HCoop / bpt / emacs.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
mule-cmds.el (select-safe-coding-system): If
[bpt/emacs.git] / lisp / international / mule-cmds.el
CommitLineData
932e8ec6 1;;; mule-cmds.el --- commands for multilingual environment -*-coding: utf-8 -*-
9ee5b744 2
ab422c4d 3;; Copyright (C) 1997-2013 Free Software Foundation, Inc.
7976eda0 4;; Copyright (C) 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004,
5df4f04c 5;; 2005, 2006, 2007, 2008, 2009, 2010, 2011
2fd125a3
KH		 6;; National Institute of Advanced Industrial Science and Technology (AIST)
7;; Registration Number H14PRO021
8f924df7 8;; Copyright (C) 2003
97941b05
KH		 9;; National Institute of Advanced Industrial Science and Technology (AIST)
10;; Registration Number H13PRO009
4ed46869 11
1f547b92 12;; Keywords: mule, i18n
4ed46869
KH		 13
14;; This file is part of GNU Emacs.
15
4936186e 16;; GNU Emacs is free software: you can redistribute it and/or modify
4ed46869 17;; it under the terms of the GNU General Public License as published by
4936186e
GM		 18;; the Free Software Foundation, either version 3 of the License, or
19;; (at your option) any later version.
4ed46869
KH		 20
21;; GNU Emacs is distributed in the hope that it will be useful,
22;; but WITHOUT ANY WARRANTY; without even the implied warranty of
23;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
24;; GNU General Public License for more details.
25
26;; You should have received a copy of the GNU General Public License
4936186e 27;; along with GNU Emacs. If not, see <http://www.gnu.org/licenses/>.
4ed46869 28
60370d40
PJ		 29;;; Commentary:
30
4ed46869
KH		 31;;; Code:
32
26b3dce6
GM		 33(defvar dos-codepage)
34(autoload 'widget-value "wid-edit")
cda74479 35
4a91d930 36(defvar mac-system-coding-system)
4a91d930 37
4ed46869
KH		 38;;; MULE related key bindings and menus.
39
64657387
SM		 40(defvar mule-keymap
41 (let ((map (make-sparse-keymap)))
42 (define-key map "f" 'set-buffer-file-coding-system)
43 (define-key map "r" 'revert-buffer-with-coding-system)
44 (define-key map "F" 'set-file-name-coding-system)
45 (define-key map "t" 'set-terminal-coding-system)
46 (define-key map "k" 'set-keyboard-coding-system)
47 (define-key map "p" 'set-buffer-process-coding-system)
48 (define-key map "x" 'set-selection-coding-system)
49 (define-key map "X" 'set-next-selection-coding-system)
50 (define-key map "\C-\\" 'set-input-method)
51 (define-key map "c" 'universal-coding-system-argument)
52 (define-key map "l" 'set-language-environment)
53 map)
33d17698 54 "Keymap for Mule (Multilingual environment) specific commands.")
4ed46869 55
8f81f784 56;; Keep "C-x C-m ..." for mule specific commands.
0709d285 57(define-key ctl-x-map "\C-m" mule-keymap)
ef8a8c8c 58
dcad02bc 59(defvar describe-language-environment-map
64657387 60 (let ((map (make-sparse-keymap "Describe Language Environment")))
1ec4b7b2
SM		 61 (bindings--define-key map
62 [Default] '(menu-item "Default" describe-specified-language-support))
64657387 63 map))
15b3e511 64
dcad02bc 65(defvar setup-language-environment-map
64657387 66 (let ((map (make-sparse-keymap "Set Language Environment")))
1ec4b7b2
SM		 67 (bindings--define-key map
68 [Default] '(menu-item "Default" setup-specified-language-environment))
64657387 69 map))
15b3e511 70
dcad02bc 71(defvar set-coding-system-map
64657387 72 (let ((map (make-sparse-keymap "Set Coding System")))
1ec4b7b2
SM		 73 (bindings--define-key map [set-buffer-process-coding-system]
74 '(menu-item "For I/O with Subprocess" set-buffer-process-coding-system
64657387
SM		 75 :visible (fboundp 'start-process)
76 :enable (get-buffer-process (current-buffer))
1ec4b7b2
SM		 77 :help "How to en/decode I/O from/to subprocess connected to this buffer"))
78 (bindings--define-key map [set-next-selection-coding-system]
79 '(menu-item "For Next X Selection" set-next-selection-coding-system
80 :visible (display-selections-p)
81 :help "How to en/decode next selection/clipboard operation"))
82 (bindings--define-key map [set-selection-coding-system]
83 '(menu-item "For X Selections/Clipboard" set-selection-coding-system
84 :visible (display-selections-p)
85 :help "How to en/decode data to/from selection/clipboard"))
86
87 (bindings--define-key map [separator-3] menu-bar-separator)
88 (bindings--define-key map [set-terminal-coding-system]
89 '(menu-item "For Terminal" set-terminal-coding-system
90 :enable (null (memq initial-window-system '(x w32 ns)))
91 :help "How to encode terminal output"))
92 (bindings--define-key map [set-keyboard-coding-system]
93 '(menu-item "For Keyboard" set-keyboard-coding-system
94 :help "How to decode keyboard input"))
72b255c7 95
1ec4b7b2
SM		 96 (bindings--define-key map [separator-2] menu-bar-separator)
97 (bindings--define-key map [set-file-name-coding-system]
98 '(menu-item "For File Name" set-file-name-coding-system
99 :help "How to decode/encode file names"))
100 (bindings--define-key map [revert-buffer-with-coding-system]
101 '(menu-item "For Reverting This File Now"
102 revert-buffer-with-coding-system
103 :enable buffer-file-name
104 :help "Revisit this file immediately using specified coding system"))
105 (bindings--define-key map [set-buffer-file-coding-system]
106 '(menu-item "For Saving This Buffer" set-buffer-file-coding-system
107 :help "How to encode this buffer when saved"))
108 (bindings--define-key map [separator-1] menu-bar-separator)
109 (bindings--define-key map [universal-coding-system-argument]
110 '(menu-item "For Next Command" universal-coding-system-argument
111 :help "Coding system to be used by next command"))
64657387
SM		 112 map))
113
114(defvar mule-menu-keymap
115 (let ((map (make-sparse-keymap "Mule (Multilingual Environment)")))
1ec4b7b2
SM		 116 (bindings--define-key map [mule-diag]
117 '(menu-item "Show All Multilingual Settings" mule-diag
118 :help "Display multilingual environment settings"))
119 (bindings--define-key map [list-character-sets]
120 '(menu-item "List Character Sets" list-character-sets
121 :help "Show table of available character sets"))
122 (bindings--define-key map [describe-coding-system]
123 '(menu-item "Describe Coding System..." describe-coding-system))
124 (bindings--define-key map [describe-input-method]
125 '(menu-item "Describe Input Method..." describe-input-method
126 :help "Keyboard layout for a specific input method"))
127 (bindings--define-key map [describe-language-environment]
128 `(menu-item "Describe Language Environment"
129 ,describe-language-environment-map
130 :help "Show multilingual settings for a specific language"))
72b255c7 131
1ec4b7b2
SM		 132 (bindings--define-key map [separator-coding-system] menu-bar-separator)
133 (bindings--define-key map [view-hello-file]
134 '(menu-item "Show Multilingual Sample Text" view-hello-file
64657387
SM		 135 :enable (file-readable-p
136 (expand-file-name "HELLO" data-directory))
1ec4b7b2
SM		 137 :help "Demonstrate various character sets"))
138 (bindings--define-key map [set-various-coding-system]
139 `(menu-item "Set Coding Systems" ,set-coding-system-map
140 :enable (default-value 'enable-multibyte-characters)))
64657387 141
1ec4b7b2
SM		 142 (bindings--define-key map [separator-input-method] menu-bar-separator)
143 (bindings--define-key map [describe-input-method]
144 '(menu-item "Describe Input Method" describe-input-method))
145 (bindings--define-key map [set-input-method]
146 '(menu-item "Select Input Method..." set-input-method))
147 (bindings--define-key map [toggle-input-method]
148 '(menu-item "Toggle Input Method" toggle-input-method))
149
150 (bindings--define-key map [separator-mule] menu-bar-separator)
151 (bindings--define-key map [set-language-environment]
152 `(menu-item "Set Language Environment" ,setup-language-environment-map))
64657387
SM		 153 map)
154 "Keymap for Mule (Multilingual environment) menu specific commands.")
cda74479 155
4ed46869
KH		 156;; This should be a single character key binding because users use it
157;; very frequently while editing multilingual text. Now we can use
158;; only two such keys: "\C-\\" and "\C-^", but the latter is not
159;; convenient because it requires shifting on most keyboards. An
160;; alternative is "\C-\]" which is now bound to `abort-recursive-edit'
161;; but it won't be used that frequently.
162(define-key global-map "\C-\\" 'toggle-input-method)
163
5f395df3
SM		 164;; This is no good because people often type Shift-SPC
165;; meaning to type SPC. -- rms.
166;; ;; Here's an alternative key binding for X users (Shift-SPACE).
167;; (define-key global-map [?\S- ] 'toggle-input-method)
b4fba33f 168
464cc130
KH		 169;;; Mule related hyperlinks.
170(defconst help-xref-mule-regexp-template
171 (purecopy (concat "\\(\\<\\("
172 "\\(coding system\\)\\|"
d0c40faa
KH		 173 "\\(input method\\)\\|"
174 "\\(character set\\)\\|"
175 "\\(charset\\)"
464cc130
KH		 176 "\\)\\s-+\\)?"
177 ;; Note starting with word-syntax character:
178 "`\\(\\sw\\(\\sw\\|\\s_\\)+\\)'")))
179
26d87040 180(defun coding-system-change-eol-conversion (coding-system eol-type)
caff3c0a 181 "Return a coding system which differs from CODING-SYSTEM in EOL conversion.
26d87040
EZ		 182The returned coding system converts end-of-line by EOL-TYPE
183but text as the same way as CODING-SYSTEM.
184EOL-TYPE should be `unix', `dos', `mac', or nil.
185If EOL-TYPE is nil, the returned coding system detects
186how end-of-line is formatted automatically while decoding.
187
188EOL-TYPE can be specified by an integer 0, 1, or 2.
189They means `unix', `dos', and `mac' respectively."
190 (if (symbolp eol-type)
191 (setq eol-type (cond ((eq eol-type 'unix) 0)
192 ((eq eol-type 'dos) 1)
193 ((eq eol-type 'mac) 2)
194 (t eol-type))))
1d5b0c66
CY		 195 ;; We call `coding-system-base' before `coding-system-eol-type',
196 ;; because the coding-system may not be initialized until then.
197 (let* ((base (coding-system-base coding-system))
198 (orig-eol-type (coding-system-eol-type coding-system)))
199 (cond ((vectorp orig-eol-type)
200 (if (not eol-type)
201 coding-system
202 (aref orig-eol-type eol-type)))
203 ((not eol-type)
204 base)
205 ((= eol-type orig-eol-type)
206 coding-system)
207 ((progn (setq orig-eol-type (coding-system-eol-type base))
208 (vectorp orig-eol-type))
209 (aref orig-eol-type eol-type)))))
26d87040
EZ		 210
211(defun coding-system-change-text-conversion (coding-system coding)
212 "Return a coding system which differs from CODING-SYSTEM in text conversion.
213The returned coding system converts text by CODING
214but end-of-line as the same way as CODING-SYSTEM.
215If CODING is nil, the returned coding system detects
216how text is formatted automatically while decoding."
1d77e15a
JR		 217 (let ((eol-type (coding-system-eol-type coding-system)))
218 (coding-system-change-eol-conversion
219 (if coding coding 'undecided)
220 (if (numberp eol-type) (aref [unix dos mac] eol-type)))))
26d87040 221
be7ca044
KH		 222;; Canonicalize the coding system name NAME by removing some prefixes
223;; and delimiter characters. Support function of
224;; coding-system-from-name.
225(defun canonicalize-coding-system-name (name)
ca4f0e9a
KH		 226 (if (string-match "^\\(ms\\|ibm\\|windows-\\)\\([0-9]+\\)$" name)
227 ;; "ms950", "ibm950", "windows-950" -> "cp950"
228 (concat "cp" (match-string 2 name))
229 (if (string-match "^iso[-_ ]?[0-9]" name)
230 ;; "iso-8859-1" -> "8859-1", "iso-2022-jp" ->"2022-jp"
231 (setq name (substring name (1- (match-end 0)))))
232 (let ((idx (string-match "[-_ /]" name)))
233 ;; Delete "-", "_", " ", "/" but do distinguish "16-be" and "16be".
234 (while idx
235 (if (and (>= idx 2)
236 (eq (string-match "16-[lb]e$" name (- idx 2))
237 (- idx 2)))
238 (setq idx (string-match "[-_ /]" name (match-end 0)))
239 (setq name (concat (substring name 0 idx) (substring name (1+ idx)))
240 idx (string-match "[-_ /]" name idx))))
241 name)))
be7ca044
KH		 242
243(defun coding-system-from-name (name)
244 "Return a coding system whose name matches with NAME (string or symbol)."
245 (let (sym)
246 (if (stringp name) (setq sym (intern name))
247 (setq sym name name (symbol-name name)))
248 (if (coding-system-p sym)
249 sym
250 (let ((eol-type
251 (if (string-match "-\\(unix\\|dos\\|mac\\)$" name)
252 (prog1 (intern (match-string 1 name))
253 (setq name (substring name 0 (match-beginning 0)))))))
254 (setq name (canonicalize-coding-system-name (downcase name)))
255 (catch 'tag
256 (dolist (elt (coding-system-list))
257 (if (string= (canonicalize-coding-system-name (symbol-name elt))
258 name)
259 (throw 'tag (if eol-type (coding-system-change-eol-conversion
260 elt eol-type)
261 elt)))))))))
262
4ed46869 263(defun toggle-enable-multibyte-characters (&optional arg)
6998e1a1 264 "Change whether this buffer uses multibyte characters.
49275d55 265With ARG, use multibyte characters if the ARG is positive.
6998e1a1
RS		 266
267Note that this command does not convert the byte contents of
268the buffer; it only changes the way those bytes are interpreted.
269In general, therefore, this command *changes* the sequence of
270characters that the current buffer contains.
271
caff3c0a
JB		 272We suggest you avoid using this command unless you know what you are
273doing. If you use it by mistake, and the buffer is now displayed
6998e1a1 274wrong, use this command again to toggle back to the right mode."
4ed46869 275 (interactive "P")
b7079457
RS		 276 (let ((new-flag
277 (if (null arg) (null enable-multibyte-characters)
278 (> (prefix-numeric-value arg) 0))))
279 (set-buffer-multibyte new-flag))
4ed46869
KH		 280 (force-mode-line-update))
281
282(defun view-hello-file ()
cf2fc7e9 283 "Display the HELLO file, which lists many languages and characters."
4ed46869 284 (interactive)
8f81f784 285 ;; We have to decode the file in any environment.
4dd1c416 286 (let ((coding-system-for-read 'iso-2022-7bit))
18ccd78a 287 (view-file (expand-file-name "HELLO" data-directory))))
4ed46869 288
9e9a77a6 289(defun universal-coding-system-argument (coding-system)
15b3e511 290 "Execute an I/O command using the specified coding system."
9e9a77a6
RS		 291 (interactive
292 (let ((default (and buffer-file-coding-system
34104362 293 (not (eq (coding-system-type buffer-file-coding-system)
a0d96cad 294 'undecided))
9e9a77a6
RS		 295 buffer-file-coding-system)))
296 (list (read-coding-system
297 (if default
5b76833f 298 (format "Coding system for following command (default %s): " default)
9e9a77a6
RS		 299 "Coding system for following command: ")
300 default))))
301 (let* ((keyseq (read-key-sequence
e14a8f4c 302 (format "Command to execute with %s:" coding-system)))
04363179
GM		 303 (cmd (key-binding keyseq))
304 prefix)
735c5b17
GM		 305 ;; read-key-sequence ignores quit, so make an explicit check.
306 ;; Like many places, this assumes quit == C-g, but it need not be.
1e4bd40d 307 (if (equal last-input-event ?\C-g)
735c5b17 308 (keyboard-quit))
4a304bb2 309 (when (memq cmd '(universal-argument digit-argument))
04363179 310 (call-interactively cmd)
a1506d29 311
04363179
GM		 312 ;; Process keys bound in `universal-argument-map'.
313 (while (progn
314 (setq keyseq (read-key-sequence nil t)
315 cmd (key-binding keyseq t))
316 (not (eq cmd 'universal-argument-other-key)))
317 (let ((current-prefix-arg prefix-arg)
61a846fb 318 ;; Have to bind `last-command-event' here so that
6b61353c 319 ;; `digit-argument', for instance, can compute the
04363179 320 ;; prefix arg.
61a846fb 321 (last-command-event (aref keyseq 0)))
04363179
GM		 322 (call-interactively cmd)))
323
1f547b92 324 ;; This is the final call to `universal-argument-other-key', which
04363179
GM		 325 ;; set's the final `prefix-arg.
326 (let ((current-prefix-arg prefix-arg))
327 (call-interactively cmd))
a1506d29 328
04363179
GM		 329 ;; Read the command to execute with the given prefix arg.
330 (setq prefix prefix-arg
331 keyseq (read-key-sequence nil t)
332 cmd (key-binding keyseq)))
333
15b3e511 334 (let ((coding-system-for-read coding-system)
04363179 335 (coding-system-for-write coding-system)
0e9ec609 336 (coding-system-require-warning t)
04363179 337 (current-prefix-arg prefix))
15b3e511
KH		 338 (message "")
339 (call-interactively cmd))))
340
de94d711 341(defun set-default-coding-systems (coding-system)
0c3154d2 342 "Set default value of various coding systems to CODING-SYSTEM.
387136f6 343This sets the following coding systems:
0c3154d2 344 o coding system of a newly created buffer
8efc03e1
KH		 345 o default coding system for subprocess I/O
346This also sets the following values:
d3e4babd 347 o default value used as `file-name-coding-system' for converting file names
caff3c0a 348 if CODING-SYSTEM is ASCII-compatible
d8cb7e3d 349 o default value for the command `set-terminal-coding-system'
d3e4babd 350 o default value for the command `set-keyboard-coding-system'
caff3c0a 351 if CODING-SYSTEM is ASCII-compatible"
de94d711
KH		 352 (check-coding-system coding-system)
353 (setq-default buffer-file-coding-system coding-system)
6b61353c 354
db9aec47 355 (if (eq system-type 'darwin)
a41118cc 356 ;; The file-name coding system on Darwin systems is always utf-8.
db9aec47 357 (setq default-file-name-coding-system 'utf-8)
597e2240 358 (if (and (default-value 'enable-multibyte-characters)
db9aec47
KH		 359 (or (not coding-system)
360 (coding-system-get coding-system 'ascii-compatible-p)))
361 (setq default-file-name-coding-system coding-system)))
970c9391 362 (setq default-terminal-coding-system coding-system)
d1f61aaf
EZ		 363 ;; Prevent default-terminal-coding-system from converting ^M to ^J.
364 (setq default-keyboard-coding-system
365 (coding-system-change-eol-conversion coding-system 'unix))
1d77e15a
JR		 366 ;; Preserve eol-type from existing default-process-coding-systems.
