Some fixes to follow coding conventions.
[bpt/emacs.git] / lisp / international / mule-cmds.el
CommitLineData
60370d40 1;;; mule-cmds.el --- commands for mulitilingual environment
4ed46869 2
4ed46869 3;; Copyright (C) 1995 Electrotechnical Laboratory, JAPAN.
fa526c4a 4;; Licensed to the Free Software Foundation.
7008ccac 5;; Copyright (C) 2000, 2001 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 25
60370d40
PJ
26;;; Commentary:
27
4ed46869
KH
28;;; Code:
29
cda74479
DL
30(eval-when-compile (defvar dos-codepage))
31
4ed46869
KH
32;;; MULE related key bindings and menus.
33
0709d285 34(defvar mule-keymap (make-sparse-keymap)
33d17698 35 "Keymap for Mule (Multilingual environment) specific commands.")
4ed46869 36
8f81f784 37;; Keep "C-x C-m ..." for mule specific commands.
0709d285 38(define-key ctl-x-map "\C-m" mule-keymap)
ef8a8c8c 39
4ed46869
KH
40(define-key mule-keymap "f" 'set-buffer-file-coding-system)
41(define-key mule-keymap "t" 'set-terminal-coding-system)
15b3e511
KH
42(define-key mule-keymap "k" 'set-keyboard-coding-system)
43(define-key mule-keymap "p" 'set-buffer-process-coding-system)
7624ebb9
KH
44(define-key mule-keymap "x" 'set-selection-coding-system)
45(define-key mule-keymap "X" 'set-next-selection-coding-system)
8b784951 46(define-key mule-keymap "\C-\\" 'set-input-method)
15b3e511 47(define-key mule-keymap "c" 'universal-coding-system-argument)
b4fba33f 48(define-key mule-keymap "l" 'set-language-environment)
4ed46869 49
281d03ec 50(define-key help-map "\C-L" 'describe-language-environment)
ac4a3a2d 51(define-key help-map "L" 'describe-language-environment)
4ed46869 52(define-key help-map "\C-\\" 'describe-input-method)
ac4a3a2d 53(define-key help-map "I" 'describe-input-method)
d0b9c3ab 54(define-key help-map "C" 'describe-coding-system)
4ed46869
KH
55(define-key help-map "h" 'view-hello-file)
56
538d88fb
EZ
57(defvar mule-menu-keymap
58 (make-sparse-keymap "Mule (Multilingual Environment)")
33d17698 59 "Keymap for Mule (Multilingual environment) menu specific commands.")
15b3e511 60
dcad02bc
EZ
61(defvar describe-language-environment-map
62 (make-sparse-keymap "Describe Language Environment"))
15b3e511 63
dcad02bc
EZ
64(defvar setup-language-environment-map
65 (make-sparse-keymap "Set Language Environment"))
15b3e511 66
dcad02bc
EZ
67(defvar set-coding-system-map
68 (make-sparse-keymap "Set Coding System"))
15b3e511 69
15b3e511 70(define-key-after mule-menu-keymap [set-language-environment]
cda74479
DL
71 (list 'menu-item "Set Language Environment" setup-language-environment-map
72 :help "Multilingual environment suitable for a specific language"))
a61f401d 73(define-key-after mule-menu-keymap [mouse-set-font]
538d88fb 74 '(menu-item "Set Font/Fontset" mouse-set-font
cda74479
DL
75 :visible (fboundp 'generate-fontset-menu)
76 :help "Select a font from list of known fonts/fontsets"))
15b3e511
KH
77(define-key-after mule-menu-keymap [separator-mule]
78 '("--")
79 t)
80(define-key-after mule-menu-keymap [toggle-input-method]
538d88fb 81 '(menu-item "Toggle Input Method" toggle-input-method)
15b3e511 82 t)
8b784951 83(define-key-after mule-menu-keymap [set-input-method]
538d88fb 84 '(menu-item "Select Input Method..." set-input-method)
15b3e511 85 t)
cda74479
DL
86(define-key-after mule-menu-keymap [describe-input-method]
87 '(menu-item "Describe Input Method" describe-input-method))
15b3e511
KH
88(define-key-after mule-menu-keymap [separator-input-method]
89 '("--")
90 t)
15b3e511 91(define-key-after mule-menu-keymap [set-various-coding-system]
cda74479
DL
92 (list 'menu-item "Set Coding Systems" set-coding-system-map
93 :enable 'enable-multibyte-characters))
538d88fb
EZ
94(define-key-after mule-menu-keymap [view-hello-file]
95 '(menu-item "Show Multi-lingual Text" view-hello-file
96 :enable (file-readable-p
97 (expand-file-name "HELLO" data-directory))
98 :help "Display file which says HELLO in many languages")
15b3e511
KH
99 t)
100(define-key-after mule-menu-keymap [separator-coding-system]
101 '("--")
102 t)
538d88fb
EZ
103(define-key-after mule-menu-keymap [describe-language-environment]
104 (list 'menu-item "Describe Language Environment"
105 describe-language-environment-map
cda74479 106 :help "Show multilingual settings for a specific language")
15b3e511 107 t)
538d88fb
EZ
108(define-key-after mule-menu-keymap [describe-input-method]
109 '(menu-item "Describe Input Method..." describe-input-method
cda74479 110 :help "Keyboard layout for a specific input method")
538d88fb
EZ
111 t)
112(define-key-after mule-menu-keymap [describe-coding-system]
113 '(menu-item "Describe Coding System..." describe-coding-system)
114 t)
cda74479
DL
115(define-key-after mule-menu-keymap [list-character-sets]
116 '(menu-item "List Character Sets" list-character-sets
117 :help "Show table of available character sets"))
538d88fb
EZ
118(define-key-after mule-menu-keymap [mule-diag]
119 '(menu-item "Show All of Mule Status" mule-diag
120 :help "Display multilingual environment settings")
15b3e511
KH
121 t)
122
123(define-key-after set-coding-system-map [set-buffer-file-coding-system]
538d88fb
EZ
124 '(menu-item "For Saving this Buffer" set-buffer-file-coding-system
125 :help "How to encode this buffer on disk")
15b3e511 126 t)
3a151e98 127(define-key-after set-coding-system-map [universal-coding-system-argument]
538d88fb
EZ
128 '(menu-item "For Next Command" universal-coding-system-argument
129 :help "Coding system to be used by next command")
3a151e98 130 t)
15b3e511 131(define-key-after set-coding-system-map [set-terminal-coding-system]
538d88fb
EZ
132 '(menu-item "For Terminal" set-terminal-coding-system
133 :enable (null (memq window-system '(x w32 mac)))
134 :help "How to encode terminal output")
15b3e511
KH
135 t)
136(define-key-after set-coding-system-map [set-keyboard-coding-system]
538d88fb
EZ
137 '(menu-item "For Keyboard" set-keyboard-coding-system
138 :help "How to decode keyboard input")
15b3e511
KH
139 t)
140(define-key-after set-coding-system-map [set-buffer-process-coding-system]
538d88fb
EZ
141 '(menu-item "For I/O with Subprocess" set-buffer-process-coding-system
142 :visible (fboundp 'start-process)
143 :enable (get-buffer-process (current-buffer))
144 :help "How to en/decode I/O from/to subprocess connected to this buffer")
15b3e511 145 t)
7624ebb9 146(define-key-after set-coding-system-map [set-selection-coding-system]
538d88fb
EZ
147 '(menu-item "For X Selections/Clipboard" set-selection-coding-system
148 :visible (display-selections-p)
149 :help "How to en/decode data to/from selection/clipboard")
7624ebb9
KH
150 t)
151(define-key-after set-coding-system-map [set-next-selection-coding-system]
538d88fb
EZ
152 '(menu-item "For Next X Selection" set-next-selection-coding-system
153 :visible (display-selections-p)
154 :help "How to en/decode next selection/clipboard operation")
7624ebb9 155 t)
15b3e511 156(define-key setup-language-environment-map
538d88fb 157 [Default] '(menu-item "Default" setup-specified-language-environment))
4ed46869 158
cda74479
DL
159(define-key describe-language-environment-map
160 [Default] '(menu-item "Default" describe-specified-language-support))
161
4ed46869
KH
162;; This should be a single character key binding because users use it
163;; very frequently while editing multilingual text. Now we can use
164;; only two such keys: "\C-\\" and "\C-^", but the latter is not
165;; convenient because it requires shifting on most keyboards. An
166;; alternative is "\C-\]" which is now bound to `abort-recursive-edit'
167;; but it won't be used that frequently.
168(define-key global-map "\C-\\" 'toggle-input-method)
169
a2ad45b9
RS
170;;; This is no good because people often type Shift-SPC
171;;; meaning to type SPC. -- rms.
172;;; ;; Here's an alternative key binding for X users (Shift-SPACE).
173;;; (define-key global-map [?\S- ] 'toggle-input-method)
b4fba33f 174
464cc130
KH
175;;; Mule related hyperlinks.
176(defconst help-xref-mule-regexp-template
177 (purecopy (concat "\\(\\<\\("
178 "\\(coding system\\)\\|"
d0c40faa
KH
179 "\\(input method\\)\\|"
180 "\\(character set\\)\\|"
181 "\\(charset\\)"
464cc130
KH
182 "\\)\\s-+\\)?"
183 ;; Note starting with word-syntax character:
184 "`\\(\\sw\\(\\sw\\|\\s_\\)+\\)'")))
185
26d87040
EZ
186(defun coding-system-change-eol-conversion (coding-system eol-type)
187 "Return a coding system which differs from CODING-SYSTEM in eol conversion.
188The returned coding system converts end-of-line by EOL-TYPE
189but text as the same way as CODING-SYSTEM.
190EOL-TYPE should be `unix', `dos', `mac', or nil.
191If EOL-TYPE is nil, the returned coding system detects
192how end-of-line is formatted automatically while decoding.
193
194EOL-TYPE can be specified by an integer 0, 1, or 2.
195They means `unix', `dos', and `mac' respectively."
196 (if (symbolp eol-type)
197 (setq eol-type (cond ((eq eol-type 'unix) 0)
198 ((eq eol-type 'dos) 1)
199 ((eq eol-type 'mac) 2)
200 (t eol-type))))
201 (let ((orig-eol-type (coding-system-eol-type coding-system)))
202 (if (vectorp orig-eol-type)
203 (if (not eol-type)
204 coding-system
205 (aref orig-eol-type eol-type))
206 (let ((base (coding-system-base coding-system)))
207 (if (not eol-type)
208 base
209 (if (= eol-type orig-eol-type)
210 coding-system
211 (setq orig-eol-type (coding-system-eol-type base))
212 (if (vectorp orig-eol-type)
213 (aref orig-eol-type eol-type))))))))
214
215(defun coding-system-change-text-conversion (coding-system coding)
216 "Return a coding system which differs from CODING-SYSTEM in text conversion.
217The returned coding system converts text by CODING
218but end-of-line as the same way as CODING-SYSTEM.
219If CODING is nil, the returned coding system detects
220how text is formatted automatically while decoding."
221 (if (not coding)
222 (coding-system-base coding-system)
223 (let ((eol-type (coding-system-eol-type coding-system)))
224 (coding-system-change-eol-conversion
225 coding
226 (if (numberp eol-type) (aref [unix dos mac] eol-type))))))
227
4ed46869 228(defun toggle-enable-multibyte-characters (&optional arg)
6998e1a1
RS
229 "Change whether this buffer uses multibyte characters.
230With arg, use multibyte characters if the arg is positive.
231
232Note that this command does not convert the byte contents of
233the buffer; it only changes the way those bytes are interpreted.
234In general, therefore, this command *changes* the sequence of
235characters that the current buffer contains.
236
237We suggest you avoid using use this command unless you know what you
238are doing. If you use it by mistake, and the buffer is now displayed
239wrong, use this command again to toggle back to the right mode."
4ed46869 240 (interactive "P")
b7079457
RS
241 (let ((new-flag
242 (if (null arg) (null enable-multibyte-characters)
243 (> (prefix-numeric-value arg) 0))))
244 (set-buffer-multibyte new-flag))
4ed46869
KH
245 (force-mode-line-update))
246
247(defun view-hello-file ()
248 "Display the HELLO file which list up many languages and characters."
249 (interactive)
8f81f784
KH
250 ;; We have to decode the file in any environment.
