(cl-transform-lambda): Recognize `declare' as well as `interactive',
[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
0269ddfb
KH
489;; Coding system also has a property `eol-type'.
490;;
491;; This property indicates how the coding system handles end-of-line
492;; format. The value is integer 0, 1, 2, or a vector of three coding
493;; systems. Each integer value 0, 1, and 2 indicates the format of
494;; end-of-line LF, CRLF, and CR respectively. A vector value
495;; indicates that the format of end-of-line should be detected
496;; automatically. Nth element of the vector is the subsidiary coding
497;; system whose `eol-type' property is N.
4ed46869 498
8057896b 499(defun coding-system-eol-type (coding-system)
0269ddfb
KH
500 "Return eol-type of CODING-SYSTEM.
501An eol-type is integer 0, 1, 2, or a vector of coding systems.
4ed46869 502
0269ddfb
KH
503Integer values 0, 1, and 2 indicate a format of end-of-line; LF,
504CRLF, and CR respectively.
505
506A vector value indicates that a format of end-of-line should be
507detected automatically. Nth element of the vector is the subsidiary
508coding system whose eol-type is N."
509 (get coding-system 'eol-type))
bd882697 510
2e729bfa
JB
511(defun coding-system-eol-type-mnemonic (coding-system)
512 "Return the string indicating end-of-line format of CODING-SYSTEM."
513 (let* ((eol-type (coding-system-eol-type coding-system))
f4f00827 514 (val (cond ((eq eol-type 0) eol-mnemonic-unix)
2e729bfa
JB
515 ((eq eol-type 1) eol-mnemonic-dos)
516 ((eq eol-type 2) eol-mnemonic-mac)
f4f00827 517 (t eol-mnemonic-undecided))))
2e729bfa
JB
518 (if (stringp val)
519 val
520 (char-to-string val))))
521
857ea15c
AS
522(defun coding-system-lessp (x y)
523 (cond ((eq x 'no-conversion) t)
524 ((eq y 'no-conversion) nil)
525 ((eq x 'emacs-mule) t)
526 ((eq y 'emacs-mule) nil)
527 ((eq x 'undecided) t)
528 ((eq y 'undecided) nil)
529 (t (let ((c1 (coding-system-mnemonic x))
530 (c2 (coding-system-mnemonic y)))
531 (or (< (downcase c1) (downcase c2))
532 (and (not (> (downcase c1) (downcase c2)))
533 (< c1 c2)))))))
534
5e2e859a
KH
535(defun coding-system-equal (coding-system-1 coding-system-2)
536 "Return t if and only if CODING-SYSTEM-1 and CODING-SYSTEM-2 are identical.
537Two coding systems are identical if two symbols are equal
538or one is an alias of the other."
539 (or (eq coding-system-1 coding-system-2)
540 (and (equal (coding-system-spec coding-system-1)
541 (coding-system-spec coding-system-2))
542 (let ((eol-type-1 (coding-system-eol-type coding-system-1))
543 (eol-type-2 (coding-system-eol-type coding-system-2)))
544 (or (eq eol-type-1 eol-type-2)
545 (and (vectorp eol-type-1) (vectorp eol-type-2)))))))
546
857ea15c 547(defun add-to-coding-system-list (coding-system)
521d4010 548 "Add CODING-SYSTEM to `coding-system-list' while keeping it sorted."
857ea15c
AS
549 (if (or (null coding-system-list)
550 (coding-system-lessp coding-system (car coding-system-list)))
551 (setq coding-system-list (cons coding-system coding-system-list))
552 (let ((len (length coding-system-list))
553 mid (tem coding-system-list))
554 (while (> len 1)
555 (setq mid (nthcdr (/ len 2) tem))
556 (if (coding-system-lessp (car mid) coding-system)
557 (setq tem mid
558 len (- len (/ len 2)))
559 (setq len (/ len 2))))
560 (setcdr tem (cons coding-system (cdr tem))))))
561
80a7463d 562(defun coding-system-list (&optional base-only)
c11a8f77
KH
563 "Return a list of all existing non-subsidiary coding systems.
