Mercurial > hg > xemacs-beta
annotate lisp/mule/mule-cmds.el @ 4590:c83cab5a4f04
Display control characters using carets, invalid-sequence-coding-system
2009-01-31 Aidan Kehoe <kehoea@parhasard.net>
* mule/mule-cmds.el (finish-set-language-environment):
Treat control characters specially in the
invalid-sequence-coding-system handling, display them using the
caret notation.
| author | Aidan Kehoe <kehoea@parhasard.net> |
|---|---|
| date | Sat, 31 Jan 2009 17:38:07 +0000 |
| parents | 774e5c7522bf |
| children | e0a8715fdb1f |
| rev | line source |
|---|---|
| 502 | 1 ;;; mule-cmds.el --- Commands for multilingual environment -*- coding: iso-2022-7bit; -*- |
| 428 | 2 |
| 3 ;; Copyright (C) 1995,1999 Electrotechnical Laboratory, JAPAN. | |
| 4 ;; Licensed to the Free Software Foundation. | |
| 5 ;; Copyright (C) 1997 MORIOKA Tomohiko | |
| 1318 | 6 ;; Copyright (C) 2000, 2001, 2002, 2003 Ben Wing. |
| 428 | 7 |
| 8 ;; Keywords: mule, multilingual | |
| 9 | |
| 10 ;; This file is part of XEmacs. | |
| 11 | |
| 12 ;; XEmacs is free software; you can redistribute it and/or modify it | |
| 13 ;; under the terms of the GNU General Public License as published by | |
| 14 ;; the Free Software Foundation; either version 2, or (at your option) | |
| 15 ;; any later version. | |
| 16 | |
| 17 ;; XEmacs is distributed in the hope that it will be useful, but | |
| 18 ;; WITHOUT ANY WARRANTY; without even the implied warranty of | |
| 19 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU | |
| 20 ;; General Public License for more details. | |
| 21 | |
| 22 ;; You should have received a copy of the GNU General Public License | |
| 23 ;; along with XEmacs; see the file COPYING. If not, write to the Free | |
| 24 ;; Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA | |
| 25 ;; 02111-1307, USA. | |
| 26 | |
| 444 | 27 ;; Note: Some of the code here is now in code-cmds.el |
| 28 | |
| 428 | 29 ;;; Code: |
| 30 | |
| 31 ;;; MULE related key bindings and menus. | |
| 32 | |
| 444 | 33 ;; Preserve the old name |
| 34 (defvaralias 'mule-keymap 'coding-keymap) | |
| 35 | |
| 428 | 36 (define-key mule-keymap "x" 'set-selection-coding-system) |
| 37 (define-key mule-keymap "X" 'set-next-selection-coding-system) | |
| 38 (define-key mule-keymap "\C-\\" 'set-input-method) | |
| 39 ;;(define-key mule-keymap "c" 'list-coding-system-briefly) ; XEmacs | |
| 40 (define-key mule-keymap "C" 'describe-coding-system) ; XEmacs | |
| 41 (define-key mule-keymap "r" 'toggle-display-direction) ; XEmacs | |
| 42 (define-key mule-keymap "l" 'set-language-environment) | |
| 43 | |
| 44 (define-key help-map "L" 'describe-language-environment) | |
| 45 (define-key help-map "\C-\\" 'describe-input-method) | |
| 46 (define-key help-map "I" 'describe-input-method) | |
| 47 (define-key help-map "h" 'view-hello-file) | |
| 48 | |
| 49 ;; Menu for XEmacs were moved to menubar-items.el. | |
| 50 | |
| 51 | |
| 52 ;; This should be a single character key binding because users use it | |
| 53 ;; very frequently while editing multilingual text. Now we can use | |
| 54 ;; only two such keys: "\C-\\" and "\C-^", but the latter is not | |
| 55 ;; convenient because it requires shifting on most keyboards. An | |
| 56 ;; alternative is "\C-\]" which is now bound to `abort-recursive-edit' | |
| 57 ;; but it won't be used that frequently. | |
| 58 (define-key global-map "\C-\\" 'toggle-input-method) | |
| 59 | |
| 771 | 60 ;; Original mapping will be altered by set-keyboard-coding-system. |
| 61 (define-key global-map [(meta \#)] 'ispell-word) ;originally "$" | |
| 62 ;; (define-key global-map [(meta {)] 'insert-parentheses) ;originally "(" | |
| 63 | |
| 428 | 64 ;;; This is no good because people often type Shift-SPC |
| 65 ;;; meaning to type SPC. -- rms. | |
| 66 ;;; ;; Here's an alternative key binding for X users (Shift-SPACE). | |
| 771 | 67 ;;; (define-key global-map '(shift space) 'toggle-input-method) |
| 428 | 68 |
| 69 (defun view-hello-file () | |
| 70 "Display the HELLO file which list up many languages and characters." | |
| 71 (interactive) | |
| 72 ;; We have to decode the file in any environment. | |
| 73 (let ((coding-system-for-read 'iso-2022-7bit)) | |
| 74 (find-file-read-only (expand-file-name "HELLO" data-directory)))) | |
| 75 | |
| 3999 | 76 (defvar system-type-file-name-coding |
| 77 '((darwin . utf-8)) | |
| 78 "A map from values of `system-type' to invariant file name coding systems. | |
| 79 Used if a give system type does not vary in the coding system it uses for | |
| 80 file names; otherwise, `language-info-alist' is consulted for this | |
| 81 information. This affects the `file-name' coding system alias, but not the | |
| 82 `file-name-coding-system' variable, which in practice is mostly ignored. ") | |
| 83 | |
| 428 | 84 |
| 771 | 85 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; |
| 86 ;;; Language Support Functions ;;; | |
| 87 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; | |
| 428 | 88 (defvar language-info-alist nil |
| 89 "Alist of language environment definitions. | |
| 90 Each element looks like: | |
| 771 | 91 (LANGUAGE-NAME . ((PROP . VALUE) ...)) |
| 428 | 92 where LANGUAGE-NAME is a string, the name of the language environment, |
| 771 | 93 PROP is a symbol denoting a property, and VALUE is the data associated |
| 94 with PROP. | |
| 95 See `set-language-info' for documentation on PROP and VALUE.") | |
| 428 | 96 |
| 771 | 97 (defun get-language-info (lang-env prop) |
| 98 "Return information listed under PROP for language environment LANG-ENV. | |
| 99 PROP is a symbol denoting a property. | |
| 100 For a list of useful values for PROP and their meanings, | |
| 101 see `set-language-info'." | |
| 428 | 102 (if (symbolp lang-env) |
| 103 (setq lang-env (symbol-name lang-env))) | |
| 104 (let ((lang-slot (assoc-ignore-case lang-env language-info-alist))) | |
| 105 (if lang-slot | |
| 771 | 106 (cdr (assq prop (cdr lang-slot)))))) |
| 428 | 107 |
| 771 | 108 (defun set-language-info (lang-env prop value) |
| 428 | 109 "Modify part of the definition of language environment LANG-ENV. |
| 771 | 110 Specifically, this stores the information VALUE under PROP |
| 428 | 111 in the definition of this language environment. |
| 771 | 112 PROP is a symbol denoting a property, and VALUE is the value of that property. |
| 113 | |
| 114 Meaningful values for PROP include | |
| 115 | |
| 116 documentation VALUE is documentation of what this language environment | |
| 117 is meant for, and how to use it. | |
| 118 | |
| 119 charset VALUE is a list of the character sets used by this | |
| 120 language environment. | |
| 121 | |
| 122 sample-text VALUE is one line of text, | |
| 123 written using those character sets, | |
| 124 appropriate for this language environment. | |
| 125 | |
| 126 setup-function VALUE is a function to call to switch to this | |
| 127 language environment. | |
| 128 | |
| 129 exit-function VALUE is a function to call to leave this | |
| 130 language environment. | |
| 131 | |
| 132 coding-system VALUE is a list of coding systems that are good | |
| 133 for saving text written in this language environment. | |
| 134 This list serves as suggestions to the user; | |
| 135 in effect, as a kind of documentation. | |
| 136 | |
| 137 coding-priority VALUE is a list of coding systems for this language | |
| 138 environment, in order of decreasing priority. | |
| 139 This is used to set up the coding system priority | |
| 140 list when you switch to this language environment. | |
| 141 | |
| 142 input-method VALUE is a default input method for this language | |
| 143 environment. | |
| 144 | |
| 145 features VALUE is a list of features requested in this | |
| 146 language environment. | |
| 147 | |
| 148 tutorial VALUE is a tutorial file name written in the language. | |
| 149 | |
| 150 locale VALUE is a list of locale expressions, which serve | |
| 151 two purposes: (1) Determining the language | |
| 152 environment from the current system locale at | |
| 153 startup, and (2) determining how to set the system | |
| 154 locale when the language environment is changed. | |
| 155 Each expression will be tried in turn, and should | |
| 156 be a string (for case (1), the string is matched | |
| 157 against the current locale using the regular | |
| 158 expression \"^STRING[^A-Za-z0-9]\"; for case (2), | |
| 159 the string is passed directly to | |
| 160 `set-current-locale' until a non-nil result is | |
| 161 returned), or a function of one argument. For | |
| 162 case (1), this argument will be a locale, and the | |
| 163 function should return t or nil to indicate | |
| 164 whether this locale matches the language | |
| 165 environment; for case (2), the argument will be | |
| 166 nil, and the function should call | |
| 167 `set-current-locale' itself and return the set | |
| 168 locale string if the locale was successfully set, | |
| 169 and nil otherwise. | |
| 428 | 170 |
| 771 | 171 NOTE: This property is *NOT* used under MS Windows; |
| 172 instead, the `mswindows-locale' property is used. | |
| 173 | |
| 174 cygwin-locale VALUE specifies a general Unix-style C library | |
| 175 locale that will be used to initialize the LANG | |
| 176 environment variable under MS Windows native, when the | |
| 177 system cannot test out the locales specified in the | |
| 178 `locale' property. This is so that Cygwin programs | |
| 179 can be run from an MS Windows native XEmacs. If not | |
| 180 specified, the last entry in `locale' will be used. | |
| 181 | |
| 182 native-coding-system VALUE is a single coding-system expression, or a | |
| 183 list of such expressions. These expressions are | |
| 184 used to compute the operating system's native | |
| 185 coding system, i.e. the coding system to be used | |
| 186 as the alias for `native' and `file-name'. This | |
| 187 specifies the coding system used for text | |
| 188 exchanged with the operating system, such as file | |
| 189 names, environment variables, subprocess | |
| 190 arguments, etc. Each expression should be either | |
| 191 a symbol naming a coding system or a function | |
| 192 (anything that is `functionp') of one argument, | |
| 193 which is passed the current locale corresponding | |
| 194 to this language environment and should return a | |
| 195 coding system or nil. Each expression is tried in | |
| 196 turn until a coding system is obtained. If there | |
| 197 is no non-nil result, or no value is specified for | |
| 198 this property, the first coding system listed | |
| 199 under the `coding-system' property is used. | |
| 200 | |
| 201 NOTE: This is *NOT* used under MS Windows. | |
| 202 Instead, `mswindows-multibyte-system-default' | |
| 203 is always used, since the system default code | |
| 204 page is what the Win32 API routines make use | |
| 205 of, and this cannot be changed. (We get around | |
| 206 this by using the Unicode versions whenever | |
| 207 possible -- i.e. on Windows NT/2000, and on | |
| 208 Windows 9x with the few API's that support | |
| 209 Unicode.) | |
| 210 | |
| 211 mswindows-locale VALUE is an element of the form MSWINDOWS-LOCALE, or | |
| 212 a list of such elements. Each element is an MS | |
| 213 Windows locale, of the form that can be passed to | |
| 214 `mswindows-set-current-locale'. This property is used | |
| 215 both to determine the current language environment at | |
| 216 startup (by matching MSWINDOWS-LOCALE against the | |
| 217 value returned by `mswindows-user-default-locale') and | |
| 218 to set the values of `set-current-locale' and | |
| 219 `mswindows-set-current-locale' when the current | |
| 220 language environment is changed. (The correct CLIB | |
| 221 locale can always be generated by passing in the | |
| 222 SUBLANG, with dashes in place of underscores, or the | |
| 223 LANG if there's no SUBLANG. The return value will be | |
| 224 the canonicalized locale, in proper CLIB form.) | |
| 225 | |
| 226 If there is no value for this property, the MS Windows | |
| 227 locale is assumed to have the same name as the | |
|
4489
b75b075a9041
Support displaying invalid UTF-8 in language-environment-specific ways.