367 ;; On non-unix-like systems in particular, these may have been set
368 ;; carefully by the user, or by the startup code, to deal with the
369 ;; users shell appropriately, so should not be altered by changing
370 ;; language environment.
371 (let ((output-coding
372 (coding-system-change-text-conversion
373 (car default-process-coding-system) coding-system))
374 (input-coding
375 (coding-system-change-text-conversion
376 (cdr default-process-coding-system) coding-system)))
377 (setq default-process-coding-system
378 (cons output-coding input-coding))))
de94d711 379
0c3154d2
KH		 380(defun prefer-coding-system (coding-system)
381 "Add CODING-SYSTEM at the front of the priority list for automatic detection.
387136f6 382This also sets the following coding systems:
0c3154d2 383 o coding system of a newly created buffer
8efc03e1
KH		 384 o default coding system for subprocess I/O
385This also sets the following values:
caff3c0a 386 o default value used as `file-name-coding-system' for converting file names
d8cb7e3d 387 o default value for the command `set-terminal-coding-system'
03c35c83
EZ		 388 o default value for the command `set-keyboard-coding-system'
389
bd3ac67e
EZ		 390If CODING-SYSTEM specifies a certain type of EOL conversion, the coding
391systems set by this function will use that type of EOL conversion.
392
d0d8885d 393A coding system that requires automatic detection of text+encoding
bb388cc5
LMI		 394\(e.g. undecided, unix) can't be preferred.
395
396To prefer, for instance, utf-8, say the following:
397
398 \(prefer-coding-system 'utf-8)"
0c3154d2
KH		 399 (interactive "zPrefer coding system: ")
400 (if (not (and coding-system (coding-system-p coding-system)))
401 (error "Invalid coding system `%s'" coding-system))
97941b05
KH		 402 (if (memq (coding-system-type coding-system) '(raw-text undecided))
403 (error "Can't prefer the coding system `%s'" coding-system))
404 (let ((base (coding-system-base coding-system))
bd3ac67e 405 (eol-type (coding-system-eol-type coding-system)))
97941b05 406 (set-coding-system-priority base)
32226619 407 (and (called-interactively-p 'interactive)
97941b05
KH		 408 (or (eq base coding-system)
409 (message "Highest priority is set to %s (base of %s)"
410 base coding-system)))
bd3ac67e 411 ;; If they asked for specific EOL conversion, honor that.
6f9dc4fd 412 (if (memq eol-type '(0 1 2))
97941b05
KH		 413 (setq base
414 (coding-system-change-eol-conversion base eol-type)))
71784361
KH		 415 (set-default-coding-systems base)
416 (if (called-interactively-p 'interactive)
417 (or (eq base default-file-name-coding-system)
418 (message "The default value of `file-name-coding-system' was not changed because the specified coding system is not suitable for file names.")))))
0c3154d2 419
b5edd1d1
KH		 420(defvar sort-coding-systems-predicate nil
421 "If non-nil, a predicate function to sort coding systems.
422
423It is called with two coding systems, and should return t if the first
424one is \"less\" than the second.
425
426The function `sort-coding-systems' use it.")
427
428(defun sort-coding-systems (codings)
429 "Sort coding system list CODINGS by a priority of each coding system.
caff3c0a 430Return the sorted list. CODINGS is modified by side effects.
b5edd1d1
KH		 431
432If a coding system is most preferred, it has the highest priority.
caff3c0a
JB		 433Otherwise, coding systems that correspond to MIME charsets have
434higher priorities. Among them, a coding system included in the
435`coding-system' key of the current language environment has higher
436priority. See also the documentation of `language-info-alist'.
b5edd1d1
KH		 437
438If the variable `sort-coding-systems-predicate' (which see) is
caff3c0a 439non-nil, it is used to sort CODINGS instead."
b5edd1d1
KH		 440 (if sort-coding-systems-predicate
441 (sort codings sort-coding-systems-predicate)
8f924df7
KH		 442 (let* ((from-priority (coding-system-priority-list))
443 (most-preferred (car from-priority))
b5edd1d1
KH		 444 (lang-preferred (get-language-info current-language-environment
445 'coding-system))
446 (func (function
447 (lambda (x)
448 (let ((base (coding-system-base x)))
da1ebad1
KH		 449 ;; We calculate the priority number 0..255 by
450 ;; using the 8 bits PMMLCEII as this:
3ecd3a56
GM		 451 ;; P: 1 if most preferred.
452 ;; MM: greater than 0 if mime-charset.
453 ;; L: 1 if one of the current lang. env.'s codings.
454 ;; C: 1 if one of codings listed in the category list.
455 ;; E: 1 if not XXX-with-esc
da1ebad1
KH		 456 ;; II: if iso-2022 based, 0..3, else 1.
457 (logior
458 (lsh (if (eq base most-preferred) 1 0) 7)
459 (lsh
8f924df7 460 (let ((mime (coding-system-get base :mime-charset)))
b439e72a
DL		 461 ;; Prefer coding systems corresponding to a
462 ;; MIME charset.
b5edd1d1 463 (if mime
b439e72a
DL		 464 ;; Lower utf-16 priority so that we
465 ;; normally prefer utf-8 to it, and put
466 ;; x-ctext below that.
d0d8885d
JB		 467 (cond ((string-match-p "utf-16"
468 (symbol-name mime))
da1ebad1 469 2)
d0d8885d 470 ((string-match-p "^x-" (symbol-name mime))
da1ebad1
KH		 471 1)
472 (t 3))
b5edd1d1 473 0))
da1ebad1
KH		 474 5)
475 (lsh (if (memq base lang-preferred) 1 0) 4)
8f924df7 476 (lsh (if (memq base from-priority) 1 0) 3)
d0d8885d
JB		 477 (lsh (if (string-match-p "-with-esc\\'"
478 (symbol-name base))
da1ebad1 479 0 1) 2)
8f924df7
KH		 480 (if (eq (coding-system-type base) 'iso-2022)
481 (let ((category (coding-system-category base)))
482 ;; For ISO based coding systems, prefer
483 ;; one that doesn't use designation nor
484 ;; locking/single shifting.
a0d96cad
KH		 485 (cond
486 ((or (eq category 'coding-category-iso-8-1)
487 (eq category 'coding-category-iso-8-2))
488 2)
489 ((or (eq category 'coding-category-iso-7-tight)
490 (eq category 'coding-category-iso-7))
491 1)
492 (t
493 0)))
494 1)
a6dfc99b 495 ))))))
b5edd1d1
KH		 496 (sort codings (function (lambda (x y)
497 (> (funcall func x) (funcall func y))))))))
54b226f7 498
3fc7dfe5 499(defun find-coding-systems-region (from to)
54b226f7 500 "Return a list of proper coding systems to encode a text between FROM and TO.
8f924df7 501
d37ef0f6 502If FROM is a string, find coding systems in that instead of the buffer.
54b226f7
KH		 503All coding systems in the list can safely encode any multibyte characters
504in the text.
505
e8dd0160 506If the text contains no multibyte characters, return a list of a single
3fc7dfe5 507element `undecided'."
b5edd1d1
KH		 508 (let ((codings (find-coding-systems-region-internal from to)))
509 (if (eq codings t)
510 ;; The text contains only ASCII characters. Any coding
511 ;; systems are safe.
512 '(undecided)
513 ;; We need copy-sequence because sorting will alter the argument.
514 (sort-coding-systems (copy-sequence codings)))))
54b226f7 515
3fc7dfe5
KH		 516(defun find-coding-systems-string (string)
517 "Return a list of proper coding systems to encode STRING.
518All coding systems in the list can safely encode any multibyte characters
519in STRING.
520
e8dd0160 521If STRING contains no multibyte characters, return a list of a single
3fc7dfe5 522element `undecided'."
b5edd1d1 523 (find-coding-systems-region string nil))
3fc7dfe5
KH		 524
525(defun find-coding-systems-for-charsets (charsets)
526 "Return a list of proper coding systems to encode characters of CHARSETS.
6053d86a 527CHARSETS is a list of character sets.
1f547b92
DL		 528
529This only finds coding systems of type `charset', whose
530`:charset-list' property includes all of CHARSETS (plus `ascii' for
d0d8885d 531ASCII-compatible coding systems). It was used in older versions of
1f547b92
DL		 532Emacs, but is unlikely to be what you really want now."
533 ;; Deal with aliases.
534 (setq charsets (mapcar (lambda (c)
535 (get-charset-property c :name))
536 charsets))
b5edd1d1
KH		 537 (cond ((or (null charsets)
538 (and (= (length charsets) 1)
539 (eq 'ascii (car charsets))))
540 '(undecided))
541 ((or (memq 'eight-bit-control charsets)
542 (memq 'eight-bit-graphic charsets))
1f547b92 543 '(raw-text utf-8-emacs))
b5edd1d1 544 (t
1f547b92
DL		 545 (let (codings)
546 (dolist (cs (coding-system-list t))
c7b4b466
DL		 547 (let ((cs-charsets (and (eq (coding-system-type cs) 'charset)
548 (coding-system-charset-list cs)))
1f547b92
DL		 549 (charsets charsets))
550 (if (coding-system-get cs :ascii-compatible-p)
551 (add-to-list 'cs-charsets 'ascii))
552 (if (catch 'ok
553 (when cs-charsets
554 (while charsets
555 (unless (memq (pop charsets) cs-charsets)
556 (throw 'ok nil)))
557 t))
558 (push cs codings))))
559 (nreverse codings)))))
54b226f7 560
51ed58ea
KH		 561(defun find-multibyte-characters (from to &optional maxcount excludes)
562 "Find multibyte characters in the region specified by FROM and TO.
563If FROM is a string, find multibyte characters in the string.
564The return value is an alist of the following format:
565 ((CHARSET COUNT CHAR ...) ...)
566where
567 CHARSET is a character set,
568 COUNT is a number of characters,
1f547b92 569 CHARs are the characters found from the character set.
51ed58ea 570Optional 3rd arg MAXCOUNT limits how many CHARs are put in the above list.
caff3c0a 571Optional 4th arg EXCLUDES is a list of character sets to be ignored."
51ed58ea
KH		 572 (let ((chars nil)
573 charset char)
574 (if (stringp from)
a0d96cad
KH		 575 (if (multibyte-string-p from)
576 (let ((idx 0))
d0d8885d 577 (while (setq idx (string-match-p "[^\000-\177]" from idx))
a0d96cad
KH		 578 (setq char (aref from idx)
579 charset (char-charset char))
580 (unless (memq charset excludes)
581 (let ((slot (assq charset chars)))
582 (if slot
583 (if (not (memq char (nthcdr 2 slot)))
584 (let ((count (nth 1 slot)))
585 (setcar (cdr slot) (1+ count))
586 (if (or (not maxcount) (< count maxcount))
587 (nconc slot (list char)))))
588 (setq chars (cons (list charset 1 char) chars)))))
589 (setq idx (1+ idx)))))
590 (if enable-multibyte-characters
591 (save-excursion
592 (goto-char from)
593 (while (re-search-forward "[^\000-\177]" to t)
594 (setq char (preceding-char)
595 charset (char-charset char))
596 (unless (memq charset excludes)
51ed58ea
KH		 597 (let ((slot (assq charset chars)))
598 (if slot
a0d96cad 599 (if (not (member char (nthcdr 2 slot)))
51ed58ea
KH		 600 (let ((count (nth 1 slot)))
601 (setcar (cdr slot) (1+ count))
602 (if (or (not maxcount) (< count maxcount))
603 (nconc slot (list char)))))
a0d96cad 604 (setq chars (cons (list charset 1 char) chars)))))))))
51ed58ea
KH		 605 (nreverse chars)))
606
738746ba
KH		 607(defun search-unencodable-char (coding-system)
608 "Search forward from point for a character that is not encodable.
609It asks which coding system to check.
610If such a character is found, set point after that character.
611Otherwise, don't move point.
612
caff3c0a
JB		 613When called from a program, the value is the position of the unencodable
614character found, or nil if all characters are encodable."
738746ba
KH		 615 (interactive
616 (list (let ((default (or buffer-file-coding-system 'us-ascii)))
617 (read-coding-system
5b76833f 618 (format "Coding-system (default %s): " default)
738746ba
KH		 619 default))))
620 (let ((pos (unencodable-char-position (point) (point-max) coding-system)))
621 (if pos
622 (goto-char (1+ pos))
623 (message "All following characters are encodable by %s" coding-system))
624 pos))
625
c83c4f60
RS		 626(defvar last-coding-system-specified nil
627 "Most recent coding system explicitly specified by the user when asked.
628This variable is set whenever Emacs asks the user which coding system
629to use in order to write a file. If you set it to nil explicitly,
630then call `write-region', then afterward this variable will be non-nil
631only if the user was explicitly asked and specified a coding system.")
632
b5edd1d1 633(defvar select-safe-coding-system-accept-default-p nil
fea6b736 634 "If non-nil, a function to control the behavior of coding system selection.
b5edd1d1
KH		 635The meaning is the same as the argument ACCEPT-DEFAULT-P of the
636function `select-safe-coding-system' (which see). This variable
637overrides that argument.")
638
9ee5b744
SM		 639(defun select-safe-coding-system-interactively (from to codings unsafe
640 &optional rejected default)
641 "Select interactively a coding system for the region FROM ... TO.
642FROM can be a string, as in `write-region'.
643CODINGS is the list of base coding systems known to be safe for this region,
644 typically obtained with `find-coding-systems-region'.
645UNSAFE is a list of coding systems known to be unsafe for this region.
646REJECTED is a list of coding systems which were safe but for some reason
647 were not recommended in the particular context.
648DEFAULT is the coding system to use by default in the query."
649 ;; At first, if some defaults are unsafe, record at most 11
650 ;; problematic characters and their positions for them by turning
651 ;; (CODING ...)
652 ;; into
653 ;; ((CODING (POS . CHAR) (POS . CHAR) ...) ...)
654 (if unsafe
655 (setq unsafe
656 (mapcar #'(lambda (coding)
657 (cons coding
658 (if (stringp from)
659 (mapcar #'(lambda (pos)
660 (cons pos (aref from pos)))
661 (unencodable-char-position
662 0 (length from) coding
663 11 from))
664 (mapcar #'(lambda (pos)
665 (cons pos (char-after pos)))
666 (unencodable-char-position
667 from to coding 11)))))
668 unsafe)))
669
670 ;; Change each safe coding system to the corresponding
671 ;; mime-charset name if it is also a coding system. Such a name
672 ;; is more friendly to users.
673 (let ((l codings)
674 mime-charset)
675 (while l
a77c22c2
KH		 676 (setq mime-charset (coding-system-get (car l) :mime-charset))
677 (if (and mime-charset (coding-system-p mime-charset)
678 (coding-system-equal (car l) mime-charset))
9ee5b744
SM		 679 (setcar l mime-charset))
680 (setq l (cdr l))))
681
682 ;; Don't offer variations with locking shift, which you
683 ;; basically never want.
684 (let (l)
685 (dolist (elt codings (setq codings (nreverse l)))
686 (unless (or (eq 'coding-category-iso-7-else
687 (coding-system-category elt))
688 (eq 'coding-category-iso-8-else
689 (coding-system-category elt)))
690 (push elt l))))
691
692 ;; Remove raw-text, emacs-mule and no-conversion unless nothing
693 ;; else is available.
694 (setq codings
695 (or (delq 'raw-text
696 (delq 'emacs-mule
697 (delq 'no-conversion codings)))
698 '(raw-text emacs-mule no-conversion)))
699
700 (let ((window-configuration (current-window-configuration))
701 (bufname (buffer-name))
702 coding-system)
703 (save-excursion
704 ;; If some defaults are unsafe, make sure the offending
705 ;; buffer is displayed.
706 (when (and unsafe (not (stringp from)))
707 (pop-to-buffer bufname)
708 (goto-char (apply 'min (mapcar #'(lambda (x) (car (cadr x)))
709 unsafe))))
710 ;; Then ask users to select one from CODINGS while showing
711 ;; the reason why none of the defaults are not used.
712 (with-output-to-temp-buffer "*Warning*"
713 (with-current-buffer standard-output
714 (if (and (null rejected) (null unsafe))
715 (insert "No default coding systems to try for "
716 (if (stringp from)
717 (format "string \"%s\"." from)
718 (format "buffer `%s'." bufname)))
719 (insert
720 "These default coding systems were tried to encode"
721 (if (stringp from)
722 (concat " \"" (if (> (length from) 10)
723 (concat (substring from 0 10) "...\"")
724 (concat from "\"")))
725 (format " text\nin the buffer `%s'" bufname))
726 ":\n")
727 (let ((pos (point))
728 (fill-prefix " "))
729 (dolist (x (append rejected unsafe))
46c04718 730 (princ " ") (princ x))
9ee5b744
SM		 731 (insert "\n")
732 (fill-region-as-paragraph pos (point)))
733 (when rejected
3fc0b26e
EZ		 734 (insert "These safely encode the text in the buffer,
735but are not recommended for encoding text in this context,
9ee5b744
SM		 736e.g., for sending an email message.\n ")
737 (dolist (x rejected)
738 (princ " ") (princ x))
739 (insert "\n"))
740 (when unsafe
3fc0b26e 741 (insert (if rejected "The other coding systems"
9ee5b744 742 "However, each of them")
3fc0b26e 743 " encountered characters it couldn't encode:\n")
9ee5b744 744 (dolist (coding unsafe)
3fc0b26e 745 (insert (format " %s cannot encode these:" (car coding)))
9ee5b744
SM		 746 (let ((i 0)
747 (func1
748 #'(lambda (bufname pos)
749 (when (buffer-live-p (get-buffer bufname))
750 (pop-to-buffer bufname)
751 (goto-char pos))))
752 (func2
753 #'(lambda (bufname pos coding)
754 (when (buffer-live-p (get-buffer bufname))
755 (pop-to-buffer bufname)
756 (if (< (point) pos)
757 (goto-char pos)
758 (forward-char 1)
759 (search-unencodable-char coding)
760 (forward-char -1))))))
761 (dolist (elt (cdr coding))
762 (insert " ")
763 (if (stringp from)
764 (insert (if (< i 10) (cdr elt) "..."))
765 (if (< i 10)
766 (insert-text-button
767 (cdr elt)
768 :type 'help-xref
4e21e5aa 769 'face 'link
9ee5b744
SM		 770 'help-echo
771 "mouse-2, RET: jump to this character"
772 'help-function func1
773 'help-args (list bufname (car elt)))
774 (insert-text-button
775 "..."
