Fix previous change.
[bpt/emacs.git] / lisp / international / mule-cmds.el
CommitLineData
8045ca80 1;;; mule-cmds.el --- commands for mulitilingual environment -*-coding: iso-2022-7bit -*-
9ee5b744
SM
2
3;; Copyright (C) 2000, 2001, 2002, 2003, 2004 Free Software Foundation, Inc.
a985cd2f 4;; Copyright (C) 1995, 2003 Electrotechnical Laboratory, JAPAN.
fa526c4a 5;; Licensed to the Free Software Foundation.
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 40(define-key mule-keymap "f" 'set-buffer-file-coding-system)
d80dee05 41(define-key mule-keymap "r" 'revert-buffer-with-coding-system)
2308d0d7 42(define-key mule-keymap "F" 'set-file-name-coding-system)
4ed46869 43(define-key mule-keymap "t" 'set-terminal-coding-system)
15b3e511
KH
44(define-key mule-keymap "k" 'set-keyboard-coding-system)
45(define-key mule-keymap "p" 'set-buffer-process-coding-system)
7624ebb9
KH
46(define-key mule-keymap "x" 'set-selection-coding-system)
47(define-key mule-keymap "X" 'set-next-selection-coding-system)
8b784951 48(define-key mule-keymap "\C-\\" 'set-input-method)
15b3e511 49(define-key mule-keymap "c" 'universal-coding-system-argument)
b4fba33f 50(define-key mule-keymap "l" 'set-language-environment)
4ed46869 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]
cda74479
DL
66 (list 'menu-item "Set Language Environment" setup-language-environment-map
67 :help "Multilingual environment suitable for a specific language"))
15b3e511
KH
68(define-key-after mule-menu-keymap [separator-mule]
69 '("--")
70 t)
71(define-key-after mule-menu-keymap [toggle-input-method]
538d88fb 72 '(menu-item "Toggle Input Method" toggle-input-method)
15b3e511 73 t)
8b784951 74(define-key-after mule-menu-keymap [set-input-method]
538d88fb 75 '(menu-item "Select Input Method..." set-input-method)
15b3e511 76 t)
cda74479
DL
77(define-key-after mule-menu-keymap [describe-input-method]
78 '(menu-item "Describe Input Method" describe-input-method))
15b3e511
KH
79(define-key-after mule-menu-keymap [separator-input-method]
80 '("--")
81 t)
15b3e511 82(define-key-after mule-menu-keymap [set-various-coding-system]
cda74479 83 (list 'menu-item "Set Coding Systems" set-coding-system-map
be9650bc 84 :enable 'default-enable-multibyte-characters))
538d88fb
EZ
85(define-key-after mule-menu-keymap [view-hello-file]
86 '(menu-item "Show Multi-lingual Text" view-hello-file
87 :enable (file-readable-p
88 (expand-file-name "HELLO" data-directory))
89 :help "Display file which says HELLO in many languages")
15b3e511
KH
90 t)
91(define-key-after mule-menu-keymap [separator-coding-system]
92 '("--")
93 t)
538d88fb
EZ
94(define-key-after mule-menu-keymap [describe-language-environment]
95 (list 'menu-item "Describe Language Environment"
96 describe-language-environment-map
cda74479 97 :help "Show multilingual settings for a specific language")
15b3e511 98 t)
538d88fb
EZ
99(define-key-after mule-menu-keymap [describe-input-method]
100 '(menu-item "Describe Input Method..." describe-input-method
cda74479 101 :help "Keyboard layout for a specific input method")
538d88fb
EZ
102 t)
103(define-key-after mule-menu-keymap [describe-coding-system]
104 '(menu-item "Describe Coding System..." describe-coding-system)
105 t)
cda74479
DL
106(define-key-after mule-menu-keymap [list-character-sets]
107 '(menu-item "List Character Sets" list-character-sets
108 :help "Show table of available character sets"))
538d88fb
EZ
109(define-key-after mule-menu-keymap [mule-diag]
110 '(menu-item "Show All of Mule Status" mule-diag
111 :help "Display multilingual environment settings")
15b3e511
KH
112 t)
113
3a151e98 114(define-key-after set-coding-system-map [universal-coding-system-argument]
538d88fb
EZ
115 '(menu-item "For Next Command" universal-coding-system-argument
116 :help "Coding system to be used by next command")
3a151e98 117 t)
d80dee05
RS
118(define-key-after set-coding-system-map [separator-1]
119 '("--")
120 t)
121(define-key-after set-coding-system-map [set-buffer-file-coding-system]
122 '(menu-item "For Saving This Buffer" set-buffer-file-coding-system
123 :help "How to encode this buffer when saved")
124 t)
125(define-key-after set-coding-system-map [revert-buffer-with-coding-system]
126 '(menu-item "For Reverting This File Now" revert-buffer-with-coding-system
127 :enable buffer-file-name
128 :help "Revisit this file immediately using specified coding system")
129 t)
2308d0d7
KH
130(define-key-after set-coding-system-map [set-file-name-coding-system]
131 '(menu-item "For File Name" set-file-name-coding-system
132 :help "How to decode/encode file names")
133 t)
d80dee05
RS
134(define-key-after set-coding-system-map [separator-2]
135 '("--")
15b3e511 136 t)
d80dee05 137
15b3e511 138(define-key-after set-coding-system-map [set-keyboard-coding-system]
538d88fb
EZ
139 '(menu-item "For Keyboard" set-keyboard-coding-system
140 :help "How to decode keyboard input")
15b3e511 141 t)
d80dee05
RS
142(define-key-after set-coding-system-map [set-terminal-coding-system]
143 '(menu-item "For Terminal" set-terminal-coding-system
144 :enable (null (memq window-system '(x w32 mac)))
145 :help "How to encode terminal output")
146 t)
147(define-key-after set-coding-system-map [separator-3]
148 '("--")
15b3e511 149 t)
7624ebb9 150(define-key-after set-coding-system-map [set-selection-coding-system]
538d88fb
EZ
151 '(menu-item "For X Selections/Clipboard" set-selection-coding-system
152 :visible (display-selections-p)
153 :help "How to en/decode data to/from selection/clipboard")
7624ebb9
KH
154 t)
155(define-key-after set-coding-system-map [set-next-selection-coding-system]
538d88fb
EZ
156 '(menu-item "For Next X Selection" set-next-selection-coding-system
157 :visible (display-selections-p)
158 :help "How to en/decode next selection/clipboard operation")
7624ebb9 159 t)
d80dee05
RS
160(define-key-after set-coding-system-map [set-buffer-process-coding-system]
161 '(menu-item "For I/O with Subprocess" set-buffer-process-coding-system
162 :visible (fboundp 'start-process)
163 :enable (get-buffer-process (current-buffer))
164 :help "How to en/decode I/O from/to subprocess connected to this buffer")
165 t)
166
167
15b3e511 168(define-key setup-language-environment-map
538d88fb 169 [Default] '(menu-item "Default" setup-specified-language-environment))
4ed46869 170
cda74479
DL
171(define-key describe-language-environment-map
172 [Default] '(menu-item "Default" describe-specified-language-support))
173
4ed46869
KH
174;; This should be a single character key binding because users use it
175;; very frequently while editing multilingual text. Now we can use
176;; only two such keys: "\C-\\" and "\C-^", but the latter is not
177;; convenient because it requires shifting on most keyboards. An
178;; alternative is "\C-\]" which is now bound to `abort-recursive-edit'
179;; but it won't be used that frequently.
180(define-key global-map "\C-\\" 'toggle-input-method)
181
5f395df3
SM
182;; This is no good because people often type Shift-SPC
183;; meaning to type SPC. -- rms.
184;; ;; Here's an alternative key binding for X users (Shift-SPACE).
185;; (define-key global-map [?\S- ] 'toggle-input-method)
b4fba33f 186
464cc130
KH
187;;; Mule related hyperlinks.
188(defconst help-xref-mule-regexp-template
189 (purecopy (concat "\\(\\<\\("
190 "\\(coding system\\)\\|"
d0c40faa
KH
191 "\\(input method\\)\\|"
192 "\\(character set\\)\\|"
193 "\\(charset\\)"
464cc130
KH
194 "\\)\\s-+\\)?"
195 ;; Note starting with word-syntax character:
196 "`\\(\\sw\\(\\sw\\|\\s_\\)+\\)'")))
197
26d87040
EZ
198(defun coding-system-change-eol-conversion (coding-system eol-type)
199 "Return a coding system which differs from CODING-SYSTEM in eol conversion.
200The returned coding system converts end-of-line by EOL-TYPE
201but text as the same way as CODING-SYSTEM.
202EOL-TYPE should be `unix', `dos', `mac', or nil.
203If EOL-TYPE is nil, the returned coding system detects
204how end-of-line is formatted automatically while decoding.
205
206EOL-TYPE can be specified by an integer 0, 1, or 2.
207They means `unix', `dos', and `mac' respectively."
208 (if (symbolp eol-type)
209 (setq eol-type (cond ((eq eol-type 'unix) 0)
210 ((eq eol-type 'dos) 1)
211 ((eq eol-type 'mac) 2)
212 (t eol-type))))
213 (let ((orig-eol-type (coding-system-eol-type coding-system)))
214 (if (vectorp orig-eol-type)
215 (if (not eol-type)
216 coding-system
217 (aref orig-eol-type eol-type))
218 (let ((base (coding-system-base coding-system)))
219 (if (not eol-type)
220 base
221 (if (= eol-type orig-eol-type)
222 coding-system
223 (setq orig-eol-type (coding-system-eol-type base))
224 (if (vectorp orig-eol-type)
225 (aref orig-eol-type eol-type))))))))
226
227(defun coding-system-change-text-conversion (coding-system coding)
228 "Return a coding system which differs from CODING-SYSTEM in text conversion.
229The returned coding system converts text by CODING
230but end-of-line as the same way as CODING-SYSTEM.
231If CODING is nil, the returned coding system detects
232how text is formatted automatically while decoding."
1d77e15a
JR
233 (let ((eol-type (coding-system-eol-type coding-system)))
234 (coding-system-change-eol-conversion
235 (if coding coding 'undecided)
236 (if (numberp eol-type) (aref [unix dos mac] eol-type)))))
26d87040 237
4ed46869 238(defun toggle-enable-multibyte-characters (&optional arg)
6998e1a1
RS
239 "Change whether this buffer uses multibyte characters.
240With arg, use multibyte characters if the arg is positive.
241
242Note that this command does not convert the byte contents of
243the buffer; it only changes the way those bytes are interpreted.
244In general, therefore, this command *changes* the sequence of
245characters that the current buffer contains.
246
247We suggest you avoid using use this command unless you know what you
248are doing. If you use it by mistake, and the buffer is now displayed
249wrong, use this command again to toggle back to the right mode."
4ed46869 250 (interactive "P")
b7079457
RS
251 (let ((new-flag
252 (if (null arg) (null enable-multibyte-characters)
253 (> (prefix-numeric-value arg) 0))))
254 (set-buffer-multibyte new-flag))
4ed46869
KH
255 (force-mode-line-update))
256
257(defun view-hello-file ()
258 "Display the HELLO file which list up many languages and characters."
259 (interactive)
8f81f784
KH
260 ;; We have to decode the file in any environment.
261 (let ((default-enable-multibyte-characters t)
95fa03b4 262 (coding-system-for-read 'iso-2022-7bit))
544dd975 263 (view-file (expand-file-name "HELLO" data-directory))))
4ed46869 264
9e9a77a6 265(defun universal-coding-system-argument (coding-system)
15b3e511 266 "Execute an I/O command using the specified coding system."
9e9a77a6
RS
267 (interactive
268 (let ((default (and buffer-file-coding-system
34104362
KH
269 (not (eq (coding-system-type buffer-file-coding-system)
270 t))
9e9a77a6
RS
271 buffer-file-coding-system)))
272 (list (read-coding-system
273 (if default
274 (format "Coding system for following command (default, %s): " default)
275 "Coding system for following command: ")
276 default))))
277 (let* ((keyseq (read-key-sequence
e14a8f4c 278 (format "Command to execute with %s:" coding-system)))
04363179
GM
279 (cmd (key-binding keyseq))
280 prefix)
281
282 (when (eq cmd 'universal-argument)
283 (call-interactively cmd)
a1506d29 284
04363179
GM
285 ;; Process keys bound in `universal-argument-map'.
286 (while (progn
287 (setq keyseq (read-key-sequence nil t)
288 cmd (key-binding keyseq t))
289 (not (eq cmd 'universal-argument-other-key)))
290 (let ((current-prefix-arg prefix-arg)
d37ef0f6 291 ;; Have to bind `last-command-char' here so that
34703cb1 292 ;; `digit-argument', for instance, can compute the
04363179
GM
293 ;; prefix arg.
294 (last-command-char (aref keyseq 0)))
295 (call-interactively cmd)))
296
34703cb1 297 ;; This is the final call to `universal-argument-other-key', which
04363179
GM
298 ;; set's the final `prefix-arg.
299 (let ((current-prefix-arg prefix-arg))
300 (call-interactively cmd))
a1506d29 301
04363179
GM
302 ;; Read the command to execute with the given prefix arg.
303 (setq prefix prefix-arg
304 keyseq (read-key-sequence nil t)
305 cmd (key-binding keyseq)))
306
15b3e511 307 (let ((coding-system-for-read coding-system)
04363179 308 (coding-system-for-write coding-system)
0e9ec609 309 (coding-system-require-warning t)
04363179 310 (current-prefix-arg prefix))
15b3e511
KH
311 (message "")
312 (call-interactively cmd))))
313
de94d711 314(defun set-default-coding-systems (coding-system)
0c3154d2 315 "Set default value of various coding systems to CODING-SYSTEM.
387136f6 316This sets the following coding systems:
0c3154d2 317 o coding system of a newly created buffer
8efc03e1
KH
318 o default coding system for subprocess I/O
319This also sets the following values:
d37ef0f6 320 o default value used as `file-name-coding-system' for converting file names.
03c35c83
EZ
321 o default value for the command `set-terminal-coding-system' (not on MSDOS)
322 o default value for the command `set-keyboard-coding-system'."
de94d711
KH
323 (check-coding-system coding-system)
324 (setq-default buffer-file-coding-system coding-system)
9a0ed120
KH
325 (if (fboundp 'ucs-set-table-for-input)
326 (dolist (buffer (buffer-list))
327 (or (local-variable-p 'buffer-file-coding-system buffer)
328 (ucs-set-table-for-input buffer))))
329
716184d4
RS
330 (if default-enable-multibyte-characters
331 (setq default-file-name-coding-system coding-system))
03c35c83
EZ
332 ;; If coding-system is nil, honor that on MS-DOS as well, so
333 ;; that they could reset the terminal coding system.
334 (unless (and (eq window-system 'pc) coding-system)
335 (setq default-terminal-coding-system coding-system))
de94d711 336 (setq default-keyboard-coding-system coding-system)
1d77e15a
JR
337 ;; Preserve eol-type from existing default-process-coding-systems.
338 ;; On non-unix-like systems in particular, these may have been set
339 ;; carefully by the user, or by the startup code, to deal with the
340 ;; users shell appropriately, so should not be altered by changing
341 ;; language environment.
342 (let ((output-coding
343 (coding-system-change-text-conversion
344 (car default-process-coding-system) coding-system))
345 (input-coding
346 (coding-system-change-text-conversion
347 (cdr default-process-coding-system) coding-system)))
348 (setq default-process-coding-system
349 (cons output-coding input-coding))))
de94d711 350
45d08cb2 351(defalias 'update-iso-coding-systems 'update-coding-systems-internal)
2598a293 352(make-obsolete 'update-iso-coding-systems 'update-coding-systems-internal "20.3")
45d08cb2 353
0c3154d2
KH
354(defun prefer-coding-system (coding-system)
355 "Add CODING-SYSTEM at the front of the priority list for automatic detection.