Aidan Kehoe <kehoea@parhasard.net>
parents:
4468
diff
changeset
|
228 language environment. |
|
b75b075a9041
Support displaying invalid UTF-8 in language-environment-specific ways.
Aidan Kehoe <kehoea@parhasard.net>
parents:
4468
diff
changeset
|
229 |
|
4490
67fbcaf3dbdc
error-sequence -> invalid-sequence
Aidan Kehoe <kehoea@parhasard.net>
parents:
4489
diff
changeset
|
230 invalid-sequence-coding-system |
|
4489
b75b075a9041
Support displaying invalid UTF-8 in language-environment-specific ways.
Aidan Kehoe <kehoea@parhasard.net>
parents:
4468
diff
changeset
|
231 VALUE is a fixed-width 8-bit coding system used to |
|
b75b075a9041
Support displaying invalid UTF-8 in language-environment-specific ways.
Aidan Kehoe <kehoea@parhasard.net>
parents:
4468
diff
changeset
|
232 display Unicode error sequences (using a face to make |
|
b75b075a9041
Support displaying invalid UTF-8 in language-environment-specific ways.
Aidan Kehoe <kehoea@parhasard.net>
parents:
4468
diff
changeset
|
233 it clear that the data is invalid). In Western Europe |
|
b75b075a9041
Support displaying invalid UTF-8 in language-environment-specific ways.
Aidan Kehoe <kehoea@parhasard.net>
parents:
4468
diff
changeset
|
234 this is normally windows-1252; in the Russia and the |
|
b75b075a9041
Support displaying invalid UTF-8 in language-environment-specific ways.
Aidan Kehoe <kehoea@parhasard.net>
parents:
4468
diff
changeset
|
235 former Soviet Union koi8-ru or windows-1251 makes more |
|
b75b075a9041
Support displaying invalid UTF-8 in language-environment-specific ways.
Aidan Kehoe <kehoea@parhasard.net>
parents:
4468
diff
changeset
|
236 sense." |
| 428 | 237 (if (symbolp lang-env) |
| 238 (setq lang-env (symbol-name lang-env))) | |
| 771 | 239 (let (lang-slot prop-slot) |
| 428 | 240 (setq lang-slot (assoc lang-env language-info-alist)) |
| 241 (if (null lang-slot) ; If no slot for the language, add it. | |
| 242 (setq lang-slot (list lang-env) | |
| 243 language-info-alist (cons lang-slot language-info-alist))) | |
| 771 | 244 (setq prop-slot (assq prop lang-slot)) |
| 245 (if (null prop-slot) ; If no slot for the prop, add it. | |
| 428 | 246 (progn |
| 771 | 247 (setq prop-slot (list prop)) |
| 248 (setcdr lang-slot (cons prop-slot (cdr lang-slot))))) | |
| 249 (setcdr prop-slot value))) | |
| 428 | 250 |
| 251 (defun set-language-info-alist (lang-env alist &optional parents) | |
| 252 "Store ALIST as the definition of language environment LANG-ENV. | |
| 771 | 253 ALIST is an alist of properties and values. See the documentation of |
| 254 `set-language-info' for the allowed properties." | |
| 428 | 255 (if (symbolp lang-env) |
| 256 (setq lang-env (symbol-name lang-env))) | |
| 771 | 257 ;; FSF has 30 lines of unbelievably ugly code to set up the menus |
| 258 ;; appropriately. We just use a filter. | |
| 259 (while alist | |
| 260 (set-language-info lang-env (car (car alist)) (cdr (car alist))) | |
| 3707 | 261 (setq alist (cdr alist))) |
| 262 lang-env) | |
| 428 | 263 |
| 264 (defun read-language-name (key prompt &optional default) | |
| 265 "Read a language environment name which has information for KEY. | |
| 266 If KEY is nil, read any language environment. | |
| 267 Prompt with PROMPT. DEFAULT is the default choice of language environment. | |
| 268 This returns a language environment name as a string." | |
| 269 (let* ((completion-ignore-case t) | |
| 270 (name (completing-read prompt | |
| 271 language-info-alist | |
| 272 (and key | |
| 273 (function (lambda (elm) (assq key elm)))) | |
| 274 t nil nil default))) | |
| 275 (if (and (> (length name) 0) | |
| 276 (or (not key) | |
| 277 (get-language-info name key))) | |
| 278 name))) | |
| 279 | |
| 280 ;;; Multilingual input methods. | |
| 281 | |
| 282 (defconst leim-list-file-name "leim-list.el" | |
| 283 "Name of LEIM list file. | |
| 284 This file contains a list of libraries of Emacs input methods (LEIM) | |
| 285 in the format of Lisp expression for registering each input method. | |
| 286 Emacs loads this file at startup time.") | |
| 287 | |
| 288 (defvar leim-list-header (format | |
| 289 ";;; %s -- list of LEIM (Library of Emacs Input Method) | |
| 290 ;; | |
| 291 ;; This file contains a list of LEIM (Library of Emacs Input Method) | |
| 292 ;; in the same directory as this file. Loading this file registers | |
| 293 ;; the whole input methods in Emacs. | |
| 294 ;; | |
| 295 ;; Each entry has the form: | |
| 296 ;; (register-input-method | |
| 297 ;; INPUT-METHOD LANGUAGE-NAME ACTIVATE-FUNC | |
| 298 ;; TITLE DESCRIPTION | |
| 299 ;; ARG ...) | |
| 300 ;; See the function `register-input-method' for the meanings of arguments. | |
| 301 ;; | |
| 302 ;; If this directory is included in load-path, Emacs automatically | |
| 303 ;; loads this file at startup time. | |
| 304 | |
| 305 " | |
| 306 leim-list-file-name) | |
| 307 "Header to be inserted in LEIM list file.") | |
| 308 | |
| 309 (defvar leim-list-entry-regexp "^(register-input-method" | |
| 310 "Regexp matching head of each entry in LEIM list file. | |
| 311 See also the variable `leim-list-header'") | |
| 312 | |
| 313 (defvar update-leim-list-functions | |
| 314 '(quail-update-leim-list-file) | |
| 315 "List of functions to call to update LEIM list file. | |
| 316 Each function is called with one arg, LEIM directory name.") | |
| 317 | |
| 318 (defun update-leim-list-file (&rest dirs) | |
| 319 "Update LEIM list file in directories DIRS." | |
| 320 (let ((functions update-leim-list-functions)) | |
| 321 (while functions | |
| 322 (apply (car functions) dirs) | |
| 323 (setq functions (cdr functions))))) | |
| 324 | |
| 325 (defvar current-input-method nil | |
| 326 "The current input method for multilingual text. | |
| 327 If nil, that means no input method is activated now.") | |
| 328 (make-variable-buffer-local 'current-input-method) | |
| 329 (put 'current-input-method 'permanent-local t) | |
| 330 | |
| 331 (defvar current-input-method-title nil | |
| 332 "Title string of the current input method shown in mode line.") | |
| 333 (make-variable-buffer-local 'current-input-method-title) | |
| 334 (put 'current-input-method-title 'permanent-local t) | |
| 335 | |
| 336 (defcustom default-input-method nil | |
| 337 "*Default input method for multilingual text (a string). | |
| 338 This is the input method activated automatically by the command | |
| 339 `toggle-input-method' (\\[toggle-input-method])." | |
| 340 :group 'mule | |
| 341 :type '(choice (const nil) string)) | |
| 342 | |
| 343 (put 'input-method-function 'permanent-local t) | |
| 344 | |
| 345 (defvar input-method-history nil | |
| 346 "History list for some commands that read input methods.") | |
| 347 (make-variable-buffer-local 'input-method-history) | |
| 348 (put 'input-method-history 'permanent-local t) | |
| 349 | |
| 350 (defvar inactivate-current-input-method-function nil | |
| 351 "Function to call for inactivating the current input method. | |
| 352 Every input method should set this to an appropriate value when activated. | |
| 353 This function is called with no argument. | |
| 354 | |
| 355 This function should never change the value of `current-input-method'. | |
| 356 It is set to nil by the function `inactivate-input-method'.") | |
| 357 (make-variable-buffer-local 'inactivate-current-input-method-function) | |
| 358 (put 'inactivate-current-input-method-function 'permanent-local t) | |
| 359 | |
| 360 (defvar describe-current-input-method-function nil | |
| 361 "Function to call for describing the current input method. | |
| 362 This function is called with no argument.") | |
| 363 (make-variable-buffer-local 'describe-current-input-method-function) | |
| 364 (put 'describe-current-input-method-function 'permanent-local t) | |
| 365 | |
| 366 (defvar input-method-alist nil | |
| 2970 | 367 "Alist mapping input method names to information used by the LEIM API. |
| 368 Elements have the form (METHOD LANGUAGE ACTIVATOR TITLE DESCRIPTION ARGS...). | |
| 369 Use `register-input-method' to add input methods to the database. See its | |
| 370 documentation for the meanings of the elements.") | |
| 428 | 371 |
| 2970 | 372 (defun register-input-method (method language |
| 373 ;; #### shouldn't be optional, but need to | |
| 374 ;; audit callers | |
| 375 &optional activator title description | |
| 376 &rest args) | |
| 377 "Register METHOD as an input method for language environment LANGUAGE. | |
| 428 | 378 |
| 2970 | 379 METHOD and LANGUAGE may be symbols or strings. |
| 380 ACTIVATOR is the function called to activate this method. METHOD (the | |
| 381 invocation name) and ARGS are passed to the function on activation. | |
| 428 | 382 TITLE is a string to show in the mode line when this method is active. |
| 383 DESCRIPTION is a string describing this method and what it is good for. | |
| 2970 | 384 Optional ARGS, if any, are stored and passed as arguments to ACTIVATOR. |
| 428 | 385 |
| 2970 | 386 When registering a new Quail input method, the input method title should be |
| 387 the one given in the third parameter of `quail-define-package' (if the values | |
| 388 are different, the string specified in this function takes precedence). | |
| 428 | 389 |
| 2970 | 390 The information provided is registered in `input-method-alist'. The commands |
| 391 `describe-input-method' and `list-input-methods' use this database to show | |
| 392 information about input methods without loading them." | |
| 393 (if (symbolp language) | |
| 394 (setq language (symbol-name language))) | |
| 395 (if (symbolp method) | |
| 396 (setq method (symbol-name method))) | |
| 397 (let ((info (append (list language activator title description) args)) | |
| 398 (slot (assoc method input-method-alist))) | |
| 428 | 399 (if slot |
| 400 (setcdr slot info) | |
| 2970 | 401 (setq slot (cons method info)) |
| 428 | 402 (setq input-method-alist (cons slot input-method-alist))))) |
| 403 | |
| 404 (defun read-input-method-name (prompt &optional default inhibit-null) | |
| 405 "Read a name of input method from a minibuffer prompting with PROMPT. | |
| 406 If DEFAULT is non-nil, use that as the default, | |
| 407 and substitute it into PROMPT at the first `%s'. | |
| 408 If INHIBIT-NULL is non-nil, null input signals an error. | |
| 409 | |
| 410 The return value is a string." | |
| 411 (if default | |
| 412 (setq prompt (format prompt default))) | |
| 413 (let* ((completion-ignore-case t) | |
| 414 ;; This binding is necessary because input-method-history is | |
| 415 ;; buffer local. | |
| 416 (input-method (completing-read prompt input-method-alist | |