776 :type 'help-xref
4e21e5aa 777 'face 'link
9ee5b744
SM		 778 'help-echo
779 "mouse-2, RET: next unencodable character"
780 'help-function func2
781 'help-args (list bufname (car elt)
782 (car coding)))))
783 (setq i (1+ i))))
784 (insert "\n"))
333f3572 785 (insert (substitute-command-keys "\
7d03c5b1 786
333f3572
KH		 787Click on a character (or switch to this window by `\\[other-window]'\n\
788and select the characters by RET) to jump to the place it appears,\n\
3fc0b26e 789where `\\[universal-argument] \\[what-cursor-position]' will give information about it.\n"))))
7d03c5b1 790 (insert (substitute-command-keys "\nSelect \
3fc0b26e
EZ		 791one of the safe coding systems listed below,\n\
792or cancel the writing with \\[keyboard-quit] and edit the buffer\n\
793 to remove or modify the problematic characters,\n\
794or specify any other coding system (and risk losing\n\
795 the problematic characters).\n\n"))
9ee5b744
SM		 796 (let ((pos (point))
797 (fill-prefix " "))
798 (dolist (x codings)
799 (princ " ") (princ x))
800 (insert "\n")
7d03c5b1 801 (fill-region-as-paragraph pos (point)))))
9ee5b744
SM		 802
803 ;; Read a coding system.
804 (setq coding-system
805 (read-coding-system
806 (format "Select coding system (default %s): " default)
807 default))
808 (setq last-coding-system-specified coding-system))
809
810 (kill-buffer "*Warning*")
811 (set-window-configuration window-configuration)
812 coding-system))
813
b5edd1d1 814(defun select-safe-coding-system (from to &optional default-coding-system
efa2c6d7 815 accept-default-p file)
d5266ddf
KH		 816 "Ask a user to select a safe coding system from candidates.
817The candidates of coding systems which can safely encode a text
b5edd1d1
KH		 818between FROM and TO are shown in a popup window. Among them, the most
819proper one is suggested as the default.
820
d0d8885d 821The list of `buffer-file-coding-system' of the current buffer, the
b56a5ae0 822default `buffer-file-coding-system', and the most preferred coding
d0d8885d
JB		 823system (if it corresponds to a MIME charset) is treated as the
824default coding system list. Among them, the first one that safely
825encodes the text is normally selected silently and returned without
826any user interaction. See also the command `prefer-coding-system'.
d37ef0f6
DL		 827
828However, the user is queried if the chosen coding system is
29d04c4f 829inconsistent with what would be selected by `find-auto-coding' from
d37ef0f6
DL		 830coding cookies &c. if the contents of the region were read from a
831file. (That could lead to data corruption in a file subsequently
832re-visited and edited.)
54b226f7 833
b5edd1d1
KH		 834Optional 3rd arg DEFAULT-CODING-SYSTEM specifies a coding system or a
835list of coding systems to be prepended to the default coding system
0e9ec609 836list. However, if DEFAULT-CODING-SYSTEM is a list and the first
ccb06340 837element is t, the cdr part is used as the default coding system list,
b56a5ae0 838i.e. current `buffer-file-coding-system', default `buffer-file-coding-system',
a985cd2f 839and the most preferred coding system are not used.
54b226f7 840
b5edd1d1
KH		 841Optional 4th arg ACCEPT-DEFAULT-P, if non-nil, is a function to
842determine the acceptability of the silently selected coding system.
843It is called with that coding system, and should return nil if it
844should not be silently selected and thus user interaction is required.
845
efa2c6d7
RS		 846Optional 5th arg FILE is the file name to use for this purpose.
847That is different from `buffer-file-name' when handling `write-region'
848\(for example).
849
d0d8885d
JB		 850The variable `select-safe-coding-system-accept-default-p', if non-nil,
851overrides ACCEPT-DEFAULT-P.
54b226f7
KH		 852
853Kludgy feature: if FROM is a string, the string is the target text,
854and TO is ignored."
64657387 855 (if (not (listp default-coding-system))
b5edd1d1
KH		 856 (setq default-coding-system (list default-coding-system)))
857
29d04c4f
KH		 858 (let ((no-other-defaults nil)
859 auto-cs)
d35deeec 860 (unless (or (stringp from) find-file-literally)
b4dc7d98 861 ;; Find an auto-coding that is specified for the current
29d04c4f
KH		 862 ;; buffer and file from the region FROM and TO.
863 (save-excursion
864 (save-restriction
865 (widen)
866 (goto-char from)
867 (setq auto-cs (find-auto-coding (or file buffer-file-name "")
868 (- to from)))
869 (if auto-cs
870 (if (coding-system-p (car auto-cs))
871 (setq auto-cs (car auto-cs))
872 (display-warning
2364df5c 873 'mule
29d04c4f
KH		 874 (format "\
875Invalid coding system `%s' is specified
876for the current buffer/file by the %s.
877It is highly recommended to fix it before writing to a file."
878 (car auto-cs)
879 (if (eq (cdr auto-cs) :coding) ":coding tag"
2364df5c
JB		 880 (format "variable `%s'" (cdr auto-cs))))
881 :warning)
29d04c4f
KH		 882 (or (yes-or-no-p "Really proceed with writing? ")
883 (error "Save aborted"))
884 (setq auto-cs nil))))))
885
0e9ec609
KH		 886 (if (eq (car default-coding-system) t)
887 (setq no-other-defaults t
888 default-coding-system (cdr default-coding-system)))
889
890 ;; Change elements of the list to (coding . base-coding).
891 (setq default-coding-system
892 (mapcar (function (lambda (x) (cons x (coding-system-base x))))
893 default-coding-system))
894
29d04c4f 895 (if (and auto-cs (not no-other-defaults))
2df48a87
SM		 896 ;; If the file has a coding cookie, use it regardless of any
897 ;; other setting.
29d04c4f 898 (let ((base (coding-system-base auto-cs)))
2df48a87
SM		 899 (unless (memq base '(nil undecided))
900 (setq default-coding-system (list (cons auto-cs base)))
901 (setq no-other-defaults t))))
29d04c4f 902
0e9ec609
KH		 903 (unless no-other-defaults
904 ;; If buffer-file-coding-system is not nil nor undecided, append it
905 ;; to the defaults.
906 (if buffer-file-coding-system
907 (let ((base (coding-system-base buffer-file-coding-system)))
908 (or (eq base 'undecided)
0e9ec609 909 (rassq base default-coding-system)
ccb06340
KH		 910 (setq default-coding-system
911 (append default-coding-system
912 (list (cons buffer-file-coding-system base)))))))
a985cd2f 913
659be258
KH		 914 (unless (and buffer-file-coding-system-explicit
915 (cdr buffer-file-coding-system-explicit))
b56a5ae0 916 ;; If default buffer-file-coding-system is not nil nor undecided,
659be258 917 ;; append it to the defaults.
b56a5ae0
SM		 918 (when (default-value 'buffer-file-coding-system)
919 (let ((base (coding-system-base
920 (default-value 'buffer-file-coding-system))))
921 (or (eq base 'undecided)
922 (rassq base default-coding-system)
923 (setq default-coding-system
924 (append default-coding-system
925 (list (cons (default-value
926 'buffer-file-coding-system)
927 base)))))))
659be258
KH		 928
929 ;; If the most preferred coding system has the property mime-charset,
930 ;; append it to the defaults.
931 (let ((preferred (coding-system-priority-list t))
932 base)
933 (and (coding-system-p preferred)
934 (setq base (coding-system-base preferred))
935 (coding-system-get preferred :mime-charset)
936 (not (rassq base default-coding-system))
937 (setq default-coding-system
938 (append default-coding-system
939 (list (cons preferred base))))))))
29d04c4f
KH		 940
941 (if select-safe-coding-system-accept-default-p
942 (setq accept-default-p select-safe-coding-system-accept-default-p))
943
bae7cd08 944 ;; Decide the eol-type from the top of the default codings,
b56a5ae0 945 ;; current buffer-file-coding-system, or default buffer-file-coding-system.
bae7cd08
KH		 946 (if default-coding-system
947 (let ((default-eol-type (coding-system-eol-type
948 (caar default-coding-system))))
949 (if (and (vectorp default-eol-type) buffer-file-coding-system)
d35deeec 950 (setq default-eol-type (coding-system-eol-type
bae7cd08 951 buffer-file-coding-system)))
b56a5ae0
SM		 952 (if (and (vectorp default-eol-type)
953 (default-value 'buffer-file-coding-system))
954 (setq default-eol-type
955 (coding-system-eol-type
956 (default-value 'buffer-file-coding-system))))
bae7cd08
KH		 957 (if (and default-eol-type (not (vectorp default-eol-type)))
958 (dolist (elt default-coding-system)
959 (setcar elt (coding-system-change-eol-conversion
960 (car elt) default-eol-type))))))
961
29d04c4f
KH		 962 (let ((codings (find-coding-systems-region from to))
963 (coding-system nil)
c381cb8d 964 (tick (if (not (stringp from)) (buffer-chars-modified-tick)))
29d04c4f 965 safe rejected unsafe)
bae7cd08
KH		 966 (if (eq (car codings) 'undecided)
967 ;; Any coding system is ok.
968 (setq coding-system (caar default-coding-system))
969 ;; Reverse the list so that elements are accumulated in safe,
970 ;; rejected, and unsafe in the correct order.
971 (setq default-coding-system (nreverse default-coding-system))
972
973 ;; Classify the defaults into safe, rejected, and unsafe.
974 (dolist (elt default-coding-system)
4d4ff4c6 975 (if (or (eq (coding-system-type (car elt)) 'undecided)
bae7cd08
KH		 976 (memq (cdr elt) codings))
977 (if (and (functionp accept-default-p)
978 (not (funcall accept-default-p (cdr elt))))
979 (push (car elt) rejected)
980 (push (car elt) safe))
981 (push (car elt) unsafe)))
982 (if safe
983 (setq coding-system (car safe))))
29d04c4f
KH		 984
985 ;; If all the defaults failed, ask a user.
bae7cd08 986 (when (not coding-system)
29d04c4f
KH		 987 (setq coding-system (select-safe-coding-system-interactively
988 from to codings unsafe rejected (car codings))))
989
29d04c4f
KH		 990 ;; Check we're not inconsistent with what `coding:' spec &c would
991 ;; give when file is re-read.
992 ;; But don't do this if we explicitly ignored the cookie
993 ;; by using `find-file-literally'.
994 (when (and auto-cs
995 (not (and
996 coding-system
997 (memq (coding-system-type coding-system) '(0 5)))))
6b61353c
KH		 998 ;; Merge coding-system and auto-cs as far as possible.
999 (if (not coding-system)
1000 (setq coding-system auto-cs)
1001 (if (not auto-cs)
1002 (setq auto-cs coding-system)
1003 (let ((eol-type-1 (coding-system-eol-type coding-system))
1004 (eol-type-2 (coding-system-eol-type auto-cs)))
1005 (if (eq (coding-system-base coding-system) 'undecided)
1006 (setq coding-system (coding-system-change-text-conversion
1007 coding-system auto-cs))
1008 (if (eq (coding-system-base auto-cs) 'undecided)
1009 (setq auto-cs (coding-system-change-text-conversion
1010 auto-cs coding-system))))
1011 (if (vectorp eol-type-1)
1012 (or (vectorp eol-type-2)
1013 (setq coding-system (coding-system-change-eol-conversion
1014 coding-system eol-type-2)))
1015 (if (vectorp eol-type-2)
1016 (setq auto-cs (coding-system-change-eol-conversion
1017 auto-cs eol-type-1)))))))
1018
1019 (if (and auto-cs
d37ef0f6
DL		 1020 ;; Don't barf if writing a compressed file, say.
1021 ;; This check perhaps isn't ideal, but is probably
1022 ;; the best thing to do.
1023 (not (auto-coding-alist-lookup (or file buffer-file-name "")))
6b61353c 1024 (not (coding-system-equal coding-system auto-cs)))
43afed8c
RS		 1025 (unless (yes-or-no-p
1026 (format "Selected encoding %s disagrees with \
1027%s specified by file contents. Really save (else edit coding cookies \
1028and try again)? " coding-system auto-cs))
29d04c4f 1029 (error "Save aborted"))))
c381cb8d 1030 (when (and tick (/= tick (buffer-chars-modified-tick)))
301b181a 1031 (error "Canceled because the buffer was modified"))
4d4ff4c6
KH		 1032 (if (and (eq (coding-system-type coding-system) 'undecided)
1033 (coding-system-get coding-system :prefer-utf-8)
1034 (< (- to from) (- (position-bytes to) (position-bytes from))))
1035 (setq coding-system
1036 (coding-system-change-text-conversion coding-system 'utf-8)))
29d04c4f 1037 coding-system)))
54b226f7
KH		 1038
1039(setq select-safe-coding-system-function 'select-safe-coding-system)
1040
46babb23
KH		 1041(defun select-message-coding-system ()
1042 "Return a coding system to encode the outgoing message of the current buffer.
1043It at first tries the first coding system found in these variables
1044in this order:
1045 (1) local value of `buffer-file-coding-system'
1046 (2) value of `sendmail-coding-system'
b5edd1d1 1047 (3) value of `default-sendmail-coding-system'
b56a5ae0 1048 (4) default value of `buffer-file-coding-system'
46babb23
KH		 1049If the found coding system can't encode the current buffer,
1050or none of them are bound to a coding system,
48e41165 1051it asks the user to select a proper coding system."
46babb23 1052 (let ((coding (or (and (local-variable-p 'buffer-file-coding-system)
b5edd1d1
KH		 1053 buffer-file-coding-system)
1054 sendmail-coding-system
1055 default-sendmail-coding-system
b56a5ae0 1056 (default-value 'buffer-file-coding-system))))
46babb23 1057 (if (eq coding 'no-conversion)
d37ef0f6 1058 ;; We should never use no-conversion for outgoing mail.
46babb23
KH		 1059 (setq coding nil))
1060 (if (fboundp select-safe-coding-system-function)
1061 (funcall select-safe-coding-system-function
b5edd1d1 1062 (point-min) (point-max) coding
209c73b0 1063 (function (lambda (x) (coding-system-get x :mime-charset))))
46babb23 1064 coding)))
4ed46869 1065\f
03c35c83 1066;;; Language support stuff.
4ed46869 1067
4ed46869 1068(defvar language-info-alist nil
2c395d56 1069 "Alist of language environment definitions.
4ed46869
KH		 1070Each element looks like:
1071 (LANGUAGE-NAME . ((KEY . INFO) ...))
2c395d56
RS		 1072where LANGUAGE-NAME is a string, the name of the language environment,
1073KEY is a symbol denoting the kind of information, and
1074INFO is the data associated with KEY.
1075Meaningful values for KEY include
1076
1077 documentation value is documentation of what this language environment
1078 is meant for, and how to use it.
7dd42fb1
KH		 1079 charset value is a list of the character sets mainly used
1080 by this language environment.
d37ef0f6
DL		 1081 sample-text value is an expression which is evalled to generate
1082 a line of text written using characters appropriate
1083 for this language environment.
2c395d56
RS		 1084 setup-function value is a function to call to switch to this
1085 language environment.
1086 exit-function value is a function to call to leave this
1087 language environment.
d0d8885d
JB		 1088 coding-system value is a list of coding systems that are good for
1089 saving text written in this language environment.
2c395d56
RS		 1090 This list serves as suggestions to the user;
1091 in effect, as a kind of documentation.
1092 coding-priority value is a list of coding systems for this language
1093 environment, in order of decreasing priority.
1094 This is used to set up the coding system priority
45d08cb2 1095 list when you switch to this language environment.
ddb5c041 1096 nonascii-translation
7dd42fb1
KH		 1097 value is a charset of dimension one to use for
1098 converting a unibyte character to multibyte
1099 and vice versa.
ddb5c041
KH		 1100 input-method value is a default input method for this language
1101 environment.
7624ebb9
KH		 1102 features value is a list of features requested in this
1103 language environment.
6b61353c 1104 ctext-non-standard-encodings
d0d8885d
JB		 1105 value is a list of non-standard encoding names used
1106 in extended segments of CTEXT. See the variable
1107 `ctext-non-standard-encodings' for more detail.
ddb5c041 1108
b9631bb2 1109The following key takes effect only when multibyte characters are
597e2240 1110globally disabled, i.e. the default value of `enable-multibyte-characters'
b9631bb2 1111is nil (which is an obsolete and deprecated use):
ddb5c041 1112
d0d8885d
JB		 1113 unibyte-display value is a coding system to encode characters for
1114 the terminal. Characters in the range of 160 to
1115 255 display not as octal escapes, but as non-ASCII
1116 characters in this language environment.")
2c395d56
RS		 1117
1118(defun get-language-info (lang-env key)
1119 "Return information listed under KEY for language environment LANG-ENV.
1120KEY is a symbol denoting the kind of information.
1121For a list of useful values for KEY and their meanings,
1122see `language-info-alist'."
1123 (if (symbolp lang-env)
1124 (setq lang-env (symbol-name lang-env)))
f15466c5 1125 (let ((lang-slot (assoc-string lang-env language-info-alist t)))
4ed46869
KH		 1126 (if lang-slot
1127 (cdr (assq key (cdr lang-slot))))))
1128
f08adf27 1129(defun set-language-info (lang-env key info)
2c395d56
RS		 1130 "Modify part of the definition of language environment LANG-ENV.
1131Specifically, this stores the information INFO under KEY
1132in the definition of this language environment.
4ed46869 1133KEY is a symbol denoting the kind of information.
2c395d56 1134INFO is the value for that information.
281d03ec 1135
2c395d56 1136For a list of useful values for KEY and their meanings,
f08adf27 1137see `language-info-alist'."
2c395d56
RS		 1138 (if (symbolp lang-env)
1139 (setq lang-env (symbol-name lang-env)))
eec5c8f9
KH		 1140 (set-language-info-internal lang-env key info)
1141 (if (equal lang-env current-language-environment)
d042f8b4 1142 (cond ((eq key 'coding-priority)
e9a8ed3c
KH		 1143 (set-language-environment-coding-systems lang-env)
1144 (set-language-environment-charset lang-env))
d042f8b4
KH		 1145 ((eq key 'input-method)
1146 (set-language-environment-input-method lang-env))
1147 ((eq key 'nonascii-translation)
1148 (set-language-environment-nonascii-translation lang-env))
1149 ((eq key 'charset)
1150 (set-language-environment-charset lang-env))
597e2240 1151 ((and (not (default-value 'enable-multibyte-characters))
d042f8b4
KH		 1152 (or (eq key 'unibyte-syntax) (eq key 'unibyte-display)))
1153 (set-language-environment-unibyte lang-env)))))
eec5c8f9
KH		 1154
1155(defun set-language-info-internal (lang-env key info)
1156 "Internal use only.
1157Arguments are the same as `set-language-info'."
4ed46869 1158 (let (lang-slot key-slot)
2c395d56 1159 (setq lang-slot (assoc lang-env language-info-alist))
4ed46869 1160 (if (null lang-slot) ; If no slot for the language, add it.
2c395d56 1161 (setq lang-slot (list lang-env)
4ed46869
KH		 1162 language-info-alist (cons lang-slot language-info-alist)))
1163 (setq key-slot (assq key lang-slot))
1164 (if (null key-slot) ; If no slot for the key, add it.
1165 (progn
1166 (setq key-slot (list key))
1167 (setcdr lang-slot (cons key-slot (cdr lang-slot)))))
590dbcba 1168 (setcdr key-slot (purecopy info))
1169bd86 1169 ;; Update the custom-type of `current-language-environment'.
590dbcba 1170 (put 'current-language-environment 'custom-type
1169bd86
MR		 1171 (cons 'choice (mapcar
1172 (lambda (lang)
756e055f
MR		 1173 (list 'const lang))
1174 (sort (mapcar 'car language-info-alist) 'string<))))))
4ed46869 1175
2c395d56
RS		 1176(defun set-language-info-alist (lang-env alist &optional parents)
1177 "Store ALIST as the definition of language environment LANG-ENV.
