HCoop / bpt / emacs.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
vc-log-incoming/vc-log-outgoing improvements for Git.
[bpt/emacs.git] / lisp / international / mule-cmds.el
CommitLineData
caff3c0a 1;;; mule-cmds.el --- commands for multilingual environment -*-coding: iso-2022-7bit -*-
9ee5b744 2
26b3dce6 3;; Copyright (C) 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006,
114f9c96 4;; 2007, 2008, 2009, 2010 Free Software Foundation, Inc.
7976eda0 5;; Copyright (C) 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004,
114f9c96 6;; 2005, 2006, 2007, 2008, 2009, 2010
2fd125a3
KH		 7;; National Institute of Advanced Industrial Science and Technology (AIST)
8;; Registration Number H14PRO021
8f924df7 9;; Copyright (C) 2003
97941b05
KH		 10;; National Institute of Advanced Industrial Science and Technology (AIST)
11;; Registration Number H13PRO009
4ed46869 12
1f547b92 13;; Keywords: mule, i18n
4ed46869
KH		 14
15;; This file is part of GNU Emacs.
16
4936186e 17;; GNU Emacs is free software: you can redistribute it and/or modify
4ed46869 18;; it under the terms of the GNU General Public License as published by
4936186e
GM		 19;; the Free Software Foundation, either version 3 of the License, or
20;; (at your option) any later version.
4ed46869
KH		 21
22;; GNU Emacs is distributed in the hope that it will be useful,
23;; but WITHOUT ANY WARRANTY; without even the implied warranty of
24;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
25;; GNU General Public License for more details.
26
27;; You should have received a copy of the GNU General Public License
4936186e 28;; along with GNU Emacs. If not, see <http://www.gnu.org/licenses/>.
4ed46869 29
60370d40
PJ		 30;;; Commentary:
31
4ed46869
KH		 32;;; Code:
33
a4723629
GM		 34(eval-when-compile (require 'cl)) ; letf
35
26b3dce6
GM		 36(defvar dos-codepage)
37(autoload 'widget-value "wid-edit")
cda74479 38
4a91d930 39(defvar mac-system-coding-system)
4a91d930 40
4ed46869
KH		 41;;; MULE related key bindings and menus.
42
64657387
SM		 43(defvar mule-keymap
44 (let ((map (make-sparse-keymap)))
45 (define-key map "f" 'set-buffer-file-coding-system)
46 (define-key map "r" 'revert-buffer-with-coding-system)
47 (define-key map "F" 'set-file-name-coding-system)
48 (define-key map "t" 'set-terminal-coding-system)
49 (define-key map "k" 'set-keyboard-coding-system)
50 (define-key map "p" 'set-buffer-process-coding-system)
51 (define-key map "x" 'set-selection-coding-system)
52 (define-key map "X" 'set-next-selection-coding-system)
53 (define-key map "\C-\\" 'set-input-method)
54 (define-key map "c" 'universal-coding-system-argument)
55 (define-key map "l" 'set-language-environment)
56 map)
33d17698 57 "Keymap for Mule (Multilingual environment) specific commands.")
4ed46869 58
8f81f784 59;; Keep "C-x C-m ..." for mule specific commands.
0709d285 60(define-key ctl-x-map "\C-m" mule-keymap)
ef8a8c8c 61
dcad02bc 62(defvar describe-language-environment-map
64657387
SM		 63 (let ((map (make-sparse-keymap "Describe Language Environment")))
64 (define-key map
8f43cbf3 65 [Default] `(menu-item ,(purecopy "Default") describe-specified-language-support))
64657387 66 map))
15b3e511 67
dcad02bc 68(defvar setup-language-environment-map
64657387
SM		 69 (let ((map (make-sparse-keymap "Set Language Environment")))
70 (define-key map
8f43cbf3 71 [Default] `(menu-item ,(purecopy "Default") setup-specified-language-environment))
64657387 72 map))
15b3e511 73
dcad02bc 74(defvar set-coding-system-map
64657387
SM		 75 (let ((map (make-sparse-keymap "Set Coding System")))
76 (define-key-after map [universal-coding-system-argument]
8f43cbf3
DN		 77 `(menu-item ,(purecopy "For Next Command") universal-coding-system-argument
78 :help ,(purecopy "Coding system to be used by next command")))
04991a1c 79 (define-key-after map [separator-1] menu-bar-separator)
64657387 80 (define-key-after map [set-buffer-file-coding-system]
8f43cbf3
DN		 81 `(menu-item ,(purecopy "For Saving This Buffer") set-buffer-file-coding-system
82 :help ,(purecopy "How to encode this buffer when saved")))
64657387 83 (define-key-after map [revert-buffer-with-coding-system]
8f43cbf3 84 `(menu-item ,(purecopy "For Reverting This File Now")
64657387
SM		 85 revert-buffer-with-coding-system
86 :enable buffer-file-name
8f43cbf3 87 :help ,(purecopy "Revisit this file immediately using specified coding system")))
64657387 88 (define-key-after map [set-file-name-coding-system]
8f43cbf3
DN		 89 `(menu-item ,(purecopy "For File Name") set-file-name-coding-system
90 :help ,(purecopy "How to decode/encode file names")))
04991a1c 91 (define-key-after map [separator-2] menu-bar-separator)
64657387
SM		 92
93 (define-key-after map [set-keyboard-coding-system]
8f43cbf3
DN		 94 `(menu-item ,(purecopy "For Keyboard") set-keyboard-coding-system
95 :help ,(purecopy "How to decode keyboard input")))
64657387 96 (define-key-after map [set-terminal-coding-system]
8f43cbf3 97 `(menu-item ,(purecopy "For Terminal") set-terminal-coding-system
64657387 98 :enable (null (memq initial-window-system '(x w32 ns)))
8f43cbf3 99 :help ,(purecopy "How to encode terminal output")))
04991a1c 100 (define-key-after map [separator-3] menu-bar-separator)
64657387
SM		 101
102 (define-key-after map [set-selection-coding-system]
8f43cbf3 103 `(menu-item ,(purecopy "For X Selections/Clipboard") set-selection-coding-system
64657387 104 :visible (display-selections-p)
8f43cbf3 105 :help ,(purecopy "How to en/decode data to/from selection/clipboard")))
64657387 106 (define-key-after map [set-next-selection-coding-system]
8f43cbf3 107 `(menu-item ,(purecopy "For Next X Selection") set-next-selection-coding-system
64657387 108 :visible (display-selections-p)
8f43cbf3 109 :help ,(purecopy "How to en/decode next selection/clipboard operation")))
64657387 110 (define-key-after map [set-buffer-process-coding-system]
8f43cbf3 111 `(menu-item ,(purecopy "For I/O with Subprocess") set-buffer-process-coding-system
64657387
SM		 112 :visible (fboundp 'start-process)
113 :enable (get-buffer-process (current-buffer))
8f43cbf3 114 :help ,(purecopy "How to en/decode I/O from/to subprocess connected to this buffer")))
64657387
SM		 115 map))
116
117(defvar mule-menu-keymap
118 (let ((map (make-sparse-keymap "Mule (Multilingual Environment)")))
119 (define-key-after map [set-language-environment]
8f43cbf3 120 `(menu-item ,(purecopy "Set Language Environment") ,setup-language-environment-map))
04991a1c 121 (define-key-after map [separator-mule] menu-bar-separator)
64657387
SM		 122
123 (define-key-after map [toggle-input-method]
8f43cbf3 124 `(menu-item ,(purecopy "Toggle Input Method") toggle-input-method))
64657387 125 (define-key-after map [set-input-method]
8f43cbf3 126 `(menu-item ,(purecopy "Select Input Method...") set-input-method))
64657387 127 (define-key-after map [describe-input-method]
8f43cbf3 128 `(menu-item ,(purecopy "Describe Input Method") describe-input-method))
04991a1c 129 (define-key-after map [separator-input-method] menu-bar-separator)
64657387
SM		 130
131 (define-key-after map [set-various-coding-system]
8f43cbf3 132 `(menu-item ,(purecopy "Set Coding Systems") ,set-coding-system-map
597e2240 133 :enable (default-value 'enable-multibyte-characters)))
64657387 134 (define-key-after map [view-hello-file]
8f43cbf3 135 `(menu-item ,(purecopy "Show Multi-lingual Text") view-hello-file
64657387
SM		 136 :enable (file-readable-p
137 (expand-file-name "HELLO" data-directory))
8f43cbf3 138 :help ,(purecopy "Display file which says HELLO in many languages")))
04991a1c 139 (define-key-after map [separator-coding-system] menu-bar-separator)
64657387
SM		 140
141 (define-key-after map [describe-language-environment]
8f43cbf3 142 `(menu-item ,(purecopy "Describe Language Environment")
71078429 143 ,describe-language-environment-map
8f43cbf3 144 :help ,(purecopy "Show multilingual settings for a specific language")))
64657387 145 (define-key-after map [describe-input-method]
8f43cbf3
DN		 146 `(menu-item ,(purecopy "Describe Input Method...") describe-input-method
147 :help ,(purecopy "Keyboard layout for a specific input method")))
64657387 148 (define-key-after map [describe-coding-system]
8f43cbf3 149 `(menu-item ,(purecopy "Describe Coding System...") describe-coding-system))
64657387 150 (define-key-after map [list-character-sets]
8f43cbf3
DN		 151 `(menu-item ,(purecopy "List Character Sets") list-character-sets
152 :help ,(purecopy "Show table of available character sets")))
64657387 153 (define-key-after map [mule-diag]
8f43cbf3
DN		 154 `(menu-item ,(purecopy "Show All of Mule Status") mule-diag
155 :help ,(purecopy "Display multilingual environment settings")))
64657387
SM		 156 map)
157 "Keymap for Mule (Multilingual environment) menu specific commands.")
cda74479 158
4ed46869
KH		 159;; This should be a single character key binding because users use it
160;; very frequently while editing multilingual text. Now we can use
161;; only two such keys: "\C-\\" and "\C-^", but the latter is not
162;; convenient because it requires shifting on most keyboards. An
163;; alternative is "\C-\]" which is now bound to `abort-recursive-edit'
164;; but it won't be used that frequently.
165(define-key global-map "\C-\\" 'toggle-input-method)
166
5f395df3
SM		 167;; This is no good because people often type Shift-SPC
168;; meaning to type SPC. -- rms.
169;; ;; Here's an alternative key binding for X users (Shift-SPACE).
170;; (define-key global-map [?\S- ] 'toggle-input-method)
b4fba33f 171
464cc130
KH		 172;;; Mule related hyperlinks.
173(defconst help-xref-mule-regexp-template
174 (purecopy (concat "\\(\\<\\("
175 "\\(coding system\\)\\|"
d0c40faa
KH		 176 "\\(input method\\)\\|"
177 "\\(character set\\)\\|"
178 "\\(charset\\)"
464cc130
KH		 179 "\\)\\s-+\\)?"
180 ;; Note starting with word-syntax character:
181 "`\\(\\sw\\(\\sw\\|\\s_\\)+\\)'")))
182
26d87040 183(defun coding-system-change-eol-conversion (coding-system eol-type)
caff3c0a 184 "Return a coding system which differs from CODING-SYSTEM in EOL conversion.
26d87040
EZ		 185The returned coding system converts end-of-line by EOL-TYPE
186but text as the same way as CODING-SYSTEM.
187EOL-TYPE should be `unix', `dos', `mac', or nil.
188If EOL-TYPE is nil, the returned coding system detects
189how end-of-line is formatted automatically while decoding.
190
191EOL-TYPE can be specified by an integer 0, 1, or 2.
192They means `unix', `dos', and `mac' respectively."
193 (if (symbolp eol-type)
194 (setq eol-type (cond ((eq eol-type 'unix) 0)
195 ((eq eol-type 'dos) 1)
196 ((eq eol-type 'mac) 2)
197 (t eol-type))))
1d5b0c66
CY		 198 ;; We call `coding-system-base' before `coding-system-eol-type',
199 ;; because the coding-system may not be initialized until then.
200 (let* ((base (coding-system-base coding-system))
201 (orig-eol-type (coding-system-eol-type coding-system)))
202 (cond ((vectorp orig-eol-type)
203 (if (not eol-type)
204 coding-system
205 (aref orig-eol-type eol-type)))
206 ((not eol-type)
207 base)
208 ((= eol-type orig-eol-type)
209 coding-system)
210 ((progn (setq orig-eol-type (coding-system-eol-type base))
211 (vectorp orig-eol-type))
212 (aref orig-eol-type eol-type)))))
26d87040
EZ		 213
214(defun coding-system-change-text-conversion (coding-system coding)
215 "Return a coding system which differs from CODING-SYSTEM in text conversion.
216The returned coding system converts text by CODING
217but end-of-line as the same way as CODING-SYSTEM.
218If CODING is nil, the returned coding system detects
219how text is formatted automatically while decoding."
1d77e15a
JR		 220 (let ((eol-type (coding-system-eol-type coding-system)))
221 (coding-system-change-eol-conversion
222 (if coding coding 'undecided)
223 (if (numberp eol-type) (aref [unix dos mac] eol-type)))))
26d87040 224
be7ca044
KH		 225;; Canonicalize the coding system name NAME by removing some prefixes
226;; and delimiter characters. Support function of
227;; coding-system-from-name.
228(defun canonicalize-coding-system-name (name)
ca4f0e9a
KH		 229 (if (string-match "^\\(ms\\|ibm\\|windows-\\)\\([0-9]+\\)$" name)
230 ;; "ms950", "ibm950", "windows-950" -> "cp950"
231 (concat "cp" (match-string 2 name))
232 (if (string-match "^iso[-_ ]?[0-9]" name)
233 ;; "iso-8859-1" -> "8859-1", "iso-2022-jp" ->"2022-jp"
234 (setq name (substring name (1- (match-end 0)))))
235 (let ((idx (string-match "[-_ /]" name)))
236 ;; Delete "-", "_", " ", "/" but do distinguish "16-be" and "16be".
237 (while idx
238 (if (and (>= idx 2)
239 (eq (string-match "16-[lb]e$" name (- idx 2))
240 (- idx 2)))
241 (setq idx (string-match "[-_ /]" name (match-end 0)))
242 (setq name (concat (substring name 0 idx) (substring name (1+ idx)))
243 idx (string-match "[-_ /]" name idx))))
244 name)))
be7ca044
KH		 245
246(defun coding-system-from-name (name)
247 "Return a coding system whose name matches with NAME (string or symbol)."
248 (let (sym)
249 (if (stringp name) (setq sym (intern name))
250 (setq sym name name (symbol-name name)))
251 (if (coding-system-p sym)
252 sym
253 (let ((eol-type
254 (if (string-match "-\\(unix\\|dos\\|mac\\)$" name)
255 (prog1 (intern (match-string 1 name))
256 (setq name (substring name 0 (match-beginning 0)))))))
257 (setq name (canonicalize-coding-system-name (downcase name)))
258 (catch 'tag
259 (dolist (elt (coding-system-list))
260 (if (string= (canonicalize-coding-system-name (symbol-name elt))
261 name)
262 (throw 'tag (if eol-type (coding-system-change-eol-conversion
263 elt eol-type)
264 elt)))))))))
265
4ed46869 266(defun toggle-enable-multibyte-characters (&optional arg)
6998e1a1 267 "Change whether this buffer uses multibyte characters.
49275d55 268With ARG, use multibyte characters if the ARG is positive.
6998e1a1
RS		 269
270Note that this command does not convert the byte contents of
271the buffer; it only changes the way those bytes are interpreted.
272In general, therefore, this command *changes* the sequence of
273characters that the current buffer contains.
274
caff3c0a
JB		 275We suggest you avoid using this command unless you know what you are
276doing. If you use it by mistake, and the buffer is now displayed
6998e1a1 277wrong, use this command again to toggle back to the right mode."
4ed46869 278 (interactive "P")
b7079457
RS		 279 (let ((new-flag
280 (if (null arg) (null enable-multibyte-characters)
281 (> (prefix-numeric-value arg) 0))))
282 (set-buffer-multibyte new-flag))
4ed46869
KH		 283 (force-mode-line-update))
284
285(defun view-hello-file ()
cf2fc7e9 286 "Display the HELLO file, which lists many languages and characters."
4ed46869 287 (interactive)
8f81f784 288 ;; We have to decode the file in any environment.
1cfb415b 289 (letf ((coding-system-for-read 'iso-2022-7bit))
a4723629 290 (view-file (expand-file-name "HELLO" data-directory))))
4ed46869 291
9e9a77a6 292(defun universal-coding-system-argument (coding-system)
15b3e511 293 "Execute an I/O command using the specified coding system."
9e9a77a6
RS		 294 (interactive
295 (let ((default (and buffer-file-coding-system
34104362 296 (not (eq (coding-system-type buffer-file-coding-system)
a0d96cad 297 'undecided))
9e9a77a6
RS		 298 buffer-file-coding-system)))
299 (list (read-coding-system
300 (if default
5b76833f 301 (format "Coding system for following command (default %s): " default)
9e9a77a6
RS		 302 "Coding system for following command: ")
303 default))))
304 (let* ((keyseq (read-key-sequence
e14a8f4c 305 (format "Command to execute with %s:" coding-system)))
04363179
GM		 306 (cmd (key-binding keyseq))
307 prefix)
735c5b17
GM		 308 ;; read-key-sequence ignores quit, so make an explicit check.
309 ;; Like many places, this assumes quit == C-g, but it need not be.
1e4bd40d 310 (if (equal last-input-event ?\C-g)
735c5b17 311 (keyboard-quit))
4a304bb2 312 (when (memq cmd '(universal-argument digit-argument))
04363179 313 (call-interactively cmd)
a1506d29 314
04363179
GM		 315 ;; Process keys bound in `universal-argument-map'.
316 (while (progn
317 (setq keyseq (read-key-sequence nil t)
318 cmd (key-binding keyseq t))
319 (not (eq cmd 'universal-argument-other-key)))
320 (let ((current-prefix-arg prefix-arg)
61a846fb 321 ;; Have to bind `last-command-event' here so that
6b61353c 322 ;; `digit-argument', for instance, can compute the
04363179 323 ;; prefix arg.
61a846fb 324 (last-command-event (aref keyseq 0)))
04363179
GM		 325 (call-interactively cmd)))
326
1f547b92 327 ;; This is the final call to `universal-argument-other-key', which
04363179
GM		 328 ;; set's the final `prefix-arg.
329 (let ((current-prefix-arg prefix-arg))
330 (call-interactively cmd))
a1506d29 331
04363179
GM		 332 ;; Read the command to execute with the given prefix arg.
333 (setq prefix prefix-arg
334 keyseq (read-key-sequence nil t)
335 cmd (key-binding keyseq)))
336
15b3e511 337 (let ((coding-system-for-read coding-system)
04363179 338 (coding-system-for-write coding-system)
0e9ec609 339 (coding-system-require-warning t)
04363179 340 (current-prefix-arg prefix))
15b3e511
KH		 341 (message "")
342 (call-interactively cmd))))
343
de94d711 344(defun set-default-coding-systems (coding-system)
0c3154d2 345 "Set default value of various coding systems to CODING-SYSTEM.
