu-item "For Keyboard" set-keyboard-coding-system
130 :help "How to decode keyboard input")
15b3e511
KH
131 t)
132(define-key-after set-coding-system-map [set-buffer-process-coding-system]
538d88fb
EZ
133 '(menu-item "For I/O with Subprocess" set-buffer-process-coding-system
134 :visible (fboundp 'start-process)
135 :enable (get-buffer-process (current-buffer))
136 :help "How to en/decode I/O from/to subprocess connected to this buffer")
15b3e511 137 t)
7624ebb9 138(define-key-after set-coding-system-map [set-selection-coding-system]
538d88fb
EZ
139 '(menu-item "For X Selections/Clipboard" set-selection-coding-system
140 :visible (display-selections-p)
141 :help "How to en/decode data to/from selection/clipboard")
7624ebb9
KH
142 t)
143(define-key-after set-coding-system-map [set-next-selection-coding-system]
538d88fb
EZ
144 '(menu-item "For Next X Selection" set-next-selection-coding-system
145 :visible (display-selections-p)
146 :help "How to en/decode next selection/clipboard operation")
7624ebb9 147 t)
15b3e511 148(define-key setup-language-environment-map
538d88fb 149 [Default] '(menu-item "Default" setup-specified-language-environment))
4ed46869 150
4ed46869
KH
151;; This should be a single character key binding because users use it
152;; very frequently while editing multilingual text. Now we can use
153;; only two such keys: "\C-\\" and "\C-^", but the latter is not
154;; convenient because it requires shifting on most keyboards. An
155;; alternative is "\C-\]" which is now bound to `abort-recursive-edit'
156;; but it won't be used that frequently.
157(define-key global-map "\C-\\" 'toggle-input-method)
158
a2ad45b9
RS
159;;; This is no good because people often type Shift-SPC
160;;; meaning to type SPC. -- rms.
161;;; ;; Here's an alternative key binding for X users (Shift-SPACE).
162;;; (define-key global-map [?\S- ] 'toggle-input-method)
b4fba33f 163
26d87040
EZ
164(defun coding-system-change-eol-conversion (coding-system eol-type)
165 "Return a coding system which differs from CODING-SYSTEM in eol conversion.
166The returned coding system converts end-of-line by EOL-TYPE
167but text as the same way as CODING-SYSTEM.
168EOL-TYPE should be `unix', `dos', `mac', or nil.
169If EOL-TYPE is nil, the returned coding system detects
170how end-of-line is formatted automatically while decoding.
171
172EOL-TYPE can be specified by an integer 0, 1, or 2.
173They means `unix', `dos', and `mac' respectively."
174 (if (symbolp eol-type)
175 (setq eol-type (cond ((eq eol-type 'unix) 0)
176 ((eq eol-type 'dos) 1)
177 ((eq eol-type 'mac) 2)
178 (t eol-type))))
179 (let ((orig-eol-type (coding-system-eol-type coding-system)))
180 (if (vectorp orig-eol-type)
181 (if (not eol-type)
182 coding-system
183 (aref orig-eol-type eol-type))
184 (let ((base (coding-system-base coding-system)))
185 (if (not eol-type)
186 base
187 (if (= eol-type orig-eol-type)
188 coding-system
189 (setq orig-eol-type (coding-system-eol-type base))
190 (if (vectorp orig-eol-type)
191 (aref orig-eol-type eol-type))))))))
192
193(defun coding-system-change-text-conversion (coding-system coding)
194 "Return a coding system which differs from CODING-SYSTEM in text conversion.
195The returned coding system converts text by CODING
196but end-of-line as the same way as CODING-SYSTEM.
197If CODING is nil, the returned coding system detects
198how text is formatted automatically while decoding."
199 (if (not coding)
200 (coding-system-base coding-system)
201 (let ((eol-type (coding-system-eol-type coding-system)))
202 (coding-system-change-eol-conversion
203 coding
204 (if (numberp eol-type) (aref [unix dos mac] eol-type))))))
205
4ed46869 206(defun toggle-enable-multibyte-characters (&optional arg)
6998e1a1
RS
207 "Change whether this buffer uses multibyte characters.
208With arg, use multibyte characters if the arg is positive.
209
210Note that this command does not convert the byte contents of
211the buffer; it only changes the way those bytes are interpreted.
212In general, therefore, this command *changes* the sequence of
213characters that the current buffer contains.
214
215We suggest you avoid using use this command unless you know what you
216are doing. If you use it by mistake, and the buffer is now displayed
217wrong, use this command again to toggle back to the right mode."
4ed46869 218 (interactive "P")
b7079457
RS
219 (let ((new-flag
220 (if (null arg) (null enable-multibyte-characters)
221 (> (prefix-numeric-value arg) 0))))
222 (set-buffer-multibyte new-flag))
4ed46869
KH
223 (force-mode-line-update))
224
225(defun view-hello-file ()
226 "Display the HELLO file which list up many languages and characters."
227 (interactive)
8f81f784
KH
228 ;; We have to decode the file in any environment.
229 (let ((default-enable-multibyte-characters t)
95fa03b4 230 (coding-system-for-read 'iso-2022-7bit))
8f81f784 231 (find-file-read-only (expand-file-name "HELLO" data-directory))))
4ed46869 232
15b3e511
KH
233(defun universal-coding-system-argument ()
234 "Execute an I/O command using the specified coding system."
235 (interactive)
34104362
KH
236 (let* ((default (and buffer-file-coding-system
237 (not (eq (coding-system-type buffer-file-coding-system)
238 t))
239 buffer-file-coding-system))
240 (coding-system (read-coding-system
241 (if default
242 (format "Coding system for following command (default, %s): " default)
243 "Coding system for following command: ")
244 default))
15b3e511 245 (keyseq (read-key-sequence
e14a8f4c 246 (format "Command to execute with %s:" coding-system)))
15b3e511
KH
247 (cmd (key-binding keyseq)))
248 (let ((coding-system-for-read coding-system)
249 (coding-system-for-write coding-system))
250 (message "")
251 (call-interactively cmd))))
252
de94d711 253(defun set-default-coding-systems (coding-system)
0c3154d2 254 "Set default value of various coding systems to CODING-SYSTEM.
387136f6 255This sets the following coding systems:
0c3154d2 256 o coding system of a newly created buffer
8efc03e1
KH
257 o default coding system for subprocess I/O
258This also sets the following values:
387136f6 259 o default value used as file-name-coding-system for converting file names.
03c35c83
EZ
260 o default value for the command `set-terminal-coding-system' (not on MSDOS)
261 o default value for the command `set-keyboard-coding-system'."
de94d711
KH
262 (check-coding-system coding-system)
263 (setq-default buffer-file-coding-system coding-system)
716184d4
RS
264 (if default-enable-multibyte-characters
265 (setq default-file-name-coding-system coding-system))
03c35c83
EZ
266 ;; If coding-system is nil, honor that on MS-DOS as well, so
267 ;; that they could reset the terminal coding system.
268 (unless (and (eq window-system 'pc) coding-system)
269 (setq default-terminal-coding-system coding-system))
de94d711
KH
270 (setq default-keyboard-coding-system coding-system)
271 (setq default-process-coding-system (cons coding-system coding-system)))
272
45d08cb2
KH
273(defalias 'update-iso-coding-systems 'update-coding-systems-internal)
274(make-obsolete 'update-iso-coding-systems 'update-coding-systems-internal)
275
0c3154d2
KH
276(defun prefer-coding-system (coding-system)
277 "Add CODING-SYSTEM at the front of the priority list for automatic detection.
387136f6 278This also sets the following coding systems:
0c3154d2 279 o coding system of a newly created buffer
8efc03e1
KH
280 o default coding system for subprocess I/O
281This also sets the following values:
387136f6 282 o default value used as file-name-coding-system for converting file names.
03c35c83
EZ
283 o default value for the command `set-terminal-coding-system' (not on MSDOS)
284 o default value for the command `set-keyboard-coding-system'
285
bd3ac67e
EZ
286If CODING-SYSTEM specifies a certain type of EOL conversion, the coding
287systems set by this function will use that type of EOL conversion.
288
03c35c83
EZ
289This command does not change the default value of terminal coding system
290for MS-DOS terminal, because DOS terminals only support a single coding
291system, and Emacs automatically sets the default to that coding system at
292startup."
0c3154d2
KH
293 (interactive "zPrefer coding system: ")
294 (if (not (and coding-system (coding-system-p coding-system)))
295 (error "Invalid coding system `%s'" coding-system))
296 (let ((coding-category (coding-system-category coding-system))
bd3ac67e
EZ
297 (base (coding-system-base coding-system))
298 (eol-type (coding-system-eol-type coding-system)))
0c3154d2
KH
299 (if (not coding-category)
300 ;; CODING-SYSTEM is no-conversion or undecided.
301 (error "Can't prefer the coding system `%s'" coding-system))
8efc03e1 302 (set coding-category (or base coding-system))
45d08cb2 303 (update-coding-systems-internal)
812cad80 304 (or (eq coding-category (car coding-category-list))
0c3154d2 305 ;; We must change the order.
812cad80 306 (set-coding-priority (list coding-category)))
8efc03e1
KH
307 (if (and base (interactive-p))
308 (message "Highest priority is set to %s (base of %s)"
309 base coding-system))
bd3ac67e 310 ;; If they asked for specific EOL conversion, honor that.
6f9dc4fd 311 (if (memq eol-type '(0 1 2))
bd3ac67e
EZ
312 (setq coding-system
313 (coding-system-change-eol-conversion base eol-type))
314 (setq coding-system base))
315 (set-default-coding-systems coding-system)))
0c3154d2 316
2c395d56 317(defun find-coding-systems-region-subset-p (list1 list2)
54b226f7
KH
318 "Return non-nil if all elements in LIST1 are included in LIST2.
319Comparison done with EQ."
320 (catch 'tag
321 (while list1
322 (or (memq (car list1) list2)
323 (throw 'tag nil))
324 (setq list1 (cdr list1)))
325 t))
326
3fc7dfe5 327(defun find-coding-systems-region (from to)
54b226f7
KH
328 "Return a list of proper coding systems to encode a text between FROM and TO.
329All coding systems in the list can safely encode any multibyte characters
330in the text.
331
e8dd0160 332If the text contains no multibyte characters, return a list of a single
3fc7dfe5
KH
333element `undecided'."