1178ALIST is an alist of KEY and INFO values. See the documentation of
98c6d6ed 1179`language-info-alist' for the meanings of KEY and INFO.
54b226f7 1180
2c395d56 1181Optional arg PARENTS is a list of parent menu names; it specifies
d37ef0f6 1182where to put this language environment in the
2c395d56
RS		 1183Describe Language Environment and Set Language Environment menus.
1184For example, (\"European\") means to put this language environment
1185in the European submenu in each of those two menus."
905a9ed3
DN		 1186 (cond ((symbolp lang-env)
1187 (setq lang-env (symbol-name lang-env)))
1188 ((stringp lang-env)
1189 (setq lang-env (purecopy lang-env))))
54b226f7
KH		 1190 (let ((describe-map describe-language-environment-map)
1191 (setup-map setup-language-environment-map))
1192 (if parents
1193 (let ((l parents)
9deed82f 1194 map parent-symbol parent prompt)
54b226f7
KH		 1195 (while l
1196 (if (symbolp (setq parent-symbol (car l)))
1197 (setq parent (symbol-name parent))
1198 (setq parent parent-symbol parent-symbol (intern parent)))
1199 (setq map (lookup-key describe-map (vector parent-symbol)))
9deed82f
EZ		 1200 ;; This prompt string is for define-prefix-command, so
1201 ;; that the map it creates will be suitable for a menu.
1202 (or map (setq prompt (format "%s Environment" parent)))
54b226f7
KH		 1203 (if (not map)
1204 (progn
1205 (setq map (intern (format "describe-%s-environment-map"
1206 (downcase parent))))
9deed82f 1207 (define-prefix-command map nil prompt)
54b226f7 1208 (define-key-after describe-map (vector parent-symbol)
64657387 1209 (cons parent map))))
54b226f7
KH		 1210 (setq describe-map (symbol-value map))
1211 (setq map (lookup-key setup-map (vector parent-symbol)))
1212 (if (not map)
1213 (progn
1214 (setq map (intern (format "setup-%s-environment-map"
1215 (downcase parent))))
9deed82f 1216 (define-prefix-command map nil prompt)
54b226f7 1217 (define-key-after setup-map (vector parent-symbol)
64657387 1218 (cons parent map))))
54b226f7
KH		 1219 (setq setup-map (symbol-value map))
1220 (setq l (cdr l)))))
f08adf27
RS		 1221
1222 ;; Set up menu items for this language env.
7624ebb9 1223 (let ((doc (assq 'documentation alist)))
f08adf27
RS		 1224 (when doc
1225 (define-key-after describe-map (vector (intern lang-env))
64657387 1226 (cons lang-env 'describe-specified-language-support))))
7624ebb9 1227 (define-key-after setup-map (vector (intern lang-env))
64657387 1228 (cons lang-env 'setup-specified-language-environment))
f08adf27 1229
eec5c8f9
KH		 1230 (dolist (elt alist)
1231 (set-language-info-internal lang-env (car elt) (cdr elt)))
d35deeec 1232
eec5c8f9
KH		 1233 (if (equal lang-env current-language-environment)
1234 (set-language-environment lang-env))))
4ed46869 1235
ae302641 1236(defun read-language-name (key prompt &optional default)
2c395d56 1237 "Read a language environment name which has information for KEY.
ddb5c041 1238If KEY is nil, read any language environment.
2c395d56
RS		 1239Prompt with PROMPT. DEFAULT is the default choice of language environment.
1240This returns a language environment name as a string."
4ed46869
KH		 1241 (let* ((completion-ignore-case t)
1242 (name (completing-read prompt
1243 language-info-alist
ddb5c041 1244 (and key
ca429a25 1245 (function (lambda (elm) (and (listp elm) (assq key elm)))))
ae302641 1246 t nil nil default)))
13e82c04 1247 (if (and (> (length name) 0)
ddb5c041
KH		 1248 (or (not key)
1249 (get-language-info name key)))
13e82c04 1250 name)))
4ed46869
KH		 1251\f
1252;;; Multilingual input methods.
d37ef0f6 1253(defgroup leim nil
d0c40faa
KH		 1254 "LEIM: Libraries of Emacs Input Methods."
1255 :group 'mule)
4ed46869 1256
d0b9c3ab
KH		 1257(defconst leim-list-file-name "leim-list.el"
1258 "Name of LEIM list file.
1259This file contains a list of libraries of Emacs input methods (LEIM)
1260in the format of Lisp expression for registering each input method.
1261Emacs loads this file at startup time.")
1262
905a9ed3 1263(defconst leim-list-header (format
316732d2 1264";;; %s -- list of LEIM (Library of Emacs Input Method) -*-coding: utf-8;-*-
d0b9c3ab 1265;;
bcd76f45
SM		 1266;; This file is automatically generated.
1267;;
d0b9c3ab 1268;; This file contains a list of LEIM (Library of Emacs Input Method)
c654de1d
DL		 1269;; methods in the same directory as this file. Loading this file
1270;; registers all the input methods in Emacs.
d0b9c3ab 1271;;
d33d5fbe 1272;; Each entry has the form:
d0b9c3ab
KH		 1273;; (register-input-method
1274;; INPUT-METHOD LANGUAGE-NAME ACTIVATE-FUNC
1275;; TITLE DESCRIPTION
1276;; ARG ...)
c654de1d 1277;; See the function `register-input-method' for the meanings of the arguments.
d0b9c3ab 1278;;
bcd76f45 1279;; If this directory is included in `load-path', Emacs automatically
d0b9c3ab
KH		 1280;; loads this file at startup time.
1281
1282"
1283 leim-list-file-name)
1284 "Header to be inserted in LEIM list file.")
1285
905a9ed3 1286(defconst leim-list-entry-regexp "^(register-input-method"
d0b9c3ab 1287 "Regexp matching head of each entry in LEIM list file.
caff3c0a 1288See also the variable `leim-list-header'.")
d0b9c3ab
KH		 1289
1290(defvar update-leim-list-functions
1291 '(quail-update-leim-list-file)
1292 "List of functions to call to update LEIM list file.
1293Each function is called with one arg, LEIM directory name.")
1294
a337fe7f
RS		 1295(defun update-leim-list-file (&rest dirs)
1296 "Update LEIM list file in directories DIRS."
d0d8885d
JB		 1297 (dolist (function update-leim-list-functions)
1298 (apply function dirs)))
d0b9c3ab 1299
4ed46869
KH		 1300(defvar current-input-method nil
1301 "The current input method for multilingual text.
96db204a 1302If nil, that means no input method is activated now.")
4ed46869
KH		 1303(make-variable-buffer-local 'current-input-method)
1304(put 'current-input-method 'permanent-local t)
1305
1306(defvar current-input-method-title nil
d0b9c3ab 1307 "Title string of the current input method shown in mode line.")
4ed46869
KH		 1308(make-variable-buffer-local 'current-input-method-title)
1309(put 'current-input-method-title 'permanent-local t)
1310
8c24b7f6
SM		 1311(define-widget 'mule-input-method-string 'string
1312 "String widget with completion for input method."
1313 :completions
1314 (lambda (string pred action)
1315 (let ((completion-ignore-case t))
1316 (complete-with-action action input-method-alist string pred)))
1317 :prompt-history 'input-method-history)
1318
b4fba33f 1319(defcustom default-input-method nil
caff3c0a 1320 "Default input method for multilingual text (a string).
b4fba33f 1321This is the input method activated automatically by the command
9b10b5a3 1322`toggle-input-method' (\\[toggle-input-method])."
1f547b92 1323 :link '(custom-manual "(emacs)Input Methods")
8861c593 1324 :group 'mule
8c24b7f6
SM		 1325 :type `(choice (const nil)
1326 mule-input-method-string)
5806e8a6 1327 :set-after '(current-language-environment))
b4fba33f 1328
0f835e87
KH		 1329(put 'input-method-function 'permanent-local t)
1330
723a427a 1331(defvar input-method-history nil
6f5d2452
EZ		 1332 "History list of input methods read from the minibuffer.
1333
1334Maximum length of the history list is determined by the value
1335of `history-length', which see.")
723a427a
KH		 1336(make-variable-buffer-local 'input-method-history)
1337(put 'input-method-history 'permanent-local t)
4ed46869 1338
e098de97
PE		 1339(define-obsolete-variable-alias
1340 'inactivate-current-input-method-function
2a1e2476 1341 'deactivate-current-input-method-function "24.3")
72b255c7
PE		 1342(defvar deactivate-current-input-method-function nil
1343 "Function to call for deactivating the current input method.
4ed46869 1344Every input method should set this to an appropriate value when activated.
f17ccaee
KH		 1345This function is called with no argument.
1346
1347This function should never change the value of `current-input-method'.
72b255c7
PE		 1348It is set to nil by the function `deactivate-input-method'.")
1349(make-variable-buffer-local 'deactivate-current-input-method-function)
1350(put 'deactivate-current-input-method-function 'permanent-local t)
4ed46869
KH		 1351
1352(defvar describe-current-input-method-function nil
1353 "Function to call for describing the current input method.
1354This function is called with no argument.")
1355(make-variable-buffer-local 'describe-current-input-method-function)
1356(put 'describe-current-input-method-function 'permanent-local t)
1357
d0b9c3ab 1358(defvar input-method-alist nil
2c395d56 1359 "Alist of input method names vs how to use them.
d0b9c3ab 1360Each element has the form:
2c395d56
RS		 1361 (INPUT-METHOD LANGUAGE-ENV ACTIVATE-FUNC TITLE DESCRIPTION ARGS...)
1362See the function `register-input-method' for the meanings of the elements.")
ac549fa5 1363;;;###autoload
6dc3311d 1364(put 'input-method-alist 'risky-local-variable t)
2c395d56 1365
f08adf27 1366(defun register-input-method (input-method lang-env &rest args)
bf42aa15 1367 "Register INPUT-METHOD as an input method for language environment LANG-ENV.
d0b9c3ab 1368
d35deeec 1369INPUT-METHOD and LANG-ENV are symbols or strings.
2c395d56
RS		 1370ACTIVATE-FUNC is a function to call to activate this method.
1371TITLE is a string to show in the mode line when this method is active.
1372DESCRIPTION is a string describing this method and what it is good for.
1373The ARGS, if any, are passed as arguments to ACTIVATE-FUNC.
205814ee
KH		 1374All told, the arguments to ACTIVATE-FUNC are INPUT-METHOD and the ARGS.
1375
c654de1d
DL		 1376This function is mainly used in the file \"leim-list.el\" which is
1377created at Emacs build time, registering all Quail input methods
0b6cadff 1378contained in the Emacs distribution.
205814ee 1379
0b6cadff 1380In case you want to register a new Quail input method by yourself, be
205814ee 1381careful to use the same input method title as given in the third
0b6cadff
DL		 1382parameter of `quail-define-package'. (If the values are different, the
1383string specified in this function takes precedence.)
205814ee
KH		 1384
1385The commands `describe-input-method' and `list-input-methods' need
0b6cadff 1386these duplicated values to show some information about input methods
d35deeec
JB		 1387without loading the relevant Quail packages.
1388\n(fn INPUT-METHOD LANG-ENV ACTIVATE-FUNC TITLE DESCRIPTION &rest ARGS)"
f08adf27 1389 (if (symbolp lang-env)
ebfa10d3
DN		 1390 (setq lang-env (symbol-name lang-env))
1391 (setq lang-env (purecopy lang-env)))
4ef06f75 1392 (if (symbolp input-method)
ebfa10d3
DN		 1393 (setq input-method (symbol-name input-method))
1394 (setq input-method (purecopy input-method)))
1395 (setq args (mapcar 'purecopy args))
f08adf27 1396 (let ((info (cons lang-env args))
d0b9c3ab
KH		 1397 (slot (assoc input-method input-method-alist)))
1398 (if slot
1399 (setcdr slot info)
1400 (setq slot (cons input-method info))
1401 (setq input-method-alist (cons slot input-method-alist)))))
1402
4d5ac029 1403(defun read-input-method-name (prompt &optional default inhibit-null)
d0b9c3ab 1404 "Read a name of input method from a minibuffer prompting with PROMPT.
4d5ac029 1405If DEFAULT is non-nil, use that as the default,
0b6cadff 1406and substitute it into PROMPT at the first `%s'.
4ef06f75
KH		 1407If INHIBIT-NULL is non-nil, null input signals an error.
1408
1409The return value is a string."
4d5ac029
RS		 1410 (if default
1411 (setq prompt (format prompt default)))
d0b9c3ab 1412 (let* ((completion-ignore-case t)
c54044ff
KH		 1413 ;; As it is quite normal to change input method in the
1414 ;; minibuffer, we must enable it even if
1415 ;; enable-recursive-minibuffers is currently nil.
1416 (enable-recursive-minibuffers t)
723a427a
KH		 1417 ;; This binding is necessary because input-method-history is
1418 ;; buffer local.
d0b9c3ab 1419 (input-method (completing-read prompt input-method-alist
87505a98
RS		 1420 nil t nil 'input-method-history
1421 default)))
bf294e6e
KH		 1422 (if (and input-method (symbolp input-method))
1423 (setq input-method (symbol-name input-method)))
d0b9c3ab
KH		 1424 (if (> (length input-method) 0)
1425 input-method
1426 (if inhibit-null
43807b77 1427 (error "No valid input method is specified")))))
d0b9c3ab 1428
d0b9c3ab 1429(defun activate-input-method (input-method)
2c395d56
RS		 1430 "Switch to input method INPUT-METHOD for the current buffer.
1431If some other input method is already active, turn it off first.
1432If INPUT-METHOD is nil, deactivate any current input method."
305a3cb6 1433 (if (and input-method (symbolp input-method))
4ef06f75 1434 (setq input-method (symbol-name input-method)))
723a427a
KH		 1435 (if (and current-input-method
1436 (not (string= current-input-method input-method)))
72b255c7 1437 (deactivate-input-method))
2c395d56 1438 (unless (or current-input-method (null input-method))
d0b9c3ab
KH		 1439 (let ((slot (assoc input-method input-method-alist)))
1440 (if (null slot)
723a427a 1441 (error "Can't activate input method `%s'" input-method))
278dd6ac 1442 (setq current-input-method-title nil)
8efc03e1
KH		 1443 (let ((func (nth 2 slot)))
1444 (if (functionp func)
1445 (apply (nth 2 slot) input-method (nthcdr 5 slot))
1446 (if (and (consp func) (symbolp (car func)) (symbolp (cdr func)))
1447 (progn
1448 (require (cdr func))
1449 (apply (car func) input-method (nthcdr 5 slot)))
1450 (error "Can't activate input method `%s'" input-method))))
d0b9c3ab 1451 (setq current-input-method input-method)
278dd6ac
KH		 1452 (or (stringp current-input-method-title)
1453 (setq current-input-method-title (nth 3 slot)))
28885c0e
KH		 1454 (unwind-protect
1455 (run-hooks 'input-method-activate-hook)
1456 (force-mode-line-update)))))
15b3e511 1457
72b255c7 1458(defun deactivate-input-method ()
f17ccaee 1459 "Turn off the current input method."
723a427a
KH		 1460 (when current-input-method
1461 (if input-method-history
1462 (unless (string= current-input-method (car input-method-history))
1463 (setq input-method-history
1464 (cons current-input-method
1465 (delete current-input-method input-method-history))))
1466 (setq input-method-history (list current-input-method)))
1467 (unwind-protect
36f8618c
KH		 1468 (progn
1469 (setq input-method-function nil
1470 current-input-method-title nil)
72b255c7 1471 (funcall deactivate-current-input-method-function))
15b3e511 1472 (unwind-protect
72b255c7
PE		 1473 (run-hooks
1474 'input-method-inactivate-hook ; for backward compatibility
1475 'input-method-deactivate-hook)
36f8618c 1476 (setq current-input-method nil)
28885c0e 1477 (force-mode-line-update)))))
4ed46869 1478
bb6eb9fc
PE		 1479(define-obsolete-function-alias
1480 'inactivate-input-method
2a1e2476 1481 'deactivate-input-method "24.3")
bb6eb9fc 1482
e893eae2 1483(defun set-input-method (input-method &optional interactive)
2c395d56 1484 "Select and activate input method INPUT-METHOD for the current buffer.
bc406911 1485This also sets the default input method to the one you specify.
402dbbd1
EZ		 1486If INPUT-METHOD is nil, this function turns off the input method, and
1487also causes you to be prompted for a name of an input method the next
1488time you invoke \\[toggle-input-method].
e893eae2
RS		 1489When called interactively, the optional arg INTERACTIVE is non-nil,
1490which marks the variable `default-input-method' as set for Custom buffers.
402dbbd1 1491
bc406911 1492To deactivate the input method interactively, use \\[toggle-input-method].
72b255c7 1493To deactivate it programmatically, use `deactivate-input-method'."
d0b9c3ab 1494 (interactive
723a427a 1495 (let* ((default (or (car input-method-history) default-input-method)))
42395763 1496 (list (read-input-method-name
87505a98 1497 (if default "Select input method (default %s): " "Select input method: ")
e893eae2
RS		 1498 default t)
1499 t)))
d0b9c3ab 1500 (activate-input-method input-method)
f4990970 1501 (setq default-input-method input-method)
e893eae2 1502 (when interactive
f4990970
PA		 1503 (customize-mark-as-set 'default-input-method))
1504 default-input-method)
4ed46869 1505
023df4cf
RS		 1506(defvar toggle-input-method-active nil
1507 "Non-nil inside `toggle-input-method'.")
1508
e893eae2 1509(defun toggle-input-method (&optional arg interactive)
f8ec20be
RS		 1510 "Enable or disable multilingual text input method for the current buffer.
1511Only one input method can be enabled at any time in a given buffer.
1512
d0d8885d
JB		 1513The normal action is to enable an input method if none was enabled,
1514and disable the current one otherwise. Which input method to enable
1515can be determined in various ways--either the one most recently used,
1516or the one specified by `default-input-method', or as a last resort
1517by reading the name of an input method in the minibuffer.
f8ec20be 1518
d0d8885d 1519With a prefix argument ARG, read an input method name with the minibuffer
f8ec20be 1520and enable that one. The default is the most recent input method specified
e893eae2 1521\(not including the currently active input method, if any).
f8ec20be 1522
d0d8885d 1523When called interactively, the optional argument INTERACTIVE is non-nil,
e893eae2
RS		 1524which marks the variable `default-input-method' as set for Custom buffers."