387136f6 346This sets the following coding systems:
0c3154d2 347 o coding system of a newly created buffer
8efc03e1
KH		 348 o default coding system for subprocess I/O
349This also sets the following values:
d3e4babd 350 o default value used as `file-name-coding-system' for converting file names
caff3c0a 351 if CODING-SYSTEM is ASCII-compatible
d8cb7e3d 352 o default value for the command `set-terminal-coding-system'
d3e4babd 353 o default value for the command `set-keyboard-coding-system'
caff3c0a 354 if CODING-SYSTEM is ASCII-compatible"
de94d711
KH		 355 (check-coding-system coding-system)
356 (setq-default buffer-file-coding-system coding-system)
6b61353c
KH		 357 (if (fboundp 'ucs-set-table-for-input)
358 (dolist (buffer (buffer-list))
359 (or (local-variable-p 'buffer-file-coding-system buffer)
360 (ucs-set-table-for-input buffer))))
361
db9aec47 362 (if (eq system-type 'darwin)
a41118cc 363 ;; The file-name coding system on Darwin systems is always utf-8.
db9aec47 364 (setq default-file-name-coding-system 'utf-8)
597e2240 365 (if (and (default-value 'enable-multibyte-characters)
db9aec47
KH		 366 (or (not coding-system)
367 (coding-system-get coding-system 'ascii-compatible-p)))
368 (setq default-file-name-coding-system coding-system)))
970c9391 369 (setq default-terminal-coding-system coding-system)
de94d711 370 (setq default-keyboard-coding-system coding-system)
1d77e15a
JR		 371 ;; Preserve eol-type from existing default-process-coding-systems.
372 ;; On non-unix-like systems in particular, these may have been set
373 ;; carefully by the user, or by the startup code, to deal with the
374 ;; users shell appropriately, so should not be altered by changing
375 ;; language environment.
376 (let ((output-coding
377 (coding-system-change-text-conversion
378 (car default-process-coding-system) coding-system))
379 (input-coding
380 (coding-system-change-text-conversion
381 (cdr default-process-coding-system) coding-system)))
382 (setq default-process-coding-system
383 (cons output-coding input-coding))))
de94d711 384
0c3154d2
KH		 385(defun prefer-coding-system (coding-system)
386 "Add CODING-SYSTEM at the front of the priority list for automatic detection.
387136f6 387This also sets the following coding systems:
0c3154d2 388 o coding system of a newly created buffer
8efc03e1
KH		 389 o default coding system for subprocess I/O
390This also sets the following values:
caff3c0a 391 o default value used as `file-name-coding-system' for converting file names
d8cb7e3d 392 o default value for the command `set-terminal-coding-system'
03c35c83
EZ		 393 o default value for the command `set-keyboard-coding-system'
394
bd3ac67e
EZ		 395If CODING-SYSTEM specifies a certain type of EOL conversion, the coding
396systems set by this function will use that type of EOL conversion.
397
d0d8885d
JB		 398A coding system that requires automatic detection of text+encoding
399\(e.g. undecided, unix) can't be preferred."
0c3154d2
KH		 400 (interactive "zPrefer coding system: ")
401 (if (not (and coding-system (coding-system-p coding-system)))
402 (error "Invalid coding system `%s'" coding-system))
97941b05
KH		 403 (if (memq (coding-system-type coding-system) '(raw-text undecided))
404 (error "Can't prefer the coding system `%s'" coding-system))
405 (let ((base (coding-system-base coding-system))
bd3ac67e 406 (eol-type (coding-system-eol-type coding-system)))
97941b05 407 (set-coding-system-priority base)
32226619 408 (and (called-interactively-p 'interactive)
97941b05
KH		 409 (or (eq base coding-system)
410 (message "Highest priority is set to %s (base of %s)"
411 base coding-system)))
bd3ac67e 412 ;; If they asked for specific EOL conversion, honor that.
6f9dc4fd 413 (if (memq eol-type '(0 1 2))
97941b05
KH		 414 (setq base
415 (coding-system-change-eol-conversion base eol-type)))
416 (set-default-coding-systems base)))
0c3154d2 417
b5edd1d1
KH		 418(defvar sort-coding-systems-predicate nil
419 "If non-nil, a predicate function to sort coding systems.
420
421It is called with two coding systems, and should return t if the first
422one is \"less\" than the second.
423
424The function `sort-coding-systems' use it.")
425
426(defun sort-coding-systems (codings)
427 "Sort coding system list CODINGS by a priority of each coding system.
caff3c0a 428Return the sorted list. CODINGS is modified by side effects.
b5edd1d1
KH		 429
430If a coding system is most preferred, it has the highest priority.
caff3c0a
JB		 431Otherwise, coding systems that correspond to MIME charsets have
432higher priorities. Among them, a coding system included in the
433`coding-system' key of the current language environment has higher
434priority. See also the documentation of `language-info-alist'.
b5edd1d1
KH		 435
436If the variable `sort-coding-systems-predicate' (which see) is
caff3c0a 437non-nil, it is used to sort CODINGS instead."
b5edd1d1
KH		 438 (if sort-coding-systems-predicate
439 (sort codings sort-coding-systems-predicate)
8f924df7
KH		 440 (let* ((from-priority (coding-system-priority-list))
441 (most-preferred (car from-priority))
b5edd1d1
KH		 442 (lang-preferred (get-language-info current-language-environment
443 'coding-system))
444 (func (function
445 (lambda (x)
446 (let ((base (coding-system-base x)))
da1ebad1
KH		 447 ;; We calculate the priority number 0..255 by
448 ;; using the 8 bits PMMLCEII as this:
3ecd3a56
GM		 449 ;; P: 1 if most preferred.
450 ;; MM: greater than 0 if mime-charset.
451 ;; L: 1 if one of the current lang. env.'s codings.
452 ;; C: 1 if one of codings listed in the category list.
453 ;; E: 1 if not XXX-with-esc
da1ebad1
KH		 454 ;; II: if iso-2022 based, 0..3, else 1.
455 (logior
456 (lsh (if (eq base most-preferred) 1 0) 7)
457 (lsh
8f924df7 458 (let ((mime (coding-system-get base :mime-charset)))
b439e72a
DL		 459 ;; Prefer coding systems corresponding to a
460 ;; MIME charset.
b5edd1d1 461 (if mime
b439e72a
DL		 462 ;; Lower utf-16 priority so that we
463 ;; normally prefer utf-8 to it, and put
464 ;; x-ctext below that.
d0d8885d
JB		 465 (cond ((string-match-p "utf-16"
466 (symbol-name mime))
da1ebad1 467 2)
d0d8885d 468 ((string-match-p "^x-" (symbol-name mime))
da1ebad1
KH		 469 1)
470 (t 3))
b5edd1d1 471 0))
da1ebad1
KH		 472 5)
473 (lsh (if (memq base lang-preferred) 1 0) 4)
8f924df7 474 (lsh (if (memq base from-priority) 1 0) 3)
d0d8885d
JB		 475 (lsh (if (string-match-p "-with-esc\\'"
476 (symbol-name base))
da1ebad1 477 0 1) 2)
8f924df7
KH		 478 (if (eq (coding-system-type base) 'iso-2022)
479 (let ((category (coding-system-category base)))
480 ;; For ISO based coding systems, prefer
481 ;; one that doesn't use designation nor
482 ;; locking/single shifting.
a0d96cad
KH		 483 (cond
484 ((or (eq category 'coding-category-iso-8-1)
485 (eq category 'coding-category-iso-8-2))
486 2)
487 ((or (eq category 'coding-category-iso-7-tight)
488 (eq category 'coding-category-iso-7))
489 1)
490 (t
491 0)))
492 1)
a6dfc99b 493 ))))))
b5edd1d1
KH		 494 (sort codings (function (lambda (x y)
495 (> (funcall func x) (funcall func y))))))))
54b226f7 496
3fc7dfe5 497(defun find-coding-systems-region (from to)
54b226f7 498 "Return a list of proper coding systems to encode a text between FROM and TO.
8f924df7 499
d37ef0f6 500If FROM is a string, find coding systems in that instead of the buffer.
54b226f7
KH		 501All coding systems in the list can safely encode any multibyte characters
502in the text.
503
e8dd0160 504If the text contains no multibyte characters, return a list of a single
3fc7dfe5 505element `undecided'."
b5edd1d1
KH		 506 (let ((codings (find-coding-systems-region-internal from to)))
507 (if (eq codings t)
508 ;; The text contains only ASCII characters. Any coding
509 ;; systems are safe.
510 '(undecided)
511 ;; We need copy-sequence because sorting will alter the argument.
512 (sort-coding-systems (copy-sequence codings)))))
54b226f7 513
3fc7dfe5
KH		 514(defun find-coding-systems-string (string)
515 "Return a list of proper coding systems to encode STRING.
516All coding systems in the list can safely encode any multibyte characters
517in STRING.
518
e8dd0160 519If STRING contains no multibyte characters, return a list of a single
3fc7dfe5 520element `undecided'."
b5edd1d1 521 (find-coding-systems-region string nil))
3fc7dfe5
KH		 522
523(defun find-coding-systems-for-charsets (charsets)
524 "Return a list of proper coding systems to encode characters of CHARSETS.
6053d86a 525CHARSETS is a list of character sets.
1f547b92
DL		 526
527This only finds coding systems of type `charset', whose
528`:charset-list' property includes all of CHARSETS (plus `ascii' for
d0d8885d 529ASCII-compatible coding systems). It was used in older versions of
1f547b92
DL		 530Emacs, but is unlikely to be what you really want now."
531 ;; Deal with aliases.
532 (setq charsets (mapcar (lambda (c)
533 (get-charset-property c :name))
534 charsets))
b5edd1d1
KH		 535 (cond ((or (null charsets)
536 (and (= (length charsets) 1)
537 (eq 'ascii (car charsets))))
538 '(undecided))
539 ((or (memq 'eight-bit-control charsets)
540 (memq 'eight-bit-graphic charsets))
1f547b92 541 '(raw-text utf-8-emacs))
b5edd1d1 542 (t
1f547b92
DL		 543 (let (codings)
544 (dolist (cs (coding-system-list t))
c7b4b466
DL		 545 (let ((cs-charsets (and (eq (coding-system-type cs) 'charset)
546 (coding-system-charset-list cs)))
1f547b92
DL		 547 (charsets charsets))
548 (if (coding-system-get cs :ascii-compatible-p)
549 (add-to-list 'cs-charsets 'ascii))
550 (if (catch 'ok
551 (when cs-charsets
552 (while charsets
553 (unless (memq (pop charsets) cs-charsets)
554 (throw 'ok nil)))
555 t))
556 (push cs codings))))
557 (nreverse codings)))))
54b226f7 558
51ed58ea
KH		 559(defun find-multibyte-characters (from to &optional maxcount excludes)
560 "Find multibyte characters in the region specified by FROM and TO.
561If FROM is a string, find multibyte characters in the string.
562The return value is an alist of the following format:
563 ((CHARSET COUNT CHAR ...) ...)
564where
565 CHARSET is a character set,
566 COUNT is a number of characters,
1f547b92 567 CHARs are the characters found from the character set.
51ed58ea 568Optional 3rd arg MAXCOUNT limits how many CHARs are put in the above list.
caff3c0a 569Optional 4th arg EXCLUDES is a list of character sets to be ignored."
51ed58ea
KH		 570 (let ((chars nil)
571 charset char)
572 (if (stringp from)
a0d96cad
KH		 573 (if (multibyte-string-p from)
574 (let ((idx 0))
d0d8885d 575 (while (setq idx (string-match-p "[^\000-\177]" from idx))
a0d96cad
KH		 576 (setq char (aref from idx)
577 charset (char-charset char))
578 (unless (memq charset excludes)
579 (let ((slot (assq charset chars)))
580 (if slot
581 (if (not (memq char (nthcdr 2 slot)))
582 (let ((count (nth 1 slot)))
583 (setcar (cdr slot) (1+ count))
584 (if (or (not maxcount) (< count maxcount))
585 (nconc slot (list char)))))
586 (setq chars (cons (list charset 1 char) chars)))))
587 (setq idx (1+ idx)))))
588 (if enable-multibyte-characters
589 (save-excursion
590 (goto-char from)
591 (while (re-search-forward "[^\000-\177]" to t)
592 (setq char (preceding-char)
593 charset (char-charset char))
594 (unless (memq charset excludes)
51ed58ea
KH		 595 (let ((slot (assq charset chars)))
596 (if slot
a0d96cad 597 (if (not (member char (nthcdr 2 slot)))
51ed58ea
KH		 598 (let ((count (nth 1 slot)))
599 (setcar (cdr slot) (1+ count))
600 (if (or (not maxcount) (< count maxcount))
601 (nconc slot (list char)))))
a0d96cad 602 (setq chars (cons (list charset 1 char) chars)))))))))
51ed58ea
KH		 603 (nreverse chars)))
604
738746ba
KH		 605(defun search-unencodable-char (coding-system)
606 "Search forward from point for a character that is not encodable.
607It asks which coding system to check.
608If such a character is found, set point after that character.
609Otherwise, don't move point.
610
caff3c0a
JB		 611When called from a program, the value is the position of the unencodable
612character found, or nil if all characters are encodable."
738746ba
KH		 613 (interactive
614 (list (let ((default (or buffer-file-coding-system 'us-ascii)))
615 (read-coding-system
5b76833f 616 (format "Coding-system (default %s): " default)
738746ba
KH		 617 default))))
618 (let ((pos (unencodable-char-position (point) (point-max) coding-system)))
619 (if pos
620 (goto-char (1+ pos))
621 (message "All following characters are encodable by %s" coding-system))
622 pos))
623
c83c4f60
RS		 624(defvar last-coding-system-specified nil
625 "Most recent coding system explicitly specified by the user when asked.
626This variable is set whenever Emacs asks the user which coding system
627to use in order to write a file. If you set it to nil explicitly,
628then call `write-region', then afterward this variable will be non-nil
629only if the user was explicitly asked and specified a coding system.")
630
b5edd1d1 631(defvar select-safe-coding-system-accept-default-p nil
fea6b736 632 "If non-nil, a function to control the behavior of coding system selection.
b5edd1d1
KH		 633The meaning is the same as the argument ACCEPT-DEFAULT-P of the
634function `select-safe-coding-system' (which see). This variable
635overrides that argument.")
636
9ee5b744
SM		 637(defun select-safe-coding-system-interactively (from to codings unsafe
638 &optional rejected default)
639 "Select interactively a coding system for the region FROM ... TO.
640FROM can be a string, as in `write-region'.
641CODINGS is the list of base coding systems known to be safe for this region,
642 typically obtained with `find-coding-systems-region'.
643UNSAFE is a list of coding systems known to be unsafe for this region.
644REJECTED is a list of coding systems which were safe but for some reason
645 were not recommended in the particular context.
646DEFAULT is the coding system to use by default in the query."
647 ;; At first, if some defaults are unsafe, record at most 11
648 ;; problematic characters and their positions for them by turning
649 ;; (CODING ...)
650 ;; into
651 ;; ((CODING (POS . CHAR) (POS . CHAR) ...) ...)
652 (if unsafe
653 (setq unsafe
654 (mapcar #'(lambda (coding)
655 (cons coding
656 (if (stringp from)
657 (mapcar #'(lambda (pos)
658 (cons pos (aref from pos)))
659 (unencodable-char-position
660 0 (length from) coding
661 11 from))
662 (mapcar #'(lambda (pos)
663 (cons pos (char-after pos)))
664 (unencodable-char-position
665 from to coding 11)))))
666 unsafe)))
667
668 ;; Change each safe coding system to the corresponding
669 ;; mime-charset name if it is also a coding system. Such a name
670 ;; is more friendly to users.
671 (let ((l codings)
672 mime-charset)
673 (while l
a77c22c2
KH		 674 (setq mime-charset (coding-system-get (car l) :mime-charset))
675 (if (and mime-charset (coding-system-p mime-charset)
676 (coding-system-equal (car l) mime-charset))
9ee5b744
SM		 677 (setcar l mime-charset))
678 (setq l (cdr l))))
679
680 ;; Don't offer variations with locking shift, which you
681 ;; basically never want.
682 (let (l)
683 (dolist (elt codings (setq codings (nreverse l)))
684 (unless (or (eq 'coding-category-iso-7-else
685 (coding-system-category elt))
686 (eq 'coding-category-iso-8-else
687 (coding-system-category elt)))
688 (push elt l))))
689
690 ;; Remove raw-text, emacs-mule and no-conversion unless nothing
691 ;; else is available.
692 (setq codings
693 (or (delq 'raw-text
694 (delq 'emacs-mule
695 (delq 'no-conversion codings)))
696 '(raw-text emacs-mule no-conversion)))
697
698 (let ((window-configuration (current-window-configuration))
699 (bufname (buffer-name))
700 coding-system)
701 (save-excursion
702 ;; If some defaults are unsafe, make sure the offending
703 ;; buffer is displayed.
704 (when (and unsafe (not (stringp from)))
705 (pop-to-buffer bufname)
706 (goto-char (apply 'min (mapcar #'(lambda (x) (car (cadr x)))
707 unsafe))))
708 ;; Then ask users to select one from CODINGS while showing
709 ;; the reason why none of the defaults are not used.
710 (with-output-to-temp-buffer "*Warning*"
711 (with-current-buffer standard-output
712 (if (and (null rejected) (null unsafe))
713 (insert "No default coding systems to try for "
714 (if (stringp from)
715 (format "string \"%s\"." from)
716 (format "buffer `%s'." bufname)))
717 (insert
718 "These default coding systems were tried to encode"
719 (if (stringp from)
720 (concat " \"" (if (> (length from) 10)
721 (concat (substring from 0 10) "...\"")
722 (concat from "\"")))
723 (format " text\nin the buffer `%s'" bufname))
724 ":\n")
725 (let ((pos (point))
726 (fill-prefix " "))
727 (dolist (x (append rejected unsafe))
46c04718 728 (princ " ") (princ x))
9ee5b744
SM		 729 (insert "\n")
730 (fill-region-as-paragraph pos (point)))
731 (when rejected
3fc0b26e
EZ		 732 (insert "These safely encode the text in the buffer,
733but are not recommended for encoding text in this context,
9ee5b744
SM		 734e.g., for sending an email message.\n ")
735 (dolist (x rejected)
736 (princ " ") (princ x))
737 (insert "\n"))
738 (when unsafe
3fc0b26e 739 (insert (if rejected "The other coding systems"
9ee5b744 740 "However, each of them")
3fc0b26e 741 " encountered characters it couldn't encode:\n")
9ee5b744 742 (dolist (coding unsafe)
3fc0b26e 743 (insert (format " %s cannot encode these:" (car coding)))
9ee5b744
SM		 744 (let ((i 0)
745 (func1
746 #'(lambda (bufname pos)
747 (when (buffer-live-p (get-buffer bufname))
748 (pop-to-buffer bufname)
749 (goto-char pos))))
750 (func2
751 #'(lambda (bufname pos coding)
752 (when (buffer-live-p (get-buffer bufname))
753 (pop-to-buffer bufname)
754 (if (< (point) pos)
755 (goto-char pos)
756 (forward-char 1)
757 (search-unencodable-char coding)
758 (forward-char -1))))))
759 (dolist (elt (cdr coding))
760 (insert " ")
761 (if (stringp from)
762 (insert (if (< i 10) (cdr elt) "..."))