387136f6 356This also sets the following coding systems:
0c3154d2 357 o coding system of a newly created buffer
8efc03e1
KH
358 o default coding system for subprocess I/O
359This also sets the following values:
c654de1d 360 o default value used as `file-name-coding-system' for converting file names.
03c35c83
EZ
361 o default value for the command `set-terminal-coding-system' (not on MSDOS)
362 o default value for the command `set-keyboard-coding-system'
363
bd3ac67e
EZ
364If CODING-SYSTEM specifies a certain type of EOL conversion, the coding
365systems set by this function will use that type of EOL conversion.
366
03c35c83
EZ
367This command does not change the default value of terminal coding system
368for MS-DOS terminal, because DOS terminals only support a single coding
369system, and Emacs automatically sets the default to that coding system at
5beaa0d1
KH
370startup.
371
c654de1d
DL
372A coding system that requires automatic detection of text
373encoding (e.g. undecided, unix) can't be preferred.
374
375See also `coding-category-list' and `coding-system-category'."
0c3154d2
KH
376 (interactive "zPrefer coding system: ")
377 (if (not (and coding-system (coding-system-p coding-system)))
378 (error "Invalid coding system `%s'" coding-system))
379 (let ((coding-category (coding-system-category coding-system))
bd3ac67e
EZ
380 (base (coding-system-base coding-system))
381 (eol-type (coding-system-eol-type coding-system)))
0c3154d2
KH
382 (if (not coding-category)
383 ;; CODING-SYSTEM is no-conversion or undecided.
384 (error "Can't prefer the coding system `%s'" coding-system))
8efc03e1 385 (set coding-category (or base coding-system))
45d08cb2 386 (update-coding-systems-internal)
812cad80 387 (or (eq coding-category (car coding-category-list))
0c3154d2 388 ;; We must change the order.
812cad80 389 (set-coding-priority (list coding-category)))
8efc03e1
KH
390 (if (and base (interactive-p))
391 (message "Highest priority is set to %s (base of %s)"
392 base coding-system))
bd3ac67e 393 ;; If they asked for specific EOL conversion, honor that.
6f9dc4fd 394 (if (memq eol-type '(0 1 2))
bd3ac67e
EZ
395 (setq coding-system
396 (coding-system-change-eol-conversion base eol-type))
397 (setq coding-system base))
398 (set-default-coding-systems coding-system)))
0c3154d2 399
b5edd1d1
KH
400(defvar sort-coding-systems-predicate nil
401 "If non-nil, a predicate function to sort coding systems.
402
403It is called with two coding systems, and should return t if the first
404one is \"less\" than the second.
405
406The function `sort-coding-systems' use it.")
407
408(defun sort-coding-systems (codings)
409 "Sort coding system list CODINGS by a priority of each coding system.
410
411If a coding system is most preferred, it has the highest priority.
412Otherwise, a coding system corresponds to some MIME charset has higher
413priorities. Among them, a coding system included in `coding-system'
414key of the current language environment has higher priorities. See
415also the documentation of `language-info-alist'.
416
417If the variable `sort-coding-systems-predicate' (which see) is
418non-nil, it is used to sort CODINGS in the different way than above."
419 (if sort-coding-systems-predicate
420 (sort codings sort-coding-systems-predicate)
da1ebad1
KH
421 (let* ((from-categories (mapcar #'(lambda (x) (symbol-value x))
422 coding-category-list))
423 (most-preferred (car from-categories))
b5edd1d1
KH
424 (lang-preferred (get-language-info current-language-environment
425 'coding-system))
426 (func (function
427 (lambda (x)
428 (let ((base (coding-system-base x)))
da1ebad1
KH
429 ;; We calculate the priority number 0..255 by
430 ;; using the 8 bits PMMLCEII as this:
431 ;; P: 1 iff most preferred.
432 ;; MM: greater than 0 iff mime-charset.
433 ;; L: 1 iff one of the current lang. env.'s codings.
434 ;; C: 1 iff one of codings listed in the category list.
435 ;; E: 1 iff not XXX-with-esc
436 ;; II: if iso-2022 based, 0..3, else 1.
437 (logior
438 (lsh (if (eq base most-preferred) 1 0) 7)
439 (lsh
440 (let ((mime (coding-system-get base 'mime-charset)))
b439e72a
DL
441 ;; Prefer coding systems corresponding to a
442 ;; MIME charset.
b5edd1d1 443 (if mime
b439e72a
DL
444 ;; Lower utf-16 priority so that we
445 ;; normally prefer utf-8 to it, and put
446 ;; x-ctext below that.
de814dd9
KH
447 (cond ((string-match "utf-16"
448 (symbol-name mime))
da1ebad1 449 2)
b439e72a 450 ((string-match "^x-" (symbol-name mime))
da1ebad1
KH
451 1)
452 (t 3))
b5edd1d1 453 0))
da1ebad1
KH
454 5)
455 (lsh (if (memq base lang-preferred) 1 0) 4)
456 (lsh (if (memq base from-categories) 1 0) 3)
457 (lsh (if (string-match "-with-esc\\'"
458 (symbol-name base))
459 0 1) 2)
460 (if (eq (coding-system-type base) 2)
461 ;; For ISO based coding systems, prefer
462 ;; one that doesn't use escape sequences.
463 (let ((flags (coding-system-flags base)))
464 (if (or (consp (aref flags 0))
465 (consp (aref flags 1))
466 (consp (aref flags 2))
467 (consp (aref flags 3)))
468 (if (or (aref flags 8) (aref flags 9))
469 0
470 1)
471 2))
472 1)))))))
b5edd1d1
KH
473 (sort codings (function (lambda (x y)
474 (> (funcall func x) (funcall func y))))))))
54b226f7 475
3fc7dfe5 476(defun find-coding-systems-region (from to)
54b226f7 477 "Return a list of proper coding systems to encode a text between FROM and TO.
d37ef0f6 478If FROM is a string, find coding systems in that instead of the buffer.
54b226f7
KH
479All coding systems in the list can safely encode any multibyte characters
480in the text.
481
e8dd0160 482If the text contains no multibyte characters, return a list of a single
3fc7dfe5 483element `undecided'."
b5edd1d1
KH
484 (let ((codings (find-coding-systems-region-internal from to)))
485 (if (eq codings t)
486 ;; The text contains only ASCII characters. Any coding
487 ;; systems are safe.
488 '(undecided)
489 ;; We need copy-sequence because sorting will alter the argument.
490 (sort-coding-systems (copy-sequence codings)))))
54b226f7 491
3fc7dfe5
KH
492(defun find-coding-systems-string (string)
493 "Return a list of proper coding systems to encode STRING.
494All coding systems in the list can safely encode any multibyte characters
495in STRING.
496
e8dd0160 497If STRING contains no multibyte characters, return a list of a single
3fc7dfe5 498element `undecided'."
b5edd1d1 499 (find-coding-systems-region string nil))
3fc7dfe5
KH
500
501(defun find-coding-systems-for-charsets (charsets)
502 "Return a list of proper coding systems to encode characters of CHARSETS.
6053d86a
KH
503CHARSETS is a list of character sets.
504It actually checks at most the first 96 characters of each charset.
505So, if a charset of dimension two is included in CHARSETS, the value may
506contain a coding system that can't encode all characters of the charset."
b5edd1d1
KH
507 (cond ((or (null charsets)
508 (and (= (length charsets) 1)
509 (eq 'ascii (car charsets))))
510 '(undecided))
511 ((or (memq 'eight-bit-control charsets)
512 (memq 'eight-bit-graphic charsets))
513 '(raw-text emacs-mule))
514 (t
515 (let ((codings t)
6053d86a 516 charset l str)
b5edd1d1
KH
517 (while (and codings charsets)
518 (setq charset (car charsets) charsets (cdr charsets))
519 (unless (eq charset 'ascii)
6053d86a
KH
520 (setq str (make-string 96 32))
521 (if (= (charset-dimension charset) 1)
522 (if (= (charset-chars charset) 96)
523 (dotimes (i 96)
524 (aset str i (make-char charset (+ i 32))))
525 (dotimes (i 94)
526 (aset str i (make-char charset (+ i 33)))))
527 (if (= (charset-chars charset) 96)
528 (dotimes (i 96)
529 (aset str i (make-char charset 32 (+ i 32))))
530 (dotimes (i 94)
531 (aset str i (make-char charset 33 (+ i 33))))))
532 (setq l (find-coding-systems-string str))
b5edd1d1
KH
533 (if (eq codings t)
534 (setq codings l)
535 (let ((ll nil))
6053d86a
KH
536 (dolist (elt codings)
537 (if (memq elt l)
538 (setq ll (cons elt ll))))
b5edd1d1 539 (setq codings ll)))))
6053d86a 540 codings))))
54b226f7 541
51ed58ea
KH
542(defun find-multibyte-characters (from to &optional maxcount excludes)
543 "Find multibyte characters in the region specified by FROM and TO.
544If FROM is a string, find multibyte characters in the string.
545The return value is an alist of the following format:
546 ((CHARSET COUNT CHAR ...) ...)
547where
548 CHARSET is a character set,
549 COUNT is a number of characters,
34703cb1 550 CHARs are the characters found from the character set.
51ed58ea 551Optional 3rd arg MAXCOUNT limits how many CHARs are put in the above list.
251d4f4b
KH
552Optional 4th arg EXCLUDE is a list of character sets to be ignored.
553
554For invalid characters, CHARs are actually strings."
51ed58ea
KH
555 (let ((chars nil)
556 charset char)
557 (if (stringp from)
558 (let ((idx 0))
559 (while (setq idx (string-match "[^\000-\177]" from idx))
560 (setq char (aref from idx)
561 charset (char-charset char))
251d4f4b
KH
562 (if (eq charset 'unknown)
563 (setq char (match-string 0)))
14333e31
KH
564 (if (or (memq charset '(unknown
565 eight-bit-control eight-bit-graphic))
251d4f4b 566 (not (or (eq excludes t) (memq charset excludes))))
51ed58ea
KH
567 (let ((slot (assq charset chars)))
568 (if slot
569 (if (not (memq char (nthcdr 2 slot)))
570 (let ((count (nth 1 slot)))
571 (setcar (cdr slot) (1+ count))
572 (if (or (not maxcount) (< count maxcount))
573 (nconc slot (list char)))))
574 (setq chars (cons (list charset 1 char) chars)))))
575 (setq idx (1+ idx))))
576 (save-excursion
577 (goto-char from)
578 (while (re-search-forward "[^\000-\177]" to t)
579 (setq char (preceding-char)
580 charset (char-charset char))
251d4f4b
KH
581 (if (eq charset 'unknown)
582 (setq char (match-string 0)))
14333e31 583 (if (or (memq charset '(unknown eight-bit-control eight-bit-graphic))
251d4f4b 584 (not (or (eq excludes t) (memq charset excludes))))
51ed58ea
KH
585 (let ((slot (assq charset chars)))
586 (if slot
251d4f4b 587 (if (not (member char (nthcdr 2 slot)))
51ed58ea
KH
588 (let ((count (nth 1 slot)))
589 (setcar (cdr slot) (1+ count))
590 (if (or (not maxcount) (< count maxcount))
591 (nconc slot (list char)))))
592 (setq chars (cons (list charset 1 char) chars))))))))
593 (nreverse chars)))
594
738746ba
KH
595
596(defun search-unencodable-char (coding-system)
597 "Search forward from point for a character that is not encodable.
598It asks which coding system to check.
599If such a character is found, set point after that character.
600Otherwise, don't move point.
601
602When called from a program, the value is a position of the found character,
603or nil if all characters are encodable."
604 (interactive
605 (list (let ((default (or buffer-file-coding-system 'us-ascii)))
606 (read-coding-system
607 (format "Coding-system (default, %s): " default)
608 default))))
609 (let ((pos (unencodable-char-position (point) (point-max) coding-system)))
610 (if pos
611 (goto-char (1+ pos))
612 (message "All following characters are encodable by %s" coding-system))
613 pos))
614
615
c83c4f60
RS
616(defvar last-coding-system-specified nil
617 "Most recent coding system explicitly specified by the user when asked.
618This variable is set whenever Emacs asks the user which coding system
619to use in order to write a file. If you set it to nil explicitly,
620then call `write-region', then afterward this variable will be non-nil
621only if the user was explicitly asked and specified a coding system.")
622
b5edd1d1
KH
623(defvar select-safe-coding-system-accept-default-p nil
624 "If non-nil, a function to control the behaviour of coding system selection.
625The meaning is the same as the argument ACCEPT-DEFAULT-P of the
626function `select-safe-coding-system' (which see). This variable
627overrides that argument.")
628
9ee5b744
SM
629(defun select-safe-coding-system-interactively (from to codings unsafe
630 &optional rejected default)
631 "Select interactively a coding system for the region FROM ... TO.
632FROM can be a string, as in `write-region'.
633CODINGS is the list of base coding systems known to be safe for this region,
634 typically obtained with `find-coding-systems-region'.
635UNSAFE is a list of coding systems known to be unsafe for this region.
636REJECTED is a list of coding systems which were safe but for some reason
637 were not recommended in the particular context.
638DEFAULT is the coding system to use by default in the query."
639 ;; At first, if some defaults are unsafe, record at most 11
640 ;; problematic characters and their positions for them by turning
641 ;; (CODING ...)
642 ;; into
643 ;; ((CODING (POS . CHAR) (POS . CHAR) ...) ...)
644 (if unsafe
645 (setq unsafe
646 (mapcar #'(lambda (coding)
647 (cons coding
648 (if (stringp from)
649 (mapcar #'(lambda (pos)
650 (cons pos (aref from pos)))
651 (unencodable-char-position
652 0 (length from) coding
653 11 from))
654 (mapcar #'(lambda (pos)
655 (cons pos (char-after pos)))
656 (unencodable-char-position
657 from to coding 11)))))
658 unsafe)))
659
660 ;; Change each safe coding system to the corresponding
661 ;; mime-charset name if it is also a coding system. Such a name
662 ;; is more friendly to users.
663 (let ((l codings)
664 mime-charset)
665 (while l
666 (setq mime-charset (coding-system-get (car l) 'mime-charset))
667 (if (and mime-charset (coding-system-p mime-charset))
668 (setcar l mime-charset))
669 (setq l (cdr l))))
670
671 ;; Don't offer variations with locking shift, which you
672 ;; basically never want.
673 (let (l)
674 (dolist (elt codings (setq codings (nreverse l)))
675 (unless (or (eq 'coding-category-iso-7-else
676 (coding-system-category elt))
677 (eq 'coding-category-iso-8-else
678 (coding-system-category elt)))
679 (push elt l))))
680
681 ;; Remove raw-text, emacs-mule and no-conversion unless nothing
682 ;; else is available.
683 (setq codings
684 (or (delq 'raw-text
685 (delq 'emacs-mule
686 (delq 'no-conversion codings)))
687 '(raw-text emacs-mule no-conversion)))
688
689 (let ((window-configuration (current-window-configuration))
690 (bufname (buffer-name))
691 coding-system)
692 (save-excursion
693 ;; If some defaults are unsafe, make sure the offending
694 ;; buffer is displayed.
695 (when (and unsafe (not (stringp from)))
696 (pop-to-buffer bufname)
697 (goto-char (apply 'min (mapcar #'(lambda (x) (car (cadr x)))
698 unsafe))))
699 ;; Then ask users to select one from CODINGS while showing
700 ;; the reason why none of the defaults are not used.