334 (find-coding-systems-for-charsets (find-charset-region from to)))
54b226f7 335
3fc7dfe5
KH
336(defun find-coding-systems-string (string)
337 "Return a list of proper coding systems to encode STRING.
338All coding systems in the list can safely encode any multibyte characters
339in STRING.
340
e8dd0160 341If STRING contains no multibyte characters, return a list of a single
3fc7dfe5
KH
342element `undecided'."
343 (find-coding-systems-for-charsets (find-charset-string string)))
344
345(defun find-coding-systems-for-charsets (charsets)
346 "Return a list of proper coding systems to encode characters of CHARSETS.
347CHARSETS is a list of character sets."
348 (if (or (null charsets)
349 (and (= (length charsets) 1)
350 (eq 'ascii (car charsets))))
351 '(undecided)
2a17b638 352 (setq charsets (delq 'composition charsets))
251d4f4b 353 (let ((l (coding-system-list 'base-only))
0e0a9ac3 354 (charset-preferred-codings
3fc7dfe5
KH
355 (mapcar (function
356 (lambda (x)
251d4f4b
KH
357 (if (eq x 'unknown)
358 'raw-text
0e0a9ac3 359 (get-charset-property x 'preferred-coding-system))))
3fc7dfe5 360 charsets))
34104362
KH
361 (priorities (mapcar (function (lambda (x) (symbol-value x)))
362 coding-category-list))
3fc7dfe5 363 codings coding safe)
251d4f4b
KH
364 (if (memq 'unknown charsets)
365 ;; The region contains invalid multibyte characters.
366 (setq l '(raw-text)))
3fc7dfe5
KH
367 (while l
368 (setq coding (car l) l (cdr l))
251d4f4b 369 (if (and (setq safe (coding-system-get coding 'safe-charsets))
3fc7dfe5 370 (or (eq safe t)
2c395d56 371 (find-coding-systems-region-subset-p charsets safe)))
3fc7dfe5 372 ;; We put the higher priority to coding systems included
0e0a9ac3 373 ;; in CHARSET-PREFERRED-CODINGS, and within them, put the
34104362 374 ;; higher priority to coding systems which support smaller
3fc7dfe5
KH
375 ;; number of charsets.
376 (let ((priority
34104362
KH
377 (+ (if (coding-system-get coding 'mime-charset) 4096 0)
378 (lsh (length (memq coding priorities)) 7)
0e0a9ac3 379 (if (memq coding charset-preferred-codings) 64 0)
34104362
KH
380 (if (> (coding-system-type coding) 0) 32 0)
381 (if (consp safe) (- 32 (length safe)) 0))))
3fc7dfe5
KH
382 (setq codings (cons (cons priority coding) codings)))))
383 (mapcar 'cdr
384 (sort codings (function (lambda (x y) (> (car x) (car y))))))
385 )))
54b226f7 386
51ed58ea
KH
387(defun find-multibyte-characters (from to &optional maxcount excludes)
388 "Find multibyte characters in the region specified by FROM and TO.
389If FROM is a string, find multibyte characters in the string.
390The return value is an alist of the following format:
391 ((CHARSET COUNT CHAR ...) ...)
392where
393 CHARSET is a character set,
394 COUNT is a number of characters,
395 CHARs are found characters of the character set.
396Optional 3rd arg MAXCOUNT limits how many CHARs are put in the above list.
251d4f4b
KH
397Optional 4th arg EXCLUDE is a list of character sets to be ignored.
398
399For invalid characters, CHARs are actually strings."
51ed58ea
KH
400 (let ((chars nil)
401 charset char)
402 (if (stringp from)
403 (let ((idx 0))
404 (while (setq idx (string-match "[^\000-\177]" from idx))
405 (setq char (aref from idx)
406 charset (char-charset char))
251d4f4b
KH
407 (if (eq charset 'unknown)
408 (setq char (match-string 0)))
409 (if (or (eq charset 'unknown)
410 (not (or (eq excludes t) (memq charset excludes))))
51ed58ea
KH
411 (let ((slot (assq charset chars)))
412 (if slot
413 (if (not (memq char (nthcdr 2 slot)))
414 (let ((count (nth 1 slot)))
415 (setcar (cdr slot) (1+ count))
416 (if (or (not maxcount) (< count maxcount))
417 (nconc slot (list char)))))
418 (setq chars (cons (list charset 1 char) chars)))))
419 (setq idx (1+ idx))))
420 (save-excursion
421 (goto-char from)
422 (while (re-search-forward "[^\000-\177]" to t)
423 (setq char (preceding-char)
424 charset (char-charset char))
251d4f4b
KH
425 (if (eq charset 'unknown)
426 (setq char (match-string 0)))
427 (if (or (eq charset 'unknown)
428 (not (or (eq excludes t) (memq charset excludes))))
51ed58ea
KH
429 (let ((slot (assq charset chars)))
430 (if slot
251d4f4b 431 (if (not (member char (nthcdr 2 slot)))
51ed58ea
KH
432 (let ((count (nth 1 slot)))
433 (setcar (cdr slot) (1+ count))
434 (if (or (not maxcount) (< count maxcount))
435 (nconc slot (list char)))))
436 (setq chars (cons (list charset 1 char) chars))))))))
437 (nreverse chars)))
438
c83c4f60
RS
439(defvar last-coding-system-specified nil
440 "Most recent coding system explicitly specified by the user when asked.
441This variable is set whenever Emacs asks the user which coding system
442to use in order to write a file. If you set it to nil explicitly,
443then call `write-region', then afterward this variable will be non-nil
444only if the user was explicitly asked and specified a coding system.")
445
54b226f7 446(defun select-safe-coding-system (from to &optional default-coding-system)
d5266ddf
KH
447 "Ask a user to select a safe coding system from candidates.
448The candidates of coding systems which can safely encode a text
449between FROM and TO are shown in a popup window.
54b226f7
KH
450
451Optional arg DEFAULT-CODING-SYSTEM specifies a coding system to be
452checked at first. If omitted, buffer-file-coding-system of the
453current buffer is used.
454
d5266ddf
KH
455If the text can be encoded safely by DEFAULT-CODING-SYSTEM, it is
456returned without any user interaction.
54b226f7
KH
457
458Kludgy feature: if FROM is a string, the string is the target text,
459and TO is ignored."
460 (or default-coding-system
461 (setq default-coding-system buffer-file-coding-system))
51ed58ea
KH
462 (let* ((charsets (if (stringp from) (find-charset-string from)
463 (find-charset-region from to)))
464 (safe-coding-systems (find-coding-systems-for-charsets charsets)))
251d4f4b
KH
465 (if (or (not enable-multibyte-characters)
466 (eq (car safe-coding-systems) 'undecided)
902e3f77 467 (eq default-coding-system 'no-conversion)
54b226f7
KH
468 (and default-coding-system
469 (memq (coding-system-base default-coding-system)
470 safe-coding-systems)))
471 default-coding-system
472
34104362
KH
473 ;; At first, change each coding system to the corresponding
474 ;; mime-charset name if it is also a coding system.
475 (let ((l safe-coding-systems)
476 mime-charset)
477 (while l
478 (setq mime-charset (coding-system-get (car l) 'mime-charset))
479 (if (and mime-charset (coding-system-p mime-charset))
480 (setcar l mime-charset))
481 (setq l (cdr l))))
482
51ed58ea
KH
483 (let ((non-safe-chars (find-multibyte-characters
484 from to 3
485 (and default-coding-system
486 (coding-system-get default-coding-system
487 'safe-charsets))))
fcdc4d67 488 show-position overlays)
51ed58ea
KH
489 (save-excursion
490 ;; Highlight characters that default-coding-system can't encode.
491 (when (integerp from)
492 (goto-char from)
493 (let ((found nil))
494 (while (and (not found)
495 (re-search-forward "[^\000-\177]" to t))
496 (setq found (assq (char-charset (preceding-char))
497 non-safe-chars))))
fcdc4d67
KH
498 (forward-line -1)
499 (setq show-position (point))
51ed58ea 500 (save-excursion
265519e3
RS
501 (while (and (< (length overlays) 256)
502 (re-search-forward "[^\000-\177]" to t))
51ed58ea
KH
503 (let* ((char (preceding-char))
504 (charset (char-charset char)))
505 (when (assq charset non-safe-chars)
506 (setq overlays (cons (make-overlay (1- (point)) (point))
507 overlays))
508 (overlay-put (car overlays) 'face 'highlight))))))
509
510 ;; At last, ask a user to select a proper coding system.
511 (unwind-protect
512 (save-window-excursion
fcdc4d67
KH
513 (when show-position
514 ;; At first, be sure to show the current buffer.
515 (set-window-buffer (selected-window) (current-buffer))
516 (set-window-start (selected-window) show-position))
517 ;; Then, show a helpful message.
51ed58ea
KH
518 (with-output-to-temp-buffer "*Warning*"
519 (save-excursion
520 (set-buffer standard-output)
521 (insert "The target text contains the following non ASCII character(s):\n")
522 (let ((len (length non-safe-chars))
523 (shown 0))
524 (while (and non-safe-chars (< shown 3))
525 (when (> (length (car non-safe-chars)) 2)
526 (setq shown (1+ shown))
527 (insert (format "%25s: " (car (car non-safe-chars))))
528 (let ((l (nthcdr 2 (car non-safe-chars))))
529 (while l
251d4f4b
KH
530 (if (or (stringp (car l)) (char-valid-p (car l)))
531 (insert (car l)))
51ed58ea
KH
532 (setq l (cdr l))))
533 (if (> (nth 1 (car non-safe-chars)) 3)
534 (insert "..."))
535 (insert "\n"))
536 (setq non-safe-chars (cdr non-safe-chars)))
537 (if (< shown len)
538 (insert (format "%27s\n" "..."))))
2e224638
SM
539 (insert (format
540"These can't be encoded safely by the coding system %s.
54b226f7
KH
541
542Please select one from the following safe coding systems:\n"
51ed58ea
KH
543 default-coding-system))
544 (let ((pos (point))
545 (fill-prefix " "))
546 (mapcar (function (lambda (x) (princ " ") (princ x)))
547 safe-coding-systems)
548 (fill-region-as-paragraph pos (point)))))
549
550 ;; Read a coding system.