763 (if (< i 10)
764 (insert-text-button
765 (cdr elt)
766 :type 'help-xref
4e21e5aa 767 'face 'link
9ee5b744
SM		 768 'help-echo
769 "mouse-2, RET: jump to this character"
770 'help-function func1
771 'help-args (list bufname (car elt)))
772 (insert-text-button
773 "..."
774 :type 'help-xref
4e21e5aa 775 'face 'link
9ee5b744
SM		 776 'help-echo
777 "mouse-2, RET: next unencodable character"
778 'help-function func2
779 'help-args (list bufname (car elt)
780 (car coding)))))
781 (setq i (1+ i))))
782 (insert "\n"))
333f3572 783 (insert (substitute-command-keys "\
7d03c5b1 784
333f3572
KH		 785Click on a character (or switch to this window by `\\[other-window]'\n\
786and select the characters by RET) to jump to the place it appears,\n\
3fc0b26e 787where `\\[universal-argument] \\[what-cursor-position]' will give information about it.\n"))))
7d03c5b1 788 (insert (substitute-command-keys "\nSelect \
3fc0b26e
EZ		 789one of the safe coding systems listed below,\n\
790or cancel the writing with \\[keyboard-quit] and edit the buffer\n\
791 to remove or modify the problematic characters,\n\
792or specify any other coding system (and risk losing\n\
793 the problematic characters).\n\n"))
9ee5b744
SM		 794 (let ((pos (point))
795 (fill-prefix " "))
796 (dolist (x codings)
797 (princ " ") (princ x))
798 (insert "\n")
7d03c5b1 799 (fill-region-as-paragraph pos (point)))))
9ee5b744
SM		 800
801 ;; Read a coding system.
802 (setq coding-system
803 (read-coding-system
804 (format "Select coding system (default %s): " default)
805 default))
806 (setq last-coding-system-specified coding-system))
807
808 (kill-buffer "*Warning*")
809 (set-window-configuration window-configuration)
810 coding-system))
811
b5edd1d1 812(defun select-safe-coding-system (from to &optional default-coding-system
efa2c6d7 813 accept-default-p file)
d5266ddf
KH		 814 "Ask a user to select a safe coding system from candidates.
815The candidates of coding systems which can safely encode a text
b5edd1d1
KH		 816between FROM and TO are shown in a popup window. Among them, the most
817proper one is suggested as the default.
818
d0d8885d 819The list of `buffer-file-coding-system' of the current buffer, the
b56a5ae0 820default `buffer-file-coding-system', and the most preferred coding
d0d8885d
JB		 821system (if it corresponds to a MIME charset) is treated as the
822default coding system list. Among them, the first one that safely
823encodes the text is normally selected silently and returned without
824any user interaction. See also the command `prefer-coding-system'.
d37ef0f6
DL		 825
826However, the user is queried if the chosen coding system is
29d04c4f 827inconsistent with what would be selected by `find-auto-coding' from
d37ef0f6
DL		 828coding cookies &c. if the contents of the region were read from a
829file. (That could lead to data corruption in a file subsequently
830re-visited and edited.)
54b226f7 831
b5edd1d1
KH		 832Optional 3rd arg DEFAULT-CODING-SYSTEM specifies a coding system or a
833list of coding systems to be prepended to the default coding system
0e9ec609 834list. However, if DEFAULT-CODING-SYSTEM is a list and the first
ccb06340 835element is t, the cdr part is used as the default coding system list,
b56a5ae0 836i.e. current `buffer-file-coding-system', default `buffer-file-coding-system',
a985cd2f 837and the most preferred coding system are not used.
54b226f7 838
b5edd1d1
KH		 839Optional 4th arg ACCEPT-DEFAULT-P, if non-nil, is a function to
840determine the acceptability of the silently selected coding system.
841It is called with that coding system, and should return nil if it
842should not be silently selected and thus user interaction is required.
843
efa2c6d7
RS		 844Optional 5th arg FILE is the file name to use for this purpose.
845That is different from `buffer-file-name' when handling `write-region'
846\(for example).
847
d0d8885d
JB		 848The variable `select-safe-coding-system-accept-default-p', if non-nil,
849overrides ACCEPT-DEFAULT-P.
54b226f7
KH		 850
851Kludgy feature: if FROM is a string, the string is the target text,
852and TO is ignored."
64657387 853 (if (not (listp default-coding-system))
b5edd1d1
KH		 854 (setq default-coding-system (list default-coding-system)))
855
29d04c4f
KH		 856 (let ((no-other-defaults nil)
857 auto-cs)
d35deeec 858 (unless (or (stringp from) find-file-literally)
b4dc7d98 859 ;; Find an auto-coding that is specified for the current
29d04c4f
KH		 860 ;; buffer and file from the region FROM and TO.
861 (save-excursion
862 (save-restriction
863 (widen)
864 (goto-char from)
865 (setq auto-cs (find-auto-coding (or file buffer-file-name "")
866 (- to from)))
867 (if auto-cs
868 (if (coding-system-p (car auto-cs))
869 (setq auto-cs (car auto-cs))
870 (display-warning
2364df5c 871 'mule
29d04c4f
KH		 872 (format "\
873Invalid coding system `%s' is specified
874for the current buffer/file by the %s.
875It is highly recommended to fix it before writing to a file."
876 (car auto-cs)
877 (if (eq (cdr auto-cs) :coding) ":coding tag"
2364df5c
JB		 878 (format "variable `%s'" (cdr auto-cs))))
879 :warning)
29d04c4f
KH		 880 (or (yes-or-no-p "Really proceed with writing? ")
881 (error "Save aborted"))
882 (setq auto-cs nil))))))
883
0e9ec609
KH		 884 (if (eq (car default-coding-system) t)
885 (setq no-other-defaults t
886 default-coding-system (cdr default-coding-system)))
887
888 ;; Change elements of the list to (coding . base-coding).
889 (setq default-coding-system
890 (mapcar (function (lambda (x) (cons x (coding-system-base x))))
891 default-coding-system))
892
29d04c4f 893 (if (and auto-cs (not no-other-defaults))
2df48a87
SM		 894 ;; If the file has a coding cookie, use it regardless of any
895 ;; other setting.
29d04c4f 896 (let ((base (coding-system-base auto-cs)))
2df48a87
SM		 897 (unless (memq base '(nil undecided))
898 (setq default-coding-system (list (cons auto-cs base)))
899 (setq no-other-defaults t))))
29d04c4f 900
0e9ec609
KH		 901 (unless no-other-defaults
902 ;; If buffer-file-coding-system is not nil nor undecided, append it
903 ;; to the defaults.
904 (if buffer-file-coding-system
905 (let ((base (coding-system-base buffer-file-coding-system)))
906 (or (eq base 'undecided)
0e9ec609 907 (rassq base default-coding-system)
ccb06340
KH		 908 (setq default-coding-system
909 (append default-coding-system
910 (list (cons buffer-file-coding-system base)))))))
a985cd2f 911
659be258
KH		 912 (unless (and buffer-file-coding-system-explicit
913 (cdr buffer-file-coding-system-explicit))
b56a5ae0 914 ;; If default buffer-file-coding-system is not nil nor undecided,
659be258 915 ;; append it to the defaults.
b56a5ae0
SM		 916 (when (default-value 'buffer-file-coding-system)
917 (let ((base (coding-system-base
918 (default-value 'buffer-file-coding-system))))
919 (or (eq base 'undecided)
920 (rassq base default-coding-system)
921 (setq default-coding-system
922 (append default-coding-system
923 (list (cons (default-value
924 'buffer-file-coding-system)
925 base)))))))
659be258
KH		 926
927 ;; If the most preferred coding system has the property mime-charset,
928 ;; append it to the defaults.
929 (let ((preferred (coding-system-priority-list t))
930 base)
931 (and (coding-system-p preferred)
932 (setq base (coding-system-base preferred))
933 (coding-system-get preferred :mime-charset)
934 (not (rassq base default-coding-system))
935 (setq default-coding-system
936 (append default-coding-system
937 (list (cons preferred base))))))))
29d04c4f
KH		 938
939 (if select-safe-coding-system-accept-default-p
940 (setq accept-default-p select-safe-coding-system-accept-default-p))
941
bae7cd08 942 ;; Decide the eol-type from the top of the default codings,
b56a5ae0 943 ;; current buffer-file-coding-system, or default buffer-file-coding-system.
bae7cd08
KH		 944 (if default-coding-system
945 (let ((default-eol-type (coding-system-eol-type
946 (caar default-coding-system))))
947 (if (and (vectorp default-eol-type) buffer-file-coding-system)
d35deeec 948 (setq default-eol-type (coding-system-eol-type
bae7cd08 949 buffer-file-coding-system)))
b56a5ae0
SM		 950 (if (and (vectorp default-eol-type)
951 (default-value 'buffer-file-coding-system))
952 (setq default-eol-type
953 (coding-system-eol-type
954 (default-value 'buffer-file-coding-system))))
bae7cd08
KH		 955 (if (and default-eol-type (not (vectorp default-eol-type)))
956 (dolist (elt default-coding-system)
957 (setcar elt (coding-system-change-eol-conversion
958 (car elt) default-eol-type))))))
959
29d04c4f
KH		 960 (let ((codings (find-coding-systems-region from to))
961 (coding-system nil)
c381cb8d 962 (tick (if (not (stringp from)) (buffer-chars-modified-tick)))
29d04c4f 963 safe rejected unsafe)
bae7cd08
KH		 964 (if (eq (car codings) 'undecided)
965 ;; Any coding system is ok.
966 (setq coding-system (caar default-coding-system))
967 ;; Reverse the list so that elements are accumulated in safe,
968 ;; rejected, and unsafe in the correct order.
969 (setq default-coding-system (nreverse default-coding-system))
970
971 ;; Classify the defaults into safe, rejected, and unsafe.
972 (dolist (elt default-coding-system)
973 (if (or (eq (car codings) 'undecided)
974 (memq (cdr elt) codings))
975 (if (and (functionp accept-default-p)
976 (not (funcall accept-default-p (cdr elt))))
977 (push (car elt) rejected)
978 (push (car elt) safe))
979 (push (car elt) unsafe)))
980 (if safe
981 (setq coding-system (car safe))))
29d04c4f
KH		 982
983 ;; If all the defaults failed, ask a user.
bae7cd08 984 (when (not coding-system)
29d04c4f
KH		 985 (setq coding-system (select-safe-coding-system-interactively
986 from to codings unsafe rejected (car codings))))
987
29d04c4f
KH		 988 ;; Check we're not inconsistent with what `coding:' spec &c would
989 ;; give when file is re-read.
990 ;; But don't do this if we explicitly ignored the cookie
991 ;; by using `find-file-literally'.
992 (when (and auto-cs
993 (not (and
994 coding-system
995 (memq (coding-system-type coding-system) '(0 5)))))
6b61353c
KH		 996 ;; Merge coding-system and auto-cs as far as possible.
997 (if (not coding-system)
998 (setq coding-system auto-cs)
999 (if (not auto-cs)
1000 (setq auto-cs coding-system)
1001 (let ((eol-type-1 (coding-system-eol-type coding-system))
1002 (eol-type-2 (coding-system-eol-type auto-cs)))
1003 (if (eq (coding-system-base coding-system) 'undecided)
1004 (setq coding-system (coding-system-change-text-conversion
1005 coding-system auto-cs))
1006 (if (eq (coding-system-base auto-cs) 'undecided)
1007 (setq auto-cs (coding-system-change-text-conversion
1008 auto-cs coding-system))))
1009 (if (vectorp eol-type-1)
1010 (or (vectorp eol-type-2)
1011 (setq coding-system (coding-system-change-eol-conversion
1012 coding-system eol-type-2)))
1013 (if (vectorp eol-type-2)
1014 (setq auto-cs (coding-system-change-eol-conversion
1015 auto-cs eol-type-1)))))))
1016
1017 (if (and auto-cs
d37ef0f6
DL		 1018 ;; Don't barf if writing a compressed file, say.
1019 ;; This check perhaps isn't ideal, but is probably
1020 ;; the best thing to do.
1021 (not (auto-coding-alist-lookup (or file buffer-file-name "")))
6b61353c 1022 (not (coding-system-equal coding-system auto-cs)))
43afed8c
RS		 1023 (unless (yes-or-no-p
1024 (format "Selected encoding %s disagrees with \
1025%s specified by file contents. Really save (else edit coding cookies \
1026and try again)? " coding-system auto-cs))
29d04c4f 1027 (error "Save aborted"))))
c381cb8d 1028 (when (and tick (/= tick (buffer-chars-modified-tick)))
86c3a9fb 1029 (error "Cancelled because the buffer was modified"))
29d04c4f 1030 coding-system)))
54b226f7
KH		 1031
1032(setq select-safe-coding-system-function 'select-safe-coding-system)
1033
46babb23
KH		 1034(defun select-message-coding-system ()
1035 "Return a coding system to encode the outgoing message of the current buffer.
1036It at first tries the first coding system found in these variables
1037in this order:
1038 (1) local value of `buffer-file-coding-system'
1039 (2) value of `sendmail-coding-system'
b5edd1d1 1040 (3) value of `default-sendmail-coding-system'
b56a5ae0 1041 (4) default value of `buffer-file-coding-system'
46babb23
KH		 1042If the found coding system can't encode the current buffer,
1043or none of them are bound to a coding system,
48e41165 1044it asks the user to select a proper coding system."
46babb23 1045 (let ((coding (or (and (local-variable-p 'buffer-file-coding-system)
b5edd1d1
KH		 1046 buffer-file-coding-system)
1047 sendmail-coding-system
1048 default-sendmail-coding-system
b56a5ae0 1049 (default-value 'buffer-file-coding-system))))
46babb23 1050 (if (eq coding 'no-conversion)
d37ef0f6 1051 ;; We should never use no-conversion for outgoing mail.
46babb23
KH		 1052 (setq coding nil))
1053 (if (fboundp select-safe-coding-system-function)
1054 (funcall select-safe-coding-system-function
b5edd1d1 1055 (point-min) (point-max) coding
209c73b0 1056 (function (lambda (x) (coding-system-get x :mime-charset))))
46babb23 1057 coding)))
4ed46869 1058\f
03c35c83 1059;;; Language support stuff.
4ed46869 1060
4ed46869 1061(defvar language-info-alist nil
2c395d56 1062 "Alist of language environment definitions.
4ed46869
KH		 1063Each element looks like:
1064 (LANGUAGE-NAME . ((KEY . INFO) ...))
2c395d56
RS		 1065where LANGUAGE-NAME is a string, the name of the language environment,
1066KEY is a symbol denoting the kind of information, and
1067INFO is the data associated with KEY.
1068Meaningful values for KEY include
1069
1070 documentation value is documentation of what this language environment
1071 is meant for, and how to use it.
7dd42fb1
KH		 1072 charset value is a list of the character sets mainly used
1073 by this language environment.
d37ef0f6
DL		 1074 sample-text value is an expression which is evalled to generate
1075 a line of text written using characters appropriate
1076 for this language environment.
2c395d56
RS		 1077 setup-function value is a function to call to switch to this
1078 language environment.
1079 exit-function value is a function to call to leave this
1080 language environment.
d0d8885d
JB		 1081 coding-system value is a list of coding systems that are good for
1082 saving text written in this language environment.
2c395d56
RS		 1083 This list serves as suggestions to the user;
1084 in effect, as a kind of documentation.
1085 coding-priority value is a list of coding systems for this language
1086 environment, in order of decreasing priority.
1087 This is used to set up the coding system priority
45d08cb2 1088 list when you switch to this language environment.
ddb5c041 1089 nonascii-translation
7dd42fb1
KH		 1090 value is a charset of dimension one to use for
1091 converting a unibyte character to multibyte
1092 and vice versa.
ddb5c041
KH		 1093 input-method value is a default input method for this language
1094 environment.
7624ebb9
KH		 1095 features value is a list of features requested in this
1096 language environment.
6b61353c 1097 ctext-non-standard-encodings
d0d8885d
JB		 1098 value is a list of non-standard encoding names used
1099 in extended segments of CTEXT. See the variable
1100 `ctext-non-standard-encodings' for more detail.
ddb5c041 1101
b9631bb2 1102The following key takes effect only when multibyte characters are
597e2240 1103globally disabled, i.e. the default value of `enable-multibyte-characters'
b9631bb2 1104is nil (which is an obsolete and deprecated use):
ddb5c041 1105
d0d8885d
JB		 1106 unibyte-display value is a coding system to encode characters for
1107 the terminal. Characters in the range of 160 to
1108 255 display not as octal escapes, but as non-ASCII
1109 characters in this language environment.")
2c395d56
RS		 1110
1111(defun get-language-info (lang-env key)
1112 "Return information listed under KEY for language environment LANG-ENV.
1113KEY is a symbol denoting the kind of information.
1114For a list of useful values for KEY and their meanings,
1115see `language-info-alist'."
1116 (if (symbolp lang-env)
1117 (setq lang-env (symbol-name lang-env)))
f15466c5 1118 (let ((lang-slot (assoc-string lang-env language-info-alist t)))
4ed46869
KH		 1119 (if lang-slot
1120 (cdr (assq key (cdr lang-slot))))))
1121
f08adf27 1122(defun set-language-info (lang-env key info)
2c395d56
RS		 1123 "Modify part of the definition of language environment LANG-ENV.
1124Specifically, this stores the information INFO under KEY
1125in the definition of this language environment.
4ed46869 1126KEY is a symbol denoting the kind of information.
2c395d56 1127INFO is the value for that information.
281d03ec 1128
2c395d56 1129For a list of useful values for KEY and their meanings,
f08adf27 1130see `language-info-alist'."
2c395d56
RS		 1131 (if (symbolp lang-env)
1132 (setq lang-env (symbol-name lang-env)))
eec5c8f9
KH		 1133 (set-language-info-internal lang-env key info)
1134 (if (equal lang-env current-language-environment)
d042f8b4 1135 (cond ((eq key 'coding-priority)
e9a8ed3c
KH		 1136 (set-language-environment-coding-systems lang-env)
1137 (set-language-environment-charset lang-env))
d042f8b4
KH		 1138 ((eq key 'input-method)
1139 (set-language-environment-input-method lang-env))
1140 ((eq key 'nonascii-translation)
1141 (set-language-environment-nonascii-translation lang-env))
1142 ((eq key 'charset)
1143 (set-language-environment-charset lang-env))
597e2240 1144 ((and (not (default-value 'enable-multibyte-characters))
d042f8b4
KH		 1145 (or (eq key 'unibyte-syntax) (eq key 'unibyte-display)))
1146 (set-language-environment-unibyte lang-env)))))
eec5c8f9
KH		 1147
1148(defun set-language-info-internal (lang-env key info)
1149 "Internal use only.
1150Arguments are the same as `set-language-info'."
4ed46869 1151 (let (lang-slot key-slot)
2c395d56 1152 (setq lang-slot (assoc lang-env language-info-alist))
4ed46869 1153 (if (null lang-slot) ; If no slot for the language, add it.
2c395d56 1154 (setq lang-slot (list lang-env)
4ed46869
KH		 1155 language-info-alist (cons lang-slot language-info-alist)))
1156 (setq key-slot (assq key lang-slot))
1157 (if (null key-slot) ; If no slot for the key, add it.
