(list-buffers-noselect): New arg BUFFER-LIST.
[bpt/emacs.git] / lisp / international / mule.el
CommitLineData
4ed46869
KH
1;;; mule.el --- basic commands for mulitilingual environment
2
4ed46869 3;; Copyright (C) 1995 Electrotechnical Laboratory, JAPAN.
58cfed09 4;; Licensed to the Free Software Foundation.
32db08f1 5;; Copyright (C) 2001, 2002 Free Software Foundation, Inc.
4ed46869
KH
6
7;; Keywords: mule, multilingual, character set, coding system
8
9;; This file is part of GNU Emacs.
10
11;; GNU Emacs is free software; you can redistribute it and/or modify
12;; it under the terms of the GNU General Public License as published by
13;; the Free Software Foundation; either version 2, or (at your option)
14;; any later version.
15
16;; GNU Emacs is distributed in the hope that it will be useful,
17;; but WITHOUT ANY WARRANTY; without even the implied warranty of
18;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
19;; GNU General Public License for more details.
20
21;; You should have received a copy of the GNU General Public License
369314dc
KH
22;; along with GNU Emacs; see the file COPYING. If not, write to the
23;; Free Software Foundation, Inc., 59 Temple Place - Suite 330,
24;; Boston, MA 02111-1307, USA.
4ed46869 25
60370d40
PJ
26;;; Commentary:
27
4ed46869
KH
28;;; Code:
29
d2542b42 30(defconst mule-version "5.0 (SAKAKI)" "\
4ed46869
KH
31Version number and name of this version of MULE (multilingual environment).")
32
35554641 33(defconst mule-version-date "1999.12.7" "\
4ed46869
KH
34Distribution date of this version of MULE (multilingual environment).")
35
36(defun load-with-code-conversion (fullname file &optional noerror nomessage)
0f69cb38
KH
37 "Execute a file of Lisp code named FILE whose absolute name is FULLNAME.
38The file contents are decoded before evaluation if necessary.
4ed46869
KH
39If optional second arg NOERROR is non-nil,
40 report no error if FILE doesn't exist.
41Print messages at start and end of loading unless
42 optional third arg NOMESSAGE is non-nil.
43Return t if file exists."
44 (if (null (file-readable-p fullname))
45 (and (null noerror)
46 (signal 'file-error (list "Cannot open load file" file)))
47 ;; Read file with code conversion, and then eval.
48 (let* ((buffer
49 ;; To avoid any autoloading, set default-major-mode to
50 ;; fundamental-mode.
88162676
RS
51 ;; So that we don't get completely screwed if the
52 ;; file is encoded in some complicated character set,
53 ;; read it with real decoding, as a multibyte buffer,
54 ;; even if this is a --unibyte Emacs session.
55 (let ((default-major-mode 'fundamental-mode)
56 (default-enable-multibyte-characters t))
4ed46869
KH
57 ;; We can't use `generate-new-buffer' because files.el
58 ;; is not yet loaded.
59 (get-buffer-create (generate-new-buffer-name " *load*"))))
db5cae4b
SM
60 (load-in-progress t)
61 (source (save-match-data (string-match "\\.el\\'" fullname))))
62 (unless nomessage
63 (if source
64 (message "Loading %s (source)..." file)
65 (message "Loading %s..." file)))
66 (when purify-flag
4c86cca0 67 (push file preloaded-file-list))
4ed46869 68 (unwind-protect
a6acd8a2 69 (let ((load-file-name fullname)
1c4cc63a 70 (set-auto-coding-for-load t)
a6acd8a2 71 (inhibit-file-name-operation nil))
4ed46869
KH
72 (save-excursion
73 (set-buffer buffer)
74 (insert-file-contents fullname)
7d276780
EZ
75 ;; If the loaded file was inserted with no-conversion or
76 ;; raw-text coding system, make the buffer unibyte.
77 ;; Otherwise, eval-buffer might try to interpret random
78 ;; binary junk as multibyte characters.
79 (if (and enable-multibyte-characters
80 (or (eq (coding-system-type last-coding-system-used) 5)
81 (eq last-coding-system-used 'no-conversion)))
82 (set-buffer-multibyte nil))
4ed46869
KH
83 ;; Make `kill-buffer' quiet.
84 (set-buffer-modified-p nil))
0f69cb38 85 ;; Have the original buffer current while we eval.
88162676
RS
86 (eval-buffer buffer nil file
87 ;; If this Emacs is running with --unibyte,
88 ;; convert multibyte strings to unibyte
89 ;; after reading them.
ba74e833 90;; (not default-enable-multibyte-characters)
8dd08b5b 91 nil t
ba74e833 92 ))
cfc70cdf
RS
93 (let (kill-buffer-hook kill-buffer-query-functions)
94 (kill-buffer buffer)))
4ed46869 95 (let ((hook (assoc file after-load-alist)))
db5cae4b
SM
96 (when hook
97 (mapcar (function eval) (cdr hook))))
98 (unless (or nomessage noninteractive)
99 (if source
100 (message "Loading %s (source)...done" file)
101 (message "Loading %s...done" file)))
4ed46869
KH
102 t)))
103
104;; API (Application Program Interface) for charsets.
105
2e02a76f 106(defsubst charset-quoted-standard-p (obj)
521d4010 107 "Return t if OBJ is a quoted symbol, and is the name of a standard charset."
2e02a76f
RS
108 (and (listp obj) (eq (car obj) 'quote)
109 (symbolp (car-safe (cdr obj)))
110 (let ((vector (get (car-safe (cdr obj)) 'charset)))
111 (and (vectorp vector)
112 (< (aref vector 0) 160)))))
4ed46869
KH
113
114(defsubst charsetp (object)
e4a7fdfc 115 "T if OBJECT is a charset."
4ed46869
KH
116 (and (symbolp object) (vectorp (get object 'charset))))
117
118(defsubst charset-info (charset)
119 "Return a vector of information of CHARSET.
120The elements of the vector are:
121 CHARSET-ID, BYTES, DIMENSION, CHARS, WIDTH, DIRECTION,
122 LEADING-CODE-BASE, LEADING-CODE-EXT,
123 ISO-FINAL-CHAR, ISO-GRAPHIC-PLANE,
124 REVERSE-CHARSET, SHORT-NAME, LONG-NAME, DESCRIPTION,
125 PLIST,
126where
127CHARSET-ID (integer) is the identification number of the charset.
aa8bb645
KH
128BYTES (integer) is the length of multi-byte form of a character in
129 the charset: one of 1, 2, 3, and 4.
4ed46869
KH
130DIMENSION (integer) is the number of bytes to represent a character of
131the charset: 1 or 2.
132CHARS (integer) is the number of characters in a dimension: 94 or 96.
4ed46869
KH
133WIDTH (integer) is the number of columns a character in the charset
134 occupies on the screen: one of 0, 1, and 2.
135DIRECTION (integer) is the rendering direction of characters in the
08b4ace5
KH
136 charset when rendering. If 0, render from left to right, else
137 render from right to left.
4ed46869
KH
138LEADING-CODE-BASE (integer) is the base leading-code for the
139 charset.
140LEADING-CODE-EXT (integer) is the extended leading-code for the
141 charset. All charsets of less than 0xA0 has the value 0.
142ISO-FINAL-CHAR (character) is the final character of the
7dd4c92d
KH
143 corresponding ISO 2022 charset. If the charset is not assigned
144 any final character, the value is -1.
4ed46869
KH
145ISO-GRAPHIC-PLANE (integer) is the graphic plane to be invoked
146 while encoding to variants of ISO 2022 coding system, one of the
147 following: 0/graphic-plane-left(GL), 1/graphic-plane-right(GR).
7dd4c92d 148 If the charset is not assigned any final character, the value is -1.
4ed46869
KH
149REVERSE-CHARSET (integer) is the charset which differs only in
150 LEFT-TO-RIGHT value from the charset. If there's no such a
151 charset, the value is -1.
152SHORT-NAME (string) is the short name to refer to the charset.
153LONG-NAME (string) is the long name to refer to the charset
154DESCRIPTION (string) is the description string of the charset.
155PLIST (property list) may contain any type of information a user
156 want to put and get by functions `put-charset-property' and
157 `get-charset-property' respectively."
158 (get charset 'charset))
159
40c81f74
PE
160;; It is better not to use backquote in this file,
161;; because that makes a bootstrapping problem
162;; if you need to recompile all the Lisp files using interpreted code.
163
4ed46869
KH
164(defmacro charset-id (charset)
165 "Return charset identification number of CHARSET."
2e02a76f 166 (if (charset-quoted-standard-p charset)
4ed46869 167 (aref (charset-info (nth 1 charset)) 0)
40c81f74 168 (list 'aref (list 'charset-info charset) 0)))
4ed46869
KH
169
170(defmacro charset-bytes (charset)
900dc6e3
KH
171 "Return bytes of CHARSET.
172See the function `charset-info' for more detail."
2e02a76f 173 (if (charset-quoted-standard-p charset)
4ed46869 174 (aref (charset-info (nth 1 charset)) 1)
40c81f74 175 (list 'aref (list 'charset-info charset) 1)))
4ed46869
KH
176
177(defmacro charset-dimension (charset)
900dc6e3
KH
178 "Return dimension of CHARSET.
179See the function `charset-info' for more detail."
2e02a76f 180 (if (charset-quoted-standard-p charset)
4ed46869 181 (aref (charset-info (nth 1 charset)) 2)
40c81f74 182 (list 'aref (list 'charset-info charset) 2)))
4ed46869
KH
183
184(defmacro charset-chars (charset)
900dc6e3
KH
185 "Return character numbers contained in a dimension of CHARSET.
186See the function `charset-info' for more detail."
2e02a76f 187 (if (charset-quoted-standard-p charset)
4ed46869 188 (aref (charset-info (nth 1 charset)) 3)
40c81f74 189 (list 'aref (list 'charset-info charset) 3)))
4ed46869
KH
190
191(defmacro charset-width (charset)
900dc6e3
KH
192 "Return width (how many column occupied on a screen) of CHARSET.
193See the function `charset-info' for more detail."
2e02a76f 194 (if (charset-quoted-standard-p charset)
4ed46869 195 (aref (charset-info (nth 1 charset)) 4)
40c81f74 196 (list 'aref (list 'charset-info charset) 4)))
4ed46869
KH
197
198(defmacro charset-direction (charset)
900dc6e3
KH
199 "Return direction of CHARSET.
200See the function `charset-info' for more detail."
2e02a76f 201 (if (charset-quoted-standard-p charset)
4ed46869 202 (aref (charset-info (nth 1 charset)) 5)
40c81f74 203 (list 'aref (list 'charset-info charset) 5)))
4ed46869
KH
204
205(defmacro charset-iso-final-char (charset)
900dc6e3
KH
206 "Return final char of CHARSET.
207See the function `charset-info' for more detail."
2e02a76f 208 (if (charset-quoted-standard-p charset)
4ed46869 209 (aref (charset-info (nth 1 charset)) 8)
40c81f74 210 (list 'aref (list 'charset-info charset) 8)))
4ed46869
KH
211
212(defmacro charset-iso-graphic-plane (charset)
900dc6e3
KH
213 "Return graphic plane of CHARSET.
214See the function `charset-info' for more detail."
2e02a76f 215 (if (charset-quoted-standard-p charset)
4ed46869 216 (aref (charset-info (nth 1 charset)) 9)
40c81f74 217 (list 'aref (list 'charset-info charset) 9)))
4ed46869
KH
218
219(defmacro charset-reverse-charset (charset)
900dc6e3
KH
220 "Return reverse charset of CHARSET.
221See the function `charset-info' for more detail."
2e02a76f 222 (if (charset-quoted-standard-p charset)
4ed46869 223 (aref (charset-info (nth 1 charset)) 10)
40c81f74 224 (list 'aref (list 'charset-info charset) 10)))
4ed46869
KH
225
226(defmacro charset-short-name (charset)
900dc6e3
KH
227 "Return short name of CHARSET.
228See the function `charset-info' for more detail."
2e02a76f 229 (if (charset-quoted-standard-p charset)
4ed46869 230 (aref (charset-info (nth 1 charset)) 11)
40c81f74 231 (list 'aref (list 'charset-info charset) 11)))
4ed46869
KH
232
233(defmacro charset-long-name (charset)
900dc6e3
KH
234 "Return long name of CHARSET.
235See the function `charset-info' for more detail."
2e02a76f 236 (if (charset-quoted-standard-p charset)
4ed46869 237 (aref (charset-info (nth 1 charset)) 12)
40c81f74 238 (list 'aref (list 'charset-info charset) 12)))
4ed46869
KH
239
240(defmacro charset-description (charset)
0611934b 241 "Return description of CHARSET.
900dc6e3 242See the function `charset-info' for more detail."
2e02a76f 243 (if (charset-quoted-standard-p charset)
4ed46869 244 (aref (charset-info (nth 1 charset)) 13)
40c81f74 245 (list 'aref (list 'charset-info charset) 13)))
4ed46869
KH
246
247(defmacro charset-plist (charset)
900dc6e3
KH
248 "Return list charset property of CHARSET.
249See the function `charset-info' for more detail."
40c81f74
PE
250 (list 'aref
251 (if (charset-quoted-standard-p charset)
252 (charset-info (nth 1 charset))
253 (list 'charset-info charset))
254 14))
4ed46869
KH
255
256(defun set-charset-plist (charset plist)
e8dd0160 257 "Set CHARSET's property list to PLIST, and return PLIST."