1525
1526 (interactive "P\np")
023df4cf
RS		 1527 (if toggle-input-method-active
1528 (error "Recursive use of `toggle-input-method'"))
7ddbb5bc 1529 (if (and current-input-method (not arg))
72b255c7 1530 (deactivate-input-method)
023df4cf
RS		 1531 (let ((toggle-input-method-active t)
1532 (default (or (car input-method-history) default-input-method)))
7ddbb5bc
RS		 1533 (if (and arg default (equal current-input-method default)
1534 (> (length input-method-history) 1))
1535 (setq default (nth 1 input-method-history)))
723a427a
KH		 1536 (activate-input-method
1537 (if (or arg (not default))
7ddbb5bc
RS		 1538 (progn
1539 (read-input-method-name
1540 (if default "Input method (default %s): " "Input method: " )
1541 default t))
723a427a 1542 default))
f4990970 1543 (unless default-input-method
d37ef0f6 1544 (prog1
f4990970 1545 (setq default-input-method current-input-method)
e893eae2 1546 (when interactive
f4990970 1547 (customize-mark-as-set 'default-input-method)))))))
d0b9c3ab 1548
26b3dce6 1549(autoload 'help-buffer "help-mode")
0855c6cd 1550
d0b9c3ab 1551(defun describe-input-method (input-method)
2c395d56 1552 "Describe input method INPUT-METHOD."
d0b9c3ab
KH		 1553 (interactive
1554 (list (read-input-method-name
5b76833f 1555 "Describe input method (default current choice): ")))
78754934 1556 (if (and input-method (symbolp input-method))
4ef06f75 1557 (setq input-method (symbol-name input-method)))
43125c28
RS		 1558 (help-setup-xref (list #'describe-input-method
1559 (or input-method current-input-method))
32226619 1560 (called-interactively-p 'interactive))
f80e2142 1561
d0b9c3ab
KH		 1562 (if (null input-method)
1563 (describe-current-input-method)
464cc130
KH		 1564 (let ((current current-input-method))
1565 (condition-case nil
1566 (progn
1567 (save-excursion
1568 (activate-input-method input-method)
1569 (describe-current-input-method))
1570 (activate-input-method current))
d37ef0f6 1571 (error
464cc130 1572 (activate-input-method current)
5f395df3 1573 (help-setup-xref (list #'describe-input-method input-method)
32226619 1574 (called-interactively-p 'interactive))
5f395df3 1575 (with-output-to-temp-buffer (help-buffer)
464cc130
KH		 1576 (let ((elt (assoc input-method input-method-alist)))
1577 (princ (format
1578 "Input method: %s (`%s' in mode line) for %s\n %s\n"
1579 input-method (nth 3 elt) (nth 1 elt) (nth 4 elt))))))))))
d0b9c3ab
KH		 1580
1581(defun describe-current-input-method ()
f80e2142
RS		 1582 "Describe the input method currently in use.
1583This is a subroutine for `describe-input-method'."
4ed46869
KH		 1584 (if current-input-method
1585 (if (and (symbolp describe-current-input-method-function)
1586 (fboundp describe-current-input-method-function))
1587 (funcall describe-current-input-method-function)
1588 (message "No way to describe the current input method `%s'"
f2979bdb 1589 current-input-method)
4ed46869 1590 (ding))
d0b9c3ab 1591 (error "No input method is activated now")))
4ed46869 1592
d3459641 1593(defun read-multilingual-string (prompt &optional initial-input input-method)
4ed46869
KH		 1594 "Read a multilingual string from minibuffer, prompting with string PROMPT.
1595The input method selected last time is activated in minibuffer.
d0d8885d
JB		 1596If optional second argument INITIAL-INPUT is non-nil, insert it in the
1597minibuffer initially.
1598Optional 3rd argument INPUT-METHOD specifies the input method to be activated
1599instead of the one selected last time. It is a symbol or a string."
88d559ec
KH		 1600 (setq input-method
1601 (or input-method
d3459641 1602 current-input-method
88d559ec
KH		 1603 default-input-method
1604 (read-input-method-name "Input method: " nil t)))
3df60841 1605 (if (and input-method (symbolp input-method))
4ef06f75 1606 (setq input-method (symbol-name input-method)))
305a3cb6
KH		 1607 (let ((prev-input-method current-input-method))
1608 (unwind-protect
1609 (progn
1610 (activate-input-method input-method)
1611 (read-string prompt initial-input nil nil t))
1612 (activate-input-method prev-input-method))))
4ed46869
KH		 1613
1614;; Variables to control behavior of input methods. All input methods
1615;; should react to these variables.
1616
8efc03e1 1617(defcustom input-method-verbose-flag 'default
caff3c0a 1618 "A flag to control extra guidance given by input methods.
8efc03e1 1619The value should be nil, t, `complex-only', or `default'.
4ed46869 1620
cb29dfb6 1621The extra guidance is done by showing list of available keys in echo
8efc03e1
KH		 1622area. When you use the input method in the minibuffer, the guidance
1623is shown at the bottom short window (split from the existing window).
c27c4ed8 1624
8efc03e1
KH		 1625If the value is t, extra guidance is always given, if the value is
1626nil, extra guidance is always suppressed.
1627
1628If the value is `complex-only', only complex input methods such as
1629`chinese-py' and `japanese' give extra guidance.
1630
1631If the value is `default', complex input methods always give extra
1632guidance, but simple input methods give it only when you are not in
1633the minibuffer.
1634
1635See also the variable `input-method-highlight-flag'."
d37ef0f6
DL		 1636 :type '(choice (const :tag "Always" t) (const :tag "Never" nil)
1637 (const complex-only) (const default))
42395763
RS		 1638 :group 'mule)
1639
1640(defcustom input-method-highlight-flag t
caff3c0a 1641 "If this flag is non-nil, input methods highlight partially-entered text.
42395763
RS		 1642For instance, while you are in the middle of a Quail input method sequence,
1643the text inserted so far is temporarily underlined.
8efc03e1
KH		 1644The underlining goes away when you finish or abort the input method sequence.
1645See also the variable `input-method-verbose-flag'."
42395763
RS		 1646 :type 'boolean
1647 :group 'mule)
4ed46869 1648
1f547b92 1649(defcustom input-method-activate-hook nil
f17ccaee
KH		 1650 "Normal hook run just after an input method is activated.
1651
1652The variable `current-input-method' keeps the input method name
1f547b92
DL		 1653just activated."
1654 :type 'hook
1655 :group 'mule)
4ed46869 1656
72b255c7
PE		 1657(define-obsolete-variable-alias
1658 'input-method-inactivate-hook
2a1e2476 1659 'input-method-deactivate-hook "24.3")
72b255c7
PE		 1660
1661(defcustom input-method-deactivate-hook nil
1662 "Normal hook run just after an input method is deactivated.
f17ccaee
KH		 1663
1664The variable `current-input-method' still keeps the input method name
72b255c7 1665just deactivated."
1f547b92 1666 :type 'hook
72b255c7 1667 :group 'mule
2a1e2476 1668 :version "24.3")
4ed46869 1669
1f547b92 1670(defcustom input-method-after-insert-chunk-hook nil
8f924df7 1671 "Normal hook run just after an input method insert some chunk of text."
1f547b92
DL		 1672 :type 'hook
1673 :group 'mule)
4ed46869 1674
dccca980 1675(defvar input-method-exit-on-first-char nil
0b6cadff 1676 "This flag controls when an input method returns.
dccca980
KH		 1677Usually, the input method does not return while there's a possibility
1678that it may find a different translation if a user types another key.
d0d8885d
JB		 1679But, if this flag is non-nil, the input method returns as soon as the
1680current key sequence gets long enough to have some valid translation.")
dccca980 1681
1f547b92 1682(defcustom input-method-use-echo-area nil
dccca980 1683 "This flag controls how an input method shows an intermediate key sequence.
39e643e2
RS		 1684Usually, the input method inserts the intermediate key sequence,
1685or candidate translations corresponding to the sequence,
1686at point in the current buffer.
1f547b92
DL		 1687But, if this flag is non-nil, it displays them in echo area instead."
1688 :type 'hook
1689 :group 'mule)
dccca980 1690
723a427a 1691(defvar input-method-exit-on-invalid-key nil
fea6b736 1692 "This flag controls the behavior of an input method on invalid key input.
723a427a
KH		 1693Usually, when a user types a key which doesn't start any character
1694handled by the input method, the key is handled by turning off the
e8dd0160 1695input method temporarily. After that key, the input method is re-enabled.
723a427a
KH		 1696But, if this flag is non-nil, the input method is never back on.")
1697
4ed46869 1698\f
1f547b92 1699(defcustom set-language-environment-hook nil
8efc03e1
KH		 1700 "Normal hook run after some language environment is set.
1701
1702When you set some hook function here, that effect usually should not
1703be inherited to another language environment. So, you had better set
1704another function in `exit-language-environment-hook' (which see) to
1f547b92
DL		 1705cancel the effect."
1706 :type 'hook
1707 :group 'mule)
8efc03e1 1708
1f547b92 1709(defcustom exit-language-environment-hook nil
8efc03e1
KH		 1710 "Normal hook run after exiting from some language environment.
1711When this hook is run, the variable `current-language-environment'
1712is still bound to the language environment being exited.
1713
e8dd0160 1714This hook is mainly used for canceling the effect of
caff3c0a 1715`set-language-environment-hook' (which see)."
1f547b92
DL		 1716 :type 'hook
1717 :group 'mule)
8efc03e1 1718
b0648a00
RS		 1719(put 'setup-specified-language-environment 'apropos-inhibit t)
1720
15b3e511 1721(defun setup-specified-language-environment ()
f08adf27 1722 "Switch to a specified language environment."
15b3e511 1723 (interactive)
f850d782 1724 (let (language-name)
15b3e511
KH		 1725 (if (and (symbolp last-command-event)
1726 (or (not (eq last-command-event 'Default))
1727 (setq last-command-event 'English))
f850d782 1728 (setq language-name (symbol-name last-command-event)))
f4990970
PA		 1729 (prog1
1730 (set-language-environment language-name)
1731 (customize-mark-as-set 'current-language-environment))
15b3e511 1732 (error "Bogus calling sequence"))))
4ed46869 1733
8861c593 1734(defcustom current-language-environment "English"
94d04df6 1735 "The last language environment specified with `set-language-environment'.
ebef6d93
KH		 1736This variable should be set only with \\[customize], which is equivalent
1737to using the function `set-language-environment'."
94d04df6 1738 :link '(custom-manual "(emacs)Language Environments")
dff1aa24 1739 :set (lambda (symbol value) (set-language-environment value))
94d04df6 1740 :get (lambda (x)
f15466c5 1741 (or (car-safe (assoc-string
94d04df6
DL		 1742 (if (symbolp current-language-environment)
1743 (symbol-name current-language-environment)
1744 current-language-environment)
f15466c5 1745 language-info-alist t))
94d04df6 1746 "English"))
990a4108
MR		 1747 ;; custom type will be updated with `set-language-info'.
1748 :type (if language-info-alist
1749 (cons 'choice (mapcar
1750 (lambda (lang)
756e055f
MR		 1751 (list 'const lang))
1752 (sort (mapcar 'car language-info-alist) 'string<)))
990a4108 1753 'string)
8861c593 1754 :initialize 'custom-initialize-default
eb9fc9e6 1755 :group 'mule)
f850d782 1756
ddb5c041
KH		 1757(defun reset-language-environment ()
1758 "Reset multilingual environment of Emacs to the default status.
1759
1760The default status is as follows:
1761
d37ef0f6 1762 The default value of `buffer-file-coding-system' is nil.
ddb5c041
KH		 1763 The default coding system for process I/O is nil.
1764 The default value for the command `set-terminal-coding-system' is nil.
1765 The default value for the command `set-keyboard-coding-system' is nil.
1766
97941b05
KH		 1767 The order of priorities of coding systems are as follows:
1768 utf-8
1769 iso-2022-7bit
1770 iso-latin-1
1771 iso-2022-7bit-lock
1772 iso-2022-8bit-ss2
1773 emacs-mule
1774 raw-text"
ddb5c041
KH		 1775 (interactive)
1776 ;; This function formerly set default-enable-multibyte-characters to t,
1777 ;; but that is incorrect. It should not alter the unibyte/multibyte choice.
1778
97941b05
KH		 1779 (set-coding-system-priority
1780 'utf-8
1781 'iso-2022-7bit
1782 'iso-latin-1
1783 'iso-2022-7bit-lock
1784 'iso-2022-8bit-ss2
1785 'emacs-mule
1786 'raw-text)
91693d18 1787
ddb5c041 1788 (set-default-coding-systems nil)
b5edd1d1 1789 (setq default-sendmail-coding-system 'iso-latin-1)
a41118cc
SM		 1790 ;; On Darwin systems, this should be utf-8, but when this file is loaded
1791 ;; utf-8 is not yet defined, so we set it in set-locale-environment instead.
787caf99 1792 (setq default-file-name-coding-system 'iso-latin-1)
1d77e15a
JR		 1793 ;; Preserve eol-type from existing default-process-coding-systems.
1794 ;; On non-unix-like systems in particular, these may have been set
1795 ;; carefully by the user, or by the startup code, to deal with the
1796 ;; users shell appropriately, so should not be altered by changing
1797 ;; language environment.
1798 (let ((output-coding
a099a2ff
JR		 1799 ;; When bootstrapping, coding-systems are not defined yet, so
1800 ;; we need to catch the error from check-coding-system.
d37ef0f6 1801 (condition-case nil
a099a2ff
JR		 1802 (coding-system-change-text-conversion
1803 (car default-process-coding-system) 'undecided)
1804 (coding-system-error 'undecided)))
1d77e15a 1805 (input-coding
a099a2ff
JR		 1806 (condition-case nil
1807 (coding-system-change-text-conversion
1808 (cdr default-process-coding-system) 'iso-latin-1)
1809 (coding-system-error 'iso-latin-1))))
1d77e15a
JR		 1810 (setq default-process-coding-system
1811 (cons output-coding input-coding)))
b5edd1d1 1812
a43977db
KH		 1813 ;; Put the highest priority to the charset iso-8859-1 to prefer the
1814 ;; registry iso8859-1 over iso8859-2 in font selection. It also
1815 ;; makes unibyte-display-via-language-environment to use iso-8859-1
1816 ;; as the unibyte charset.
1817 (set-charset-priority 'iso-8859-1)
1818
ddb5c041
KH		 1819 ;; Don't alter the terminal and keyboard coding systems here.
1820 ;; The terminal still supports the same coding system
1821 ;; that it supported a minute ago.
1b8dc791
SM		 1822 ;; (set-terminal-coding-system-internal nil)
1823 ;; (set-keyboard-coding-system-internal nil)
ddb5c041 1824
33de15f4
SM		 1825 ;; Back in Emacs-20, it was necessary to provide some fallback implicit
1826 ;; conversion, because almost no packages handled coding-system issues.
1827 ;; Nowadays it'd just paper over bugs.
1828 ;; (set-unibyte-charset 'iso-8859-1)
1829 )
ddb5c041 1830
0c47a7c8
KH		 1831(reset-language-environment)
1832
97c57fb2 1833(defun set-display-table-and-terminal-coding-system (language-name &optional coding-system display)
40c81f74
PE		 1834 "Set up the display table and terminal coding system for LANGUAGE-NAME."
1835 (let ((coding (get-language-info language-name 'unibyte-display)))
735b7c87
KH		 1836 (if (and coding
1837 (or (not coding-system)
1838 (coding-system-equal coding coding-system)))
40c81f74 1839 (standard-display-european-internal)
a67ae60e
EZ		 1840 ;; The following 2 lines undo the 8-bit display that we set up
1841 ;; in standard-display-european-internal, which see. This is in
1842 ;; case the user has used standard-display-european earlier in
970c9391 1843 ;; this session.
d73a7bb8
JPW		 1844 (when standard-display-table
1845 (dotimes (i 128)
b221615b 1846 (aset standard-display-table (+ i 128) nil))))
970c9391 1847 (set-terminal-coding-system (or coding-system coding) display)))
40c81f74 1848
166246f7 1849(defun set-language-environment (language-name)
9c758578 1850 "Set up multilingual environment for using LANGUAGE-NAME.
6c05d680 1851This sets the coding system priority and the default input method
8861c593
RS		 1852and sometimes other things. LANGUAGE-NAME should be a string
1853which is the name of a language environment. For example, \"Latin-1\"
ca7d96c8
GM		 1854specifies the character set for the major languages of Western Europe.
1855
1856If there is a prior value for `current-language-environment', this
1857runs the hook `exit-language-environment-hook'. After setting up
1858the new language environment, it runs `set-language-environment-hook'."
8efc03e1 1859 (interactive (list (read-language-name
ddb5c041 1860 nil
5b76833f 1861 "Set language environment (default English): ")))
4ef06f75
KH		 1862 (if language-name
1863 (if (symbolp language-name)
1864 (setq language-name (symbol-name language-name)))
1865 (setq language-name "English"))
f15466c5 1866 (let ((slot (assoc-string language-name language-info-alist t)))
95498fd0 1867 (unless slot
f850d782 1868 (error "Language environment not defined: %S" language-name))
95498fd0 1869 (setq language-name (car slot)))
8efc03e1
KH		 1870 (if current-language-environment
1871 (let ((func (get-language-info current-language-environment
1872 'exit-function)))
e63645c2 1873 (run-hooks 'exit-language-environment-hook)
5f395df3 1874 (if (functionp func) (funcall func))))
ddb5c041 1875
e9a8ed3c
KH		 1876 (reset-language-environment)
1877 ;; The features might set up coding systems.
ddb5c041
KH		 1878 (let ((required-features (get-language-info language-name 'features)))
1879 (while required-features
1880 (require (car required-features))
1881 (setq required-features (cdr required-features))))
6b61353c 1882
e9a8ed3c
KH		 1883 (setq current-language-environment language-name)
1884
1885 (set-language-environment-coding-systems language-name)
1886 (set-language-environment-input-method language-name)
1887 (set-language-environment-nonascii-translation language-name)
1888 (set-language-environment-charset language-name)
1889 ;; Unibyte setups if necessary.
597e2240 1890 (unless (default-value 'enable-multibyte-characters)
e9a8ed3c
KH		 1891 (set-language-environment-unibyte language-name))
1892
ddb5c041 1893 (let ((func (get-language-info language-name 'setup-function)))
5f395df3 1894 (if (functionp func)
ddb5c041 1895 (funcall func)))
e9a8ed3c 1896
8aeebac2 1897 (setq current-iso639-language
e9aaa1db
KH		 1898 (or (get-language-info language-name 'iso639-language)
1899 current-iso639-language))
8aeebac2 1900
8efc03e1 1901 (run-hooks 'set-language-environment-hook)
f850d782 1902 (force-mode-line-update t))
4ed46869 1903
c3869589 1904(define-widget 'charset 'symbol
aaa448c9 1905 "An Emacs charset."
1644d5b9 1906 :tag "Charset"
8c24b7f6
SM		 1907 :completions
1908 (lambda (string pred action)
1909 (let ((completion-ignore-case t))
1910 (completion-table-with-predicate
1911 obarray #'charsetp 'strict string pred action)))
c3869589
DL		 1912 :value 'ascii
1913 :validate (lambda (widget)
1914 (unless (charsetp (widget-value widget))
1915 (widget-put widget :error (format "Invalid charset: %S"
1916 (widget-value widget)))
1917 widget))
1918 :prompt-history 'charset-history)
1919
1920(defcustom language-info-custom-alist nil
1921 "Customizations of language environment parameters.
1922Value is an alist with elements like those of `language-info-alist'.
1923These are used to set values in `language-info-alist' which replace
1924the defaults. A typical use is replacing the default input method for
1925the environment. Use \\[describe-language-environment] to find the environment's settings.