551 (let* ((safe-names (mapcar (lambda (x) (list (symbol-name x)))
552 safe-coding-systems))
553 (name (completing-read
554 (format "Select coding system (default %s): "
555 (car safe-coding-systems))
556 safe-names nil t nil nil
557 (car (car safe-names)))))
558 (setq last-coding-system-specified (intern name))
559 (if (integerp (coding-system-eol-type default-coding-system))
560 (setq last-coding-system-specified
561 (coding-system-change-eol-conversion
562 last-coding-system-specified
563 (coding-system-eol-type default-coding-system))))
564 last-coding-system-specified))
565 (kill-buffer "*Warning*")
566 (while overlays
567 (delete-overlay (car overlays))
568 (setq overlays (cdr overlays)))))))))
54b226f7
KH
569
570(setq select-safe-coding-system-function 'select-safe-coding-system)
571
46babb23
KH
572(defun select-message-coding-system ()
573 "Return a coding system to encode the outgoing message of the current buffer.
574It at first tries the first coding system found in these variables
575in this order:
576 (1) local value of `buffer-file-coding-system'
577 (2) value of `sendmail-coding-system'
578 (3) value of `default-buffer-file-coding-system'
579 (4) value of `default-sendmail-coding-system'
580If the found coding system can't encode the current buffer,
581or none of them are bound to a coding system,
48e41165 582it asks the user to select a proper coding system."
46babb23
KH
583 (let ((coding (or (and (local-variable-p 'buffer-file-coding-system)
584 buffer-file-coding-system)
585 sendmail-coding-system
586 default-buffer-file-coding-system
587 default-sendmail-coding-system)))
588 (if (eq coding 'no-conversion)
589 ;; We should never use no-conversion for outgoing mails.
590 (setq coding nil))
591 (if (fboundp select-safe-coding-system-function)
592 (funcall select-safe-coding-system-function
593 (point-min) (point-max) coding)
594 coding)))
4ed46869 595\f
03c35c83 596;;; Language support stuff.
4ed46869 597
4ed46869 598(defvar language-info-alist nil
2c395d56 599 "Alist of language environment definitions.
4ed46869
KH
600Each element looks like:
601 (LANGUAGE-NAME . ((KEY . INFO) ...))
2c395d56
RS
602where LANGUAGE-NAME is a string, the name of the language environment,
603KEY is a symbol denoting the kind of information, and
604INFO is the data associated with KEY.
605Meaningful values for KEY include
606
607 documentation value is documentation of what this language environment
608 is meant for, and how to use it.
609 charset value is a list of the character sets used by this
610 language environment.
611 sample-text value is one line of text,
612 written using those character sets,
613 appropriate for this language environment.
614 setup-function value is a function to call to switch to this
615 language environment.
616 exit-function value is a function to call to leave this
617 language environment.
618 coding-system value is a list of coding systems that are good
619 for saving text written in this language environment.
620 This list serves as suggestions to the user;
621 in effect, as a kind of documentation.
622 coding-priority value is a list of coding systems for this language
623 environment, in order of decreasing priority.
624 This is used to set up the coding system priority
45d08cb2 625 list when you switch to this language environment.
ddb5c041 626 nonascii-translation
7624ebb9 627 value is a translation table to be set in the
45d08cb2 628 variable `nonascii-translation-table' in this
7624ebb9
KH
629 language environment, or a character set from
630 which `nonascii-insert-offset' is calculated.
ddb5c041
KH
631 input-method value is a default input method for this language
632 environment.
7624ebb9
KH
633 features value is a list of features requested in this
634 language environment.
ddb5c041
KH
635
636The following keys take effect only when multibyte characters are
637globally disabled, i.e. the value of `default-enable-multibyte-characters'
638is nil.
639
640 unibyte-syntax value is a library name to load to set
e8dd0160 641 unibyte 8-bit character syntaxes for this
ddb5c041
KH
642 language environment.
643
644 unibyte-display value is a coding system to encode characters
645 for the terminal. Characters in the range
646 of 160 to 255 display not as octal escapes,
647 but as non-ASCII characters in this language
648 environment.")
2c395d56
RS
649
650(defun get-language-info (lang-env key)
651 "Return information listed under KEY for language environment LANG-ENV.
652KEY is a symbol denoting the kind of information.
653For a list of useful values for KEY and their meanings,
654see `language-info-alist'."
655 (if (symbolp lang-env)
656 (setq lang-env (symbol-name lang-env)))
657 (let ((lang-slot (assoc-ignore-case lang-env language-info-alist)))
4ed46869
KH
658 (if lang-slot
659 (cdr (assq key (cdr lang-slot))))))
660
f08adf27 661(defun set-language-info (lang-env key info)
2c395d56
RS
662 "Modify part of the definition of language environment LANG-ENV.
663Specifically, this stores the information INFO under KEY
664in the definition of this language environment.
4ed46869 665KEY is a symbol denoting the kind of information.
2c395d56 666INFO is the value for that information.
281d03ec 667
2c395d56 668For a list of useful values for KEY and their meanings,
f08adf27 669see `language-info-alist'."
2c395d56
RS
670 (if (symbolp lang-env)
671 (setq lang-env (symbol-name lang-env)))
4ed46869 672 (let (lang-slot key-slot)
2c395d56 673 (setq lang-slot (assoc lang-env language-info-alist))
4ed46869 674 (if (null lang-slot) ; If no slot for the language, add it.
2c395d56 675 (setq lang-slot (list lang-env)
4ed46869
KH
676 language-info-alist (cons lang-slot language-info-alist)))
677 (setq key-slot (assq key lang-slot))
678 (if (null key-slot) ; If no slot for the key, add it.
679 (progn
680 (setq key-slot (list key))
681 (setcdr lang-slot (cons key-slot (cdr lang-slot)))))
f08adf27 682 (setcdr key-slot info)))
4ed46869 683
2c395d56
RS
684(defun set-language-info-alist (lang-env alist &optional parents)
685 "Store ALIST as the definition of language environment LANG-ENV.
686ALIST is an alist of KEY and INFO values. See the documentation of
7624ebb9 687`set-language-info' for the meanings of KEY and INFO.
54b226f7 688
2c395d56
RS
689Optional arg PARENTS is a list of parent menu names; it specifies
690where to put this language environment in the
691Describe Language Environment and Set Language Environment menus.
692For example, (\"European\") means to put this language environment
693in the European submenu in each of those two menus."
694 (if (symbolp lang-env)
695 (setq lang-env (symbol-name lang-env)))
54b226f7
KH
696 (let ((describe-map describe-language-environment-map)
697 (setup-map setup-language-environment-map))
698 (if parents
699 (let ((l parents)
700 map parent-symbol parent)
701 (while l
702 (if (symbolp (setq parent-symbol (car l)))
703 (setq parent (symbol-name parent))
704 (setq parent parent-symbol parent-symbol (intern parent)))
705 (setq map (lookup-key describe-map (vector parent-symbol)))
706 (if (not map)
707 (progn
708 (setq map (intern (format "describe-%s-environment-map"
709 (downcase parent))))
710 (define-prefix-command map)
711 (define-key-after describe-map (vector parent-symbol)
712 (cons parent map) t)))
713 (setq describe-map (symbol-value map))
714 (setq map (lookup-key setup-map (vector parent-symbol)))
715 (if (not map)
716 (progn
717 (setq map (intern (format "setup-%s-environment-map"
718 (downcase parent))))
719 (define-prefix-command map)
720 (define-key-after setup-map (vector parent-symbol)
721 (cons parent map) t)))
722 (setq setup-map (symbol-value map))
723 (setq l (cdr l)))))
f08adf27
RS
724
725 ;; Set up menu items for this language env.
7624ebb9 726 (let ((doc (assq 'documentation alist)))
f08adf27
RS
727 (when doc
728 (define-key-after describe-map (vector (intern lang-env))
7624ebb9
KH
729 (cons lang-env 'describe-specified-language-support) t)))
730 (define-key-after setup-map (vector (intern lang-env))
731 (cons lang-env 'setup-specified-language-environment) t)
f08adf27 732
54b226f7 733 (while alist
f08adf27 734 (set-language-info lang-env (car (car alist)) (cdr (car alist)))
54b226f7 735 (setq alist (cdr alist)))))
4ed46869 736
ae302641 737(defun read-language-name (key prompt &optional default)
2c395d56 738 "Read a language environment name which has information for KEY.
ddb5c041 739If KEY is nil, read any language environment.
2c395d56
RS
740Prompt with PROMPT. DEFAULT is the default choice of language environment.
741This returns a language environment name as a string."
4ed46869
KH
742 (let* ((completion-ignore-case t)
743 (name (completing-read prompt
744 language-info-alist
ddb5c041
KH
745 (and key
746 (function (lambda (elm) (assq key elm))))
ae302641 747 t nil nil default)))
13e82c04 748 (if (and (> (length name) 0)
ddb5c041
KH
749 (or (not key)
750 (get-language-info name key)))
13e82c04 751 name)))
4ed46869
KH
752\f
753;;; Multilingual input methods.
754
d0b9c3ab
KH
755(defconst leim-list-file-name "leim-list.el"
756 "Name of LEIM list file.
757This file contains a list of libraries of Emacs input methods (LEIM)
758in the format of Lisp expression for registering each input method.
759Emacs loads this file at startup time.")
760
2e224638
SM
761(defvar leim-list-header (format
762";;; %s -- list of LEIM (Library of Emacs Input Method)
d0b9c3ab
KH
763;;
764;; This file contains a list of LEIM (Library of Emacs Input Method)
e8dd0160 765;; in the same directory as this file. Loading this file registers
d0b9c3ab
KH
766;; the whole input methods in Emacs.
767;;
d33d5fbe 768;; Each entry has the form:
d0b9c3ab
KH
769;; (register-input-method
770;; INPUT-METHOD LANGUAGE-NAME ACTIVATE-FUNC
771;; TITLE DESCRIPTION
772;; ARG ...)
773;; See the function `register-input-method' for the meanings of arguments.
774;;
775;; If this directory is included in load-path, Emacs automatically
776;; loads this file at startup time.
777
778"
779 leim-list-file-name)
780 "Header to be inserted in LEIM list file.")
781
e55e92ee 782(defvar leim-list-entry-regexp "^(register-input-method"
d0b9c3ab
KH
783 "Regexp matching head of each entry in LEIM list file.