4ed46869
KH
258 (aset (charset-info charset) 14 plist))
259
d3e7e7cf
EZ
260(defun make-char (charset &optional code1 code2)
261 "Return a character of CHARSET whose position codes are CODE1 and CODE2.
f98e2797 262CODE1 and CODE2 are optional, but if you don't supply
a9fb0b58
KH
263sufficient position codes, return a generic character which stands for
264all characters or group of characters in the character set.
d3e7e7cf 265A generic character can be used to index a char table (e.g. syntax-table).
9ff05eae
KH
266
267Such character sets as ascii, eight-bit-control, and eight-bit-graphic
268don't have corresponding generic characters. If CHARSET is one of
269them and you don't supply CODE1, return the character of the smallest
270code in CHARSET.
271
6fb7766c
DL
272If CODE1 or CODE2 are invalid (out of range), this function signals an
273error. However, the eighth bit of both CODE1 and CODE2 is zeroed
274before they are used to index CHARSET. Thus you may use, say, the
275actual ISO 8859 character code rather than subtracting 128, as you
276would need to index the corresponding Emacs charset."
636799f2 277 (make-char-internal (charset-id charset) code1 code2))
a6acd8a2
KH
278
279(put 'make-char 'byte-compile
521d4010 280 (function
a6acd8a2
KH
281 (lambda (form)
282 (let ((charset (nth 1 form)))
283 (if (charset-quoted-standard-p charset)
284 (byte-compile-normal-call
285 (cons 'make-char-internal
286 (cons (charset-id (nth 1 charset)) (nthcdr 2 form))))
287 (byte-compile-normal-call
288 (cons 'make-char-internal
289 (cons (list 'charset-id charset) (nthcdr 2 form)))))))))
4ed46869 290
d3675a42 291(defun charset-list ()
900dc6e3
KH
292 "Return list of charsets ever defined.
293
d3675a42 294This function is provided for backward compatibility.
900dc6e3 295Now we have the variable `charset-list'."
d3675a42
KH
296 charset-list)
297
13d5617d
KH
298(defsubst generic-char-p (char)
299 "Return t if and only if CHAR is a generic character.
521d4010 300See also the documentation of `make-char'."
1d935669
KH
301 (and (>= char 0400)
302 (let ((l (split-char char)))
303 (and (or (= (nth 1 l) 0) (eq (nth 2 l) 0))
304 (not (eq (car l) 'composition))))))
69eba008 305
cc57cc54 306(defun decode-char (ccs code-point &optional restriction)
7f341749 307 "Return character specified by coded character set CCS and CODE-POINT in it.
cc57cc54 308Return nil if such a character is not supported.
e76938e7 309Currently the only supported coded character set is `ucs' (ISO/IEC
9fb1b1a8 31010646: Universal Multi-Octet Coded Character Set), and the result is
0c76a98d
KH
311translated through the translation-table named
312`utf-translation-table-for-decode' or the translation-hash-table named
313`utf-subst-table-for-decode'.
cc57cc54
KH
314
315Optional argument RESTRICTION specifies a way to map the pair of CCS
e7fea150 316and CODE-POINT to a character. Currently not supported and just ignored."
9fb1b1a8
DL
317 (cond
318 ((eq ccs 'ucs)
8458c8b9 319 (or (utf-lookup-subst-table-for-decode code-point)
0c76a98d
KH
320 (let ((c (cond
321 ((< code-point 160)
322 code-point)
323 ((< code-point 256)
324 (make-char 'latin-iso8859-1 code-point))
325 ((< code-point #x2500)
326 (setq code-point (- code-point #x0100))
327 (make-char 'mule-unicode-0100-24ff
328 (+ (/ code-point 96) 32) (+ (% code-point 96) 32)))
329 ((< code-point #x3400)
330 (setq code-point (- code-point #x2500))
331 (make-char 'mule-unicode-2500-33ff
332 (+ (/ code-point 96) 32) (+ (% code-point 96) 32)))
333 ((and (>= code-point #xe000) (< code-point #x10000))
334 (setq code-point (- code-point #xe000))
335 (make-char 'mule-unicode-e000-ffff
336 (+ (/ code-point 96) 32)
337 (+ (% code-point 96) 32))))))
338 (when c
339 (or (aref (get 'utf-translation-table-for-decode
340 'translation-table) c)
341 c)))))))
cc57cc54
KH
342
343(defun encode-char (char ccs &optional restriction)
7f341749 344 "Return code-point in coded character set CCS that corresponds to CHAR.
cc57cc54 345Return nil if CHAR is not included in CCS.
e76938e7 346Currently the only supported coded character set is `ucs' (ISO/IEC
9fb1b1a8 34710646: Universal Multi-Octet Coded Character Set), and CHAR is first
0c76a98d
KH
348translated through the translation-table named
349`utf-translation-table-for-encode' or the translation-hash-table named
350`utf-subst-table-for-encode'.
7f341749
KH
351
352CHAR should be in one of these charsets:
cc57cc54 353 ascii, latin-iso8859-1, mule-unicode-0100-24ff, mule-unicode-2500-33ff,
6ba9b20b 354 mule-unicode-e000-ffff, eight-bit-control
cc57cc54
KH
355Otherwise, return nil.
356
357Optional argument RESTRICTION specifies a way to map CHAR to a
358code-point in CCS. Currently not supported and just ignored."
359 (let* ((split (split-char char))
9fb1b1a8
DL
360 (charset (car split))
361 trans)
cc57cc54 362 (cond ((eq ccs 'ucs)
8458c8b9 363 (or (utf-lookup-subst-table-for-encode char)
0c76a98d
KH
364 (let ((table (get 'utf-translation-table-for-encode
365 'translation-table)))
366 (setq trans (aref table char))
367 (if trans
368 (setq split (split-char trans)
369 charset (car split)))
370 (cond ((eq charset 'ascii)
bb55c6c9 371 (or trans char))
0c76a98d
KH
372 ((eq charset 'latin-iso8859-1)
373 (+ (nth 1 split) 128))
374 ((eq charset 'mule-unicode-0100-24ff)
375 (+ #x0100 (+ (* (- (nth 1 split) 32) 96)
376 (- (nth 2 split) 32))))
377 ((eq charset 'mule-unicode-2500-33ff)
378 (+ #x2500 (+ (* (- (nth 1 split) 32) 96)
379 (- (nth 2 split) 32))))
380 ((eq charset 'mule-unicode-e000-ffff)
381 (+ #xe000 (+ (* (- (nth 1 split) 32) 96)
382 (- (nth 2 split) 32))))
383 ((eq charset 'eight-bit-control)
384 char))))))))
cc57cc54 385
0269ddfb 386\f
e76938e7 387;; Coding system stuff
4ed46869 388
8057896b 389;; Coding system is a symbol that has the property `coding-system'.
4ed46869 390;;
8057896b
KH
391;; The value of the property `coding-system' is a vector of the
392;; following format:
d3675a42 393;; [TYPE MNEMONIC DOC-STRING PLIST FLAGS]
8057896b 394;; We call this vector as coding-spec. See comments in src/coding.c
521d4010 395;; for more detail.
8057896b
KH
396
397(defconst coding-spec-type-idx 0)
398(defconst coding-spec-mnemonic-idx 1)
399(defconst coding-spec-doc-string-idx 2)
0269ddfb 400(defconst coding-spec-plist-idx 3)
8057896b
KH
401(defconst coding-spec-flags-idx 4)
402
a6acd8a2
KH
403;; PLIST is a property list of a coding system. To share PLIST among
404;; alias coding systems, a coding system has PLIST in coding-spec
405;; instead of having it in normal property list of Lisp symbol.
406;; Here's a list of coding system properties currently being used.
0269ddfb
KH
407;;
408;; o coding-category
409;;
410;; The value is a coding category the coding system belongs to. The
521d4010 411;; function `make-coding-system' sets this value automatically
1b46a680 412;; unless its argument PROPERTIES specifies this property.
4ed46869 413;;
0269ddfb 414;; o alias-coding-systems
4ed46869 415;;
0269ddfb
KH
416;; The value is a list of coding systems of the same alias group. The
417;; first element is the coding system made at first, which we call as
1b46a680
KH
418;; `base coding system'. The function `make-coding-system' sets this
419;; value automatically and `define-coding-system-alias' updates it.
0269ddfb 420;;
3c948dde
KH
421;; See the documentation of make-coding-system for the meanings of the
422;; following properties.
4ed46869 423;;
3c948dde 424;; o post-read-conversion
0269ddfb 425;; o pre-write-conversion
f967223b 426;; o translation-table-for-decode
f967223b 427;; o translation-table-for-encode
c11a8f77 428;; o safe-chars
a6acd8a2 429;; o safe-charsets
b25eef20 430;; o mime-charset
c76b5c99 431;; o valid-codes (meaningful only for a coding system based on CCL)
c76b5c99 432
0269ddfb 433
0269ddfb 434(defsubst coding-system-spec (coding-system)
521d4010 435 "Return coding-spec of CODING-SYSTEM."
0269ddfb 436 (get (check-coding-system coding-system) 'coding-system))
4ed46869 437
4ed46869 438(defun coding-system-type (coding-system)
0269ddfb
KH
439 "Return the coding type of CODING-SYSTEM.
440A coding type is an integer value indicating the encoding method
441of CODING-SYSTEM. See the function `make-coding-system' for more detail."
2653e2a7 442 (aref (coding-system-spec coding-system) coding-spec-type-idx))
4ed46869 443
4ed46869 444(defun coding-system-mnemonic (coding-system)
0269ddfb 445 "Return the mnemonic character of CODING-SYSTEM.
6e2c8840
KH
446The mnemonic character of a coding system is used in mode line
447to indicate the coding system. If the arg is nil, return ?-."
448 (let ((spec (coding-system-spec coding-system)))
449 (if spec (aref spec coding-spec-mnemonic-idx) ?-)))
4ed46869 450
8057896b 451(defun coding-system-doc-string (coding-system)
0269ddfb 452 "Return the documentation string for CODING-SYSTEM."
2653e2a7 453 (aref (coding-system-spec coding-system) coding-spec-doc-string-idx))
4ed46869 454
d3675a42 455(defun coding-system-plist (coding-system)
0269ddfb 456 "Return the property list of CODING-SYSTEM."
2653e2a7 457 (aref (coding-system-spec coding-system) coding-spec-plist-idx))
d3675a42 458
4ed46869 459(defun coding-system-flags (coding-system)
0269ddfb
KH
460 "Return `flags' of CODING-SYSTEM.
461A `flags' of a coding system is a vector of length 32 indicating detailed
462information of a coding system. See the function `make-coding-system'
463for more detail."
2653e2a7 464 (aref (coding-system-spec coding-system) coding-spec-flags-idx))
0269ddfb
KH
465
466(defun coding-system-get (coding-system prop)
467 "Extract a value from CODING-SYSTEM's property list for property PROP."
468 (plist-get (coding-system-plist coding-system) prop))
469
470(defun coding-system-put (coding-system prop val)
471 "Change value in CODING-SYSTEM's property list PROP to VAL."
472 (let ((plist (coding-system-plist coding-system)))
473 (if plist
474 (plist-put plist prop val)
2653e2a7
KH
475 (aset (coding-system-spec coding-system) coding-spec-plist-idx
476 (list prop val)))))
0269ddfb
KH
477
478(defun coding-system-category (coding-system)
72f73748
DL
479 "Return the coding category of CODING-SYSTEM.
480See also `coding-category-list'."
0269ddfb
KH
481 (coding-system-get coding-system 'coding-category))
482
483(defun coding-system-base (coding-system)
484 "Return the base coding system of CODING-SYSTEM.
d9e3229d
KH
485A base coding system is what made by `make-coding-system'.
486Any alias nor subsidiary coding systems are not base coding system."
0269ddfb
KH
487 (car (coding-system-get coding-system 'alias-coding-systems)))
488
489(defalias 'coding-system-parent 'coding-system-base)
2598a293 490(make-obsolete 'coding-system-parent 'coding-system-base "20.3")
0269ddfb
KH
491
492;; Coding system also has a property `eol-type'.
493;;
494;; This property indicates how the coding system handles end-of-line
495;; format. The value is integer 0, 1, 2, or a vector of three coding
496;; systems. Each integer value 0, 1, and 2 indicates the format of
497;; end-of-line LF, CRLF, and CR respectively. A vector value
498;; indicates that the format of end-of-line should be detected
499;; automatically. Nth element of the vector is the subsidiary coding
500;; system whose `eol-type' property is N.
4ed46869 501
8057896b 502(defun coding-system-eol-type (coding-system)
0269ddfb
KH
503 "Return eol-type of CODING-SYSTEM.
504An eol-type is integer 0, 1, 2, or a vector of coding systems.
4ed46869 505
0269ddfb
KH
506Integer values 0, 1, and 2 indicate a format of end-of-line; LF,
507CRLF, and CR respectively.
508
509A vector value indicates that a format of end-of-line should be
510detected automatically. Nth element of the vector is the subsidiary
511coding system whose eol-type is N."
512 (get coding-system 'eol-type))
bd882697 513
2e729bfa
JB
514(defun coding-system-eol-type-mnemonic (coding-system)
515 "Return the string indicating end-of-line format of CODING-SYSTEM."