564If optional arg BASE-ONLY is non-nil, only base coding systems are listed.
565The value doesn't include subsidiary coding systems which are what
566made from bases and aliases automatically for various end-of-line
567formats (e.g. iso-latin-1-unix, koi8-r-dos)."
80a7463d
KH
568 (let* ((codings (copy-sequence coding-system-list))
569 (tail (cons nil codings)))
570 ;; Remove subsidiary coding systems (eol variants) and alias
571 ;; coding systems (if necessary).
572 (while (cdr tail)
573 (let* ((coding (car (cdr tail)))
574 (aliases (coding-system-get coding 'alias-coding-systems)))
575 (if (or
576 ;; CODING is an eol variant if not in ALIASES.
577 (not (memq coding aliases))
578 ;; CODING is an alias if it is not car of ALIASES.
579 (and base-only (not (eq coding (car aliases)))))
580 (setcdr tail (cdr (cdr tail)))
581 (setq tail (cdr tail)))))
582 codings))
583
6fb7766c
DL
584(defun map-charset-chars (func charset)
585 "Use FUNC to map over all characters in CHARSET for side effects.
586FUNC is a function of two args, the start and end (inclusive) of a
587character code range. Thus FUNC should iterate over [START, END]."
588 (let* ((dim (charset-dimension charset))
589 (chars (charset-chars charset))
590 (start (if (= chars 94)
591 33
592 32)))
593 (if (= dim 1)
594 (funcall func
595 (make-char charset start)
596 (make-char charset (+ start chars -1)))
597 (dotimes (i chars)
598 (funcall func
599 (make-char charset (+ i start) start)
600 (make-char charset (+ i start) (+ start chars -1)))))))
601
9857367f 602(defalias 'register-char-codings 'ignore "")
620956ca 603(make-obsolete 'register-char-codings
9857367f 604 "it exists just for backward compatibility, and does nothing."
620956ca
KH
605 "21.3")
606
607(defconst char-coding-system-table nil
608 "This is an obsolete variable.
609It exists just for backward compatibility, and the value is always nil.")
c11a8f77 610
6e9722b0 611(defun make-subsidiary-coding-system (coding-system)
521d4010 612 "Make subsidiary coding systems (eol-type variants) of CODING-SYSTEM."
0269ddfb
KH
613 (let ((coding-spec (coding-system-spec coding-system))
614 (subsidiaries (vector (intern (format "%s-unix" coding-system))
8057896b
KH
615 (intern (format "%s-dos" coding-system))
616 (intern (format "%s-mac" coding-system))))
0269ddfb
KH
617 (i 0)
618 temp)
8057896b 619 (while (< i 3)
0269ddfb 620 (put (aref subsidiaries i) 'coding-system coding-spec)
8057896b 621 (put (aref subsidiaries i) 'eol-type i)
857ea15c 622 (add-to-coding-system-list (aref subsidiaries i))
0269ddfb
KH
623 (setq coding-system-alist
624 (cons (list (symbol-name (aref subsidiaries i)))
625 coding-system-alist))
8057896b
KH
626 (setq i (1+ i)))
627 subsidiaries))
4ed46869 628
50c29104
KH
629(defun transform-make-coding-system-args (name type &optional doc-string props)
630 "For internal use only.
631Transform XEmacs style args for `make-coding-system' to Emacs style.
632Value is a list of transformed arguments."
633 (let ((mnemonic (string-to-char (or (plist-get props 'mnemonic) "?")))
634 (eol-type (plist-get props 'eol-type))
635 properties tmp)
636 (cond
637 ((eq eol-type 'lf) (setq eol-type 'unix))
638 ((eq eol-type 'crlf) (setq eol-type 'dos))
639 ((eq eol-type 'cr) (setq eol-type 'mac)))
640 (if (setq tmp (plist-get props 'post-read-conversion))
641 (setq properties (plist-put properties 'post-read-conversion tmp)))
642 (if (setq tmp (plist-get props 'pre-write-conversion))
643 (setq properties (plist-put properties 'pre-write-conversion tmp)))
644 (cond
f4a012a6
KH
645 ((eq type 'shift-jis)
646 `(,name 1 ,mnemonic ,doc-string () ,properties ,eol-type))
647 ((eq type 'iso2022) ; This is not perfect.