701 (with-output-to-temp-buffer "*Warning*"
702 (with-current-buffer standard-output
703 (if (and (null rejected) (null unsafe))
704 (insert "No default coding systems to try for "
705 (if (stringp from)
706 (format "string \"%s\"." from)
707 (format "buffer `%s'." bufname)))
708 (insert
709 "These default coding systems were tried to encode"
710 (if (stringp from)
711 (concat " \"" (if (> (length from) 10)
712 (concat (substring from 0 10) "...\"")
713 (concat from "\"")))
714 (format " text\nin the buffer `%s'" bufname))
715 ":\n")
716 (let ((pos (point))
717 (fill-prefix " "))
718 (dolist (x (append rejected unsafe))
719 (princ " ") (princ (car x)))
720 (insert "\n")
721 (fill-region-as-paragraph pos (point)))
722 (when rejected
723 (insert "These safely encodes the target text,
724but it is not recommended for encoding text in this context,
725e.g., for sending an email message.\n ")
726 (dolist (x rejected)
727 (princ " ") (princ x))
728 (insert "\n"))
729 (when unsafe
730 (insert (if rejected "And the others"
731 "However, each of them")
732 " encountered these problematic characters:\n")
733 (dolist (coding unsafe)
734 (insert (format " %s:" (car coding)))
735 (let ((i 0)
736 (func1
737 #'(lambda (bufname pos)
738 (when (buffer-live-p (get-buffer bufname))
739 (pop-to-buffer bufname)
740 (goto-char pos))))
741 (func2
742 #'(lambda (bufname pos coding)
743 (when (buffer-live-p (get-buffer bufname))
744 (pop-to-buffer bufname)
745 (if (< (point) pos)
746 (goto-char pos)
747 (forward-char 1)
748 (search-unencodable-char coding)
749 (forward-char -1))))))
750 (dolist (elt (cdr coding))
751 (insert " ")
752 (if (stringp from)
753 (insert (if (< i 10) (cdr elt) "..."))
754 (if (< i 10)
755 (insert-text-button
756 (cdr elt)
757 :type 'help-xref
758 'help-echo
759 "mouse-2, RET: jump to this character"
760 'help-function func1
761 'help-args (list bufname (car elt)))
762 (insert-text-button
763 "..."
764 :type 'help-xref
765 'help-echo
766 "mouse-2, RET: next unencodable character"
767 'help-function func2
768 'help-args (list bufname (car elt)
769 (car coding)))))
770 (setq i (1+ i))))
771 (insert "\n"))
772 (insert "\
773The first problematic character is at point in the displayed buffer,\n"
774 (substitute-command-keys "\
775and \\[universal-argument] \\[what-cursor-position] will give information about it.\n"))))
776 (insert "\nSelect \
777one of the following safe coding systems, or edit the buffer:\n")
778 (let ((pos (point))
779 (fill-prefix " "))
780 (dolist (x codings)
781 (princ " ") (princ x))
782 (insert "\n")
783 (fill-region-as-paragraph pos (point)))
784 (insert "Or specify any other coding system
785at the risk of losing the problematic characters.\n")))
786
787 ;; Read a coding system.
788 (setq coding-system
789 (read-coding-system
790 (format "Select coding system (default %s): " default)
791 default))
792 (setq last-coding-system-specified coding-system))
793
794 (kill-buffer "*Warning*")
795 (set-window-configuration window-configuration)
796 coding-system))
797
b5edd1d1 798(defun select-safe-coding-system (from to &optional default-coding-system
efa2c6d7 799 accept-default-p file)
d5266ddf
KH
800 "Ask a user to select a safe coding system from candidates.
801The candidates of coding systems which can safely encode a text
b5edd1d1
KH
802between FROM and TO are shown in a popup window. Among them, the most
803proper one is suggested as the default.
804
a985cd2f
KH
805The list of `buffer-file-coding-system' of the current buffer,
806the `default-buffer-file-coding-system', and the
b5edd1d1
KH
807most preferred coding system (if it corresponds to a MIME charset) is
808treated as the default coding system list. Among them, the first one
d37ef0f6
DL
809that safely encodes the text is normally selected silently and
810returned without any user interaction. See also the command
811`prefer-coding-system'.
812
813However, the user is queried if the chosen coding system is
814inconsistent with what would be selected by `set-auto-coding' from
815coding cookies &c. if the contents of the region were read from a
816file. (That could lead to data corruption in a file subsequently
817re-visited and edited.)
54b226f7 818
b5edd1d1
KH
819Optional 3rd arg DEFAULT-CODING-SYSTEM specifies a coding system or a
820list of coding systems to be prepended to the default coding system
0e9ec609
KH
821list. However, if DEFAULT-CODING-SYSTEM is a list and the first
822element is t, the cdr part is used as the defualt coding system list,
a985cd2f
KH
823i.e. `buffer-file-coding-system', `default-buffer-file-coding-system',
824and the most preferred coding system are not used.
54b226f7 825
b5edd1d1
KH
826Optional 4th arg ACCEPT-DEFAULT-P, if non-nil, is a function to
827determine the acceptability of the silently selected coding system.
828It is called with that coding system, and should return nil if it
829should not be silently selected and thus user interaction is required.
830
efa2c6d7
RS
831Optional 5th arg FILE is the file name to use for this purpose.
832That is different from `buffer-file-name' when handling `write-region'
833\(for example).
834
b5edd1d1
KH
835The variable `select-safe-coding-system-accept-default-p', if
836non-nil, overrides ACCEPT-DEFAULT-P.
54b226f7
KH
837
838Kludgy feature: if FROM is a string, the string is the target text,
839and TO is ignored."
b5edd1d1
KH
840 (if (and default-coding-system
841 (not (listp default-coding-system)))
842 (setq default-coding-system (list default-coding-system)))
843
0e9ec609
KH
844 (let ((no-other-defaults nil))
845 (if (eq (car default-coding-system) t)
846 (setq no-other-defaults t
847 default-coding-system (cdr default-coding-system)))
848
849 ;; Change elements of the list to (coding . base-coding).
850 (setq default-coding-system
851 (mapcar (function (lambda (x) (cons x (coding-system-base x))))
852 default-coding-system))
853
a985cd2f
KH
854 ;; From now on, the list of defaults is reversed.
855 (setq default-coding-system (nreverse default-coding-system))
856
0e9ec609
KH
857 (unless no-other-defaults
858 ;; If buffer-file-coding-system is not nil nor undecided, append it
859 ;; to the defaults.
860 (if buffer-file-coding-system
861 (let ((base (coding-system-base buffer-file-coding-system)))
862 (or (eq base 'undecided)
0e9ec609 863 (rassq base default-coding-system)
a985cd2f
KH
864 (push (cons buffer-file-coding-system base)
865 default-coding-system))))
866
867 ;; If default-buffer-file-coding-system is not nil nor undecided,
868 ;; append it to the defaults.
869 (if default-buffer-file-coding-system
870 (let ((base (coding-system-base default-buffer-file-coding-system)))
871 (or (eq base 'undecided)
872 (rassq base default-coding-system)
873 (push (cons default-buffer-file-coding-system base)
874 default-coding-system))))
0e9ec609
KH
875
876 ;; If the most preferred coding system has the property mime-charset,
877 ;; append it to the defaults.
878 (let ((tail coding-category-list)
879 preferred base)
a985cd2f 880 (while (and tail (not (setq preferred (symbol-value (car tail)))))
0e9ec609
KH
881 (setq tail (cdr tail)))
882 (and (coding-system-p preferred)
883 (setq base (coding-system-base preferred))
884 (coding-system-get preferred 'mime-charset)
0e9ec609 885 (not (rassq base default-coding-system))
a985cd2f
KH
886 (push (cons preferred base)
887 default-coding-system)))))
b5edd1d1
KH
888
889 (if select-safe-coding-system-accept-default-p
890 (setq accept-default-p select-safe-coding-system-accept-default-p))
891
892 (let ((codings (find-coding-systems-region from to))
893 (coding-system nil)
c0d3ed97 894 safe rejected unsafe)
b5edd1d1
KH
895 (if (eq (car codings) 'undecided)
896 ;; Any coding system is ok.
897 (setq coding-system t)
c0d3ed97
KH
898 ;; Classify the defaults into safe, rejected, and unsafe.
899 (dolist (elt default-coding-system)
900 (if (memq (cdr elt) codings)
901 (if (and (functionp accept-default-p)
902 (not (funcall accept-default-p (cdr elt))))
903 (push (car elt) rejected)
904 (push (car elt) safe))
905 (push (car elt) unsafe)))
906 (if safe
a985cd2f 907 (setq coding-system (car safe))))
c0d3ed97 908
b5edd1d1 909 ;; If all the defaults failed, ask a user.
c0d3ed97 910 (when (not coding-system)
9ee5b744
SM
911 (setq coding-system (select-safe-coding-system-interactively
912 from to codings unsafe rejected (car codings))))
b5edd1d1
KH
913
914 (if (vectorp (coding-system-eol-type coding-system))
915 (let ((eol (coding-system-eol-type buffer-file-coding-system)))
916 (if (numberp eol)
917 (setq coding-system
918 (coding-system-change-eol-conversion coding-system eol)))))
919
920 (if (eq coding-system t)
921 (setq coding-system buffer-file-coding-system))
43afed8c
RS
922 ;; Check we're not inconsistent with what `coding:' spec &c would
923 ;; give when file is re-read.
9e7638c3
RS
924 ;; But don't do this if we explicitly ignored the cookie
925 ;; by using `find-file-literally'.
926 (unless (or (stringp from) find-file-literally)
efa2c6d7
RS
927 (let ((auto-cs (save-excursion
928 (save-restriction
929 (widen)
930 (narrow-to-region from to)
43afed8c 931 (goto-char (point-min))
efa2c6d7 932 (set-auto-coding (or file buffer-file-name "")
43afed8c 933 (buffer-size))))))
6aa69bd7
KH
934 ;; Merge coding-system and auto-cs as far as possible.
935 (if (not coding-system)
936 (setq coding-system auto-cs)
937 (if (not auto-cs)
938 (setq auto-cs coding-system)
939 (let ((eol-type-1 (coding-system-eol-type coding-system))
940 (eol-type-2 (coding-system-eol-type auto-cs)))
941 (if (eq (coding-system-base coding-system) 'undecided)
942 (setq coding-system (coding-system-change-text-conversion
943 coding-system auto-cs))
944 (if (eq (coding-system-base auto-cs) 'undecided)
945 (setq auto-cs (coding-system-change-text-conversion
946 auto-cs coding-system))))
947 (if (vectorp eol-type-1)
948 (or (vectorp eol-type-2)
949 (setq coding-system (coding-system-change-eol-conversion
950 coding-system eol-type-2)))
951 (if (vectorp eol-type-2)
952 (setq auto-cs (coding-system-change-eol-conversion
953 auto-cs eol-type-1)))))))
954
955 (if (and auto-cs
d37ef0f6
DL
956 ;; Don't barf if writing a compressed file, say.
957 ;; This check perhaps isn't ideal, but is probably
958 ;; the best thing to do.
959 (not (auto-coding-alist-lookup (or file buffer-file-name "")))
6aa69bd7 960 (not (coding-system-equal coding-system auto-cs)))
43afed8c
RS
961 (unless (yes-or-no-p
962 (format "Selected encoding %s disagrees with \
963%s specified by file contents. Really save (else edit coding cookies \
964and try again)? " coding-system auto-cs))
41bfe57a
RS
965 (error "Save aborted")))))
966 coding-system))
54b226f7
KH
967
968(setq select-safe-coding-system-function 'select-safe-coding-system)
969
46babb23
KH
970(defun select-message-coding-system ()
971 "Return a coding system to encode the outgoing message of the current buffer.
972It at first tries the first coding system found in these variables
973in this order:
974 (1) local value of `buffer-file-coding-system'
975 (2) value of `sendmail-coding-system'
b5edd1d1
KH
976 (3) value of `default-sendmail-coding-system'
977 (4) value of `default-buffer-file-coding-system'
46babb23
KH
978If the found coding system can't encode the current buffer,
979or none of them are bound to a coding system,
48e41165 980it asks the user to select a proper coding system."
46babb23 981 (let ((coding (or (and (local-variable-p 'buffer-file-coding-system)
b5edd1d1
KH
982 buffer-file-coding-system)
983 sendmail-coding-system
984 default-sendmail-coding-system
985 default-buffer-file-coding-system)))
46babb23 986 (if (eq coding 'no-conversion)
d37ef0f6 987 ;; We should never use no-conversion for outgoing mail.
46babb23
KH
988 (setq coding nil))
989 (if (fboundp select-safe-coding-system-function)
990 (funcall select-safe-coding-system-function
b5edd1d1
KH
991 (point-min) (point-max) coding
992 (function (lambda (x) (coding-system-get x 'mime-charset))))
46babb23 993 coding)))
4ed46869 994\f
03c35c83 995;;; Language support stuff.
4ed46869 996
4ed46869 997(defvar language-info-alist nil
2c395d56 998 "Alist of language environment definitions.
4ed46869
KH
999Each element looks like:
1000 (LANGUAGE-NAME . ((KEY . INFO) ...))
2c395d56
RS
1001where LANGUAGE-NAME is a string, the name of the language environment,
1002KEY is a symbol denoting the kind of information, and
1003INFO is the data associated with KEY.
1004Meaningful values for KEY include
1005
1006 documentation value is documentation of what this language environment
1007 is meant for, and how to use it.
1008 charset value is a list of the character sets used by this
1009 language environment.
d37ef0f6
DL
1010 sample-text value is an expression which is evalled to generate
1011 a line of text written using characters appropriate
1012 for this language environment.
2c395d56
RS
1013 setup-function value is a function to call to switch to this
1014 language environment.
1015 exit-function value is a function to call to leave this
1016 language environment.
1017 coding-system value is a list of coding systems that are good
1018 for saving text written in this language environment.
1019 This list serves as suggestions to the user;
1020 in effect, as a kind of documentation.
1021 coding-priority value is a list of coding systems for this language
1022 environment, in order of decreasing priority.
1023 This is used to set up the coding system priority
45d08cb2 1024 list when you switch to this language environment.
ddb5c041 1025 nonascii-translation
7624ebb9 1026 value is a translation table to be set in the
45d08cb2 1027 variable `nonascii-translation-table' in this
7624ebb9
KH
1028 language environment, or a character set from
1029 which `nonascii-insert-offset' is calculated.
ddb5c041
KH
1030 input-method value is a default input method for this language
1031 environment.
7624ebb9
KH
1032 features value is a list of features requested in this
1033 language environment.
0077e394
KH
1034 ctext-non-standard-encodings
1035 value is a list of non-standard encoding
1036 names used in extended segments of CTEXT.
1037 See the variable
1038 `ctext-non-standard-encodings' for more
1039 detail.
ddb5c041
KH
1040
1041The following keys take effect only when multibyte characters are
1042globally disabled, i.e. the value of `default-enable-multibyte-characters'
1043is nil.
1044
1045 unibyte-syntax value is a library name to load to set
e8dd0160 1046 unibyte 8-bit character syntaxes for this
ddb5c041
KH
1047 language environment.
1048
1049 unibyte-display value is a coding system to encode characters
1050 for the terminal. Characters in the range
1051 of 160 to 255 display not as octal escapes,
1052 but as non-ASCII characters in this language
1053 environment.")
2c395d56
RS
1054
1055(defun get-language-info (lang-env key)
1056 "Return information listed under KEY for language environment LANG-ENV.
1057KEY is a symbol denoting the kind of information.
1058For a list of useful values for KEY and their meanings,
1059see `language-info-alist'."