1926
1927This option is intended for use at startup. Removing items doesn't
1928remove them from the language info until you next restart Emacs.
1929
d0d8885d
JB		 1930Setting this variable directly does not take effect.
1931See `set-language-info-alist' for use in programs."
c3869589 1932 :group 'mule
8589dc17 1933 :version "23.1"
c3869589
DL		 1934 :set (lambda (s v)
1935 (custom-set-default s v)
1936 ;; Can't do this before language environments are set up.
1937 (when v
1938 ;; modify language-info-alist
1939 (dolist (elt v)
1940 (set-language-info-alist (car elt) (cdr elt)))
1941 ;; re-set the environment in case its parameters changed
1942 (set-language-environment current-language-environment)))
1943 :type `(alist
1944 :key-type (string :tag "Language environment"
bcd70d97 1945 :completions
8c24b7f6
SM		 1946 (lambda (string pred action)
1947 (let ((completion-ignore-case t))
1948 (complete-with-action
1949 action language-info-alist string pred))))
c3869589
DL		 1950 :value-type
1951 (alist :key-type symbol
1952 :options ((documentation string)
1953 (charset (repeat charset))
1954 (sample-text string)
1955 (setup-function function)
1956 (exit-function function)
1957 (coding-system (repeat coding-system))
1958 (coding-priority (repeat coding-system))
1959 (nonascii-translation charset)
8c24b7f6 1960 (input-method mule-input-method-string)
c3869589
DL		 1961 (features (repeat symbol))
1962 (unibyte-display coding-system)))))
1963
aa360da1
GM		 1964(declare-function x-server-vendor "xfns.c" (&optional terminal))
1965(declare-function x-server-version "xfns.c" (&optional terminal))
1966
51a8fc1d
RS		 1967(defun standard-display-european-internal ()
1968 ;; Actually set up direct output of non-ASCII characters.
03c35c83
EZ		 1969 (standard-display-8bit (if (eq window-system 'pc) 128 160) 255)
1970 ;; Unibyte Emacs on MS-DOS wants to display all 8-bit characters with
1971 ;; the native font, and codes 160 and 146 stand for something very
1972 ;; different there.
597e2240
GM		 1973 (or (and (eq window-system 'pc) (not (default-value
1974 'enable-multibyte-characters)))
03c35c83 1975 (progn
6b626913
SM		 1976 ;; Most X fonts used to do the wrong thing for latin-1 code 160.
1977 (unless (and (eq window-system 'x)
1978 ;; XFree86 4 has fixed the fonts.
1979 (string= "The XFree86 Project, Inc" (x-server-vendor))
1980 (> (aref (number-to-string (nth 2 (x-server-version))) 0)
1981 ?3))
1982 ;; Make non-line-break space display as a plain space.
d419e1d9 1983 (aset standard-display-table (unibyte-char-to-multibyte 160) [32]))
1492f7ac 1984 ;; Most Windows programs send out apostrophes as \222. Most X fonts
03c35c83 1985 ;; don't contain a character at that position. Map it to the ASCII
5f395df3
SM		 1986 ;; apostrophe. [This is actually RIGHT SINGLE QUOTATION MARK,
1987 ;; U+2019, normally from the windows-1252 character set. XFree 4
1988 ;; fonts probably have the appropriate glyph at this position,
1989 ;; so they could use standard-display-8bit. It's better to use a
1990 ;; proper windows-1252 coding system. --fx]
d419e1d9 1991 (aset standard-display-table (unibyte-char-to-multibyte 146) [39]))))
03c35c83 1992
e9a8ed3c
KH		 1993(defun set-language-environment-coding-systems (language-name)
1994 "Do various coding system setups for language environment LANGUAGE-NAME."
54b226f7 1995 (let* ((priority (get-language-info language-name 'coding-priority))
e9a8ed3c 1996 (default-coding (car priority))
b56a5ae0 1997 ;; If the default buffer-file-coding-system is nil, don't use
207422da 1998 ;; coding-system-eol-type, because it treats nil as
b56a5ae0 1999 ;; `no-conversion'. The default buffer-file-coding-system is set
207422da
EZ		 2000 ;; to nil by reset-language-environment, and in that case we
2001 ;; want to have here the native EOL type for each platform.
2002 ;; FIXME: there should be a common code that runs both on
2003 ;; startup and here to set the default EOL type correctly.
2004 ;; Right now, DOS/Windows platforms set this on dos-w32.el,
2005 ;; which works only as long as the order of loading files at
2006 ;; dump time and calling functions at startup is not modified
2007 ;; significantly, i.e. as long as this function is called
b56a5ae0 2008 ;; _after_ the default buffer-file-coding-system was set by
207422da
EZ		 2009 ;; dos-w32.el.
2010 (eol-type
b56a5ae0
SM		 2011 (coding-system-eol-type
2012 (or (default-value 'buffer-file-coding-system)
2013 (if (memq system-type '(windows-nt ms-dos)) 'dos 'unix)))))
97941b05
KH		 2014 (when priority
2015 (set-default-coding-systems
2016 (if (memq eol-type '(0 1 2 unix dos mac))
2017 (coding-system-change-eol-conversion default-coding eol-type)
2018 default-coding))
2019 (setq default-sendmail-coding-system default-coding)
2020 (apply 'set-coding-system-priority priority))))
54b226f7 2021
d042f8b4
KH		 2022(defun set-language-environment-input-method (language-name)
2023 "Do various input method setups for language environment LANGUAGE-NAME."
2024 (let ((input-method (get-language-info language-name 'input-method)))
2025 (when input-method
2026 (setq default-input-method input-method)
2027 (if input-method-history
2028 (setq input-method-history
2029 (cons input-method
2030 (delete input-method input-method-history)))))))
2031
2032(defun set-language-environment-nonascii-translation (language-name)
2033 "Do unibyte/multibyte translation setup for language environment LANGUAGE-NAME."
e9a8ed3c
KH		 2034 ;; Note: For DOS, we assumed that the charset cpXXX is already
2035 ;; defined.
2036 (let ((nonascii (get-language-info language-name 'nonascii-translation)))
2037 (if (eq window-system 'pc)
970c9391 2038 (setq nonascii (intern (format "cp%d" dos-codepage))))
e9a8ed3c
KH		 2039 (or (and (charsetp nonascii)
2040 (get-charset-property nonascii :ascii-compatible-p))
2041 (setq nonascii 'iso-8859-1))
33de15f4
SM		 2042 ;; Back in Emacs-20, it was necessary to provide some fallback implicit
2043 ;; conversion, because almost no packages handled coding-system issues.
2044 ;; Nowadays it'd just paper over bugs.
2045 ;; (set-unibyte-charset nonascii)
2046 ))
d042f8b4
KH		 2047
2048(defun set-language-environment-charset (language-name)
2049 "Do various charset setups for language environment LANGUAGE-NAME."
e9a8ed3c
KH		 2050 ;; Put higher priorities to such charsets that are supported by the
2051 ;; coding systems of higher priorities in this environment.
2052 (let ((charsets (get-language-info language-name 'charset)))
2053 (dolist (coding (get-language-info language-name 'coding-priority))
bf974dc9
KH		 2054 (let ((list (coding-system-charset-list coding)))
2055 (if (consp list)
2056 (setq charsets (append charsets list)))))
e9a8ed3c
KH		 2057 (if charsets
2058 (apply 'set-charset-priority charsets))))
d042f8b4
KH		 2059
2060(defun set-language-environment-unibyte (language-name)
2061 "Do various unibyte-mode setups for language environment LANGUAGE-NAME."
d042f8b4
KH		 2062 (set-display-table-and-terminal-coding-system language-name))
2063
cc4d2366 2064(defun princ-list (&rest args)
caff3c0a 2065 "Print all arguments with `princ', then print \"\\n\"."
59f7af81 2066 (declare (obsolete "use mapc and princ instead." "23.3"))
f3bb9e16 2067 (mapc #'princ args)
4ed46869
KH		 2068 (princ "\n"))
2069
b0648a00
RS		 2070(put 'describe-specified-language-support 'apropos-inhibit t)
2071
6b61353c 2072;; Print language-specific information such as input methods,
13e82c04 2073;; charsets, and coding systems. This function is intended to be
48082651 2074;; called from the menu:
281d03ec 2075;; [menu-bar mule describe-language-environment LANGUAGE]
48082651
KH		 2076;; and should not run it by `M-x describe-current-input-method-function'.
2077(defun describe-specified-language-support ()
96db204a 2078 "Describe how Emacs supports the specified language environment."
48082651 2079 (interactive)
281d03ec 2080 (let (language-name)
48082651 2081 (if (not (and (symbolp last-command-event)
cda74479
DL		 2082 (or (not (eq last-command-event 'Default))
2083 (setq last-command-event 'English))
281d03ec 2084 (setq language-name (symbol-name last-command-event))))
3aa5f34b 2085 (error "This command should only be called from the menu bar"))
281d03ec
RS		 2086 (describe-language-environment language-name)))
2087
2088(defun describe-language-environment (language-name)
2089 "Describe how Emacs supports language environment LANGUAGE-NAME."
78754934
KH		 2090 (interactive
2091 (list (read-language-name
2092 'documentation
5b76833f 2093 "Describe language environment (default current choice): ")))
f850d782
RS		 2094 (if (null language-name)
2095 (setq language-name current-language-environment))
281d03ec
RS		 2096 (if (or (null language-name)
2097 (null (get-language-info language-name 'documentation)))
2098 (error "No documentation for the specified language"))
4ef06f75
KH		 2099 (if (symbolp language-name)
2100 (setq language-name (symbol-name language-name)))
ef5a4730
KH		 2101 (dolist (feature (get-language-info language-name 'features))
2102 (require feature))
6b626913 2103 (let ((doc (get-language-info language-name 'documentation)))
c3034e84 2104 (help-setup-xref (list #'describe-language-environment language-name)
32226619 2105 (called-interactively-p 'interactive))
c3034e84 2106 (with-output-to-temp-buffer (help-buffer)
9a529312 2107 (with-current-buffer standard-output
464cc130
KH		 2108 (insert language-name " language environment\n\n")
2109 (if (stringp doc)
2110 (insert doc "\n\n"))
e036b0a6
KH		 2111 (condition-case nil
2112 (let ((str (eval (get-language-info language-name 'sample-text))))
2113 (if (stringp str)
fd0dd4c3
KH		 2114 (insert "Sample text:\n "
2115 (replace-regexp-in-string "\n" "\n " str)
2116 "\n\n")))
e036b0a6 2117 (error nil))
464cc130 2118 (let ((input-method (get-language-info language-name 'input-method))
acd1b9bd
KH		 2119 (l (copy-sequence input-method-alist))
2120 (first t))
2121 (when (and input-method
2122 (setq input-method (assoc input-method l)))
2123 (insert "Input methods (default " (car input-method) ")\n")
2124 (setq l (cons input-method (delete input-method l))
2125 first nil))
2126 (dolist (elt l)
2127 (when (or (eq input-method elt)
2128 (eq t (compare-strings language-name nil nil
2129 (nth 1 elt) nil nil t)))
2130 (when first
2131 (insert "Input methods:\n")
2132 (setq first nil))
2133 (insert " " (car elt))
2134 (search-backward (car elt))
2135 (help-xref-button 0 'help-input-method (car elt))
464cc130 2136 (goto-char (point-max))
2fa7e202 2137 (insert " (\""
acd1b9bd
KH		 2138 (if (stringp (nth 3 elt)) (nth 3 elt) (car (nth 3 elt)))
2139 "\" in mode line)\n")))
2140 (or first
2141 (insert "\n")))
464cc130
KH		 2142 (insert "Character sets:\n")
2143 (let ((l (get-language-info language-name 'charset)))
2144 (if (null l)
2145 (insert " nothing specific to " language-name "\n")
2146 (while l
2147 (insert " " (symbol-name (car l)))
2148 (search-backward (symbol-name (car l)))
467412aa 2149 (help-xref-button 0 'help-character-set (car l))
464cc130
KH		 2150 (goto-char (point-max))
2151 (insert ": " (charset-description (car l)) "\n")
2152 (setq l (cdr l)))))
2153 (insert "\n")
2154 (insert "Coding systems:\n")
2155 (let ((l (get-language-info language-name 'coding-system)))
2156 (if (null l)
2157 (insert " nothing specific to " language-name "\n")
2158 (while l
2159 (insert " " (symbol-name (car l)))
2160 (search-backward (symbol-name (car l)))
467412aa 2161 (help-xref-button 0 'help-coding-system (car l))
464cc130
KH		 2162 (goto-char (point-max))
2163 (insert " (`"
2164 (coding-system-mnemonic (car l))
2165 "' in mode line):\n\t"
2166 (coding-system-doc-string (car l))
2167 "\n")
0855c6cd 2168 (let ((aliases (coding-system-aliases (car l))))
464cc130
KH		 2169 (when aliases
2170 (insert "\t(alias:")
2171 (while aliases
2172 (insert " " (symbol-name (car aliases)))
2173 (setq aliases (cdr aliases)))
2174 (insert ")\n")))
c3034e84 2175 (setq l (cdr l)))))))))
4ed46869 2176\f
40c81f74
PE		 2177;;; Locales.
2178
0d7c5bb9
DL		 2179(defvar locale-translation-file-name nil
2180 "File name for the system's file of locale-name aliases, or nil if none.")
40c81f74 2181
5f395df3
SM		 2182;; The following definitions might as well be marked as constants and
2183;; purecopied, since they're normally used on startup, and probably
2184;; should reflect the facilities of the base Emacs.
2185(defconst locale-language-names
2186 (purecopy
2187 '(
40c81f74
PE		 2188 ;; Locale names of the form LANGUAGE[_TERRITORY][.CODESET][@MODIFIER]
2189 ;; as specified in the Single Unix Spec, Version 2.
2190 ;; LANGUAGE is a language code taken from ISO 639:1988 (E/F)
2191 ;; with additions from ISO 639/RA Newsletter No.1/1989;
5f395df3
SM		 2192 ;; see Internet RFC 2165 (1997-06) and
2193 ;; http://www.evertype.com/standards/iso639/iso639-en.html
2194 ;; TERRITORY is a country code taken from ISO 3166
2195 ;; http://www.din.de/gremien/nas/nabd/iso3166ma/codlstp1/en_listp1.html.
40c81f74 2196 ;; CODESET and MODIFIER are implementation-dependent.
5f395df3 2197
6b61353c
KH		 2198 ;; jasonr comments: MS Windows uses three letter codes for
2199 ;; languages instead of the two letter ISO codes that POSIX
2200 ;; uses. In most cases the first two letters are the same, so
2201 ;; most of the regexps in locale-language-names work. Japanese
2202 ;; and Chinese are exceptions, which are listed in the
2203 ;; non-standard section at the bottom of locale-language-names.
2204
8dedddd5
KH		 2205 ("aa_DJ" . "Latin-1") ; Afar
2206 ("aa" . "UTF-8")
2207 ;; ab Abkhazian
6ececc4d 2208 ("af" . "Latin-1") ; Afrikaans
8dedddd5
KH		 2209 ("am" "Ethiopic" utf-8) ; Amharic
2210 ("an" . "Latin-9") ; Aragonese
b60f961f 2211 ("ar" . "Arabic")
40c81f74
PE		 2212 ; as Assamese
2213 ; ay Aymara
8dedddd5 2214 ("az" . "UTF-8") ; Azerbaijani
40c81f74 2215 ; ba Bashkir
8dedddd5
KH		 2216 ("be" "Belarusian" cp1251) ; Belarusian [Byelorussian until early 1990s]
2217 ("bg" "Bulgarian" cp1251) ; Bulgarian
40c81f74
PE		 2218 ; bh Bihari
2219 ; bi Bislama
8dedddd5 2220 ("bn" . "UTF-8") ; Bengali, Bangla
40c81f74
PE		 2221 ("bo" . "Tibetan")
2222 ("br" . "Latin-1") ; Breton
d37ef0f6 2223 ("bs" . "Latin-2") ; Bosnian
8dedddd5 2224 ("byn" . "UTF-8") ; Bilin; Blin
40c81f74
PE		 2225 ("ca" . "Latin-1") ; Catalan
2226 ; co Corsican
8dedddd5
KH		 2227 ("cs" "Czech" iso-8859-2)
2228 ("cy" "Welsh" iso-8859-14)
40c81f74 2229 ("da" . "Latin-1") ; Danish
8dedddd5 2230 ("de" "German" iso-8859-1)
569a6374 2231 ; dv Divehi
40c81f74 2232 ; dz Bhutani
8dedddd5 2233 ("el" "Greek" iso-8859-7)
6ececc4d 2234 ;; Users who specify "en" explicitly typically want Latin-1, not ASCII.
e11cf111
DL		 2235 ;; That's actually what the GNU locales define, modulo things like
2236 ;; en_IN -- fx.