516 (let* ((eol-type (coding-system-eol-type coding-system))
f4f00827 517 (val (cond ((eq eol-type 0) eol-mnemonic-unix)
2e729bfa
JB
518 ((eq eol-type 1) eol-mnemonic-dos)
519 ((eq eol-type 2) eol-mnemonic-mac)
f4f00827 520 (t eol-mnemonic-undecided))))
2e729bfa
JB
521 (if (stringp val)
522 val
523 (char-to-string val))))
524
857ea15c
AS
525(defun coding-system-lessp (x y)
526 (cond ((eq x 'no-conversion) t)
527 ((eq y 'no-conversion) nil)
528 ((eq x 'emacs-mule) t)
529 ((eq y 'emacs-mule) nil)
530 ((eq x 'undecided) t)
531 ((eq y 'undecided) nil)
532 (t (let ((c1 (coding-system-mnemonic x))
533 (c2 (coding-system-mnemonic y)))
534 (or (< (downcase c1) (downcase c2))
535 (and (not (> (downcase c1) (downcase c2)))
536 (< c1 c2)))))))
537
5e2e859a
KH
538(defun coding-system-equal (coding-system-1 coding-system-2)
539 "Return t if and only if CODING-SYSTEM-1 and CODING-SYSTEM-2 are identical.
540Two coding systems are identical if two symbols are equal
541or one is an alias of the other."
542 (or (eq coding-system-1 coding-system-2)
543 (and (equal (coding-system-spec coding-system-1)
544 (coding-system-spec coding-system-2))
545 (let ((eol-type-1 (coding-system-eol-type coding-system-1))
546 (eol-type-2 (coding-system-eol-type coding-system-2)))
547 (or (eq eol-type-1 eol-type-2)
548 (and (vectorp eol-type-1) (vectorp eol-type-2)))))))
549
857ea15c 550(defun add-to-coding-system-list (coding-system)
521d4010 551 "Add CODING-SYSTEM to `coding-system-list' while keeping it sorted."
857ea15c
AS
552 (if (or (null coding-system-list)
553 (coding-system-lessp coding-system (car coding-system-list)))
554 (setq coding-system-list (cons coding-system coding-system-list))
555 (let ((len (length coding-system-list))
556 mid (tem coding-system-list))
557 (while (> len 1)
558 (setq mid (nthcdr (/ len 2) tem))
559 (if (coding-system-lessp (car mid) coding-system)
560 (setq tem mid
561 len (- len (/ len 2)))
562 (setq len (/ len 2))))
563 (setcdr tem (cons coding-system (cdr tem))))))
564
80a7463d 565(defun coding-system-list (&optional base-only)
c11a8f77
KH
566 "Return a list of all existing non-subsidiary coding systems.
567If optional arg BASE-ONLY is non-nil, only base coding systems are listed.
568The value doesn't include subsidiary coding systems which are what
569made from bases and aliases automatically for various end-of-line
570formats (e.g. iso-latin-1-unix, koi8-r-dos)."
80a7463d
KH
571 (let* ((codings (copy-sequence coding-system-list))
572 (tail (cons nil codings)))
573 ;; Remove subsidiary coding systems (eol variants) and alias
574 ;; coding systems (if necessary).
575 (while (cdr tail)
576 (let* ((coding (car (cdr tail)))
577 (aliases (coding-system-get coding 'alias-coding-systems)))
578 (if (or
579 ;; CODING is an eol variant if not in ALIASES.
580 (not (memq coding aliases))
581 ;; CODING is an alias if it is not car of ALIASES.
582 (and base-only (not (eq coding (car aliases)))))
583 (setcdr tail (cdr (cdr tail)))
584 (setq tail (cdr tail)))))
585 codings))
586
6fb7766c
DL
587(defun map-charset-chars (func charset)
588 "Use FUNC to map over all characters in CHARSET for side effects.
589FUNC is a function of two args, the start and end (inclusive) of a
590character code range. Thus FUNC should iterate over [START, END]."
591 (let* ((dim (charset-dimension charset))
592 (chars (charset-chars charset))
593 (start (if (= chars 94)
594 33
595 32)))
596 (if (= dim 1)
597 (funcall func
598 (make-char charset start)
599 (make-char charset (+ start chars -1)))
600 (dotimes (i chars)
601 (funcall func
602 (make-char charset (+ i start) start)
603 (make-char charset (+ i start) (+ start chars -1)))))))
604
9857367f 605(defalias 'register-char-codings 'ignore "")
620956ca 606(make-obsolete 'register-char-codings
9857367f 607 "it exists just for backward compatibility, and does nothing."
620956ca
KH
608 "21.3")
609
610(defconst char-coding-system-table nil
611 "This is an obsolete variable.
612It exists just for backward compatibility, and the value is always nil.")
c11a8f77 613
6e9722b0 614(defun make-subsidiary-coding-system (coding-system)
521d4010 615 "Make subsidiary coding systems (eol-type variants) of CODING-SYSTEM."
0269ddfb
KH
616 (let ((coding-spec (coding-system-spec coding-system))
617 (subsidiaries (vector (intern (format "%s-unix" coding-system))
8057896b
KH
618 (intern (format "%s-dos" coding-system))
619 (intern (format "%s-mac" coding-system))))
0269ddfb
KH
620 (i 0)
621 temp)
8057896b 622 (while (< i 3)
0269ddfb 623 (put (aref subsidiaries i) 'coding-system coding-spec)
8057896b 624 (put (aref subsidiaries i) 'eol-type i)
857ea15c 625 (add-to-coding-system-list (aref subsidiaries i))
0269ddfb
KH
626 (setq coding-system-alist
627 (cons (list (symbol-name (aref subsidiaries i)))
628 coding-system-alist))
8057896b
KH
629 (setq i (1+ i)))
630 subsidiaries))
4ed46869 631
50c29104
KH
632(defun transform-make-coding-system-args (name type &optional doc-string props)
633 "For internal use only.
634Transform XEmacs style args for `make-coding-system' to Emacs style.
635Value is a list of transformed arguments."
636 (let ((mnemonic (string-to-char (or (plist-get props 'mnemonic) "?")))
637 (eol-type (plist-get props 'eol-type))
638 properties tmp)
639 (cond
640 ((eq eol-type 'lf) (setq eol-type 'unix))
641 ((eq eol-type 'crlf) (setq eol-type 'dos))
642 ((eq eol-type 'cr) (setq eol-type 'mac)))
643 (if (setq tmp (plist-get props 'post-read-conversion))
644 (setq properties (plist-put properties 'post-read-conversion tmp)))
645 (if (setq tmp (plist-get props 'pre-write-conversion))
646 (setq properties (plist-put properties 'pre-write-conversion tmp)))
647 (cond
f4a012a6
KH
648 ((eq type 'shift-jis)
649 `(,name 1 ,mnemonic ,doc-string () ,properties ,eol-type))
650 ((eq type 'iso2022) ; This is not perfect.
651 (if (plist-get props 'escape-quoted)
652 (error "escape-quoted is not supported: %S"
653 `(,name ,type ,doc-string ,props)))
654 (let ((g0 (plist-get props 'charset-g0))
655 (g1 (plist-get props 'charset-g1))
656 (g2 (plist-get props 'charset-g2))
657 (g3 (plist-get props 'charset-g3))
658 (use-roman
659 (and
660 (eq (cadr (assoc 'latin-jisx0201
661 (plist-get props 'input-charset-conversion)))
662 'ascii)
663 (eq (cadr (assoc 'ascii
664 (plist-get props 'output-charset-conversion)))
665 'latin-jisx0201)))
666 (use-oldjis
667 (and
668 (eq (cadr (assoc 'japanese-jisx0208-1978
669 (plist-get props 'input-charset-conversion)))
670 'japanese-jisx0208)
671 (eq (cadr (assoc 'japanese-jisx0208
672 (plist-get props 'output-charset-conversion)))
673 'japanese-jisx0208-1978))))
674 (if (charsetp g0)
675 (if (plist-get props 'force-g0-on-output)
676 (setq g0 `(nil ,g0))
677 (setq g0 `(,g0 t))))
678 (if (charsetp g1)
679 (if (plist-get props 'force-g1-on-output)
680 (setq g1 `(nil ,g1))
681 (setq g1 `(,g1 t))))
682 (if (charsetp g2)
683 (if (plist-get props 'force-g2-on-output)
684 (setq g2 `(nil ,g2))
685 (setq g2 `(,g2 t))))
686 (if (charsetp g3)
687 (if (plist-get props 'force-g3-on-output)
688 (setq g3 `(nil ,g3))
689 (setq g3 `(,g3 t))))
690 `(,name 2 ,mnemonic ,doc-string
691 (,g0 ,g1 ,g2 ,g3
692 ,(plist-get props 'short)
693 ,(not (plist-get props 'no-ascii-eol))
694 ,(not (plist-get props 'no-ascii-cntl))
695 ,(plist-get props 'seven)
696 t
697 ,(not (plist-get props 'lock-shift))
698 ,use-roman
699 ,use-oldjis
700 ,(plist-get props 'no-iso6429)
701 nil nil nil nil)
702 ,properties ,eol-type)))
703 ((eq type 'big5)
704 `(,name 3 ,mnemonic ,doc-string () ,properties ,eol-type))
50c29104 705 ((eq type 'ccl)
f4a012a6 706 `(,name 4 ,mnemonic ,doc-string
50c29104 707 (,(plist-get props 'decode) . ,(plist-get props 'encode))
f4a012a6 708 ,properties ,eol-type))
50c29104 709 (t
f4a012a6 710 (error "unsupported XEmacs style make-coding-style arguments: %S"
50c29104
KH
711 `(,name ,type ,doc-string ,props))))))
712
8057896b 713(defun make-coding-system (coding-system type mnemonic doc-string
1b46a680
KH
714 &optional
715 flags
716 properties
717 eol-type)
3bb1accb 718 "Define a new coding system CODING-SYSTEM (symbol).
521d4010 719Remaining arguments are TYPE, MNEMONIC, DOC-STRING, FLAGS (optional),
d9e3229d 720and PROPERTIES (optional) which construct a coding-spec of CODING-SYSTEM
a6acd8a2 721in the following format:
0269ddfb 722 [TYPE MNEMONIC DOC-STRING PLIST FLAGS]
3bb1accb
KH
723
724TYPE is an integer value indicating the type of the coding system as follows:
4ed46869 725 0: Emacs internal format,
d9f6dfe6 726 1: Shift-JIS (or MS-Kanji) used mainly on Japanese PCs,
4ed46869 727 2: ISO-2022 including many variants,
d9f6dfe6 728 3: Big5 used mainly on Chinese PCs,
cd9c3177 729 4: private, CCL programs provide encoding/decoding algorithm,
521d4010 730 5: Raw-text, which means that text contains random 8-bit codes.
0269ddfb 731
3bb1accb 732MNEMONIC is a character to be displayed on mode line for the coding system.
0269ddfb 733
3bb1accb 734DOC-STRING is a documentation string for the coding system.
0269ddfb 735
3bb1accb 736FLAGS specifies more detailed information of the coding system as follows:
d6d6d592 737
3bb1accb 738 If TYPE is 2 (ISO-2022), FLAGS is a list of these elements:
4ed46869
KH
739 CHARSET0, CHARSET1, CHARSET2, CHARSET3, SHORT-FORM,
740 ASCII-EOL, ASCII-CNTL, SEVEN, LOCKING-SHIFT, SINGLE-SHIFT,
d6d6d592 741 USE-ROMAN, USE-OLDJIS, NO-ISO6429, INIT-BOL, DESIGNATION-BOL,
850101ed 742 SAFE, ACCEPT-LATIN-EXTRA-CODE.
4ed46869
KH
743 CHARSETn are character sets initially designated to Gn graphic registers.
744 If CHARSETn is nil, Gn is never used.
745 If CHARSETn is t, Gn can be used but nothing designated initially.
746 If CHARSETn is a list of character sets, those character sets are
747 designated to Gn on output, but nothing designated to Gn initially.
421e3b4e 748 But, character set `ascii' can be designated only to G0.
4ed46869
KH
749 SHORT-FORM non-nil means use short designation sequence on output.
750 ASCII-EOL non-nil means designate ASCII to g0 at end of line on output.
751 ASCII-CNTL non-nil means designate ASCII to g0 before control codes and
752 SPACE on output.
753 SEVEN non-nil means use 7-bit code only on output.
754 LOCKING-SHIFT non-nil means use locking-shift.
755 SINGLE-SHIFT non-nil means use single-shift.
756 USE-ROMAN non-nil means designate JIS0201-1976-Roman instead of ASCII.
757 USE-OLDJIS non-nil means designate JIS0208-1976 instead of JIS0208-1983.
758 NO-ISO6429 non-nil means not use ISO6429's direction specification.
69eba008
KH
759 INIT-BOL non-nil means any designation state is assumed to be reset
760 to initial at each beginning of line on output.
761 DESIGNATION-BOL non-nil means designation sequences should be placed
762 at beginning of line on output.
a6acd8a2 763 SAFE non-nil means convert unsafe characters to `?' on output.
c11a8f77
KH
764 Characters not specified in the property `safe-charsets' nor
765 `safe-chars' are unsafe.