1060 (if (symbolp lang-env)
1061 (setq lang-env (symbol-name lang-env)))
f15466c5 1062 (let ((lang-slot (assoc-string lang-env language-info-alist t)))
4ed46869
KH
1063 (if lang-slot
1064 (cdr (assq key (cdr lang-slot))))))
1065
f08adf27 1066(defun set-language-info (lang-env key info)
2c395d56
RS
1067 "Modify part of the definition of language environment LANG-ENV.
1068Specifically, this stores the information INFO under KEY
1069in the definition of this language environment.
4ed46869 1070KEY is a symbol denoting the kind of information.
2c395d56 1071INFO is the value for that information.
281d03ec 1072
2c395d56 1073For a list of useful values for KEY and their meanings,
f08adf27 1074see `language-info-alist'."
2c395d56
RS
1075 (if (symbolp lang-env)
1076 (setq lang-env (symbol-name lang-env)))
4ed46869 1077 (let (lang-slot key-slot)
2c395d56 1078 (setq lang-slot (assoc lang-env language-info-alist))
4ed46869 1079 (if (null lang-slot) ; If no slot for the language, add it.
2c395d56 1080 (setq lang-slot (list lang-env)
4ed46869
KH
1081 language-info-alist (cons lang-slot language-info-alist)))
1082 (setq key-slot (assq key lang-slot))
1083 (if (null key-slot) ; If no slot for the key, add it.
1084 (progn
1085 (setq key-slot (list key))
1086 (setcdr lang-slot (cons key-slot (cdr lang-slot)))))
590dbcba 1087 (setcdr key-slot (purecopy info))
1169bd86 1088 ;; Update the custom-type of `current-language-environment'.
590dbcba 1089 (put 'current-language-environment 'custom-type
1169bd86
MR
1090 (cons 'choice (mapcar
1091 (lambda (lang)
756e055f
MR
1092 (list 'const lang))
1093 (sort (mapcar 'car language-info-alist) 'string<))))))
4ed46869 1094
2c395d56
RS
1095(defun set-language-info-alist (lang-env alist &optional parents)
1096 "Store ALIST as the definition of language environment LANG-ENV.
1097ALIST is an alist of KEY and INFO values. See the documentation of
98c6d6ed 1098`language-info-alist' for the meanings of KEY and INFO.
54b226f7 1099
2c395d56 1100Optional arg PARENTS is a list of parent menu names; it specifies
d37ef0f6 1101where to put this language environment in the
2c395d56
RS
1102Describe Language Environment and Set Language Environment menus.
1103For example, (\"European\") means to put this language environment
1104in the European submenu in each of those two menus."
1105 (if (symbolp lang-env)
1106 (setq lang-env (symbol-name lang-env)))
54b226f7
KH
1107 (let ((describe-map describe-language-environment-map)
1108 (setup-map setup-language-environment-map))
1109 (if parents
1110 (let ((l parents)
9deed82f 1111 map parent-symbol parent prompt)
54b226f7
KH
1112 (while l
1113 (if (symbolp (setq parent-symbol (car l)))
1114 (setq parent (symbol-name parent))
1115 (setq parent parent-symbol parent-symbol (intern parent)))
1116 (setq map (lookup-key describe-map (vector parent-symbol)))
9deed82f
EZ
1117 ;; This prompt string is for define-prefix-command, so
1118 ;; that the map it creates will be suitable for a menu.
1119 (or map (setq prompt (format "%s Environment" parent)))
54b226f7
KH
1120 (if (not map)
1121 (progn
1122 (setq map (intern (format "describe-%s-environment-map"
1123 (downcase parent))))
9deed82f 1124 (define-prefix-command map nil prompt)
54b226f7
KH
1125 (define-key-after describe-map (vector parent-symbol)
1126 (cons parent map) t)))
1127 (setq describe-map (symbol-value map))
1128 (setq map (lookup-key setup-map (vector parent-symbol)))
1129 (if (not map)
1130 (progn
1131 (setq map (intern (format "setup-%s-environment-map"
1132 (downcase parent))))
9deed82f 1133 (define-prefix-command map nil prompt)
54b226f7
KH
1134 (define-key-after setup-map (vector parent-symbol)
1135 (cons parent map) t)))
1136 (setq setup-map (symbol-value map))
1137 (setq l (cdr l)))))
f08adf27
RS
1138
1139 ;; Set up menu items for this language env.
7624ebb9 1140 (let ((doc (assq 'documentation alist)))
f08adf27
RS
1141 (when doc
1142 (define-key-after describe-map (vector (intern lang-env))
7624ebb9
KH
1143 (cons lang-env 'describe-specified-language-support) t)))
1144 (define-key-after setup-map (vector (intern lang-env))
1145 (cons lang-env 'setup-specified-language-environment) t)
f08adf27 1146
54b226f7 1147 (while alist
f08adf27 1148 (set-language-info lang-env (car (car alist)) (cdr (car alist)))
54b226f7 1149 (setq alist (cdr alist)))))
4ed46869 1150
ae302641 1151(defun read-language-name (key prompt &optional default)
2c395d56 1152 "Read a language environment name which has information for KEY.
ddb5c041 1153If KEY is nil, read any language environment.
2c395d56
RS
1154Prompt with PROMPT. DEFAULT is the default choice of language environment.
1155This returns a language environment name as a string."
4ed46869
KH
1156 (let* ((completion-ignore-case t)
1157 (name (completing-read prompt
1158 language-info-alist
ddb5c041 1159 (and key
ca429a25 1160 (function (lambda (elm) (and (listp elm) (assq key elm)))))
ae302641 1161 t nil nil default)))
13e82c04 1162 (if (and (> (length name) 0)
ddb5c041
KH
1163 (or (not key)
1164 (get-language-info name key)))
13e82c04 1165 name)))
4ed46869
KH
1166\f
1167;;; Multilingual input methods.
d37ef0f6 1168(defgroup leim nil
d0c40faa
KH
1169 "LEIM: Libraries of Emacs Input Methods."
1170 :group 'mule)
4ed46869 1171
d0b9c3ab
KH
1172(defconst leim-list-file-name "leim-list.el"
1173 "Name of LEIM list file.
1174This file contains a list of libraries of Emacs input methods (LEIM)
1175in the format of Lisp expression for registering each input method.
1176Emacs loads this file at startup time.")
1177
2e224638 1178(defvar leim-list-header (format
d37ef0f6 1179";;; %s -- list of LEIM (Library of Emacs Input Method) -*-coding: iso-2022-7bit;-*-
d0b9c3ab
KH
1180;;
1181;; This file contains a list of LEIM (Library of Emacs Input Method)
c654de1d
DL
1182;; methods in the same directory as this file. Loading this file
1183;; registers all the input methods in Emacs.
d0b9c3ab 1184;;
d33d5fbe 1185;; Each entry has the form:
d0b9c3ab
KH
1186;; (register-input-method
1187;; INPUT-METHOD LANGUAGE-NAME ACTIVATE-FUNC
1188;; TITLE DESCRIPTION
1189;; ARG ...)
c654de1d 1190;; See the function `register-input-method' for the meanings of the arguments.
d0b9c3ab
KH
1191;;
1192;; If this directory is included in load-path, Emacs automatically
1193;; loads this file at startup time.
1194
1195"
1196 leim-list-file-name)
1197 "Header to be inserted in LEIM list file.")
1198
e55e92ee 1199(defvar leim-list-entry-regexp "^(register-input-method"
d0b9c3ab
KH
1200 "Regexp matching head of each entry in LEIM list file.
1201See also the variable `leim-list-header'")
1202
1203(defvar update-leim-list-functions
1204 '(quail-update-leim-list-file)
1205 "List of functions to call to update LEIM list file.
1206Each function is called with one arg, LEIM directory name.")
1207
a337fe7f
RS
1208(defun update-leim-list-file (&rest dirs)
1209 "Update LEIM list file in directories DIRS."
d0b9c3ab
KH
1210 (let ((functions update-leim-list-functions))
1211 (while functions
a337fe7f 1212 (apply (car functions) dirs)
d0b9c3ab
KH
1213 (setq functions (cdr functions)))))
1214
4ed46869
KH
1215(defvar current-input-method nil
1216 "The current input method for multilingual text.
96db204a 1217If nil, that means no input method is activated now.")
4ed46869
KH
1218(make-variable-buffer-local 'current-input-method)
1219(put 'current-input-method 'permanent-local t)
1220
1221(defvar current-input-method-title nil
d0b9c3ab 1222 "Title string of the current input method shown in mode line.")
4ed46869
KH
1223(make-variable-buffer-local 'current-input-method-title)
1224(put 'current-input-method-title 'permanent-local t)
1225
b4fba33f 1226(defcustom default-input-method nil
8861c593 1227 "*Default input method for multilingual text (a string).
b4fba33f 1228This is the input method activated automatically by the command
9b10b5a3 1229`toggle-input-method' (\\[toggle-input-method])."
34703cb1 1230 :link '(custom-manual "(emacs)Input Methods")
8861c593 1231 :group 'mule
34703cb1
DL
1232 :type '(choice (const nil) (string
1233 :completion-ignore-case t
1234 :complete-function widget-string-complete
1235 :completion-alist input-method-alist
1236 :prompt-history input-method-history))
5806e8a6 1237 :set-after '(current-language-environment))
b4fba33f 1238
0f835e87
KH
1239(put 'input-method-function 'permanent-local t)
1240
723a427a
KH
1241(defvar input-method-history nil
1242 "History list for some commands that read input methods.")
1243(make-variable-buffer-local 'input-method-history)
1244(put 'input-method-history 'permanent-local t)
4ed46869
KH
1245
1246(defvar inactivate-current-input-method-function nil
1247 "Function to call for inactivating the current input method.
1248Every input method should set this to an appropriate value when activated.
f17ccaee
KH
1249This function is called with no argument.
1250
1251This function should never change the value of `current-input-method'.
1252It is set to nil by the function `inactivate-input-method'.")
4ed46869
KH
1253(make-variable-buffer-local 'inactivate-current-input-method-function)
1254(put 'inactivate-current-input-method-function 'permanent-local t)
1255
1256(defvar describe-current-input-method-function nil
1257 "Function to call for describing the current input method.
1258This function is called with no argument.")
1259(make-variable-buffer-local 'describe-current-input-method-function)
1260(put 'describe-current-input-method-function 'permanent-local t)
1261
d0b9c3ab 1262(defvar input-method-alist nil
2c395d56 1263 "Alist of input method names vs how to use them.
d0b9c3ab 1264Each element has the form:
2c395d56
RS
1265 (INPUT-METHOD LANGUAGE-ENV ACTIVATE-FUNC TITLE DESCRIPTION ARGS...)
1266See the function `register-input-method' for the meanings of the elements.")
1267
f08adf27 1268(defun register-input-method (input-method lang-env &rest args)
bf42aa15 1269 "Register INPUT-METHOD as an input method for language environment LANG-ENV.
f08adf27 1270INPUT-METHOD and LANG-ENV are symbols or strings.
d0b9c3ab 1271
d0b9c3ab 1272The remaining arguments are:
2c395d56
RS
1273 ACTIVATE-FUNC, TITLE, DESCRIPTION, and ARGS...
1274ACTIVATE-FUNC is a function to call to activate this method.
1275TITLE is a string to show in the mode line when this method is active.
1276DESCRIPTION is a string describing this method and what it is good for.
1277The ARGS, if any, are passed as arguments to ACTIVATE-FUNC.
205814ee
KH
1278All told, the arguments to ACTIVATE-FUNC are INPUT-METHOD and the ARGS.
1279
c654de1d
DL
1280This function is mainly used in the file \"leim-list.el\" which is
1281created at Emacs build time, registering all Quail input methods
0b6cadff 1282contained in the Emacs distribution.
205814ee 1283
0b6cadff 1284In case you want to register a new Quail input method by yourself, be
205814ee 1285careful to use the same input method title as given in the third
0b6cadff
DL
1286parameter of `quail-define-package'. (If the values are different, the
1287string specified in this function takes precedence.)
205814ee
KH
1288
1289The commands `describe-input-method' and `list-input-methods' need
0b6cadff
DL
1290these duplicated values to show some information about input methods
1291without loading the relevant Quail packages."
f08adf27
RS
1292 (if (symbolp lang-env)
1293 (setq lang-env (symbol-name lang-env)))
4ef06f75
KH
1294 (if (symbolp input-method)
1295 (setq input-method (symbol-name input-method)))
f08adf27 1296 (let ((info (cons lang-env args))
d0b9c3ab
KH
1297 (slot (assoc input-method input-method-alist)))
1298 (if slot
1299 (setcdr slot info)
1300 (setq slot (cons input-method info))
1301 (setq input-method-alist (cons slot input-method-alist)))))
1302
4d5ac029 1303(defun read-input-method-name (prompt &optional default inhibit-null)
d0b9c3ab 1304 "Read a name of input method from a minibuffer prompting with PROMPT.
4d5ac029 1305If DEFAULT is non-nil, use that as the default,
0b6cadff 1306and substitute it into PROMPT at the first `%s'.
4ef06f75
KH
1307If INHIBIT-NULL is non-nil, null input signals an error.
1308
1309The return value is a string."
4d5ac029
RS
1310 (if default
1311 (setq prompt (format prompt default)))
d0b9c3ab 1312 (let* ((completion-ignore-case t)
c54044ff
KH
1313 ;; As it is quite normal to change input method in the
1314 ;; minibuffer, we must enable it even if
1315 ;; enable-recursive-minibuffers is currently nil.
1316 (enable-recursive-minibuffers t)
723a427a
KH
1317 ;; This binding is necessary because input-method-history is
1318 ;; buffer local.
d0b9c3ab 1319 (input-method (completing-read prompt input-method-alist
87505a98
RS
1320 nil t nil 'input-method-history
1321 default)))
bf294e6e
KH
1322 (if (and input-method (symbolp input-method))
1323 (setq input-method (symbol-name input-method)))
d0b9c3ab
KH
1324 (if (> (length input-method) 0)
1325 input-method
1326 (if inhibit-null
43807b77 1327 (error "No valid input method is specified")))))
d0b9c3ab 1328
d0b9c3ab 1329(defun activate-input-method (input-method)
2c395d56
RS
1330 "Switch to input method INPUT-METHOD for the current buffer.
1331If some other input method is already active, turn it off first.
1332If INPUT-METHOD is nil, deactivate any current input method."
305a3cb6 1333 (if (and input-method (symbolp input-method))
4ef06f75 1334 (setq input-method (symbol-name input-method)))
723a427a
KH
1335 (if (and current-input-method
1336 (not (string= current-input-method input-method)))
305a3cb6 1337 (inactivate-input-method))
2c395d56 1338 (unless (or current-input-method (null input-method))
d0b9c3ab
KH
1339 (let ((slot (assoc input-method input-method-alist)))
1340 (if (null slot)
723a427a 1341 (error "Can't activate input method `%s'" input-method))
278dd6ac 1342 (setq current-input-method-title nil)
8efc03e1
KH
1343 (let ((func (nth 2 slot)))
1344 (if (functionp func)
1345 (apply (nth 2 slot) input-method (nthcdr 5 slot))
1346 (if (and (consp func) (symbolp (car func)) (symbolp (cdr func)))
1347 (progn
1348 (require (cdr func))
1349 (apply (car func) input-method (nthcdr 5 slot)))
1350 (error "Can't activate input method `%s'" input-method))))
d0b9c3ab 1351 (setq current-input-method input-method)
278dd6ac
KH
1352 (or (stringp current-input-method-title)
1353 (setq current-input-method-title (nth 3 slot)))
28885c0e
KH
1354 (unwind-protect
1355 (run-hooks 'input-method-activate-hook)
1356 (force-mode-line-update)))))
15b3e511 1357
15b3e511 1358(defun inactivate-input-method ()
f17ccaee 1359 "Turn off the current input method."