8dedddd5 2237 ("en_IN" "English" utf-8) ; glibc uses utf-8 for English in India
eec5c8f9 2238 ("en" "English" iso-8859-1) ; English
58802cd7 2239 ("eo" . "Esperanto") ; Esperanto
8dedddd5
KH		 2240 ("es" "Spanish" iso-8859-1)
2241 ("et" . "Latin-1") ; Estonian
40c81f74 2242 ("eu" . "Latin-1") ; Basque
8dedddd5 2243 ("fa" . "UTF-8") ; Persian
40c81f74 2244 ("fi" . "Latin-1") ; Finnish
8dedddd5 2245 ("fj" . "Latin-1") ; Fiji
40c81f74 2246 ("fo" . "Latin-1") ; Faroese
8dedddd5 2247 ("fr" "French" iso-8859-1) ; French
40c81f74 2248 ("fy" . "Latin-1") ; Frisian
6ececc4d 2249 ("ga" . "Latin-1") ; Irish Gaelic (new orthography)
8dedddd5
KH		 2250 ("gd" . "Latin-9") ; Scots Gaelic
2251 ("gez" "Ethiopic" utf-8) ; Geez
2252 ("gl" . "Latin-1") ; Gallegan; Galician
40c81f74 2253 ; gn Guarani
8dedddd5
KH		 2254 ("gu" . "UTF-8") ; Gujarati
2255 ("gv" . "Latin-1") ; Manx Gaelic
40c81f74 2256 ; ha Hausa
8dedddd5
KH		 2257 ("he" "Hebrew" iso-8859-8)
2258 ("hi" "Devanagari" utf-8) ; Hindi
2259 ("hr" "Croatian" iso-8859-2) ; Croatian
40c81f74
PE		 2260 ("hu" . "Latin-2") ; Hungarian
2261 ; hy Armenian
2262 ; ia Interlingua
2263 ("id" . "Latin-1") ; Indonesian
2264 ; ie Interlingue
2265 ; ik Inupiak
2266 ("is" . "Latin-1") ; Icelandic
8dedddd5 2267 ("it" "Italian" iso-8859-1) ; Italian
40c81f74 2268 ; iu Inuktitut
8dedddd5
KH		 2269 ("iw" "Hebrew" iso-8859-8)
2270 ("ja" "Japanese" euc-jp)
40c81f74 2271 ; jw Javanese
8dedddd5 2272 ("ka" "Georgian" georgian-ps) ; Georgian
40c81f74 2273 ; kk Kazakh
6ececc4d 2274 ("kl" . "Latin-1") ; Greenlandic
40c81f74 2275 ; km Cambodian
8dedddd5
KH		 2276 ("kn" "Kannada" utf-8)
2277 ("ko" "Korean" euc-kr)
40c81f74
PE		 2278 ; ks Kashmiri
2279 ; ku Kurdish
5f395df3 2280 ("kw" . "Latin-1") ; Cornish
40c81f74
PE		 2281 ; ky Kirghiz
2282 ("la" . "Latin-1") ; Latin
5f395df3 2283 ("lb" . "Latin-1") ; Luxemburgish
8dedddd5 2284 ("lg" . "Laint-6") ; Ganda
40c81f74 2285 ; ln Lingala
8dedddd5
KH		 2286 ("lo" "Lao" utf-8) ; Laothian
2287 ("lt" "Lithuanian" iso-8859-13)
9c20a8d5 2288 ("lv" . "Latvian") ; Latvian, Lettish
40c81f74 2289 ; mg Malagasy
5f395df3 2290 ("mi" . "Latin-7") ; Maori
8dedddd5
KH		 2291 ("mk" "Cyrillic-ISO" iso-8859-5) ; Macedonian
2292 ("ml" "Malayalam" utf-8)
2293 ("mn" . "UTF-8") ; Mongolian
40c81f74 2294 ; mo Moldavian
8dedddd5 2295 ("mr" "Devanagari" utf-8) ; Marathi
5f395df3 2296 ("ms" . "Latin-1") ; Malay
40c81f74
PE		 2297 ("mt" . "Latin-3") ; Maltese
2298 ; my Burmese
2299 ; na Nauru
8dedddd5
KH		 2300 ("nb" . "Latin-1") ; Norwegian
2301 ("ne" "Devanagari" utf-8) ; Nepali
2302 ("nl" "Dutch" iso-8859-1)
40c81f74 2303 ("no" . "Latin-1") ; Norwegian
5f395df3 2304 ("oc" . "Latin-1") ; Occitan
8dedddd5
KH		 2305 ("om_ET" . "UTF-8") ; (Afan) Oromo
2306 ("om" . "Latin-1") ; (Afan) Oromo
40c81f74 2307 ; or Oriya
8dedddd5 2308 ("pa" . "UTF-8") ; Punjabi
40c81f74
PE		 2309 ("pl" . "Latin-2") ; Polish
2310 ; ps Pashto, Pushto
2311 ("pt" . "Latin-1") ; Portuguese
2312 ; qu Quechua
6ececc4d 2313 ("rm" . "Latin-1") ; Rhaeto-Romanic
40c81f74 2314 ; rn Kirundi
8dedddd5
KH		 2315 ("ro" "Romanian" iso-8859-2)
2316 ("ru_RU" "Russian" iso-8859-5)
2317 ("ru_UA" "Russian" koi8-u)
40c81f74
PE		 2318 ; rw Kinyarwanda
2319 ("sa" . "Devanagari") ; Sanskrit
2320 ; sd Sindhi
8dedddd5 2321 ("se" . "UTF-8") ; Northern Sami
40c81f74
PE		 2322 ; sg Sangho
2323 ("sh" . "Latin-2") ; Serbo-Croatian
2324 ; si Sinhalese
8dedddd5
KH		 2325 ("sid" . "UTF-8") ; Sidamo
2326 ("sk" "Slovak" iso-8859-2)
2327 ("sl" "Slovenian" iso-8859-2)
40c81f74
PE		 2328 ; sm Samoan
2329 ; sn Shona
8dedddd5
KH		 2330 ("so_ET" "UTF-8") ; Somali
2331 ("so" "Latin-1") ; Somali
6ececc4d 2332 ("sq" . "Latin-1") ; Albanian
40c81f74
PE		 2333 ("sr" . "Latin-2") ; Serbian (Latin alphabet)
2334 ; ss Siswati
8dedddd5 2335 ("st" . "Latin-1") ; Sesotho
40c81f74 2336 ; su Sundanese
8dedddd5 2337 ("sv" "Swedish" iso-8859-1) ; Swedish
40c81f74 2338 ("sw" . "Latin-1") ; Swahili
8dedddd5
KH		 2339 ("ta" "Tamil" utf-8)
2340 ("te" . "UTF-8") ; Telugu
2341 ("tg" "Tajik" koi8-t)
2342 ("th" "Thai" tis-620)
2343 ("ti" "Ethiopic" utf-8) ; Tigrinya
2344 ("tig_ER" . "UTF-8") ; Tigre
40c81f74 2345 ; tk Turkmen
6ececc4d 2346 ("tl" . "Latin-1") ; Tagalog
40c81f74
PE		 2347 ; tn Setswana
2348 ; to Tonga
8dedddd5 2349 ("tr" "Turkish" iso-8859-9)
40c81f74 2350 ; ts Tsonga
8dedddd5 2351 ("tt" . "UTF-8") ; Tatar
40c81f74
PE		 2352 ; tw Twi
2353 ; ug Uighur
8dedddd5
KH		 2354 ("uk" "Ukrainian" koi8-u)
2355 ("ur" . "UTF-8") ; Urdu
2356 ("uz_UZ@cyrillic" . "UTF-8"); Uzbek
5f395df3 2357 ("uz" . "Latin-1") ; Uzbek
8dedddd5 2358 ("vi" "Vietnamese" utf-8)
40c81f74 2359 ; vo Volapuk
d37ef0f6 2360 ("wa" . "Latin-1") ; Walloon
40c81f74 2361 ; wo Wolof
8dedddd5 2362 ("xh" . "Latin-1") ; Xhosa
9c20a8d5 2363 ("yi" . "Windows-1255") ; Yiddish
40c81f74
PE		 2364 ; yo Yoruba
2365 ; za Zhuang
8dedddd5 2366 ("zh_HK" . "Chinese-Big5")
05b81b42 2367 ; zh_HK/BIG5-HKSCS \
8dedddd5 2368 ("zh_TW" . "Chinese-Big5")
05b81b42
KH		 2369 ("zh_CN.GB2312" "Chinese-GB")
2370 ("zh_CN.GBK" "Chinese-GBK")
2371 ("zh_CN.GB18030" "Chinese-GB18030")
f21605fa 2372 ("zh_CN.UTF-8" . "Chinese-GBK")
8dedddd5 2373 ("zh_CN" . "Chinese-GB")
40c81f74 2374 ("zh" . "Chinese-GB")
8dedddd5 2375 ("zu" . "Latin-1") ; Zulu
40c81f74
PE		 2376
2377 ;; ISO standard locales
2378 ("c$" . "ASCII")
2379 ("posix$" . "ASCII")
2380
40c81f74
PE		 2381 ;; The "IPA" Emacs language environment does not correspond
2382 ;; to any ISO 639 code, so let it stand for itself.
2383 ("ipa$" . "IPA")
2384
2385 ;; Nonstandard or obsolete language codes
2386 ("cz" . "Czech") ; e.g. Solaris 2.6
2387 ("ee" . "Latin-4") ; Estonian, e.g. X11R6.4
2388 ("iw" . "Hebrew") ; e.g. X11R6.4
f1282c7f 2389 ("sp" . "Cyrillic-ISO") ; Serbian (Cyrillic alphabet), e.g. X11R6.4
40c81f74 2390 ("su" . "Latin-1") ; Finnish, e.g. Solaris 2.6
2e86ceaa 2391 ("jp" . "Japanese") ; e.g. MS Windows
f21605fa 2392 ("chs" . "Chinese-GBK") ; MS Windows Chinese Simplified
86d9e628 2393 ("cht" . "Chinese-BIG5") ; MS Windows Chinese Traditional
569a6374
EZ		 2394 ("gbz" . "UTF-8") ; MS Windows Dari Persian
2395 ("div" . "UTF-8") ; MS Windows Divehi (Maldives)
2396 ("wee" . "Latin-2") ; MS Windows Lower Sorbian
2397 ("wen" . "Latin-2") ; MS Windows Upper Sorbian
367ca50f 2398 ))
8dedddd5 2399 "Alist of locale regexps vs the corresponding languages and coding systems.
caff3c0a 2400Each element has this form:
8dedddd5
KH		 2401 \(LOCALE-REGEXP LANG-ENV CODING-SYSTEM)
2402The first element whose LOCALE-REGEXP matches the start of a
caff3c0a 2403downcased locale specifies the LANG-ENV \(language environment)
8dedddd5
KH		 2404and CODING-SYSTEM corresponding to that locale. If there is no
2405appropriate language environment, the element may have this form:
2406 \(LOCALE-REGEXP . LANG-ENV)
2407In this case, LANG-ENV is one of generic language environments for an
2408specific encoding such as \"Latin-1\" and \"UTF-8\".")
40c81f74 2409
5f395df3
SM		 2410(defconst locale-charset-language-names
2411 (purecopy
2412 '((".*8859[-_]?1\\>" . "Latin-1")
2413 (".*8859[-_]?2\\>" . "Latin-2")
2414 (".*8859[-_]?3\\>" . "Latin-3")
2415 (".*8859[-_]?4\\>" . "Latin-4")
2416 (".*8859[-_]?9\\>" . "Latin-5")
2417 (".*8859[-_]?14\\>" . "Latin-8")
2418 (".*8859[-_]?15\\>" . "Latin-9")
6b61353c
KH		 2419 (".*utf\\(?:-?8\\)?\\>" . "UTF-8")
2420 ;; utf-8@euro exists, so put this last. (@euro really specifies
2421 ;; the currency, rather than the charset.)
e522f07b 2422 (".*@euro\\>" . "Latin-9")))
6ececc4d
PE		 2423 "List of pairs of locale regexps and charset language names.
2424The first element whose locale regexp matches the start of a downcased locale
6b61353c 2425specifies the language name whose charset corresponds to that locale.
8dedddd5 2426This language name is used if the locale is not listed in
caff3c0a 2427`locale-language-names'.")
6ececc4d 2428
5f395df3
SM		 2429(defconst locale-preferred-coding-systems
2430 (purecopy
8dedddd5
KH		 2431 '((".*8859[-_]?1\\>" . iso-8859-1)
2432 (".*8859[-_]?2\\>" . iso-8859-2)
2433 (".*8859[-_]?3\\>" . iso-8859-3)
2434 (".*8859[-_]?4\\>" . iso-8859-4)
2435 (".*8859[-_]?9\\>" . iso-8859-9)
2436 (".*8859[-_]?14\\>" . iso-8859-14)
2437 (".*8859[-_]?15\\>" . iso-8859-15)
2438 (".*utf\\(?:-?8\\)?" . utf-8)
2439 ;; utf-8@euro exists, so put this after utf-8. (@euro really
2440 ;; specifies the currency, rather than the charset.)
2441 (".*@euro" . iso-8859-15)
2442 ("koi8-?r" . koi8-r)
2443 ("koi8-?u" . koi8-u)
2444 ("tcvn" . tcvn)
38c05d07 2445 ("big5[-_]?hkscs" . big5-hkscs)
8dedddd5
KH		 2446 ("big5" . big5)
2447 ("euc-?tw" . euc-tw)
d6be7497 2448 ("euc-?cn" . euc-cn)
38c05d07
KH		 2449 ("gb2312" . gb2312)
2450 ("gbk" . gbk)
2451 ("gb18030" . gb18030)
8dedddd5 2452 ("ja.*[._]euc" . japanese-iso-8bit)
5f395df3
SM		 2453 ("ja.*[._]jis7" . iso-2022-jp)
2454 ("ja.*[._]pck" . japanese-shift-jis)
2455 ("ja.*[._]sjis" . japanese-shift-jis)
69210880 2456 ("jpn" . japanese-shift-jis) ; MS-Windows uses this.
8dedddd5 2457 ))
6ececc4d
PE		 2458 "List of pairs of locale regexps and preferred coding systems.
2459The first element whose locale regexp matches the start of a downcased locale
8dedddd5
KH		 2460specifies the coding system to prefer when using that locale.
2461This coding system is used if the locale specifies a specific charset.")
40c81f74
PE		 2462
2463(defun locale-name-match (key alist)
2464 "Search for KEY in ALIST, which should be a list of regexp-value pairs.
2465Return the value corresponding to the first regexp that matches the
2466start of KEY, or nil if there is no match."
2467 (let (element)
2468 (while (and alist (not element))
d0d8885d 2469 (if (string-match-p (concat "\\`\\(?:" (car (car alist)) "\\)") key)
40c81f74
PE		 2470 (setq element (car alist)))
2471 (setq alist (cdr alist)))
2472 (cdr element)))
2473
58b78d5b 2474(defun locale-charset-match-p (charset1 charset2)
91ed0599 2475 "Whether charset names (strings) CHARSET1 and CHARSET2 are equivalent.
58b78d5b
DL		 2476Matching is done ignoring case and any hyphens and underscores in the
2477names. E.g. `ISO_8859-1' and `iso88591' both match `iso-8859-1'."
2478 (setq charset1 (replace-regexp-in-string "[-_]" "" charset1))
2479 (setq charset2 (replace-regexp-in-string "[-_]" "" charset2))
2480 (eq t (compare-strings charset1 nil nil charset2 nil nil t)))
2481
6b61353c
KH		 2482(defvar locale-charset-alist nil
2483 "Coding system alist keyed on locale-style charset name.
2484Used by `locale-charset-to-coding-system'.")
2485
2486(defun locale-charset-to-coding-system (charset)
2487 "Find coding system corresponding to CHARSET.
2488CHARSET is any sort of non-Emacs charset name, such as might be used
2489in a locale codeset, or elsewhere. It is matched to a coding system
2490first by case-insensitive lookup in `locale-charset-alist'. Then
2491matches are looked for in the coding system list, treating case and
2492the characters `-' and `_' as insignificant. The coding system base
2493is returned. Thus, for instance, if charset \"ISO8859-2\",
2494`iso-latin-2' is returned."
f15466c5 2495 (or (car (assoc-string charset locale-charset-alist t))
6b61353c
KH		 2496 (let ((cs coding-system-alist)
2497 c)
2498 (while (and (not c) cs)
2499 (if (locale-charset-match-p charset (caar cs))
2500 (setq c (intern (caar cs)))
2501 (pop cs)))
2502 (if c (coding-system-base c)))))
2503
2504;; Fixme: This ought to deal with the territory part of the locale
2505;; too, for setting things such as calendar holidays, ps-print paper
2506;; size, spelling dictionary.
2507
44ee1bdf
GM		 2508(defun locale-translate (locale)
2509 "Expand LOCALE according to `locale-translation-file-name', if possible.
2510For example, translate \"swedish\" into \"sv_SE.ISO8859-1\"."
2511 (if locale-translation-file-name
2512 (with-temp-buffer
2513 (set-buffer-multibyte nil)
2514 (insert-file-contents locale-translation-file-name)
2515 (if (re-search-forward
2516 (concat "^" (regexp-quote locale) ":?[ \t]+") nil t)
2517 (buffer-substring (point) (line-end-position))
2518 locale))
2519 locale))
2520
36ab8612 2521(defun set-locale-environment (&optional locale-name frame)
9c758578 2522 "Set up multilingual environment for using LOCALE-NAME.
758f07de
RS		 2523This sets the language environment, the coding system priority,
2524the default input method and sometimes other things.
2525
58b78d5b 2526LOCALE-NAME should be a string which is the name of a locale supported
6b61353c 2527by the system. Often it is of the form xx_XX.CODE, where xx is a
58b78d5b
DL		 2528language, XX is a country, and CODE specifies a character set and
2529coding system. For example, the locale name \"ja_JP.EUC\" might name
2530a locale for Japanese in Japan using the `japanese-iso-8bit'
2531coding-system. The name may also have a modifier suffix, e.g. `@euro'
2532or `@cyrillic'.
40c81f74 2533
758f07de 2534If LOCALE-NAME is nil, its value is taken from the environment
d37ef0f6 2535variables LC_ALL, LC_CTYPE and LANG (the first one that is set).
40c81f74
PE		 2536
2537The locale names supported by your system can typically be found in a
0812c1e8 2538directory named `/usr/share/locale' or `/usr/lib/locale'. LOCALE-NAME
758f07de 2539will be translated according to the table specified by
0812c1e8
DL		 2540`locale-translation-file-name'.
2541
36ab8612
MB		 2542If FRAME is non-nil, only set the keyboard coding system and the
2543terminal coding system for the terminal of that frame, and don't
2544touch session-global parameters like the language environment.
97c57fb2 2545
0812c1e8
DL		 2546See also `locale-charset-language-names', `locale-language-names',
2547`locale-preferred-coding-systems' and `locale-coding-system'."
758f07de 2548 (interactive "sSet environment for locale: ")
a1506d29 2549
0d7c5bb9
DL		 2550 ;; Do this at runtime for the sake of binaries possibly transported
2551 ;; to a system without X.
2552 (setq locale-translation-file-name
2553 (let ((files
6bba8c70
KH		 2554 '("/usr/share/X11/locale/locale.alias" ; e.g. X11R7
2555 "/usr/lib/X11/locale/locale.alias" ; e.g. X11R6.4
6b61353c 2556 "/usr/X11R6/lib/X11/locale/locale.alias" ; XFree86, e.g. RedHat 4.2
0d7c5bb9
DL		 2557 "/usr/openwin/lib/locale/locale.alias" ; e.g. Solaris 2.6
2558 ;;
2559 ;; The following name appears after the X-related names above,
2560 ;; since the X-related names are what X actually uses.
2561 "/usr/share/locale/locale.alias" ; GNU/Linux sans X
2562 )))
2563 (while (and files (not (file-exists-p (car files))))
2564 (setq files (cdr files)))
2565 (car files)))
2566
7008ccac
GM		 2567 (let ((locale locale-name))
2568
2569 (unless locale
2570 ;; Use the first of these three environment variables
2571 ;; that has a nonempty value.
2572 (let ((vars '("LC_ALL" "LC_CTYPE" "LANG")))
db6bd804
DL		 2573 (while (and vars
2574 (= 0 (length locale))) ; nil or empty string
36ab8612 2575 (setq locale (getenv (pop vars) frame)))))
7008ccac
GM		 2576
2577 (when locale
44ee1bdf 2578 (setq locale (locale-translate locale))
7008ccac
GM		 2579
2580 ;; Leave the system locales alone if the caller did not specify
2581 ;; an explicit locale name, as their defaults are set from
2582 ;; LC_MESSAGES and LC_TIME, not LC_CTYPE, and the user might not
2583 ;; want to set them to the same value as LC_CTYPE.