784See also the variable `leim-list-header'")
785
786(defvar update-leim-list-functions
787 '(quail-update-leim-list-file)
788 "List of functions to call to update LEIM list file.
789Each function is called with one arg, LEIM directory name.")
790
a337fe7f
RS
791(defun update-leim-list-file (&rest dirs)
792 "Update LEIM list file in directories DIRS."
d0b9c3ab
KH
793 (let ((functions update-leim-list-functions))
794 (while functions
a337fe7f 795 (apply (car functions) dirs)
d0b9c3ab
KH
796 (setq functions (cdr functions)))))
797
4ed46869
KH
798(defvar current-input-method nil
799 "The current input method for multilingual text.
96db204a 800If nil, that means no input method is activated now.")
4ed46869
KH
801(make-variable-buffer-local 'current-input-method)
802(put 'current-input-method 'permanent-local t)
803
804(defvar current-input-method-title nil
d0b9c3ab 805 "Title string of the current input method shown in mode line.")
4ed46869
KH
806(make-variable-buffer-local 'current-input-method-title)
807(put 'current-input-method-title 'permanent-local t)
808
b4fba33f 809(defcustom default-input-method nil
8861c593 810 "*Default input method for multilingual text (a string).
b4fba33f 811This is the input method activated automatically by the command
9b10b5a3 812`toggle-input-method' (\\[toggle-input-method])."
8861c593 813 :group 'mule
5806e8a6
GM
814 :type '(choice (const nil) string)
815 :set-after '(current-language-environment))
b4fba33f 816
0f835e87
KH
817(put 'input-method-function 'permanent-local t)
818
723a427a
KH
819(defvar input-method-history nil
820 "History list for some commands that read input methods.")
821(make-variable-buffer-local 'input-method-history)
822(put 'input-method-history 'permanent-local t)
4ed46869
KH
823
824(defvar inactivate-current-input-method-function nil
825 "Function to call for inactivating the current input method.
826Every input method should set this to an appropriate value when activated.
f17ccaee
KH
827This function is called with no argument.
828
829This function should never change the value of `current-input-method'.
830It is set to nil by the function `inactivate-input-method'.")
4ed46869
KH
831(make-variable-buffer-local 'inactivate-current-input-method-function)
832(put 'inactivate-current-input-method-function 'permanent-local t)
833
834(defvar describe-current-input-method-function nil
835 "Function to call for describing the current input method.
836This function is called with no argument.")
837(make-variable-buffer-local 'describe-current-input-method-function)
838(put 'describe-current-input-method-function 'permanent-local t)
839
d0b9c3ab 840(defvar input-method-alist nil
2c395d56 841 "Alist of input method names vs how to use them.
d0b9c3ab 842Each element has the form:
2c395d56
RS
843 (INPUT-METHOD LANGUAGE-ENV ACTIVATE-FUNC TITLE DESCRIPTION ARGS...)
844See the function `register-input-method' for the meanings of the elements.")
845
f08adf27 846(defun register-input-method (input-method lang-env &rest args)
2c395d56 847 "Register INPUT-METHOD as an input method for language environment ENV.
f08adf27 848INPUT-METHOD and LANG-ENV are symbols or strings.
d0b9c3ab 849
d0b9c3ab 850The remaining arguments are:
2c395d56
RS
851 ACTIVATE-FUNC, TITLE, DESCRIPTION, and ARGS...
852ACTIVATE-FUNC is a function to call to activate this method.
853TITLE is a string to show in the mode line when this method is active.
854DESCRIPTION is a string describing this method and what it is good for.
855The ARGS, if any, are passed as arguments to ACTIVATE-FUNC.
205814ee
KH
856All told, the arguments to ACTIVATE-FUNC are INPUT-METHOD and the ARGS.
857
858This function is mainly used in the file \"leim-list.el\" which is
859created at building time of emacs, registering all quail input methods
860contained in the emacs distribution.
861
862In case you want to register a new quail input method by yourself, be
863careful to use the same input method title as given in the third
864parameter of `quail-define-package' (if the values are different, the
865string specified in this function takes precedence).
866
867The commands `describe-input-method' and `list-input-methods' need
868this duplicated values to show some information about input methods
869without loading the affected quail packages."
f08adf27
RS
870 (if (symbolp lang-env)
871 (setq lang-env (symbol-name lang-env)))
4ef06f75
KH
872 (if (symbolp input-method)
873 (setq input-method (symbol-name input-method)))
f08adf27 874 (let ((info (cons lang-env args))
d0b9c3ab
KH
875 (slot (assoc input-method input-method-alist)))
876 (if slot
877 (setcdr slot info)
878 (setq slot (cons input-method info))
879 (setq input-method-alist (cons slot input-method-alist)))))
880
4d5ac029 881(defun read-input-method-name (prompt &optional default inhibit-null)
d0b9c3ab 882 "Read a name of input method from a minibuffer prompting with PROMPT.
4d5ac029
RS
883If DEFAULT is non-nil, use that as the default,
884 and substitute it into PROMPT at the first `%s'.
4ef06f75
KH
885If INHIBIT-NULL is non-nil, null input signals an error.
886
887The return value is a string."
4d5ac029
RS
888 (if default
889 (setq prompt (format prompt default)))
d0b9c3ab 890 (let* ((completion-ignore-case t)
723a427a
KH
891 ;; This binding is necessary because input-method-history is
892 ;; buffer local.
d0b9c3ab 893 (input-method (completing-read prompt input-method-alist
87505a98
RS
894 nil t nil 'input-method-history
895 default)))
bf294e6e
KH
896 (if (and input-method (symbolp input-method))
897 (setq input-method (symbol-name input-method)))
d0b9c3ab
KH
898 (if (> (length input-method) 0)
899 input-method
900 (if inhibit-null
43807b77 901 (error "No valid input method is specified")))))
d0b9c3ab 902
d0b9c3ab 903(defun activate-input-method (input-method)
2c395d56
RS
904 "Switch to input method INPUT-METHOD for the current buffer.
905If some other input method is already active, turn it off first.
906If INPUT-METHOD is nil, deactivate any current input method."
305a3cb6 907 (if (and input-method (symbolp input-method))
4ef06f75 908 (setq input-method (symbol-name input-method)))
723a427a
KH
909 (if (and current-input-method
910 (not (string= current-input-method input-method)))
305a3cb6 911 (inactivate-input-method))
2c395d56 912 (unless (or current-input-method (null input-method))
d0b9c3ab
KH
913 (let ((slot (assoc input-method input-method-alist)))
914 (if (null slot)
723a427a 915 (error "Can't activate input method `%s'" input-method))
8efc03e1
KH
916 (let ((func (nth 2 slot)))
917 (if (functionp func)
918 (apply (nth 2 slot) input-method (nthcdr 5 slot))
919 (if (and (consp func) (symbolp (car func)) (symbolp (cdr func)))
920 (progn
921 (require (cdr func))
922 (apply (car func) input-method (nthcdr 5 slot)))
923 (error "Can't activate input method `%s'" input-method))))
d0b9c3ab 924 (setq current-input-method input-method)
723a427a 925 (setq current-input-method-title (nth 3 slot))
28885c0e
KH
926 (unwind-protect
927 (run-hooks 'input-method-activate-hook)
928 (force-mode-line-update)))))
15b3e511 929
15b3e511 930(defun inactivate-input-method ()
f17ccaee 931 "Turn off the current input method."
723a427a
KH
932 (when current-input-method
933 (if input-method-history
934 (unless (string= current-input-method (car input-method-history))
935 (setq input-method-history
936 (cons current-input-method
937 (delete current-input-method input-method-history))))
938 (setq input-method-history (list current-input-method)))
939 (unwind-protect
940 (funcall inactivate-current-input-method-function)
15b3e511 941 (unwind-protect
723a427a
KH
942 (run-hooks 'input-method-inactivate-hook)
943 (setq current-input-method nil
28885c0e
KH
944 current-input-method-title nil)
945 (force-mode-line-update)))))
4ed46869 946
8b784951 947(defun set-input-method (input-method)
2c395d56
RS
948 "Select and activate input method INPUT-METHOD for the current buffer.
949This also sets the default input method to the one you specify."
d0b9c3ab 950 (interactive
723a427a 951 (let* ((default (or (car input-method-history) default-input-method)))
42395763 952 (list (read-input-method-name
87505a98 953 (if default "Select input method (default %s): " "Select input method: ")
42395763 954 default t))))
d0b9c3ab 955 (activate-input-method input-method)
42395763 956 (setq default-input-method input-method))
4ed46869
KH
957
958(defun toggle-input-method (&optional arg)
15b3e511 959 "Turn on or off a multilingual text input method for the current buffer.
723a427a 960
f2979bdb
KH
961With no prefix argument, if an input method is currently activated,
962turn it off. Otherwise, activate an input method -- the one most
963recently used, or the one specified in `default-input-method', or
964the one read from the minibuffer.
723a427a 965
f2979bdb
KH
966With a prefix argument, read an input method from the minibuffer and
967turn it on.
723a427a 968
f2979bdb
KH
969The default is to use the most recent input method specified
970\(not including the currently active input method, if any)."
4ed46869 971 (interactive "P")
7ddbb5bc
RS
972 (if (and current-input-method (not arg))
973 (inactivate-input-method)
974 (let ((default (or (car input-method-history) default-input-method)))
975 (if (and arg default (equal current-input-method default)
976 (> (length input-method-history) 1))
977 (setq default (nth 1 input-method-history)))
723a427a
KH
978 (activate-input-method
979 (if (or arg (not default))
7ddbb5bc
RS
980 (progn
981 (read-input-method-name
982 (if default "Input method (default %s): " "Input method: " )
983 default t))
723a427a
KH
984 default))
985 (or default-input-method
986 (setq default-input-method current-input-method)))))
d0b9c3ab
KH
987
988(defun describe-input-method (input-method)
2c395d56 989 "Describe input method INPUT-METHOD."
d0b9c3ab
KH
990 (interactive
991 (list (read-input-method-name
992 "Describe input method (default, current choice): ")))
78754934 993 (if (and input-method (symbolp input-method))
4ef06f75 994 (setq input-method (symbol-name input-method)))
d0b9c3ab
KH
995 (if (null input-method)
996 (describe-current-input-method)
997 (with-output-to-temp-buffer "*Help*"
998 (let ((elt (assoc input-method input-method-alist)))
999 (princ (format "Input method: %s (`%s' in mode line) for %s\n %s\n"
1000 input-method (nth 3 elt) (nth 1 elt) (nth 4 elt)))))))
1001
1002(defun describe-current-input-method ()
96db204a 1003 "Describe the input method currently in use."