723a427a
KH
1360 (when current-input-method
1361 (if input-method-history
1362 (unless (string= current-input-method (car input-method-history))
1363 (setq input-method-history
1364 (cons current-input-method
1365 (delete current-input-method input-method-history))))
1366 (setq input-method-history (list current-input-method)))
1367 (unwind-protect
1368 (funcall inactivate-current-input-method-function)
15b3e511 1369 (unwind-protect
723a427a
KH
1370 (run-hooks 'input-method-inactivate-hook)
1371 (setq current-input-method nil
2b1ec973 1372 input-method-function nil
28885c0e
KH
1373 current-input-method-title nil)
1374 (force-mode-line-update)))))
4ed46869 1375
e893eae2 1376(defun set-input-method (input-method &optional interactive)
2c395d56 1377 "Select and activate input method INPUT-METHOD for the current buffer.
bc406911 1378This also sets the default input method to the one you specify.
402dbbd1
EZ
1379If INPUT-METHOD is nil, this function turns off the input method, and
1380also causes you to be prompted for a name of an input method the next
1381time you invoke \\[toggle-input-method].
e893eae2
RS
1382When called interactively, the optional arg INTERACTIVE is non-nil,
1383which marks the variable `default-input-method' as set for Custom buffers.
402dbbd1 1384
bc406911
EZ
1385To deactivate the input method interactively, use \\[toggle-input-method].
1386To deactivate it programmatically, use \\[inactivate-input-method]."
d0b9c3ab 1387 (interactive
723a427a 1388 (let* ((default (or (car input-method-history) default-input-method)))
42395763 1389 (list (read-input-method-name
87505a98 1390 (if default "Select input method (default %s): " "Select input method: ")
e893eae2
RS
1391 default t)
1392 t)))
d0b9c3ab 1393 (activate-input-method input-method)
f4990970 1394 (setq default-input-method input-method)
e893eae2 1395 (when interactive
f4990970
PA
1396 (customize-mark-as-set 'default-input-method))
1397 default-input-method)
4ed46869 1398
e893eae2 1399(defun toggle-input-method (&optional arg interactive)
f8ec20be
RS
1400 "Enable or disable multilingual text input method for the current buffer.
1401Only one input method can be enabled at any time in a given buffer.
1402
1403The normal action is to enable an input method if none was
1404enabled, and disable the current one otherwise. Which input method
1405to enable can be determined in various ways--either the one most
1406recently used, or the one specified by `default-input-method', or
1407as a last resort by reading the name of an input method in the
1408minibuffer.
1409
1410With a prefix argument, read an input method name with the minibuffer
1411and enable that one. The default is the most recent input method specified
e893eae2 1412\(not including the currently active input method, if any).
f8ec20be 1413
e893eae2
RS
1414When called interactively, the optional arg INTERACTIVE is non-nil,
1415which marks the variable `default-input-method' as set for Custom buffers."
1416
1417 (interactive "P\np")
7ddbb5bc
RS
1418 (if (and current-input-method (not arg))
1419 (inactivate-input-method)
1420 (let ((default (or (car input-method-history) default-input-method)))
1421 (if (and arg default (equal current-input-method default)
1422 (> (length input-method-history) 1))
1423 (setq default (nth 1 input-method-history)))
723a427a
KH
1424 (activate-input-method
1425 (if (or arg (not default))
7ddbb5bc
RS
1426 (progn
1427 (read-input-method-name
1428 (if default "Input method (default %s): " "Input method: " )
1429 default t))
723a427a 1430 default))
f4990970 1431 (unless default-input-method
d37ef0f6 1432 (prog1
f4990970 1433 (setq default-input-method current-input-method)
e893eae2 1434 (when interactive
f4990970 1435 (customize-mark-as-set 'default-input-method)))))))
d0b9c3ab
KH
1436
1437(defun describe-input-method (input-method)
2c395d56 1438 "Describe input method INPUT-METHOD."
d0b9c3ab
KH
1439 (interactive
1440 (list (read-input-method-name
1441 "Describe input method (default, current choice): ")))
78754934 1442 (if (and input-method (symbolp input-method))
4ef06f75 1443 (setq input-method (symbol-name input-method)))
43125c28
RS
1444 (help-setup-xref (list #'describe-input-method
1445 (or input-method current-input-method))
f80e2142
RS
1446 (interactive-p))
1447
d0b9c3ab
KH
1448 (if (null input-method)
1449 (describe-current-input-method)
464cc130
KH
1450 (let ((current current-input-method))
1451 (condition-case nil
1452 (progn
1453 (save-excursion
1454 (activate-input-method input-method)
1455 (describe-current-input-method))
1456 (activate-input-method current))
d37ef0f6 1457 (error
464cc130 1458 (activate-input-method current)
5f395df3
SM
1459 (help-setup-xref (list #'describe-input-method input-method)
1460 (interactive-p))
1461 (with-output-to-temp-buffer (help-buffer)
464cc130
KH
1462 (let ((elt (assoc input-method input-method-alist)))
1463 (princ (format
1464 "Input method: %s (`%s' in mode line) for %s\n %s\n"
1465 input-method (nth 3 elt) (nth 1 elt) (nth 4 elt))))))))))
d0b9c3ab
KH
1466
1467(defun describe-current-input-method ()
f80e2142
RS
1468 "Describe the input method currently in use.
1469This is a subroutine for `describe-input-method'."
4ed46869
KH
1470 (if current-input-method
1471 (if (and (symbolp describe-current-input-method-function)
1472 (fboundp describe-current-input-method-function))
1473 (funcall describe-current-input-method-function)
1474 (message "No way to describe the current input method `%s'"
f2979bdb 1475 current-input-method)
4ed46869 1476 (ding))
d0b9c3ab 1477 (error "No input method is activated now")))
4ed46869 1478
d3459641 1479(defun read-multilingual-string (prompt &optional initial-input input-method)
4ed46869
KH
1480 "Read a multilingual string from minibuffer, prompting with string PROMPT.
1481The input method selected last time is activated in minibuffer.
15b3e511 1482If optional second arg INITIAL-INPUT is non-nil, insert it in the minibuffer
d0b9c3ab
KH
1483initially.
1484Optional 3rd argument INPUT-METHOD specifies the input method
4ef06f75
KH
1485to be activated instead of the one selected last time. It is a symbol
1486or a string."
88d559ec
KH
1487 (setq input-method
1488 (or input-method
d3459641 1489 current-input-method
88d559ec
KH
1490 default-input-method
1491 (read-input-method-name "Input method: " nil t)))
3df60841 1492 (if (and input-method (symbolp input-method))
4ef06f75 1493 (setq input-method (symbol-name input-method)))
305a3cb6
KH
1494 (let ((prev-input-method current-input-method))
1495 (unwind-protect
1496 (progn
1497 (activate-input-method input-method)
1498 (read-string prompt initial-input nil nil t))
1499 (activate-input-method prev-input-method))))
4ed46869
KH
1500
1501;; Variables to control behavior of input methods. All input methods
1502;; should react to these variables.
1503
8efc03e1
KH
1504(defcustom input-method-verbose-flag 'default
1505 "*A flag to control extra guidance given by input methods.
1506The value should be nil, t, `complex-only', or `default'.
4ed46869 1507
cb29dfb6 1508The extra guidance is done by showing list of available keys in echo
8efc03e1
KH
1509area. When you use the input method in the minibuffer, the guidance
1510is shown at the bottom short window (split from the existing window).
c27c4ed8 1511
8efc03e1
KH
1512If the value is t, extra guidance is always given, if the value is
1513nil, extra guidance is always suppressed.
1514
1515If the value is `complex-only', only complex input methods such as
1516`chinese-py' and `japanese' give extra guidance.
1517
1518If the value is `default', complex input methods always give extra
1519guidance, but simple input methods give it only when you are not in
1520the minibuffer.
1521
1522See also the variable `input-method-highlight-flag'."
d37ef0f6
DL
1523 :type '(choice (const :tag "Always" t) (const :tag "Never" nil)
1524 (const complex-only) (const default))
42395763
RS
1525 :group 'mule)
1526
1527(defcustom input-method-highlight-flag t
1528 "*If this flag is non-nil, input methods highlight partially-entered text.
1529For instance, while you are in the middle of a Quail input method sequence,
1530the text inserted so far is temporarily underlined.
8efc03e1
KH
1531The underlining goes away when you finish or abort the input method sequence.
1532See also the variable `input-method-verbose-flag'."
42395763
RS
1533 :type 'boolean
1534 :group 'mule)
4ed46869
KH
1535
1536(defvar input-method-activate-hook nil
f17ccaee
KH
1537 "Normal hook run just after an input method is activated.
1538
1539The variable `current-input-method' keeps the input method name
1540just activated.")
4ed46869
KH
1541
1542(defvar input-method-inactivate-hook nil
f17ccaee
KH
1543 "Normal hook run just after an input method is inactivated.
1544
1545The variable `current-input-method' still keeps the input method name
4d0e6a11 1546just inactivated.")
4ed46869
KH
1547
1548(defvar input-method-after-insert-chunk-hook nil
1549 "Normal hook run just after an input method insert some chunk of text.")
1550
dccca980 1551(defvar input-method-exit-on-first-char nil
0b6cadff 1552 "This flag controls when an input method returns.
dccca980
KH
1553Usually, the input method does not return while there's a possibility
1554that it may find a different translation if a user types another key.
39e643e2
RS
1555But, it this flag is non-nil, the input method returns as soon as
1556the current key sequence gets long enough to have some valid translation.")
dccca980
KH
1557
1558(defvar input-method-use-echo-area nil
1559 "This flag controls how an input method shows an intermediate key sequence.
39e643e2
RS
1560Usually, the input method inserts the intermediate key sequence,
1561or candidate translations corresponding to the sequence,
1562at point in the current buffer.
1563But, if this flag is non-nil, it displays them in echo area instead.")
dccca980 1564
723a427a
KH
1565(defvar input-method-exit-on-invalid-key nil
1566 "This flag controls the behaviour of an input method on invalid key input.
1567Usually, when a user types a key which doesn't start any character
1568handled by the input method, the key is handled by turning off the
e8dd0160 1569input method temporarily. After that key, the input method is re-enabled.
723a427a
KH
1570But, if this flag is non-nil, the input method is never back on.")
1571
4ed46869 1572\f
8efc03e1
KH
1573(defvar set-language-environment-hook nil
1574 "Normal hook run after some language environment is set.
1575
1576When you set some hook function here, that effect usually should not
1577be inherited to another language environment. So, you had better set
1578another function in `exit-language-environment-hook' (which see) to
1579cancel the effect.")
1580
1581(defvar exit-language-environment-hook nil
1582 "Normal hook run after exiting from some language environment.
1583When this hook is run, the variable `current-language-environment'
1584is still bound to the language environment being exited.
1585
e8dd0160 1586This hook is mainly used for canceling the effect of
8efc03e1
KH
1587`set-language-environment-hook' (which-see).")
1588
b0648a00
RS
1589(put 'setup-specified-language-environment 'apropos-inhibit t)
1590
15b3e511 1591(defun setup-specified-language-environment ()
f08adf27 1592 "Switch to a specified language environment."
15b3e511 1593 (interactive)
f850d782 1594 (let (language-name)
15b3e511
KH
1595 (if (and (symbolp last-command-event)
1596 (or (not (eq last-command-event 'Default))
1597 (setq last-command-event 'English))
f850d782 1598 (setq language-name (symbol-name last-command-event)))
f4990970
PA
1599 (prog1
1600 (set-language-environment language-name)
1601 (customize-mark-as-set 'current-language-environment))
15b3e511 1602 (error "Bogus calling sequence"))))
4ed46869 1603
8861c593 1604(defcustom current-language-environment "English"
94d04df6 1605 "The last language environment specified with `set-language-environment'.
ebef6d93
KH
1606This variable should be set only with \\[customize], which is equivalent
1607to using the function `set-language-environment'."
94d04df6 1608 :link '(custom-manual "(emacs)Language Environments")
dff1aa24 1609 :set (lambda (symbol value) (set-language-environment value))
94d04df6 1610 :get (lambda (x)
f15466c5 1611 (or (car-safe (assoc-string
94d04df6
DL
1612 (if (symbolp current-language-environment)
1613 (symbol-name current-language-environment)
1614 current-language-environment)
f15466c5 1615 language-info-alist t))
94d04df6 1616 "English"))
990a4108
MR
1617 ;; custom type will be updated with `set-language-info'.
1618 :type (if language-info-alist
1619 (cons 'choice (mapcar
1620 (lambda (lang)
756e055f
MR
1621 (list 'const lang))
1622 (sort (mapcar 'car language-info-alist) 'string<)))
990a4108 1623 'string)
8861c593 1624 :initialize 'custom-initialize-default
eb9fc9e6 1625 :group 'mule)
f850d782 1626
ddb5c041
KH
1627(defun reset-language-environment ()
1628 "Reset multilingual environment of Emacs to the default status.
1629
1630The default status is as follows:
1631
d37ef0f6 1632 The default value of `buffer-file-coding-system' is nil.
ddb5c041
KH
1633 The default coding system for process I/O is nil.
1634 The default value for the command `set-terminal-coding-system' is nil.
1635 The default value for the command `set-keyboard-coding-system' is nil.
1636
1637 The order of priorities of coding categories and the coding system
1638 bound to each category are as follows
1639 coding category coding system
1640 --------------------------------------------------
ddb5c041 1641 coding-category-iso-8-1 iso-latin-1
370bd9cf
KH
1642 coding-category-iso-8-2 iso-latin-1
1643 coding-category-utf-8 mule-utf-8
b510f863
KH
1644 coding-category-utf-16-be mule-utf-16be-with-signature
1645 coding-category-utf-16-le mule-utf-16le-with-signature
ddb5c041
KH
1646 coding-category-iso-7-tight iso-2022-jp
1647 coding-category-iso-7 iso-2022-7bit
1648 coding-category-iso-7-else iso-2022-7bit-lock
1649 coding-category-iso-8-else iso-2022-8bit-ss2
1650 coding-category-emacs-mule emacs-mule
1651 coding-category-raw-text raw-text
1652 coding-category-sjis japanese-shift-jis
1653 coding-category-big5 chinese-big5
1654 coding-category-ccl nil
370bd9cf 1655 coding-category-binary no-conversion"
ddb5c041
KH
1656 (interactive)
1657 ;; This function formerly set default-enable-multibyte-characters to t,
1658 ;; but that is incorrect. It should not alter the unibyte/multibyte choice.