251 (let ((default-enable-multibyte-characters t)
95fa03b4 252 (coding-system-for-read 'iso-2022-7bit))
8f81f784 253 (find-file-read-only (expand-file-name "HELLO" data-directory))))
4ed46869 254
15b3e511
KH
255(defun universal-coding-system-argument ()
256 "Execute an I/O command using the specified coding system."
257 (interactive)
34104362
KH
258 (let* ((default (and buffer-file-coding-system
259 (not (eq (coding-system-type buffer-file-coding-system)
260 t))
261 buffer-file-coding-system))
262 (coding-system (read-coding-system
263 (if default
264 (format "Coding system for following command (default, %s): " default)
265 "Coding system for following command: ")
266 default))
15b3e511 267 (keyseq (read-key-sequence
e14a8f4c 268 (format "Command to execute with %s:" coding-system)))
04363179
GM
269 (cmd (key-binding keyseq))
270 prefix)
271
272 (when (eq cmd 'universal-argument)
273 (call-interactively cmd)
274
275 ;; Process keys bound in `universal-argument-map'.
276 (while (progn
277 (setq keyseq (read-key-sequence nil t)
278 cmd (key-binding keyseq t))
279 (not (eq cmd 'universal-argument-other-key)))
280 (let ((current-prefix-arg prefix-arg)
281 ;; Have to bind `last-command-char' here so that
282 ;; `digit-argument', for isntance, can compute the
283 ;; prefix arg.
284 (last-command-char (aref keyseq 0)))
285 (call-interactively cmd)))
286
287 ;; This is the final call to `univeral-argument-other-key', which
288 ;; set's the final `prefix-arg.
289 (let ((current-prefix-arg prefix-arg))
290 (call-interactively cmd))
291
292 ;; Read the command to execute with the given prefix arg.
293 (setq prefix prefix-arg
294 keyseq (read-key-sequence nil t)
295 cmd (key-binding keyseq)))
296
15b3e511 297 (let ((coding-system-for-read coding-system)
04363179
GM
298 (coding-system-for-write coding-system)
299 (current-prefix-arg prefix))
15b3e511
KH
300 (message "")
301 (call-interactively cmd))))
302
de94d711 303(defun set-default-coding-systems (coding-system)
0c3154d2 304 "Set default value of various coding systems to CODING-SYSTEM.
387136f6 305This sets the following coding systems:
0c3154d2 306 o coding system of a newly created buffer
8efc03e1
KH
307 o default coding system for subprocess I/O
308This also sets the following values:
387136f6 309 o default value used as file-name-coding-system for converting file names.
03c35c83
EZ
310 o default value for the command `set-terminal-coding-system' (not on MSDOS)
311 o default value for the command `set-keyboard-coding-system'."
de94d711
KH
312 (check-coding-system coding-system)
313 (setq-default buffer-file-coding-system coding-system)
716184d4
RS
314 (if default-enable-multibyte-characters
315 (setq default-file-name-coding-system coding-system))
03c35c83
EZ
316 ;; If coding-system is nil, honor that on MS-DOS as well, so
317 ;; that they could reset the terminal coding system.
318 (unless (and (eq window-system 'pc) coding-system)
319 (setq default-terminal-coding-system coding-system))
de94d711
KH
320 (setq default-keyboard-coding-system coding-system)
321 (setq default-process-coding-system (cons coding-system coding-system)))
322
45d08cb2 323(defalias 'update-iso-coding-systems 'update-coding-systems-internal)
2598a293 324(make-obsolete 'update-iso-coding-systems 'update-coding-systems-internal "20.3")
45d08cb2 325
0c3154d2
KH
326(defun prefer-coding-system (coding-system)
327 "Add CODING-SYSTEM at the front of the priority list for automatic detection.
387136f6 328This also sets the following coding systems:
0c3154d2 329 o coding system of a newly created buffer
8efc03e1
KH
330 o default coding system for subprocess I/O
331This also sets the following values:
c654de1d 332 o default value used as `file-name-coding-system' for converting file names.
03c35c83
EZ
333 o default value for the command `set-terminal-coding-system' (not on MSDOS)
334 o default value for the command `set-keyboard-coding-system'
335
bd3ac67e
EZ
336If CODING-SYSTEM specifies a certain type of EOL conversion, the coding
337systems set by this function will use that type of EOL conversion.
338
03c35c83
EZ
339This command does not change the default value of terminal coding system
340for MS-DOS terminal, because DOS terminals only support a single coding
341system, and Emacs automatically sets the default to that coding system at
5beaa0d1
KH
342startup.
343
c654de1d
DL
344A coding system that requires automatic detection of text
345encoding (e.g. undecided, unix) can't be preferred.
346
347See also `coding-category-list' and `coding-system-category'."
0c3154d2
KH
348 (interactive "zPrefer coding system: ")
349 (if (not (and coding-system (coding-system-p coding-system)))
350 (error "Invalid coding system `%s'" coding-system))
351 (let ((coding-category (coding-system-category coding-system))
bd3ac67e
EZ
352 (base (coding-system-base coding-system))
353 (eol-type (coding-system-eol-type coding-system)))
0c3154d2
KH
354 (if (not coding-category)
355 ;; CODING-SYSTEM is no-conversion or undecided.
356 (error "Can't prefer the coding system `%s'" coding-system))
8efc03e1 357 (set coding-category (or base coding-system))
45d08cb2 358 (update-coding-systems-internal)
812cad80 359 (or (eq coding-category (car coding-category-list))
0c3154d2 360 ;; We must change the order.
812cad80 361 (set-coding-priority (list coding-category)))
8efc03e1
KH
362 (if (and base (interactive-p))
363 (message "Highest priority is set to %s (base of %s)"
364 base coding-system))
bd3ac67e 365 ;; If they asked for specific EOL conversion, honor that.
6f9dc4fd 366 (if (memq eol-type '(0 1 2))
bd3ac67e
EZ
367 (setq coding-system
368 (coding-system-change-eol-conversion base eol-type))
369 (setq coding-system base))
370 (set-default-coding-systems coding-system)))
0c3154d2 371
b5edd1d1
KH
372(defvar sort-coding-systems-predicate nil
373 "If non-nil, a predicate function to sort coding systems.
374
375It is called with two coding systems, and should return t if the first
376one is \"less\" than the second.
377
378The function `sort-coding-systems' use it.")
379
380(defun sort-coding-systems (codings)
381 "Sort coding system list CODINGS by a priority of each coding system.
382
383If a coding system is most preferred, it has the highest priority.
384Otherwise, a coding system corresponds to some MIME charset has higher
385priorities. Among them, a coding system included in `coding-system'
386key of the current language environment has higher priorities. See
387also the documentation of `language-info-alist'.
388
389If the variable `sort-coding-systems-predicate' (which see) is
390non-nil, it is used to sort CODINGS in the different way than above."
391 (if sort-coding-systems-predicate
392 (sort codings sort-coding-systems-predicate)
393 (let* ((most-preferred (symbol-value (car coding-category-list)))
394 (lang-preferred (get-language-info current-language-environment
395 'coding-system))
396 (func (function
397 (lambda (x)
398 (let ((base (coding-system-base x)))
399 (+ (if (eq base most-preferred) 64 0)
400 (let ((mime (coding-system-get base 'mime-charset)))
401 (if mime
402 (if (string-match "^x-" (symbol-name mime))
403 16 32)
404 0))
405 (if (memq base lang-preferred) 8 0)
406 (if (string-match "-with-esc$" (symbol-name base))
407 0 4)
408 (if (eq (coding-system-type base) 2)
409 ;; For ISO based coding systems, prefer
410 ;; one that doesn't use escape sequences.
411 (let ((flags (coding-system-flags base)))
412 (if (or (consp (aref flags 0))
413 (consp (aref flags 1))
414 (consp (aref flags 2))
415 (consp (aref flags 3)))
416 (if (or (aref flags 8) (aref flags 9))
417 0
418 1)
419 2))
420 1)))))))
421 (sort codings (function (lambda (x y)
422 (> (funcall func x) (funcall func y))))))))
54b226f7 423
3fc7dfe5 424(defun find-coding-systems-region (from to)
54b226f7
KH
425 "Return a list of proper coding systems to encode a text between FROM and TO.
426All coding systems in the list can safely encode any multibyte characters
427in the text.
428
e8dd0160 429If the text contains no multibyte characters, return a list of a single
3fc7dfe5 430element `undecided'."
b5edd1d1
KH
431 (let ((codings (find-coding-systems-region-internal from to)))
432 (if (eq codings t)
433 ;; The text contains only ASCII characters. Any coding
434 ;; systems are safe.
435 '(undecided)
436 ;; We need copy-sequence because sorting will alter the argument.
437 (sort-coding-systems (copy-sequence codings)))))
54b226f7 438
3fc7dfe5
KH
439(defun find-coding-systems-string (string)
440 "Return a list of proper coding systems to encode STRING.
441All coding systems in the list can safely encode any multibyte characters
442in STRING.
443
e8dd0160 444If STRING contains no multibyte characters, return a list of a single
3fc7dfe5 445element `undecided'."
b5edd1d1 446 (find-coding-systems-region string nil))
3fc7dfe5
KH
447
448(defun find-coding-systems-for-charsets (charsets)
449 "Return a list of proper coding systems to encode characters of CHARSETS.
450CHARSETS is a list of character sets."
b5edd1d1
KH
451 (cond ((or (null charsets)
452 (and (= (length charsets) 1)
453 (eq 'ascii (car charsets))))
454 '(undecided))
455 ((or (memq 'eight-bit-control charsets)
456 (memq 'eight-bit-graphic charsets))
457 '(raw-text emacs-mule))
458 (t
459 (let ((codings t)
460 charset l ll)
461 (while (and codings charsets)
462 (setq charset (car charsets) charsets (cdr charsets))
463 (unless (eq charset 'ascii)
464 (setq l (aref char-coding-system-table (make-char charset)))
465 (if (eq codings t)
466 (setq codings l)
467 (let ((ll nil))
468 (while codings
469 (if (memq (car codings) l)
470 (setq ll (cons (car codings) ll)))
471 (setq codings (cdr codings)))
472 (setq codings ll)))))
473 (append codings
474 (char-table-extra-slot char-coding-system-table 0))))))
54b226f7 475
51ed58ea
KH
476(defun find-multibyte-characters (from to &optional maxcount excludes)
477 "Find multibyte characters in the region specified by FROM and TO.
478If FROM is a string, find multibyte characters in the string.
479The return value is an alist of the following format:
480 ((CHARSET COUNT CHAR ...) ...)
481where
482 CHARSET is a character set,
483 COUNT is a number of characters,
484 CHARs are found characters of the character set.
485Optional 3rd arg MAXCOUNT limits how many CHARs are put in the above list.
251d4f4b
KH
486Optional 4th arg EXCLUDE is a list of character sets to be ignored.
487
488For invalid characters, CHARs are actually strings."
51ed58ea
KH
489 (let ((chars nil)
490 charset char)
491 (if (stringp from)
492 (let ((idx 0))
493 (while (setq idx (string-match "[^\000-\177]" from idx))
494 (setq char (aref from idx)
495 charset (char-charset char))
251d4f4b
KH
496 (if (eq charset 'unknown)
497 (setq char (match-string 0)))
14333e31
KH
498 (if (or (memq charset '(unknown
499 eight-bit-control eight-bit-graphic))
251d4f4b 500 (not (or (eq excludes t) (memq charset excludes))))
51ed58ea
KH
501 (let ((slot (assq charset chars)))
502 (if slot
503 (if (not (memq char (nthcdr 2 slot)))
504 (let ((count (nth 1 slot)))
505 (setcar (cdr slot) (1+ count))
506 (if (or (not maxcount) (< count maxcount))
507 (nconc slot (list char)))))
508 (setq chars (cons (list charset 1 char) chars)))))
509 (setq idx (1+ idx))))
510 (save-excursion
511 (goto-char from)
512 (while (re-search-forward "[^\000-\177]" to t)
513 (setq char (preceding-char)
514 charset (char-charset char))
251d4f4b
KH
515 (if (eq charset 'unknown)
516 (setq char (match-string 0)))
14333e31 517 (if (or (memq charset '(unknown eight-bit-control eight-bit-graphic))
251d4f4b 518 (not (or (eq excludes t) (memq charset excludes))))
51ed58ea
KH
519 (let ((slot (assq charset chars)))
520 (if slot
251d4f4b 521 (if (not (member char (nthcdr 2 slot)))
51ed58ea
KH
522 (let ((count (nth 1 slot)))
523 (setcar (cdr slot) (1+ count))
524 (if (or (not maxcount) (< count maxcount))
525 (nconc slot (list char)))))
526 (setq chars (cons (list charset 1 char) chars))))))))
527 (nreverse chars)))
528
c83c4f60
RS
529(defvar last-coding-system-specified nil
530 "Most recent coding system explicitly specified by the user when asked.