4ed46869
KH
1004 (if current-input-method
1005 (if (and (symbolp describe-current-input-method-function)
1006 (fboundp describe-current-input-method-function))
1007 (funcall describe-current-input-method-function)
1008 (message "No way to describe the current input method `%s'"
f2979bdb 1009 current-input-method)
4ed46869 1010 (ding))
d0b9c3ab 1011 (error "No input method is activated now")))
4ed46869 1012
d3459641 1013(defun read-multilingual-string (prompt &optional initial-input input-method)
4ed46869
KH
1014 "Read a multilingual string from minibuffer, prompting with string PROMPT.
1015The input method selected last time is activated in minibuffer.
15b3e511 1016If optional second arg INITIAL-INPUT is non-nil, insert it in the minibuffer
d0b9c3ab
KH
1017initially.
1018Optional 3rd argument INPUT-METHOD specifies the input method
4ef06f75
KH
1019to be activated instead of the one selected last time. It is a symbol
1020or a string."
88d559ec
KH
1021 (setq input-method
1022 (or input-method
d3459641 1023 current-input-method
88d559ec
KH
1024 default-input-method
1025 (read-input-method-name "Input method: " nil t)))
3df60841 1026 (if (and input-method (symbolp input-method))
4ef06f75 1027 (setq input-method (symbol-name input-method)))
305a3cb6
KH
1028 (let ((prev-input-method current-input-method))
1029 (unwind-protect
1030 (progn
1031 (activate-input-method input-method)
1032 (read-string prompt initial-input nil nil t))
1033 (activate-input-method prev-input-method))))
4ed46869
KH
1034
1035;; Variables to control behavior of input methods. All input methods
1036;; should react to these variables.
1037
8efc03e1
KH
1038(defcustom input-method-verbose-flag 'default
1039 "*A flag to control extra guidance given by input methods.
1040The value should be nil, t, `complex-only', or `default'.
4ed46869 1041
cb29dfb6 1042The extra guidance is done by showing list of available keys in echo
8efc03e1
KH
1043area. When you use the input method in the minibuffer, the guidance
1044is shown at the bottom short window (split from the existing window).
c27c4ed8 1045
8efc03e1
KH
1046If the value is t, extra guidance is always given, if the value is
1047nil, extra guidance is always suppressed.
1048
1049If the value is `complex-only', only complex input methods such as
1050`chinese-py' and `japanese' give extra guidance.
1051
1052If the value is `default', complex input methods always give extra
1053guidance, but simple input methods give it only when you are not in
1054the minibuffer.
1055
1056See also the variable `input-method-highlight-flag'."
1057 :type '(choice (const t) (const nil) (const complex-only) (const default))
42395763
RS
1058 :group 'mule)
1059
1060(defcustom input-method-highlight-flag t
1061 "*If this flag is non-nil, input methods highlight partially-entered text.
1062For instance, while you are in the middle of a Quail input method sequence,
1063the text inserted so far is temporarily underlined.
8efc03e1
KH
1064The underlining goes away when you finish or abort the input method sequence.
1065See also the variable `input-method-verbose-flag'."
42395763
RS
1066 :type 'boolean
1067 :group 'mule)
4ed46869
KH
1068
1069(defvar input-method-activate-hook nil
f17ccaee
KH
1070 "Normal hook run just after an input method is activated.
1071
1072The variable `current-input-method' keeps the input method name
1073just activated.")
4ed46869
KH
1074
1075(defvar input-method-inactivate-hook nil
f17ccaee
KH
1076 "Normal hook run just after an input method is inactivated.
1077
1078The variable `current-input-method' still keeps the input method name
4d0e6a11 1079just inactivated.")
4ed46869
KH
1080
1081(defvar input-method-after-insert-chunk-hook nil
1082 "Normal hook run just after an input method insert some chunk of text.")
1083
dccca980
KH
1084(defvar input-method-exit-on-first-char nil
1085 "This flag controls a timing when an input method returns.
1086Usually, the input method does not return while there's a possibility
1087that it may find a different translation if a user types another key.
39e643e2
RS
1088But, it this flag is non-nil, the input method returns as soon as
1089the current key sequence gets long enough to have some valid translation.")
dccca980
KH
1090
1091(defvar input-method-use-echo-area nil
1092 "This flag controls how an input method shows an intermediate key sequence.
39e643e2
RS
1093Usually, the input method inserts the intermediate key sequence,
1094or candidate translations corresponding to the sequence,
1095at point in the current buffer.
1096But, if this flag is non-nil, it displays them in echo area instead.")
dccca980 1097
723a427a
KH
1098(defvar input-method-exit-on-invalid-key nil
1099 "This flag controls the behaviour of an input method on invalid key input.
1100Usually, when a user types a key which doesn't start any character
1101handled by the input method, the key is handled by turning off the
e8dd0160 1102input method temporarily. After that key, the input method is re-enabled.
723a427a
KH
1103But, if this flag is non-nil, the input method is never back on.")
1104
4ed46869 1105\f
8efc03e1
KH
1106(defvar set-language-environment-hook nil
1107 "Normal hook run after some language environment is set.
1108
1109When you set some hook function here, that effect usually should not
1110be inherited to another language environment. So, you had better set
1111another function in `exit-language-environment-hook' (which see) to
1112cancel the effect.")
1113
1114(defvar exit-language-environment-hook nil
1115 "Normal hook run after exiting from some language environment.
1116When this hook is run, the variable `current-language-environment'
1117is still bound to the language environment being exited.
1118
e8dd0160 1119This hook is mainly used for canceling the effect of
8efc03e1
KH
1120`set-language-environment-hook' (which-see).")
1121
b0648a00
RS
1122(put 'setup-specified-language-environment 'apropos-inhibit t)
1123
15b3e511 1124(defun setup-specified-language-environment ()
f08adf27 1125 "Switch to a specified language environment."
15b3e511 1126 (interactive)
f850d782 1127 (let (language-name)
15b3e511
KH
1128 (if (and (symbolp last-command-event)
1129 (or (not (eq last-command-event 'Default))
1130 (setq last-command-event 'English))
f850d782
RS
1131 (setq language-name (symbol-name last-command-event)))
1132 (set-language-environment language-name)
15b3e511 1133 (error "Bogus calling sequence"))))
4ed46869 1134
8861c593 1135(defcustom current-language-environment "English"
94d04df6 1136 "The last language environment specified with `set-language-environment'.
ebef6d93
KH
1137This variable should be set only with \\[customize], which is equivalent
1138to using the function `set-language-environment'."
94d04df6 1139 :link '(custom-manual "(emacs)Language Environments")
dff1aa24 1140 :set (lambda (symbol value) (set-language-environment value))
94d04df6
DL
1141 :get (lambda (x)
1142 (or (car-safe (assoc-ignore-case
1143 (if (symbolp current-language-environment)
1144 (symbol-name current-language-environment)
1145 current-language-environment)
1146 language-info-alist))
1147 "English"))
1148 :type (cons 'choice (mapcar (lambda (lang)
1149 (list 'const (car lang)))
1150 language-info-alist))
8861c593
RS
1151 :initialize 'custom-initialize-default
1152 :group 'mule
1153 :type 'string)
f850d782 1154
ddb5c041
KH
1155(defun reset-language-environment ()
1156 "Reset multilingual environment of Emacs to the default status.
1157
1158The default status is as follows:
1159
1160 The default value of buffer-file-coding-system is nil.
1161 The default coding system for process I/O is nil.
1162 The default value for the command `set-terminal-coding-system' is nil.
1163 The default value for the command `set-keyboard-coding-system' is nil.
1164
1165 The order of priorities of coding categories and the coding system
1166 bound to each category are as follows
1167 coding category coding system
1168 --------------------------------------------------
1169 coding-category-iso-8-2 iso-latin-1
1170 coding-category-iso-8-1 iso-latin-1
1171 coding-category-iso-7-tight iso-2022-jp
1172 coding-category-iso-7 iso-2022-7bit
1173 coding-category-iso-7-else iso-2022-7bit-lock
1174 coding-category-iso-8-else iso-2022-8bit-ss2
1175 coding-category-emacs-mule emacs-mule
1176 coding-category-raw-text raw-text
1177 coding-category-sjis japanese-shift-jis
1178 coding-category-big5 chinese-big5
1179 coding-category-ccl nil
e8dd0160 1180 coding-category-binary no-conversion
ddb5c041
KH
1181"
1182 (interactive)
1183 ;; This function formerly set default-enable-multibyte-characters to t,
1184 ;; but that is incorrect. It should not alter the unibyte/multibyte choice.
1185
1186 (setq coding-category-iso-7-tight 'iso-2022-jp
1187 coding-category-iso-7 'iso-2022-7bit
1188 coding-category-iso-8-1 'iso-latin-1
1189 coding-category-iso-8-2 'iso-latin-1
1190 coding-category-iso-7-else 'iso-2022-7bit-lock
1191 coding-category-iso-8-else 'iso-2022-8bit-ss2
1192 coding-category-emacs-mule 'emacs-mule
1193 coding-category-raw-text 'raw-text
1194 coding-category-sjis 'japanese-shift-jis
1195 coding-category-big5 'chinese-big5
9bfcd269
KH
1196 coding-category-utf-8 nil
1197 coding-category-utf-16-be nil
1198 coding-category-utf-16-le nil
ddb5c041
KH
1199 coding-category-ccl nil
1200 coding-category-binary 'no-conversion)
1201
1202 (set-coding-priority
1203 '(coding-category-iso-8-1
1204 coding-category-iso-8-2
1205 coding-category-iso-7-tight
1206 coding-category-iso-7
1207 coding-category-iso-7-else
1208 coding-category-iso-8-else
1209 coding-category-emacs-mule
1210 coding-category-raw-text
1211 coding-category-sjis
1212 coding-category-big5
1213 coding-category-ccl
9bfcd269
KH
1214 coding-category-binary
1215 coding-category-utf-16-be
1216 coding-category-utf-16-le
1217 coding-category-utf-8))
ddb5c041 1218
91693d18
KH
1219 (update-coding-systems-internal)
1220
ddb5c041
KH
1221 (set-default-coding-systems nil)
1222 ;; Don't alter the terminal and keyboard coding systems here.