648 (if (plist-get props 'escape-quoted)
649 (error "escape-quoted is not supported: %S"
650 `(,name ,type ,doc-string ,props)))
651 (let ((g0 (plist-get props 'charset-g0))
652 (g1 (plist-get props 'charset-g1))
653 (g2 (plist-get props 'charset-g2))
654 (g3 (plist-get props 'charset-g3))
655 (use-roman
656 (and
657 (eq (cadr (assoc 'latin-jisx0201
658 (plist-get props 'input-charset-conversion)))
659 'ascii)
660 (eq (cadr (assoc 'ascii
661 (plist-get props 'output-charset-conversion)))
662 'latin-jisx0201)))
663 (use-oldjis
664 (and
665 (eq (cadr (assoc 'japanese-jisx0208-1978
666 (plist-get props 'input-charset-conversion)))
667 'japanese-jisx0208)
668 (eq (cadr (assoc 'japanese-jisx0208
669 (plist-get props 'output-charset-conversion)))
670 'japanese-jisx0208-1978))))
671 (if (charsetp g0)
672 (if (plist-get props 'force-g0-on-output)
673 (setq g0 `(nil ,g0))
674 (setq g0 `(,g0 t))))
675 (if (charsetp g1)
676 (if (plist-get props 'force-g1-on-output)
677 (setq g1 `(nil ,g1))
678 (setq g1 `(,g1 t))))
679 (if (charsetp g2)
680 (if (plist-get props 'force-g2-on-output)
681 (setq g2 `(nil ,g2))
682 (setq g2 `(,g2 t))))
683 (if (charsetp g3)
684 (if (plist-get props 'force-g3-on-output)
685 (setq g3 `(nil ,g3))
686 (setq g3 `(,g3 t))))
687 `(,name 2 ,mnemonic ,doc-string
688 (,g0 ,g1 ,g2 ,g3
689 ,(plist-get props 'short)
690 ,(not (plist-get props 'no-ascii-eol))
691 ,(not (plist-get props 'no-ascii-cntl))
692 ,(plist-get props 'seven)
693 t
694 ,(not (plist-get props 'lock-shift))
695 ,use-roman
696 ,use-oldjis
697 ,(plist-get props 'no-iso6429)
698 nil nil nil nil)
699 ,properties ,eol-type)))
700 ((eq type 'big5)
701 `(,name 3 ,mnemonic ,doc-string () ,properties ,eol-type))
50c29104 702 ((eq type 'ccl)
f4a012a6 703 `(,name 4 ,mnemonic ,doc-string
50c29104 704 (,(plist-get props 'decode) . ,(plist-get props 'encode))
f4a012a6 705 ,properties ,eol-type))
50c29104 706 (t
f4a012a6 707 (error "unsupported XEmacs style make-coding-style arguments: %S"
50c29104
KH
708 `(,name ,type ,doc-string ,props))))))
709
8057896b 710(defun make-coding-system (coding-system type mnemonic doc-string
1b46a680
KH
711 &optional
712 flags
713 properties
714 eol-type)
3bb1accb 715 "Define a new coding system CODING-SYSTEM (symbol).
521d4010 716Remaining arguments are TYPE, MNEMONIC, DOC-STRING, FLAGS (optional),
d9e3229d 717and PROPERTIES (optional) which construct a coding-spec of CODING-SYSTEM
a6acd8a2 718in the following format:
0269ddfb 719 [TYPE MNEMONIC DOC-STRING PLIST FLAGS]
3bb1accb
KH
720
721TYPE is an integer value indicating the type of the coding system as follows:
4ed46869 722 0: Emacs internal format,
d9f6dfe6 723 1: Shift-JIS (or MS-Kanji) used mainly on Japanese PCs,
4ed46869 724 2: ISO-2022 including many variants,
d9f6dfe6 725 3: Big5 used mainly on Chinese PCs,
cd9c3177 726 4: private, CCL programs provide encoding/decoding algorithm,
521d4010 727 5: Raw-text, which means that text contains random 8-bit codes.