2584 (when locale-name
2585 (setq system-messages-locale locale)
e9aaa1db
KH		 2586 (setq system-time-locale locale))
2587
2588 (if (string-match "^[a-z][a-z]" locale)
2589 (setq current-iso639-language (intern (match-string 0 locale)))))
7008ccac 2590
44ee1bdf
GM		 2591 (setq woman-locale
2592 (or system-messages-locale
739b7764 2593 (let ((msglocale (getenv "LC_MESSAGES" frame)))
44ee1bdf
GM		 2594 (if (zerop (length msglocale))
2595 locale
2596 (locale-translate msglocale)))))
2597
2598 (when locale
7008ccac
GM		 2599 (setq locale (downcase locale))
2600
2601 (let ((language-name
2602 (locale-name-match locale locale-language-names))
2603 (charset-language-name
2604 (locale-name-match locale locale-charset-language-names))
cdb1af30 2605 (default-eol-type (coding-system-eol-type
b56a5ae0 2606 (default-value 'buffer-file-coding-system)))
7008ccac 2607 (coding-system
6b61353c
KH		 2608 (or (locale-name-match locale locale-preferred-coding-systems)
2609 (when locale
2610 (if (string-match "\\.\\([^@]+\\)" locale)
2611 (locale-charset-to-coding-system
9e2a2647 2612 (match-string 1 locale)))))))
7008ccac 2613
8dedddd5
KH		 2614 (if (consp language-name)
2615 ;; locale-language-names specify both lang-env and coding.
2616 ;; But, what specified in locale-preferred-coding-systems
2617 ;; has higher priority.
2618 (setq coding-system (or coding-system
2619 (nth 1 language-name))
2620 language-name (car language-name))
2621 ;; Otherwise, if locale is not listed in locale-language-names,
2622 ;; use what listed in locale-charset-language-names.
2623 (if (not language-name)
2624 (setq language-name charset-language-name)))
7008ccac 2625
cdb1af30
EZ		 2626 ;; If a specific EOL conversion was specified in the default
2627 ;; buffer-file-coding-system, preserve it in the coding system
2628 ;; we will be using from now on.
8bca692e
EZ		 2629 (if (and (memq default-eol-type '(0 1 2 unix dos mac))
2630 coding-system
2631 (coding-system-p coding-system))
cdb1af30
EZ		 2632 (setq coding-system (coding-system-change-eol-conversion
2633 coding-system default-eol-type)))
2634
7008ccac
GM		 2635 (when language-name
2636
2637 ;; Set up for this character set. This is now the right way
2638 ;; to do it for both unibyte and multibyte modes.
36ab8612 2639 (unless frame
97c57fb2 2640 (set-language-environment language-name))
7008ccac 2641
597e2240 2642 ;; If the default enable-multibyte-characters is nil,
7008ccac
GM		 2643 ;; we are using single-byte characters,
2644 ;; so the display table and terminal coding system are irrelevant.
597e2240 2645 (when (default-value 'enable-multibyte-characters)
ff76e074 2646 (set-display-table-and-terminal-coding-system
36ab8612 2647 language-name coding-system frame))
7008ccac 2648
166ce29f
DL		 2649 ;; Set the `keyboard-coding-system' if appropriate (tty
2650 ;; only). At least X and MS Windows can generate
2651 ;; multilingual input.
68bba4e4 2652 ;; XXX This was disabled unless `window-system', but that
fffa137c 2653 ;; leads to buggy behavior when a tty frame is opened
68bba4e4
KL		 2654 ;; later. Setting the keyboard coding system has no adverse
2655 ;; effect on X, so let's do it anyway. -- Lorentey
2656 (let ((kcs (or coding-system
2657 (car (get-language-info language-name
2658 'coding-system)))))
36ab8612 2659 (if kcs (set-keyboard-coding-system kcs frame)))
dc2be2fa 2660
36ab8612 2661 (unless frame
97c57fb2
KL		 2662 (setq locale-coding-system
2663 (car (get-language-info language-name 'coding-priority)))))
7008ccac 2664
36ab8612 2665 (when (and (not frame)
97c57fb2 2666 coding-system
8dedddd5
KH		 2667 (not (coding-system-equal coding-system
2668 locale-coding-system)))
7008ccac 2669 (prefer-coding-system coding-system)
218e7ce3
KH		 2670 ;; Fixme: perhaps prefer-coding-system should set this too.
2671 ;; But it's not the time to do such a fundamental change.
2672 (setq default-sendmail-coding-system coding-system)
80e3310b 2673 (setq locale-coding-system coding-system))))
e76ef161 2674
52c7f9ee 2675 ;; On Windows, override locale-coding-system,
26245233 2676 ;; default-file-name-coding-system, keyboard-coding-system,
20ba0cb4 2677 ;; terminal-coding-system with system codepage.
0fda9b75
DC		 2678 (when (and (eq system-type 'windows-nt)
2679 (boundp 'w32-ansi-code-page))
20ba0cb4
EZ		 2680 (let ((code-page-coding (intern (format "cp%d" w32-ansi-code-page))))
2681 (when (coding-system-p code-page-coding)
2682 (unless frame (setq locale-coding-system code-page-coding))
2683 (set-keyboard-coding-system code-page-coding frame)
2684 (set-terminal-coding-system code-page-coding frame)
20ba0cb4 2685 (setq default-file-name-coding-system code-page-coding))))
893b49bb 2686
a41118cc 2687 (when (eq system-type 'darwin)
ff76e074
SM		 2688 ;; On Darwin, file names are always encoded in utf-8, no matter
2689 ;; the locale.
2690 (setq default-file-name-coding-system 'utf-8)
2691 ;; Mac OS X's Terminal.app by default uses utf-8 regardless of
2692 ;; the locale.
2693 (when (and (null window-system)
36ab8612 2694 (equal (getenv "TERM_PROGRAM" frame) "Apple_Terminal"))
ff76e074
SM		 2695 (set-terminal-coding-system 'utf-8)
2696 (set-keyboard-coding-system 'utf-8)))
a41118cc 2697
e76ef161 2698 ;; Default to A4 paper if we're not in a C, POSIX or US locale.
3479c806 2699 ;; (See comments in Flocale_info.)
36ab8612 2700 (unless frame
97c57fb2
KL		 2701 (let ((locale locale)
2702 (paper (locale-info 'paper)))
2703 (if paper
2704 ;; This will always be null at the time of writing.
2705 (cond
2706 ((equal paper '(216 279))
2707 (setq ps-paper-type 'letter))
2708 ((equal paper '(210 297))
2709 (setq ps-paper-type 'a4)))
2710 (let ((vars '("LC_ALL" "LC_PAPER" "LANG")))
2711 (while (and vars (= 0 (length locale)))
36ab8612 2712 (setq locale (getenv (pop vars) frame))))
97c57fb2
KL		 2713 (when locale
2714 ;; As of glibc 2.2.5, these are the only US Letter locales,
2715 ;; and the rest are A4.
2716 (setq ps-paper-type
2717 (or (locale-name-match locale '(("c$" . letter)
2718 ("posix$" . letter)
2719 (".._us" . letter)
2720 (".._pr" . letter)
2721 (".._ca" . letter)
2722 ("enu$" . letter) ; Windows
2723 ("esu$" . letter)
2724 ("enc$" . letter)
2725 ("frc$" . letter)))
2726 'a4)))))))
e76ef161 2727 nil)
40c81f74 2728\f
16400c32
KH		 2729;;; Character property
2730
16400c32
KH		 2731(put 'char-code-property-table 'char-table-extra-slots 5)
2732
2733(defun define-char-code-property (name table &optional docstring)
2734 "Define NAME as a character code property given by TABLE.
2735TABLE is a char-table of purpose `char-code-property-table' with
2736these extra slots:
2737 1st: NAME.
2738 2nd: Function to call to get a property value of a character.
caff3c0a 2739 It is called with three arguments CHAR, VAL, and TABLE, where
16400c32
KH		 2740 CHAR is a character, VAL is the value of (aref TABLE CHAR).
2741 3rd: Function to call to put a property value of a character.
2742 It is called with the same arguments as above.
2743 4th: Function to call to get a description string of a property value.
2744 It is called with one argument VALUE, a property value.
2745 5th: Data used by the above functions.
2746
2747TABLE may be a name of file to load to build a char-table. The
2748file should contain a call of `define-char-code-property' with a
2749char-table of the above format as the argument TABLE.
2750
2751TABLE may also be nil, in which case no property value is pre-assigned.
2752
49275d55 2753Optional 3rd argument DOCSTRING is a documentation string of the property.
16400c32
KH		 2754
2755See also the documentation of `get-char-code-property' and
2756`put-char-code-property'."
2757 (or (symbolp name)
2758 (error "Not a symbol: %s" name))
2759 (if (char-table-p table)
2760 (or (and (eq (char-table-subtype table) 'char-code-property-table)
2761 (eq (char-table-extra-slot table 0) name))
2762 (error "Invalid char-table: %s" table))
2763 (or (stringp table)
2764 (error "Not a char-table nor a file name: %s" table)))
ebfa10d3 2765 (if (stringp table) (setq table (purecopy table)))
16400c32
KH		 2766 (let ((slot (assq name char-code-property-alist)))
2767 (if slot
2768 (setcdr slot table)
2769 (setq char-code-property-alist
2770 (cons (cons name table) char-code-property-alist))))
905a9ed3 2771 (put name 'char-code-property-documentation (purecopy docstring)))
4ed46869
KH		 2772
2773(defvar char-code-property-table
2774 (make-char-table 'char-code-property-table)
2775 "Char-table containing a property list of each character code.
16400c32 2776This table is used for properties not listed in `char-code-property-alist'.
4ed46869 2777See also the documentation of `get-char-code-property' and
96db204a 2778`put-char-code-property'.")
4ed46869
KH		 2779
2780(defun get-char-code-property (char propname)
16400c32 2781 "Return the value of CHAR's PROPNAME property."
c805dec0
KH		 2782 (let ((table (unicode-property-table-internal propname)))
2783 (if table
2784 (let ((func (char-table-extra-slot table 1)))
16400c32 2785 (if (functionp func)
c805dec0
KH		 2786 (funcall func char (aref table char) table)
2787 (get-unicode-property-internal table char)))
16400c32 2788 (plist-get (aref char-code-property-table char) propname))))
4ed46869
KH		 2789
2790(defun put-char-code-property (char propname value)
16400c32 2791 "Store CHAR's PROPNAME property with VALUE.
4ed46869 2792It can be retrieved with `(get-char-code-property CHAR PROPNAME)'."
c805dec0
KH		 2793 (let ((table (unicode-property-table-internal propname)))
2794 (if table
2795 (let ((func (char-table-extra-slot table 2)))
16400c32
KH		 2796 (if (functionp func)
2797 (funcall func char value table)
c805dec0 2798 (put-unicode-property-internal table char value)))
16400c32
KH		 2799 (let* ((plist (aref char-code-property-table char))
2800 (x (plist-put plist propname value)))
2801 (or (eq x plist)
2802 (aset char-code-property-table char x))))
2803 value))
2804
2805(defun char-code-property-description (prop value)
2806 "Return a description string of character property PROP's value VALUE.
2807If there's no description string for VALUE, return nil."
c805dec0
KH		 2808 (let ((table (unicode-property-table-internal prop)))
2809 (if table
2810 (let ((func (char-table-extra-slot table 3)))
16400c32
KH		 2811 (if (functionp func)
2812 (funcall func value))))))
4ed46869 2813
a127b764
KH		 2814\f
2815;; Pretty description of encoded string
2816
2817;; Alist of ISO 2022 control code vs the corresponding mnemonic string.
2f7f4bee 2818(defconst iso-2022-control-alist
a127b764
KH		 2819 '((?\x1b . "ESC")
2820 (?\x0e . "SO")
2821 (?\x0f . "SI")
2822 (?\x8e . "SS2")
2823 (?\x8f . "SS3")
2824 (?\x9b . "CSI")))
2825
2826(defun encoded-string-description (str coding-system)
2827 "Return a pretty description of STR that is encoded by CODING-SYSTEM."
2828 (setq str (string-as-unibyte str))
993b2a7d 2829 (mapconcat
a6dfc99b 2830 (if (and coding-system (eq (coding-system-type coding-system) 'iso-2022))
993b2a7d
KH		 2831 ;; Try to get a pretty description for ISO 2022 escape sequences.
2832 (function (lambda (x) (or (cdr (assq x iso-2022-control-alist))
695effcc
JL		 2833 (format "#x%02X" x))))
2834 (function (lambda (x) (format "#x%02X" x))))
993b2a7d 2835 str " "))
a127b764 2836
430c6ced 2837(defun encode-coding-char (char coding-system &optional charset)
a127b764 2838 "Encode CHAR by CODING-SYSTEM and return the resulting string.
430c6ced
KH		 2839If CODING-SYSTEM can't safely encode CHAR, return nil.
2840The 3rd optional argument CHARSET, if non-nil, is a charset preferred
2841on encoding."
b529ed1e
KH		 2842 (let* ((str1 (string-as-multibyte (string char)))
2843 (str2 (string-as-multibyte (string char char)))
2844 (found (find-coding-systems-string str1))
a127b764 2845 enc1 enc2 i1 i2)
b529ed1e
KH		 2846 (if (and (consp found)
2847 (eq (car found) 'undecided))
2848 str1
2849 (when (memq (coding-system-base coding-system) found)
2850 ;; We must find the encoded string of CHAR. But, just encoding
2851 ;; CHAR will put extra control sequences (usually to designate
2852 ;; ASCII charset) at the tail if type of CODING is ISO 2022.
2853 ;; To exclude such tailing bytes, we at first encode one-char
2854 ;; string and two-char string, then check how many bytes at the
2855 ;; tail of both encoded strings are the same.
2856
430c6ced
KH		 2857 (when charset
2858 (put-text-property 0 1 'charset charset str1)
2859 (put-text-property 0 2 'charset charset str2))
b529ed1e
KH		 2860 (setq enc1 (encode-coding-string str1 coding-system)
2861 i1 (length enc1)
2862 enc2 (encode-coding-string str2 coding-system)
2863 i2 (length enc2))
2864 (while (and (> i1 0) (= (aref enc1 (1- i1)) (aref enc2 (1- i2))))
2865 (setq i1 (1- i1) i2 (1- i2)))
2866
2867 ;; Now (substring enc1 i1) and (substring enc2 i2) are the same,
2868 ;; and they are the extra control sequences at the tail to
2869 ;; exclude.
2870 (substring enc2 0 i2)))))
a127b764 2871
d2f613a0
DL		 2872;; Backwards compatibility. These might be better with :init-value t,
2873;; but that breaks loadup.
2874(define-minor-mode unify-8859-on-encoding-mode
f82cb659 2875 "Exists only for backwards compatibility."
d2f613a0
DL		 2876 :group 'mule
2877 :global t)
f82cb659
GM		 2878;; Doc said "obsolete" in 23.1, this statement only added in 24.1.
2879(make-obsolete 'unify-8859-on-encoding-mode "don't use it." "23.1")
2880
d2f613a0 2881(define-minor-mode unify-8859-on-decoding-mode
f82cb659 2882 "Exists only for backwards compatibility."
d2f613a0
DL		 2883 :group 'mule
2884 :global t)
f82cb659
GM		 2885;; Doc said "obsolete" in 23.1, this statement only added in 24.1.
2886(make-obsolete 'unify-8859-on-decoding-mode "don't use it." "23.1")
a127b764 2887
18ccd78a
JB		 2888(defvar nonascii-insert-offset 0)
2889(make-obsolete-variable 'nonascii-insert-offset "do not use it." "23.1")
2890(defvar nonascii-translation-table nil)
2891(make-obsolete-variable 'nonascii-translation-table "do not use it." "23.1")
7dd42fb1 2892
838d78d4
JL		 2893(defvar ucs-names nil
2894 "Alist of cached (CHAR-NAME . CHAR-CODE) pairs.")
2895
2896(defun ucs-names ()
2897 "Return alist of (CHAR-NAME . CHAR-CODE) pairs cached in `ucs-names'."
2898 (or ucs-names
ae63e572
KH		 2899 (let ((bmp-ranges
2900 '((#x0000 . #x33FF)
bf3e70eb 2901 ;; (#x3400 . #x4DBF) CJK Ideographs Extension A
ae63e572 2902 (#x4DC0 . #x4DFF)
bf3e70eb
JB		 2903 ;; (#x4E00 . #x9FFF) CJK Unified Ideographs
2904 (#xA000 . #xD7FF)
ae63e572
KH		 2905 ;; (#xD800 . #xFAFF) Surrogate/Private
2906 (#xFB00 . #xFFFD)))
2907 (upper-ranges
2908 '((#x10000 . #x134FF)
bf3e70eb
JB		 2909 ;; (#x13500 . #x167FF) unused
2910 (#x16800 . #x16A3F)
2911 ;; (#x16A40 . #x1AFFF) unused
2912 (#x1B000 . #x1B0FF)
2913 ;; (#x1B100 . #x1CFFF) unused
ae63e572 2914 (#x1D000 . #x1FFFF)
bf3e70eb 2915 ;; (#x20000 . #xDFFFF) CJK Ideograph Extension A, B, etc, unused
ae63e572
KH		 2916 (#xE0000 . #xE01FF)))
2917 (gc-cons-threshold 10000000)
2918 c end name names)
2919 (dolist (range bmp-ranges)
2920 (setq c (car range)
2921 end (cdr range))
2922 (while (<= c end)
2923 (if (setq name (get-char-code-property c 'name))
2924 (push (cons name c) names))
2925 (if (setq name (get-char-code-property c 'old-name))
2926 (push (cons name c) names))
2927 (setq c (1+ c))))
2928 (dolist (range upper-ranges)
2929 (setq c (car range)
2930 end (cdr range))
2931 (while (<= c end)
2932 (if (setq name (get-char-code-property c 'name))
2933 (push (cons name c) names))
2934 (setq c (1+ c))))
e0727873 2935 (setq ucs-names names))))
838d78d4 2936
838d78d4
JL		 2937(defun read-char-by-name (prompt)
2938 "Read a character by its Unicode name or hex number string.
c0a67379 2939Display PROMPT and read a string that represents a character by its
a167e505
EZ		 2940Unicode property `name' or `old-name'.
2941
2942This function returns the character as a number.
2943
2944You can type a few of the first letters of the Unicode name and
2945use completion. If you type a substring of the Unicode name
2946preceded by an asterisk `*' and use completion, it will show all
2947the characters whose names include that substring, not necessarily
2948at the beginning of the name.
2949
2950This function also accepts a hexadecimal number of Unicode code
2951point or a number in hash notation, e.g. #o21430 for octal,
2952#x2318 for hex, or #10r8984 for decimal."
279f9b06 2953 (let* ((enable-recursive-minibuffers t)
fb9f7146 2954 (completion-ignore-case t)
279f9b06
JL		 2955 (input
2956 (completing-read
2957 prompt
2958 (lambda (string pred action)
fb9f7146
GM		 2959 (if (eq action 'metadata)
2960 '(metadata (category . unicode-name))
2961 (complete-with-action action (ucs-names) string pred)))))
34b4b7eb
CY		 2962 (char
2963 (cond
2964 ((string-match-p "\\`[0-9a-fA-F]+\\'" input)
2965 (string-to-number input 16))
2966 ((string-match-p "\\`#" input)
2967 (read input))
2968 (t
2969 (cdr (assoc-string input (ucs-names) t))))))
2970 (unless (characterp char)
2971 (error "Invalid character"))
2972 char))
838d78d4 2973
2a1e2476 2974(define-obsolete-function-alias 'ucs-insert 'insert-char "24.3")
ddfc8813 2975(define-key ctl-x-map "8\r" 'insert-char)
a127b764 2976
4ed46869 2977;;; mule-cmds.el ends here
Unnamed repository; edit this file 'description' to name the repository.