| 417 nil t nil 'input-method-history | |
| 418 default))) | |
| 419 (if (and input-method (symbolp input-method)) | |
| 420 (setq input-method (symbol-name input-method))) | |
| 421 (if (> (length input-method) 0) | |
| 422 input-method | |
| 423 (if inhibit-null | |
| 424 (error "No valid input method is specified"))))) | |
| 425 | |
| 426 (defun activate-input-method (input-method) | |
| 427 "Switch to input method INPUT-METHOD for the current buffer. | |
| 428 If some other input method is already active, turn it off first. | |
| 429 If INPUT-METHOD is nil, deactivate any current input method." | |
| 430 (if (and input-method (symbolp input-method)) | |
| 431 (setq input-method (symbol-name input-method))) | |
| 432 (if (and current-input-method | |
| 433 (not (string= current-input-method input-method))) | |
| 434 (inactivate-input-method)) | |
| 435 (unless (or current-input-method (null input-method)) | |
| 436 (let ((slot (assoc input-method input-method-alist))) | |
| 437 (if (null slot) | |
| 438 (error "Can't activate input method `%s'" input-method)) | |
| 439 (let ((func (nth 2 slot))) | |
| 440 (if (functionp func) | |
| 441 (apply (nth 2 slot) input-method (nthcdr 5 slot)) | |
| 442 (if (and (consp func) (symbolp (car func)) (symbolp (cdr func))) | |
| 443 (progn | |
| 444 (require (cdr func)) | |
| 445 (apply (car func) input-method (nthcdr 5 slot))) | |
| 446 (error "Can't activate input method `%s'" input-method)))) | |
| 447 (setq current-input-method input-method) | |
| 448 (setq current-input-method-title (nth 3 slot)) | |
| 449 (unwind-protect | |
| 450 (run-hooks 'input-method-activate-hook) | |
| 451 (force-mode-line-update))))) | |
| 452 | |
| 453 (defun inactivate-input-method () | |
| 454 "Turn off the current input method." | |
| 455 (when current-input-method | |
| 456 (if input-method-history | |
| 457 (unless (string= current-input-method (car input-method-history)) | |
| 458 (setq input-method-history | |
| 459 (cons current-input-method | |
| 460 (delete current-input-method input-method-history)))) | |
| 461 (setq input-method-history (list current-input-method))) | |
| 462 (unwind-protect | |
| 463 (funcall inactivate-current-input-method-function) | |
| 464 (unwind-protect | |
| 465 (run-hooks 'input-method-inactivate-hook) | |
| 466 (setq current-input-method nil | |
| 467 current-input-method-title nil) | |
| 468 (force-mode-line-update))))) | |
| 469 | |
| 470 (defun set-input-method (input-method) | |
| 471 "Select and activate input method INPUT-METHOD for the current buffer. | |
| 472 This also sets the default input method to the one you specify." | |
| 473 (interactive | |
| 474 (let* ((default (or (car input-method-history) default-input-method))) | |
| 475 (list (read-input-method-name | |
| 476 (if default "Select input method (default %s): " "Select input method: ") | |
| 477 default t)))) | |
| 478 (activate-input-method input-method) | |
| 479 (setq default-input-method input-method)) | |
| 480 | |
| 481 (defun toggle-input-method (&optional arg) | |
| 482 "Turn on or off a multilingual text input method for the current buffer. | |
| 483 | |
| 484 With no prefix argument, if an input method is currently activated, | |
| 485 turn it off. Otherwise, activate an input method -- the one most | |
| 486 recently used, or the one specified in `default-input-method', or | |
| 487 the one read from the minibuffer. | |
| 488 | |
| 489 With a prefix argument, read an input method from the minibuffer and | |
| 490 turn it on. | |
| 491 | |
| 492 The default is to use the most recent input method specified | |
| 493 \(not including the currently active input method, if any)." | |
| 494 (interactive "P") | |
| 495 (if (and current-input-method (not arg)) | |
| 496 (inactivate-input-method) | |
| 497 (let ((default (or (car input-method-history) default-input-method))) | |
| 498 (if (and arg default (equal current-input-method default) | |
| 499 (> (length input-method-history) 1)) | |
| 500 (setq default (nth 1 input-method-history))) | |
| 501 (activate-input-method | |
| 502 (if (or arg (not default)) | |
| 503 (progn | |
| 504 (read-input-method-name | |
| 505 (if default "Input method (default %s): " "Input method: " ) | |
| 506 default t)) | |
| 507 default)) | |
| 508 (or default-input-method | |
| 509 (setq default-input-method current-input-method))))) | |
| 510 | |
| 511 (defun describe-input-method (input-method) | |
| 512 "Describe input method INPUT-METHOD." | |
| 513 (interactive | |
| 514 (list (read-input-method-name | |
| 515 "Describe input method (default, current choice): "))) | |
| 516 (if (and input-method (symbolp input-method)) | |
| 517 (setq input-method (symbol-name input-method))) | |
| 518 (if (null input-method) | |
| 519 (describe-current-input-method) | |
| 520 (with-output-to-temp-buffer "*Help*" | |
| 521 (let ((elt (assoc input-method input-method-alist))) | |
| 522 (princ (format "Input method: %s (`%s' in mode line) for %s\n %s\n" | |
| 523 input-method (nth 3 elt) (nth 1 elt) (nth 4 elt))))))) | |
| 524 | |
| 525 (defun describe-current-input-method () | |
| 526 "Describe the input method currently in use." | |
| 527 (if current-input-method | |
| 528 (if (and (symbolp describe-current-input-method-function) | |
| 529 (fboundp describe-current-input-method-function)) | |
| 530 (funcall describe-current-input-method-function) | |
| 531 (message "No way to describe the current input method `%s'" | |
| 532 current-input-method) | |
| 533 (ding)) | |
| 534 (error "No input method is activated now"))) | |
| 535 | |
| 536 (defun read-multilingual-string (prompt &optional initial-input input-method) | |
| 537 "Read a multilingual string from minibuffer, prompting with string PROMPT. | |
| 538 The input method selected last time is activated in minibuffer. | |
| 539 If optional second arg INITIAL-INPUT is non-nil, insert it in the minibuffer | |
| 540 initially. | |
| 541 Optional 3rd argument INPUT-METHOD specifies the input method | |
| 542 to be activated instead of the one selected last time. It is a symbol | |
| 543 or a string." | |
| 544 (setq input-method | |
| 545 (or input-method | |
| 546 current-input-method | |
| 547 default-input-method | |
| 548 (read-input-method-name "Input method: " nil t))) | |
| 549 (if (and input-method (symbolp input-method)) | |
| 550 (setq input-method (symbol-name input-method))) | |
| 551 (let ((prev-input-method current-input-method)) | |
| 552 (unwind-protect | |
| 553 (progn | |
| 554 (activate-input-method input-method) | |
| 555 ;; FSF Emacs | |
| 556 ;; (read-string prompt initial-input nil nil t) | |
| 557 (read-string prompt initial-input nil)) | |
| 558 (activate-input-method prev-input-method)))) | |
| 559 | |
| 560 ;; Variables to control behavior of input methods. All input methods | |
| 561 ;; should react to these variables. | |
| 562 | |
| 563 (defcustom input-method-verbose-flag 'default | |
| 564 "*A flag to control extra guidance given by input methods. | |
| 565 The value should be nil, t, `complex-only', or `default'. | |
| 566 | |
| 567 The extra guidance is done by showing list of available keys in echo | |
| 568 area. When you use the input method in the minibuffer, the guidance | |
| 569 is shown at the bottom short window (split from the existing window). | |
| 570 | |
| 571 If the value is t, extra guidance is always given, if the value is | |
| 572 nil, extra guidance is always suppressed. | |
| 573 | |
| 574 If the value is `complex-only', only complex input methods such as | |
| 575 `chinese-py' and `japanese' give extra guidance. | |
| 576 | |
| 577 If the value is `default', complex input methods always give extra | |
| 578 guidance, but simple input methods give it only when you are not in | |
| 579 the minibuffer. | |
| 580 | |
| 581 See also the variable `input-method-highlight-flag'." | |
| 582 :type '(choice (const t) (const nil) (const complex-only) (const default)) | |
| 583 :group 'mule) | |
| 584 | |
| 585 (defcustom input-method-highlight-flag t | |
| 586 "*If this flag is non-nil, input methods highlight partially-entered text. | |
| 587 For instance, while you are in the middle of a Quail input method sequence, | |
| 588 the text inserted so far is temporarily underlined. | |
| 589 The underlining goes away when you finish or abort the input method sequence. | |
| 590 See also the variable `input-method-verbose-flag'." | |
| 591 :type 'boolean | |
| 592 :group 'mule) | |
| 593 | |
| 594 (defvar input-method-activate-hook nil | |
| 595 "Normal hook run just after an input method is activated. | |
| 596 | |
| 597 The variable `current-input-method' keeps the input method name | |
| 598 just activated.") | |
| 599 | |
| 600 (defvar input-method-inactivate-hook nil | |
| 601 "Normal hook run just after an input method is inactivated. | |
| 602 | |
| 603 The variable `current-input-method' still keeps the input method name | |
| 604 just inactivated.") | |
| 605 | |
| 606 (defvar input-method-after-insert-chunk-hook nil | |
| 607 "Normal hook run just after an input method insert some chunk of text.") | |
| 608 | |
| 609 (defvar input-method-exit-on-first-char nil | |
| 610 "This flag controls a timing when an input method returns. | |
| 611 Usually, the input method does not return while there's a possibility | |
| 612 that it may find a different translation if a user types another key. | |
| 613 But, it this flag is non-nil, the input method returns as soon as | |
| 614 the current key sequence gets long enough to have some valid translation.") | |
| 615 | |
| 616 (defvar input-method-use-echo-area nil | |
| 617 "This flag controls how an input method shows an intermediate key sequence. | |
| 618 Usually, the input method inserts the intermediate key sequence, | |
| 619 or candidate translations corresponding to the sequence, | |
| 620 at point in the current buffer. | |
| 621 But, if this flag is non-nil, it displays them in echo area instead.") | |
| 622 | |
| 623 (defvar input-method-exit-on-invalid-key nil | |
| 442 | 624 "This flag controls the behavior of an input method on invalid key input. |
| 428 | 625 Usually, when a user types a key which doesn't start any character |
| 626 handled by the input method, the key is handled by turning off the | |
| 627 input method temporarily. After that key, the input method is re-enabled. | |
| 628 But, if this flag is non-nil, the input method is never back on.") | |
| 629 | |
| 630 | |
| 631 (defvar set-language-environment-hook nil | |
| 632 "Normal hook run after some language environment is set. | |
| 633 | |