1158 (progn
1159 (setq key-slot (list key))
1160 (setcdr lang-slot (cons key-slot (cdr lang-slot)))))
590dbcba 1161 (setcdr key-slot (purecopy info))
1169bd86 1162 ;; Update the custom-type of `current-language-environment'.
590dbcba 1163 (put 'current-language-environment 'custom-type
1169bd86
MR		 1164 (cons 'choice (mapcar
1165 (lambda (lang)
756e055f
MR		 1166 (list 'const lang))
1167 (sort (mapcar 'car language-info-alist) 'string<))))))
4ed46869 1168
2c395d56
RS		 1169(defun set-language-info-alist (lang-env alist &optional parents)
1170 "Store ALIST as the definition of language environment LANG-ENV.
1171ALIST is an alist of KEY and INFO values. See the documentation of
98c6d6ed 1172`language-info-alist' for the meanings of KEY and INFO.
54b226f7 1173
2c395d56 1174Optional arg PARENTS is a list of parent menu names; it specifies
d37ef0f6 1175where to put this language environment in the
2c395d56
RS		 1176Describe Language Environment and Set Language Environment menus.
1177For example, (\"European\") means to put this language environment
1178in the European submenu in each of those two menus."
905a9ed3
DN		 1179 (cond ((symbolp lang-env)
1180 (setq lang-env (symbol-name lang-env)))
1181 ((stringp lang-env)
1182 (setq lang-env (purecopy lang-env))))
54b226f7
KH		 1183 (let ((describe-map describe-language-environment-map)
1184 (setup-map setup-language-environment-map))
1185 (if parents
1186 (let ((l parents)
9deed82f 1187 map parent-symbol parent prompt)
54b226f7
KH		 1188 (while l
1189 (if (symbolp (setq parent-symbol (car l)))
1190 (setq parent (symbol-name parent))
1191 (setq parent parent-symbol parent-symbol (intern parent)))
1192 (setq map (lookup-key describe-map (vector parent-symbol)))
9deed82f
EZ		 1193 ;; This prompt string is for define-prefix-command, so
1194 ;; that the map it creates will be suitable for a menu.
1195 (or map (setq prompt (format "%s Environment" parent)))
54b226f7
KH		 1196 (if (not map)
1197 (progn
1198 (setq map (intern (format "describe-%s-environment-map"
1199 (downcase parent))))
9deed82f 1200 (define-prefix-command map nil prompt)
54b226f7 1201 (define-key-after describe-map (vector parent-symbol)
64657387 1202 (cons parent map))))
54b226f7
KH		 1203 (setq describe-map (symbol-value map))
1204 (setq map (lookup-key setup-map (vector parent-symbol)))
1205 (if (not map)
1206 (progn
1207 (setq map (intern (format "setup-%s-environment-map"
1208 (downcase parent))))
9deed82f 1209 (define-prefix-command map nil prompt)
54b226f7 1210 (define-key-after setup-map (vector parent-symbol)
64657387 1211 (cons parent map))))
54b226f7
KH		 1212 (setq setup-map (symbol-value map))
1213 (setq l (cdr l)))))
f08adf27
RS		 1214
1215 ;; Set up menu items for this language env.
7624ebb9 1216 (let ((doc (assq 'documentation alist)))
f08adf27
RS		 1217 (when doc
1218 (define-key-after describe-map (vector (intern lang-env))
64657387 1219 (cons lang-env 'describe-specified-language-support))))
7624ebb9 1220 (define-key-after setup-map (vector (intern lang-env))
64657387 1221 (cons lang-env 'setup-specified-language-environment))
f08adf27 1222
eec5c8f9
KH		 1223 (dolist (elt alist)
1224 (set-language-info-internal lang-env (car elt) (cdr elt)))
d35deeec 1225
eec5c8f9
KH		 1226 (if (equal lang-env current-language-environment)
1227 (set-language-environment lang-env))))
4ed46869 1228
ae302641 1229(defun read-language-name (key prompt &optional default)
2c395d56 1230 "Read a language environment name which has information for KEY.
ddb5c041 1231If KEY is nil, read any language environment.
2c395d56
RS		 1232Prompt with PROMPT. DEFAULT is the default choice of language environment.
1233This returns a language environment name as a string."
4ed46869
KH		 1234 (let* ((completion-ignore-case t)
1235 (name (completing-read prompt
1236 language-info-alist
ddb5c041 1237 (and key
ca429a25 1238 (function (lambda (elm) (and (listp elm) (assq key elm)))))
ae302641 1239 t nil nil default)))
13e82c04 1240 (if (and (> (length name) 0)
ddb5c041
KH		 1241 (or (not key)
1242 (get-language-info name key)))
13e82c04 1243 name)))
4ed46869
KH		 1244\f
1245;;; Multilingual input methods.
d37ef0f6 1246(defgroup leim nil
d0c40faa
KH		 1247 "LEIM: Libraries of Emacs Input Methods."
1248 :group 'mule)
4ed46869 1249
d0b9c3ab
KH		 1250(defconst leim-list-file-name "leim-list.el"
1251 "Name of LEIM list file.
1252This file contains a list of libraries of Emacs input methods (LEIM)
1253in the format of Lisp expression for registering each input method.
1254Emacs loads this file at startup time.")
1255
905a9ed3 1256(defconst leim-list-header (format
316732d2 1257";;; %s -- list of LEIM (Library of Emacs Input Method) -*-coding: utf-8;-*-
d0b9c3ab 1258;;
bcd76f45
SM		 1259;; This file is automatically generated.
1260;;
d0b9c3ab 1261;; This file contains a list of LEIM (Library of Emacs Input Method)
c654de1d
DL		 1262;; methods in the same directory as this file. Loading this file
1263;; registers all the input methods in Emacs.
d0b9c3ab 1264;;
d33d5fbe 1265;; Each entry has the form:
d0b9c3ab
KH		 1266;; (register-input-method
1267;; INPUT-METHOD LANGUAGE-NAME ACTIVATE-FUNC
1268;; TITLE DESCRIPTION
1269;; ARG ...)
c654de1d 1270;; See the function `register-input-method' for the meanings of the arguments.
d0b9c3ab 1271;;
bcd76f45 1272;; If this directory is included in `load-path', Emacs automatically
d0b9c3ab
KH		 1273;; loads this file at startup time.
1274
1275"
1276 leim-list-file-name)
1277 "Header to be inserted in LEIM list file.")
1278
905a9ed3 1279(defconst leim-list-entry-regexp "^(register-input-method"
d0b9c3ab 1280 "Regexp matching head of each entry in LEIM list file.
caff3c0a 1281See also the variable `leim-list-header'.")
d0b9c3ab
KH		 1282
1283(defvar update-leim-list-functions
1284 '(quail-update-leim-list-file)
1285 "List of functions to call to update LEIM list file.
1286Each function is called with one arg, LEIM directory name.")
1287
a337fe7f
RS		 1288(defun update-leim-list-file (&rest dirs)
1289 "Update LEIM list file in directories DIRS."
d0d8885d
JB		 1290 (dolist (function update-leim-list-functions)
1291 (apply function dirs)))
d0b9c3ab 1292
4ed46869
KH		 1293(defvar current-input-method nil
1294 "The current input method for multilingual text.
96db204a 1295If nil, that means no input method is activated now.")
4ed46869
KH		 1296(make-variable-buffer-local 'current-input-method)
1297(put 'current-input-method 'permanent-local t)
1298
1299(defvar current-input-method-title nil
d0b9c3ab 1300 "Title string of the current input method shown in mode line.")
4ed46869
KH		 1301(make-variable-buffer-local 'current-input-method-title)
1302(put 'current-input-method-title 'permanent-local t)
1303
b4fba33f 1304(defcustom default-input-method nil
caff3c0a 1305 "Default input method for multilingual text (a string).
b4fba33f 1306This is the input method activated automatically by the command
9b10b5a3 1307`toggle-input-method' (\\[toggle-input-method])."
1f547b92 1308 :link '(custom-manual "(emacs)Input Methods")
8861c593 1309 :group 'mule
d398dba6
DL		 1310 :type '(choice (const nil) (string
1311 :completion-ignore-case t
1312 :complete-function widget-string-complete
1313 :completion-alist input-method-alist
1314 :prompt-history input-method-history))
5806e8a6 1315 :set-after '(current-language-environment))
b4fba33f 1316
0f835e87
KH		 1317(put 'input-method-function 'permanent-local t)
1318
723a427a 1319(defvar input-method-history nil
6f5d2452
EZ		 1320 "History list of input methods read from the minibuffer.
1321
1322Maximum length of the history list is determined by the value
1323of `history-length', which see.")
723a427a
KH		 1324(make-variable-buffer-local 'input-method-history)
1325(put 'input-method-history 'permanent-local t)
4ed46869
KH		 1326
1327(defvar inactivate-current-input-method-function nil
1328 "Function to call for inactivating the current input method.
1329Every input method should set this to an appropriate value when activated.
f17ccaee
KH		 1330This function is called with no argument.
1331
1332This function should never change the value of `current-input-method'.
1333It is set to nil by the function `inactivate-input-method'.")
4ed46869
KH		 1334(make-variable-buffer-local 'inactivate-current-input-method-function)
1335(put 'inactivate-current-input-method-function 'permanent-local t)
1336
1337(defvar describe-current-input-method-function nil
1338 "Function to call for describing the current input method.
1339This function is called with no argument.")
1340(make-variable-buffer-local 'describe-current-input-method-function)
1341(put 'describe-current-input-method-function 'permanent-local t)
1342
d0b9c3ab 1343(defvar input-method-alist nil
2c395d56 1344 "Alist of input method names vs how to use them.
d0b9c3ab 1345Each element has the form:
2c395d56
RS		 1346 (INPUT-METHOD LANGUAGE-ENV ACTIVATE-FUNC TITLE DESCRIPTION ARGS...)
1347See the function `register-input-method' for the meanings of the elements.")
ac549fa5 1348;;;###autoload
6dc3311d 1349(put 'input-method-alist 'risky-local-variable t)
2c395d56 1350
f08adf27 1351(defun register-input-method (input-method lang-env &rest args)
bf42aa15 1352 "Register INPUT-METHOD as an input method for language environment LANG-ENV.
d0b9c3ab 1353
d35deeec 1354INPUT-METHOD and LANG-ENV are symbols or strings.
2c395d56
RS		 1355ACTIVATE-FUNC is a function to call to activate this method.
1356TITLE is a string to show in the mode line when this method is active.
1357DESCRIPTION is a string describing this method and what it is good for.
1358The ARGS, if any, are passed as arguments to ACTIVATE-FUNC.
205814ee
KH		 1359All told, the arguments to ACTIVATE-FUNC are INPUT-METHOD and the ARGS.
1360
c654de1d
DL		 1361This function is mainly used in the file \"leim-list.el\" which is
1362created at Emacs build time, registering all Quail input methods
0b6cadff 1363contained in the Emacs distribution.
205814ee 1364
0b6cadff 1365In case you want to register a new Quail input method by yourself, be
205814ee 1366careful to use the same input method title as given in the third
0b6cadff
DL		 1367parameter of `quail-define-package'. (If the values are different, the
1368string specified in this function takes precedence.)
205814ee
KH		 1369
1370The commands `describe-input-method' and `list-input-methods' need
0b6cadff 1371these duplicated values to show some information about input methods
d35deeec
JB		 1372without loading the relevant Quail packages.
1373\n(fn INPUT-METHOD LANG-ENV ACTIVATE-FUNC TITLE DESCRIPTION &rest ARGS)"
f08adf27 1374 (if (symbolp lang-env)
ebfa10d3
DN		 1375 (setq lang-env (symbol-name lang-env))
1376 (setq lang-env (purecopy lang-env)))
4ef06f75 1377 (if (symbolp input-method)
ebfa10d3
DN		 1378 (setq input-method (symbol-name input-method))
1379 (setq input-method (purecopy input-method)))
1380 (setq args (mapcar 'purecopy args))
f08adf27 1381 (let ((info (cons lang-env args))
d0b9c3ab
KH		 1382 (slot (assoc input-method input-method-alist)))
1383 (if slot
1384 (setcdr slot info)
1385 (setq slot (cons input-method info))
1386 (setq input-method-alist (cons slot input-method-alist)))))
1387
4d5ac029 1388(defun read-input-method-name (prompt &optional default inhibit-null)
d0b9c3ab 1389 "Read a name of input method from a minibuffer prompting with PROMPT.
4d5ac029 1390If DEFAULT is non-nil, use that as the default,
0b6cadff 1391and substitute it into PROMPT at the first `%s'.
4ef06f75
KH		 1392If INHIBIT-NULL is non-nil, null input signals an error.
1393
1394The return value is a string."
4d5ac029
RS		 1395 (if default
1396 (setq prompt (format prompt default)))
d0b9c3ab 1397 (let* ((completion-ignore-case t)
c54044ff
KH		 1398 ;; As it is quite normal to change input method in the
1399 ;; minibuffer, we must enable it even if
1400 ;; enable-recursive-minibuffers is currently nil.
1401 (enable-recursive-minibuffers t)
723a427a
KH		 1402 ;; This binding is necessary because input-method-history is
1403 ;; buffer local.
d0b9c3ab 1404 (input-method (completing-read prompt input-method-alist
87505a98
RS		 1405 nil t nil 'input-method-history
1406 default)))
bf294e6e
KH		 1407 (if (and input-method (symbolp input-method))
1408 (setq input-method (symbol-name input-method)))
d0b9c3ab
KH		 1409 (if (> (length input-method) 0)
1410 input-method
1411 (if inhibit-null
43807b77 1412 (error "No valid input method is specified")))))
d0b9c3ab 1413
d0b9c3ab 1414(defun activate-input-method (input-method)
2c395d56
RS		 1415 "Switch to input method INPUT-METHOD for the current buffer.
1416If some other input method is already active, turn it off first.
1417If INPUT-METHOD is nil, deactivate any current input method."
305a3cb6 1418 (if (and input-method (symbolp input-method))
4ef06f75 1419 (setq input-method (symbol-name input-method)))
723a427a
KH		 1420 (if (and current-input-method
1421 (not (string= current-input-method input-method)))
305a3cb6 1422 (inactivate-input-method))
2c395d56 1423 (unless (or current-input-method (null input-method))
d0b9c3ab
KH		 1424 (let ((slot (assoc input-method input-method-alist)))
1425 (if (null slot)
723a427a 1426 (error "Can't activate input method `%s'" input-method))
278dd6ac 1427 (setq current-input-method-title nil)
8efc03e1
KH		 1428 (let ((func (nth 2 slot)))
1429 (if (functionp func)
1430 (apply (nth 2 slot) input-method (nthcdr 5 slot))
1431 (if (and (consp func) (symbolp (car func)) (symbolp (cdr func)))
1432 (progn
1433 (require (cdr func))
1434 (apply (car func) input-method (nthcdr 5 slot)))
1435 (error "Can't activate input method `%s'" input-method))))
d0b9c3ab 1436 (setq current-input-method input-method)
278dd6ac
KH		 1437 (or (stringp current-input-method-title)
1438 (setq current-input-method-title (nth 3 slot)))
28885c0e
KH		 1439 (unwind-protect
1440 (run-hooks 'input-method-activate-hook)
1441 (force-mode-line-update)))))
15b3e511 1442
15b3e511 1443(defun inactivate-input-method ()
f17ccaee 1444 "Turn off the current input method."
723a427a
KH		 1445 (when current-input-method
1446 (if input-method-history
1447 (unless (string= current-input-method (car input-method-history))
1448 (setq input-method-history
1449 (cons current-input-method
1450 (delete current-input-method input-method-history))))
1451 (setq input-method-history (list current-input-method)))
1452 (unwind-protect
36f8618c
KH		 1453 (progn
1454 (setq input-method-function nil
1455 current-input-method-title nil)
1456 (funcall inactivate-current-input-method-function))
15b3e511 1457 (unwind-protect
723a427a 1458 (run-hooks 'input-method-inactivate-hook)
36f8618c 1459 (setq current-input-method nil)
28885c0e 1460 (force-mode-line-update)))))
4ed46869 1461
e893eae2 1462(defun set-input-method (input-method &optional interactive)
2c395d56 1463 "Select and activate input method INPUT-METHOD for the current buffer.
bc406911 1464This also sets the default input method to the one you specify.
402dbbd1
EZ		 1465If INPUT-METHOD is nil, this function turns off the input method, and
1466also causes you to be prompted for a name of an input method the next
1467time you invoke \\[toggle-input-method].
e893eae2
RS		 1468When called interactively, the optional arg INTERACTIVE is non-nil,
1469which marks the variable `default-input-method' as set for Custom buffers.
402dbbd1 1470
bc406911 1471To deactivate the input method interactively, use \\[toggle-input-method].
caff3c0a 1472To deactivate it programmatically, use `inactivate-input-method'."
d0b9c3ab 1473 (interactive
723a427a 1474 (let* ((default (or (car input-method-history) default-input-method)))
42395763 1475 (list (read-input-method-name
87505a98 1476 (if default "Select input method (default %s): " "Select input method: ")
e893eae2
RS		 1477 default t)
1478 t)))
d0b9c3ab 1479 (activate-input-method input-method)
f4990970 1480 (setq default-input-method input-method)
e893eae2 1481 (when interactive
f4990970
PA		 1482 (customize-mark-as-set 'default-input-method))
1483 default-input-method)
4ed46869 1484
023df4cf
RS		 1485(defvar toggle-input-method-active nil
1486 "Non-nil inside `toggle-input-method'.")
1487
e893eae2 1488(defun toggle-input-method (&optional arg interactive)
f8ec20be
RS		 1489 "Enable or disable multilingual text input method for the current buffer.
1490Only one input method can be enabled at any time in a given buffer.
1491
d0d8885d
JB		 1492The normal action is to enable an input method if none was enabled,
1493and disable the current one otherwise. Which input method to enable
1494can be determined in various ways--either the one most recently used,
1495or the one specified by `default-input-method', or as a last resort
1496by reading the name of an input method in the minibuffer.
f8ec20be 1497
d0d8885d 1498With a prefix argument ARG, read an input method name with the minibuffer
f8ec20be 1499and enable that one. The default is the most recent input method specified
e893eae2 1500\(not including the currently active input method, if any).
f8ec20be 1501
d0d8885d 1502When called interactively, the optional argument INTERACTIVE is non-nil,
e893eae2
RS		 1503which marks the variable `default-input-method' as set for Custom buffers."