1223 ;; The terminal still supports the same coding system
1224 ;; that it supported a minute ago.
1225;;; (set-terminal-coding-system-internal nil)
1226;;; (set-keyboard-coding-system-internal nil)
1227
1228 (setq nonascii-translation-table nil
1229 nonascii-insert-offset 0))
1230
40c81f74
PE
1231(defun set-display-table-and-terminal-coding-system (language-name)
1232 "Set up the display table and terminal coding system for LANGUAGE-NAME."
1233 (let ((coding (get-language-info language-name 'unibyte-display)))
1234 (if coding
1235 (standard-display-european-internal)
1236 (standard-display-default (if (eq window-system 'pc) 128 160) 255)
1237 (aset standard-display-table 146 nil))
1238 (or (eq window-system 'pc)
1239 (set-terminal-coding-system coding))))
1240
166246f7 1241(defun set-language-environment (language-name)
6c05d680
RS
1242 "Set up multi-lingual environment for using LANGUAGE-NAME.
1243This sets the coding system priority and the default input method
8861c593
RS
1244and sometimes other things. LANGUAGE-NAME should be a string
1245which is the name of a language environment. For example, \"Latin-1\"
1246specifies the character set for the major languages of Western Europe."
8efc03e1 1247 (interactive (list (read-language-name
ddb5c041 1248 nil
8efc03e1 1249 "Set language environment (default, English): ")))
4ef06f75
KH
1250 (if language-name
1251 (if (symbolp language-name)
1252 (setq language-name (symbol-name language-name)))
1253 (setq language-name "English"))
ddb5c041 1254 (or (assoc-ignore-case language-name language-info-alist)
f850d782 1255 (error "Language environment not defined: %S" language-name))
8efc03e1
KH
1256 (if current-language-environment
1257 (let ((func (get-language-info current-language-environment
1258 'exit-function)))
e63645c2
KH
1259 (run-hooks 'exit-language-environment-hook)
1260 (if (fboundp func) (funcall func))))
03c35c83
EZ
1261 (let ((default-eol-type (coding-system-eol-type
1262 default-buffer-file-coding-system)))
1263 (reset-language-environment)
ddb5c041 1264
03c35c83
EZ
1265 (setq current-language-environment language-name)
1266 (set-language-environment-coding-systems language-name default-eol-type))
ddb5c041
KH
1267 (let ((input-method (get-language-info language-name 'input-method)))
1268 (when input-method
1269 (setq default-input-method input-method)
1270 (if input-method-history
1271 (setq input-method-history
1272 (cons input-method
1273 (delete input-method input-method-history))))))
ec241f58
EZ
1274 (let ((nonascii (get-language-info language-name 'nonascii-translation))
1275 (dos-table
4e2ac2d9
EZ
1276 (if (eq window-system 'pc)
1277 (intern
1278 (concat "cp" dos-codepage "-nonascii-translation-table")))))
03c35c83
EZ
1279 (cond
1280 ((char-table-p nonascii)
1281 (setq nonascii-translation-table nonascii))
ec241f58 1282 ((and (eq window-system 'pc) (boundp dos-table))
03c35c83
EZ
1283 ;; DOS terminals' default is to use a special non-ASCII translation
1284 ;; table as appropriate for the installed codepage.
ec241f58 1285 (setq nonascii-translation-table (symbol-value dos-table)))
03c35c83
EZ
1286 ((charsetp nonascii)
1287 (setq nonascii-insert-offset (- (make-char nonascii) 128)))))
ddb5c041 1288
45d08cb2
KH
1289 (setq charset-origin-alist
1290 (get-language-info language-name 'charset-origin-alist))
ddb5c041 1291
63283a8f 1292 ;; Unibyte setups if necessary.
ddb5c041 1293 (unless default-enable-multibyte-characters
63283a8f 1294 ;; Syntax and case table.
ddb5c041
KH
1295 (let ((syntax (get-language-info language-name 'unibyte-syntax)))
1296 (if syntax
1297 (let ((set-case-syntax-set-multibyte nil))
63283a8f
KH
1298 (load syntax nil t))
1299 ;; No information for syntax and case. Reset to the defaults.
1300 (let ((syntax-table (standard-syntax-table))
1301 (case-table (standard-case-table))
03c35c83 1302 (ch (if (eq window-system 'pc) 128 160)))
63283a8f
KH
1303 (while (< ch 256)
1304 (modify-syntax-entry ch " " syntax-table)
1305 (aset case-table ch ch)
1306 (setq ch (1+ ch)))
1307 (set-char-table-extra-slot case-table 0 nil)
1308 (set-char-table-extra-slot case-table 1 nil)
1309 (set-char-table-extra-slot case-table 2 nil))
1310 (set-standard-case-table (standard-case-table))
1311 (let ((list (buffer-list)))
1312 (while list
1313 (with-current-buffer (car list)
1314 (set-case-table (standard-case-table)))
1315 (setq list (cdr list))))))
40c81f74 1316 (set-display-table-and-terminal-coding-system language-name))
ddb5c041
KH
1317
1318 (let ((required-features (get-language-info language-name 'features)))
1319 (while required-features
1320 (require (car required-features))
1321 (setq required-features (cdr required-features))))
1322 (let ((func (get-language-info language-name 'setup-function)))
1323 (if (fboundp func)
1324 (funcall func)))
8efc03e1 1325 (run-hooks 'set-language-environment-hook)
f850d782 1326 (force-mode-line-update t))
4ed46869 1327
51a8fc1d
RS
1328(defun standard-display-european-internal ()
1329 ;; Actually set up direct output of non-ASCII characters.
03c35c83
EZ
1330 (standard-display-8bit (if (eq window-system 'pc) 128 160) 255)
1331 ;; Unibyte Emacs on MS-DOS wants to display all 8-bit characters with
1332 ;; the native font, and codes 160 and 146 stand for something very
1333 ;; different there.
1334 (or (and (eq window-system 'pc) (not default-enable-multibyte-characters))
1335 (progn
1336 ;; Make non-line-break space display as a plain space.
1337 ;; Most X fonts do the wrong thing for code 160.
1338 (aset standard-display-table 160 [32])
1339 ;; Most Windows programs send out apostrophe's as \222. Most X fonts
1340 ;; don't contain a character at that position. Map it to the ASCII
1341 ;; apostrophe.
1342 (aset standard-display-table 146 [39]))))
1343
1344(defun set-language-environment-coding-systems (language-name
1345 &optional eol-type)
1346 "Do various coding system setups for language environment LANGUAGE-NAME.
1347
1348The optional arg EOL-TYPE specifies the eol-type of the default value
1349of buffer-file-coding-system set by this function."
54b226f7
KH
1350 (let* ((priority (get-language-info language-name 'coding-priority))
1351 (default-coding (car priority)))
1352 (if priority
1353 (let ((categories (mapcar 'coding-system-category priority)))
03c35c83
EZ
1354 (set-default-coding-systems
1355 (if (memq eol-type '(0 1 2 unix dos mac))
1356 (coding-system-change-eol-conversion default-coding eol-type)
1357 default-coding))
46babb23 1358 (setq default-sendmail-coding-system default-coding)
54b226f7
KH
1359 (set-coding-priority categories)
1360 (while priority
1361 (set (car categories) (car priority))
1362 (setq priority (cdr priority) categories (cdr categories)))
45d08cb2 1363 (update-coding-systems-internal)))))
54b226f7 1364
4ed46869
KH
1365;; Print all arguments with `princ', then print "\n".
1366(defsubst princ-list (&rest args)
1367 (while args (princ (car args)) (setq args (cdr args)))
1368 (princ "\n"))
1369
b0648a00
RS
1370(put 'describe-specified-language-support 'apropos-inhibit t)
1371
48082651 1372;; Print a language specific information such as input methods,
13e82c04 1373;; charsets, and coding systems. This function is intended to be
48082651 1374;; called from the menu:
281d03ec 1375;; [menu-bar mule describe-language-environment LANGUAGE]
48082651
KH
1376;; and should not run it by `M-x describe-current-input-method-function'.
1377(defun describe-specified-language-support ()
96db204a 1378 "Describe how Emacs supports the specified language environment."
48082651 1379 (interactive)
281d03ec 1380 (let (language-name)
48082651 1381 (if (not (and (symbolp last-command-event)
281d03ec 1382 (setq language-name (symbol-name last-command-event))))
48082651 1383 (error "Bogus calling sequence"))
281d03ec
RS
1384 (describe-language-environment language-name)))
1385
1386(defun describe-language-environment (language-name)
1387 "Describe how Emacs supports language environment LANGUAGE-NAME."