850101ed
RS
766 ACCEPT-LATIN-EXTRA-CODE non-nil means code-detection routine accepts
767 a code specified in `latin-extra-code-table' (which see) as a valid
768 code of the coding system.
d6d6d592 769
3bb1accb
KH
770 If TYPE is 4 (private), FLAGS should be a cons of CCL programs, for
771 decoding and encoding. CCL programs should be specified by their
772 symbols.
d9e3229d 773
a2852e45
KH
774PROPERTIES is an alist of properties vs the corresponding values. The
775following properties are recognized:
776
777 o post-read-conversion
0bca779a 778
a2852e45
KH
779 The value is a function to call after some text is inserted and
780 decoded by the coding system itself and before any functions in
521d4010
DL
781 `after-insert-functions' are called. The argument of this
782 function is the same as for a function in
783 `after-insert-file-functions', i.e. LENGTH of the text inserted,
784 with point at the head of the text to be decoded.
0bca779a 785
a2852e45 786 o pre-write-conversion
0bca779a 787
a2852e45
KH
788 The value is a function to call after all functions in
789 `write-region-annotate-functions' and `buffer-file-format' are
790 called, and before the text is encoded by the coding system itself.
a284eea3
DL
791 The arguments to this function are the same as those of a function
792 in `write-region-annotate-functions', i.e. FROM and TO, specifying
521d4010 793 a region of text.
0bca779a 794
a2852e45 795 o translation-table-for-decode
0bca779a 796
a2852e45
KH
797 The value is a translation table to be applied on decoding. See
798 the function `make-translation-table' for the format of translation
521d4010 799 table. This is not applicable to type 4 (CCL-based) coding systems.
0bca779a 800
a2852e45 801 o translation-table-for-encode
0bca779a 802
521d4010
DL
803 The value is a translation table to be applied on encoding. This is
804 not applicable to type 4 (CCL-based) coding systems.
0bca779a 805
a2852e45 806 o safe-chars
0bca779a 807
a2852e45
KH
808 The value is a char table. If a character has non-nil value in it,
809 the character is safely supported by the coding system. This
810 overrides the specification of safe-charsets.
811
812 o safe-charsets
521d4010 813
a2852e45
KH
814 The value is a list of charsets safely supported by the coding
815 system. The value t means that all charsets Emacs handles are
816 supported. Even if some charset is not in this list, it doesn't
521d4010
DL
817 mean that the charset can't be encoded in the coding system;
818 it just means that some other receiver of text encoded
a2852e45 819 in the coding system won't be able to handle that charset.
0bca779a 820
a2852e45 821 o mime-charset
0bca779a 822
d9f6dfe6 823 The value is a symbol whose name is the `MIME-charset' parameter of
a2852e45 824 the coding system.
0bca779a 825
3b6cc699
DL
826 o mime-text-unsuitable
827
828 A non-nil value means the `mime-charset' property names a charset
829 which is unsuitable for the top-level media type \"text\".
830
a2852e45 831 o valid-codes (meaningful only for a coding system based on CCL)
0bca779a 832
a2852e45
KH
833 The value is a list to indicate valid byte ranges of the encoded
834 file. Each element of the list is an integer or a cons of integer.
835 In the former case, the integer value is a valid byte code. In the
6fb7766c 836 latter case, the integers specify the range of valid byte codes.
a2852e45 837
1e7f6327
KH
838 o composition (meaningful only when TYPE is 0 or 2)
839
0ea0b6d2
DL
840 If the value is non-nil, the coding system preserves composition
841 information.
1e7f6327 842
d9e3229d
KH
843These properties are set in PLIST, a property list. This function
844also sets properties `coding-category' and `alias-coding-systems'
845automatically.
4ed46869 846
1b46a680
KH
847EOL-TYPE specifies the EOL type of the coding-system in one of the
848following formats:
849
850 o symbol (unix, dos, or mac)
851
852 The symbol `unix' means Unix-like EOL (LF), `dos' means
853 DOS-like EOL (CRLF), and `mac' means MAC-like EOL (CR).
854
855 o number (0, 1, or 2)
856
857 The number 0, 1, and 2 mean UNIX, DOS, and MAC-like EOL
858 respectively.
859
860 o vector of coding-systems of length 3
861
862 The EOL type is detected automatically for the coding system.
863 And, according to the detected EOL type, one of the coding
864 systems in the vector is selected. Elements of the vector
4d3a9228 865 corresponds to Unix-like EOL, DOS-like EOL, and Mac-like EOL
1b46a680
KH
866 in this order.
867
3bb1accb
KH
868Kludgy features for backward compatibility:
869
8701. If TYPE is 4 and car or cdr of FLAGS is a vector, the vector is
871treated as a compiled CCL code.
872
8732. If PROPERTIES is just a list of character sets, the list is set as
874a value of `safe-charsets' in PLIST."
50c29104
KH
875
876 ;; For compatiblity with XEmacs, we check the type of TYPE. If it
c3d0ee51
EZ
877 ;; is a symbol, perhaps, this function is called with XEmacs-style
878 ;; arguments. Here, try to transform that kind of arguments to
50c29104
KH
879 ;; Emacs style.
880 (if (symbolp type)
881 (let ((args (transform-make-coding-system-args coding-system type
882 mnemonic doc-string)))
883 (setq coding-system (car args)
1053cc93 884 type (nth 1 args)
50c29104
KH
885 mnemonic (nth 2 args)
886 doc-string (nth 3 args)
887 flags (nth 4 args)
888 properties (nth 5 args)
889 eol-type (nth 6 args))))
890
0269ddfb 891 ;; Set a value of `coding-system' property.
6e9722b0 892 (let ((coding-spec (make-vector 5 nil))
d9e3229d
KH
893 (no-initial-designation t)
894 (no-alternative-designation t)
e6cddb1a 895 (accept-latin-extra-code nil)
6e9722b0 896 coding-category)
cd9c3177 897 (if (or (not (integerp type)) (< type 0) (> type 5))
0269ddfb 898 (error "TYPE argument must be 0..5"))
8057896b 899 (if (or (not (integerp mnemonic)) (<= mnemonic ? ) (> mnemonic 127))
521d4010 900 (error "MNEMONIC argument must be an ASCII printable character"))
0269ddfb
KH
901 (aset coding-spec coding-spec-type-idx type)
902 (aset coding-spec coding-spec-mnemonic-idx mnemonic)
903 (aset coding-spec coding-spec-doc-string-idx
e127a722 904 (purecopy (if (stringp doc-string) doc-string "")))
6e9722b0
KH
905 (cond ((= type 0)
906 (setq coding-category 'coding-category-emacs-mule))
907 ((= type 1)
908 (setq coding-category 'coding-category-sjis))
909 ((= type 2) ; ISO2022
4ed46869 910 (let ((i 0)
6e9722b0 911 (vec (make-vector 32 nil))
05134257
KH
912 (g1-designation nil)
913 (fl flags))
4ed46869 914 (while (< i 4)
05134257 915 (let ((charset (car fl)))
6e9722b0
KH
916 (if (and no-initial-designation
917 (> i 0)
918 (or (charsetp charset)
919 (and (consp charset)
920 (charsetp (car charset)))))
921 (setq no-initial-designation nil))
922 (if (charsetp charset)
923 (if (= i 1) (setq g1-designation charset))
924 (if (consp charset)
925 (let ((tail charset)
926 elt)
927 (while tail
928 (setq elt (car tail))
d9e3229d
KH
929 (if (eq elt t)
930 (setq no-alternative-designation nil)
931 (if (and elt (not (charsetp elt)))
932 (error "Invalid charset: %s" elt)))
6e9722b0
KH
933 (setq tail (cdr tail)))
934 (setq g1-designation (car charset)))
d9e3229d
KH
935 (if charset
936 (if (eq charset t)
937 (setq no-alternative-designation nil)
938 (error "Invalid charset: %s" charset)))))
4ed46869 939 (aset vec i charset))
05134257
KH
940 (setq fl (cdr fl) i (1+ i)))
941 (while (and (< i 32) fl)
942 (aset vec i (car fl))
e6cddb1a
KH
943 (if (and (= i 16) ; ACCEPT-LATIN-EXTRA-CODE
944 (car fl))
945 (setq accept-latin-extra-code t))
05134257 946 (setq fl (cdr fl) i (1+ i)))
6e9722b0 947 (aset coding-spec 4 vec)
6e9722b0
KH
948 (setq coding-category
949 (if (aref vec 8) ; Use locking-shift.
dc64cd19
KH
950 (or (and (aref vec 7) 'coding-category-iso-7-else)
951 'coding-category-iso-8-else)
6e9722b0
KH
952 (if (aref vec 7) ; 7-bit only.
953 (if (aref vec 9) ; Use single-shift.
dc64cd19 954 'coding-category-iso-7-else
d9e3229d
KH
955 (if no-alternative-designation
956 'coding-category-iso-7-tight
957 'coding-category-iso-7))
3ad911d8
KH
958 (if (or no-initial-designation
959 (not no-alternative-designation))
dc64cd19 960 'coding-category-iso-8-else
6e9722b0
KH
961 (if (and (charsetp g1-designation)
962 (= (charset-dimension g1-designation) 2))
963 'coding-category-iso-8-2
964 'coding-category-iso-8-1)))))))
965 ((= type 3)
966 (setq coding-category 'coding-category-big5))
967 ((= type 4) ; private
c76b5c99 968 (setq coding-category 'coding-category-ccl)
3bb1accb
KH
969 (if (not (consp flags))
970 (error "Invalid FLAGS argument for TYPE 4 (CCL)")
971 (let ((decoder (check-ccl-program
972 (car flags)
973 (intern (format "%s-decoder" coding-system))))
974 (encoder (check-ccl-program
975 (cdr flags)
976 (intern (format "%s-encoder" coding-system)))))
977 (if (and decoder encoder)
978 (aset coding-spec 4 (cons decoder encoder))
979 (error "Invalid FLAGS argument for TYPE 4 (CCL)")))))
cd9c3177
KH
980 (t ; i.e. (= type 5)
981 (setq coding-category 'coding-category-raw-text)))
0269ddfb
KH
982
983 (let ((plist (list 'coding-category coding-category
d9e3229d 984 'alias-coding-systems (list coding-system))))
0269ddfb 985 (if no-initial-designation
d9e3229d
KH
986 (plist-put plist 'no-initial-designation t))
987 (if (and properties
988 (or (eq properties t)
989 (not (consp (car properties)))))
990 ;; In the old version, the arg PROPERTIES is a list to be
991 ;; set in PLIST as a value of property `safe-charsets'.
c11a8f77
KH
992 (setq properties (list (cons 'safe-charsets properties))))
993 ;; In the current version PROPERTIES is a property list.
994 ;; Reflect it into PLIST one by one while handling safe-chars
995 ;; specially.
996 (let ((safe-charsets (cdr (assq 'safe-charsets properties)))
997 (safe-chars (cdr (assq 'safe-chars properties)))
998 (l properties)
999 prop val)
1000 ;; If only safe-charsets is specified, make a char-table from
1001 ;; it, and store that char-table as the value of `safe-chars'.
1002 (if (and (not safe-chars) safe-charsets)
1003 (let (charset)
1004 (if (eq safe-charsets t)
1005 (setq safe-chars t)
1006 (setq safe-chars (make-char-table 'safe-chars))
1007 (while safe-charsets
1008 (setq charset (car safe-charsets)
1009 safe-charsets (cdr safe-charsets))
1010 (cond ((eq charset 'ascii)) ; just ignore
1011 ((eq charset 'eight-bit-control)
1012 (let ((i 128))
1013 (while (< i 160)
1014 (aset safe-chars i t)
1015 (setq i (1+ i)))))
1016 ((eq charset 'eight-bit-graphic)
1017 (let ((i 160))
1018 (while (< i 256)
1019 (aset safe-chars i t)
1020 (setq i (1+ i)))))
1021 (t
e6cddb1a
KH
1022 (aset safe-chars (make-char charset) t))))
1023 (if accept-latin-extra-code
1024 (let ((i 128))
1025 (while (< i 160)
1026 (if (aref latin-extra-code-table i)
1027 (aset safe-chars i t))
1028 (setq i (1+ i))))))
c11a8f77
KH
1029 (setq l (cons (cons 'safe-chars safe-chars) l))))
1030 (while l
1031 (setq prop (car (car l)) val (cdr (car l)) l (cdr l))
1032 (if (eq prop 'safe-chars)
1033 (progn
cc57cc54
KH
1034 (if (and (symbolp val)
1035 (get val 'translation-table))
1036 (setq safe-chars (get val 'translation-table)))
cc57cc54 1037 (setq val safe-chars)))
c11a8f77 1038 (plist-put plist prop val)))
1b46a680
KH
1039 ;; The property `coding-category' may have been set differently
1040 ;; through PROPERTIES.
1041 (setq coding-category (plist-get plist 'coding-category))
0269ddfb 1042 (aset coding-spec coding-spec-plist-idx plist))
6e9722b0 1043 (put coding-system 'coding-system coding-spec)
6e9722b0
KH
1044 (put coding-category 'coding-systems
1045 (cons coding-system (get coding-category 'coding-systems))))
4ed46869 1046
1b46a680 1047 ;; Next, set a value of `eol-type' property.
af873cd1 1048 (if (not eol-type)
1b46a680
KH
1049 ;; If EOL-TYPE is nil, set a vector of subsidiary coding
1050 ;; systems, each corresponds to a coding system for the detected
1051 ;; EOL format.