1504
1505 (interactive "P\np")
023df4cf
RS		 1506 (if toggle-input-method-active
1507 (error "Recursive use of `toggle-input-method'"))
7ddbb5bc
RS		 1508 (if (and current-input-method (not arg))
1509 (inactivate-input-method)
023df4cf
RS		 1510 (let ((toggle-input-method-active t)
1511 (default (or (car input-method-history) default-input-method)))
7ddbb5bc
RS		 1512 (if (and arg default (equal current-input-method default)
1513 (> (length input-method-history) 1))
1514 (setq default (nth 1 input-method-history)))
723a427a
KH		 1515 (activate-input-method
1516 (if (or arg (not default))
7ddbb5bc
RS		 1517 (progn
1518 (read-input-method-name
1519 (if default "Input method (default %s): " "Input method: " )
1520 default t))
723a427a 1521 default))
f4990970 1522 (unless default-input-method
d37ef0f6 1523 (prog1
f4990970 1524 (setq default-input-method current-input-method)
e893eae2 1525 (when interactive
f4990970 1526 (customize-mark-as-set 'default-input-method)))))))
d0b9c3ab 1527
26b3dce6 1528(autoload 'help-buffer "help-mode")
0855c6cd 1529
d0b9c3ab 1530(defun describe-input-method (input-method)
2c395d56 1531 "Describe input method INPUT-METHOD."
d0b9c3ab
KH		 1532 (interactive
1533 (list (read-input-method-name
5b76833f 1534 "Describe input method (default current choice): ")))
78754934 1535 (if (and input-method (symbolp input-method))
4ef06f75 1536 (setq input-method (symbol-name input-method)))
43125c28
RS		 1537 (help-setup-xref (list #'describe-input-method
1538 (or input-method current-input-method))
32226619 1539 (called-interactively-p 'interactive))
f80e2142 1540
d0b9c3ab
KH		 1541 (if (null input-method)
1542 (describe-current-input-method)
464cc130
KH		 1543 (let ((current current-input-method))
1544 (condition-case nil
1545 (progn
1546 (save-excursion
1547 (activate-input-method input-method)
1548 (describe-current-input-method))
1549 (activate-input-method current))
d37ef0f6 1550 (error
464cc130 1551 (activate-input-method current)
5f395df3 1552 (help-setup-xref (list #'describe-input-method input-method)
32226619 1553 (called-interactively-p 'interactive))
5f395df3 1554 (with-output-to-temp-buffer (help-buffer)
464cc130
KH		 1555 (let ((elt (assoc input-method input-method-alist)))
1556 (princ (format
1557 "Input method: %s (`%s' in mode line) for %s\n %s\n"
1558 input-method (nth 3 elt) (nth 1 elt) (nth 4 elt))))))))))
d0b9c3ab
KH		 1559
1560(defun describe-current-input-method ()
f80e2142
RS		 1561 "Describe the input method currently in use.
1562This is a subroutine for `describe-input-method'."
4ed46869
KH		 1563 (if current-input-method
1564 (if (and (symbolp describe-current-input-method-function)
1565 (fboundp describe-current-input-method-function))
1566 (funcall describe-current-input-method-function)
1567 (message "No way to describe the current input method `%s'"
f2979bdb 1568 current-input-method)
4ed46869 1569 (ding))
d0b9c3ab 1570 (error "No input method is activated now")))
4ed46869 1571
d3459641 1572(defun read-multilingual-string (prompt &optional initial-input input-method)
4ed46869
KH		 1573 "Read a multilingual string from minibuffer, prompting with string PROMPT.
1574The input method selected last time is activated in minibuffer.
d0d8885d
JB		 1575If optional second argument INITIAL-INPUT is non-nil, insert it in the
1576minibuffer initially.
1577Optional 3rd argument INPUT-METHOD specifies the input method to be activated
1578instead of the one selected last time. It is a symbol or a string."
88d559ec
KH		 1579 (setq input-method
1580 (or input-method
d3459641 1581 current-input-method
88d559ec
KH		 1582 default-input-method
1583 (read-input-method-name "Input method: " nil t)))
3df60841 1584 (if (and input-method (symbolp input-method))
4ef06f75 1585 (setq input-method (symbol-name input-method)))
305a3cb6
KH		 1586 (let ((prev-input-method current-input-method))
1587 (unwind-protect
1588 (progn
1589 (activate-input-method input-method)
1590 (read-string prompt initial-input nil nil t))
1591 (activate-input-method prev-input-method))))
4ed46869
KH		 1592
1593;; Variables to control behavior of input methods. All input methods
1594;; should react to these variables.
1595
8efc03e1 1596(defcustom input-method-verbose-flag 'default
caff3c0a 1597 "A flag to control extra guidance given by input methods.
8efc03e1 1598The value should be nil, t, `complex-only', or `default'.
4ed46869 1599
cb29dfb6 1600The extra guidance is done by showing list of available keys in echo
8efc03e1
KH		 1601area. When you use the input method in the minibuffer, the guidance
1602is shown at the bottom short window (split from the existing window).
c27c4ed8 1603
8efc03e1
KH		 1604If the value is t, extra guidance is always given, if the value is
1605nil, extra guidance is always suppressed.
1606
1607If the value is `complex-only', only complex input methods such as
1608`chinese-py' and `japanese' give extra guidance.
1609
1610If the value is `default', complex input methods always give extra
1611guidance, but simple input methods give it only when you are not in
1612the minibuffer.
1613
1614See also the variable `input-method-highlight-flag'."
d37ef0f6
DL		 1615 :type '(choice (const :tag "Always" t) (const :tag "Never" nil)
1616 (const complex-only) (const default))
42395763
RS		 1617 :group 'mule)
1618
1619(defcustom input-method-highlight-flag t
caff3c0a 1620 "If this flag is non-nil, input methods highlight partially-entered text.
42395763
RS		 1621For instance, while you are in the middle of a Quail input method sequence,
1622the text inserted so far is temporarily underlined.
8efc03e1
KH		 1623The underlining goes away when you finish or abort the input method sequence.
1624See also the variable `input-method-verbose-flag'."
42395763
RS		 1625 :type 'boolean
1626 :group 'mule)
4ed46869 1627
1f547b92 1628(defcustom input-method-activate-hook nil
f17ccaee
KH		 1629 "Normal hook run just after an input method is activated.
1630
1631The variable `current-input-method' keeps the input method name
1f547b92
DL		 1632just activated."
1633 :type 'hook
1634 :group 'mule)
4ed46869 1635
1f547b92 1636(defcustom input-method-inactivate-hook nil
f17ccaee
KH		 1637 "Normal hook run just after an input method is inactivated.
1638
1639The variable `current-input-method' still keeps the input method name
1f547b92
DL		 1640just inactivated."
1641 :type 'hook
1642 :group 'mule)
4ed46869 1643
1f547b92 1644(defcustom input-method-after-insert-chunk-hook nil
8f924df7 1645 "Normal hook run just after an input method insert some chunk of text."
1f547b92
DL		 1646 :type 'hook
1647 :group 'mule)
4ed46869 1648
dccca980 1649(defvar input-method-exit-on-first-char nil
0b6cadff 1650 "This flag controls when an input method returns.
dccca980
KH		 1651Usually, the input method does not return while there's a possibility
1652that it may find a different translation if a user types another key.
d0d8885d
JB		 1653But, if this flag is non-nil, the input method returns as soon as the
1654current key sequence gets long enough to have some valid translation.")
dccca980 1655
1f547b92 1656(defcustom input-method-use-echo-area nil
dccca980 1657 "This flag controls how an input method shows an intermediate key sequence.
39e643e2
RS		 1658Usually, the input method inserts the intermediate key sequence,
1659or candidate translations corresponding to the sequence,
1660at point in the current buffer.
1f547b92
DL		 1661But, if this flag is non-nil, it displays them in echo area instead."
1662 :type 'hook
1663 :group 'mule)
dccca980 1664
723a427a 1665(defvar input-method-exit-on-invalid-key nil
fea6b736 1666 "This flag controls the behavior of an input method on invalid key input.
723a427a
KH		 1667Usually, when a user types a key which doesn't start any character
1668handled by the input method, the key is handled by turning off the
e8dd0160 1669input method temporarily. After that key, the input method is re-enabled.
723a427a
KH		 1670But, if this flag is non-nil, the input method is never back on.")
1671
4ed46869 1672\f
1f547b92 1673(defcustom set-language-environment-hook nil
8efc03e1
KH		 1674 "Normal hook run after some language environment is set.
1675
1676When you set some hook function here, that effect usually should not
1677be inherited to another language environment. So, you had better set
1678another function in `exit-language-environment-hook' (which see) to
1f547b92
DL		 1679cancel the effect."
1680 :type 'hook
1681 :group 'mule)
8efc03e1 1682
1f547b92 1683(defcustom exit-language-environment-hook nil
8efc03e1
KH		 1684 "Normal hook run after exiting from some language environment.
1685When this hook is run, the variable `current-language-environment'
1686is still bound to the language environment being exited.
1687
e8dd0160 1688This hook is mainly used for canceling the effect of
caff3c0a 1689`set-language-environment-hook' (which see)."
1f547b92
DL		 1690 :type 'hook
1691 :group 'mule)
8efc03e1 1692
b0648a00
RS		 1693(put 'setup-specified-language-environment 'apropos-inhibit t)
1694
15b3e511 1695(defun setup-specified-language-environment ()
f08adf27 1696 "Switch to a specified language environment."
15b3e511 1697 (interactive)
f850d782 1698 (let (language-name)
15b3e511
KH		 1699 (if (and (symbolp last-command-event)
1700 (or (not (eq last-command-event 'Default))
1701 (setq last-command-event 'English))
f850d782 1702 (setq language-name (symbol-name last-command-event)))
f4990970
PA		 1703 (prog1
1704 (set-language-environment language-name)
1705 (customize-mark-as-set 'current-language-environment))
15b3e511 1706 (error "Bogus calling sequence"))))
4ed46869 1707
8861c593 1708(defcustom current-language-environment "English"
94d04df6 1709 "The last language environment specified with `set-language-environment'.
ebef6d93
KH		 1710This variable should be set only with \\[customize], which is equivalent
1711to using the function `set-language-environment'."
94d04df6 1712 :link '(custom-manual "(emacs)Language Environments")
dff1aa24 1713 :set (lambda (symbol value) (set-language-environment value))
94d04df6 1714 :get (lambda (x)
f15466c5 1715 (or (car-safe (assoc-string
94d04df6
DL		 1716 (if (symbolp current-language-environment)
1717 (symbol-name current-language-environment)
1718 current-language-environment)
f15466c5 1719 language-info-alist t))
94d04df6 1720 "English"))
990a4108
MR		 1721 ;; custom type will be updated with `set-language-info'.
1722 :type (if language-info-alist
1723 (cons 'choice (mapcar
1724 (lambda (lang)
756e055f
MR		 1725 (list 'const lang))
1726 (sort (mapcar 'car language-info-alist) 'string<)))
990a4108 1727 'string)
8861c593 1728 :initialize 'custom-initialize-default
eb9fc9e6 1729 :group 'mule)
f850d782 1730
ddb5c041
KH		 1731(defun reset-language-environment ()
1732 "Reset multilingual environment of Emacs to the default status.
1733
1734The default status is as follows:
1735
d37ef0f6 1736 The default value of `buffer-file-coding-system' is nil.
ddb5c041
KH		 1737 The default coding system for process I/O is nil.
1738 The default value for the command `set-terminal-coding-system' is nil.
1739 The default value for the command `set-keyboard-coding-system' is nil.
1740
97941b05
KH		 1741 The order of priorities of coding systems are as follows:
1742 utf-8
1743 iso-2022-7bit
1744 iso-latin-1
1745 iso-2022-7bit-lock
1746 iso-2022-8bit-ss2
1747 emacs-mule
1748 raw-text"
ddb5c041
KH		 1749 (interactive)
1750 ;; This function formerly set default-enable-multibyte-characters to t,
1751 ;; but that is incorrect. It should not alter the unibyte/multibyte choice.
1752
97941b05
KH		 1753 (set-coding-system-priority
1754 'utf-8
1755 'iso-2022-7bit
1756 'iso-latin-1
1757 'iso-2022-7bit-lock
1758 'iso-2022-8bit-ss2
1759 'emacs-mule
1760 'raw-text)
91693d18 1761
ddb5c041 1762 (set-default-coding-systems nil)
b5edd1d1 1763 (setq default-sendmail-coding-system 'iso-latin-1)
a41118cc
SM		 1764 ;; On Darwin systems, this should be utf-8, but when this file is loaded
1765 ;; utf-8 is not yet defined, so we set it in set-locale-environment instead.
787caf99 1766 (setq default-file-name-coding-system 'iso-latin-1)
1d77e15a
JR		 1767 ;; Preserve eol-type from existing default-process-coding-systems.
1768 ;; On non-unix-like systems in particular, these may have been set
1769 ;; carefully by the user, or by the startup code, to deal with the
1770 ;; users shell appropriately, so should not be altered by changing
1771 ;; language environment.
1772 (let ((output-coding
a099a2ff
JR		 1773 ;; When bootstrapping, coding-systems are not defined yet, so
1774 ;; we need to catch the error from check-coding-system.
d37ef0f6 1775 (condition-case nil
a099a2ff
JR		 1776 (coding-system-change-text-conversion
1777 (car default-process-coding-system) 'undecided)
1778 (coding-system-error 'undecided)))
1d77e15a 1779 (input-coding
a099a2ff
JR		 1780 (condition-case nil
1781 (coding-system-change-text-conversion
1782 (cdr default-process-coding-system) 'iso-latin-1)
1783 (coding-system-error 'iso-latin-1))))
1d77e15a
JR		 1784 (setq default-process-coding-system
1785 (cons output-coding input-coding)))
b5edd1d1 1786
a43977db
KH		 1787 ;; Put the highest priority to the charset iso-8859-1 to prefer the
1788 ;; registry iso8859-1 over iso8859-2 in font selection. It also
1789 ;; makes unibyte-display-via-language-environment to use iso-8859-1
1790 ;; as the unibyte charset.
1791 (set-charset-priority 'iso-8859-1)
1792
ddb5c041
KH		 1793 ;; Don't alter the terminal and keyboard coding systems here.
1794 ;; The terminal still supports the same coding system
1795 ;; that it supported a minute ago.
1b8dc791
SM		 1796 ;; (set-terminal-coding-system-internal nil)
1797 ;; (set-keyboard-coding-system-internal nil)
ddb5c041 1798
33de15f4
SM		 1799 ;; Back in Emacs-20, it was necessary to provide some fallback implicit
1800 ;; conversion, because almost no packages handled coding-system issues.
1801 ;; Nowadays it'd just paper over bugs.
1802 ;; (set-unibyte-charset 'iso-8859-1)
1803 )
ddb5c041 1804
0c47a7c8
KH		 1805(reset-language-environment)
1806
97c57fb2 1807(defun set-display-table-and-terminal-coding-system (language-name &optional coding-system display)
40c81f74
PE		 1808 "Set up the display table and terminal coding system for LANGUAGE-NAME."
1809 (let ((coding (get-language-info language-name 'unibyte-display)))
735b7c87
KH		 1810 (if (and coding
1811 (or (not coding-system)
1812 (coding-system-equal coding coding-system)))
40c81f74 1813 (standard-display-european-internal)
a67ae60e
EZ		 1814 ;; The following 2 lines undo the 8-bit display that we set up
1815 ;; in standard-display-european-internal, which see. This is in
1816 ;; case the user has used standard-display-european earlier in
970c9391 1817 ;; this session.
d73a7bb8
JPW		 1818 (when standard-display-table
1819 (dotimes (i 128)
b221615b 1820 (aset standard-display-table (+ i 128) nil))))
970c9391 1821 (set-terminal-coding-system (or coding-system coding) display)))
40c81f74 1822
166246f7 1823(defun set-language-environment (language-name)
6c05d680
RS		 1824 "Set up multi-lingual environment for using LANGUAGE-NAME.
1825This sets the coding system priority and the default input method
8861c593
RS		 1826and sometimes other things. LANGUAGE-NAME should be a string
1827which is the name of a language environment. For example, \"Latin-1\"
1828specifies the character set for the major languages of Western Europe."
8efc03e1 1829 (interactive (list (read-language-name
ddb5c041 1830 nil
5b76833f 1831 "Set language environment (default English): ")))
4ef06f75
KH		 1832 (if language-name
1833 (if (symbolp language-name)
1834 (setq language-name (symbol-name language-name)))
1835 (setq language-name "English"))
f15466c5 1836 (let ((slot (assoc-string language-name language-info-alist t)))
95498fd0 1837 (unless slot
f850d782 1838 (error "Language environment not defined: %S" language-name))
95498fd0 1839 (setq language-name (car slot)))
8efc03e1
KH		 1840 (if current-language-environment
1841 (let ((func (get-language-info current-language-environment
1842 'exit-function)))
e63645c2 1843 (run-hooks 'exit-language-environment-hook)
5f395df3 1844 (if (functionp func) (funcall func))))
ddb5c041 1845
e9a8ed3c
KH		 1846 (reset-language-environment)
1847 ;; The features might set up coding systems.
ddb5c041
KH		 1848 (let ((required-features (get-language-info language-name 'features)))
1849 (while required-features
1850 (require (car required-features))
1851 (setq required-features (cdr required-features))))
6b61353c 1852
e9a8ed3c
KH		 1853 (setq current-language-environment language-name)
1854
1855 (set-language-environment-coding-systems language-name)
1856 (set-language-environment-input-method language-name)
1857 (set-language-environment-nonascii-translation language-name)
1858 (set-language-environment-charset language-name)
1859 ;; Unibyte setups if necessary.
597e2240 1860 (unless (default-value 'enable-multibyte-characters)
e9a8ed3c
KH		 1861 (set-language-environment-unibyte language-name))
1862
ddb5c041 1863 (let ((func (get-language-info language-name 'setup-function)))
5f395df3 1864 (if (functionp func)
ddb5c041 1865 (funcall func)))
e9a8ed3c 1866
8aeebac2 1867 (setq current-iso639-language
e9aaa1db
KH		 1868 (or (get-language-info language-name 'iso639-language)
1869 current-iso639-language))
8aeebac2 1870
8efc03e1 1871 (run-hooks 'set-language-environment-hook)
f850d782 1872 (force-mode-line-update t))
4ed46869 1873
c3869589 1874(define-widget 'charset 'symbol
aaa448c9 1875 "An Emacs charset."
1644d5b9 1876 :tag "Charset"
c3869589
DL		 1877 :complete-function (lambda ()
1878 (interactive)
1879 (lisp-complete-symbol 'charsetp))
1880 :completion-ignore-case t
1881 :value 'ascii
1882 :validate (lambda (widget)
1883 (unless (charsetp (widget-value widget))
1884 (widget-put widget :error (format "Invalid charset: %S"
1885 (widget-value widget)))
1886 widget))
1887 :prompt-history 'charset-history)
1888
1889(defcustom language-info-custom-alist nil
1890 "Customizations of language environment parameters.
1891Value is an alist with elements like those of `language-info-alist'.
1892These are used to set values in `language-info-alist' which replace
1893the defaults. A typical use is replacing the default input method for
1894the environment. Use \\[describe-language-environment] to find the environment's settings.
1895
1896This option is intended for use at startup. Removing items doesn't
1897remove them from the language info until you next restart Emacs.
1898
d0d8885d
JB		 1899Setting this variable directly does not take effect.
1900See `set-language-info-alist' for use in programs."
c3869589 1901 :group 'mule
8589dc17 1902 :version "23.1"
c3869589
DL		 1903 :set (lambda (s v)
1904 (custom-set-default s v)
1905 ;; Can't do this before language environments are set up.