531This variable is set whenever Emacs asks the user which coding system
532to use in order to write a file. If you set it to nil explicitly,
533then call `write-region', then afterward this variable will be non-nil
534only if the user was explicitly asked and specified a coding system.")
535
b5edd1d1
KH
536(defvar select-safe-coding-system-accept-default-p nil
537 "If non-nil, a function to control the behaviour of coding system selection.
538The meaning is the same as the argument ACCEPT-DEFAULT-P of the
539function `select-safe-coding-system' (which see). This variable
540overrides that argument.")
541
542(defun select-safe-coding-system (from to &optional default-coding-system
543 accept-default-p)
d5266ddf
KH
544 "Ask a user to select a safe coding system from candidates.
545The candidates of coding systems which can safely encode a text
b5edd1d1
KH
546between FROM and TO are shown in a popup window. Among them, the most
547proper one is suggested as the default.
548
549The list of `buffer-file-coding-system' of the current buffer and the
550most preferred coding system (if it corresponds to a MIME charset) is
551treated as the default coding system list. Among them, the first one
552that safely encodes the text is silently selected and returned without
553any user interaction. See also the command `prefer-coding-system'.
54b226f7 554
b5edd1d1
KH
555Optional 3rd arg DEFAULT-CODING-SYSTEM specifies a coding system or a
556list of coding systems to be prepended to the default coding system
557list.
54b226f7 558
b5edd1d1
KH
559Optional 4th arg ACCEPT-DEFAULT-P, if non-nil, is a function to
560determine the acceptability of the silently selected coding system.
561It is called with that coding system, and should return nil if it
562should not be silently selected and thus user interaction is required.
563
564The variable `select-safe-coding-system-accept-default-p', if
565non-nil, overrides ACCEPT-DEFAULT-P.
54b226f7
KH
566
567Kludgy feature: if FROM is a string, the string is the target text,
568and TO is ignored."
b5edd1d1
KH
569 (if (and default-coding-system
570 (not (listp default-coding-system)))
571 (setq default-coding-system (list default-coding-system)))
572
573 ;; Change elements of the list to (coding . base-coding).
574 (setq default-coding-system
575 (mapcar (function (lambda (x) (cons x (coding-system-base x))))
576 default-coding-system))
577
578 ;; If buffer-file-coding-system is not nil nor undecided, append it
579 ;; to the defaults.
580 (if buffer-file-coding-system
581 (let ((base (coding-system-base buffer-file-coding-system)))
582 (or (eq base 'undecided)
583 (assq buffer-file-coding-system default-coding-system)
584 (rassq base default-coding-system)
e56d7900 585 (setq default-coding-system
b5edd1d1
KH
586 (append default-coding-system
587 (list (cons buffer-file-coding-system base)))))))
588
589 ;; If the most preferred coding system has the property mime-charset,
590 ;; append it to the defaults.
c24e49a8
KH
591 (let ((tail coding-category-list)
592 preferred base)
593 (while (and tail
87ad2c90 594 (not (setq preferred (symbol-value (car tail)))))
c24e49a8
KH
595 (setq tail (cdr tail)))
596 (and (coding-system-p preferred)
597 (setq base (coding-system-base preferred))
598 (coding-system-get preferred 'mime-charset)
b5edd1d1
KH
599 (not (assq preferred default-coding-system))
600 (not (rassq base default-coding-system))
601 (setq default-coding-system
602 (append default-coding-system (list (cons preferred base))))))
603
604 (if select-safe-coding-system-accept-default-p
605 (setq accept-default-p select-safe-coding-system-accept-default-p))
606
607 (let ((codings (find-coding-systems-region from to))
608 (coding-system nil)
609 (l default-coding-system))
610 (if (eq (car codings) 'undecided)
611 ;; Any coding system is ok.
612 (setq coding-system t)
613 ;; Try the defaults.
614 (while (and l (not coding-system))
615 (if (memq (cdr (car l)) codings)
616 (setq coding-system (car (car l)))
617 (setq l (cdr l))))
618 (if (and coding-system accept-default-p)
619 (or (funcall accept-default-p coding-system)
620 (setq coding-system (list coding-system)))))
621
622 ;; If all the defaults failed, ask a user.
623 (when (or (not coding-system) (consp coding-system))
34104362 624 ;; At first, change each coding system to the corresponding
b5edd1d1
KH
625 ;; mime-charset name if it is also a coding system. Such a name
626 ;; is more friendly to users.
627 (let ((l codings)
34104362
KH
628 mime-charset)
629 (while l
630 (setq mime-charset (coding-system-get (car l) 'mime-charset))
631 (if (and mime-charset (coding-system-p mime-charset))
632 (setcar l mime-charset))
633 (setq l (cdr l))))
634
b5edd1d1
KH
635 ;; Then ask users to select one form CODINGS.
636 (unwind-protect
637 (save-window-excursion
638 (with-output-to-temp-buffer "*Warning*"
639 (save-excursion
640 (set-buffer standard-output)
a672b7cd
KH
641 (if (not default-coding-system)
642 (insert "No default coding systems to try.")
643 (insert "These default coding systems were tried")
644 (if (stringp from)
645 (insert " to encode \""
646 (if (> (length from) 10)
647 (substring from 0 10)
648 from)
649 "...\""))
650 (insert ":\n")
651 (let ((pos (point))
652 (fill-prefix " "))
653 (mapcar (function (lambda (x)
654 (princ " ") (princ (car x))))
655 default-coding-system)
656 (insert "\n")
657 (fill-region-as-paragraph pos (point)))
658 (insert
659 (if (consp coding-system)
660 (concat (format "%s safely encodes the target text,\n"
661 (car coding-system))
662 "but it is not recommended for encoding text in this context,\n"
663 "e.g., for sending an email message.\n")
664 "However, none of them safely encodes the target text.\n")))
b5edd1d1 665 (insert (if (consp coding-system)
2f1fa038
EZ
666 "\nSelect the above, or "
667 "\nSelect ")
668 "one of the following safe coding systems:\n")
b5edd1d1
KH
669 (let ((pos (point))
670 (fill-prefix " "))
671 (mapcar (function (lambda (x) (princ " ") (princ x)))
672 codings)
673 (insert "\n")
674 (fill-region-as-paragraph pos (point)))))
675
676 ;; Read a coding system.
677 (if (consp coding-system)
678 (setq codings (cons (car coding-system) codings)))
679 (let* ((safe-names (mapcar (lambda (x) (list (symbol-name x)))
680 codings))
681 (name (completing-read
682 (format "Select coding system (default %s): "
683 (car codings))
684 safe-names nil t nil nil
685 (car (car safe-names)))))
686 (setq last-coding-system-specified (intern name)
687 coding-system last-coding-system-specified)))
688 (kill-buffer "*Warning*")))
689
690 (if (vectorp (coding-system-eol-type coding-system))
691 (let ((eol (coding-system-eol-type buffer-file-coding-system)))
692 (if (numberp eol)
693 (setq coding-system
694 (coding-system-change-eol-conversion coding-system eol)))))
695
696 (if (eq coding-system t)
697 (setq coding-system buffer-file-coding-system))
e56d7900 698 coding-system))
54b226f7
KH
699
700(setq select-safe-coding-system-function 'select-safe-coding-system)
701
46babb23
KH
702(defun select-message-coding-system ()
703 "Return a coding system to encode the outgoing message of the current buffer.
704It at first tries the first coding system found in these variables
705in this order:
706 (1) local value of `buffer-file-coding-system'
707 (2) value of `sendmail-coding-system'
b5edd1d1
KH
708 (3) value of `default-sendmail-coding-system'
709 (4) value of `default-buffer-file-coding-system'
46babb23
KH
710If the found coding system can't encode the current buffer,
711or none of them are bound to a coding system,
48e41165 712it asks the user to select a proper coding system."
46babb23 713 (let ((coding (or (and (local-variable-p 'buffer-file-coding-system)
b5edd1d1
KH
714 buffer-file-coding-system)
715 sendmail-coding-system
716 default-sendmail-coding-system
717 default-buffer-file-coding-system)))
46babb23
KH
718 (if (eq coding 'no-conversion)
719 ;; We should never use no-conversion for outgoing mails.
720 (setq coding nil))
721 (if (fboundp select-safe-coding-system-function)
722 (funcall select-safe-coding-system-function
b5edd1d1
KH
723 (point-min) (point-max) coding
724 (function (lambda (x) (coding-system-get x 'mime-charset))))
46babb23 725 coding)))
4ed46869 726\f
03c35c83 727;;; Language support stuff.
4ed46869 728
4ed46869 729(defvar language-info-alist nil
2c395d56 730 "Alist of language environment definitions.
4ed46869
KH
731Each element looks like:
732 (LANGUAGE-NAME . ((KEY . INFO) ...))
2c395d56
RS
733where LANGUAGE-NAME is a string, the name of the language environment,
734KEY is a symbol denoting the kind of information, and
735INFO is the data associated with KEY.
736Meaningful values for KEY include
737
738 documentation value is documentation of what this language environment
739 is meant for, and how to use it.
740 charset value is a list of the character sets used by this
741 language environment.
742 sample-text value is one line of text,
743 written using those character sets,
744 appropriate for this language environment.
745 setup-function value is a function to call to switch to this
746 language environment.
747 exit-function value is a function to call to leave this
748 language environment.
749 coding-system value is a list of coding systems that are good
750 for saving text written in this language environment.
751 This list serves as suggestions to the user;
752 in effect, as a kind of documentation.
753 coding-priority value is a list of coding systems for this language
754 environment, in order of decreasing priority.
755 This is used to set up the coding system priority
45d08cb2 756 list when you switch to this language environment.
ddb5c041 757 nonascii-translation
7624ebb9 758 value is a translation table to be set in the
45d08cb2 759 variable `nonascii-translation-table' in this
7624ebb9
KH
760 language environment, or a character set from
761 which `nonascii-insert-offset' is calculated.
ddb5c041
KH
762 input-method value is a default input method for this language
763 environment.
7624ebb9
KH
764 features value is a list of features requested in this
765 language environment.
ddb5c041
KH
766
767The following keys take effect only when multibyte characters are
768globally disabled, i.e. the value of `default-enable-multibyte-characters'
769is nil.
770
771 unibyte-syntax value is a library name to load to set
e8dd0160 772 unibyte 8-bit character syntaxes for this
ddb5c041
KH
773 language environment.
774
775 unibyte-display value is a coding system to encode characters
776 for the terminal. Characters in the range
777 of 160 to 255 display not as octal escapes,
778 but as non-ASCII characters in this language
779 environment.")
2c395d56
RS
780
781(defun get-language-info (lang-env key)
782 "Return information listed under KEY for language environment LANG-ENV.
783KEY is a symbol denoting the kind of information.
784For a list of useful values for KEY and their meanings,
785see `language-info-alist'."
786 (if (symbolp lang-env)
787 (setq lang-env (symbol-name lang-env)))
788 (let ((lang-slot (assoc-ignore-case lang-env language-info-alist)))
4ed46869
KH
789 (if lang-slot
790 (cdr (assq key (cdr lang-slot))))))
791
f08adf27 792(defun set-language-info (lang-env key info)
2c395d56
RS
793 "Modify part of the definition of language environment LANG-ENV.
794Specifically, this stores the information INFO under KEY
795in the definition of this language environment.
4ed46869 796KEY is a symbol denoting the kind of information.
2c395d56 797INFO is the value for that information.
281d03ec 798
2c395d56 799For a list of useful values for KEY and their meanings,
f08adf27 800see `language-info-alist'."
2c395d56
RS
801 (if (symbolp lang-env)
802 (setq lang-env (symbol-name lang-env)))
4ed46869 803 (let (lang-slot key-slot)
2c395d56 804 (setq lang-slot (assoc lang-env language-info-alist))
4ed46869 805 (if (null lang-slot) ; If no slot for the language, add it.