78754934
KH
1388 (interactive
1389 (list (read-language-name
1390 'documentation
8adfa8be 1391 "Describe language environment (default, current choice): ")))
f850d782
RS
1392 (if (null language-name)
1393 (setq language-name current-language-environment))
281d03ec
RS
1394 (if (or (null language-name)
1395 (null (get-language-info language-name 'documentation)))
1396 (error "No documentation for the specified language"))
4ef06f75
KH
1397 (if (symbolp language-name)
1398 (setq language-name (symbol-name language-name)))
281d03ec 1399 (let ((doc (get-language-info language-name 'documentation)))
48082651 1400 (with-output-to-temp-buffer "*Help*"
a33c9d6f 1401 (princ-list language-name " language environment" "\n")
13e82c04 1402 (if (stringp doc)
d0b9c3ab
KH
1403 (progn
1404 (princ-list doc)
1405 (terpri)))
15b3e511
KH
1406 (let ((str (get-language-info language-name 'sample-text)))
1407 (if (stringp str)
1408 (progn
281d03ec 1409 (princ "Sample text:\n")
d0b9c3ab
KH
1410 (princ-list " " str)
1411 (terpri))))
ddb5c041 1412 (let ((input-method (get-language-info language-name 'input-method))
00a069c6 1413 (l (copy-sequence input-method-alist)))
ddb5c041
KH
1414 (princ "Input methods")
1415 (when input-method
1416 (princ (format " (default, %s)" input-method))
1417 (setq input-method (assoc input-method input-method-alist))
1418 (setq l (cons input-method (delete input-method l))))
1419 (princ ":\n")
15b3e511 1420 (while l
d0b9c3ab
KH
1421 (if (string= language-name (nth 1 (car l)))
1422 (princ-list " " (car (car l))
1423 (format " (`%s' in mode line)" (nth 3 (car l)))))
15b3e511 1424 (setq l (cdr l))))
281d03ec
RS
1425 (terpri)
1426 (princ "Character sets:\n")
15b3e511
KH
1427 (let ((l (get-language-info language-name 'charset)))
1428 (if (null l)
1429 (princ-list " nothing specific to " language-name)
1430 (while l
1431 (princ-list " " (car l) ": "
1432 (charset-description (car l)))
1433 (setq l (cdr l)))))
281d03ec
RS
1434 (terpri)
1435 (princ "Coding systems:\n")
15b3e511
KH
1436 (let ((l (get-language-info language-name 'coding-system)))
1437 (if (null l)
1438 (princ-list " nothing specific to " language-name)
48082651 1439 (while l
281d03ec
RS
1440 (princ (format " %s (`%c' in mode line):\n\t%s\n"
1441 (car l)
1442 (coding-system-mnemonic (car l))
a904b20b 1443 (coding-system-doc-string (car l))))
8efc03e1
KH
1444 (let ((aliases (coding-system-get (car l) 'alias-coding-systems)))
1445 (when aliases
1446 (princ "\t")
1447 (princ (cons 'alias: (cdr aliases)))
1448 (terpri)))
15b3e511 1449 (setq l (cdr l))))))))
4ed46869 1450\f
40c81f74
PE
1451;;; Locales.
1452
1453(defvar locale-translation-file-name
1454 (let ((files '("/usr/lib/X11/locale/locale.alias" ; e.g. X11R6.4
1455 "/usr/X11R6/lib/X11/locale/locale.alias" ; e.g. RedHat 4.2
1456 "/usr/openwin/lib/locale/locale.alias" ; e.g. Solaris 2.6
1457 ;;
1458 ;; The following name appears after the X-related names above,
1459 ;; since the X-related names are what X actually uses.
1460 "/usr/share/locale/locale.alias" ; GNU/Linux sans X
1461 )))
1462 (while (and files (not (file-exists-p (car files))))
1463 (setq files (cdr files)))
1464 (car files))
1465 "*File name for the system's file of locale-name aliases, or nil if none.")
1466
1467(defvar locale-language-names
1468 '(
1469 ;; UTF-8 is not yet implemented.
1470 ;; Put this first, so that e.g. "ko.UTF-8" does not match "ko" below.
1471 (".*[._]utf" . nil)
1472
1473 ;; Locale names of the form LANGUAGE[_TERRITORY][.CODESET][@MODIFIER]
1474 ;; as specified in the Single Unix Spec, Version 2.
1475 ;; LANGUAGE is a language code taken from ISO 639:1988 (E/F)
1476 ;; with additions from ISO 639/RA Newsletter No.1/1989;
1477 ;; see Internet RFC 2165 (1997-06).
1478 ;; TERRITORY is a country code taken from ISO 3166.
1479 ;; CODESET and MODIFIER are implementation-dependent.
1480 ;;
1481 ; aa Afar
1482 ; ab Abkhazian
6ececc4d 1483 ("af" . "Latin-1") ; Afrikaans
40c81f74
PE
1484 ("am" . "Ethiopic") ; Amharic
1485 ; ar Arabic
1486 ; as Assamese
1487 ; ay Aymara
1488 ; az Azerbaijani
1489 ; ba Bashkir
6ececc4d
PE
1490 ("be" . "Latin-5") ; Byelorussian
1491 ("bg" . "Latin-5") ; Bulgarian
40c81f74
PE
1492 ; bh Bihari
1493 ; bi Bislama
1494 ; bn Bengali, Bangla
1495 ("bo" . "Tibetan")
1496 ("br" . "Latin-1") ; Breton
1497 ("ca" . "Latin-1") ; Catalan
1498 ; co Corsican
1499 ("cs" . "Czech")
6ececc4d 1500 ("cy" . "Latin-8") ; Welsh
40c81f74
PE
1501 ("da" . "Latin-1") ; Danish
1502 ("de" . "German")
1503 ; dz Bhutani
1504 ("el" . "Greek")
6ececc4d
PE
1505 ;; Users who specify "en" explicitly typically want Latin-1, not ASCII.
1506 ("en" . "Latin-1") ; English
40c81f74
PE
1507 ("eo" . "Latin-3") ; Esperanto
1508 ("es" . "Latin-1") ; Spanish
1509 ("et" . "Latin-4") ; Estonian
1510 ("eu" . "Latin-1") ; Basque
1511 ; fa Persian
1512 ("fi" . "Latin-1") ; Finnish
1513 ; fj Fiji
1514 ("fo" . "Latin-1") ; Faroese
1515 ("fr" . "Latin-1") ; French
1516 ("fy" . "Latin-1") ; Frisian
6ececc4d
PE
1517 ("ga" . "Latin-1") ; Irish Gaelic (new orthography)
1518 ("gd" . "Latin-1") ; Scots Gaelic
1519 ("gl" . "Latin-1") ; Galician
40c81f74
PE
1520 ; gn Guarani
1521 ; gu Gujarati
1522 ; ha Hausa
1523 ("he" . "Hebrew")
1524 ("hi" . "Devanagari") ; Hindi
1525 ("hr" . "Latin-2") ; Croatian
1526 ("hu" . "Latin-2") ; Hungarian
1527 ; hy Armenian
1528 ; ia Interlingua
1529 ("id" . "Latin-1") ; Indonesian
1530 ; ie Interlingue
1531 ; ik Inupiak
1532 ("is" . "Latin-1") ; Icelandic
1533 ("it" . "Latin-1") ; Italian
1534 ; iu Inuktitut
1535 ("ja" . "Japanese")
1536 ; jw Javanese
1537 ; ka Georgian
1538 ; kk Kazakh
6ececc4d 1539 ("kl" . "Latin-1") ; Greenlandic
40c81f74
PE
1540 ; km Cambodian
1541 ; kn Kannada
1542 ("ko" . "Korean")
1543 ; ks Kashmiri
1544 ; ku Kurdish
1545 ; ky Kirghiz
1546 ("la" . "Latin-1") ; Latin
1547 ; ln Lingala
1548 ("lo" . "Lao") ; Laothian
1549 ("lt" . "Latin-4") ; Lithuanian
1550 ("lv" . "Latin-4") ; Latvian, Lettish
1551 ; mg Malagasy
1552 ; mi Maori
6ececc4d 1553 ("mk" . "Latin-5") ; Macedonian
40c81f74
PE
1554 ; ml Malayalam
1555 ; mn Mongolian
1556 ; mo Moldavian
1557 ("mr" . "Devanagari") ; Marathi
1558 ; ms Malay
1559 ("mt" . "Latin-3") ; Maltese
1560 ; my Burmese
1561 ; na Nauru
1562 ("ne" . "Devanagari") ; Nepali
1563 ("nl" . "Latin-1") ; Dutch
1564 ("no" . "Latin-1") ; Norwegian
1565 ; oc Occitan
1566 ; om (Afan) Oromo
1567 ; or Oriya
1568 ; pa Punjabi
1569 ("pl" . "Latin-2") ; Polish
1570 ; ps Pashto, Pushto
1571 ("pt" . "Latin-1") ; Portuguese
1572 ; qu Quechua
6ececc4d 1573 ("rm" . "Latin-1") ; Rhaeto-Romanic
40c81f74
PE
1574 ; rn Kirundi
1575 ("ro" . "Romanian")
1576 ("ru.*[_.]koi8" . "Cyrillic-KOI8") ; Russian
6ececc4d 1577 ("ru" . "Latin-5") ; Russian
40c81f74
PE
1578 ; rw Kinyarwanda
1579 ("sa" . "Devanagari") ; Sanskrit
1580 ; sd Sindhi
1581 ; sg Sangho
1582 ("sh" . "Latin-2") ; Serbo-Croatian
1583 ; si Sinhalese
1584 ("sk" . "Slovak")
1585 ("sl" . "Slovenian")
1586 ; sm Samoan
1587 ; sn Shona
1588 ; so Somali
6ececc4d 1589 ("sq" . "Latin-1") ; Albanian
40c81f74
PE
1590 ("sr" . "Latin-2") ; Serbian (Latin alphabet)
1591 ; ss Siswati
1592 ; st Sesotho
1593 ; su Sundanese
1594 ("sv" . "Latin-1") ; Swedish
1595 ("sw" . "Latin-1") ; Swahili
1596 ; ta Tamil
1597 ; te Telugu
1598 ; tg Tajik
1599 ("th" . "Thai")
1600 ; ti Tigrinya
1601 ; tk Turkmen
6ececc4d 1602 ("tl" . "Latin-1") ; Tagalog
40c81f74
PE
1603 ; tn Setswana
1604 ; to Tonga
1605 ("tr" . "Latin-5") ; Turkish
1606 ; ts Tsonga
1607 ; tt Tatar
1608 ; tw Twi
1609 ; ug Uighur
6ececc4d 1610 ("uk" . "Latin-5") ; Ukrainian
40c81f74
PE
1611 ; ur Urdu
1612 ; uz Uzbek
1613 ("vi" . "Vietnamese")
1614 ; vo Volapuk
1615 ; wo Wolof
1616 ; xh Xhosa
1617 ; yi Yiddish
1618 ; yo Yoruba
1619 ; za Zhuang
1620 ("zh.*[._]big5" . "Chinese-BIG5")
1621 ("zh.*[._]gbk" . nil) ; Solaris 2.7; has gbk-0 as well as GB 2312.1980-0
1622 ("zh_tw" . "Chinese-CNS")
1623 ("zh" . "Chinese-GB")
1624 ; zu Zulu
1625
1626 ;; ISO standard locales
1627 ("c$" . "ASCII")
1628 ("posix$" . "ASCII")
1629
40c81f74
PE
1630 ;; The "IPA" Emacs language environment does not correspond
1631 ;; to any ISO 639 code, so let it stand for itself.