1052 (setq eol-type (make-subsidiary-coding-system coding-system)))
1053 (setq eol-type
1054 (cond ((or (eq eol-type 'unix) (null eol-type))
1055 0)
1056 ((eq eol-type 'dos)
1057 1)
1058 ((eq eol-type 'mac)
1059 2)
1060 ((or (and (vectorp eol-type)
1061 (= (length eol-type) 3))
1062 (and (numberp eol-type)
1063 (and (>= eol-type 0)
1064 (<= eol-type 2))))
1065 eol-type)
1066 (t
1067 (error "Invalid EOL-TYPE spec:%S" eol-type))))
1068 (put coding-system 'eol-type eol-type)
0269ddfb 1069
620956ca
KH
1070 (define-coding-system-internal coding-system)
1071
0269ddfb
KH
1072 ;; At last, register CODING-SYSTEM in `coding-system-list' and
1073 ;; `coding-system-alist'.
857ea15c 1074 (add-to-coding-system-list coding-system)
0269ddfb 1075 (setq coding-system-alist (cons (list (symbol-name coding-system))
d9e3229d 1076 coding-system-alist))
05134257
KH
1077
1078 ;; For a coding system of cateogory iso-8-1 and iso-8-2, create
1079 ;; XXX-with-esc variants.
1080 (let ((coding-category (coding-system-category coding-system)))
1081 (if (or (eq coding-category 'coding-category-iso-8-1)
1082 (eq coding-category 'coding-category-iso-8-2))
1083 (let ((esc (intern (concat (symbol-name coding-system) "-with-esc")))
c11a8f77
KH
1084 (doc (format "Same as %s but can handle any charsets by ISO's escape sequences." coding-system))
1085 (safe-charsets (assq 'safe-charsets properties))
1086 (mime-charset (assq 'mime-charset properties)))
1087 (if safe-charsets
1088 (setcdr safe-charsets t)
1089 (setq properties (cons (cons 'safe-charsets t) properties)))
1090 (if mime-charset
1091 (setcdr mime-charset nil))
05134257
KH
1092 (make-coding-system esc type mnemonic doc
1093 (if (listp (car flags))
1094 (cons (append (car flags) '(t)) (cdr flags))
1095 (cons (list (car flags) t) (cdr flags)))
c11a8f77 1096 properties))))
05134257 1097
d9e3229d 1098 coding-system)
8057896b 1099
5470d5b6
KH
1100(put 'safe-chars 'char-table-extra-slots 0)
1101
a42763dc 1102(defun define-coding-system-alias (alias coding-system)
358d28fb 1103 "Define ALIAS as an alias for coding system CODING-SYSTEM."
0269ddfb 1104 (put alias 'coding-system (coding-system-spec coding-system))
857ea15c 1105 (add-to-coding-system-list alias)
0269ddfb
KH
1106 (setq coding-system-alist (cons (list (symbol-name alias))
1107 coding-system-alist))
1108 (let ((eol-type (coding-system-eol-type coding-system)))
1109 (if (vectorp eol-type)
708fc465
KH
1110 (progn
1111 (nconc (coding-system-get alias 'alias-coding-systems) (list alias))
1112 (put alias 'eol-type (make-subsidiary-coding-system alias)))
0269ddfb 1113 (put alias 'eol-type eol-type))))
4ed46869 1114
bbdea948
RS
1115(defun merge-coding-systems (first second)
1116 "Fill in any unspecified aspects of coding system FIRST from SECOND.
1117Return the resulting coding system."
1118 (let ((base (coding-system-base second))
1119 (eol (coding-system-eol-type second)))
1120 ;; If FIRST doesn't specify text conversion, merge with that of SECOND.
1121 (if (eq (coding-system-base first) 'undecided)
1122 (setq first (coding-system-change-text-conversion first base)))
1123 ;; If FIRST doesn't specify eol conversion, merge with that of SECOND.
1124 (if (and (vectorp (coding-system-eol-type first))
1125 (numberp eol) (>= eol 0) (<= eol 2))
1126 (setq first (coding-system-change-eol-conversion
1127 first eol)))
1128 first))
1129
2b96ab33
KH
1130(defun autoload-coding-system (symbol form)
1131 "Define SYMBOL as a coding-system that is defined on demand.
1132
1133FROM is a form to evaluate to define the coding-system."
28849ffc
KH
1134 (put symbol 'coding-system-define-form form)
1135 (setq coding-system-alist (cons (list (symbol-name symbol))
1136 coding-system-alist)))
2b96ab33 1137
4ed46869 1138(defun set-buffer-file-coding-system (coding-system &optional force)
358d28fb
RS
1139 "Set the file coding-system of the current buffer to CODING-SYSTEM.
1140This means that when you save the buffer, it will be converted
1141according to CODING-SYSTEM. For a list of possible values of CODING-SYSTEM,
1142use \\[list-coding-systems].
1143
bbdea948
RS
1144If CODING-SYSTEM leaves the text conversion unspecified, or if it
1145leaves the end-of-line conversion unspecified, FORCE controls what to
1146do. If FORCE is nil, get the unspecified aspect (or aspects) from the
1147buffer's previous `buffer-file-coding-system' value (if it is
3b6cc699 1148specified there). Otherwise, leave it unspecified.
aeef8f07
KH
1149
1150This marks the buffer modified so that the succeeding \\[save-buffer]
1151surely saves the buffer with CODING-SYSTEM. From a program, if you
1152don't want to mark the buffer modified, just set the variable
1153`buffer-file-coding-system' directly."
bbdea948 1154 (interactive "zCoding system for saving file (default, nil): \nP")
4ed46869 1155 (check-coding-system coding-system)
36d455c4 1156 (if (and coding-system buffer-file-coding-system (null force))
bbdea948
RS
1157 (setq coding-system
1158 (merge-coding-systems coding-system buffer-file-coding-system)))
4ed46869 1159 (setq buffer-file-coding-system coding-system)
38a1356d
RS
1160 ;; This is in case of an explicit call. Normally, `normal-mode' and
1161 ;; `set-buffer-major-mode-hook' take care of setting the table.
1162 (if (fboundp 'ucs-set-table-for-input) ; don't lose when building
1163 (ucs-set-table-for-input))
4ed46869
KH
1164 (set-buffer-modified-p t)
1165 (force-mode-line-update))
1166
bbdea948
RS
1167(defun revert-buffer-with-coding-system (coding-system &optional force)
1168 "Visit the current buffer's file again using coding system CODING-SYSTEM.
1169For a list of possible values of CODING-SYSTEM, use \\[list-coding-systems].
1170
1171If CODING-SYSTEM leaves the text conversion unspecified, or if it
1172leaves the end-of-line conversion unspecified, FORCE controls what to
1173do. If FORCE is nil, get the unspecified aspect (or aspects) from the
1174buffer's previous `buffer-file-coding-system' value (if it is
1175specified there). Otherwise, determine it from the file contents as
1176usual for visiting a file."
1177 (interactive "zCoding system for visited file (default, nil): \nP")
1178 (check-coding-system coding-system)
1179 (if (and coding-system buffer-file-coding-system (null force))
1180 (setq coding-system
1181 (merge-coding-systems coding-system buffer-file-coding-system)))
1182 (let ((coding-system-for-read coding-system))
1183 (revert-buffer)))
1184
701414e3
KH
1185(defun set-file-name-coding-system (coding-system)
1186 "Set coding system for decoding and encoding file names to CODING-SYSTEM.
1187It actually just set the variable `file-name-coding-system' (which
1188see) to CODING-SYSTEM."
1189 (interactive "zCoding system for file names (default, nil): ")
1190 (check-coding-system coding-system)
1191 (setq file-name-coding-system coding-system))
1192
358d28fb
RS
1193(defvar default-terminal-coding-system nil
1194 "Default value for the terminal coding system.
1195This is normally set according to the selected language environment.
1196See also the command `set-terminal-coding-system'.")
1197
df100398
KH
1198(defun set-terminal-coding-system (coding-system)
1199 "Set coding system of your terminal to CODING-SYSTEM.
358d28fb
RS
1200All text output to the terminal will be encoded
1201with the specified coding system.
1202For a list of possible values of CODING-SYSTEM, use \\[list-coding-systems].
1203The default is determined by the selected language environment
1204or by the previous use of this command."
1205 (interactive
2e02a76f
RS
1206 (list (let ((default (if (and (not (terminal-coding-system))
1207 default-terminal-coding-system)
1208 default-terminal-coding-system)))
1209 (read-coding-system
1210 (format "Coding system for terminal display (default, %s): "
1211 default)
1212 default))))
358d28fb
RS
1213 (if (and (not coding-system)
1214 (not (terminal-coding-system)))
1215 (setq coding-system default-terminal-coding-system))
1216 (if coding-system
521d4010 1217 (setq default-terminal-coding-system coding-system))
df100398
KH
1218 (set-terminal-coding-system-internal coding-system)
1219 (redraw-frame (selected-frame)))
1220
358d28fb
RS
1221(defvar default-keyboard-coding-system nil
1222 "Default value of the keyboard coding system.
1223This is normally set according to the selected language environment.
1224See also the command `set-keyboard-coding-system'.")
1225
df100398 1226(defun set-keyboard-coding-system (coding-system)
358d28fb
RS
1227 "Set coding system for keyboard input to CODING-SYSTEM.
1228In addition, this command enables Encoded-kbd minor mode.
6d34f495
DL
1229\(If CODING-SYSTEM is nil, Encoded-kbd mode is turned off -- see
1230`encoded-kbd-mode'.)
358d28fb
RS
1231For a list of possible values of CODING-SYSTEM, use \\[list-coding-systems].
1232The default is determined by the selected language environment
1233or by the previous use of this command."
1234 (interactive
2e02a76f
RS
1235 (list (let ((default (if (and (not (keyboard-coding-system))
1236 default-keyboard-coding-system)
1237 default-keyboard-coding-system)))
1238 (read-coding-system
1239 (format "Coding system for keyboard input (default, %s): "
1240 default)
1241 default))))
358d28fb
RS
1242 (if (and (not coding-system)
1243 (not (keyboard-coding-system)))
1244 (setq coding-system default-keyboard-coding-system))
1245 (if coding-system
1246 (setq default-keyboard-coding-system coding-system))
df100398 1247 (set-keyboard-coding-system-internal coding-system)
b23bad0b 1248 (setq keyboard-coding-system coding-system)
df100398
KH
1249 (encoded-kbd-mode (if coding-system 1 0)))
1250
6d34f495
DL
1251(defcustom keyboard-coding-system nil
1252 "Specify coding system for keyboard input.
1253If you set this on a terminal which can't distinguish Meta keys from
12548-bit characters, you will have to use ESC to type Meta characters.
1255See Info node `Specify Coding' and Info node `Single-Byte Character Support'.
1256
237e5993
DL
1257On non-windowing terminals, this is set from the locale by default.
1258
6d34f495 1259Setting this variable directly does not take effect;
c4131d96 1260use either \\[customize] or \\[set-keyboard-coding-system]."
6d34f495
DL
1261 :type '(coding-system :tag "Coding system")
1262 :link '(info-link "(emacs)Specify Coding")
1263 :link '(info-link "(emacs)Single-Byte Character Support")
1264 :set (lambda (symbol value)
1265 ;; Don't load encoded-kbd-mode unnecessarily.
1266 (if (or value (boundp 'encoded-kbd-mode))
1267 (set-keyboard-coding-system value)
1268 (set-default 'keyboard-coding-system nil))) ; must initialize
237e5993 1269 :version "21.4"
6d34f495
DL
1270 :group 'keyboard
1271 :group 'mule)
1272
df100398 1273(defun set-buffer-process-coding-system (decoding encoding)
358d28fb 1274 "Set coding systems for the process associated with the current buffer.
df100398 1275DECODING is the coding system to be used to decode input from the process,
358d28fb
RS
1276ENCODING is the coding system to be used to encode output to the process.
1277
1278For a list of possible values of CODING-SYSTEM, use \\[list-coding-systems]."
4ed46869 1279 (interactive
83911021 1280 "zCoding-system for output from the process: \nzCoding-system for input to the process: ")
4ed46869
KH
1281 (let ((proc (get-buffer-process (current-buffer))))
1282 (if (null proc)
521d4010 1283 (error "No process")
df100398
KH
1284 (check-coding-system decoding)
1285 (check-coding-system encoding)
1286 (set-process-coding-system proc decoding encoding)))
4ed46869
KH
1287 (force-mode-line-update))
1288
d0b99881
RS
1289(defalias 'set-clipboard-coding-system 'set-selection-coding-system)
1290
14915c37 1291(defun set-selection-coding-system (coding-system)
8c52d564 1292 "Make CODING-SYSTEM used for communicating with other X clients.
b25eef20
KH
1293When sending or receiving text via cut_buffer, selection, and clipboard,
1294the text is encoded or decoded by CODING-SYSTEM."
a03b3ce1 1295 (interactive "zCoding system for X selection: ")
b25eef20 1296 (check-coding-system coding-system)
14915c37 1297 (setq selection-coding-system coding-system))
b25eef20 1298
e8dd0160 1299;; Coding system lastly specified by the command
a03b3ce1
KH
1300;; set-next-selection-coding-system.
1301(defvar last-next-selection-coding-system nil)
1302
1303(defun set-next-selection-coding-system (coding-system)
1304 "Make CODING-SYSTEM used for the next communication with other X clients.
1305This setting is effective for the next communication only."
1306 (interactive
1307 (list (read-coding-system
1308 (if last-next-selection-coding-system
1309 (format "Coding system for the next X selection (default, %S): "
1310 last-next-selection-coding-system)
1311 "Coding system for the next X selection: ")
1312 last-next-selection-coding-system)))
1313 (if coding-system
1314 (setq last-next-selection-coding-system coding-system)
1315 (setq coding-system last-next-selection-coding-system))
1316 (check-coding-system coding-system)
1317
1318 (setq next-selection-coding-system coding-system))
1319
4ed46869 1320(defun set-coding-priority (arg)
521d4010
DL
1321 "Set priority of coding categories according to ARG.