| 634 When you set some hook function here, that effect usually should not | |
| 635 be inherited to another language environment. So, you had better set | |
| 636 another function in `exit-language-environment-hook' (which see) to | |
| 637 cancel the effect.") | |
| 638 | |
| 639 (defvar exit-language-environment-hook nil | |
| 640 "Normal hook run after exiting from some language environment. | |
| 641 When this hook is run, the variable `current-language-environment' | |
| 642 is still bound to the language environment being exited. | |
| 643 | |
| 644 This hook is mainly used for canceling the effect of | |
| 645 `set-language-environment-hook' (which-see).") | |
| 646 | |
| 771 | 647 ;; bogus FSF function setup-specified-language-support. |
| 428 | 648 |
| 649 (defcustom current-language-environment "English" | |
| 650 "The last language environment specified with `set-language-environment'. | |
| 651 This variable should be set only with \\[customize], which is equivalent | |
| 652 to using the function `set-language-environment'." | |
| 653 :link '(custom-manual "(emacs)Language Environments") | |
| 654 :set (lambda (symbol value) (set-language-environment value)) | |
| 655 :get (lambda (x) | |
| 656 (or (car-safe (assoc-ignore-case | |
| 657 (if (symbolp current-language-environment) | |
| 658 (symbol-name current-language-environment) | |
| 659 current-language-environment) | |
| 660 language-info-alist)) | |
| 661 "English")) | |
| 662 :type (cons 'choice (mapcar (lambda (lang) | |
| 663 (list 'const (car lang))) | |
| 664 language-info-alist)) | |
| 665 :initialize 'custom-initialize-default | |
| 666 :group 'mule | |
| 667 :type 'string) | |
| 668 | |
| 669 (defun set-language-environment (language-name) | |
| 670 "Set up multi-lingual environment for using LANGUAGE-NAME. | |
| 771 | 671 This sets the coding system autodetection priority, the default buffer |
| 672 coding system, the default input method, the system locale, and other | |
| 673 relevant language properties. LANGUAGE-NAME should be a string, the | |
| 674 name of a language environment. For example, \"Latin-1\" specifies | |
| 675 the language environment for the major languages of Western Europe." | |
| 428 | 676 (interactive (list (read-language-name |
| 677 nil | |
| 678 "Set language environment (default, English): "))) | |
| 679 (if language-name | |
| 680 (if (symbolp language-name) | |
| 681 (setq language-name (symbol-name language-name))) | |
| 682 (setq language-name "English")) | |
| 683 (or (assoc-ignore-case language-name language-info-alist) | |
| 771 | 684 (error 'invalid-argument "Language environment not defined" |
| 685 language-name)) | |
| 428 | 686 (if current-language-environment |
| 687 (let ((func (get-language-info current-language-environment | |
| 688 'exit-function))) | |
| 689 (run-hooks 'exit-language-environment-hook) | |
| 4145 | 690 (if (functionp func) (funcall func)))) |
| 771 | 691 (setq current-language-environment language-name) |
| 428 | 692 (let ((default-eol-type (coding-system-eol-type |
| 693 default-buffer-file-coding-system))) | |
| 771 | 694 (reset-coding-categories-to-default) |
| 695 (set-locale-for-language-environment language-name) | |
| 696 (set-language-environment-coding-systems language-name default-eol-type)) | |
| 697 | |
| 698 (finish-set-language-environment language-name)) | |
| 428 | 699 |
| 771 | 700 (defun finish-set-language-environment (language-name) |
| 701 ;; Internal function. Only what's here is called at startup, once the | |
| 702 ;; first language environment is determined. The above stuff was already | |
| 703 ;; taken care of very early in the startup sequence, in a special | |
| 704 ;; fashion. | |
| 428 | 705 (let ((input-method (get-language-info language-name 'input-method))) |
| 706 (when input-method | |
| 707 (setq default-input-method input-method) | |
| 708 (if input-method-history | |
| 709 (setq input-method-history | |
| 710 (cons input-method | |
| 711 (delete input-method input-method-history)))))) | |
| 712 ;; (let ((nonascii (get-language-info language-name 'nonascii-translation)) | |
| 713 ;; (dos-table | |
| 714 ;; (if (eq window-system 'pc) | |
| 715 ;; (intern | |
| 716 ;; (concat "cp" dos-codepage "-nonascii-translation-table"))))) | |
| 717 ;; (cond | |
| 718 ;; ((char-table-p nonascii) | |
| 719 ;; (setq nonascii-translation-table nonascii)) | |
| 720 ;; ((and (eq window-system 'pc) (boundp dos-table)) | |
| 721 ;; ;; DOS terminals' default is to use a special non-ASCII translation | |
| 722 ;; ;; table as appropriate for the installed codepage. | |
| 723 ;; (setq nonascii-translation-table (symbol-value dos-table))) | |
| 724 ;; ((charsetp nonascii) | |
| 725 ;; (setq nonascii-insert-offset (- (make-char nonascii) 128))))) | |
| 726 | |
| 727 ;; (setq charset-origin-alist | |
| 728 ;; (get-language-info language-name 'charset-origin-alist)) | |
| 729 | |
| 730 ;; Unibyte setups if necessary. | |
| 731 ;; (unless default-enable-multibyte-characters | |
| 732 ;; ;; Syntax and case table. | |
| 733 ;; (let ((syntax (get-language-info language-name 'unibyte-syntax))) | |
| 734 ;; (if syntax | |
| 735 ;; (let ((set-case-syntax-set-multibyte nil)) | |
| 736 ;; (load syntax nil t)) | |
| 737 ;; ;; No information for syntax and case. Reset to the defaults. | |
| 738 ;; (let ((syntax-table (standard-syntax-table)) | |
| 739 ;; (case-table (standard-case-table)) | |
| 740 ;; (ch (if (eq window-system 'pc) 128 160))) | |
| 741 ;; (while (< ch 256) | |
| 742 ;; (modify-syntax-entry ch " " syntax-table) | |
| 743 ;; (aset case-table ch ch) | |
| 744 ;; (setq ch (1+ ch))) | |
| 745 ;; (set-char-table-extra-slot case-table 0 nil) | |
| 746 ;; (set-char-table-extra-slot case-table 1 nil) | |
| 747 ;; (set-char-table-extra-slot case-table 2 nil)) | |
| 748 ;; (set-standard-case-table (standard-case-table)) | |
| 749 ;; (let ((list (buffer-list))) | |
| 750 ;; (while list | |
| 751 ;; (with-current-buffer (car list) | |
| 752 ;; (set-case-table (standard-case-table))) | |
| 753 ;; (setq list (cdr list)))))) | |
| 754 ;; ;; Display table and coding system for terminal. | |
| 755 ;; (let ((coding (get-language-info language-name 'unibyte-display))) | |
| 756 ;; (if coding | |
| 757 ;; (standard-display-european-internal) | |
| 758 ;; (standard-display-default (if (eq window-system 'pc) 128 160) 255) | |
| 759 ;; (aset standard-display-table 146 nil)) | |
| 760 ;; (or (eq window-system 'pc) | |
| 761 ;; (set-terminal-coding-system coding)))) | |
| 762 | |
| 763 (let ((required-features (get-language-info language-name 'features))) | |
| 764 (while required-features | |
| 765 (require (car required-features)) | |
| 766 (setq required-features (cdr required-features)))) | |
| 767 (let ((func (get-language-info language-name 'setup-function))) | |
| 4145 | 768 (if (functionp func) |
| 428 | 769 (funcall func))) |
| 952 | 770 |
|
4490
67fbcaf3dbdc
error-sequence -> invalid-sequence
Aidan Kehoe <kehoea@parhasard.net>
parents:
4489
diff
changeset
|
771 (let ((invalid-sequence-coding-system |
|
67fbcaf3dbdc
error-sequence -> invalid-sequence
Aidan Kehoe <kehoea@parhasard.net>
parents:
4489
diff
changeset
|
772 (get-language-info language-name 'invalid-sequence-coding-system)) |
|
4489
b75b075a9041
Support displaying invalid UTF-8 in language-environment-specific ways.
Aidan Kehoe <kehoea@parhasard.net>
parents:
4468
diff
changeset
|
773 (disp-table (specifier-instance current-display-table)) |
|
4590
c83cab5a4f04
Display control characters using carets, invalid-sequence-coding-system
Aidan Kehoe <kehoea@parhasard.net>
parents:
4576
diff
changeset
|
774 glyph string) |
|
4490
67fbcaf3dbdc
error-sequence -> invalid-sequence
Aidan Kehoe <kehoea@parhasard.net>
parents:
4489
diff
changeset
|
775 (when (consp invalid-sequence-coding-system) |
|
67fbcaf3dbdc
error-sequence -> invalid-sequence
Aidan Kehoe <kehoea@parhasard.net>
parents:
4489
diff
changeset
|
776 (setq invalid-sequence-coding-system |
|
67fbcaf3dbdc
error-sequence -> invalid-sequence
Aidan Kehoe <kehoea@parhasard.net>
parents:
4489
diff
changeset
|
777 (car invalid-sequence-coding-system))) |
|
4489
b75b075a9041
Support displaying invalid UTF-8 in language-environment-specific ways.
Aidan Kehoe <kehoea@parhasard.net>
parents:
4468
diff
changeset
|
778 (map-char-table |
|
b75b075a9041
Support displaying invalid UTF-8 in language-environment-specific ways.
Aidan Kehoe <kehoea@parhasard.net>
parents:
4468
diff
changeset
|
779 #'(lambda (key entry) |
|
4590
c83cab5a4f04
Display control characters using carets, invalid-sequence-coding-system
Aidan Kehoe <kehoea@parhasard.net>
parents:
4576
diff
changeset
|
780 (setq string (decode-coding-string (string entry) |
|
c83cab5a4f04
Display control characters using carets, invalid-sequence-coding-system
Aidan Kehoe <kehoea@parhasard.net>
parents:
4576
diff
changeset
|
781 invalid-sequence-coding-system)) |
|
c83cab5a4f04
Display control characters using carets, invalid-sequence-coding-system
Aidan Kehoe <kehoea@parhasard.net>
parents:
4576
diff
changeset
|
782 ;; Treat control characters specially: |
|
c83cab5a4f04
Display control characters using carets, invalid-sequence-coding-system
Aidan Kehoe <kehoea@parhasard.net>
parents:
4576
diff
changeset
|
783 (when (string-match "^[\x00-\x1f\x80-\x9f]$" string) |
|
c83cab5a4f04
Display control characters using carets, invalid-sequence-coding-system
Aidan Kehoe <kehoea@parhasard.net>
parents:
4576
diff
changeset
|
784 (setq string (format "^%c" (+ ?@ (aref string 0))))) |
|
c83cab5a4f04
Display control characters using carets, invalid-sequence-coding-system
Aidan Kehoe <kehoea@parhasard.net>
parents:
4576
diff
changeset
|
785 (setq glyph (make-glyph (vector 'string :data string))) |
|
4490
67fbcaf3dbdc
error-sequence -> invalid-sequence
Aidan Kehoe <kehoea@parhasard.net>
parents:
4489
diff
changeset
|
786 (set-glyph-face glyph 'unicode-invalid-sequence-warning-face) |
|
4489
b75b075a9041
Support displaying invalid UTF-8 in language-environment-specific ways.
Aidan Kehoe <kehoea@parhasard.net>
parents:
4468
diff
changeset
|
787 (put-char-table key glyph disp-table) |
|
b75b075a9041
Support displaying invalid UTF-8 in language-environment-specific ways.
Aidan Kehoe <kehoea@parhasard.net>
parents:
4468
diff
changeset
|
788 nil) |
|
b75b075a9041
Support displaying invalid UTF-8 in language-environment-specific ways.
Aidan Kehoe <kehoea@parhasard.net>
parents:
4468
diff
changeset
|
789 unicode-error-default-translation-table)) |
|
b75b075a9041
Support displaying invalid UTF-8 in language-environment-specific ways.