1659
1660 (setq coding-category-iso-7-tight 'iso-2022-jp
1661 coding-category-iso-7 'iso-2022-7bit
1662 coding-category-iso-8-1 'iso-latin-1
1663 coding-category-iso-8-2 'iso-latin-1
1664 coding-category-iso-7-else 'iso-2022-7bit-lock
1665 coding-category-iso-8-else 'iso-2022-8bit-ss2
1666 coding-category-emacs-mule 'emacs-mule
1667 coding-category-raw-text 'raw-text
1668 coding-category-sjis 'japanese-shift-jis
1669 coding-category-big5 'chinese-big5
b510f863
KH
1670 coding-category-utf-16-be 'mule-utf-16be-with-signature
1671 coding-category-utf-16-le 'mule-utf-16le-with-signature
b909cfc0
KH
1672 coding-category-utf-8 'mule-utf-8
1673 coding-category-ccl nil
1674 coding-category-binary 'no-conversion)
ddb5c041
KH
1675
1676 (set-coding-priority
1677 '(coding-category-iso-8-1
1678 coding-category-iso-8-2
370bd9cf
KH
1679 coding-category-utf-8
1680 coding-category-utf-16-be
1681 coding-category-utf-16-le
ddb5c041
KH
1682 coding-category-iso-7-tight
1683 coding-category-iso-7
1684 coding-category-iso-7-else
1685 coding-category-iso-8-else
d37ef0f6 1686 coding-category-emacs-mule
ddb5c041
KH
1687 coding-category-raw-text
1688 coding-category-sjis
1689 coding-category-big5
1690 coding-category-ccl
370bd9cf 1691 coding-category-binary))
ddb5c041 1692
91693d18
KH
1693 (update-coding-systems-internal)
1694
ddb5c041 1695 (set-default-coding-systems nil)
b5edd1d1 1696 (setq default-sendmail-coding-system 'iso-latin-1)
787caf99 1697 (setq default-file-name-coding-system 'iso-latin-1)
1d77e15a
JR
1698 ;; Preserve eol-type from existing default-process-coding-systems.
1699 ;; On non-unix-like systems in particular, these may have been set
1700 ;; carefully by the user, or by the startup code, to deal with the
1701 ;; users shell appropriately, so should not be altered by changing
1702 ;; language environment.
1703 (let ((output-coding
a099a2ff
JR
1704 ;; When bootstrapping, coding-systems are not defined yet, so
1705 ;; we need to catch the error from check-coding-system.
d37ef0f6 1706 (condition-case nil
a099a2ff
JR
1707 (coding-system-change-text-conversion
1708 (car default-process-coding-system) 'undecided)
1709 (coding-system-error 'undecided)))
1d77e15a 1710 (input-coding
a099a2ff
JR
1711 (condition-case nil
1712 (coding-system-change-text-conversion
1713 (cdr default-process-coding-system) 'iso-latin-1)
1714 (coding-system-error 'iso-latin-1))))
1d77e15a
JR
1715 (setq default-process-coding-system
1716 (cons output-coding input-coding)))
b5edd1d1 1717
ddb5c041
KH
1718 ;; Don't alter the terminal and keyboard coding systems here.
1719 ;; The terminal still supports the same coding system
1720 ;; that it supported a minute ago.
1b8dc791
SM
1721 ;; (set-terminal-coding-system-internal nil)
1722 ;; (set-keyboard-coding-system-internal nil)
ddb5c041
KH
1723
1724 (setq nonascii-translation-table nil
0077e394
KH
1725 nonascii-insert-offset 0)
1726
5d26489f
EZ
1727 ;; Don't invoke fontset-related functions if fontsets aren't
1728 ;; supported in this build of Emacs.
1729 (and (fboundp 'fontset-list)
1730 (set-overriding-fontspec-internal nil)))
ddb5c041 1731
0c47a7c8
KH
1732(reset-language-environment)
1733
40c81f74
PE
1734(defun set-display-table-and-terminal-coding-system (language-name)
1735 "Set up the display table and terminal coding system for LANGUAGE-NAME."
1736 (let ((coding (get-language-info language-name 'unibyte-display)))
1737 (if coding
1738 (standard-display-european-internal)
a67ae60e
EZ
1739 ;; The following 2 lines undo the 8-bit display that we set up
1740 ;; in standard-display-european-internal, which see. This is in
1741 ;; case the user has used standard-display-european earlier in
1742 ;; this session. (The MS-DOS port doesn't use that setup, so it
1743 ;; doesn't need to undo it.)
d73a7bb8
JPW
1744 (when standard-display-table
1745 (dotimes (i 128)
b221615b 1746 (aset standard-display-table (+ i 128) nil))))
40c81f74 1747 (or (eq window-system 'pc)
d73a7bb8 1748 (set-terminal-coding-system coding))))
40c81f74 1749
166246f7 1750(defun set-language-environment (language-name)
6c05d680
RS
1751 "Set up multi-lingual environment for using LANGUAGE-NAME.
1752This sets the coding system priority and the default input method
8861c593
RS
1753and sometimes other things. LANGUAGE-NAME should be a string
1754which is the name of a language environment. For example, \"Latin-1\"
1755specifies the character set for the major languages of Western Europe."
8efc03e1 1756 (interactive (list (read-language-name
ddb5c041 1757 nil
8efc03e1 1758 "Set language environment (default, English): ")))
4ef06f75
KH
1759 (if language-name
1760 (if (symbolp language-name)
1761 (setq language-name (symbol-name language-name)))
1762 (setq language-name "English"))
f15466c5 1763 (let ((slot (assoc-string language-name language-info-alist t)))
95498fd0 1764 (unless slot
f850d782 1765 (error "Language environment not defined: %S" language-name))
95498fd0 1766 (setq language-name (car slot)))
8efc03e1
KH
1767 (if current-language-environment
1768 (let ((func (get-language-info current-language-environment
1769 'exit-function)))
e63645c2 1770 (run-hooks 'exit-language-environment-hook)
5f395df3 1771 (if (functionp func) (funcall func))))
03c35c83
EZ
1772 (let ((default-eol-type (coding-system-eol-type
1773 default-buffer-file-coding-system)))
1774 (reset-language-environment)
ddb5c041 1775
e522f07b 1776 ;; The features might set up coding systems.
9c20a8d5
DL
1777 (let ((required-features (get-language-info language-name 'features)))
1778 (while required-features
1779 (require (car required-features))
1780 (setq required-features (cdr required-features))))
1781
03c35c83
EZ
1782 (setq current-language-environment language-name)
1783 (set-language-environment-coding-systems language-name default-eol-type))
ddb5c041
KH
1784 (let ((input-method (get-language-info language-name 'input-method)))
1785 (when input-method
1786 (setq default-input-method input-method)
1787 (if input-method-history
1788 (setq input-method-history
1789 (cons input-method
1790 (delete input-method input-method-history))))))
ec241f58
EZ
1791 (let ((nonascii (get-language-info language-name 'nonascii-translation))
1792 (dos-table
4e2ac2d9
EZ
1793 (if (eq window-system 'pc)
1794 (intern
d9c0a50e 1795 (format "cp%d-nonascii-translation-table" dos-codepage)))))
03c35c83
EZ
1796 (cond
1797 ((char-table-p nonascii)
1798 (setq nonascii-translation-table nonascii))
ec241f58 1799 ((and (eq window-system 'pc) (boundp dos-table))
03c35c83
EZ
1800 ;; DOS terminals' default is to use a special non-ASCII translation
1801 ;; table as appropriate for the installed codepage.
ec241f58 1802 (setq nonascii-translation-table (symbol-value dos-table)))
03c35c83
EZ
1803 ((charsetp nonascii)
1804 (setq nonascii-insert-offset (- (make-char nonascii) 128)))))
ddb5c041 1805
63283a8f 1806 ;; Unibyte setups if necessary.
ddb5c041 1807 (unless default-enable-multibyte-characters
63283a8f 1808 ;; Syntax and case table.
ddb5c041
KH
1809 (let ((syntax (get-language-info language-name 'unibyte-syntax)))
1810 (if syntax
1811 (let ((set-case-syntax-set-multibyte nil))
63283a8f
KH
1812 (load syntax nil t))
1813 ;; No information for syntax and case. Reset to the defaults.
1814 (let ((syntax-table (standard-syntax-table))
1815 (case-table (standard-case-table))
03c35c83 1816 (ch (if (eq window-system 'pc) 128 160)))
63283a8f
KH
1817 (while (< ch 256)
1818 (modify-syntax-entry ch " " syntax-table)
1819 (aset case-table ch ch)
1820 (setq ch (1+ ch)))
1821 (set-char-table-extra-slot case-table 0 nil)
1822 (set-char-table-extra-slot case-table 1 nil)
1823 (set-char-table-extra-slot case-table 2 nil))
1824 (set-standard-case-table (standard-case-table))
1825 (let ((list (buffer-list)))
1826 (while list
1827 (with-current-buffer (car list)
1828 (set-case-table (standard-case-table)))
1829 (setq list (cdr list))))))
40c81f74 1830 (set-display-table-and-terminal-coding-system language-name))
ddb5c041
KH
1831
1832 (let ((required-features (get-language-info language-name 'features)))
1833 (while required-features
1834 (require (car required-features))
1835 (setq required-features (cdr required-features))))
0077e394 1836
5d26489f
EZ
1837 ;; Don't invoke fontset-related functions if fontsets aren't
1838 ;; supported in this build of Emacs.
1839 (when (fboundp 'fontset-list)
1840 (let ((overriding-fontspec (get-language-info language-name
1841 'overriding-fontspec)))
1842 (if overriding-fontspec
1843 (set-overriding-fontspec-internal overriding-fontspec))))
0077e394 1844
ddb5c041 1845 (let ((func (get-language-info language-name 'setup-function)))
5f395df3 1846 (if (functionp func)
ddb5c041 1847 (funcall func)))
60d334c3
KH
1848 (if (and utf-translate-cjk-mode
1849 utf-translate-cjk-lang-env
1850 (not (eq utf-translate-cjk-lang-env language-name))
1851 (catch 'tag
1852 (dolist (charset (get-language-info language-name 'charset))
1853 (if (memq charset utf-translate-cjk-charsets)
1854 (throw 'tag t)))
1855 nil))
1856 (utf-translate-cjk-load-tables))
8efc03e1 1857 (run-hooks 'set-language-environment-hook)
f850d782 1858 (force-mode-line-update t))
4ed46869 1859
51a8fc1d
RS
1860(defun standard-display-european-internal ()
1861 ;; Actually set up direct output of non-ASCII characters.
03c35c83
EZ
1862 (standard-display-8bit (if (eq window-system 'pc) 128 160) 255)
1863 ;; Unibyte Emacs on MS-DOS wants to display all 8-bit characters with
1864 ;; the native font, and codes 160 and 146 stand for something very
1865 ;; different there.
1866 (or (and (eq window-system 'pc) (not default-enable-multibyte-characters))
1867 (progn
1868 ;; Make non-line-break space display as a plain space.
1869 ;; Most X fonts do the wrong thing for code 160.
1870 (aset standard-display-table 160 [32])
1492f7ac
DL
1871 ;; With luck, non-Latin-1 fonts are more recent and so don't
1872 ;; have this bug.
e64db02d 1873 (aset standard-display-table (make-char 'latin-iso8859-1 160) [32])
1492f7ac 1874 ;; Most Windows programs send out apostrophes as \222. Most X fonts
03c35c83 1875 ;; don't contain a character at that position. Map it to the ASCII
5f395df3
SM
1876 ;; apostrophe. [This is actually RIGHT SINGLE QUOTATION MARK,
1877 ;; U+2019, normally from the windows-1252 character set. XFree 4
1878 ;; fonts probably have the appropriate glyph at this position,
1879 ;; so they could use standard-display-8bit. It's better to use a
1880 ;; proper windows-1252 coding system. --fx]
e64db02d
DL
1881 (aset standard-display-table 146 [39])
1882 ;; XFree86 4 has changed most of the fonts from their designed
1883 ;; versions such that `' no longer appears as balanced quotes.
1884 ;; Assume it has iso10646 fonts installed, so we can display
1885 ;; balanced quotes.
1886 (when (and (eq window-system 'x)
1887 (string= "The XFree86 Project, Inc" (x-server-vendor))
1888 (> (aref (number-to-string (nth 2 (x-server-version))) 0)
1889 ?3))
d0d071d2
KH
1890 ;; We suppress these setting for the moment because the
1891 ;; above assumption is wrong.
9ee5b744
SM
1892 ;; (aset standard-display-table ?' [?’])
1893 ;; (aset standard-display-table ?` [?‘])
e64db02d
DL
1894 ;; The fonts don't have the relevant bug.
1895 (aset standard-display-table 160 nil)
1896 (aset standard-display-table (make-char 'latin-iso8859-1 160)
1897 nil)))))
03c35c83
EZ
1898
1899(defun set-language-environment-coding-systems (language-name
1900 &optional eol-type)
1901 "Do various coding system setups for language environment LANGUAGE-NAME.
1902
1903The optional arg EOL-TYPE specifies the eol-type of the default value
d37ef0f6 1904of `buffer-file-coding-system' set by this function."
54b226f7
KH
1905 (let* ((priority (get-language-info language-name 'coding-priority))
1906 (default-coding (car priority)))
1907 (if priority
1908 (let ((categories (mapcar 'coding-system-category priority)))
03c35c83
EZ
1909 (set-default-coding-systems
1910 (if (memq eol-type '(0 1 2 unix dos mac))
1911 (coding-system-change-eol-conversion default-coding eol-type)
1912 default-coding))
46babb23 1913 (setq default-sendmail-coding-system default-coding)
54b226f7
KH
1914 (set-coding-priority categories)
1915 (while priority
1916 (set (car categories) (car priority))
1917 (setq priority (cdr priority) categories (cdr categories)))
45d08cb2 1918 (update-coding-systems-internal)))))
54b226f7 1919
4ed46869 1920(defsubst princ-list (&rest args)
d37ef0f6 1921 "Print all arguments with `princ', then print \"\n\"."
4ed46869
KH
1922 (while args (princ (car args)) (setq args (cdr args)))
1923 (princ "\n"))
1924
b0648a00
RS
1925(put 'describe-specified-language-support 'apropos-inhibit t)
1926
34703cb1 1927;; Print language-specific information such as input methods,
13e82c04 1928;; charsets, and coding systems. This function is intended to be
48082651 1929;; called from the menu:
281d03ec 1930;; [menu-bar mule describe-language-environment LANGUAGE]
48082651
KH
1931;; and should not run it by `M-x describe-current-input-method-function'.
1932(defun describe-specified-language-support ()
96db204a 1933 "Describe how Emacs supports the specified language environment."
48082651 1934 (interactive)
281d03ec 1935 (let (language-name)
48082651 1936 (if (not (and (symbolp last-command-event)
cda74479
DL
1937 (or (not (eq last-command-event 'Default))
1938 (setq last-command-event 'English))
281d03ec 1939 (setq language-name (symbol-name last-command-event))))
48082651 1940 (error "Bogus calling sequence"))
281d03ec
RS
1941 (describe-language-environment language-name)))
1942
1943(defun describe-language-environment (language-name)
1944 "Describe how Emacs supports language environment LANGUAGE-NAME."