1322ARG is a list of coding categories ordered by priority."
d9e3229d
KH
1323 (let ((l arg)
1324 (current-list (copy-sequence coding-category-list)))
e8dd0160 1325 ;; Check the validity of ARG while deleting coding categories in
d9e3229d
KH
1326 ;; ARG from CURRENT-LIST. We assume that CODING-CATEGORY-LIST
1327 ;; contains all coding categories.
1328 (while l
1329 (if (or (null (get (car l) 'coding-category-index))
1330 (null (memq (car l) current-list)))
1331 (error "Invalid or duplicated element in argument: %s" arg))
1332 (setq current-list (delq (car l) current-list))
1333 (setq l (cdr l)))
4ed46869 1334 ;; Update `coding-category-list' and return it.
2feaf204
KH
1335 (setq coding-category-list (append arg current-list))
1336 (set-coding-priority-internal)))
4ed46869 1337
835cbadb
EZ
1338;;; X selections
1339
cc926903 1340(defvar ctext-non-standard-encodings-alist
6c805023
KH
1341 '(("big5-0" big5 2 (chinese-big5-1 chinese-big5-2))
1342 ("ISO8859-14" iso-8859-14 1 latin-iso8859-14)
1343 ("ISO8859-15" iso-8859-15 1 latin-iso8859-15))
1344 "Alist of non-standard encoding names vs the corresponding usages in CTEXT.
1345
1346It controls how extended segments of a compound text are handled
1347by the coding system `compound-text-with-extensions'.
1348
1349Each element has the form (ENCODING-NAME CODING-SYSTEM N-OCTET CHARSET).
1350
1351ENCODING-NAME is an encoding name of an \"extended segments\".
1352
1353CODING-SYSTEM is the coding-system to encode (or decode) the
1354characters into (or from) the extended segment.
1355
1356N-OCTET is the number of octets (bytes) that encodes a character
1357in the segment. It can be 0 (meaning the number of octets per
1358character is variable), 1, 2, 3, or 4.
1359
1360CHARSET is a charater set containing characters that are encoded
1361in the segment. It can be a list of character sets. It can also
1362be a char-table, in which case characters that have non-nil value
1363in the char-table are the target.
1364
1365On decoding CTEXT, all encoding names listed here are recognized.
1366
1367On encoding CTEXT, encoding names in the variable
1368`ctext-non-standard-encodings' (which see) and in the information
1369listed for the current language environment under the key
1370`ctext-non-standard-encodings' are used.")
1371
1372(defvar ctext-non-standard-encodings
1373 '("big5-0")
1374 "List of non-standard encoding names used in extended segments of CTEXT.
1375Each element must be one of the names listed in the variable
1376`ctext-non-standard-encodings-alist' (which see).")
cc926903
KH
1377
1378(defvar ctext-non-standard-encodings-regexp
1379 (string-to-multibyte
1380 (concat
1381 ;; For non-standard encodings.
1382 "\\(\e%/[0-4][\200-\377][\200-\377]\\([^\002]+\\)\002\\)"
1383 "\\|"
1384 ;; For UTF-8 encoding.
1385 "\\(\e%G[^\e]*\e%@\\)")))
835cbadb
EZ
1386
1387;; Functions to support "Non-Standard Character Set Encodings" defined
6c805023
KH
1388;; by the COMPOUND-TEXT spec. They also support "The UTF-8 encoding"
1389;; described in the section 7 of the documentation of COMPOUND-TEXT
1390;; distributed with XFree86.
5c88a01e 1391
835cbadb
EZ
1392(defun ctext-post-read-conversion (len)
1393 "Decode LEN characters encoded as Compound Text with Extended Segments."
835cbadb 1394 (save-match-data
cc926903
KH
1395 (save-restriction
1396 (let ((case-fold-search nil)
1397 (in-workbuf (string= (buffer-name) " *code-converting-work*"))
1398 last-coding-system-used
1399 pos bytes)
1400 (or in-workbuf
1401 (narrow-to-region (point) (+ (point) len)))
cc926903
KH
1402 (if in-workbuf
1403 (set-buffer-multibyte t))
1404 (while (re-search-forward ctext-non-standard-encodings-regexp
1405 nil 'move)
1406 (setq pos (match-beginning 0))
1407 (if (match-beginning 1)
1408 ;; ESC % / [0-4] M L --ENCODING-NAME-- \002 --BYTES--
1409 (let* ((M (char-after (+ pos 4)))
1410 (L (char-after (+ pos 5)))
1411 (encoding (match-string 2))
9857367f 1412 (encoding-info (assoc-string
6c805023 1413 encoding
c12bc1fb 1414 ctext-non-standard-encodings-alist t))
6c805023
KH
1415 (coding (if encoding-info
1416 (nth 1 encoding-info)
1417 (setq encoding (intern (downcase encoding)))
1418 (and (coding-system-p encoding)
1419 encoding))))
cc926903
KH
1420 (setq bytes (- (+ (* (- M 128) 128) (- L 128))
1421 (- (point) (+ pos 6))))
1422 (when coding
1423 (delete-region pos (point))
1424 (forward-char bytes)
1425 (decode-coding-region (- (point) bytes) (point) coding)))
1426 ;; ESC % G --UTF-8-BYTES-- ESC % @
6c805023
KH
1427 (delete-char -3)
1428 (delete-region pos (+ pos 3))
1429 (decode-coding-region pos (point) 'utf-8))))
cc926903
KH
1430 (goto-char (point-min))
1431 (- (point-max) (point)))))
835cbadb 1432
6c805023
KH
1433;; Return a char table of extended segment usage for each character.
1434;; Each value of the char table is nil, one of the elements of
1435;; `ctext-non-standard-encodings-alist', or the symbol `utf-8'.
1436
1437(defun ctext-non-standard-encodings-table ()
1438 (let ((table (make-char-table 'translation-table)))
1439 (aset table (make-char 'mule-unicode-0100-24ff) 'utf-8)
1440 (aset table (make-char 'mule-unicode-2500-33ff) 'utf-8)
1441 (aset table (make-char 'mule-unicode-e000-ffff) 'utf-8)
1442 (dolist (encoding (reverse
1443 (append
1444 (get-language-info current-language-environment
1445 'ctext-non-standard-encodings)
1446 ctext-non-standard-encodings)))
1447 (let* ((slot (assoc encoding ctext-non-standard-encodings-alist))
1448 (charset (nth 3 slot)))
1449 (if charset
1450 (cond ((charsetp charset)
1451 (aset table (make-char charset) slot))
1452 ((listp charset)
1453 (dolist (elt charset)
1454 (aset table (make-char elt) slot)))
1455 ((char-table-p charset)
9857367f 1456 (map-char-table #'(lambda (k v)
6c805023
KH
1457 (if (and v (> k 128)) (aset table k slot)))
1458 charset))))))
1459 table))
835cbadb
EZ
1460
1461(defun ctext-pre-write-conversion (from to)
5dde3c71
EZ
1462 "Encode characters between FROM and TO as Compound Text w/Extended Segments.
1463
1464If FROM is a string, or if the current buffer is not the one set up for us
cc926903 1465by encode-coding-string, generate a new temp buffer, insert the
5dde3c71 1466text, and convert it in the temporary buffer. Otherwise, convert in-place."
835cbadb 1467 (save-match-data
cc926903
KH
1468 ;; Setup a working buffer if necessary.
1469 (cond ((stringp from)
1470 (let ((buf (current-buffer)))
1471 (set-buffer (generate-new-buffer " *temp"))
1472 (set-buffer-multibyte (multibyte-string-p from))
1473 (insert from)))
1474 ((not (string= (buffer-name) " *code-converting-work*"))
1475 (let ((buf (current-buffer))
1476 (multibyte enable-multibyte-characters))
1477 (set-buffer (generate-new-buffer " *temp"))
1478 (set-buffer-multibyte multibyte)
1479 (insert-buffer-substring buf from to))))
1480
1481 ;; Now we can encode the whole buffer.
6c805023 1482 (let ((encoding-table (ctext-non-standard-encodings-table))
cc926903 1483 last-coding-system-used
6c805023
KH
1484 last-pos last-encoding-info
1485 encoding-info end-pos)
1486 (goto-char (setq last-pos (point-min)))
1487 (setq end-pos (point-marker))
1488 (while (re-search-forward "[^\000-\177]+" nil t)
1489 ;; Found a sequence of non-ASCII characters.
1490 (setq last-pos (match-beginning 0)
1491 last-encoding-info (aref encoding-table (char-after last-pos)))
1492 (set-marker end-pos (match-end 0))
1493 (goto-char (1+ last-pos))
1494 (catch 'tag
1495 (while t
1496 (setq encoding-info
1497 (if (< (point) end-pos)
1498 (aref encoding-table (following-char))))
1499 (unless (eq last-encoding-info encoding-info)
1500 (cond ((consp last-encoding-info)
1501 ;; Encode the previous range using an extended
1502 ;; segment.
1503 (let ((encoding-name (car last-encoding-info))
1504 (coding-system (nth 1 last-encoding-info))
1505 (noctets (nth 2 last-encoding-info))
1506 len)
1507 (encode-coding-region last-pos (point) coding-system)
1508 (setq len (+ (length encoding-name) 1
1509 (- (point) last-pos)))
1510 (save-excursion
1511 (goto-char last-pos)
9857367f 1512 (insert (string-to-multibyte
6c805023
KH
1513 (format "\e%%/%d%c%c%s\ 2"
1514 noctets
1515 (+ (/ len 128) 128)
1516 (+ (% len 128) 128)
1517 encoding-name))))))
1518 ((eq last-encoding-info 'utf-8)
1519 ;; Encode the previous range using UTF-8 encoding
1520 ;; extention.
1521 (encode-coding-region last-pos (point) 'mule-utf-8)
1522 (save-excursion
1523 (goto-char last-pos)
1524 (insert "\e%G"))
1525 (insert "\e%@")))
1526 (setq last-pos (point)
1527 last-encoding-info encoding-info))
1528 (if (< (point) end-pos)
1529 (forward-char 1)
1530 (throw 'tag nil)))))
1531 (set-marker end-pos nil)
cc926903 1532 (goto-char (point-min))))
5dde3c71 1533 ;; Must return nil, as build_annotations_2 expects that.
835cbadb
EZ
1534 nil)
1535
4ed46869
KH
1536;;; FILE I/O
1537
e76938e7 1538(defcustom auto-coding-alist
0bca779a 1539 '(("\\.\\(arc\\|zip\\|lzh\\|zoo\\|jar\\|sx[dmicw]\\|tar\\|tgz\\)\\'" . no-conversion)
6caef2da 1540 ("\\.\\(gz\\|Z\\|bz\\|bz2\\|gpg\\)\\'" . no-conversion)
45885400 1541 ("/#[^/]+#\\'" . emacs-mule))
835f49b8
KH
1542 "Alist of filename patterns vs corresponding coding systems.
1543Each element looks like (REGEXP . CODING-SYSTEM).
558b0c86 1544A file whose name matches REGEXP is decoded by CODING-SYSTEM on reading.
835f49b8 1545
7fed493a
RS
1546The settings in this alist take priority over `coding:' tags
1547in the file (see the function `set-auto-coding')
e76938e7
DL
1548and the contents of `file-coding-system-alist'."
1549 :group 'files
1550 :group 'mule
1551 :type '(repeat (cons (regexp :tag "File name regexp")
1552 (symbol :tag "Coding system"))))
835f49b8 1553
502522b2
GM
1554(defcustom auto-coding-regexp-alist
1555 '(("^BABYL OPTIONS:[ \t]*-\\*-[ \t]*rmail[ \t]*-\\*-" . no-conversion))
1556 "Alist of patterns vs corresponding coding systems.
1557Each element looks like (REGEXP . CODING-SYSTEM).
1558A file whose first bytes match REGEXP is decoded by CODING-SYSTEM on reading.
1559
1560The settings in this alist take priority over `coding:' tags
1561in the file (see the function `set-auto-coding')
1562and the contents of `file-coding-system-alist'."
1563 :group 'files
1564 :group 'mule
1565 :type '(repeat (cons (regexp :tag "Regexp")
1566 (symbol :tag "Coding system"))))
1567
d9f6dfe6 1568;; See the bottom of this file for built-in auto coding functions.
447404a3
CW
1569(defcustom auto-coding-functions '(sgml-xml-auto-coding-function
1570 sgml-html-meta-auto-coding-function)
d9f6dfe6
CW
1571 "A list of functions which attempt to determine a coding system.
1572
66643502
RS
1573Each function in this list should be written to operate on the
1574current buffer, but should not modify it in any way. The buffer
1575will contain undecoded text of parts of the file. Each function
1576should take one argument, SIZE, which says how many
1577characters (starting from point) it should look at.
1578
1579If one of these functions succeeds in determining a coding
1580system, it should return that coding system. Otherwise, it
1581should return nil.
1582
1583If a file has a `coding:' tag, that takes precedence over these
1584functions, so they won't be called at all."
d9f6dfe6
CW
1585 :group 'files
1586 :group 'mule
1587 :type '(repeat function))
1588
1c4cc63a
KH
1589(defvar set-auto-coding-for-load nil
1590 "Non-nil means look for `load-coding' property instead of `coding'.