1906 (when v
1907 ;; modify language-info-alist
1908 (dolist (elt v)
1909 (set-language-info-alist (car elt) (cdr elt)))
1910 ;; re-set the environment in case its parameters changed
1911 (set-language-environment current-language-environment)))
1912 :type `(alist
1913 :key-type (string :tag "Language environment"
1914 :completion-ignore-case t
1915 :complete-function widget-string-complete
1916 :completion-alist language-info-alist)
1917 :value-type
1918 (alist :key-type symbol
1919 :options ((documentation string)
1920 (charset (repeat charset))
1921 (sample-text string)
1922 (setup-function function)
1923 (exit-function function)
1924 (coding-system (repeat coding-system))
1925 (coding-priority (repeat coding-system))
1926 (nonascii-translation charset)
1927 (input-method
1928 (string
1929 :completion-ignore-case t
1930 :complete-function widget-string-complete
1931 :completion-alist input-method-alist
1932 :prompt-history input-method-history))
1933 (features (repeat symbol))
1934 (unibyte-display coding-system)))))
1935
aa360da1
GM		 1936(declare-function x-server-vendor "xfns.c" (&optional terminal))
1937(declare-function x-server-version "xfns.c" (&optional terminal))
1938
51a8fc1d
RS		 1939(defun standard-display-european-internal ()
1940 ;; Actually set up direct output of non-ASCII characters.
03c35c83
EZ		 1941 (standard-display-8bit (if (eq window-system 'pc) 128 160) 255)
1942 ;; Unibyte Emacs on MS-DOS wants to display all 8-bit characters with
1943 ;; the native font, and codes 160 and 146 stand for something very
1944 ;; different there.
597e2240
GM		 1945 (or (and (eq window-system 'pc) (not (default-value
1946 'enable-multibyte-characters)))
03c35c83 1947 (progn
6b626913
SM		 1948 ;; Most X fonts used to do the wrong thing for latin-1 code 160.
1949 (unless (and (eq window-system 'x)
1950 ;; XFree86 4 has fixed the fonts.
1951 (string= "The XFree86 Project, Inc" (x-server-vendor))
1952 (> (aref (number-to-string (nth 2 (x-server-version))) 0)
1953 ?3))
1954 ;; Make non-line-break space display as a plain space.
1955 (aset standard-display-table 160 [32]))
1492f7ac 1956 ;; Most Windows programs send out apostrophes as \222. Most X fonts
03c35c83 1957 ;; don't contain a character at that position. Map it to the ASCII
5f395df3
SM		 1958 ;; apostrophe. [This is actually RIGHT SINGLE QUOTATION MARK,
1959 ;; U+2019, normally from the windows-1252 character set. XFree 4
1960 ;; fonts probably have the appropriate glyph at this position,
1961 ;; so they could use standard-display-8bit. It's better to use a
1962 ;; proper windows-1252 coding system. --fx]
6b626913 1963 (aset standard-display-table 146 [39]))))
03c35c83 1964
e9a8ed3c
KH		 1965(defun set-language-environment-coding-systems (language-name)
1966 "Do various coding system setups for language environment LANGUAGE-NAME."
54b226f7 1967 (let* ((priority (get-language-info language-name 'coding-priority))
e9a8ed3c 1968 (default-coding (car priority))
b56a5ae0 1969 ;; If the default buffer-file-coding-system is nil, don't use
207422da 1970 ;; coding-system-eol-type, because it treats nil as
b56a5ae0 1971 ;; `no-conversion'. The default buffer-file-coding-system is set
207422da
EZ		 1972 ;; to nil by reset-language-environment, and in that case we
1973 ;; want to have here the native EOL type for each platform.
1974 ;; FIXME: there should be a common code that runs both on
1975 ;; startup and here to set the default EOL type correctly.
1976 ;; Right now, DOS/Windows platforms set this on dos-w32.el,
1977 ;; which works only as long as the order of loading files at
1978 ;; dump time and calling functions at startup is not modified
1979 ;; significantly, i.e. as long as this function is called
b56a5ae0 1980 ;; _after_ the default buffer-file-coding-system was set by
207422da
EZ		 1981 ;; dos-w32.el.
1982 (eol-type
b56a5ae0
SM		 1983 (coding-system-eol-type
1984 (or (default-value 'buffer-file-coding-system)
1985 (if (memq system-type '(windows-nt ms-dos)) 'dos 'unix)))))
97941b05
KH		 1986 (when priority
1987 (set-default-coding-systems
1988 (if (memq eol-type '(0 1 2 unix dos mac))
1989 (coding-system-change-eol-conversion default-coding eol-type)
1990 default-coding))
1991 (setq default-sendmail-coding-system default-coding)
1992 (apply 'set-coding-system-priority priority))))
54b226f7 1993
d042f8b4
KH		 1994(defun set-language-environment-input-method (language-name)
1995 "Do various input method setups for language environment LANGUAGE-NAME."
1996 (let ((input-method (get-language-info language-name 'input-method)))
1997 (when input-method
1998 (setq default-input-method input-method)
1999 (if input-method-history
2000 (setq input-method-history
2001 (cons input-method
2002 (delete input-method input-method-history)))))))
2003
2004(defun set-language-environment-nonascii-translation (language-name)
2005 "Do unibyte/multibyte translation setup for language environment LANGUAGE-NAME."
e9a8ed3c
KH		 2006 ;; Note: For DOS, we assumed that the charset cpXXX is already
2007 ;; defined.
2008 (let ((nonascii (get-language-info language-name 'nonascii-translation)))
2009 (if (eq window-system 'pc)
970c9391 2010 (setq nonascii (intern (format "cp%d" dos-codepage))))
e9a8ed3c
KH		 2011 (or (and (charsetp nonascii)
2012 (get-charset-property nonascii :ascii-compatible-p))
2013 (setq nonascii 'iso-8859-1))
33de15f4
SM		 2014 ;; Back in Emacs-20, it was necessary to provide some fallback implicit
2015 ;; conversion, because almost no packages handled coding-system issues.
2016 ;; Nowadays it'd just paper over bugs.
2017 ;; (set-unibyte-charset nonascii)
2018 ))
d042f8b4
KH		 2019
2020(defun set-language-environment-charset (language-name)
2021 "Do various charset setups for language environment LANGUAGE-NAME."
e9a8ed3c
KH		 2022 ;; Put higher priorities to such charsets that are supported by the
2023 ;; coding systems of higher priorities in this environment.
2024 (let ((charsets (get-language-info language-name 'charset)))
2025 (dolist (coding (get-language-info language-name 'coding-priority))
bf974dc9
KH		 2026 (let ((list (coding-system-charset-list coding)))
2027 (if (consp list)
2028 (setq charsets (append charsets list)))))
e9a8ed3c
KH		 2029 (if charsets
2030 (apply 'set-charset-priority charsets))))
d042f8b4
KH		 2031
2032(defun set-language-environment-unibyte (language-name)
2033 "Do various unibyte-mode setups for language environment LANGUAGE-NAME."
d042f8b4
KH		 2034 (set-display-table-and-terminal-coding-system language-name))
2035
4ed46869 2036(defsubst princ-list (&rest args)
caff3c0a 2037 "Print all arguments with `princ', then print \"\\n\"."
4ed46869
KH		 2038 (while args (princ (car args)) (setq args (cdr args)))
2039 (princ "\n"))
2040
b0648a00
RS		 2041(put 'describe-specified-language-support 'apropos-inhibit t)
2042
6b61353c 2043;; Print language-specific information such as input methods,
13e82c04 2044;; charsets, and coding systems. This function is intended to be
48082651 2045;; called from the menu:
281d03ec 2046;; [menu-bar mule describe-language-environment LANGUAGE]
48082651
KH		 2047;; and should not run it by `M-x describe-current-input-method-function'.
2048(defun describe-specified-language-support ()
96db204a 2049 "Describe how Emacs supports the specified language environment."
48082651 2050 (interactive)
281d03ec 2051 (let (language-name)
48082651 2052 (if (not (and (symbolp last-command-event)
cda74479
DL		 2053 (or (not (eq last-command-event 'Default))
2054 (setq last-command-event 'English))
281d03ec 2055 (setq language-name (symbol-name last-command-event))))
48082651 2056 (error "Bogus calling sequence"))
281d03ec
RS		 2057 (describe-language-environment language-name)))
2058
2059(defun describe-language-environment (language-name)
2060 "Describe how Emacs supports language environment LANGUAGE-NAME."
78754934
KH		 2061 (interactive
2062 (list (read-language-name
2063 'documentation
5b76833f 2064 "Describe language environment (default current choice): ")))
f850d782
RS		 2065 (if (null language-name)
2066 (setq language-name current-language-environment))
281d03ec
RS		 2067 (if (or (null language-name)
2068 (null (get-language-info language-name 'documentation)))
2069 (error "No documentation for the specified language"))
4ef06f75
KH		 2070 (if (symbolp language-name)
2071 (setq language-name (symbol-name language-name)))
ef5a4730
KH		 2072 (dolist (feature (get-language-info language-name 'features))
2073 (require feature))
6b626913 2074 (let ((doc (get-language-info language-name 'documentation)))
c3034e84 2075 (help-setup-xref (list #'describe-language-environment language-name)
32226619 2076 (called-interactively-p 'interactive))
c3034e84 2077 (with-output-to-temp-buffer (help-buffer)
9a529312 2078 (with-current-buffer standard-output
464cc130
KH		 2079 (insert language-name " language environment\n\n")
2080 (if (stringp doc)
2081 (insert doc "\n\n"))
e036b0a6
KH		 2082 (condition-case nil
2083 (let ((str (eval (get-language-info language-name 'sample-text))))
2084 (if (stringp str)
fd0dd4c3
KH		 2085 (insert "Sample text:\n "
2086 (replace-regexp-in-string "\n" "\n " str)
2087 "\n\n")))
e036b0a6 2088 (error nil))
464cc130 2089 (let ((input-method (get-language-info language-name 'input-method))
acd1b9bd
KH		 2090 (l (copy-sequence input-method-alist))
2091 (first t))
2092 (when (and input-method
2093 (setq input-method (assoc input-method l)))
2094 (insert "Input methods (default " (car input-method) ")\n")
2095 (setq l (cons input-method (delete input-method l))
2096 first nil))
2097 (dolist (elt l)
2098 (when (or (eq input-method elt)
2099 (eq t (compare-strings language-name nil nil
2100 (nth 1 elt) nil nil t)))
2101 (when first
2102 (insert "Input methods:\n")
2103 (setq first nil))
2104 (insert " " (car elt))
2105 (search-backward (car elt))
2106 (help-xref-button 0 'help-input-method (car elt))
464cc130 2107 (goto-char (point-max))
2fa7e202 2108 (insert " (\""
acd1b9bd
KH		 2109 (if (stringp (nth 3 elt)) (nth 3 elt) (car (nth 3 elt)))
2110 "\" in mode line)\n")))
2111 (or first
2112 (insert "\n")))
464cc130
KH		 2113 (insert "Character sets:\n")
2114 (let ((l (get-language-info language-name 'charset)))
2115 (if (null l)
2116 (insert " nothing specific to " language-name "\n")
2117 (while l
2118 (insert " " (symbol-name (car l)))
2119 (search-backward (symbol-name (car l)))
467412aa 2120 (help-xref-button 0 'help-character-set (car l))
464cc130
KH		 2121 (goto-char (point-max))
2122 (insert ": " (charset-description (car l)) "\n")
2123 (setq l (cdr l)))))
2124 (insert "\n")
2125 (insert "Coding systems:\n")
2126 (let ((l (get-language-info language-name 'coding-system)))
2127 (if (null l)
2128 (insert " nothing specific to " language-name "\n")
2129 (while l
2130 (insert " " (symbol-name (car l)))
2131 (search-backward (symbol-name (car l)))
467412aa 2132 (help-xref-button 0 'help-coding-system (car l))
464cc130
KH		 2133 (goto-char (point-max))
2134 (insert " (`"
2135 (coding-system-mnemonic (car l))
2136 "' in mode line):\n\t"
2137 (coding-system-doc-string (car l))
2138 "\n")
0855c6cd 2139 (let ((aliases (coding-system-aliases (car l))))
464cc130
KH		 2140 (when aliases
2141 (insert "\t(alias:")
2142 (while aliases
2143 (insert " " (symbol-name (car aliases)))
2144 (setq aliases (cdr aliases)))
2145 (insert ")\n")))
c3034e84 2146 (setq l (cdr l)))))))))
4ed46869 2147\f
40c81f74
PE		 2148;;; Locales.
2149
0d7c5bb9
DL		 2150(defvar locale-translation-file-name nil
2151 "File name for the system's file of locale-name aliases, or nil if none.")
40c81f74 2152
5f395df3
SM		 2153;; The following definitions might as well be marked as constants and
2154;; purecopied, since they're normally used on startup, and probably
2155;; should reflect the facilities of the base Emacs.
2156(defconst locale-language-names
2157 (purecopy
2158 '(
40c81f74
PE		 2159 ;; Locale names of the form LANGUAGE[_TERRITORY][.CODESET][@MODIFIER]
2160 ;; as specified in the Single Unix Spec, Version 2.
2161 ;; LANGUAGE is a language code taken from ISO 639:1988 (E/F)
2162 ;; with additions from ISO 639/RA Newsletter No.1/1989;
5f395df3
SM		 2163 ;; see Internet RFC 2165 (1997-06) and
2164 ;; http://www.evertype.com/standards/iso639/iso639-en.html
2165 ;; TERRITORY is a country code taken from ISO 3166
2166 ;; http://www.din.de/gremien/nas/nabd/iso3166ma/codlstp1/en_listp1.html.
40c81f74 2167 ;; CODESET and MODIFIER are implementation-dependent.
5f395df3 2168
6b61353c
KH		 2169 ;; jasonr comments: MS Windows uses three letter codes for
2170 ;; languages instead of the two letter ISO codes that POSIX
2171 ;; uses. In most cases the first two letters are the same, so
2172 ;; most of the regexps in locale-language-names work. Japanese
2173 ;; and Chinese are exceptions, which are listed in the
2174 ;; non-standard section at the bottom of locale-language-names.
2175
8dedddd5
KH		 2176 ("aa_DJ" . "Latin-1") ; Afar
2177 ("aa" . "UTF-8")
2178 ;; ab Abkhazian
6ececc4d 2179 ("af" . "Latin-1") ; Afrikaans
8dedddd5
KH		 2180 ("am" "Ethiopic" utf-8) ; Amharic
2181 ("an" . "Latin-9") ; Aragonese
5f395df3 2182 ; ar Arabic glibc uses 8859-6
40c81f74
PE		 2183 ; as Assamese
2184 ; ay Aymara
8dedddd5 2185 ("az" . "UTF-8") ; Azerbaijani
40c81f74 2186 ; ba Bashkir
8dedddd5
KH		 2187 ("be" "Belarusian" cp1251) ; Belarusian [Byelorussian until early 1990s]
2188 ("bg" "Bulgarian" cp1251) ; Bulgarian
40c81f74
PE		 2189 ; bh Bihari
2190 ; bi Bislama
8dedddd5 2191 ("bn" . "UTF-8") ; Bengali, Bangla
40c81f74
PE		 2192 ("bo" . "Tibetan")
2193 ("br" . "Latin-1") ; Breton
d37ef0f6 2194 ("bs" . "Latin-2") ; Bosnian
8dedddd5 2195 ("byn" . "UTF-8") ; Bilin; Blin
40c81f74
PE		 2196 ("ca" . "Latin-1") ; Catalan
2197 ; co Corsican
8dedddd5
KH		 2198 ("cs" "Czech" iso-8859-2)
2199 ("cy" "Welsh" iso-8859-14)
40c81f74 2200 ("da" . "Latin-1") ; Danish
8dedddd5 2201 ("de" "German" iso-8859-1)
569a6374 2202 ; dv Divehi
40c81f74 2203 ; dz Bhutani
8dedddd5 2204 ("el" "Greek" iso-8859-7)
6ececc4d 2205 ;; Users who specify "en" explicitly typically want Latin-1, not ASCII.
e11cf111
DL		 2206 ;; That's actually what the GNU locales define, modulo things like
2207 ;; en_IN -- fx.