78754934
KH
1945 (interactive
1946 (list (read-language-name
1947 'documentation
8adfa8be 1948 "Describe language environment (default, current choice): ")))
f850d782
RS
1949 (if (null language-name)
1950 (setq language-name current-language-environment))
281d03ec
RS
1951 (if (or (null language-name)
1952 (null (get-language-info language-name 'documentation)))
1953 (error "No documentation for the specified language"))
4ef06f75
KH
1954 (if (symbolp language-name)
1955 (setq language-name (symbol-name language-name)))
ef5a4730
KH
1956 (dolist (feature (get-language-info language-name 'features))
1957 (require feature))
464cc130
KH
1958 (let ((doc (get-language-info language-name 'documentation))
1959 pos)
c3034e84
SM
1960 (help-setup-xref (list #'describe-language-environment language-name)
1961 (interactive-p))
1962 (with-output-to-temp-buffer (help-buffer)
464cc130
KH
1963 (save-excursion
1964 (set-buffer standard-output)
1965 (insert language-name " language environment\n\n")
1966 (if (stringp doc)
1967 (insert doc "\n\n"))
e036b0a6
KH
1968 (condition-case nil
1969 (let ((str (eval (get-language-info language-name 'sample-text))))
1970 (if (stringp str)
1971 (insert "Sample text:\n " str "\n\n")))
1972 (error nil))
464cc130
KH
1973 (let ((input-method (get-language-info language-name 'input-method))
1974 (l (copy-sequence input-method-alist)))
1975 (insert "Input methods")
1976 (when input-method
1977 (insert " (default, " input-method ")")
1978 (setq input-method (assoc input-method input-method-alist))
1979 (setq l (cons input-method (delete input-method l))))
1980 (insert ":\n")
48082651 1981 (while l
464cc130
KH
1982 (when (string= language-name (nth 1 (car l)))
1983 (insert " " (car (car l)))
1984 (search-backward (car (car l)))
467412aa 1985 (help-xref-button 0 'help-input-method (car (car l)))
464cc130 1986 (goto-char (point-max))
2fa7e202
KH
1987 (insert " (\""
1988 (if (stringp (nth 3 (car l)))
1989 (nth 3 (car l))
1990 (car (nth 3 (car l))))
1991 "\" in mode line)\n"))
464cc130
KH
1992 (setq l (cdr l)))
1993 (insert "\n"))
1994 (insert "Character sets:\n")
1995 (let ((l (get-language-info language-name 'charset)))
1996 (if (null l)
1997 (insert " nothing specific to " language-name "\n")
1998 (while l
1999 (insert " " (symbol-name (car l)))
2000 (search-backward (symbol-name (car l)))
467412aa 2001 (help-xref-button 0 'help-character-set (car l))
464cc130
KH
2002 (goto-char (point-max))
2003 (insert ": " (charset-description (car l)) "\n")
2004 (setq l (cdr l)))))
2005 (insert "\n")
2006 (insert "Coding systems:\n")
2007 (let ((l (get-language-info language-name 'coding-system)))
2008 (if (null l)
2009 (insert " nothing specific to " language-name "\n")
2010 (while l
2011 (insert " " (symbol-name (car l)))
2012 (search-backward (symbol-name (car l)))
467412aa 2013 (help-xref-button 0 'help-coding-system (car l))
464cc130
KH
2014 (goto-char (point-max))
2015 (insert " (`"
2016 (coding-system-mnemonic (car l))
2017 "' in mode line):\n\t"
2018 (coding-system-doc-string (car l))
2019 "\n")
2020 (let ((aliases (coding-system-get (car l)
2021 'alias-coding-systems)))
2022 (when aliases
2023 (insert "\t(alias:")
2024 (while aliases
2025 (insert " " (symbol-name (car aliases)))
2026 (setq aliases (cdr aliases)))
2027 (insert ")\n")))
c3034e84 2028 (setq l (cdr l)))))))))
4ed46869 2029\f
40c81f74
PE
2030;;; Locales.
2031
0d7c5bb9
DL
2032(defvar locale-translation-file-name nil
2033 "File name for the system's file of locale-name aliases, or nil if none.")
40c81f74 2034
5f395df3
SM
2035;; The following definitions might as well be marked as constants and
2036;; purecopied, since they're normally used on startup, and probably
2037;; should reflect the facilities of the base Emacs.
2038(defconst locale-language-names
2039 (purecopy
2040 '(
40c81f74
PE
2041 ;; Locale names of the form LANGUAGE[_TERRITORY][.CODESET][@MODIFIER]
2042 ;; as specified in the Single Unix Spec, Version 2.
2043 ;; LANGUAGE is a language code taken from ISO 639:1988 (E/F)
2044 ;; with additions from ISO 639/RA Newsletter No.1/1989;
5f395df3
SM
2045 ;; see Internet RFC 2165 (1997-06) and
2046 ;; http://www.evertype.com/standards/iso639/iso639-en.html
2047 ;; TERRITORY is a country code taken from ISO 3166
2048 ;; http://www.din.de/gremien/nas/nabd/iso3166ma/codlstp1/en_listp1.html.
40c81f74 2049 ;; CODESET and MODIFIER are implementation-dependent.
5f395df3 2050
0363e09a
DL
2051 ;; jasonr comments: MS Windows uses three letter codes for
2052 ;; languages instead of the two letter ISO codes that POSIX
2053 ;; uses. In most cases the first two letters are the same, so
2054 ;; most of the regexps in locale-language-names work. Japanese
2055 ;; and Chinese are exceptions, which are listed in the
2056 ;; non-standard section at the bottom of locale-language-names.
2057
40c81f74
PE
2058 ; aa Afar
2059 ; ab Abkhazian
6ececc4d 2060 ("af" . "Latin-1") ; Afrikaans
40c81f74 2061 ("am" . "Ethiopic") ; Amharic
5f395df3 2062 ; ar Arabic glibc uses 8859-6
40c81f74
PE
2063 ; as Assamese
2064 ; ay Aymara
2065 ; az Azerbaijani
2066 ; ba Bashkir
e522f07b 2067 ("be" . "Belarusian") ; Belarusian [Byelorussian until early 1990s]
9c20a8d5 2068 ("bg" . "Bulgarian") ; Bulgarian
40c81f74
PE
2069 ; bh Bihari
2070 ; bi Bislama
2071 ; bn Bengali, Bangla
2072 ("bo" . "Tibetan")
2073 ("br" . "Latin-1") ; Breton
d37ef0f6 2074 ("bs" . "Latin-2") ; Bosnian
40c81f74
PE
2075 ("ca" . "Latin-1") ; Catalan
2076 ; co Corsican
2077 ("cs" . "Czech")
e76ef161 2078 ("cy" . "Welsh") ; Welsh [glibc uses Latin-8. Did this change?]
40c81f74
PE
2079 ("da" . "Latin-1") ; Danish
2080 ("de" . "German")
2081 ; dz Bhutani
2082 ("el" . "Greek")
6ececc4d 2083 ;; Users who specify "en" explicitly typically want Latin-1, not ASCII.
e11cf111
DL
2084 ;; That's actually what the GNU locales define, modulo things like
2085 ;; en_IN -- fx.
6ececc4d 2086 ("en" . "Latin-1") ; English
40c81f74 2087 ("eo" . "Latin-3") ; Esperanto
8c4b6822 2088 ("es" . "Spanish")
40c81f74
PE
2089 ("et" . "Latin-4") ; Estonian
2090 ("eu" . "Latin-1") ; Basque
d37ef0f6 2091 ; fa Persian glibc uses utf-8
40c81f74
PE
2092 ("fi" . "Latin-1") ; Finnish
2093 ; fj Fiji
2094 ("fo" . "Latin-1") ; Faroese
8d738cfd 2095 ("fr" . "French") ; French
40c81f74 2096 ("fy" . "Latin-1") ; Frisian
6ececc4d
PE
2097 ("ga" . "Latin-1") ; Irish Gaelic (new orthography)
2098 ("gd" . "Latin-1") ; Scots Gaelic
2099 ("gl" . "Latin-1") ; Galician
40c81f74
PE
2100 ; gn Guarani
2101 ; gu Gujarati
d37ef0f6 2102 ("gv" . "Latin-8") ; Manx Gaelic glibc uses 8859-1
40c81f74
PE
2103 ; ha Hausa
2104 ("he" . "Hebrew")
5f395df3 2105 ("hi" . "Devanagari") ; Hindi glibc uses utf-8
03cb0324 2106 ("hr" . "Croatian") ; Croatian
40c81f74
PE
2107 ("hu" . "Latin-2") ; Hungarian
2108 ; hy Armenian
2109 ; ia Interlingua
2110 ("id" . "Latin-1") ; Indonesian
2111 ; ie Interlingue
2112 ; ik Inupiak
2113 ("is" . "Latin-1") ; Icelandic
e76ef161 2114 ("it" . "Italian") ; Italian
40c81f74
PE
2115 ; iu Inuktitut
2116 ("ja" . "Japanese")
2117 ; jw Javanese
9c20a8d5 2118 ("ka" . "Georgian") ; Georgian
40c81f74 2119 ; kk Kazakh
6ececc4d 2120 ("kl" . "Latin-1") ; Greenlandic
40c81f74
PE
2121 ; km Cambodian
2122 ; kn Kannada
2123 ("ko" . "Korean")
2124 ; ks Kashmiri
2125 ; ku Kurdish
5f395df3 2126 ("kw" . "Latin-1") ; Cornish
40c81f74
PE
2127 ; ky Kirghiz
2128 ("la" . "Latin-1") ; Latin
5f395df3 2129 ("lb" . "Latin-1") ; Luxemburgish
40c81f74
PE
2130 ; ln Lingala
2131 ("lo" . "Lao") ; Laothian
9c20a8d5
DL
2132 ("lt" . "Lithuanian")
2133 ("lv" . "Latvian") ; Latvian, Lettish
40c81f74 2134 ; mg Malagasy
5f395df3 2135 ("mi" . "Latin-7") ; Maori
f1282c7f 2136 ("mk" . "Cyrillic-ISO") ; Macedonian
40c81f74
PE
2137 ; ml Malayalam
2138 ; mn Mongolian
2139 ; mo Moldavian
5f395df3
SM
2140 ("mr" . "Devanagari") ; Marathi glibc uses utf-8
2141 ("ms" . "Latin-1") ; Malay
40c81f74
PE
2142 ("mt" . "Latin-3") ; Maltese
2143 ; my Burmese
2144 ; na Nauru
2145 ("ne" . "Devanagari") ; Nepali
8c4b6822 2146 ("nl" . "Dutch")
40c81f74 2147 ("no" . "Latin-1") ; Norwegian
5f395df3 2148 ("oc" . "Latin-1") ; Occitan
40c81f74
PE
2149 ; om (Afan) Oromo
2150 ; or Oriya
2151 ; pa Punjabi
2152 ("pl" . "Latin-2") ; Polish
2153 ; ps Pashto, Pushto
2154 ("pt" . "Latin-1") ; Portuguese
2155 ; qu Quechua
6ececc4d 2156 ("rm" . "Latin-1") ; Rhaeto-Romanic
40c81f74
PE
2157 ; rn Kirundi
2158 ("ro" . "Romanian")
e76ef161 2159 ("ru.*[_.]koi8" . "Russian")
f1282c7f 2160 ("ru" . "Cyrillic-ISO") ; Russian
40c81f74
PE
2161 ; rw Kinyarwanda
2162 ("sa" . "Devanagari") ; Sanskrit
2163 ; sd Sindhi
5f395df3 2164 ; se Northern Sami
40c81f74
PE
2165 ; sg Sangho
2166 ("sh" . "Latin-2") ; Serbo-Croatian
2167 ; si Sinhalese
2168 ("sk" . "Slovak")
2169 ("sl" . "Slovenian")
2170 ; sm Samoan
2171 ; sn Shona
2172 ; so Somali
6ececc4d 2173 ("sq" . "Latin-1") ; Albanian
40c81f74 2174 ("sr" . "Latin-2") ; Serbian (Latin alphabet)
e522f07b 2175 ("sr_YU@cyrillic" . "Cyrillic-ISO") ; per glibc
40c81f74
PE
2176 ; ss Siswati
2177 ; st Sesotho
2178 ; su Sundanese
03cb0324 2179 ("sv" . "Swedish") ; Swedish
40c81f74 2180 ("sw" . "Latin-1") ; Swahili
5f395df3
SM
2181 ; ta Tamil glibc uses utf-8
2182 ; te Telugu glibc uses utf-8
e522f07b 2183 ("tg" . "Tajik")
40c81f74
PE
2184 ("th" . "Thai")
2185 ; ti Tigrinya
2186 ; tk Turkmen
6ececc4d 2187 ("tl" . "Latin-1") ; Tagalog
40c81f74
PE
2188 ; tn Setswana
2189 ; to Tonga
e76ef161 2190 ("tr" . "Turkish")
40c81f74
PE
2191 ; ts Tsonga
2192 ; tt Tatar
2193 ; tw Twi
2194 ; ug Uighur
9c20a8d5 2195 ("uk" . "Ukrainian") ; Ukrainian
5f395df3
SM
2196 ; ur Urdu glibc uses utf-8
2197 ("uz" . "Latin-1") ; Uzbek
2198 ("vi" . "Vietnamese") ; glibc uses utf-8
40c81f74 2199 ; vo Volapuk
d37ef0f6 2200 ("wa" . "Latin-1") ; Walloon
40c81f74
PE
2201 ; wo Wolof
2202 ; xh Xhosa
9c20a8d5 2203 ("yi" . "Windows-1255") ; Yiddish
40c81f74
PE
2204 ; yo Yoruba
2205 ; za Zhuang
5f395df3
SM
2206
2207 ; glibc:
2208 ; zh_CN.GB18030/GB18030 \
2209 ; zh_CN.GBK/GBK \
2210 ; zh_HK/BIG5-HKSCS \
5f395df3 2211
40c81f74
PE
2212 ("zh.*[._]big5" . "Chinese-BIG5")
2213 ("zh.*[._]gbk" . nil) ; Solaris 2.7; has gbk-0 as well as GB 2312.1980-0
e76ef161
DL
2214 ("zh_tw" . "Chinese-CNS") ; glibc uses big5
2215 ("zh_tw[._]euc-tw" . "Chinese-EUC-TW")
40c81f74
PE
2216 ("zh" . "Chinese-GB")
2217 ; zu Zulu
2218
2219 ;; ISO standard locales
2220 ("c$" . "ASCII")
2221 ("posix$" . "ASCII")
2222
40c81f74
PE
2223 ;; The "IPA" Emacs language environment does not correspond
2224 ;; to any ISO 639 code, so let it stand for itself.
2225 ("ipa$" . "IPA")
2226
2227 ;; Nonstandard or obsolete language codes
2228 ("cz" . "Czech") ; e.g. Solaris 2.6
2229 ("ee" . "Latin-4") ; Estonian, e.g. X11R6.4
2230 ("iw" . "Hebrew") ; e.g. X11R6.4
f1282c7f 2231 ("sp" . "Cyrillic-ISO") ; Serbian (Cyrillic alphabet), e.g. X11R6.4
40c81f74 2232 ("su" . "Latin-1") ; Finnish, e.g. Solaris 2.6
2e86ceaa 2233 ("jp" . "Japanese") ; e.g. MS Windows
86d9e628
JR
2234 ("chs" . "Chinese-GB") ; MS Windows Chinese Simplified
2235 ("cht" . "Chinese-BIG5") ; MS Windows Chinese Traditional
367ca50f 2236 ))
40c81f74 2237 "List of pairs of locale regexps and language names.
6ececc4d
PE
2238The first element whose locale regexp matches the start of a downcased locale
2239specifies the language name corresponding to that locale.
40c81f74
PE
2240If the language name is nil, there is no corresponding language environment.")
2241
5f395df3
SM
2242(defconst locale-charset-language-names
2243 (purecopy
2244 '((".*8859[-_]?1\\>" . "Latin-1")
2245 (".*8859[-_]?2\\>" . "Latin-2")
2246 (".*8859[-_]?3\\>" . "Latin-3")
2247 (".*8859[-_]?4\\>" . "Latin-4")
2248 (".*8859[-_]?9\\>" . "Latin-5")
2249 (".*8859[-_]?14\\>" . "Latin-8")
2250 (".*8859[-_]?15\\>" . "Latin-9")
34703cb1 2251 (".*utf\\(?:-?8\\)?\\>" . "UTF-8")
e522f07b
DL
2252 ;; utf-8@euro exists, so put this last. (@euro really specifies
2253 ;; the currency, rather than the charset.)
2254 (".*@euro\\>" . "Latin-9")))
6ececc4d
PE
2255 "List of pairs of locale regexps and charset language names.
2256The first element whose locale regexp matches the start of a downcased locale
34703cb1 2257specifies the language name whose charset corresponds to that locale.
6ececc4d
PE
2258This language name is used if its charsets disagree with the charsets of
2259the language name that would otherwise be used for this locale.")