Aidan Kehoe <kehoea@parhasard.net>
parents:
4468
diff
changeset
|
790 |
| 952 | 791 ;; Fit the charsets preferences in unicode conversions for the |
| 792 ;; language environment. | |
| 1285 | 793 (set-language-unicode-precedence-list |
| 794 (get-language-info language-name 'charset)) | |
| 952 | 795 |
| 428 | 796 (run-hooks 'set-language-environment-hook) |
| 797 (force-mode-line-update t)) | |
| 798 | |
| 799 ;; (defun standard-display-european-internal () | |
| 800 ;; ;; Actually set up direct output of non-ASCII characters. | |
| 801 ;; (standard-display-8bit (if (eq window-system 'pc) 128 160) 255) | |
| 802 ;; ;; Unibyte Emacs on MS-DOS wants to display all 8-bit characters with | |
| 803 ;; ;; the native font, and codes 160 and 146 stand for something very | |
| 804 ;; ;; different there. | |
| 805 ;; (or (and (eq window-system 'pc) (not default-enable-multibyte-characters)) | |
| 806 ;; (progn | |
| 807 ;; ;; Make non-line-break space display as a plain space. | |
| 808 ;; ;; Most X fonts do the wrong thing for code 160. | |
| 809 ;; (aset standard-display-table 160 [32]) | |
| 810 ;; ;; Most Windows programs send out apostrophe's as \222. Most X fonts | |
| 811 ;; ;; don't contain a character at that position. Map it to the ASCII | |
| 812 ;; ;; apostrophe. | |
| 813 ;; (aset standard-display-table 146 [39])))) | |
| 814 | |
| 771 | 815 ;; bogus FSF function describe-specified-language-support. |
| 428 | 816 |
| 817 (defun describe-language-environment (language-name) | |
| 818 "Describe how Emacs supports language environment LANGUAGE-NAME." | |
| 819 (interactive | |
| 820 (list (read-language-name | |
| 821 'documentation | |
| 822 "Describe language environment (default, current choice): "))) | |
| 823 (if (null language-name) | |
| 824 (setq language-name current-language-environment)) | |
| 825 (if (or (null language-name) | |
| 826 (null (get-language-info language-name 'documentation))) | |
| 827 (error "No documentation for the specified language")) | |
| 828 (if (symbolp language-name) | |
| 829 (setq language-name (symbol-name language-name))) | |
| 830 (let ((doc (get-language-info language-name 'documentation))) | |
| 771 | 831 (flet ((princ-list (&rest args) |
| 832 (while args (princ (car args)) (setq args (cdr args))) | |
| 833 (princ "\n"))) | |
| 834 (with-output-to-temp-buffer "*Help*" | |
| 835 (princ-list language-name " language environment" "\n") | |
| 836 (if (stringp doc) | |
| 428 | 837 (progn |
| 771 | 838 (princ-list doc) |
| 839 (terpri))) | |
| 840 (let ((str (get-language-info language-name 'sample-text))) | |
| 841 (if (stringp str) | |
| 842 (progn | |
| 843 (princ "Sample text:\n") | |
| 844 (princ-list " " str) | |
| 845 (terpri)))) | |
| 846 (let ((input-method (get-language-info language-name 'input-method)) | |
| 847 (l (copy-sequence input-method-alist))) | |
| 848 (princ "Input methods") | |
| 849 (when input-method | |
| 850 (princ (format " (default, %s)" input-method)) | |
| 851 (setq input-method (assoc input-method input-method-alist)) | |
| 852 (setq l (cons input-method (delete input-method l)))) | |
| 853 (princ ":\n") | |
| 428 | 854 (while l |
| 771 | 855 (if (string= language-name (nth 1 (car l))) |
| 856 (princ-list " " (car (car l)) | |
| 857 (format " (`%s' in mode line)" (nth 3 (car l))))) | |
| 858 (setq l (cdr l)))) | |
| 859 (terpri) | |
| 860 (princ "Character sets:\n") | |
| 861 (let ((l (get-language-info language-name 'charset))) | |
| 862 (if (null l) | |
| 863 (princ-list " nothing specific to " language-name) | |
| 864 (while l | |
| 865 (princ-list " " (car l) ": " | |
| 866 (charset-description (car l))) | |
| 867 (setq l (cdr l))))) | |
| 868 (terpri) | |
| 869 (princ "Coding systems:\n") | |
| 870 (let ((l (get-language-info language-name 'coding-system))) | |
| 871 (if (null l) | |
| 872 (princ-list " nothing specific to " language-name) | |
| 873 (while l | |
| 874 (princ ; (format " %s (`%c' in mode line):\n\t%s\n" | |
| 875 ;; In XEmacs, `coding-system-mnemonic' returns string. | |
| 876 (format " %s (`%s' in English, `%s' in mode line):\n\t%s\n" | |
| 877 (car l) | |
| 878 (coding-system-description (car l)) | |
| 879 (coding-system-mnemonic (car l)) | |
| 880 (or (coding-system-documentation (car l)) | |
| 881 "Not documented.")) ) | |
| 882 ;; (let ((aliases (coding-system-get (car l) 'alias-coding-systems))) | |
| 883 ;; (when aliases | |
| 884 ;; (princ "\t") | |
| 885 ;; (princ (cons 'alias: (cdr aliases))) | |
| 886 ;; (terpri))) | |
| 887 (setq l (cdr l))))))))) | |
| 428 | 888 |
| 889 ;;; Charset property | |
| 890 | |
| 891 ;; (defsubst get-charset-property (charset propname) | |
| 892 ;; "Return the value of CHARSET's PROPNAME property. | |
| 893 ;; This is the last value stored with | |
| 894 ;; `(put-charset-property CHARSET PROPNAME VALUE)'." | |
| 895 ;; (plist-get (charset-plist charset) propname)) | |
| 896 | |
| 897 ;; (defsubst put-charset-property (charset propname value) | |
| 898 ;; "Store CHARSETS's PROPNAME property with value VALUE. | |
| 899 ;; It can be retrieved with `(get-charset-property CHARSET PROPNAME)'." | |
| 900 ;; (set-charset-plist charset | |
| 901 ;; (plist-put (charset-plist charset) propname value))) | |
| 902 | |
| 903 (defvar char-code-property-table | |
| 904 (make-char-table 'generic) | |
| 905 "Char-table containing a property list of each character code. | |
| 906 | |
| 907 See also the documentation of `get-char-code-property' and | |
| 908 `put-char-code-property'") | |
| 909 ;; (let ((plist (aref char-code-property-table char))) | |
| 910 (defun get-char-code-property (char propname) | |
| 911 "Return the value of CHAR's PROPNAME property in `char-code-property-table'." | |
| 912 (let ((plist (get-char-table char char-code-property-table))) | |
| 913 (if (listp plist) | |
| 914 (car (cdr (memq propname plist)))))) | |
| 915 | |
| 916 (defun put-char-code-property (char propname value) | |
| 917 "Store CHAR's PROPNAME property with VALUE in `char-code-property-table'. | |
| 918 It can be retrieved with `(get-char-code-property CHAR PROPNAME)'." | |
| 919 (let ((plist (get-char-table char char-code-property-table))) | |
| 920 (if plist | |
| 921 (let ((slot (memq propname plist))) | |
| 922 (if slot | |
| 923 (setcar (cdr slot) value) | |
| 924 (nconc plist (list propname value)))) | |
| 925 (put-char-table char (list propname value) char-code-property-table) | |
| 926 ))) | |
| 927 | |
| 928 | |
| 929 ;; Pretty description of encoded string | |
| 930 | |
| 931 ;; Alist of ISO 2022 control code vs the corresponding mnemonic string. | |
|
4468
a78d697ccd2c
Import and extend GNU's descr-text.el, supporting prefix argument for C-x =
Aidan Kehoe <kehoea@parhasard.net>
parents:
4292
diff
changeset
|
932 (defvar iso-2022-control-alist |
|
a78d697ccd2c
Import and extend GNU's descr-text.el, supporting prefix argument for C-x =
Aidan Kehoe <kehoea@parhasard.net>
parents:
4292
diff
changeset
|
933 '((?\x1b . "ESC") |
|
a78d697ccd2c
Import and extend GNU's descr-text.el, supporting prefix argument for C-x =
Aidan Kehoe <kehoea@parhasard.net>
parents:
4292
diff
changeset
|
934 (?\x0e . "SO") |
|
a78d697ccd2c
Import and extend GNU's descr-text.el, supporting prefix argument for C-x =
Aidan Kehoe <kehoea@parhasard.net>
parents:
4292
diff
changeset
|
935 (?\x0f . "SI") |
|
a78d697ccd2c
Import and extend GNU's descr-text.el, supporting prefix argument for C-x =
Aidan Kehoe <kehoea@parhasard.net>
parents:
4292
diff
changeset
|
936 (?\x8e . "SS2") |
|
a78d697ccd2c
Import and extend GNU's descr-text.el, supporting prefix argument for C-x =
Aidan Kehoe <kehoea@parhasard.net>
parents:
4292
diff
changeset
|
937 (?\x8f . "SS3") |
|
a78d697ccd2c
Import and extend GNU's descr-text.el, supporting prefix argument for C-x =
Aidan Kehoe <kehoea@parhasard.net>
parents:
4292
diff
changeset
|
938 (?\x9b . "CSI"))) |
| 428 | 939 |
|
4468
a78d697ccd2c
Import and extend GNU's descr-text.el, supporting prefix argument for C-x =
Aidan Kehoe <kehoea@parhasard.net>
parents:
4292
diff
changeset
|
940 (defun encoded-string-description (str coding-system) |
|
a78d697ccd2c
Import and extend GNU's descr-text.el, supporting prefix argument for C-x =
Aidan Kehoe <kehoea@parhasard.net>
parents:
4292
diff
changeset
|
941 "Return a pretty description of STR that is encoded by CODING-SYSTEM." |
|
a78d697ccd2c
Import and extend GNU's descr-text.el, supporting prefix argument for C-x =
Aidan Kehoe <kehoea@parhasard.net>
parents:
4292
diff
changeset
|
942 ; (setq str (string-as-unibyte str)) |
|
a78d697ccd2c
Import and extend GNU's descr-text.el, supporting prefix argument for C-x =
Aidan Kehoe <kehoea@parhasard.net>
parents:
4292
diff
changeset
|
943 (mapconcat |
|
a78d697ccd2c
Import and extend GNU's descr-text.el, supporting prefix argument for C-x =
Aidan Kehoe <kehoea@parhasard.net>
parents:
4292
diff
changeset
|
944 (if (and coding-system (eq (coding-system-type coding-system) 'iso2022)) |
|
a78d697ccd2c
Import and extend GNU's descr-text.el, supporting prefix argument for C-x =
Aidan Kehoe <kehoea@parhasard.net>
parents:
4292
diff
changeset
|
945 ;; Try to get a pretty description for ISO 2022 escape sequences. |
|
a78d697ccd2c
Import and extend GNU's descr-text.el, supporting prefix argument for C-x =
Aidan Kehoe <kehoea@parhasard.net>
parents:
4292
diff
changeset
|
946 (function (lambda (x) (or (cdr (assq x iso-2022-control-alist)) |
|
a78d697ccd2c
Import and extend GNU's descr-text.el, supporting prefix argument for C-x =
Aidan Kehoe <kehoea@parhasard.net>
parents:
4292
diff
changeset
|
947 (format "#x%02X" x)))) |
|
a78d697ccd2c
Import and extend GNU's descr-text.el, supporting prefix argument for C-x =
Aidan Kehoe <kehoea@parhasard.net>
parents:
4292
diff
changeset
|
948 (function (lambda (x) (format "#x%02X" x)))) |
|
a78d697ccd2c
Import and extend GNU's descr-text.el, supporting prefix argument for C-x =
Aidan Kehoe <kehoea@parhasard.net>
parents:
4292
diff
changeset
|
949 str " ")) |
| 428 | 950 |
| 951 ;; (defun encode-coding-char (char coding-system) | |
| 952 ;; "Encode CHAR by CODING-SYSTEM and return the resulting string. | |
| 953 ;; If CODING-SYSTEM can't safely encode CHAR, return nil." | |
| 954 ;; (if (cmpcharp char) | |
| 955 ;; (setq char (car (decompose-composite-char char 'list)))) | |
| 956 ;; (let ((str1 (char-to-string char)) | |
| 957 ;; (str2 (make-string 2 char)) | |
| 958 ;; (safe-charsets (and coding-system | |
| 959 ;; (coding-system-get coding-system 'safe-charsets))) | |
| 960 ;; enc1 enc2 i1 i2) | |
| 961 ;; (when (or (eq safe-charsets t) | |
| 962 ;; (memq (char-charset char) safe-charsets)) | |
| 963 ;; ;; We must find the encoded string of CHAR. But, just encoding | |
| 964 ;; ;; CHAR will put extra control sequences (usually to designate | |
| 440 | 965 ;; ;; ASCII charset) at the tail if type of CODING is ISO 2022. |
| 428 | 966 ;; ;; To exclude such tailing bytes, we at first encode one-char |
| 967 ;; ;; string and two-char string, then check how many bytes at the | |
| 968 ;; ;; tail of both encoded strings are the same. | |
| 969 ;; | |
| 970 ;; (setq enc1 (string-as-unibyte (encode-coding-string str1 coding-system)) | |
| 971 ;; i1 (length enc1) | |
| 972 ;; enc2 (string-as-unibyte (encode-coding-string str2 coding-system)) | |
| 973 ;; i2 (length enc2)) | |
| 974 ;; (while (and (> i1 0) (= (aref enc1 (1- i1)) (aref enc2 (1- i2)))) | |
| 975 ;; (setq i1 (1- i1) i2 (1- i2))) | |