8dedddd5 2208 ("en_IN" "English" utf-8) ; glibc uses utf-8 for English in India
eec5c8f9 2209 ("en" "English" iso-8859-1) ; English
58802cd7 2210 ("eo" . "Esperanto") ; Esperanto
8dedddd5
KH		 2211 ("es" "Spanish" iso-8859-1)
2212 ("et" . "Latin-1") ; Estonian
40c81f74 2213 ("eu" . "Latin-1") ; Basque
8dedddd5 2214 ("fa" . "UTF-8") ; Persian
40c81f74 2215 ("fi" . "Latin-1") ; Finnish
8dedddd5 2216 ("fj" . "Latin-1") ; Fiji
40c81f74 2217 ("fo" . "Latin-1") ; Faroese
8dedddd5 2218 ("fr" "French" iso-8859-1) ; French
40c81f74 2219 ("fy" . "Latin-1") ; Frisian
6ececc4d 2220 ("ga" . "Latin-1") ; Irish Gaelic (new orthography)
8dedddd5
KH		 2221 ("gd" . "Latin-9") ; Scots Gaelic
2222 ("gez" "Ethiopic" utf-8) ; Geez
2223 ("gl" . "Latin-1") ; Gallegan; Galician
40c81f74 2224 ; gn Guarani
8dedddd5
KH		 2225 ("gu" . "UTF-8") ; Gujarati
2226 ("gv" . "Latin-1") ; Manx Gaelic
40c81f74 2227 ; ha Hausa
8dedddd5
KH		 2228 ("he" "Hebrew" iso-8859-8)
2229 ("hi" "Devanagari" utf-8) ; Hindi
2230 ("hr" "Croatian" iso-8859-2) ; Croatian
40c81f74
PE		 2231 ("hu" . "Latin-2") ; Hungarian
2232 ; hy Armenian
2233 ; ia Interlingua
2234 ("id" . "Latin-1") ; Indonesian
2235 ; ie Interlingue
2236 ; ik Inupiak
2237 ("is" . "Latin-1") ; Icelandic
8dedddd5 2238 ("it" "Italian" iso-8859-1) ; Italian
40c81f74 2239 ; iu Inuktitut
8dedddd5
KH		 2240 ("iw" "Hebrew" iso-8859-8)
2241 ("ja" "Japanese" euc-jp)
40c81f74 2242 ; jw Javanese
8dedddd5 2243 ("ka" "Georgian" georgian-ps) ; Georgian
40c81f74 2244 ; kk Kazakh
6ececc4d 2245 ("kl" . "Latin-1") ; Greenlandic
40c81f74 2246 ; km Cambodian
8dedddd5
KH		 2247 ("kn" "Kannada" utf-8)
2248 ("ko" "Korean" euc-kr)
40c81f74
PE		 2249 ; ks Kashmiri
2250 ; ku Kurdish
5f395df3 2251 ("kw" . "Latin-1") ; Cornish
40c81f74
PE		 2252 ; ky Kirghiz
2253 ("la" . "Latin-1") ; Latin
5f395df3 2254 ("lb" . "Latin-1") ; Luxemburgish
8dedddd5 2255 ("lg" . "Laint-6") ; Ganda
40c81f74 2256 ; ln Lingala
8dedddd5
KH		 2257 ("lo" "Lao" utf-8) ; Laothian
2258 ("lt" "Lithuanian" iso-8859-13)
9c20a8d5 2259 ("lv" . "Latvian") ; Latvian, Lettish
40c81f74 2260 ; mg Malagasy
5f395df3 2261 ("mi" . "Latin-7") ; Maori
8dedddd5
KH		 2262 ("mk" "Cyrillic-ISO" iso-8859-5) ; Macedonian
2263 ("ml" "Malayalam" utf-8)
2264 ("mn" . "UTF-8") ; Mongolian
40c81f74 2265 ; mo Moldavian
8dedddd5 2266 ("mr" "Devanagari" utf-8) ; Marathi
5f395df3 2267 ("ms" . "Latin-1") ; Malay
40c81f74
PE		 2268 ("mt" . "Latin-3") ; Maltese
2269 ; my Burmese
2270 ; na Nauru
8dedddd5
KH		 2271 ("nb" . "Latin-1") ; Norwegian
2272 ("ne" "Devanagari" utf-8) ; Nepali
2273 ("nl" "Dutch" iso-8859-1)
40c81f74 2274 ("no" . "Latin-1") ; Norwegian
5f395df3 2275 ("oc" . "Latin-1") ; Occitan
8dedddd5
KH		 2276 ("om_ET" . "UTF-8") ; (Afan) Oromo
2277 ("om" . "Latin-1") ; (Afan) Oromo
40c81f74 2278 ; or Oriya
8dedddd5 2279 ("pa" . "UTF-8") ; Punjabi
40c81f74
PE		 2280 ("pl" . "Latin-2") ; Polish
2281 ; ps Pashto, Pushto
2282 ("pt" . "Latin-1") ; Portuguese
2283 ; qu Quechua
6ececc4d 2284 ("rm" . "Latin-1") ; Rhaeto-Romanic
40c81f74 2285 ; rn Kirundi
8dedddd5
KH		 2286 ("ro" "Romanian" iso-8859-2)
2287 ("ru_RU" "Russian" iso-8859-5)
2288 ("ru_UA" "Russian" koi8-u)
40c81f74
PE		 2289 ; rw Kinyarwanda
2290 ("sa" . "Devanagari") ; Sanskrit
2291 ; sd Sindhi
8dedddd5 2292 ("se" . "UTF-8") ; Northern Sami
40c81f74
PE		 2293 ; sg Sangho
2294 ("sh" . "Latin-2") ; Serbo-Croatian
2295 ; si Sinhalese
8dedddd5
KH		 2296 ("sid" . "UTF-8") ; Sidamo
2297 ("sk" "Slovak" iso-8859-2)
2298 ("sl" "Slovenian" iso-8859-2)
40c81f74
PE		 2299 ; sm Samoan
2300 ; sn Shona
8dedddd5
KH		 2301 ("so_ET" "UTF-8") ; Somali
2302 ("so" "Latin-1") ; Somali
6ececc4d 2303 ("sq" . "Latin-1") ; Albanian
40c81f74
PE		 2304 ("sr" . "Latin-2") ; Serbian (Latin alphabet)
2305 ; ss Siswati
8dedddd5 2306 ("st" . "Latin-1") ; Sesotho
40c81f74 2307 ; su Sundanese
8dedddd5 2308 ("sv" "Swedish" iso-8859-1) ; Swedish
40c81f74 2309 ("sw" . "Latin-1") ; Swahili
8dedddd5
KH		 2310 ("ta" "Tamil" utf-8)
2311 ("te" . "UTF-8") ; Telugu
2312 ("tg" "Tajik" koi8-t)
2313 ("th" "Thai" tis-620)
2314 ("ti" "Ethiopic" utf-8) ; Tigrinya
2315 ("tig_ER" . "UTF-8") ; Tigre
40c81f74 2316 ; tk Turkmen
6ececc4d 2317 ("tl" . "Latin-1") ; Tagalog
40c81f74
PE		 2318 ; tn Setswana
2319 ; to Tonga
8dedddd5 2320 ("tr" "Turkish" iso-8859-9)
40c81f74 2321 ; ts Tsonga
8dedddd5 2322 ("tt" . "UTF-8") ; Tatar
40c81f74
PE		 2323 ; tw Twi
2324 ; ug Uighur
8dedddd5
KH		 2325 ("uk" "Ukrainian" koi8-u)
2326 ("ur" . "UTF-8") ; Urdu
2327 ("uz_UZ@cyrillic" . "UTF-8"); Uzbek
5f395df3 2328 ("uz" . "Latin-1") ; Uzbek
8dedddd5 2329 ("vi" "Vietnamese" utf-8)
40c81f74 2330 ; vo Volapuk
d37ef0f6 2331 ("wa" . "Latin-1") ; Walloon
40c81f74 2332 ; wo Wolof
8dedddd5 2333 ("xh" . "Latin-1") ; Xhosa
9c20a8d5 2334 ("yi" . "Windows-1255") ; Yiddish
40c81f74
PE		 2335 ; yo Yoruba
2336 ; za Zhuang
8dedddd5 2337 ("zh_HK" . "Chinese-Big5")
05b81b42 2338 ; zh_HK/BIG5-HKSCS \
8dedddd5 2339 ("zh_TW" . "Chinese-Big5")
05b81b42
KH		 2340 ("zh_CN.GB2312" "Chinese-GB")
2341 ("zh_CN.GBK" "Chinese-GBK")
2342 ("zh_CN.GB18030" "Chinese-GB18030")
f21605fa 2343 ("zh_CN.UTF-8" . "Chinese-GBK")
8dedddd5 2344 ("zh_CN" . "Chinese-GB")
40c81f74 2345 ("zh" . "Chinese-GB")
8dedddd5 2346 ("zu" . "Latin-1") ; Zulu
40c81f74
PE		 2347
2348 ;; ISO standard locales
2349 ("c$" . "ASCII")
2350 ("posix$" . "ASCII")
2351
40c81f74
PE		 2352 ;; The "IPA" Emacs language environment does not correspond
2353 ;; to any ISO 639 code, so let it stand for itself.
2354 ("ipa$" . "IPA")
2355
2356 ;; Nonstandard or obsolete language codes
2357 ("cz" . "Czech") ; e.g. Solaris 2.6
2358 ("ee" . "Latin-4") ; Estonian, e.g. X11R6.4
2359 ("iw" . "Hebrew") ; e.g. X11R6.4
f1282c7f 2360 ("sp" . "Cyrillic-ISO") ; Serbian (Cyrillic alphabet), e.g. X11R6.4
40c81f74 2361 ("su" . "Latin-1") ; Finnish, e.g. Solaris 2.6
2e86ceaa 2362 ("jp" . "Japanese") ; e.g. MS Windows
f21605fa 2363 ("chs" . "Chinese-GBK") ; MS Windows Chinese Simplified
86d9e628 2364 ("cht" . "Chinese-BIG5") ; MS Windows Chinese Traditional
569a6374
EZ		 2365 ("gbz" . "UTF-8") ; MS Windows Dari Persian
2366 ("div" . "UTF-8") ; MS Windows Divehi (Maldives)
2367 ("wee" . "Latin-2") ; MS Windows Lower Sorbian
2368 ("wen" . "Latin-2") ; MS Windows Upper Sorbian
367ca50f 2369 ))
8dedddd5 2370 "Alist of locale regexps vs the corresponding languages and coding systems.
caff3c0a 2371Each element has this form:
8dedddd5
KH		 2372 \(LOCALE-REGEXP LANG-ENV CODING-SYSTEM)
2373The first element whose LOCALE-REGEXP matches the start of a
caff3c0a 2374downcased locale specifies the LANG-ENV \(language environment)
8dedddd5
KH		 2375and CODING-SYSTEM corresponding to that locale. If there is no
2376appropriate language environment, the element may have this form:
2377 \(LOCALE-REGEXP . LANG-ENV)
2378In this case, LANG-ENV is one of generic language environments for an
2379specific encoding such as \"Latin-1\" and \"UTF-8\".")
40c81f74 2380
5f395df3
SM		 2381(defconst locale-charset-language-names
2382 (purecopy
2383 '((".*8859[-_]?1\\>" . "Latin-1")
2384 (".*8859[-_]?2\\>" . "Latin-2")
2385 (".*8859[-_]?3\\>" . "Latin-3")
2386 (".*8859[-_]?4\\>" . "Latin-4")
2387 (".*8859[-_]?9\\>" . "Latin-5")
2388 (".*8859[-_]?14\\>" . "Latin-8")
2389 (".*8859[-_]?15\\>" . "Latin-9")
6b61353c
KH		 2390 (".*utf\\(?:-?8\\)?\\>" . "UTF-8")
2391 ;; utf-8@euro exists, so put this last. (@euro really specifies
2392 ;; the currency, rather than the charset.)
e522f07b 2393 (".*@euro\\>" . "Latin-9")))
6ececc4d
PE		 2394 "List of pairs of locale regexps and charset language names.
2395The first element whose locale regexp matches the start of a downcased locale
6b61353c 2396specifies the language name whose charset corresponds to that locale.
8dedddd5 2397This language name is used if the locale is not listed in
caff3c0a 2398`locale-language-names'.")
6ececc4d 2399
5f395df3
SM		 2400(defconst locale-preferred-coding-systems
2401 (purecopy
8dedddd5
KH		 2402 '((".*8859[-_]?1\\>" . iso-8859-1)
2403 (".*8859[-_]?2\\>" . iso-8859-2)
2404 (".*8859[-_]?3\\>" . iso-8859-3)
2405 (".*8859[-_]?4\\>" . iso-8859-4)
2406 (".*8859[-_]?9\\>" . iso-8859-9)
2407 (".*8859[-_]?14\\>" . iso-8859-14)
2408 (".*8859[-_]?15\\>" . iso-8859-15)
2409 (".*utf\\(?:-?8\\)?" . utf-8)
2410 ;; utf-8@euro exists, so put this after utf-8. (@euro really
2411 ;; specifies the currency, rather than the charset.)
2412 (".*@euro" . iso-8859-15)
2413 ("koi8-?r" . koi8-r)
2414 ("koi8-?u" . koi8-u)
2415 ("tcvn" . tcvn)
38c05d07 2416 ("big5[-_]?hkscs" . big5-hkscs)
8dedddd5
KH		 2417 ("big5" . big5)
2418 ("euc-?tw" . euc-tw)
d6be7497 2419 ("euc-?cn" . euc-cn)
38c05d07
KH		 2420 ("gb2312" . gb2312)
2421 ("gbk" . gbk)
2422 ("gb18030" . gb18030)
8dedddd5 2423 ("ja.*[._]euc" . japanese-iso-8bit)
5f395df3
SM		 2424 ("ja.*[._]jis7" . iso-2022-jp)
2425 ("ja.*[._]pck" . japanese-shift-jis)
2426 ("ja.*[._]sjis" . japanese-shift-jis)
69210880 2427 ("jpn" . japanese-shift-jis) ; MS-Windows uses this.
8dedddd5 2428 ))
6ececc4d
PE		 2429 "List of pairs of locale regexps and preferred coding systems.
2430The first element whose locale regexp matches the start of a downcased locale
8dedddd5
KH		 2431specifies the coding system to prefer when using that locale.
2432This coding system is used if the locale specifies a specific charset.")
40c81f74
PE		 2433
2434(defun locale-name-match (key alist)
2435 "Search for KEY in ALIST, which should be a list of regexp-value pairs.
2436Return the value corresponding to the first regexp that matches the
2437start of KEY, or nil if there is no match."
2438 (let (element)
2439 (while (and alist (not element))
d0d8885d 2440 (if (string-match-p (concat "\\`\\(?:" (car (car alist)) "\\)") key)
40c81f74
PE		 2441 (setq element (car alist)))
2442 (setq alist (cdr alist)))
2443 (cdr element)))
2444
58b78d5b 2445(defun locale-charset-match-p (charset1 charset2)
91ed0599 2446 "Whether charset names (strings) CHARSET1 and CHARSET2 are equivalent.
58b78d5b
DL		 2447Matching is done ignoring case and any hyphens and underscores in the
2448names. E.g. `ISO_8859-1' and `iso88591' both match `iso-8859-1'."
2449 (setq charset1 (replace-regexp-in-string "[-_]" "" charset1))
2450 (setq charset2 (replace-regexp-in-string "[-_]" "" charset2))
2451 (eq t (compare-strings charset1 nil nil charset2 nil nil t)))
2452
6b61353c
KH		 2453(defvar locale-charset-alist nil
2454 "Coding system alist keyed on locale-style charset name.
2455Used by `locale-charset-to-coding-system'.")
2456
2457(defun locale-charset-to-coding-system (charset)
2458 "Find coding system corresponding to CHARSET.
2459CHARSET is any sort of non-Emacs charset name, such as might be used
2460in a locale codeset, or elsewhere. It is matched to a coding system
2461first by case-insensitive lookup in `locale-charset-alist'. Then
2462matches are looked for in the coding system list, treating case and
2463the characters `-' and `_' as insignificant. The coding system base
2464is returned. Thus, for instance, if charset \"ISO8859-2\",
2465`iso-latin-2' is returned."
f15466c5 2466 (or (car (assoc-string charset locale-charset-alist t))
6b61353c
KH		 2467 (let ((cs coding-system-alist)
2468 c)
2469 (while (and (not c) cs)
2470 (if (locale-charset-match-p charset (caar cs))
2471 (setq c (intern (caar cs)))
2472 (pop cs)))
2473 (if c (coding-system-base c)))))
2474
2475;; Fixme: This ought to deal with the territory part of the locale
2476;; too, for setting things such as calendar holidays, ps-print paper
2477;; size, spelling dictionary.
2478
44ee1bdf
GM		 2479(defun locale-translate (locale)
2480 "Expand LOCALE according to `locale-translation-file-name', if possible.
2481For example, translate \"swedish\" into \"sv_SE.ISO8859-1\"."
2482 (if locale-translation-file-name
2483 (with-temp-buffer
2484 (set-buffer-multibyte nil)
2485 (insert-file-contents locale-translation-file-name)
2486 (if (re-search-forward
2487 (concat "^" (regexp-quote locale) ":?[ \t]+") nil t)
2488 (buffer-substring (point) (line-end-position))
2489 locale))
2490 locale))
2491
36ab8612 2492(defun set-locale-environment (&optional locale-name frame)
40c81f74 2493 "Set up multi-lingual environment for using LOCALE-NAME.
758f07de
RS		 2494This sets the language environment, the coding system priority,
2495the default input method and sometimes other things.
2496
58b78d5b 2497LOCALE-NAME should be a string which is the name of a locale supported
6b61353c 2498by the system. Often it is of the form xx_XX.CODE, where xx is a
58b78d5b
DL		 2499language, XX is a country, and CODE specifies a character set and
2500coding system. For example, the locale name \"ja_JP.EUC\" might name
2501a locale for Japanese in Japan using the `japanese-iso-8bit'
2502coding-system. The name may also have a modifier suffix, e.g. `@euro'
2503or `@cyrillic'.
40c81f74 2504
758f07de 2505If LOCALE-NAME is nil, its value is taken from the environment
d37ef0f6 2506variables LC_ALL, LC_CTYPE and LANG (the first one that is set).
40c81f74
PE		 2507
2508The locale names supported by your system can typically be found in a
0812c1e8 2509directory named `/usr/share/locale' or `/usr/lib/locale'. LOCALE-NAME
758f07de 2510will be translated according to the table specified by
0812c1e8
DL		 2511`locale-translation-file-name'.
2512
36ab8612
MB		 2513If FRAME is non-nil, only set the keyboard coding system and the
2514terminal coding system for the terminal of that frame, and don't
2515touch session-global parameters like the language environment.
97c57fb2 2516
0812c1e8
DL		 2517See also `locale-charset-language-names', `locale-language-names',
2518`locale-preferred-coding-systems' and `locale-coding-system'."
758f07de 2519 (interactive "sSet environment for locale: ")
a1506d29 2520
0d7c5bb9
DL		 2521 ;; Do this at runtime for the sake of binaries possibly transported
2522 ;; to a system without X.
2523 (setq locale-translation-file-name
2524 (let ((files
6bba8c70
KH		 2525 '("/usr/share/X11/locale/locale.alias" ; e.g. X11R7
2526 "/usr/lib/X11/locale/locale.alias" ; e.g. X11R6.4
6b61353c 2527 "/usr/X11R6/lib/X11/locale/locale.alias" ; XFree86, e.g. RedHat 4.2
0d7c5bb9
DL		 2528 "/usr/openwin/lib/locale/locale.alias" ; e.g. Solaris 2.6
2529 ;;
2530 ;; The following name appears after the X-related names above,
2531 ;; since the X-related names are what X actually uses.
2532 "/usr/share/locale/locale.alias" ; GNU/Linux sans X
2533 )))
2534 (while (and files (not (file-exists-p (car files))))
2535 (setq files (cdr files)))
2536 (car files)))
2537
7008ccac
GM		 2538 (let ((locale locale-name))
2539
2540 (unless locale
2541 ;; Use the first of these three environment variables
2542 ;; that has a nonempty value.
2543 (let ((vars '("LC_ALL" "LC_CTYPE" "LANG")))
db6bd804
DL		 2544 (while (and vars
2545 (= 0 (length locale))) ; nil or empty string
36ab8612 2546 (setq locale (getenv (pop vars) frame)))))
7008ccac
GM		 2547
2548 (when locale
44ee1bdf 2549 (setq locale (locale-translate locale))
7008ccac
GM		 2550
2551 ;; Leave the system locales alone if the caller did not specify
2552 ;; an explicit locale name, as their defaults are set from
2553 ;; LC_MESSAGES and LC_TIME, not LC_CTYPE, and the user might not
2554 ;; want to set them to the same value as LC_CTYPE.
2555 (when locale-name
2556 (setq system-messages-locale locale)
e9aaa1db
KH		 2557 (setq system-time-locale locale))
2558
2559 (if (string-match "^[a-z][a-z]" locale)
2560 (setq current-iso639-language (intern (match-string 0 locale)))))
7008ccac 2561
44ee1bdf
GM		 2562 (setq woman-locale
2563 (or system-messages-locale
739b7764 2564 (let ((msglocale (getenv "LC_MESSAGES" frame)))
44ee1bdf
GM		 2565 (if (zerop (length msglocale))
2566 locale
2567 (locale-translate msglocale)))))
2568
2569 (when locale
7008ccac
GM		 2570 (setq locale (downcase locale))
2571
2572 (let ((language-name
2573 (locale-name-match locale locale-language-names))
2574 (charset-language-name
2575 (locale-name-match locale locale-charset-language-names))
cdb1af30 2576 (default-eol-type (coding-system-eol-type
b56a5ae0 2577 (default-value 'buffer-file-coding-system)))
7008ccac 2578 (coding-system
6b61353c
KH		 2579 (or (locale-name-match locale locale-preferred-coding-systems)
2580 (when locale
2581 (if (string-match "\\.\\([^@]+\\)" locale)
2582 (locale-charset-to-coding-system
9e2a2647 2583 (match-string 1 locale)))))))
7008ccac 2584
8dedddd5
KH		 2585 (if (consp language-name)
2586 ;; locale-language-names specify both lang-env and coding.