1591This is used for loading and byte-compiling Emacs Lisp files.")
1592
8a592131
RS
1593(defun auto-coding-alist-lookup (filename)
1594 "Return the coding system specified by `auto-coding-alist' for FILENAME."
1595 (let ((alist auto-coding-alist)
c60ee5e7 1596 (case-fold-search (memq system-type '(vax-vms windows-nt ms-dos cygwin)))
8a592131
RS
1597 coding-system)
1598 (while (and alist (not coding-system))
1599 (if (string-match (car (car alist)) filename)
1600 (setq coding-system (cdr (car alist)))
1601 (setq alist (cdr alist))))
1602 coding-system))
1603
835f49b8
KH
1604(defun set-auto-coding (filename size)
1605 "Return coding system for a file FILENAME of which SIZE bytes follow point.
1c4cc63a
KH
1606These bytes should include at least the first 1k of the file
1607and the last 3k of the file, but the middle may be omitted.
63561304 1608
d21ba5e0
DL
1609The function checks FILENAME against the variable `auto-coding-alist'.
1610If FILENAME doesn't match any entries in the variable, it checks the
502522b2 1611contents of the current buffer following point against
447404a3 1612`auto-coding-regexp-alist'. If no match is found, it checks for a
502522b2 1613`coding:' tag in the first one or two lines following point. If no
d21ba5e0 1614`coding:' tag is found, it checks any local variables list in the last
447404a3 16153K bytes out of the SIZE bytes. Finally, if none of these methods
d21ba5e0
DL
1616succeed, it checks to see if any function in `auto-coding-functions'
1617gives a match.
63561304 1618
d21ba5e0
DL
1619The return value is the specified coding system, or nil if nothing is
1620specified.
87aba788 1621
ba74e833 1622The variable `set-auto-coding-function' (which see) is set to this
87aba788 1623function by default."
502522b2 1624 (or (auto-coding-alist-lookup filename)
447404a3
CW
1625 ;; Try using `auto-coding-regexp-alist'.
1626 (save-excursion
1627 (let ((alist auto-coding-regexp-alist)
1628 coding-system)
1629 (while (and alist (not coding-system))
1630 (let ((regexp (car (car alist))))
1631 (when (re-search-forward regexp (+ (point) size) t)
1632 (setq coding-system (cdr (car alist)))))
0bca779a 1633 (setq alist (cdr alist)))
447404a3 1634 coding-system))
502522b2
GM
1635 (let* ((case-fold-search t)
1636 (head-start (point))
1637 (head-end (+ head-start (min size 1024)))
1638 (tail-start (+ head-start (max (- size 3072) 0)))
1639 (tail-end (+ head-start size))
1640 coding-system head-found tail-found pos)
1641 ;; Try a short cut by searching for the string "coding:"
1642 ;; and for "unibyte:" at the head and tail of SIZE bytes.
1643 (setq head-found (or (search-forward "coding:" head-end t)
1644 (search-forward "unibyte:" head-end t)))
1645 (if (and head-found (> head-found tail-start))
1646 ;; Head and tail are overlapped.
1647 (setq tail-found head-found)
1648 (goto-char tail-start)
1649 (setq tail-found (or (search-forward "coding:" tail-end t)
1650 (search-forward "unibyte:" tail-end t))))
1651
1652 ;; At first check the head.
1653 (when head-found
1654 (goto-char head-start)
6b66d028
RS
1655 (setq head-end (set-auto-mode-1))
1656 (setq head-start (point))
1d8e9a7c 1657 (when (and head-end (< head-found head-end))
835f49b8 1658 (goto-char head-start)
502522b2
GM
1659 (when (and set-auto-coding-for-load
1660 (re-search-forward
6b66d028 1661 "\\(.*;\\)?[ \t]*unibyte:[ \t]*\\([^ ;]+\\)"
502522b2
GM
1662 head-end t))
1663 (setq coding-system 'raw-text))
1664 (when (and (not coding-system)
1665 (re-search-forward
6b66d028 1666 "\\(.*;\\)?[ \t]*coding:[ \t]*\\([^ ;]+\\)"
502522b2
GM
1667 head-end t))
1668 (setq coding-system (intern (match-string 2)))
1669 (or (coding-system-p coding-system)
1670 (setq coding-system nil)))))
1671
1672 ;; If no coding: tag in the head, check the tail.
c251286b
KH
1673 ;; Here we must pay attention to the case that the end-of-line
1674 ;; is just "\r" and we can't use "^" nor "$" in regexp.
502522b2
GM
1675 (when (and tail-found (not coding-system))
1676 (goto-char tail-start)
c251286b 1677 (re-search-forward "[\r\n]\^L" nil t)
502522b2 1678 (if (re-search-forward
9857367f 1679 "[\r\n]\\([^[\r\n]*\\)[ \t]*Local Variables:[ \t]*\\([^\r\n]*\\)[\r\n]"
c251286b
KH
1680 tail-end t)
1681 ;; The prefix is what comes before "local variables:" in its
1682 ;; line. The suffix is what comes after "local variables:"
502522b2
GM
1683 ;; in its line.
1684 (let* ((prefix (regexp-quote (match-string 1)))
1685 (suffix (regexp-quote (match-string 2)))
1686 (re-coding
1687 (concat
c251286b 1688 "[\r\n]" prefix
cfe98f50
GM
1689 ;; N.B. without the \n below, the regexp can
1690 ;; eat newlines.
c251286b
KH
1691 "[ \t]*coding[ \t]*:[ \t]*\\([^ \t\r\n]+\\)[ \t]*"
1692 suffix "[\r\n]"))
502522b2
GM
1693 (re-unibyte
1694 (concat
c251286b
KH
1695 "[\r\n]" prefix
1696 "[ \t]*unibyte[ \t]*:[ \t]*\\([^ \t\r\n]+\\)[ \t]*"
1697 suffix "[\r\n]"))
502522b2 1698 (re-end
9857367f 1699 (concat "[\r\n]" prefix "[ \t]*End *:[ \t]*" suffix
c251286b
KH
1700 "[\r\n]?"))
1701 (pos (1- (point))))
1702 (forward-char -1) ; skip back \r or \n.
502522b2
GM
1703 (re-search-forward re-end tail-end 'move)
1704 (setq tail-end (point))
1705 (goto-char pos)
1706 (when (and set-auto-coding-for-load
1707 (re-search-forward re-unibyte tail-end t))
1708 (setq coding-system 'raw-text))
1709 (when (and (not coding-system)
1710 (re-search-forward re-coding tail-end t))
1711 (setq coding-system (intern (match-string 1)))
1712 (or (coding-system-p coding-system)
1713 (setq coding-system nil))))))
447404a3
CW
1714 coding-system)
1715 ;; Finally, try all the `auto-coding-functions'.
1716 (let ((funcs auto-coding-functions)
1717 (coding-system nil))
1718 (while (and funcs (not coding-system))
1719 (setq coding-system (condition-case e
1720 (save-excursion
1721 (goto-char (point-min))
1722 (funcall (pop funcs) size))
1723 (error nil))))
502522b2 1724 coding-system)))
63561304
KH
1725
1726(setq set-auto-coding-function 'set-auto-coding)
87aba788 1727
c36b37ed 1728(defun after-insert-file-set-coding (inserted)
872a0a6f
RS
1729 "Set `buffer-file-coding-system' of current buffer after text is inserted.
1730INSERTED is the number of characters that were inserted, as figured
1731in the situation before this function. Return the number of characters
1732inserted, as figured in the situation after. The two numbers can be
1733different if the buffer has become unibyte."
4ed46869
KH
1734 (if last-coding-system-used
1735 (let ((coding-system
1736 (find-new-buffer-file-coding-system last-coding-system-used))
1737 (modified-p (buffer-modified-p)))
0269ddfb 1738 (when coding-system
71983219 1739 (set-buffer-file-coding-system coding-system t)
d0c26c63
KH
1740 (if (and enable-multibyte-characters
1741 (or (eq coding-system 'no-conversion)
ccb77b4e 1742 (eq (coding-system-type coding-system) 5))
136e48e4
KH
1743 ;; If buffer was unmodified and the size is the
1744 ;; same as INSERTED, we must be visiting it.
1745 (not modified-p)
1746 (= (buffer-size) inserted))
ccb77b4e
RS
1747 ;; For coding systems no-conversion and raw-text...,
1748 ;; edit the buffer as unibyte.
872a0a6f 1749 (let ((pos-marker (copy-marker (+ (point) inserted))))
d0c26c63 1750 (set-buffer-multibyte nil)
872a0a6f 1751 (setq inserted (- pos-marker (point)))))
0269ddfb 1752 (set-buffer-modified-p modified-p))))
d0c26c63 1753 inserted)
4ed46869 1754
8057896b 1755;; The coding-spec and eol-type of coding-system returned is decided
4ed46869
KH
1756;; independently in the following order.
1757;; 1. That of buffer-file-coding-system locally bound.
1758;; 2. That of CODING.
1759
1760(defun find-new-buffer-file-coding-system (coding)
1761 "Return a coding system for a buffer when a file of CODING is inserted.
a73a8c89
KH
1762The local variable `buffer-file-coding-system' of the current buffer
1763is set to the returned value.
509064c5 1764Return nil if there's no need to set `buffer-file-coding-system'."
4ed46869 1765 (let (local-coding local-eol
b685f8d6 1766 found-coding found-eol
4ed46869
KH
1767 new-coding new-eol)
1768 (if (null coding)
1769 ;; Nothing found about coding.
1770 nil
1771
b685f8d6
RS
1772 ;; Get information of `buffer-file-coding-system' in LOCAL-EOL
1773 ;; and LOCAL-CODING.
1774 (setq local-eol (coding-system-eol-type buffer-file-coding-system))
1775 (if (null (numberp local-eol))
1776 ;; But eol-type is not yet set.
1777 (setq local-eol nil))
0269ddfb
KH
1778 (if (and buffer-file-coding-system
1779 (not (eq (coding-system-type buffer-file-coding-system) t)))
1780 ;; This is not `undecided'.
1781 (setq local-coding (coding-system-base buffer-file-coding-system)))
b685f8d6
RS
1782
1783 (if (and (local-variable-p 'buffer-file-coding-system)
1784 local-eol local-coding)
4ed46869
KH
1785 ;; The current buffer has already set full coding-system, we
1786 ;; had better not change it.
1787 nil
1788
8057896b 1789 (setq found-eol (coding-system-eol-type coding))
4ed46869 1790 (if (null (numberp found-eol))
be02cd54
EZ
1791 ;; But eol-type is not found.
1792 ;; If EOL conversions are inhibited, force unix eol-type.
1793 (setq found-eol (if inhibit-eol-conversion 0)))
c76b5c99
KH
1794 (if (eq (coding-system-type coding) t)
1795 (setq found-coding 'undecided)
1796 (setq found-coding (coding-system-base coding)))
1797
1798 (if (and (not found-eol) (eq found-coding 'undecided))
1799 ;; No valid coding information found.
1800 nil
1801
1802 ;; Some coding information (eol or text) found.
1803
1804 ;; The local setting takes precedence over the found one.
1805 (setq new-coding (if (local-variable-p 'buffer-file-coding-system)
1806 (or local-coding found-coding)
1807 (or found-coding local-coding)))
1808 (setq new-eol (if (local-variable-p 'buffer-file-coding-system)
1809 (or local-eol found-eol)
1810 (or found-eol local-eol)))
1811
1812 (let ((eol-type (coding-system-eol-type new-coding)))
1813 (if (and (numberp new-eol) (vectorp eol-type))
1814 (aref eol-type new-eol)
1815 new-coding)))))))
4ed46869 1816
fe831d33
GV
1817(defun modify-coding-system-alist (target-type regexp coding-system)
1818 "Modify one of look up tables for finding a coding system on I/O operation.
8c453b46
RS
1819There are three of such tables, `file-coding-system-alist',
1820`process-coding-system-alist', and `network-coding-system-alist'.
fe831d33
GV
1821
1822TARGET-TYPE specifies which of them to modify.
8c453b46
RS
1823If it is `file', it affects `file-coding-system-alist' (which see).
1824If it is `process', it affects `process-coding-system-alist' (which see).
e8dd0160 1825If it is `network', it affects `network-coding-system-alist' (which see).
fe831d33
GV
1826
1827REGEXP is a regular expression matching a target of I/O operation.
1828The target is a file name if TARGET-TYPE is `file', a program name if
1829TARGET-TYPE is `process', or a network service name or a port number
1830to connect to if TARGET-TYPE is `network'.