| 976 ;; | |
| 977 ;; ;; Now (substring enc1 i1) and (substring enc2 i2) are the same, | |
| 978 ;; ;; and they are the extra control sequences at the tail to | |
| 979 ;; ;; exclude. | |
| 980 ;; (substring enc2 0 i2)))) | |
| 981 | |
| 771 | 982 |
| 983 ;; #### The following section is utter junk from mule-misc.el. | |
| 984 ;; I've deleted everything that's not referenced in mule-packages and | |
| 985 ;; not in FSF 20.6; there's no point in keeping old namespace-polluting | |
| 986 ;; Mule 2.3 crap around. --ben | |
| 987 | |
| 988 (defvar self-insert-after-hook nil | |
| 989 "Hook to run when extended self insertion command exits. Should take | |
| 990 two arguments START and END corresponding to character position.") | |
| 991 | |
| 992 (make-variable-buffer-local 'self-insert-after-hook) | |
| 993 | |
| 994 (defun delete-text-in-column (from to) | |
| 995 "Delete the text between column FROM and TO (exclusive) of the current line. | |
| 996 Nil of FORM or TO means the current column. | |
| 997 | |
| 998 If there's a character across the borders, the character is replaced | |
| 999 with the same width of spaces before deleting." | |
| 1000 (save-excursion | |
| 1001 (let (p1 p2) | |
| 1002 (if from | |
| 1003 (progn | |
| 1004 (setq p1 (move-to-column from)) | |
| 1005 (if (> p1 from) | |
| 1006 (progn | |
| 1007 (delete-char -1) | |
| 1008 (insert-char ? (- p1 (current-column))) | |
| 1009 (forward-char (- from p1)))))) | |
| 1010 (setq p1 (point)) | |
| 1011 (if to | |
| 1012 (progn | |
| 1013 (setq p2 (move-to-column to)) | |
| 1014 (if (> p2 to) | |
| 1015 (progn | |
| 1016 (delete-char -1) | |
| 1017 (insert-char ? (- p2 (current-column))) | |
| 1018 (forward-char (- to p2)))))) | |
| 1019 (setq p2 (point)) | |
| 1020 (delete-region p1 p2)))) | |
| 1021 | |
| 1022 (defun cancel-undo-boundary () | |
| 1023 "Cancel undo boundary." | |
| 1024 (if (and (consp buffer-undo-list) | |
| 1025 (null (car buffer-undo-list))) | |
| 1026 (setq buffer-undo-list (cdr buffer-undo-list)))) | |
| 1027 | |
| 1028 (defun define-egg-environment (env-sym doc-string enable-function) | |
| 1029 "Define a new language environment for egg, named by ENV-SYM. | |
| 1030 DOC-STRING should be a string describing the environment. | |
| 1031 ENABLE-FUNCTION should be a function of no arguments that will be called | |
| 1032 when the language environment is made current." | |
| 1033 (put env-sym 'egg-environ-doc-string doc-string) | |
| 1034 (put env-sym 'set-egg-environ enable-function)) | |
| 1035 | |
| 1036 | |
| 1037 ;; Init code. | |
| 1038 | |
| 1039 ;; auto-language-alist deleted. We have a more sophisticated system, | |
| 1040 ;; with the locales stored in the language data. | |
| 1041 | |
| 4145 | 1042 ;; Initialised in mule/general-late.el, which is compiled after all the |
| 1043 ;; language support--and, thus, minority Chinese coding systems and so | |
| 1044 ;; on--has been loaded. | |
| 3707 | 1045 (defvar posix-charset-to-coding-system-hash nil |
| 1046 "A map from the POSIX locale charset versions of the defined coding | |
| 1047 systems' names, with all alpha-numeric characters removed, to the actual | |
| 1048 coding system names. Used at startup when working out which coding system | |
| 1049 should be the default for the locale. ") | |
| 1050 | |
| 1051 (defun parse-posix-locale-string (locale-string) | |
| 1052 "Return values \(LANGUAGE REGION CHARSET MODIFIERS\) given LOCALE-STRING. | |
| 1053 | |
| 1054 LOCALE-STRING should be a POSIX locale. If it cannot be parsed as such, this | |
| 1055 function returns nil. " | |
| 1056 (let (language region charset modifiers locinfo) | |
| 1057 (setq locale-string (downcase locale-string)) | |
| 1058 (cond ((string-match | |
| 1059 #r"^\([a-z0-9]\{2,2\}\)\(_[a-z0-9]\{2,2\}\)?\(\.[^@]*\)?\(@.*\)?$" | |
| 1060 locale-string) | |
| 1061 (setq language (match-string 1 locale-string) | |
| 1062 region (match-string 2 locale-string) | |
| 1063 charset (match-string 3 locale-string) | |
| 1064 modifiers (match-string 4 locale-string) | |
| 1065 region (and region (replace-in-string region "^_" "")) | |
| 1066 charset (and charset (replace-in-string charset #r"^\." "")) | |
| 1067 modifiers (and modifiers | |
| 1068 (replace-in-string modifiers "^@" ""))) | |
| 1069 (when (and modifiers (equal modifiers "euro") (null charset)) | |
| 1070 ;; Not ideal for Latvian, say, but I don't have any locales | |
| 1071 ;; where the @euro modifier doesn't mean ISO-8859-15 in the 956 | |
| 1072 ;; I have. | |
| 1073 (setq charset "iso-8859-15")) | |
| 4073 | 1074 (when (and modifiers (equal modifiers "cyrillic") (null charset)) |
| 1075 ;; Feedback wanted! | |
| 1076 (setq charset "windows-1251")) | |
| 3707 | 1077 (values language region charset modifiers)) |
| 1078 ((and (string-match "^[a-z0-9]+$" locale-string) | |
| 1079 (assoc-ignore-case locale-string language-info-alist)) | |
| 1080 (setq language (get-language-info locale-string 'locale) | |
| 1081 language (if (listp language) (car language) language)) | |
| 1082 (values language region charset modifiers)) | |
| 1083 ((string-match #r"^\([a-z0-9]+\)\.\([a-z0-9]+\)$" locale-string) | |
| 1084 (when (assoc-ignore-case | |
| 1085 (setq locinfo (match-string 1 locale-string)) | |
| 1086 language-info-alist) | |
| 1087 (setq language (get-language-info locinfo 'locale) | |
| 1088 language (if (listp language) (car language) language))) | |
| 1089 (setq charset (match-string 2 locale-string)) | |
| 1090 (values language region charset modifiers))))) | |
| 1091 | |
| 1092 (defun create-variant-language-environment (langenv coding-system) | |
| 1093 "Create a variant of LANGENV with CODING-SYSTEM as its coding systems. | |
| 1094 | |
| 1095 The coding systems in question are those described in the | |
| 1096 `set-language-info' docstring with the property names of | |
| 1097 `native-coding-system' and `coding-system'. The name of the new language | |
| 1098 environment is the name of the old language environment, followed by | |
| 1099 CODING-SYSTEM in parentheses. Returns the name of the new language | |
| 3767 | 1100 environment. |
| 1101 | |
| 1102 This function also modifies the `coding-priority' of a language | |
| 3707 | 1103 environment. " |
| 1104 (check-coding-system coding-system) | |
| 1105 (if (symbolp langenv) (setq langenv (symbol-name langenv))) | |
| 1106 (unless (setq langenv | |
| 1107 (assoc-ignore-case langenv language-info-alist)) | |
| 1108 (error 'wrong-type-argument "Not a known language environment")) | |
| 1109 (set-language-info-alist | |
| 1110 (if (string-match " ([^)]+)$" (car langenv)) | |
| 1111 (replace-match (format " (%s)" | |
| 1112 (upcase (symbol-name | |
| 1113 (coding-system-name coding-system)))) | |
| 3737 | 1114 nil nil (car langenv)) |
| 3707 | 1115 (format "%s (%s)" (car langenv) |
| 1116 (upcase (symbol-name (coding-system-name coding-system))))) | |
| 1117 (destructive-plist-to-alist | |
| 3767 | 1118 (plist-put |
| 1119 (plist-put | |
| 4292 | 1120 (plist-put |
| 1121 (plist-put (alist-to-plist (cdr langenv)) 'native-coding-system | |
| 1122 coding-system) | |
| 1123 'coding-system (cons coding-system | |
| 1124 (cdr (assoc 'coding-system (cdr langenv))))) | |
| 1125 'coding-priority (cons coding-system | |
| 1126 (cdr (assq 'coding-priority (cdr langenv))))) | |
| 1127 ;; The tutorial coding system is important; otherwise the tutorial file | |
| 1128 ;; gets loaded in the variant coding system. | |
| 1129 'tutorial-coding-system | |
| 1130 (or (car-safe (cdr-safe (assoc 'tutorial-coding-system (cdr langenv)))) | |
| 1131 (car-safe (cdr-safe (assoc 'coding-system (cdr langenv))))))))) | |
| 801 | 1132 |
| 771 | 1133 (defun get-language-environment-from-locale (locale) |
| 1134 "Convert LOCALE into a language environment. | |
| 1135 LOCALE is a C library locale string, as returned by `current-locale'. | |
| 1136 Uses the `locale' property of the language environment." | |
| 3707 | 1137 (block langenv |
| 4103 | 1138 (multiple-value-bind (language ignored-arg charset ignored-arg) |
| 3707 | 1139 (parse-posix-locale-string locale) |
| 1140 (let ((case-fold-search t) | |
| 1141 (desired-coding-system | |
| 1142 (and charset (gethash (replace-in-string charset "[^a-z0-9]" "") | |
| 1143 posix-charset-to-coding-system-hash))) | |
| 3767 | 1144 lang locs given-coding-system) |
| 3707 | 1145 (dolist (langcons language-info-alist) |
| 1146 (setq lang (car langcons) | |
| 1147 locs (get-language-info lang 'locale)) | |
| 1148 (dolist (loc (if (listp locs) locs (list locs))) | |
| 1149 (cond ((functionp loc) | |
| 1150 (if (funcall loc locale) | |
| 1151 (return-from langenv lang))) | |
| 1152 ((stringp loc) | |
| 1153 (when (or (equal loc language) | |
| 1154 (string-match | |
| 1155 (format "^%s\\([^A-Za-z0-9]\\|$\\)" loc) | |
| 1156 locale)) | |
| 1157 (if (or (null desired-coding-system) | |
| 1158 (and desired-coding-system | |
| 3767 | 1159 (or (eq desired-coding-system |
| 1160 (setq given-coding-system | |
| 1161 (get-language-info | |
| 1162 lang | |
| 1163 'native-coding-system))) | |
| 1164 (and (listp given-coding-system) | |
| 1165 (memq desired-coding-system | |
| 1166 given-coding-system))))) | |
| 3707 | 1167 (return-from langenv lang) |
| 1168 (return-from langenv | |
| 1169 (create-variant-language-environment | |
| 1170 lang desired-coding-system)))))))))))) | |
| 771 | 1171 |
| 1172 (defun mswindows-get-language-environment-from-locale (ms-locale) | |
| 1173 "Convert MS-LOCALE (an MS Windows locale) into a language environment. | |
| 1174 MS-LOCALE is in the format recognized by `set-mswindows-current-locale' -- | |
| 1175 i.e. a language string or a cons (LANG . SUBLANG). Note: This is NOT the | |
| 1176 same as the C library locale format (see `set-current-locale')! | |
| 1177 | |
| 1178 This looks up the `mswindows-locale' property of all language environments; | |
| 1179 if nothing matching is found, it looks for a language environment with the | |
| 1180 same name (modulo case differences) as the LANG part of the locale." | |
| 1181 (or (consp ms-locale) (setq ms-locale (cons ms-locale "DEFAULT"))) | |
| 1182 (or (block langenv | |
| 1183 (dolist (langcons language-info-alist) | |
| 1184 (let* ((lang (car langcons)) | |
| 1185 (mswlocs (get-language-info lang 'mswindows-locale)) | |
| 1186 (mswlocs (if (and (consp mswlocs) | |
| 1187 (listp (cdr mswlocs))) | |
| 1188 mswlocs (list mswlocs)))) | |
| 1189 (dolist (loc mswlocs) | |
| 1190 (or (consp loc) (setq loc (cons loc "DEFAULT"))) | |
| 1191 (if (equalp loc ms-locale) | |
| 1192 (return-from langenv lang)))))) | |
| 1193 (dolist (langcons language-info-alist) | |
| 1194 (let* ((lang (car langcons))) | |
| 1195 (if (equalp lang (car ms-locale)) | |
| 1196 (return-from nil lang)))))) | |
| 1197 | |
| 1198 (defun get-native-coding-system-from-language-environment (langenv locale) | |
| 1199 "Return the native coding system appropriate for LANGENV. | |
| 1200 LANGENV is a string naming a language environment. May use the LOCALE | |
| 1201 \(which should be the C library LOCALE corresponding to LANGENV) to | |
| 1202 determine the correct coding system. (For example, in the Japanese language | |
| 1203 environment, there are multiple encodings in use: euc-jp, shift-jis, jis7, | |
| 1204 jis8, iso-2022-jp, etc. The LOCALE may tell which one is correct.) | |
| 1205 | |
| 1206 Specifically: Under X, the returned value is determined from these two. | |