2260
5f395df3
SM
2261(defconst locale-preferred-coding-systems
2262 (purecopy
2263 '(("ja.*[._]euc" . japanese-iso-8bit)
2264 ("ja.*[._]jis7" . iso-2022-jp)
2265 ("ja.*[._]pck" . japanese-shift-jis)
2266 ("ja.*[._]sjis" . japanese-shift-jis)
69210880 2267 ("jpn" . japanese-shift-jis) ; MS-Windows uses this.
9c20a8d5 2268 (".*[._]utf" . utf-8)))
6ececc4d
PE
2269 "List of pairs of locale regexps and preferred coding systems.
2270The first element whose locale regexp matches the start of a downcased locale
2271specifies the coding system to prefer when using that locale.")
40c81f74
PE
2272
2273(defun locale-name-match (key alist)
2274 "Search for KEY in ALIST, which should be a list of regexp-value pairs.
2275Return the value corresponding to the first regexp that matches the
2276start of KEY, or nil if there is no match."
2277 (let (element)
2278 (while (and alist (not element))
ddf6b836 2279 (if (string-match (concat "\\`\\(?:" (car (car alist)) "\\)") key)
40c81f74
PE
2280 (setq element (car alist)))
2281 (setq alist (cdr alist)))
2282 (cdr element)))
2283
34703cb1
DL
2284(defun locale-charset-match-p (charset1 charset2)
2285 "Whether charset names (strings) CHARSET1 and CHARSET2 are equivalent.
2286Matching is done ignoring case and any hyphens and underscores in the
2287names. E.g. `ISO_8859-1' and `iso88591' both match `iso-8859-1'."
2288 (setq charset1 (replace-regexp-in-string "[-_]" "" charset1))
2289 (setq charset2 (replace-regexp-in-string "[-_]" "" charset2))
2290 (eq t (compare-strings charset1 nil nil charset2 nil nil t)))
2291
2292(defvar locale-charset-alist nil
2293 "Coding system alist keyed on locale-style charset name.
2294Used by `locale-charset-to-coding-system'.")
2295
2296(defun locale-charset-to-coding-system (charset)
2297 "Find coding system corresponding to CHARSET.
2298CHARSET is any sort of non-Emacs charset name, such as might be used
2299in a locale codeset, or elsewhere. It is matched to a coding system
2300first by case-insensitive lookup in `locale-charset-alist'. Then
2301matches are looked for in the coding system list, treating case and
2302the characters `-' and `_' as insignificant. The coding system base
2303is returned. Thus, for instance, if charset \"ISO8859-2\",
2304`iso-latin-2' is returned."
f15466c5 2305 (or (car (assoc-string charset locale-charset-alist t))
34703cb1
DL
2306 (let ((cs coding-system-alist)
2307 c)
2308 (while (and (not c) cs)
2309 (if (locale-charset-match-p charset (caar cs))
2310 (setq c (intern (caar cs)))
2311 (pop cs)))
0363e09a 2312 (if c (coding-system-base c)))))
34703cb1
DL
2313
2314;; Fixme: This ought to deal with the territory part of the locale
2315;; too, for setting things such as calendar holidays, ps-print paper
2316;; size, spelling dictionary.
2317
758f07de 2318(defun set-locale-environment (&optional locale-name)
40c81f74 2319 "Set up multi-lingual environment for using LOCALE-NAME.
758f07de
RS
2320This sets the language environment, the coding system priority,
2321the default input method and sometimes other things.
2322
34703cb1
DL
2323LOCALE-NAME should be a string which is the name of a locale supported
2324by the system. Often it is of the form xx_XX.CODE, where xx is a
2325language, XX is a country, and CODE specifies a character set and
2326coding system. For example, the locale name \"ja_JP.EUC\" might name
2327a locale for Japanese in Japan using the `japanese-iso-8bit'
2328coding-system. The name may also have a modifier suffix, e.g. `@euro'
2329or `@cyrillic'.
40c81f74 2330
758f07de 2331If LOCALE-NAME is nil, its value is taken from the environment
d37ef0f6 2332variables LC_ALL, LC_CTYPE and LANG (the first one that is set).
40c81f74
PE
2333
2334The locale names supported by your system can typically be found in a
0812c1e8 2335directory named `/usr/share/locale' or `/usr/lib/locale'. LOCALE-NAME
758f07de 2336will be translated according to the table specified by
0812c1e8
DL
2337`locale-translation-file-name'.
2338
2339See also `locale-charset-language-names', `locale-language-names',
2340`locale-preferred-coding-systems' and `locale-coding-system'."
758f07de 2341 (interactive "sSet environment for locale: ")
a1506d29 2342
0d7c5bb9
DL
2343 ;; Do this at runtime for the sake of binaries possibly transported
2344 ;; to a system without X.
2345 (setq locale-translation-file-name
2346 (let ((files
2347 '("/usr/lib/X11/locale/locale.alias" ; e.g. X11R6.4
34703cb1 2348 "/usr/X11R6/lib/X11/locale/locale.alias" ; XFree86, e.g. RedHat 4.2
0d7c5bb9
DL
2349 "/usr/openwin/lib/locale/locale.alias" ; e.g. Solaris 2.6
2350 ;;
2351 ;; The following name appears after the X-related names above,
2352 ;; since the X-related names are what X actually uses.
2353 "/usr/share/locale/locale.alias" ; GNU/Linux sans X
2354 )))
2355 (while (and files (not (file-exists-p (car files))))
2356 (setq files (cdr files)))
2357 (car files)))
2358
7008ccac
GM
2359 (let ((locale locale-name))
2360
2361 (unless locale
2362 ;; Use the first of these three environment variables
2363 ;; that has a nonempty value.
2364 (let ((vars '("LC_ALL" "LC_CTYPE" "LANG")))
db6bd804
DL
2365 (while (and vars
2366 (= 0 (length locale))) ; nil or empty string
2367 (setq locale (getenv (pop vars))))))
7008ccac
GM
2368
2369 (when locale
2370
2371 ;; Translate "swedish" into "sv_SE.ISO8859-1", and so on,
2372 ;; using the translation file that many systems have.
2373 (when locale-translation-file-name
2374 (with-temp-buffer
2375 (insert-file-contents locale-translation-file-name)
2376 (when (re-search-forward
2377 (concat "^" (regexp-quote locale) ":?[ \t]+") nil t)
2378 (setq locale (buffer-substring (point) (line-end-position))))))
2379
2380 ;; Leave the system locales alone if the caller did not specify
2381 ;; an explicit locale name, as their defaults are set from
2382 ;; LC_MESSAGES and LC_TIME, not LC_CTYPE, and the user might not
2383 ;; want to set them to the same value as LC_CTYPE.
2384 (when locale-name
2385 (setq system-messages-locale locale)
2386 (setq system-time-locale locale))
2387
2388 (setq locale (downcase locale))
2389
2390 (let ((language-name
2391 (locale-name-match locale locale-language-names))
2392 (charset-language-name
2393 (locale-name-match locale locale-charset-language-names))
2394 (coding-system
34703cb1
DL
2395 (or (locale-name-match locale locale-preferred-coding-systems)
2396 (when locale
2397 (if (string-match "\\.\\([^@]+\\)" locale)
2398 (locale-charset-to-coding-system
2399 (match-string 1 locale)))))))
7008ccac 2400
5f395df3 2401 ;; Give preference to charset-language-name over language-name.
7008ccac
GM
2402 (if (and charset-language-name
2403 (not
2404 (equal (get-language-info language-name 'charset)
2405 (get-language-info charset-language-name 'charset))))
2406 (setq language-name charset-language-name))
2407
2408 (when language-name
2409
2410 ;; Set up for this character set. This is now the right way
2411 ;; to do it for both unibyte and multibyte modes.
2412 (set-language-environment language-name)
2413
2414 ;; If default-enable-multibyte-characters is nil,
2415 ;; we are using single-byte characters,
2416 ;; so the display table and terminal coding system are irrelevant.
2417 (when default-enable-multibyte-characters
2418 (set-display-table-and-terminal-coding-system language-name))
2419
166ce29f
DL
2420 ;; Set the `keyboard-coding-system' if appropriate (tty
2421 ;; only). At least X and MS Windows can generate
2422 ;; multilingual input.
2423 (unless window-system
2424 (let ((kcs (or coding-system
2425 (car (get-language-info language-name
2426 'coding-system)))))
2427 (if kcs (set-keyboard-coding-system kcs))))
ddf6b836 2428
7008ccac
GM
2429 (setq locale-coding-system
2430 (car (get-language-info language-name 'coding-priority))))
2431
2432 (when coding-system
2433 (prefer-coding-system coding-system)
e76ef161
DL
2434 (setq locale-coding-system coding-system))))
2435
893b49bb
JR
2436 ;; On Windows, override locale-coding-system, keyboard-coding-system,
2437 ;; selection-coding-system with system codepage.
5eb94383 2438 (when (boundp 'w32-ansi-code-page)
893b49bb
JR
2439 (let ((code-page-coding (intern (format "cp%d" w32-ansi-code-page))))
2440 (when (coding-system-p code-page-coding)
2441 (setq locale-coding-system code-page-coding)
2442 (set-selection-coding-system code-page-coding)
2443 (set-keyboard-coding-system code-page-coding)
2444 (set-terminal-coding-system code-page-coding))))
2445
e76ef161 2446 ;; Default to A4 paper if we're not in a C, POSIX or US locale.
3479c806 2447 ;; (See comments in Flocale_info.)
e76ef161 2448 (let ((locale locale)
3479c806 2449 (paper (locale-info 'paper)))
e76ef161
DL
2450 (if paper
2451 ;; This will always be null at the time of writing.
2452 (cond
2453 ((equal paper '(216 279))
2454 (setq ps-paper-type 'letter))
2455 ((equal paper '(210 297))
2456 (setq ps-paper-type 'a4)))
2457 (let ((vars '("LC_ALL" "LC_PAPER" "LANG")))
2458 (while (and vars (= 0 (length locale)))
2459 (setq locale (getenv (pop vars)))))
2460 (when locale
2461 ;; As of glibc 2.2.5, these are the only US Letter locales,
2462 ;; and the rest are A4.
2463 (setq ps-paper-type
2464 (or (locale-name-match locale '(("c$" . letter)
2465 ("posix$" . letter)
2466 (".._us" . letter)
2467 (".._pr" . letter)
893b49bb
JR
2468 (".._ca" . letter)
2469 ("enu$" . letter) ; Windows
2470 ("esu$" . letter)
2471 ("enc$" . letter)
2472 ("frc$" . letter)))
e76ef161
DL
2473 'a4))))))
2474 nil)
40c81f74 2475\f
4ed46869
KH
2476;;; Charset property
2477
251d4f4b 2478(defun get-charset-property (charset propname)
4ed46869
KH
2479 "Return the value of CHARSET's PROPNAME property.
2480This is the last value stored with
96db204a 2481 (put-charset-property CHARSET PROPNAME VALUE)."
251d4f4b
KH
2482 (and (not (eq charset 'composition))
2483 (plist-get (charset-plist charset) propname)))
4ed46869 2484
251d4f4b 2485(defun put-charset-property (charset propname value)
4ed46869
KH
2486 "Store CHARSETS's PROPNAME property with value VALUE.
2487It can be retrieved with `(get-charset-property CHARSET PROPNAME)'."
1300d43f
KH
2488 (or (eq charset 'composition)
2489 (set-charset-plist charset
2490 (plist-put (charset-plist charset) propname value))))
4ed46869
KH
2491
2492;;; Character code property
2493(put 'char-code-property-table 'char-table-extra-slots 0)
2494
2495(defvar char-code-property-table
2496 (make-char-table 'char-code-property-table)
2497 "Char-table containing a property list of each character code.
2498
2499See also the documentation of `get-char-code-property' and
96db204a 2500`put-char-code-property'.")
4ed46869
KH
2501
2502(defun get-char-code-property (char propname)
2503 "Return the value of CHAR's PROPNAME property in `char-code-property-table'."
2504 (let ((plist (aref char-code-property-table char)))
2505 (if (listp plist)
2506 (car (cdr (memq propname plist))))))
2507
2508(defun put-char-code-property (char propname value)
2509 "Store CHAR's PROPNAME property with VALUE in `char-code-property-table'.
2510It can be retrieved with `(get-char-code-property CHAR PROPNAME)'."
2511 (let ((plist (aref char-code-property-table char)))
2512 (if plist
2513 (let ((slot (memq propname plist)))
2514 (if slot
2515 (setcar (cdr slot) value)
2516 (nconc plist (list propname value))))
2517 (aset char-code-property-table char (list propname value)))))
2518
a127b764
KH
2519\f
2520;; Pretty description of encoded string
2521
2522;; Alist of ISO 2022 control code vs the corresponding mnemonic string.
2523(defvar iso-2022-control-alist
2524 '((?\x1b . "ESC")
2525 (?\x0e . "SO")
2526 (?\x0f . "SI")
2527 (?\x8e . "SS2")
2528 (?\x8f . "SS3")
2529 (?\x9b . "CSI")))
2530
2531(defun encoded-string-description (str coding-system)
2532 "Return a pretty description of STR that is encoded by CODING-SYSTEM."
2533 (setq str (string-as-unibyte str))
993b2a7d 2534 (mapconcat
8c9d55a9 2535 (if (and coding-system (eq (coding-system-type coding-system) 2))
993b2a7d
KH
2536 ;; Try to get a pretty description for ISO 2022 escape sequences.
2537 (function (lambda (x) (or (cdr (assq x iso-2022-control-alist))
a9f60091 2538 (format "0x%02X" x))))
5de75f53 2539 (function (lambda (x) (format "0x%02X" x))))
993b2a7d 2540 str " "))
a127b764
KH
2541
2542(defun encode-coding-char (char coding-system)
2543 "Encode CHAR by CODING-SYSTEM and return the resulting string.
2544If CODING-SYSTEM can't safely encode CHAR, return nil."
bd953173
KH
2545 (let ((str1 (string-as-multibyte (char-to-string char)))
2546 (str2 (string-as-multibyte (make-string 2 char)))
a89f541b
KH
2547 (safe-chars (and coding-system
2548 (coding-system-get coding-system 'safe-chars)))
8c9d55a9 2549 (charset (char-charset char))
a127b764 2550 enc1 enc2 i1 i2)
a89f541b 2551 (when (or (eq safe-chars t)
8c9d55a9 2552 (eq charset 'ascii)
a89f541b 2553 (and safe-chars (aref safe-chars char)))
a127b764
KH
2554 ;; We must find the encoded string of CHAR. But, just encoding
2555 ;; CHAR will put extra control sequences (usually to designate
34703cb1 2556 ;; ASCII charset) at the tail if type of CODING is ISO 2022.
a127b764
KH
2557 ;; To exclude such tailing bytes, we at first encode one-char
2558 ;; string and two-char string, then check how many bytes at the
2559 ;; tail of both encoded strings are the same.
2560
bd953173 2561 (setq enc1 (encode-coding-string str1 coding-system)
a127b764 2562 i1 (length enc1)
bd953173 2563 enc2 (encode-coding-string str2 coding-system)
a127b764
KH
2564 i2 (length enc2))
2565 (while (and (> i1 0) (= (aref enc1 (1- i1)) (aref enc2 (1- i2))))
2566 (setq i1 (1- i1) i2 (1- i2)))
2567
2568 ;; Now (substring enc1 i1) and (substring enc2 i2) are the same,
2569 ;; and they are the extra control sequences at the tail to
2570 ;; exclude.
2571 (substring enc2 0 i2))))
2572
2573
9ee5b744 2574;; arch-tag: b382c432-4b36-460e-bf4c-05efd0bb18dc
4ed46869 2575;;; mule-cmds.el ends here