2c395d56 806 (setq lang-slot (list lang-env)
4ed46869
KH
807 language-info-alist (cons lang-slot language-info-alist)))
808 (setq key-slot (assq key lang-slot))
809 (if (null key-slot) ; If no slot for the key, add it.
810 (progn
811 (setq key-slot (list key))
812 (setcdr lang-slot (cons key-slot (cdr lang-slot)))))
cda74479 813 (setcdr key-slot (purecopy info))))
4ed46869 814
2c395d56
RS
815(defun set-language-info-alist (lang-env alist &optional parents)
816 "Store ALIST as the definition of language environment LANG-ENV.
817ALIST is an alist of KEY and INFO values. See the documentation of
98c6d6ed 818`language-info-alist' for the meanings of KEY and INFO.
54b226f7 819
2c395d56
RS
820Optional arg PARENTS is a list of parent menu names; it specifies
821where to put this language environment in the
822Describe Language Environment and Set Language Environment menus.
823For example, (\"European\") means to put this language environment
824in the European submenu in each of those two menus."
825 (if (symbolp lang-env)
826 (setq lang-env (symbol-name lang-env)))
54b226f7
KH
827 (let ((describe-map describe-language-environment-map)
828 (setup-map setup-language-environment-map))
829 (if parents
830 (let ((l parents)
9deed82f 831 map parent-symbol parent prompt)
54b226f7
KH
832 (while l
833 (if (symbolp (setq parent-symbol (car l)))
834 (setq parent (symbol-name parent))
835 (setq parent parent-symbol parent-symbol (intern parent)))
836 (setq map (lookup-key describe-map (vector parent-symbol)))
9deed82f
EZ
837 ;; This prompt string is for define-prefix-command, so
838 ;; that the map it creates will be suitable for a menu.
839 (or map (setq prompt (format "%s Environment" parent)))
54b226f7
KH
840 (if (not map)
841 (progn
842 (setq map (intern (format "describe-%s-environment-map"
843 (downcase parent))))
9deed82f 844 (define-prefix-command map nil prompt)
54b226f7
KH
845 (define-key-after describe-map (vector parent-symbol)
846 (cons parent map) t)))
847 (setq describe-map (symbol-value map))
848 (setq map (lookup-key setup-map (vector parent-symbol)))
849 (if (not map)
850 (progn
851 (setq map (intern (format "setup-%s-environment-map"
852 (downcase parent))))
9deed82f 853 (define-prefix-command map nil prompt)
54b226f7
KH
854 (define-key-after setup-map (vector parent-symbol)
855 (cons parent map) t)))
856 (setq setup-map (symbol-value map))
857 (setq l (cdr l)))))
f08adf27
RS
858
859 ;; Set up menu items for this language env.
7624ebb9 860 (let ((doc (assq 'documentation alist)))
f08adf27
RS
861 (when doc
862 (define-key-after describe-map (vector (intern lang-env))
7624ebb9
KH
863 (cons lang-env 'describe-specified-language-support) t)))
864 (define-key-after setup-map (vector (intern lang-env))
865 (cons lang-env 'setup-specified-language-environment) t)
f08adf27 866
54b226f7 867 (while alist
f08adf27 868 (set-language-info lang-env (car (car alist)) (cdr (car alist)))
54b226f7 869 (setq alist (cdr alist)))))
4ed46869 870
ae302641 871(defun read-language-name (key prompt &optional default)
2c395d56 872 "Read a language environment name which has information for KEY.
ddb5c041 873If KEY is nil, read any language environment.
2c395d56
RS
874Prompt with PROMPT. DEFAULT is the default choice of language environment.
875This returns a language environment name as a string."
4ed46869
KH
876 (let* ((completion-ignore-case t)
877 (name (completing-read prompt
878 language-info-alist
ddb5c041
KH
879 (and key
880 (function (lambda (elm) (assq key elm))))
ae302641 881 t nil nil default)))
13e82c04 882 (if (and (> (length name) 0)
ddb5c041
KH
883 (or (not key)
884 (get-language-info name key)))
13e82c04 885 name)))
4ed46869
KH
886\f
887;;; Multilingual input methods.
d0c40faa
KH
888(defgroup leim nil
889 "LEIM: Libraries of Emacs Input Methods."
890 :group 'mule)
4ed46869 891
d0b9c3ab
KH
892(defconst leim-list-file-name "leim-list.el"
893 "Name of LEIM list file.
894This file contains a list of libraries of Emacs input methods (LEIM)
895in the format of Lisp expression for registering each input method.
896Emacs loads this file at startup time.")
897
2e224638
SM
898(defvar leim-list-header (format
899";;; %s -- list of LEIM (Library of Emacs Input Method)
d0b9c3ab
KH
900;;
901;; This file contains a list of LEIM (Library of Emacs Input Method)
c654de1d
DL
902;; methods in the same directory as this file. Loading this file
903;; registers all the input methods in Emacs.
d0b9c3ab 904;;
d33d5fbe 905;; Each entry has the form:
d0b9c3ab
KH
906;; (register-input-method
907;; INPUT-METHOD LANGUAGE-NAME ACTIVATE-FUNC
908;; TITLE DESCRIPTION
909;; ARG ...)
c654de1d 910;; See the function `register-input-method' for the meanings of the arguments.
d0b9c3ab
KH
911;;
912;; If this directory is included in load-path, Emacs automatically
913;; loads this file at startup time.
914
915"
916 leim-list-file-name)
917 "Header to be inserted in LEIM list file.")
918
e55e92ee 919(defvar leim-list-entry-regexp "^(register-input-method"
d0b9c3ab
KH
920 "Regexp matching head of each entry in LEIM list file.
921See also the variable `leim-list-header'")
922
923(defvar update-leim-list-functions
924 '(quail-update-leim-list-file)
925 "List of functions to call to update LEIM list file.
926Each function is called with one arg, LEIM directory name.")
927
a337fe7f
RS
928(defun update-leim-list-file (&rest dirs)
929 "Update LEIM list file in directories DIRS."
d0b9c3ab
KH
930 (let ((functions update-leim-list-functions))
931 (while functions
a337fe7f 932 (apply (car functions) dirs)
d0b9c3ab
KH
933 (setq functions (cdr functions)))))
934
4ed46869
KH
935(defvar current-input-method nil
936 "The current input method for multilingual text.
96db204a 937If nil, that means no input method is activated now.")
4ed46869
KH
938(make-variable-buffer-local 'current-input-method)
939(put 'current-input-method 'permanent-local t)
940
941(defvar current-input-method-title nil
d0b9c3ab 942 "Title string of the current input method shown in mode line.")
4ed46869
KH
943(make-variable-buffer-local 'current-input-method-title)
944(put 'current-input-method-title 'permanent-local t)
945
b4fba33f 946(defcustom default-input-method nil
8861c593 947 "*Default input method for multilingual text (a string).
b4fba33f 948This is the input method activated automatically by the command
9b10b5a3 949`toggle-input-method' (\\[toggle-input-method])."
8861c593 950 :group 'mule
5806e8a6
GM
951 :type '(choice (const nil) string)
952 :set-after '(current-language-environment))
b4fba33f 953
0f835e87
KH
954(put 'input-method-function 'permanent-local t)
955
723a427a
KH
956(defvar input-method-history nil
957 "History list for some commands that read input methods.")
958(make-variable-buffer-local 'input-method-history)
959(put 'input-method-history 'permanent-local t)
4ed46869
KH
960
961(defvar inactivate-current-input-method-function nil
962 "Function to call for inactivating the current input method.
963Every input method should set this to an appropriate value when activated.
f17ccaee
KH
964This function is called with no argument.
965
966This function should never change the value of `current-input-method'.
967It is set to nil by the function `inactivate-input-method'.")
4ed46869
KH
968(make-variable-buffer-local 'inactivate-current-input-method-function)
969(put 'inactivate-current-input-method-function 'permanent-local t)
970
971(defvar describe-current-input-method-function nil
972 "Function to call for describing the current input method.
973This function is called with no argument.")
974(make-variable-buffer-local 'describe-current-input-method-function)
975(put 'describe-current-input-method-function 'permanent-local t)
976
d0b9c3ab 977(defvar input-method-alist nil
2c395d56 978 "Alist of input method names vs how to use them.
d0b9c3ab 979Each element has the form:
2c395d56
RS
980 (INPUT-METHOD LANGUAGE-ENV ACTIVATE-FUNC TITLE DESCRIPTION ARGS...)
981See the function `register-input-method' for the meanings of the elements.")
982
f08adf27 983(defun register-input-method (input-method lang-env &rest args)
bf42aa15 984 "Register INPUT-METHOD as an input method for language environment LANG-ENV.
f08adf27 985INPUT-METHOD and LANG-ENV are symbols or strings.
d0b9c3ab 986
d0b9c3ab 987The remaining arguments are:
2c395d56
RS
988 ACTIVATE-FUNC, TITLE, DESCRIPTION, and ARGS...
989ACTIVATE-FUNC is a function to call to activate this method.
990TITLE is a string to show in the mode line when this method is active.
991DESCRIPTION is a string describing this method and what it is good for.
992The ARGS, if any, are passed as arguments to ACTIVATE-FUNC.
205814ee
KH
993All told, the arguments to ACTIVATE-FUNC are INPUT-METHOD and the ARGS.
994
c654de1d
DL
995This function is mainly used in the file \"leim-list.el\" which is
996created at Emacs build time, registering all Quail input methods
0b6cadff 997contained in the Emacs distribution.
205814ee 998
0b6cadff 999In case you want to register a new Quail input method by yourself, be
205814ee 1000careful to use the same input method title as given in the third
0b6cadff
DL
1001parameter of `quail-define-package'. (If the values are different, the
1002string specified in this function takes precedence.)
205814ee
KH
1003
1004The commands `describe-input-method' and `list-input-methods' need
0b6cadff
DL
1005these duplicated values to show some information about input methods
1006without loading the relevant Quail packages."
f08adf27
RS
1007 (if (symbolp lang-env)
1008 (setq lang-env (symbol-name lang-env)))
4ef06f75
KH
1009 (if (symbolp input-method)
1010 (setq input-method (symbol-name input-method)))
f08adf27 1011 (let ((info (cons lang-env args))
d0b9c3ab
KH
1012 (slot (assoc input-method input-method-alist)))
1013 (if slot
1014 (setcdr slot info)
1015 (setq slot (cons input-method info))
1016 (setq input-method-alist (cons slot input-method-alist)))))
1017
4d5ac029 1018(defun read-input-method-name (prompt &optional default inhibit-null)
d0b9c3ab 1019 "Read a name of input method from a minibuffer prompting with PROMPT.
4d5ac029 1020If DEFAULT is non-nil, use that as the default,
0b6cadff 1021and substitute it into PROMPT at the first `%s'.
4ef06f75
KH
1022If INHIBIT-NULL is non-nil, null input signals an error.
1023
1024The return value is a string."
4d5ac029
RS
1025 (if default
1026 (setq prompt (format prompt default)))
d0b9c3ab 1027 (let* ((completion-ignore-case t)
c54044ff
KH
1028 ;; As it is quite normal to change input method in the
1029 ;; minibuffer, we must enable it even if
1030 ;; enable-recursive-minibuffers is currently nil.
1031 (enable-recursive-minibuffers t)
723a427a
KH
1032 ;; This binding is necessary because input-method-history is
1033 ;; buffer local.
d0b9c3ab 1034 (input-method (completing-read prompt input-method-alist
87505a98
RS
1035 nil t nil 'input-method-history
1036 default)))
bf294e6e
KH
1037 (if (and input-method (symbolp input-method))
1038 (setq input-method (symbol-name input-method)))
d0b9c3ab
KH
1039 (if (> (length input-method) 0)
1040 input-method
1041 (if inhibit-null
43807b77 1042 (error "No valid input method is specified")))))
d0b9c3ab 1043
d0b9c3ab 1044(defun activate-input-method (input-method)
2c395d56
RS
1045 "Switch to input method INPUT-METHOD for the current buffer.
1046If some other input method is already active, turn it off first.
1047If INPUT-METHOD is nil, deactivate any current input method."