| 1207 Under MS Windows, the native coding system must be set from the default | |
| 1208 system locale and is not influenced by LOCALE. (In other words, a program | |
| 1209 can't set the text encoding used to communicate with the OS. To get around | |
| 1210 this, we use Unicode whenever available, i.e. on Windows NT always and on | |
| 1318 | 1211 Windows 9x whenever a Unicode version of a system call is available.)" |
| 771 | 1212 (if (eq system-type 'windows-nt) |
| 1213 ;; should not apply to Cygwin, I don't think | |
| 1214 'mswindows-multibyte-system-default | |
| 1215 (let ((ncod (get-language-info langenv 'native-coding-system))) | |
| 1216 (if (or (functionp ncod) (not (listp ncod))) | |
| 1217 (setq ncod (list ncod))) | |
| 1218 (let ((native | |
| 1219 (dolist (try-native ncod) | |
| 1220 (let ((result | |
| 1221 (if (functionp try-native) | |
| 1222 (funcall try-native locale) | |
| 1223 try-native))) | |
| 1224 (if result (return result)))))) | |
| 1225 (or native (car (get-language-info langenv 'coding-system)) | |
| 1226 'raw-text))))) | |
| 1227 | |
| 1228 (defun get-coding-system-from-locale (locale) | |
| 1229 "Return the coding system corresponding to a locale string." | |
| 1230 (get-native-coding-system-from-language-environment | |
| 1231 (get-language-environment-from-locale locale) locale)) | |
| 1232 | |
| 1233 (defvar mswindows-langenv-to-locale-table (make-hash-table) | |
| 1234 "Table mapping language environments to associated MS Windows locales. | |
| 1235 There may be more than one MS Windows locale that maps to a given language | |
| 1236 environment, so once we've made the mapping, we record it here when we need | |
| 1237 to make the reverse mapping. For example, all MS Windows locales with | |
| 1238 language ENGLISH will map to language environment English, and when the | |
| 1239 user starts up in such a locale, switches to another language environment | |
| 1240 and then back to English, we want the same locale again.") | |
| 1241 | |
| 1242 (defun set-locale-for-language-environment (langenv) | |
| 1243 "Sets the current system locale as appropriate for LANGENV. | |
| 1244 LANGENV is a language environment. The locale is determined by looking at | |
| 1245 the 'locale (or maybe 'mswindows-locale) property of LANGENV, and then | |
| 1246 setting it using `set-current-locale' and maybe also | |
| 1247 `mswindows-set-current-locale'. Also sets the LANG environment variable. | |
| 1248 Returns non-nil if successfully set the locale(s)." | |
| 1249 (flet ((mswindows-get-and-set-locale-from-langenv (langenv) | |
| 1250 ;; find the mswindows locale for the langenv, make it current, | |
| 1251 ;; and return it. first we check the langenv-to-locale table | |
| 1252 ;; ... | |
| 1253 (let ((ms-locale | |
| 1254 (gethash langenv mswindows-langenv-to-locale-table))) | |
| 1255 (if ms-locale (progn | |
| 778 | 1256 (declare-fboundp (mswindows-set-current-locale |
| 1257 ms-locale)) | |
| 771 | 1258 ms-locale) |
| 1259 ;; ... if not, see if the langenv specifies any locale(s). | |
| 1260 ;; if not, construct one from the langenv name. | |
| 1261 (let* ((mslocs (get-language-info langenv 'mswindows-locale)) | |
| 1262 (mslocs (or mslocs (cons (upcase langenv) "DEFAULT"))) | |
| 1263 (mslocs (if (and (consp mslocs) | |
| 1264 (listp (cdr mslocs))) | |
| 1265 mslocs (list mslocs)))) | |
| 1266 (dolist (msloc mslocs) | |
| 1267 ;; Sometimes a language with DEFAULT is different from | |
| 1268 ;; with SYS_DEFAULT, and on my system | |
| 1269 ;; (set-current-locale "chinese") is NOT the same as | |
| 1270 ;; (set-current-locale "chinese-default")! The latter | |
| 1271 ;; gives Taiwan (DEFAULT), the former PRC (SYS_DEFAULT). | |
| 1272 ;; In the interests of consistency, we always use DEFAULT. | |
| 1273 (or (consp msloc) (setq msloc (cons msloc "DEFAULT"))) | |
| 1274 (when (condition-case nil | |
| 1275 (progn | |
| 778 | 1276 (declare-fboundp (mswindows-set-current-locale |
| 1277 msloc)) | |
| 771 | 1278 t) |
| 1279 (error nil)) | |
| 1280 (return msloc)))))))) | |
| 1281 (if (eq system-type 'windows-nt) | |
| 1282 (let ((ms-locale (mswindows-get-and-set-locale-from-langenv langenv))) | |
| 1283 (when ms-locale | |
| 1284 ;; also need to set the clib locale. | |
| 1285 (or (set-current-locale | |
| 1286 ;; if the locale is '("DUTCH" . "DUTCH_BELGIAN"), | |
| 1287 ;; try "DUTCH-BELGIAN". (Case is insignificant; | |
| 1288 ;; "dutch-belgian" works just as well.) This type | |
| 1289 ;; of transformation should always work, and you | |
| 1290 ;; get back the canonicalized version -- in this | |
| 1291 ;; case "Dutch_Belgium.1252". Note the futility of | |
| 1292 ;; trying to construct "Belgium" directly from | |
| 1293 ;; "BELGIAN". | |
| 1294 ;; | |
| 1295 ;; BUT ... We actually have to be trickier. | |
| 1296 ;; ("SPANISH" . "SPANISH_DOMINICAN_REPUBLIC") needs | |
| 1297 ;; to end up as "SPANISH-DOMINICAN REPUBLIC"; any | |
| 1298 ;; other punctuation makes it fail (you either get | |
| 1299 ;; Spain for the country, or nil). | |
| 1300 ;; | |
| 1301 ;; assume it's DEFAULT or NEUTRAL (or something else | |
| 1302 ;; without the language in it?) and prepend the | |
| 1303 ;; language. | |
| 1304 (if (string-match "_" (cdr ms-locale)) | |
| 1305 (replace-in-string | |
| 1306 (replace-match "-" nil nil (cdr ms-locale)) "_" " ") | |
| 1307 (format "%s-%s" (car ms-locale) (cdr ms-locale)))) | |
| 1308 ;; ???? huh ???? if failure, just try the language | |
| 1309 ;; name. | |
| 1310 (set-current-locale (car ms-locale)))) | |
| 1311 ;; also set LANG, for the benefit of Cygwin subprocesses. | |
| 1312 (let* ((cygloc (or (get-language-info langenv 'cygwin-locale) | |
| 1313 (get-language-info langenv 'locale))) | |
| 1314 (cygloc (if (listp cygloc) (car (last cygloc)) cygloc))) | |
| 1315 (if (and cygloc (stringp cygloc)) (setenv "LANG" cygloc))) | |
| 1316 (not (null ms-locale))) | |
| 1317 | |
| 1318 ;; not MS Windows native. | |
| 1319 | |
| 1320 ;; Cygwin is as usual an unholy mixture -- C library locales | |
| 1321 ;; that follow Unix conventions, but also MS Windows locales. | |
| 1322 ;; So set the MS Windows locale, and then try to find a Unix | |
| 1323 ;; locale. | |
| 1324 (when (eq system-type 'cygwin32) | |
| 1325 (mswindows-get-and-set-locale-from-langenv langenv)) | |
| 1326 (let ((locs (get-language-info langenv 'locale))) | |
| 1327 (dolist (loc (if (listp locs) locs (list locs))) | |
| 1328 (let ((retval | |
| 1329 (cond ((functionp loc) (funcall loc nil)) | |
| 1330 ((stringp loc) (set-current-locale loc)) | |
| 1331 (t nil)))) | |
| 1332 (when retval | |
| 1333 (setenv "LANG" retval) | |
| 1334 (return t)))))))) | |
| 1335 | |
| 1336 (defun set-language-environment-coding-systems (language-name | |
| 1337 &optional eol-type) | |
| 1338 "Do various coding system setups for language environment LANGUAGE-NAME. | |
| 1339 This function assumes that the locale for LANGUAGE-NAME has been set using | |
| 1340 `set-current-locale'. | |
| 1341 | |
| 1342 The optional arg EOL-TYPE specifies the eol-type of the default value | |
| 1343 of buffer-file-coding-system set by this function." | |
| 1344 | |
| 1345 ;; The following appeared as the third paragraph of the doc string for this | |
| 778 | 1346 ;; function, but it's not in FSF 21.1, and it's not true, since we call |
| 771 | 1347 ;; reset-coding-categories-to-default before calling this function. #### |
| 1348 ;; Should we rethink this? | |
| 1349 | |
| 1350 ; Note that `coding-priority-list' is not reset first; thus changing language | |
| 1351 ; environment allows recognition of coding systems from previously set language | |
| 1352 ; environments. (This will not work if the desired coding systems are from the | |
| 1353 ; same category. E.g., starting with a Hebrew language environment, ISO 8859-8 | |
| 1354 ; will be recognized. If you shift to Russian, ISO 8859-8 will be shadowed by | |
| 1355 ; ISO 8859-5, and cannot be automatically recognized without resetting the | |
| 1356 ; language environment to Hebrew. However, if you shift from Japanese to | |
| 1357 ; Russian, ISO-2022-JP will continue to be automatically recognized, since | |
| 1358 ; ISO-8859-5 and ISO-2022-JP are different coding categories.)" | |
| 1359 | |
| 1360 (flet ((maybe-change-coding-system-with-eol (codesys eol-type) | |
| 1361 ;; if the EOL type specifies a specific type of ending, | |
| 1362 ;; then add that ending onto the given CODESYS; otherwise, | |
| 1363 ;; return CODESYS unchanged. | |
| 1364 (if (memq eol-type '(lf crlf cr unix dos mac)) | |
| 1365 (coding-system-change-eol-conversion codesys eol-type) | |
| 1366 codesys))) | |
| 1367 | |
| 1368 ;; initialize category mappings and priority list. | |
| 1369 (let* ((priority (get-language-info language-name 'coding-priority)) | |
| 1370 (default-coding (car priority))) | |
| 1371 (if priority | |
| 1372 (let ((categories (mapcar 'coding-system-category priority)) | |
| 1373 category checked-categories) | |
| 1374 (while priority | |
| 1375 (unless (memq (setq category (car categories)) checked-categories) | |
| 1376 (set-coding-category-system category (car priority)) | |
| 1377 (setq checked-categories (cons category checked-categories))) | |
| 1378 (setq priority (cdr priority) | |
| 1379 categories (cdr categories))) | |
| 1380 (set-coding-priority-list (nreverse checked-categories)) | |
| 1381 )) | |
| 1382 | |
| 1383 ;; set the default buffer coding system from the first element of the | |
| 1384 ;; list in the `coding-priority' property, under Unix. Under Windows, it | |
| 1385 ;; should stay at `mswindows-multibyte', which will reference the current | |
| 3707 | 1386 ;; code page. ([Does it really make sense to set the Unix default |
| 771 | 1387 ;; that way? NOTE also that it's not the same as the native coding |
| 1388 ;; system for the locale, which is correct -- the form we choose for text | |
| 3707 | 1389 ;; files should not necessarily have any relevance to whether we're in a |
| 1390 ;; Shift-JIS, EUC-JP, JIS, or other Japanese locale.]) | |
| 1391 ;; | |
| 1392 ;; On Unix--with the exception of Mac OS X--there is no way to | |
| 1393 ;; know for certain what coding system to use for file names, and | |
| 1394 ;; the environment is the best guess. If a particular user's | |
| 1395 ;; preferences differ from this, then that particular user needs | |
| 1396 ;; to edit ~/.xemacs/init.el. Aidan Kehoe, Sun Nov 26 18:11:31 CET | |
| 1397 ;; 2006. OS X uses an almost-normal-form version of UTF-8. | |
| 1398 | |
| 771 | 1399 (unless (memq system-type '(windows-nt cygwin32)) |
| 1400 (set-default-buffer-file-coding-system | |
| 1401 (maybe-change-coding-system-with-eol default-coding eol-type)))) | |
| 1402 ;; (setq default-sendmail-coding-system default-coding) | |
| 1403 | |
| 1318 | 1404 ;; set the native coding system and the default process-output system. |
| 771 | 1405 (let ((native (get-native-coding-system-from-language-environment |
| 1406 language-name (current-locale)))) | |
| 1318 | 1407 |
| 771 | 1408 (condition-case nil |
| 1318 | 1409 (define-coding-system-alias 'native |
| 1410 (maybe-change-coding-system-with-eol native eol-type)) | |
| 771 | 1411 (error |
| 1412 (warn "Invalid native-coding-system %s in language environment %s" | |
| 1413 native language-name))) | |
|
4576
774e5c7522bf
Preserve the relation btw. file-name-coding-system & the 'file-name c-s alias.