1632 ("ipa$" . "IPA")
1633
1634 ;; Nonstandard or obsolete language codes
1635 ("cz" . "Czech") ; e.g. Solaris 2.6
1636 ("ee" . "Latin-4") ; Estonian, e.g. X11R6.4
1637 ("iw" . "Hebrew") ; e.g. X11R6.4
6ececc4d 1638 ("sp" . "Latin-5") ; Serbian (Cyrillic alphabet), e.g. X11R6.4
40c81f74
PE
1639 ("su" . "Latin-1") ; Finnish, e.g. Solaris 2.6
1640 )
1641 "List of pairs of locale regexps and language names.
6ececc4d
PE
1642The first element whose locale regexp matches the start of a downcased locale
1643specifies the language name corresponding to that locale.
40c81f74
PE
1644If the language name is nil, there is no corresponding language environment.")
1645
6ececc4d
PE
1646(defvar locale-charset-language-names
1647 '((".*8859[-_]?1\\>" . "Latin-1")
1648 (".*8859[-_]?2\\>" . "Latin-2")
1649 (".*8859[-_]?3\\>" . "Latin-3")
1650 (".*8859[-_]?4\\>" . "Latin-4")
1651 (".*8859[-_]?9\\>" . "Latin-5")
1652 (".*8859[-_]?14\\>" . "Latin-8")
1653 (".*8859[-_]?15\\>" . "Latin-9")
1654 )
1655 "List of pairs of locale regexps and charset language names.
1656The first element whose locale regexp matches the start of a downcased locale
1657specifies the language name whose charsets corresponds to that locale.
1658This language name is used if its charsets disagree with the charsets of
1659the language name that would otherwise be used for this locale.")
1660
40c81f74
PE
1661(defvar locale-preferred-coding-systems
1662 '(("ja.*[._]euc" . japanese-iso-8bit)
1663 ("ja.*[._]jis7" . iso-2022-jp)
1664 ("ja.*[._]pck" . japanese-shift-jis)
1665 ("ja.*[._]sjis" . japanese-shift-jis)
40c81f74 1666 )
6ececc4d
PE
1667 "List of pairs of locale regexps and preferred coding systems.
1668The first element whose locale regexp matches the start of a downcased locale
1669specifies the coding system to prefer when using that locale.")
40c81f74
PE
1670
1671(defun locale-name-match (key alist)
1672 "Search for KEY in ALIST, which should be a list of regexp-value pairs.
1673Return the value corresponding to the first regexp that matches the
1674start of KEY, or nil if there is no match."
1675 (let (element)
1676 (while (and alist (not element))
1677 (if (string-match (concat "^\\(" (car (car alist)) "\\)") key)
1678 (setq element (car alist)))
1679 (setq alist (cdr alist)))
1680 (cdr element)))
1681
1682(defun set-locale-environment (locale-name)
1683 "Set up multi-lingual environment for using LOCALE-NAME.
1684This sets the coding system priority and the default input method
1685and sometimes other things. LOCALE-NAME should be a string
1686which is the name of a locale supported by the system;
1687often it is of the form xx_XX.CODE, where xx is a language,
1688XX is a country, and CODE specifies a character set and coding system.
1689For example, the locale name \"ja_JP.EUC\" might name a locale
1690for Japanese in Japan using the `japanese-iso-8bit' coding-system.
1691
1692If LOCALE-NAME is nil, its value is taken from the environment.
1693
1694The locale names supported by your system can typically be found in a
1695directory named `/usr/share/locale' or `/usr/lib/locale'."
1696
1697 (unless locale-name
1698 ;; Use the first of these three environment variables
1699 ;; that has a nonempty value.
1700 (let ((vars '("LC_ALL" "LC_CTYPE" "LANG")))
1701 (while (and vars (not (setq locale-name (getenv (car vars)))))
1702 (setq vars (cdr vars)))))
1703
1704 (when locale-name
1705
1706 ;; Translate "swedish" into "sv_SE.ISO8859-1", and so on,
1707 ;; using the translation file that many systems have.
1708 (when locale-translation-file-name
1709 (with-temp-buffer
1710 (insert-file-contents locale-translation-file-name)
1711 (when (re-search-forward
1712 (concat "^" (regexp-quote locale-name) ":?[ \t]+") nil t)
1713 (setq locale-name (buffer-substring (point) (line-end-position))))))
1714
1715 (setq locale-name (downcase locale-name))
1716
6ececc4d
PE
1717 (let ((language-name
1718 (locale-name-match locale-name locale-language-names))
1719 (charset-language-name
1720 (locale-name-match locale-name locale-charset-language-names))
1721 (coding-system
1722 (locale-name-match locale-name locale-preferred-coding-systems)))
1723
1724 (if (and charset-language-name
1725 (not
1726 (equal (get-language-info language-name 'charset)
1727 (get-language-info charset-language-name 'charset))))
1728 (setq language-name charset-language-name))
1729
40c81f74
PE
1730 (when language-name
1731
1732 ;; Set up for this character set. This is now the right way
1733 ;; to do it for both unibyte and multibyte modes.
1734 (set-language-environment language-name)
1735
1736 ;; If default-enable-multibyte-characters is nil,
1737 ;; we are using single-byte characters,
1738 ;; so the display table and terminal coding system are irrelevant.
1739 (when default-enable-multibyte-characters
1740 (set-display-table-and-terminal-coding-system language-name))
1741
1742 (setq locale-coding-system
1743 (car (get-language-info language-name 'coding-priority))))
1744
1745 (when coding-system
1746 (prefer-coding-system coding-system)
1747 (setq locale-coding-system coding-system)))))
1748\f
4ed46869
KH
1749;;; Charset property
1750
251d4f4b 1751(defun get-charset-property (charset propname)
4ed46869
KH
1752 "Return the value of CHARSET's PROPNAME property.
1753This is the last value stored with
96db204a 1754 (put-charset-property CHARSET PROPNAME VALUE)."
251d4f4b
KH
1755 (and (not (eq charset 'composition))
1756 (plist-get (charset-plist charset) propname)))
4ed46869 1757
251d4f4b 1758(defun put-charset-property (charset propname value)
4ed46869
KH
1759 "Store CHARSETS's PROPNAME property with value VALUE.
1760It can be retrieved with `(get-charset-property CHARSET PROPNAME)'."
1300d43f
KH
1761 (or (eq charset 'composition)
1762 (set-charset-plist charset
1763 (plist-put (charset-plist charset) propname value))))
4ed46869
KH
1764
1765;;; Character code property
1766(put 'char-code-property-table 'char-table-extra-slots 0)
1767
1768(defvar char-code-property-table
1769 (make-char-table 'char-code-property-table)
1770 "Char-table containing a property list of each character code.
1771
1772See also the documentation of `get-char-code-property' and
96db204a 1773`put-char-code-property'.")
4ed46869
KH
1774
1775(defun get-char-code-property (char propname)
1776 "Return the value of CHAR's PROPNAME property in `char-code-property-table'."
1777 (let ((plist (aref char-code-property-table char)))
1778 (if (listp plist)
1779 (car (cdr (memq propname plist))))))
1780
1781(defun put-char-code-property (char propname value)
1782 "Store CHAR's PROPNAME property with VALUE in `char-code-property-table'.
1783It can be retrieved with `(get-char-code-property CHAR PROPNAME)'."
1784 (let ((plist (aref char-code-property-table char)))
1785 (if plist
1786 (let ((slot (memq propname plist)))
1787 (if slot
1788 (setcar (cdr slot) value)
1789 (nconc plist (list propname value))))
1790 (aset char-code-property-table char (list propname value)))))
1791
a127b764
KH
1792\f
1793;; Pretty description of encoded string
1794
1795;; Alist of ISO 2022 control code vs the corresponding mnemonic string.
1796(defvar iso-2022-control-alist
1797 '((?\x1b . "ESC")
1798 (?\x0e . "SO")
1799 (?\x0f . "SI")
1800 (?\x8e . "SS2")
1801 (?\x8f . "SS3")
1802 (?\x9b . "CSI")))
1803
1804(defun encoded-string-description (str coding-system)
1805 "Return a pretty description of STR that is encoded by CODING-SYSTEM."
1806 (setq str (string-as-unibyte str))
1807 (let ((char (aref str 0))
1808 desc)
1809 (when (< char 128)
1810 (setq desc (or (cdr (assq char iso-2022-control-alist))
1811 (char-to-string char)))
1812 (let ((i 1)
1813 (len (length str)))
1814 (while (< i len)
1815 (setq char (aref str i))
1816 (if (>= char 128)
1817 (setq desc nil i len)
1818 (setq desc (concat desc " "
1819 (or (cdr (assq char iso-2022-control-alist))
1820 (char-to-string char)))
1821 i (1+ i))))))
1822 (or desc
1823 (mapconcat (function (lambda (x) (format "0x%02x" x))) str " "))))
1824
1825(defun encode-coding-char (char coding-system)
1826 "Encode CHAR by CODING-SYSTEM and return the resulting string.
1827If CODING-SYSTEM can't safely encode CHAR, return nil."
a127b764
KH
1828 (let ((str1 (char-to-string char))
1829 (str2 (make-string 2 char))
1830 (safe-charsets (and coding-system
1831 (coding-system-get coding-system 'safe-charsets)))
1832 enc1 enc2 i1 i2)
1833 (when (or (eq safe-charsets t)
1834 (memq (char-charset char) safe-charsets))
1835 ;; We must find the encoded string of CHAR. But, just encoding
1836 ;; CHAR will put extra control sequences (usually to designate
1837 ;; ASCII charaset) at the tail if type of CODING is ISO 2022.
1838 ;; To exclude such tailing bytes, we at first encode one-char
1839 ;; string and two-char string, then check how many bytes at the
1840 ;; tail of both encoded strings are the same.
1841
1842 (setq enc1 (string-as-unibyte (encode-coding-string str1 coding-system))
1843 i1 (length enc1)
1844 enc2 (string-as-unibyte (encode-coding-string str2 coding-system))
1845 i2 (length enc2))
1846 (while (and (> i1 0) (= (aref enc1 (1- i1)) (aref enc2 (1- i2))))
1847 (setq i1 (1- i1) i2 (1- i2)))
1848
1849 ;; Now (substring enc1 i1) and (substring enc2 i2) are the same,
1850 ;; and they are the extra control sequences at the tail to
1851 ;; exclude.
1852 (substring enc2 0 i2))))
1853
1854
4ed46869 1855;;; mule-cmds.el ends here