305a3cb6 1048 (if (and input-method (symbolp input-method))
4ef06f75 1049 (setq input-method (symbol-name input-method)))
723a427a
KH
1050 (if (and current-input-method
1051 (not (string= current-input-method input-method)))
305a3cb6 1052 (inactivate-input-method))
2c395d56 1053 (unless (or current-input-method (null input-method))
d0b9c3ab
KH
1054 (let ((slot (assoc input-method input-method-alist)))
1055 (if (null slot)
723a427a 1056 (error "Can't activate input method `%s'" input-method))
278dd6ac 1057 (setq current-input-method-title nil)
8efc03e1
KH
1058 (let ((func (nth 2 slot)))
1059 (if (functionp func)
1060 (apply (nth 2 slot) input-method (nthcdr 5 slot))
1061 (if (and (consp func) (symbolp (car func)) (symbolp (cdr func)))
1062 (progn
1063 (require (cdr func))
1064 (apply (car func) input-method (nthcdr 5 slot)))
1065 (error "Can't activate input method `%s'" input-method))))
d0b9c3ab 1066 (setq current-input-method input-method)
278dd6ac
KH
1067 (or (stringp current-input-method-title)
1068 (setq current-input-method-title (nth 3 slot)))
28885c0e
KH
1069 (unwind-protect
1070 (run-hooks 'input-method-activate-hook)
1071 (force-mode-line-update)))))
15b3e511 1072
15b3e511 1073(defun inactivate-input-method ()
f17ccaee 1074 "Turn off the current input method."
723a427a
KH
1075 (when current-input-method
1076 (if input-method-history
1077 (unless (string= current-input-method (car input-method-history))
1078 (setq input-method-history
1079 (cons current-input-method
1080 (delete current-input-method input-method-history))))
1081 (setq input-method-history (list current-input-method)))
1082 (unwind-protect
1083 (funcall inactivate-current-input-method-function)
15b3e511 1084 (unwind-protect
723a427a
KH
1085 (run-hooks 'input-method-inactivate-hook)
1086 (setq current-input-method nil
2b1ec973 1087 input-method-function nil
28885c0e
KH
1088 current-input-method-title nil)
1089 (force-mode-line-update)))))
4ed46869 1090
8b784951 1091(defun set-input-method (input-method)
2c395d56 1092 "Select and activate input method INPUT-METHOD for the current buffer.
bc406911 1093This also sets the default input method to the one you specify.
402dbbd1
EZ
1094If INPUT-METHOD is nil, this function turns off the input method, and
1095also causes you to be prompted for a name of an input method the next
1096time you invoke \\[toggle-input-method].
1097
bc406911
EZ
1098To deactivate the input method interactively, use \\[toggle-input-method].
1099To deactivate it programmatically, use \\[inactivate-input-method]."
d0b9c3ab 1100 (interactive
723a427a 1101 (let* ((default (or (car input-method-history) default-input-method)))
42395763 1102 (list (read-input-method-name
87505a98 1103 (if default "Select input method (default %s): " "Select input method: ")
42395763 1104 default t))))
d0b9c3ab 1105 (activate-input-method input-method)
42395763 1106 (setq default-input-method input-method))
4ed46869
KH
1107
1108(defun toggle-input-method (&optional arg)
f8ec20be
RS
1109 "Enable or disable multilingual text input method for the current buffer.
1110Only one input method can be enabled at any time in a given buffer.
1111
1112The normal action is to enable an input method if none was
1113enabled, and disable the current one otherwise. Which input method
1114to enable can be determined in various ways--either the one most
1115recently used, or the one specified by `default-input-method', or
1116as a last resort by reading the name of an input method in the
1117minibuffer.
1118
1119With a prefix argument, read an input method name with the minibuffer
1120and enable that one. The default is the most recent input method specified
f2979bdb 1121\(not including the currently active input method, if any)."
f8ec20be 1122
4ed46869 1123 (interactive "P")
7ddbb5bc
RS
1124 (if (and current-input-method (not arg))
1125 (inactivate-input-method)
1126 (let ((default (or (car input-method-history) default-input-method)))
1127 (if (and arg default (equal current-input-method default)
1128 (> (length input-method-history) 1))
1129 (setq default (nth 1 input-method-history)))
723a427a
KH
1130 (activate-input-method
1131 (if (or arg (not default))
7ddbb5bc
RS
1132 (progn
1133 (read-input-method-name
1134 (if default "Input method (default %s): " "Input method: " )
1135 default t))
723a427a
KH
1136 default))
1137 (or default-input-method
1138 (setq default-input-method current-input-method)))))
d0b9c3ab
KH
1139
1140(defun describe-input-method (input-method)
2c395d56 1141 "Describe input method INPUT-METHOD."
d0b9c3ab
KH
1142 (interactive
1143 (list (read-input-method-name
1144 "Describe input method (default, current choice): ")))
78754934 1145 (if (and input-method (symbolp input-method))
4ef06f75 1146 (setq input-method (symbol-name input-method)))
d0b9c3ab
KH
1147 (if (null input-method)
1148 (describe-current-input-method)
464cc130
KH
1149 (let ((current current-input-method))
1150 (condition-case nil
1151 (progn
1152 (save-excursion
1153 (activate-input-method input-method)
1154 (describe-current-input-method))
1155 (activate-input-method current))
1156 (error
1157 (activate-input-method current)
1158 (with-output-to-temp-buffer "*Help*"
1159 (let ((elt (assoc input-method input-method-alist)))
1160 (princ (format
1161 "Input method: %s (`%s' in mode line) for %s\n %s\n"
1162 input-method (nth 3 elt) (nth 1 elt) (nth 4 elt))))))))))
d0b9c3ab
KH
1163
1164(defun describe-current-input-method ()
96db204a 1165 "Describe the input method currently in use."
4ed46869
KH
1166 (if current-input-method
1167 (if (and (symbolp describe-current-input-method-function)
1168 (fboundp describe-current-input-method-function))
1169 (funcall describe-current-input-method-function)
1170 (message "No way to describe the current input method `%s'"
f2979bdb 1171 current-input-method)
4ed46869 1172 (ding))
d0b9c3ab 1173 (error "No input method is activated now")))
4ed46869 1174
d3459641 1175(defun read-multilingual-string (prompt &optional initial-input input-method)
4ed46869
KH
1176 "Read a multilingual string from minibuffer, prompting with string PROMPT.
1177The input method selected last time is activated in minibuffer.
15b3e511 1178If optional second arg INITIAL-INPUT is non-nil, insert it in the minibuffer
d0b9c3ab
KH
1179initially.
1180Optional 3rd argument INPUT-METHOD specifies the input method
4ef06f75
KH
1181to be activated instead of the one selected last time. It is a symbol
1182or a string."
88d559ec
KH
1183 (setq input-method
1184 (or input-method
d3459641 1185 current-input-method
88d559ec
KH
1186 default-input-method
1187 (read-input-method-name "Input method: " nil t)))
3df60841 1188 (if (and input-method (symbolp input-method))
4ef06f75 1189 (setq input-method (symbol-name input-method)))
305a3cb6
KH
1190 (let ((prev-input-method current-input-method))
1191 (unwind-protect
1192 (progn
1193 (activate-input-method input-method)
1194 (read-string prompt initial-input nil nil t))
1195 (activate-input-method prev-input-method))))
4ed46869
KH
1196
1197;; Variables to control behavior of input methods. All input methods
1198;; should react to these variables.
1199
8efc03e1
KH
1200(defcustom input-method-verbose-flag 'default
1201 "*A flag to control extra guidance given by input methods.
1202The value should be nil, t, `complex-only', or `default'.
4ed46869 1203
cb29dfb6 1204The extra guidance is done by showing list of available keys in echo
8efc03e1
KH
1205area. When you use the input method in the minibuffer, the guidance
1206is shown at the bottom short window (split from the existing window).
c27c4ed8 1207
8efc03e1
KH
1208If the value is t, extra guidance is always given, if the value is
1209nil, extra guidance is always suppressed.
1210
1211If the value is `complex-only', only complex input methods such as
1212`chinese-py' and `japanese' give extra guidance.
1213
1214If the value is `default', complex input methods always give extra
1215guidance, but simple input methods give it only when you are not in
1216the minibuffer.
1217
1218See also the variable `input-method-highlight-flag'."
1219 :type '(choice (const t) (const nil) (const complex-only) (const default))
42395763
RS
1220 :group 'mule)
1221
1222(defcustom input-method-highlight-flag t
1223 "*If this flag is non-nil, input methods highlight partially-entered text.
1224For instance, while you are in the middle of a Quail input method sequence,
1225the text inserted so far is temporarily underlined.
8efc03e1
KH
1226The underlining goes away when you finish or abort the input method sequence.
1227See also the variable `input-method-verbose-flag'."
42395763
RS
1228 :type 'boolean
1229 :group 'mule)
4ed46869
KH
1230
1231(defvar input-method-activate-hook nil
f17ccaee
KH
1232 "Normal hook run just after an input method is activated.
1233
1234The variable `current-input-method' keeps the input method name
1235just activated.")
4ed46869
KH
1236
1237(defvar input-method-inactivate-hook nil
f17ccaee
KH
1238 "Normal hook run just after an input method is inactivated.
1239
1240The variable `current-input-method' still keeps the input method name
4d0e6a11 1241just inactivated.")
4ed46869
KH
1242
1243(defvar input-method-after-insert-chunk-hook nil
1244 "Normal hook run just after an input method insert some chunk of text.")
1245
dccca980 1246(defvar input-method-exit-on-first-char nil
0b6cadff 1247 "This flag controls when an input method returns.
dccca980
KH
1248Usually, the input method does not return while there's a possibility
1249that it may find a different translation if a user types another key.
39e643e2
RS
1250But, it this flag is non-nil, the input method returns as soon as
1251the current key sequence gets long enough to have some valid translation.")
dccca980
KH
1252
1253(defvar input-method-use-echo-area nil
1254 "This flag controls how an input method shows an intermediate key sequence.
39e643e2
RS
1255Usually, the input method inserts the intermediate key sequence,
1256or candidate translations corresponding to the sequence,
1257at point in the current buffer.
1258But, if this flag is non-nil, it displays them in echo area instead.")
dccca980 1259
723a427a
KH
1260(defvar input-method-exit-on-invalid-key nil
1261 "This flag controls the behaviour of an input method on invalid key input.
1262Usually, when a user types a key which doesn't start any character
1263handled by the input method, the key is handled by turning off the
e8dd0160 1264input method temporarily. After that key, the input method is re-enabled.
723a427a
KH
1265But, if this flag is non-nil, the input method is never back on.")
1266
4ed46869 1267\f
8efc03e1
KH
1268(defvar set-language-environment-hook nil
1269 "Normal hook run after some language environment is set.
1270
1271When you set some hook function here, that effect usually should not
1272be inherited to another language environment. So, you had better set
1273another function in `exit-language-environment-hook' (which see) to
1274cancel the effect.")
1275
1276(defvar exit-language-environment-hook nil
1277 "Normal hook run after exiting from some language environment.
1278When this hook is run, the variable `current-language-environment'
1279is still bound to the language environment being exited.
1280
e8dd0160 1281This hook is mainly used for canceling the effect of
8efc03e1
KH
1282`set-language-environment-hook' (which-see).")
1283
b0648a00
RS
1284(put 'setup-specified-language-environment 'apropos-inhibit t)
1285
15b3e511 1286(defun setup-specified-language-environment ()
f08adf27 1287 "Switch to a specified language environment."
15b3e511 1288 (interactive)
f850d782 1289 (let (language-name)
15b3e511
KH
1290 (if (and (symbolp last-command-event)
1291 (or (not (eq last-command-event 'Default))
1292 (setq last-command-event 'English))
f850d782
RS
1293 (setq language-name (symbol-name last-command-event)))
1294 (set-language-environment language-name)
15b3e511 1295 (error "Bogus calling sequence"))))
4ed46869 1296
8861c593 1297(defcustom current-language-environment "English"
94d04df6 1298 "The last language environment specified with `set-language-environment'.
ebef6d93
KH
1299This variable should be set only with \\[customize], which is equivalent
1300to using the function `set-language-environment'."