2587 ;; But, what specified in locale-preferred-coding-systems
2588 ;; has higher priority.
2589 (setq coding-system (or coding-system
2590 (nth 1 language-name))
2591 language-name (car language-name))
2592 ;; Otherwise, if locale is not listed in locale-language-names,
2593 ;; use what listed in locale-charset-language-names.
2594 (if (not language-name)
2595 (setq language-name charset-language-name)))
7008ccac 2596
cdb1af30
EZ		 2597 ;; If a specific EOL conversion was specified in the default
2598 ;; buffer-file-coding-system, preserve it in the coding system
2599 ;; we will be using from now on.
8bca692e
EZ		 2600 (if (and (memq default-eol-type '(0 1 2 unix dos mac))
2601 coding-system
2602 (coding-system-p coding-system))
cdb1af30
EZ		 2603 (setq coding-system (coding-system-change-eol-conversion
2604 coding-system default-eol-type)))
2605
7008ccac
GM		 2606 (when language-name
2607
2608 ;; Set up for this character set. This is now the right way
2609 ;; to do it for both unibyte and multibyte modes.
36ab8612 2610 (unless frame
97c57fb2 2611 (set-language-environment language-name))
7008ccac 2612
597e2240 2613 ;; If the default enable-multibyte-characters is nil,
7008ccac
GM		 2614 ;; we are using single-byte characters,
2615 ;; so the display table and terminal coding system are irrelevant.
597e2240 2616 (when (default-value 'enable-multibyte-characters)
ff76e074 2617 (set-display-table-and-terminal-coding-system
36ab8612 2618 language-name coding-system frame))
7008ccac 2619
166ce29f
DL		 2620 ;; Set the `keyboard-coding-system' if appropriate (tty
2621 ;; only). At least X and MS Windows can generate
2622 ;; multilingual input.
68bba4e4 2623 ;; XXX This was disabled unless `window-system', but that
fffa137c 2624 ;; leads to buggy behavior when a tty frame is opened
68bba4e4
KL		 2625 ;; later. Setting the keyboard coding system has no adverse
2626 ;; effect on X, so let's do it anyway. -- Lorentey
2627 (let ((kcs (or coding-system
2628 (car (get-language-info language-name
2629 'coding-system)))))
36ab8612 2630 (if kcs (set-keyboard-coding-system kcs frame)))
dc2be2fa 2631
36ab8612 2632 (unless frame
97c57fb2
KL		 2633 (setq locale-coding-system
2634 (car (get-language-info language-name 'coding-priority)))))
7008ccac 2635
36ab8612 2636 (when (and (not frame)
97c57fb2 2637 coding-system
8dedddd5
KH		 2638 (not (coding-system-equal coding-system
2639 locale-coding-system)))
7008ccac 2640 (prefer-coding-system coding-system)
218e7ce3
KH		 2641 ;; Fixme: perhaps prefer-coding-system should set this too.
2642 ;; But it's not the time to do such a fundamental change.
2643 (setq default-sendmail-coding-system coding-system)
80e3310b 2644 (setq locale-coding-system coding-system))))
e76ef161 2645
52c7f9ee 2646 ;; On Windows, override locale-coding-system,
26245233
JR		 2647 ;; default-file-name-coding-system, keyboard-coding-system,
2648 ;; terminal-coding-system with system codepage.
5eb94383 2649 (when (boundp 'w32-ansi-code-page)
893b49bb
JR		 2650 (let ((code-page-coding (intern (format "cp%d" w32-ansi-code-page))))
2651 (when (coding-system-p code-page-coding)
36ab8612
MB		 2652 (unless frame (setq locale-coding-system code-page-coding))
2653 (set-keyboard-coding-system code-page-coding frame)
d4aa48db 2654 (set-terminal-coding-system code-page-coding frame)
59938af3
EZ		 2655 ;; Set default-file-name-coding-system last, so that Emacs
2656 ;; doesn't try to use cpNNNN when it defines keyboard and
2657 ;; terminal encoding. That's because the above two lines
2658 ;; will want to load code-pages.el, where cpNNNN are
2659 ;; defined; if default-file-name-coding-system were set to
2660 ;; cpNNNN while these two lines run, Emacs will want to use
2661 ;; it for encoding the file name it wants to load. And that
2662 ;; will fail, since cpNNNN is not yet usable until
2663 ;; code-pages.el finishes loading.
2664 (setq default-file-name-coding-system code-page-coding))))
893b49bb 2665
a41118cc 2666 (when (eq system-type 'darwin)
ff76e074
SM		 2667 ;; On Darwin, file names are always encoded in utf-8, no matter
2668 ;; the locale.
2669 (setq default-file-name-coding-system 'utf-8)
2670 ;; Mac OS X's Terminal.app by default uses utf-8 regardless of
2671 ;; the locale.
2672 (when (and (null window-system)
36ab8612 2673 (equal (getenv "TERM_PROGRAM" frame) "Apple_Terminal"))
ff76e074
SM		 2674 (set-terminal-coding-system 'utf-8)
2675 (set-keyboard-coding-system 'utf-8)))
a41118cc 2676
e76ef161 2677 ;; Default to A4 paper if we're not in a C, POSIX or US locale.
3479c806 2678 ;; (See comments in Flocale_info.)
36ab8612 2679 (unless frame
97c57fb2
KL		 2680 (let ((locale locale)
2681 (paper (locale-info 'paper)))
2682 (if paper
2683 ;; This will always be null at the time of writing.
2684 (cond
2685 ((equal paper '(216 279))
2686 (setq ps-paper-type 'letter))
2687 ((equal paper '(210 297))
2688 (setq ps-paper-type 'a4)))
2689 (let ((vars '("LC_ALL" "LC_PAPER" "LANG")))
2690 (while (and vars (= 0 (length locale)))
36ab8612 2691 (setq locale (getenv (pop vars) frame))))
97c57fb2
KL		 2692 (when locale
2693 ;; As of glibc 2.2.5, these are the only US Letter locales,
2694 ;; and the rest are A4.
2695 (setq ps-paper-type
2696 (or (locale-name-match locale '(("c$" . letter)
2697 ("posix$" . letter)
2698 (".._us" . letter)
2699 (".._pr" . letter)
2700 (".._ca" . letter)
2701 ("enu$" . letter) ; Windows
2702 ("esu$" . letter)
2703 ("enc$" . letter)
2704 ("frc$" . letter)))
2705 'a4)))))))
e76ef161 2706 nil)
40c81f74 2707\f
16400c32
KH		 2708;;; Character property
2709
2710;; Each element has the form (PROP . TABLE).
2711;; PROP is a symbol representing a character property.
2712;; TABLE is a char-table containing the property value for each character.
2713;; TABLE may be a name of file to load to build a char-table.
2714;; Don't modify this variable directly but use `define-char-code-property'.
2715
2716(defvar char-code-property-alist nil
2717 "Alist of character property name vs char-table containing property values.
2718Internal use only.")
2719
2720(put 'char-code-property-table 'char-table-extra-slots 5)
2721
2722(defun define-char-code-property (name table &optional docstring)
2723 "Define NAME as a character code property given by TABLE.
2724TABLE is a char-table of purpose `char-code-property-table' with
2725these extra slots:
2726 1st: NAME.
2727 2nd: Function to call to get a property value of a character.
caff3c0a 2728 It is called with three arguments CHAR, VAL, and TABLE, where
16400c32
KH		 2729 CHAR is a character, VAL is the value of (aref TABLE CHAR).
2730 3rd: Function to call to put a property value of a character.
2731 It is called with the same arguments as above.
2732 4th: Function to call to get a description string of a property value.
2733 It is called with one argument VALUE, a property value.
2734 5th: Data used by the above functions.
2735
2736TABLE may be a name of file to load to build a char-table. The
2737file should contain a call of `define-char-code-property' with a
2738char-table of the above format as the argument TABLE.
2739
2740TABLE may also be nil, in which case no property value is pre-assigned.
2741
49275d55 2742Optional 3rd argument DOCSTRING is a documentation string of the property.
16400c32
KH		 2743
2744See also the documentation of `get-char-code-property' and
2745`put-char-code-property'."
2746 (or (symbolp name)
2747 (error "Not a symbol: %s" name))
2748 (if (char-table-p table)
2749 (or (and (eq (char-table-subtype table) 'char-code-property-table)
2750 (eq (char-table-extra-slot table 0) name))
2751 (error "Invalid char-table: %s" table))
2752 (or (stringp table)
2753 (error "Not a char-table nor a file name: %s" table)))
ebfa10d3 2754 (if (stringp table) (setq table (purecopy table)))
16400c32
KH		 2755 (let ((slot (assq name char-code-property-alist)))
2756 (if slot
2757 (setcdr slot table)
2758 (setq char-code-property-alist
2759 (cons (cons name table) char-code-property-alist))))
905a9ed3 2760 (put name 'char-code-property-documentation (purecopy docstring)))
4ed46869
KH		 2761
2762(defvar char-code-property-table
2763 (make-char-table 'char-code-property-table)
2764 "Char-table containing a property list of each character code.
16400c32 2765This table is used for properties not listed in `char-code-property-alist'.
4ed46869 2766See also the documentation of `get-char-code-property' and
96db204a 2767`put-char-code-property'.")
4ed46869
KH		 2768
2769(defun get-char-code-property (char propname)
16400c32
KH		 2770 "Return the value of CHAR's PROPNAME property."
2771 (let ((slot (assq propname char-code-property-alist)))
2772 (if slot
2773 (let (table value func)
2774 (if (stringp (cdr slot))
d0d8885d 2775 (load (cdr slot) nil t))
16400c32
KH		 2776 (setq table (cdr slot)
2777 value (aref table char)
2778 func (char-table-extra-slot table 1))
2779 (if (functionp func)
2780 (setq value (funcall func char value table)))
2781 value)
2782 (plist-get (aref char-code-property-table char) propname))))
4ed46869
KH		 2783
2784(defun put-char-code-property (char propname value)
16400c32 2785 "Store CHAR's PROPNAME property with VALUE.
4ed46869 2786It can be retrieved with `(get-char-code-property CHAR PROPNAME)'."
16400c32
KH		 2787 (let ((slot (assq propname char-code-property-alist)))
2788 (if slot
2789 (let (table func)
2790 (if (stringp (cdr slot))
d0d8885d 2791 (load (cdr slot) nil t))
16400c32
KH		 2792 (setq table (cdr slot)
2793 func (char-table-extra-slot table 2))
2794 (if (functionp func)
2795 (funcall func char value table)
2796 (aset table char value)))
2797 (let* ((plist (aref char-code-property-table char))
2798 (x (plist-put plist propname value)))
2799 (or (eq x plist)
2800 (aset char-code-property-table char x))))
2801 value))
2802
2803(defun char-code-property-description (prop value)
2804 "Return a description string of character property PROP's value VALUE.
2805If there's no description string for VALUE, return nil."
2806 (let ((slot (assq prop char-code-property-alist)))
2807 (if slot
2808 (let (table func)
2809 (if (stringp (cdr slot))
d0d8885d 2810 (load (cdr slot) nil t))
16400c32
KH		 2811 (setq table (cdr slot)
2812 func (char-table-extra-slot table 3))
2813 (if (functionp func)
2814 (funcall func value))))))
4ed46869 2815
a127b764
KH		 2816\f
2817;; Pretty description of encoded string
2818
2819;; Alist of ISO 2022 control code vs the corresponding mnemonic string.
2f7f4bee 2820(defconst iso-2022-control-alist
a127b764
KH		 2821 '((?\x1b . "ESC")
2822 (?\x0e . "SO")
2823 (?\x0f . "SI")
2824 (?\x8e . "SS2")
2825 (?\x8f . "SS3")
2826 (?\x9b . "CSI")))
2827
2828(defun encoded-string-description (str coding-system)
2829 "Return a pretty description of STR that is encoded by CODING-SYSTEM."
2830 (setq str (string-as-unibyte str))
993b2a7d 2831 (mapconcat
a6dfc99b 2832 (if (and coding-system (eq (coding-system-type coding-system) 'iso-2022))
993b2a7d
KH		 2833 ;; Try to get a pretty description for ISO 2022 escape sequences.
2834 (function (lambda (x) (or (cdr (assq x iso-2022-control-alist))
695effcc
JL		 2835 (format "#x%02X" x))))
2836 (function (lambda (x) (format "#x%02X" x))))
993b2a7d 2837 str " "))
a127b764 2838
430c6ced 2839(defun encode-coding-char (char coding-system &optional charset)
a127b764 2840 "Encode CHAR by CODING-SYSTEM and return the resulting string.
430c6ced
KH		 2841If CODING-SYSTEM can't safely encode CHAR, return nil.
2842The 3rd optional argument CHARSET, if non-nil, is a charset preferred
2843on encoding."
b529ed1e
KH		 2844 (let* ((str1 (string-as-multibyte (string char)))
2845 (str2 (string-as-multibyte (string char char)))
2846 (found (find-coding-systems-string str1))
a127b764 2847 enc1 enc2 i1 i2)
b529ed1e
KH		 2848 (if (and (consp found)
2849 (eq (car found) 'undecided))
2850 str1
2851 (when (memq (coding-system-base coding-system) found)
2852 ;; We must find the encoded string of CHAR. But, just encoding
2853 ;; CHAR will put extra control sequences (usually to designate
2854 ;; ASCII charset) at the tail if type of CODING is ISO 2022.
2855 ;; To exclude such tailing bytes, we at first encode one-char
2856 ;; string and two-char string, then check how many bytes at the
2857 ;; tail of both encoded strings are the same.
2858
430c6ced
KH		 2859 (when charset
2860 (put-text-property 0 1 'charset charset str1)
2861 (put-text-property 0 2 'charset charset str2))
b529ed1e
KH		 2862 (setq enc1 (encode-coding-string str1 coding-system)
2863 i1 (length enc1)
2864 enc2 (encode-coding-string str2 coding-system)
2865 i2 (length enc2))
2866 (while (and (> i1 0) (= (aref enc1 (1- i1)) (aref enc2 (1- i2))))
2867 (setq i1 (1- i1) i2 (1- i2)))
2868
2869 ;; Now (substring enc1 i1) and (substring enc2 i2) are the same,
2870 ;; and they are the extra control sequences at the tail to
2871 ;; exclude.
2872 (substring enc2 0 i2)))))
a127b764 2873
d2f613a0
DL		 2874;; Backwards compatibility. These might be better with :init-value t,
2875;; but that breaks loadup.
2876(define-minor-mode unify-8859-on-encoding-mode
2877 "Obsolete."
2878 :group 'mule
2879 :global t)
2880(define-minor-mode unify-8859-on-decoding-mode
2881 "Obsolete."
2882 :group 'mule
2883 :global t)
a127b764 2884
7dd42fb1
KH		 2885(defvar nonascii-insert-offset 0 "This variable is obsolete.")
2886(defvar nonascii-translation-table nil "This variable is obsolete.")
2887
838d78d4
JL		 2888(defvar ucs-names nil
2889 "Alist of cached (CHAR-NAME . CHAR-CODE) pairs.")
2890
2891(defun ucs-names ()
2892 "Return alist of (CHAR-NAME . CHAR-CODE) pairs cached in `ucs-names'."
2893 (or ucs-names
ae63e572
KH		 2894 (let ((bmp-ranges
2895 '((#x0000 . #x33FF)
2896 ;; (#x3400 . #x4DBF) CJK Ideograph Extension A
2897 (#x4DC0 . #x4DFF)
2898 ;; (#x4E00 . #x9FFF) CJK Ideograph
2899 (#xA000 . #x0D7FF)
2900 ;; (#xD800 . #xFAFF) Surrogate/Private
2901 (#xFB00 . #xFFFD)))
2902 (upper-ranges
2903 '((#x10000 . #x134FF)
2904 ;; (#x13500 . #x1CFFF) unsed
2905 (#x1D000 . #x1FFFF)
2906 ;; (#x20000 . #xDFFFF) CJK Ideograph Extension A, B, etc, unsed
2907 (#xE0000 . #xE01FF)))
2908 (gc-cons-threshold 10000000)
2909 c end name names)
2910 (dolist (range bmp-ranges)
2911 (setq c (car range)
2912 end (cdr range))
2913 (while (<= c end)
2914 (if (setq name (get-char-code-property c 'name))
2915 (push (cons name c) names))
2916 (if (setq name (get-char-code-property c 'old-name))
2917 (push (cons name c) names))
2918 (setq c (1+ c))))
2919 (dolist (range upper-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 (setq c (1+ c))))
e0727873 2926 (setq ucs-names names))))
838d78d4
JL		 2927
2928(defvar ucs-completions (lazy-completion-table ucs-completions ucs-names)
2929 "Lazy completion table for completing on Unicode character names.")
bdf66e8d 2930(put 'ucs-completions 'risky-local-variable t)
838d78d4
JL		 2931
2932(defun read-char-by-name (prompt)
2933 "Read a character by its Unicode name or hex number string.
c0a67379
JL		 2934Display PROMPT and read a string that represents a character by its
2935Unicode property `name' or `old-name'. You can type a few of first
2936letters of the Unicode name and use completion. This function also
2937accepts a hexadecimal number of Unicode code point or a number in
2938hash notation, e.g. #o21430 for octal, #x2318 for hex, or #10r8984
2939for decimal. Returns a character as a number."
838d78d4
JL		 2940 (let* ((completion-ignore-case t)
2941 (input (completing-read prompt ucs-completions)))
c0a67379 2942 (cond
d0d8885d 2943 ((string-match-p "^[0-9a-fA-F]+$" input)
c0a67379 2944 (string-to-number input 16))
d0d8885d 2945 ((string-match-p "^#" input)
c0a67379
JL		 2946 (read input))
2947 (t
d6ff002e 2948 (cdr (assoc-string input (ucs-names) t))))))
838d78d4 2949
dcbac02a
JL		 2950(defun ucs-insert (character &optional count inherit)
2951 "Insert COUNT copies of CHARACTER of the given Unicode code point.
c0a67379 2952Interactively, prompts for a Unicode character name or a hex number
dcbac02a
JL		 2953using `read-char-by-name'.
2954The optional third arg INHERIT (non-nil when called interactively),
2955says to inherit text properties from adjoining text, if those
2956properties are sticky."
2957 (interactive
2958 (list (read-char-by-name "Unicode (name or hex): ")
2959 (prefix-numeric-value current-prefix-arg)
2960 t))
2961 (unless count (setq count 1))
2962 (if (stringp character)
2963 (setq character (string-to-number character 16)))
c0a67379 2964 (cond
dcbac02a
JL		 2965 ((not (integerp character))
2966 (error "Not a Unicode character code: %S" character))
2967 ((or (< character 0) (> character #x10FFFF))
2968 (error "Not a Unicode character code: 0x%X" character)))
2969 (if inherit
2970 (dotimes (i count) (insert-and-inherit character))
2971 (dotimes (i count) (insert character))))
cb9e47dd 2972
c0a67379 2973(define-key ctl-x-map "8\r" 'ucs-insert)
a127b764 2974
9ee5b744 2975;; arch-tag: b382c432-4b36-460e-bf4c-05efd0bb18dc
4ed46869 2976;;; mule-cmds.el ends here
Unnamed repository; edit this file 'description' to name the repository.