Aidan Kehoe <kehoea@parhasard.net>
parents:
4490
diff
changeset
|
1414 ;; These variables have magic handlers to make setting them equivalent |
|
774e5c7522bf
Preserve the relation btw. file-name-coding-system & the 'file-name c-s alias.
Aidan Kehoe <kehoea@parhasard.net>
parents:
4490
diff
changeset
|
1415 ;; to setting the file-name, terminal and keyboard coding system |
|
774e5c7522bf
Preserve the relation btw. file-name-coding-system & the 'file-name c-s alias.
Aidan Kehoe <kehoea@parhasard.net>
parents:
4490
diff
changeset
|
1416 ;; aliases. See coding.el. |
|
774e5c7522bf
Preserve the relation btw. file-name-coding-system & the 'file-name c-s alias.
Aidan Kehoe <kehoea@parhasard.net>
parents:
4490
diff
changeset
|
1417 (setq file-name-coding-system |
|
774e5c7522bf
Preserve the relation btw. file-name-coding-system & the 'file-name c-s alias.
Aidan Kehoe <kehoea@parhasard.net>
parents:
4490
diff
changeset
|
1418 (or |
|
774e5c7522bf
Preserve the relation btw. file-name-coding-system & the 'file-name c-s alias.
Aidan Kehoe <kehoea@parhasard.net>
parents:
4490
diff
changeset
|
1419 (let ((fncs (assq system-type system-type-file-name-coding))) |
|
774e5c7522bf
Preserve the relation btw. file-name-coding-system & the 'file-name c-s alias.
Aidan Kehoe <kehoea@parhasard.net>
parents:
4490
diff
changeset
|
1420 (and fncs (cdr fncs))) |
|
774e5c7522bf
Preserve the relation btw. file-name-coding-system & the 'file-name c-s alias.
Aidan Kehoe <kehoea@parhasard.net>
parents:
4490
diff
changeset
|
1421 native) |
|
774e5c7522bf
Preserve the relation btw. file-name-coding-system & the 'file-name c-s alias.
Aidan Kehoe <kehoea@parhasard.net>
parents:
4490
diff
changeset
|
1422 ;; Set the default keyboard and terminal coding systems to the |
|
774e5c7522bf
Preserve the relation btw. file-name-coding-system & the 'file-name c-s alias.
Aidan Kehoe <kehoea@parhasard.net>
parents:
4490
diff
changeset
|
1423 ;; native coding system of the language environment. |
|
774e5c7522bf
Preserve the relation btw. file-name-coding-system & the 'file-name c-s alias.
Aidan Kehoe <kehoea@parhasard.net>
parents:
4490
diff
changeset
|
1424 keyboard-coding-system native |
| 3142 | 1425 terminal-coding-system native) |
| 1426 | |
| 1427 ;; And do the same for any TTYs. | |
| 1428 (dolist (con (console-list)) | |
| 1429 (when (eq 'tty (device-type (car (console-device-list con)))) | |
| 1430 ;; Calling set-input-mode at the same time would be a sane thing | |
| 1431 ;; to do here. I would prefer to default to accepting eight bit | |
| 1432 ;; input and not using the top bit for Meta. | |
| 1433 (set-console-tty-coding-system con native))) | |
| 1434 | |
| 1318 | 1435 ;; process output should not have EOL conversion. under MS Windows |
| 1436 ;; and Cygwin, this screws things up (`cmd' is fine with just LF and | |
| 1437 ;; `bash' chokes on CR-LF). | |
| 771 | 1438 (setq default-process-coding-system |
| 1318 | 1439 (cons (car default-process-coding-system) native))))) |
| 771 | 1440 |
| 1441 (defun init-locale-at-early-startup () | |
| 1442 "Don't call this." | |
| 1443 ;; Called directly from the C code in intl.c, very early in the startup | |
| 1444 ;; sequence. Don't call this!!! The main purpose is to set things up | |
| 1445 ;; so that non-ASCII strings of all sorts (e.g. file names, command-line | |
| 1446 ;; arguments, environment variables) can be correctly processed during | |
| 1447 ;; the rest of the startup sequence. As a result, this will almost | |
| 1448 ;; certainly be the FIRST Lisp code called when a dumped XEmacs is run, | |
| 1449 ;; and it's called before ANY of the external environment is initialized. | |
| 1450 ;; Thus, it cannot interact at all with the outside world, make any | |
| 1451 ;; system calls, etc! (Except for `set-current-locale'.) | |
| 1452 ;; | |
| 1453 ;; NOTE: The following are the basic settings we have to deal with when | |
| 1454 ;; changing the language environment; | |
| 1455 ;; | |
| 1456 ;; -- current C library locale | |
| 1457 ;; -- under MS Windows, current MS Windows locale | |
| 1458 ;; -- LANG environment variable | |
| 1459 ;; -- native/file-name coding systems | |
| 1460 ;; -- subprocess write coding system (cdr of default-process-coding-system) | |
| 1461 ;; -- coding categories (for detection) | |
| 1462 | |
| 1463 (let (langenv) | |
| 1464 ;; under ms windows (any): | |
| 1465 (if (memq system-type '(windows-nt cygwin32)) | |
| 778 | 1466 (let ((userdef (declare-fboundp (mswindows-user-default-locale))) |
| 1467 (sysdef (declare-fboundp (mswindows-system-default-locale)))) | |
| 771 | 1468 ;; (1) current langenv comes from user-default locale. |
| 1469 (setq langenv (mswindows-get-language-environment-from-locale | |
| 1470 userdef)) | |
| 1471 ;; (2) init the langenv-to-locale table. | |
| 1472 (puthash (mswindows-get-language-environment-from-locale sysdef) | |
| 1473 sysdef mswindows-langenv-to-locale-table) | |
| 1474 ;; user-default second in langenv-to-locale table so it will | |
| 1475 ;; override the system-default if the two are different but both | |
| 1476 ;; map to the same language environment | |
| 1477 (puthash langenv userdef mswindows-langenv-to-locale-table) | |
| 1478 ;; (3) setup C lib locale, MS Windows locale, LANG environment | |
| 1479 ;; variable. Note that under Cygwin we are ignoring the | |
| 1480 ;; passed-in LANG environment variable for the moment -- it's | |
| 1481 ;; usually wrong anyway and just says "C". #### Perhaps we | |
| 1482 ;; should reconsider. | |
| 1483 (and langenv (set-locale-for-language-environment langenv)) | |
| 1484 ;; (4) override current MS Windows locale with the user-default | |
| 1485 ;; locale. Always init the MS Windows locale from the | |
| 1486 ;; user-default locale even if the langenv doesn't correspond; | |
| 1487 ;; we might not be able to find a langenv for the user-default | |
| 1488 ;; locale but we should still use the right code page, etc. | |
| 778 | 1489 (declare-fboundp (mswindows-set-current-locale userdef))) |
| 771 | 1490 ;; Unix: |
| 3173 | 1491 (let (locstring) |
| 1492 ;; Init the POSIX locale from the environment--this calls the C | |
| 1493 ;; library's setlocale(3). | |
| 1494 (set-current-locale "") | |
| 1495 ;; Can't let locstring be the result of (set-current-locale "") | |
| 1496 ;; because that can return a more detailed string than we know how | |
| 1497 ;; to handle. | |
| 1498 (setq locstring (current-locale) | |
| 1499 ;; assume C lib locale and LANG env var are set correctly. | |
| 1500 ;; use them to find the langenv. | |
| 1501 langenv | |
| 1502 (and locstring (get-language-environment-from-locale | |
| 1503 locstring))))) | |
| 771 | 1504 ;; All systems: |
| 1505 (unless langenv (setq langenv "English")) | |
| 1506 (setq current-language-environment langenv) | |
| 1507 ;; Setup various coding systems and categories. | |
| 1508 (let ((default-eol-type (coding-system-eol-type | |
| 1509 default-buffer-file-coding-system))) | |
| 1510 (reset-language-environment) | |
| 1511 (set-language-environment-coding-systems langenv default-eol-type)))) | |
| 1512 | |
| 1513 (defun init-mule-at-startup () | |
| 1514 "Initialize MULE environment at startup. Don't call this." | |
| 1515 | |
| 2367 | 1516 (when (not load-unicode-tables-at-dump-time) |
| 1517 (load-unicode-tables)) | |
| 771 | 1518 |
| 2367 | 1519 ;; This is called (currently; might be moved earlier) from startup.el, |
| 1520 ;; after the basic GUI systems have been initialized, and just before the | |
| 1521 ;; init file gets read in. It needs to finish up initializing the | |
| 1522 ;; current language environment. Very early in the startup procedure we | |
| 1523 ;; determined the default language environment from the locale, and | |
| 1524 ;; bootstrapped the native, file-name and process I/O coding systems. | |
| 1525 ;; Now we need to do it over `the right away'. | |
| 771 | 1526 (finish-set-language-environment current-language-environment) |
| 1527 | |
| 1528 ;; Load a (localizable) locale-specific init file, if it exists. | |
| 1529 ;; We now use the language environment name, NOT the locale, | |
| 1530 ;; whose name varies from system to system. | |
| 1531 (load (format "%s%s/locale-start" | |
| 1532 (locate-data-directory "start-files") | |
| 1533 current-language-environment) | |
| 1534 t t) | |
| 1535 | |
| 1536 ;; #### the rest is junk that should be deleted. | |
| 1537 | |
| 1538 (when current-language-environment | |
| 1539 ;; rman seems to be incompatible with encoded text | |
| 1540 (setq Manual-use-rosetta-man nil)) | |
| 1541 | |
| 1542 ;; Register available input methods by loading LEIM list file. | |
| 1543 (load "leim-list.el" 'noerror 'nomessage 'nosuffix) | |
| 1544 ) | |
| 1545 | |
| 1546 ;; Code deleted: init-mule-tm (Enable the tm package by default) | |
| 428 | 1547 |
| 1548 ;;; mule-cmds.el ends here |