94d04df6 1301 :link '(custom-manual "(emacs)Language Environments")
dff1aa24 1302 :set (lambda (symbol value) (set-language-environment value))
94d04df6
DL
1303 :get (lambda (x)
1304 (or (car-safe (assoc-ignore-case
1305 (if (symbolp current-language-environment)
1306 (symbol-name current-language-environment)
1307 current-language-environment)
1308 language-info-alist))
1309 "English"))
1310 :type (cons 'choice (mapcar (lambda (lang)
1311 (list 'const (car lang)))
1312 language-info-alist))
8861c593
RS
1313 :initialize 'custom-initialize-default
1314 :group 'mule
1315 :type 'string)
f850d782 1316
ddb5c041
KH
1317(defun reset-language-environment ()
1318 "Reset multilingual environment of Emacs to the default status.
1319
1320The default status is as follows:
1321
1322 The default value of buffer-file-coding-system is nil.
1323 The default coding system for process I/O is nil.
1324 The default value for the command `set-terminal-coding-system' is nil.
1325 The default value for the command `set-keyboard-coding-system' is nil.
1326
1327 The order of priorities of coding categories and the coding system
1328 bound to each category are as follows
1329 coding category coding system
1330 --------------------------------------------------
1331 coding-category-iso-8-2 iso-latin-1
1332 coding-category-iso-8-1 iso-latin-1
1333 coding-category-iso-7-tight iso-2022-jp
1334 coding-category-iso-7 iso-2022-7bit
1335 coding-category-iso-7-else iso-2022-7bit-lock
1336 coding-category-iso-8-else iso-2022-8bit-ss2
1337 coding-category-emacs-mule emacs-mule
1338 coding-category-raw-text raw-text
1339 coding-category-sjis japanese-shift-jis
1340 coding-category-big5 chinese-big5
1341 coding-category-ccl nil
e8dd0160 1342 coding-category-binary no-conversion
631f62fd
KH
1343 coding-category-utf-16-be nil
1344 coding-category-utf-16-le nil
c654de1d 1345 coding-category-utf-8 mule-utf-8"
ddb5c041
KH
1346 (interactive)
1347 ;; This function formerly set default-enable-multibyte-characters to t,
1348 ;; but that is incorrect. It should not alter the unibyte/multibyte choice.
1349
1350 (setq coding-category-iso-7-tight 'iso-2022-jp
1351 coding-category-iso-7 'iso-2022-7bit
1352 coding-category-iso-8-1 'iso-latin-1
1353 coding-category-iso-8-2 'iso-latin-1
1354 coding-category-iso-7-else 'iso-2022-7bit-lock
1355 coding-category-iso-8-else 'iso-2022-8bit-ss2
1356 coding-category-emacs-mule 'emacs-mule
1357 coding-category-raw-text 'raw-text
1358 coding-category-sjis 'japanese-shift-jis
1359 coding-category-big5 'chinese-big5
631f62fd 1360 coding-category-utf-16-be nil
b909cfc0
KH
1361 coding-category-utf-16-le nil
1362 coding-category-utf-8 'mule-utf-8
1363 coding-category-ccl nil
1364 coding-category-binary 'no-conversion)
ddb5c041
KH
1365
1366 (set-coding-priority
1367 '(coding-category-iso-8-1
1368 coding-category-iso-8-2
1369 coding-category-iso-7-tight
1370 coding-category-iso-7
1371 coding-category-iso-7-else
1372 coding-category-iso-8-else
1373 coding-category-emacs-mule
1374 coding-category-raw-text
1375 coding-category-sjis
1376 coding-category-big5
1377 coding-category-ccl
9bfcd269
KH
1378 coding-category-binary
1379 coding-category-utf-16-be
b909cfc0
KH
1380 coding-category-utf-16-le
1381 coding-category-utf-8))
ddb5c041 1382
91693d18
KH
1383 (update-coding-systems-internal)
1384
ddb5c041 1385 (set-default-coding-systems nil)
b5edd1d1 1386 (setq default-sendmail-coding-system 'iso-latin-1)
0c47a7c8 1387 (setq default-process-coding-system '(undecided . iso-latin-1))
b5edd1d1 1388
ddb5c041
KH
1389 ;; Don't alter the terminal and keyboard coding systems here.
1390 ;; The terminal still supports the same coding system
1391 ;; that it supported a minute ago.
1392;;; (set-terminal-coding-system-internal nil)
1393;;; (set-keyboard-coding-system-internal nil)
1394
1395 (setq nonascii-translation-table nil
1396 nonascii-insert-offset 0))
1397
0c47a7c8
KH
1398(reset-language-environment)
1399
40c81f74
PE
1400(defun set-display-table-and-terminal-coding-system (language-name)
1401 "Set up the display table and terminal coding system for LANGUAGE-NAME."
1402 (let ((coding (get-language-info language-name 'unibyte-display)))
1403 (if coding
1404 (standard-display-european-internal)
1405 (standard-display-default (if (eq window-system 'pc) 128 160) 255)
1406 (aset standard-display-table 146 nil))
1407 (or (eq window-system 'pc)
1408 (set-terminal-coding-system coding))))
1409
166246f7 1410(defun set-language-environment (language-name)
6c05d680
RS
1411 "Set up multi-lingual environment for using LANGUAGE-NAME.
1412This sets the coding system priority and the default input method
8861c593
RS
1413and sometimes other things. LANGUAGE-NAME should be a string
1414which is the name of a language environment. For example, \"Latin-1\"
1415specifies the character set for the major languages of Western Europe."
8efc03e1 1416 (interactive (list (read-language-name
ddb5c041 1417 nil
8efc03e1 1418 "Set language environment (default, English): ")))
4ef06f75
KH
1419 (if language-name
1420 (if (symbolp language-name)
1421 (setq language-name (symbol-name language-name)))
1422 (setq language-name "English"))
ddb5c041 1423 (or (assoc-ignore-case language-name language-info-alist)
f850d782 1424 (error "Language environment not defined: %S" language-name))
8efc03e1
KH
1425 (if current-language-environment
1426 (let ((func (get-language-info current-language-environment
1427 'exit-function)))
e63645c2
KH
1428 (run-hooks 'exit-language-environment-hook)
1429 (if (fboundp func) (funcall func))))
03c35c83
EZ
1430 (let ((default-eol-type (coding-system-eol-type
1431 default-buffer-file-coding-system)))
1432 (reset-language-environment)
ddb5c041 1433
03c35c83
EZ
1434 (setq current-language-environment language-name)
1435 (set-language-environment-coding-systems language-name default-eol-type))
ddb5c041
KH
1436 (let ((input-method (get-language-info language-name 'input-method)))
1437 (when input-method
1438 (setq default-input-method input-method)
1439 (if input-method-history
1440 (setq input-method-history
1441 (cons input-method
1442 (delete input-method input-method-history))))))
ec241f58
EZ
1443 (let ((nonascii (get-language-info language-name 'nonascii-translation))
1444 (dos-table
4e2ac2d9
EZ
1445 (if (eq window-system 'pc)
1446 (intern
d9c0a50e 1447 (format "cp%d-nonascii-translation-table" dos-codepage)))))
03c35c83
EZ
1448 (cond
1449 ((char-table-p nonascii)
1450 (setq nonascii-translation-table nonascii))
ec241f58 1451 ((and (eq window-system 'pc) (boundp dos-table))
03c35c83
EZ
1452 ;; DOS terminals' default is to use a special non-ASCII translation
1453 ;; table as appropriate for the installed codepage.
ec241f58 1454 (setq nonascii-translation-table (symbol-value dos-table)))
03c35c83
EZ
1455 ((charsetp nonascii)
1456 (setq nonascii-insert-offset (- (make-char nonascii) 128)))))
ddb5c041 1457
63283a8f 1458 ;; Unibyte setups if necessary.
ddb5c041 1459 (unless default-enable-multibyte-characters
63283a8f 1460 ;; Syntax and case table.
ddb5c041
KH
1461 (let ((syntax (get-language-info language-name 'unibyte-syntax)))
1462 (if syntax
1463 (let ((set-case-syntax-set-multibyte nil))
63283a8f
KH
1464 (load syntax nil t))
1465 ;; No information for syntax and case. Reset to the defaults.
1466 (let ((syntax-table (standard-syntax-table))
1467 (case-table (standard-case-table))
03c35c83 1468 (ch (if (eq window-system 'pc) 128 160)))
63283a8f
KH
1469 (while (< ch 256)
1470 (modify-syntax-entry ch " " syntax-table)
1471 (aset case-table ch ch)
1472 (setq ch (1+ ch)))
1473 (set-char-table-extra-slot case-table 0 nil)
1474 (set-char-table-extra-slot case-table 1 nil)
1475 (set-char-table-extra-slot case-table 2 nil))
1476 (set-standard-case-table (standard-case-table))
1477 (let ((list (buffer-list)))
1478 (while list
1479 (with-current-buffer (car list)
1480 (set-case-table (standard-case-table)))
1481 (setq list (cdr list))))))
40c81f74 1482 (set-display-table-and-terminal-coding-system language-name))
ddb5c041
KH
1483
1484 (let ((required-features (get-language-info language-name 'features)))
1485 (while required-features
1486 (require (car required-features))
1487 (setq required-features (cdr required-features))))
1488 (let ((func (get-language-info language-name 'setup-function)))
1489 (if (fboundp func)
1490 (funcall func)))
8efc03e1 1491 (run-hooks 'set-language-environment-hook)
f850d782 1492 (force-mode-line-update t))
4ed46869 1493
51a8fc1d
RS
1494(defun standard-display-european-internal ()
1495 ;; Actually set up direct output of non-ASCII characters.
03c35c83
EZ
1496 (standard-display-8bit (if (eq window-system 'pc) 128 160) 255)
1497 ;; Unibyte Emacs on MS-DOS wants to display all 8-bit characters with
1498 ;; the native font, and codes 160 and 146 stand for something very
1499 ;; different there.
1500 (or (and (eq window-system 'pc) (not default-enable-multibyte-characters))
1501 (progn
1502 ;; Make non-line-break space display as a plain space.
1503 ;; Most X fonts do the wrong thing for code 160.
1504 (aset standard-display-table 160 [32])
1492f7ac
DL
1505 ;; With luck, non-Latin-1 fonts are more recent and so don't
1506 ;; have this bug.
1507 (aset standard-display-table 2208 [32]) ; Latin-1 NBSP
1508 ;; Most Windows programs send out apostrophes as \222. Most X fonts
03c35c83 1509 ;; don't contain a character at that position. Map it to the ASCII
1492f7ac
DL
1510 ;; apostrophe. [This is actually RIGHT SINGLE QUOTATION MARK
1511 ;; from the cp1252, aka Windows-1252 character set. --fx]
03c35c83
EZ
1512 (aset standard-display-table 146 [39]))))
1513
1514(defun set-language-environment-coding-systems (language-name
1515 &optional eol-type)
1516 "Do various coding system setups for language environment LANGUAGE-NAME.
1517
1518The optional arg EOL-TYPE specifies the eol-type of the default value
1519of buffer-file-coding-system set by this function."
54b226f7
KH
1520 (let* ((priority (get-language-info language-name 'coding-priority))
1521 (default-coding (car priority)))
1522 (if priority
1523 (let ((categories (mapcar 'coding-system-category priority)))
03c35c83
EZ
1524 (set-default-coding-systems
1525 (if (memq eol-type '(0 1 2 unix dos mac))
1526 (coding-system-change-eol-conversion default-coding eol-type)
1527 default-coding))
46babb23 1528 (setq default-sendmail-coding-system default-coding)
54b226f7
KH
1529 (set-coding-priority categories)
1530 (while priority
1531 (set (car categories) (car priority))
1532 (setq priority (cdr priority) categories (cdr categories)))
45d08cb2 1533 (update-coding-systems-internal)))))
54b226f7 1534
4ed46869
KH
1535;; Print all arguments with `princ', then print "\n".
1536(defsubst princ-list (&rest args)
1537 (while args (princ (car args)) (setq args (cdr args)))
1538 (princ "\n"))
1539
b0648a00
RS
1540(put 'describe-specified-language-support 'apropos-inhibit t)
1541
48082651 1542;; Print a language specific information such as input methods,
13e82c04 1543;; charsets, and coding systems. This function is intended to be
48082651 1544;; called from the menu:
281d03ec 1545;; [menu-bar mule describe-language-environment LANGUAGE]
48082651
KH
1546;; and should not run it by `M-x describe-current-input-method-function'.