1831
1832CODING-SYSTEM is a coding system to perform code conversion on the I/O
8c453b46
RS
1833operation, or a cons cell (DECODING . ENCODING) specifying the coding systems
1834for decoding and encoding respectively,
1835or a function symbol which, when called, returns such a cons cell."
fe831d33
GV
1836 (or (memq target-type '(file process network))
1837 (error "Invalid target type: %s" target-type))
1838 (or (stringp regexp)
1839 (and (eq target-type 'network) (integerp regexp))
1840 (error "Invalid regular expression: %s" regexp))
1841 (if (symbolp coding-system)
1842 (if (not (fboundp coding-system))
1843 (progn
1844 (check-coding-system coding-system)
1845 (setq coding-system (cons coding-system coding-system))))
1846 (check-coding-system (car coding-system))
1847 (check-coding-system (cdr coding-system)))
1848 (cond ((eq target-type 'file)
1849 (let ((slot (assoc regexp file-coding-system-alist)))
1850 (if slot
1851 (setcdr slot coding-system)
1852 (setq file-coding-system-alist
1853 (cons (cons regexp coding-system)
1854 file-coding-system-alist)))))
1855 ((eq target-type 'process)
1856 (let ((slot (assoc regexp process-coding-system-alist)))
1857 (if slot
1858 (setcdr slot coding-system)
1859 (setq process-coding-system-alist
1860 (cons (cons regexp coding-system)
1861 process-coding-system-alist)))))
1862 (t
1863 (let ((slot (assoc regexp network-coding-system-alist)))
1864 (if slot
1865 (setcdr slot coding-system)
1866 (setq network-coding-system-alist
1867 (cons (cons regexp coding-system)
1868 network-coding-system-alist)))))))
1869
db046b7d
KH
1870(defun decode-coding-inserted-region (from to filename
1871 &optional visit beg end replace)
f29387e8
KH
1872 "Decode the region between FROM and TO as if it is read from file FILENAME.
1873Optional arguments VISIT, BEG, END, and REPLACE are the same as those
1874of the function `insert-file-contents'."
1875 (save-excursion
1876 (save-restriction
1877 (narrow-to-region from to)
1878 (goto-char (point-min))
1879 (let ((coding coding-system-for-read))
1880 (or coding
1881 (setq coding (funcall set-auto-coding-function
1882 filename (- (point-max) (point-min)))))
1883 (or coding
b69ff51b
KH
1884 (setq coding (car (find-operation-coding-system
1885 'insert-file-contents
1886 filename visit beg end replace))))
f29387e8
KH
1887 (if (coding-system-p coding)
1888 (or enable-multibyte-characters
1889 (setq coding
1890 (coding-system-change-text-conversion coding 'raw-text)))
1891 (setq coding nil))
1892 (if coding
b12e19b2
KH
1893 (decode-coding-region (point-min) (point-max) coding)
1894 (setq last-coding-system-used coding))))))
f29387e8 1895
b25eef20 1896(defun make-translation-table (&rest args)
a284eea3 1897 "Make a translation table from arguments.
d38b07f9 1898A translation table is a char table intended for character
a284eea3
DL
1899translation in CCL programs.
1900
d38b07f9 1901Each argument is a list of elements of the form (FROM . TO), where FROM
a284eea3 1902is a character to be translated to TO.
13d5617d 1903
452fdb31 1904FROM can be a generic character (see `make-char'). In this case, TO is
d38b07f9 1905a generic character containing the same number of characters, or an
452fdb31 1906ordinary character. If FROM and TO are both generic characters, all
b25eef20 1907characters belonging to FROM are translated to characters belonging to TO
4e003d37
KH
1908without changing their position code(s).
1909
1910The arguments and forms in each argument are processed in the given
1911order, and if a previous form already translates TO to some other
1912character, say TO-ALT, FROM is also translated to TO-ALT."
f967223b 1913 (let ((table (make-char-table 'translation-table))
a73a8c89
KH
1914 revlist)
1915 (while args
1916 (let ((elts (car args)))
1917 (while elts
13d5617d
KH
1918 (let* ((from (car (car elts)))
1919 (from-i 0) ; degree of freedom of FROM
1920 (from-rev (nreverse (split-char from)))
1921 (to (cdr (car elts)))
1922 (to-i 0) ; degree of freedom of TO
1923 (to-rev (nreverse (split-char to))))
1924 ;; Check numbers of heading 0s in FROM-REV and TO-REV.
1925 (while (eq (car from-rev) 0)
1926 (setq from-i (1+ from-i) from-rev (cdr from-rev)))
1927 (while (eq (car to-rev) 0)
1928 (setq to-i (1+ to-i) to-rev (cdr to-rev)))
1929 (if (and (/= from-i to-i) (/= to-i 0))
1930 (error "Invalid character pair (%d . %d)" from to))
b25eef20
KH
1931 ;; If we have already translated TO to TO-ALT, FROM should
1932 ;; also be translated to TO-ALT. But, this is only if TO
1933 ;; is a generic character or TO-ALT is not a generic
13d5617d
KH
1934 ;; character.
1935 (let ((to-alt (aref table to)))
1936 (if (and to-alt
1937 (or (> to-i 0) (not (generic-char-p to-alt))))
1938 (setq to to-alt)))
1939 (if (> from-i 0)
1940 (set-char-table-default table from to)
1941 (aset table from to))
b25eef20
KH
1942 ;; If we have already translated some chars to FROM, they
1943 ;; should also be translated to TO.
a73a8c89
KH
1944 (let ((l (assq from revlist)))
1945 (if l
1946 (let ((ch (car l)))
1947 (setcar l to)
1948 (setq l (cdr l))
1949 (while l
1950 (aset table ch to)
1951 (setq l (cdr l)) ))))
1952 ;; Now update REVLIST.
1953 (let ((l (assq to revlist)))
1954 (if l
1955 (setcdr l (cons from (cdr l)))
1956 (setq revlist (cons (list to from) revlist)))))
1957 (setq elts (cdr elts))))
1958 (setq args (cdr args)))
1959 ;; Return TABLE just created.
1960 table))
1961
c76b5c99
KH
1962(defun make-translation-table-from-vector (vec)
1963 "Make translation table from decoding vector VEC.
9e3b6057
DL
1964VEC is an array of 256 elements to map unibyte codes to multibyte
1965characters. Elements may be nil for undefined code points.
c76b5c99
KH
1966See also the variable `nonascii-translation-table'."
1967 (let ((table (make-char-table 'translation-table))
1968 (rev-table (make-char-table 'translation-table))
c76b5c99 1969 ch)
9e3b6057 1970 (dotimes (i 256)
c76b5c99 1971 (setq ch (aref vec i))
9e3b6057
DL
1972 (when ch
1973 (aset table i ch)
1974 (if (>= ch 256)
1975 (aset rev-table ch i))))
c76b5c99
KH
1976 (set-char-table-extra-slot table 0 rev-table)
1977 table))
1978
f967223b 1979(defun define-translation-table (symbol &rest args)
a284eea3
DL
1980 "Define SYMBOL as the name of translation table made by ARGS.
1981This sets up information so that the table can be used for
1982translations in a CCL program.
b25eef20 1983
a284eea3
DL
1984If the first element of ARGS is a char-table whose purpose is
1985`translation-table', just define SYMBOL to name it. (Note that this
1986function does not bind SYMBOL.)
007c79c8 1987
a284eea3 1988Any other ARGS should be suitable as arguments of the function
007c79c8 1989`make-translation-table' (which see).
b25eef20 1990
452fdb31 1991This function sets properties `translation-table' and
521d4010
DL
1992`translation-table-id' of SYMBOL to the created table itself and the
1993identification number of the table respectively. It also registers
1994the table in `translation-table-vector'."
007c79c8
KH
1995 (let ((table (if (and (char-table-p (car args))
1996 (eq (char-table-subtype (car args))
1997 'translation-table))
1998 (car args)
1999 (apply 'make-translation-table args)))
f967223b 2000 (len (length translation-table-vector))
d9e3229d 2001 (id 0)
b25eef20 2002 (done nil))
f967223b 2003 (put symbol 'translation-table table)
b25eef20
KH
2004 (while (not done)
2005 (if (>= id len)
f967223b
KH
2006 (setq translation-table-vector
2007 (vconcat translation-table-vector (make-vector len nil))))
2008 (let ((slot (aref translation-table-vector id)))
b25eef20
KH
2009 (if (or (not slot)
2010 (eq (car slot) symbol))
2011 (progn
f967223b 2012 (aset translation-table-vector id (cons symbol table))
007c79c8
KH
2013 (setq done t))
2014 (setq id (1+ id)))))
f967223b 2015 (put symbol 'translation-table-id id)
d9e3229d
KH
2016 id))
2017
0e86dca1
KH
2018(defun translate-region (start end table)
2019 "From START to END, translate characters according to TABLE.
2020TABLE is a string or a char-table.
2021If TABLE is a string, the Nth character in it is the mapping
2022for the character with code N.
2023If TABLE is a char-table, the element for character N is the mapping
2024for the character with code N.
2025It returns the number of characters changed."
2026 (interactive
2027 (list (region-beginning)
2028 (region-end)
2029 (let (table l)
2030 (dotimes (i (length translation-table-vector))
2031 (if (consp (aref translation-table-vector i))
2032 (push (list (symbol-name
2033 (car (aref translation-table-vector i)))) l)))
2034 (if (not l)
2035 (error "No translation table defined"))
2036 (while (not table)
2037 (setq table (completing-read "Translation table: " l nil t)))
2038 (intern table))))
2039 (if (symbolp table)
2040 (let ((val (get table 'translation-table)))
2041 (or (char-table-p val)
2042 (error "Invalid translation table name: %s" table))
2043 (setq table val)))
2044 (translate-region-internal start end table))
2045
35554641
KH
2046(put 'with-category-table 'lisp-indent-function 1)
2047
ef6e365d
JPW
2048(defmacro with-category-table (table &rest body)
2049 "Evaluate BODY with category table of current buffer set to TABLE.
2050The category table of the current buffer is saved, BODY is evaluated,
2051then the saved table is restored, even in case of an abnormal exit.
2052Value is what BODY returns."
2053 (let ((old-table (make-symbol "old-table"))
2054 (old-buffer (make-symbol "old-buffer")))
2055 `(let ((,old-table (category-table))
2056 (,old-buffer (current-buffer)))
2057 (unwind-protect
2058 (progn
2059 (set-category-table ,table)
2060 ,@body)
2061 (save-current-buffer
2062 (set-buffer ,old-buffer)
2063 (set-category-table ,old-table))))))
35554641 2064
394e4eb0
DL
2065(defun define-translation-hash-table (symbol table)
2066 "Define SYMBOL as the name of the hash translation TABLE for use in CCL.
2067
2068Analogous to `define-translation-table', but updates
2069`translation-hash-table-vector' and the table is for use in the CCL
2070`lookup-integer' and `lookup-character' functions."
2071 (unless (and (symbolp symbol)
2072 (hash-table-p table))
2073 (error "Bad args to define-translation-hash-table"))
2074 (let ((len (length translation-hash-table-vector))
2075 (id 0)
2076 done)
2077 (put symbol 'translation-hash-table table)
2078 (while (not done)
2079 (if (>= id len)
2080 (setq translation-hash-table-vector
2081 (vconcat translation-hash-table-vector [nil])))
2082 (let ((slot (aref translation-hash-table-vector id)))
2083 (if (or (not slot)
2084 (eq (car slot) symbol))
2085 (progn
2086 (aset translation-hash-table-vector id (cons symbol table))
2087 (setq done t))
2088 (setq id (1+ id)))))
2089 (put symbol 'translation-hash-table-id id)
2090 id))
2091
69eba008
KH
2092;;; Initialize some variables.
2093
2094(put 'use-default-ascent 'char-table-extra-slots 0)
2095(setq use-default-ascent (make-char-table 'use-default-ascent))
d6d6d592
KH
2096(put 'ignore-relative-composition 'char-table-extra-slots 0)
2097(setq ignore-relative-composition
2098 (make-char-table 'ignore-relative-composition))
69eba008 2099
d9f6dfe6
CW
2100
2101;;; Built-in auto-coding-functions:
2102
2103(defun sgml-xml-auto-coding-function (size)
2104 "Determine whether the buffer is XML, and if so, its encoding.
2105This function is intended to be added to `auto-coding-functions'."
c069d3ac
SM
2106 (setq size (+ (point) size))
2107 (when (re-search-forward "\\`[[:space:]\n]*<\\?xml" size t)
d9f6dfe6
CW
2108 (let ((end (save-excursion
2109 ;; This is a hack.
2f4e9c47 2110 (re-search-forward "\"\\s-*\\?>" size t))))
d9f6dfe6
CW
2111 (when end
2112 (if (re-search-forward "encoding=\"\\(.+?\\)\"" end t)
447404a3
CW
2113 (let* ((match (match-string 1))
2114 (sym (intern (downcase match))))
2115 (if (coding-system-p sym)
2116 sym
2117 (message "Warning: unknown coding system \"%s\"" match)
2118 nil))
d9f6dfe6
CW
2119 'utf-8)))))
2120
447404a3
CW
2121(defun sgml-html-meta-auto-coding-function (size)
2122 "If the buffer has an HTML meta tag, use it to determine encoding.
2123This function is intended to be added to `auto-coding-functions'."
c069d3ac 2124 (setq size (min (+ (point) size)
447404a3
CW
2125 ;; Only search forward 10 lines
2126 (save-excursion
2127 (forward-line 10)
2128 (point))))
81668eb2 2129 (when (and (search-forward "<html" size t)
447404a3
CW
2130 (re-search-forward "<meta\\s-+http-equiv=\"content-type\"\\s-+content=\"text/\\sw+;\\s-*charset=\\(.+?\\)\"" size t))
2131 (let* ((match (match-string 1))
2132 (sym (intern (downcase match))))
2133 (if (coding-system-p sym)
2134 sym
2135 (message "Warning: unknown coding system \"%s\"" match)
2136 nil))))
0bca779a 2137
69eba008 2138;;;
4ed46869
KH
2139(provide 'mule)
2140
ab5796a9 2141;;; arch-tag: 9aebaa6e-0e8a-40a9-b857-cb5d04a39e7c
4ed46869 2142;;; mule.el ends here