1547(defun describe-specified-language-support ()
96db204a 1548 "Describe how Emacs supports the specified language environment."
48082651 1549 (interactive)
281d03ec 1550 (let (language-name)
48082651 1551 (if (not (and (symbolp last-command-event)
cda74479
DL
1552 (or (not (eq last-command-event 'Default))
1553 (setq last-command-event 'English))
281d03ec 1554 (setq language-name (symbol-name last-command-event))))
48082651 1555 (error "Bogus calling sequence"))
281d03ec
RS
1556 (describe-language-environment language-name)))
1557
1558(defun describe-language-environment (language-name)
1559 "Describe how Emacs supports language environment LANGUAGE-NAME."
78754934
KH
1560 (interactive
1561 (list (read-language-name
1562 'documentation
8adfa8be 1563 "Describe language environment (default, current choice): ")))
f850d782
RS
1564 (if (null language-name)
1565 (setq language-name current-language-environment))
281d03ec
RS
1566 (if (or (null language-name)
1567 (null (get-language-info language-name 'documentation)))
1568 (error "No documentation for the specified language"))
4ef06f75
KH
1569 (if (symbolp language-name)
1570 (setq language-name (symbol-name language-name)))
464cc130
KH
1571 (let ((doc (get-language-info language-name 'documentation))
1572 pos)
48082651 1573 (with-output-to-temp-buffer "*Help*"
464cc130
KH
1574 (save-excursion
1575 (set-buffer standard-output)
1576 (insert language-name " language environment\n\n")
1577 (if (stringp doc)
1578 (insert doc "\n\n"))
e036b0a6
KH
1579 (condition-case nil
1580 (let ((str (eval (get-language-info language-name 'sample-text))))
1581 (if (stringp str)
1582 (insert "Sample text:\n " str "\n\n")))
1583 (error nil))
464cc130
KH
1584 (let ((input-method (get-language-info language-name 'input-method))
1585 (l (copy-sequence input-method-alist)))
1586 (insert "Input methods")
1587 (when input-method
1588 (insert " (default, " input-method ")")
1589 (setq input-method (assoc input-method input-method-alist))
1590 (setq l (cons input-method (delete input-method l))))
1591 (insert ":\n")
48082651 1592 (while l
464cc130
KH
1593 (when (string= language-name (nth 1 (car l)))
1594 (insert " " (car (car l)))
1595 (search-backward (car (car l)))
1596 (help-xref-button 0 #'describe-input-method (car (car l))
1597 "mouse-2, RET: describe this input method")
1598 (goto-char (point-max))
2fa7e202
KH
1599 (insert " (\""
1600 (if (stringp (nth 3 (car l)))
1601 (nth 3 (car l))
1602 (car (nth 3 (car l))))
1603 "\" in mode line)\n"))
464cc130
KH
1604 (setq l (cdr l)))
1605 (insert "\n"))
1606 (insert "Character sets:\n")
1607 (let ((l (get-language-info language-name 'charset)))
1608 (if (null l)
1609 (insert " nothing specific to " language-name "\n")
1610 (while l
1611 (insert " " (symbol-name (car l)))
1612 (search-backward (symbol-name (car l)))
1613 (help-xref-button 0 #'describe-character-set (car l)
1614 "mouse-2, RET: describe this character set")
1615 (goto-char (point-max))
1616 (insert ": " (charset-description (car l)) "\n")
1617 (setq l (cdr l)))))
1618 (insert "\n")
1619 (insert "Coding systems:\n")
1620 (let ((l (get-language-info language-name 'coding-system)))
1621 (if (null l)
1622 (insert " nothing specific to " language-name "\n")
1623 (while l
1624 (insert " " (symbol-name (car l)))
1625 (search-backward (symbol-name (car l)))
1626 (help-xref-button 0 #'describe-coding-system (car l)
1627 "mouse-2, RET: describe this coding system")
1628 (goto-char (point-max))
1629 (insert " (`"
1630 (coding-system-mnemonic (car l))
1631 "' in mode line):\n\t"
1632 (coding-system-doc-string (car l))
1633 "\n")
1634 (let ((aliases (coding-system-get (car l)
1635 'alias-coding-systems)))
1636 (when aliases
1637 (insert "\t(alias:")
1638 (while aliases
1639 (insert " " (symbol-name (car aliases)))
1640 (setq aliases (cdr aliases)))
1641 (insert ")\n")))
1642 (setq l (cdr l)))))
1643 (help-setup-xref (list #'describe-language-environment language-name)
1644 (interactive-p))))))
4ed46869 1645\f
40c81f74
PE
1646;;; Locales.
1647
0d7c5bb9
DL
1648(defvar locale-translation-file-name nil
1649 "File name for the system's file of locale-name aliases, or nil if none.")
40c81f74
PE
1650
1651(defvar locale-language-names
1652 '(
1653 ;; UTF-8 is not yet implemented.
1654 ;; Put this first, so that e.g. "ko.UTF-8" does not match "ko" below.
1655 (".*[._]utf" . nil)
1656
1657 ;; Locale names of the form LANGUAGE[_TERRITORY][.CODESET][@MODIFIER]
1658 ;; as specified in the Single Unix Spec, Version 2.
1659 ;; LANGUAGE is a language code taken from ISO 639:1988 (E/F)
1660 ;; with additions from ISO 639/RA Newsletter No.1/1989;
1661 ;; see Internet RFC 2165 (1997-06).
1662 ;; TERRITORY is a country code taken from ISO 3166.
1663 ;; CODESET and MODIFIER are implementation-dependent.
1664 ;;
1665 ; aa Afar
1666 ; ab Abkhazian
6ececc4d 1667 ("af" . "Latin-1") ; Afrikaans
40c81f74
PE
1668 ("am" . "Ethiopic") ; Amharic
1669 ; ar Arabic
1670 ; as Assamese
1671 ; ay Aymara
1672 ; az Azerbaijani
1673 ; ba Bashkir
6ececc4d
PE
1674 ("be" . "Latin-5") ; Byelorussian
1675 ("bg" . "Latin-5") ; Bulgarian
40c81f74
PE
1676 ; bh Bihari
1677 ; bi Bislama
1678 ; bn Bengali, Bangla
1679 ("bo" . "Tibetan")
1680 ("br" . "Latin-1") ; Breton
1681 ("ca" . "Latin-1") ; Catalan
1682 ; co Corsican
1683 ("cs" . "Czech")
6ececc4d 1684 ("cy" . "Latin-8") ; Welsh
40c81f74
PE
1685 ("da" . "Latin-1") ; Danish
1686 ("de" . "German")
1687 ; dz Bhutani
1688 ("el" . "Greek")
6ececc4d
PE
1689 ;; Users who specify "en" explicitly typically want Latin-1, not ASCII.
1690 ("en" . "Latin-1") ; English
40c81f74 1691 ("eo" . "Latin-3") ; Esperanto
8c4b6822 1692 ("es" . "Spanish")
40c81f74
PE
1693 ("et" . "Latin-4") ; Estonian
1694 ("eu" . "Latin-1") ; Basque
1695 ; fa Persian
1696 ("fi" . "Latin-1") ; Finnish
1697 ; fj Fiji
1698 ("fo" . "Latin-1") ; Faroese
1699 ("fr" . "Latin-1") ; French
1700 ("fy" . "Latin-1") ; Frisian
6ececc4d
PE
1701 ("ga" . "Latin-1") ; Irish Gaelic (new orthography)
1702 ("gd" . "Latin-1") ; Scots Gaelic
1703 ("gl" . "Latin-1") ; Galician
40c81f74
PE
1704 ; gn Guarani
1705 ; gu Gujarati
1706 ; ha Hausa
1707 ("he" . "Hebrew")
1708 ("hi" . "Devanagari") ; Hindi
1709 ("hr" . "Latin-2") ; Croatian
1710 ("hu" . "Latin-2") ; Hungarian
1711 ; hy Armenian
1712 ; ia Interlingua
1713 ("id" . "Latin-1") ; Indonesian
1714 ; ie Interlingue
1715 ; ik Inupiak
1716 ("is" . "Latin-1") ; Icelandic
1717 ("it" . "Latin-1") ; Italian
1718 ; iu Inuktitut
1719 ("ja" . "Japanese")
1720 ; jw Javanese
1721 ; ka Georgian
1722 ; kk Kazakh
6ececc4d 1723 ("kl" . "Latin-1") ; Greenlandic
40c81f74
PE
1724 ; km Cambodian
1725 ; kn Kannada
1726 ("ko" . "Korean")
1727 ; ks Kashmiri
1728 ; ku Kurdish
1729 ; ky Kirghiz
1730 ("la" . "Latin-1") ; Latin
1731 ; ln Lingala
1732 ("lo" . "Lao") ; Laothian
1733 ("lt" . "Latin-4") ; Lithuanian
1734 ("lv" . "Latin-4") ; Latvian, Lettish
1735 ; mg Malagasy
1736 ; mi Maori
6ececc4d 1737 ("mk" . "Latin-5") ; Macedonian
40c81f74
PE
1738 ; ml Malayalam
1739 ; mn Mongolian
1740 ; mo Moldavian
1741 ("mr" . "Devanagari") ; Marathi
1742 ; ms Malay
1743 ("mt" . "Latin-3") ; Maltese
1744 ; my Burmese
1745 ; na Nauru
1746 ("ne" . "Devanagari") ; Nepali
8c4b6822 1747 ("nl" . "Dutch")
40c81f74
PE
1748 ("no" . "Latin-1") ; Norwegian
1749 ; oc Occitan
1750 ; om (Afan) Oromo
1751 ; or Oriya
1752 ; pa Punjabi
1753 ("pl" . "Latin-2") ; Polish
1754 ; ps Pashto, Pushto
1755 ("pt" . "Latin-1") ; Portuguese
1756 ; qu Quechua
6ececc4d 1757 ("rm" . "Latin-1") ; Rhaeto-Romanic
40c81f74
PE
1758 ; rn Kirundi
1759 ("ro" . "Romanian")
1760 ("ru.*[_.]koi8" . "Cyrillic-KOI8") ; Russian
6ececc4d 1761 ("ru" . "Latin-5") ; Russian
40c81f74
PE
1762 ; rw Kinyarwanda
1763 ("sa" . "Devanagari") ; Sanskrit
1764 ; sd Sindhi
1765 ; sg Sangho
1766 ("sh" . "Latin-2") ; Serbo-Croatian
1767 ; si Sinhalese
1768 ("sk" . "Slovak")
1769 ("sl" . "Slovenian")
1770 ; sm Samoan
1771 ; sn Shona
1772 ; so Somali
6ececc4d 1773 ("sq" . "Latin-1") ; Albanian
40c81f74
PE
1774 ("sr" . "Latin-2") ; Serbian (Latin alphabet)
1775 ; ss Siswati
1776 ; st Sesotho
1777 ; su Sundanese
1778 ("sv" . "Latin-1") ; Swedish
1779 ("sw" . "Latin-1") ; Swahili
1780 ; ta Tamil
1781 ; te Telugu
1782 ; tg Tajik
1783 ("th" . "Thai")
1784 ; ti Tigrinya
1785 ; tk Turkmen
6ececc4d 1786 ("tl" . "Latin-1") ; Tagalog
40c81f74
PE
1787 ; tn Setswana
1788 ; to Tonga
1789 ("tr" . "Latin-5") ; Turkish
1790 ; ts Tsonga
1791 ; tt Tatar
1792 ; tw Twi
1793 ; ug Uighur
6ececc4d 1794 ("uk" . "Latin-5") ; Ukrainian
40c81f74
PE
1795 ; ur Urdu
1796 ; uz Uzbek
1797 ("vi" . "Vietnamese")
1798 ; vo Volapuk
1799 ; wo Wolof
1800 ; xh Xhosa
1801 ; yi Yiddish
1802 ; yo Yoruba
1803 ; za Zhuang
1804 ("zh.*[._]big5" . "Chinese-BIG5")
1805 ("zh.*[._]gbk" . nil) ; Solaris 2.7; has gbk-0 as well as GB 2312.1980-0
1806 ("zh_tw" . "Chinese-CNS")
1807 ("zh" . "Chinese-GB")
1808 ; zu Zulu
1809
1810 ;; ISO standard locales
1811 ("c$" . "ASCII")
1812 ("posix$" . "ASCII")
1813
40c81f74
PE
1814 ;; The "IPA" Emacs language environment does not correspond
1815 ;; to any ISO 639 code, so let it stand for itself.
1816 ("ipa$" . "IPA")
1817
1818 ;; Nonstandard or obsolete language codes
1819 ("cz" . "Czech") ; e.g. Solaris 2.6
1820 ("ee" . "Latin-4") ; Estonian, e.g. X11R6.4
1821 ("iw" . "Hebrew") ; e.g. X11R6.4
6ececc4d 1822 ("sp" . "Latin-5") ; Serbian (Cyrillic alphabet), e.g. X11R6.4
40c81f74 1823 ("su" . "Latin-1") ; Finnish, e.g. Solaris 2.6
2e86ceaa 1824 ("jp" . "Japanese") ; e.g. MS Windows
40c81f74
PE
1825 )
1826 "List of pairs of locale regexps and language names.
6ececc4d
PE
1827The first element whose locale regexp matches the start of a downcased locale
1828specifies the language name corresponding to that locale.
40c81f74
PE
1829If the language name is nil, there is no corresponding language environment.")
1830
6ececc4d
PE
1831(defvar locale-charset-language-names
1832 '((".*8859[-_]?1\\>" . "Latin-1")
1833 (".*8859[-_]?2\\>" . "Latin-2")
1834 (".*8859[-_]?3\\>" . "Latin-3")
1835 (".*8859[-_]?4\\>" . "Latin-4")
1836 (".*8859[-_]?9\\>" . "Latin-5")
1837 (".*8859[-_]?14\\>" . "Latin-8")
1838 (".*8859[-_]?15\\>" . "Latin-9")
da645c53 1839 (".*@euro\\>" . "Latin-9")
6ececc4d
PE
1840 )
1841 "List of pairs of locale regexps and charset language names.
1842The first element whose locale regexp matches the start of a downcased locale
1843specifies the language name whose charsets corresponds to that locale.
1844This language name is used if its charsets disagree with the charsets of
1845the language name that would otherwise be used for this locale.")
1846
40c81f74
PE
1847(defvar locale-preferred-coding-systems
1848 '(("ja.*[._]euc" . japanese-iso-8bit)
1849 ("ja.*[._]jis7" . iso-2022-jp)
1850 ("ja.*[._]pck" . japanese-shift-jis)
1851 ("ja.*[._]sjis" . japanese-shift-jis)
40c81f74 1852 )
6ececc4d
PE
1853 "List of pairs of locale regexps and preferred coding systems.
1854The first element whose locale regexp matches the start of a downcased locale
1855specifies the coding system to prefer when using that locale.")
40c81f74
PE
1856
1857(defun locale-name-match (key alist)
1858 "Search for KEY in ALIST, which should be a list of regexp-value pairs.
1859Return the value corresponding to the first regexp that matches the
1860start of KEY, or nil if there is no match."
1861 (let (element)
1862 (while (and alist (not element))
1863 (if (string-match (concat "^\\(" (car (car alist)) "\\)") key)
1864 (setq element (car alist)))
1865 (setq alist (cdr alist)))
1866 (cdr element)))
1867
1868(defun set-locale-environment (locale-name)
1869 "Set up multi-lingual environment for using LOCALE-NAME.
1870This sets the coding system priority and the default input method
1871and sometimes other things. LOCALE-NAME should be a string
1872which is the name of a locale supported by the system;
1873often it is of the form xx_XX.CODE, where xx is a language,
1874XX is a country, and CODE specifies a character set and coding system.
1875For example, the locale name \"ja_JP.EUC\" might name a locale
1876for Japanese in Japan using the `japanese-iso-8bit' coding-system.
1877
1878If LOCALE-NAME is nil, its value is taken from the environment.
1879
1880The locale names supported by your system can typically be found in a
0812c1e8
DL
1881directory named `/usr/share/locale' or `/usr/lib/locale'. LOCALE-NAME
1882may be translated according to the table specified by
1883`locale-translation-file-name'.
1884
1885See also `locale-charset-language-names', `locale-language-names',
1886`locale-preferred-coding-systems' and `locale-coding-system'."
40c81f74 1887
0d7c5bb9
DL
1888 ;; Do this at runtime for the sake of binaries possibly transported
1889 ;; to a system without X.
1890 (setq locale-translation-file-name
1891 (let ((files
1892 '("/usr/lib/X11/locale/locale.alias" ; e.g. X11R6.4
1893 "/usr/X11R6/lib/X11/locale/locale.alias" ; e.g. RedHat 4.2
1894 "/usr/openwin/lib/locale/locale.alias" ; e.g. Solaris 2.6
1895 ;;
1896 ;; The following name appears after the X-related names above,
1897 ;; since the X-related names are what X actually uses.
1898 "/usr/share/locale/locale.alias" ; GNU/Linux sans X
1899 )))
1900 (while (and files (not (file-exists-p (car files))))
1901 (setq files (cdr files)))
1902 (car files)))
1903
7008ccac
GM
1904 (let ((locale locale-name))
1905
1906 (unless locale
1907 ;; Use the first of these three environment variables
1908 ;; that has a nonempty value.
1909 (let ((vars '("LC_ALL" "LC_CTYPE" "LANG")))
1910 (while (and vars (not (setq locale (getenv (car vars)))))
1911 (setq vars (cdr vars)))))
1912
1913 (when locale
1914
1915 ;; Translate "swedish" into "sv_SE.ISO8859-1", and so on,
1916 ;; using the translation file that many systems have.
1917 (when locale-translation-file-name
1918 (with-temp-buffer
1919 (insert-file-contents locale-translation-file-name)
1920 (when (re-search-forward
1921 (concat "^" (regexp-quote locale) ":?[ \t]+") nil t)
1922 (setq locale (buffer-substring (point) (line-end-position))))))
1923
1924 ;; Leave the system locales alone if the caller did not specify
1925 ;; an explicit locale name, as their defaults are set from
1926 ;; LC_MESSAGES and LC_TIME, not LC_CTYPE, and the user might not
1927 ;; want to set them to the same value as LC_CTYPE.
1928 (when locale-name
1929 (setq system-messages-locale locale)
1930 (setq system-time-locale locale))
1931
1932 (setq locale (downcase locale))
1933
1934 (let ((language-name
1935 (locale-name-match locale locale-language-names))
1936 (charset-language-name
1937 (locale-name-match locale locale-charset-language-names))
1938 (coding-system
1939 (locale-name-match locale locale-preferred-coding-systems)))
1940
1941 (if (and charset-language-name
1942 (not
1943 (equal (get-language-info language-name 'charset)
1944 (get-language-info charset-language-name 'charset))))
1945 (setq language-name charset-language-name))
1946
1947 (when language-name
1948
1949 ;; Set up for this character set. This is now the right way
1950 ;; to do it for both unibyte and multibyte modes.
1951 (set-language-environment language-name)
1952
1953 ;; If default-enable-multibyte-characters is nil,
1954 ;; we are using single-byte characters,
1955 ;; so the display table and terminal coding system are irrelevant.
1956 (when default-enable-multibyte-characters
1957 (set-display-table-and-terminal-coding-system language-name))
1958
1959 (setq locale-coding-system
1960 (car (get-language-info language-name 'coding-priority))))
1961
1962 (when coding-system
1963 (prefer-coding-system coding-system)
1964 (setq locale-coding-system coding-system))))))
40c81f74 1965\f
4ed46869
KH
1966;;; Charset property
1967
251d4f4b 1968(defun get-charset-property (charset propname)
4ed46869
KH
1969 "Return the value of CHARSET's PROPNAME property.
1970This is the last value stored with
96db204a 1971 (put-charset-property CHARSET PROPNAME VALUE)."
251d4f4b
KH
1972 (and (not (eq charset 'composition))
1973 (plist-get (charset-plist charset) propname)))
4ed46869 1974
251d4f4b 1975(defun put-charset-property (charset propname value)
4ed46869
KH
1976 "Store CHARSETS's PROPNAME property with value VALUE.
1977It can be retrieved with `(get-charset-property CHARSET PROPNAME)'."
1300d43f
KH
1978 (or (eq charset 'composition)
1979 (set-charset-plist charset
1980 (plist-put (charset-plist charset) propname value))))
4ed46869
KH
1981
1982;;; Character code property
1983(put 'char-code-property-table 'char-table-extra-slots 0)
1984
1985(defvar char-code-property-table
1986 (make-char-table 'char-code-property-table)
1987 "Char-table containing a property list of each character code.
1988
1989See also the documentation of `get-char-code-property' and
96db204a 1990`put-char-code-property'.")
4ed46869
KH
1991
1992(defun get-char-code-property (char propname)
1993 "Return the value of CHAR's PROPNAME property in `char-code-property-table'."
1994 (let ((plist (aref char-code-property-table char)))
1995 (if (listp plist)
1996 (car (cdr (memq propname plist))))))
1997
1998(defun put-char-code-property (char propname value)
1999 "Store CHAR's PROPNAME property with VALUE in `char-code-property-table'.
2000It can be retrieved with `(get-char-code-property CHAR PROPNAME)'."
2001 (let ((plist (aref char-code-property-table char)))
2002 (if plist
2003 (let ((slot (memq propname plist)))
2004 (if slot
2005 (setcar (cdr slot) value)
2006 (nconc plist (list propname value))))
2007 (aset char-code-property-table char (list propname value)))))
2008
a127b764
KH
2009\f
2010;; Pretty description of encoded string
2011
2012;; Alist of ISO 2022 control code vs the corresponding mnemonic string.
2013(defvar iso-2022-control-alist
2014 '((?\x1b . "ESC")
2015 (?\x0e . "SO")
2016 (?\x0f . "SI")
2017 (?\x8e . "SS2")
2018 (?\x8f . "SS3")
2019 (?\x9b . "CSI")))
2020
2021(defun encoded-string-description (str coding-system)
2022 "Return a pretty description of STR that is encoded by CODING-SYSTEM."
2023 (setq str (string-as-unibyte str))
993b2a7d 2024 (mapconcat
8c9d55a9 2025 (if (and coding-system (eq (coding-system-type coding-system) 2))
993b2a7d
KH
2026 ;; Try to get a pretty description for ISO 2022 escape sequences.
2027 (function (lambda (x) (or (cdr (assq x iso-2022-control-alist))
2028 (format "%02X" x))))
5de75f53 2029 (function (lambda (x) (format "0x%02X" x))))
993b2a7d 2030 str " "))
a127b764
KH
2031
2032(defun encode-coding-char (char coding-system)
2033 "Encode CHAR by CODING-SYSTEM and return the resulting string.
2034If CODING-SYSTEM can't safely encode CHAR, return nil."
bd953173
KH
2035 (let ((str1 (string-as-multibyte (char-to-string char)))
2036 (str2 (string-as-multibyte (make-string 2 char)))
a89f541b
KH
2037 (safe-chars (and coding-system
2038 (coding-system-get coding-system 'safe-chars)))
8c9d55a9 2039 (charset (char-charset char))
a127b764 2040 enc1 enc2 i1 i2)
a89f541b 2041 (when (or (eq safe-chars t)
8c9d55a9 2042 (eq charset 'ascii)
a89f541b 2043 (and safe-chars (aref safe-chars char)))
a127b764
KH
2044 ;; We must find the encoded string of CHAR. But, just encoding
2045 ;; CHAR will put extra control sequences (usually to designate
2046 ;; ASCII charaset) at the tail if type of CODING is ISO 2022.
2047 ;; To exclude such tailing bytes, we at first encode one-char
2048 ;; string and two-char string, then check how many bytes at the
2049 ;; tail of both encoded strings are the same.
2050
bd953173 2051 (setq enc1 (encode-coding-string str1 coding-system)
a127b764 2052 i1 (length enc1)
bd953173 2053 enc2 (encode-coding-string str2 coding-system)
a127b764
KH
2054 i2 (length enc2))
2055 (while (and (> i1 0) (= (aref enc1 (1- i1)) (aref enc2 (1- i2))))
2056 (setq i1 (1- i1) i2 (1- i2)))
2057
2058 ;; Now (substring enc1 i1) and (substring enc2 i2) are the same,
2059 ;; and they are the extra control sequences at the tail to
2060 ;; exclude.
2061 (substring enc2 0 i2))))
2062
2063
4ed46869 2064;;; mule-cmds.el ends here