0269ddfb 728
3bb1accb 729MNEMONIC is a character to be displayed on mode line for the coding system.
0269ddfb 730
3bb1accb 731DOC-STRING is a documentation string for the coding system.
0269ddfb 732
3bb1accb 733FLAGS specifies more detailed information of the coding system as follows:
d6d6d592 734
3bb1accb 735 If TYPE is 2 (ISO-2022), FLAGS is a list of these elements:
4ed46869
KH
736 CHARSET0, CHARSET1, CHARSET2, CHARSET3, SHORT-FORM,
737 ASCII-EOL, ASCII-CNTL, SEVEN, LOCKING-SHIFT, SINGLE-SHIFT,
d6d6d592 738 USE-ROMAN, USE-OLDJIS, NO-ISO6429, INIT-BOL, DESIGNATION-BOL,
850101ed 739 SAFE, ACCEPT-LATIN-EXTRA-CODE.
4ed46869
KH
740 CHARSETn are character sets initially designated to Gn graphic registers.
741 If CHARSETn is nil, Gn is never used.
742 If CHARSETn is t, Gn can be used but nothing designated initially.
743 If CHARSETn is a list of character sets, those character sets are
744 designated to Gn on output, but nothing designated to Gn initially.
421e3b4e 745 But, character set `ascii' can be designated only to G0.
4ed46869
KH
746 SHORT-FORM non-nil means use short designation sequence on output.
747 ASCII-EOL non-nil means designate ASCII to g0 at end of line on output.
748 ASCII-CNTL non-nil means designate ASCII to g0 before control codes and
749 SPACE on output.
750 SEVEN non-nil means use 7-bit code only on output.
751 LOCKING-SHIFT non-nil means use locking-shift.
752 SINGLE-SHIFT non-nil means use single-shift.
753 USE-ROMAN non-nil means designate JIS0201-1976-Roman instead of ASCII.
754 USE-OLDJIS non-nil means designate JIS0208-1976 instead of JIS0208-1983.
755 NO-ISO6429 non-nil means not use ISO6429's direction specification.
69eba008
KH
756 INIT-BOL non-nil means any designation state is assumed to be reset
757 to initial at each beginning of line on output.
758 DESIGNATION-BOL non-nil means designation sequences should be placed
759 at beginning of line on output.
a6acd8a2 760 SAFE non-nil means convert unsafe characters to `?' on output.
c11a8f77
KH
761 Characters not specified in the property `safe-charsets' nor
762 `safe-chars' are unsafe.
850101ed
RS
763 ACCEPT-LATIN-EXTRA-CODE non-nil means code-detection routine accepts
764 a code specified in `latin-extra-code-table' (which see) as a valid
765 code of the coding system.
d6d6d592 766
3bb1accb
KH
767 If TYPE is 4 (private), FLAGS should be a cons of CCL programs, for
768 decoding and encoding. CCL programs should be specified by their
769 symbols.
d9e3229d 770
a2852e45
KH
771PROPERTIES is an alist of properties vs the corresponding values. The
772following properties are recognized:
773
774 o post-read-conversion
0bca779a 775
a2852e45
KH
776 The value is a function to call after some text is inserted and
777 decoded by the coding system itself and before any functions in
521d4010
DL
778 `after-insert-functions' are called. The argument of this
779 function is the same as for a function in
780 `after-insert-file-functions', i.e. LENGTH of the text inserted,
781 with point at the head of the text to be decoded.
0bca779a 782
a2852e45 783 o pre-write-conversion
0bca779a 784
a2852e45
KH
785 The value is a function to call after all functions in
786 `write-region-annotate-functions' and `buffer-file-format' are
787 called, and before the text is encoded by the coding system itself.
a284eea3
DL
788 The arguments to this function are the same as those of a function
789 in `write-region-annotate-functions', i.e. FROM and TO, specifying
521d4010 790 a region of text.
0bca779a 791
a2852e45 792 o translation-table-for-decode
0bca779a 793
a2852e45
KH
794 The value is a translation table to be applied on decoding. See
795 the function `make-translation-table' for the format of translation
521d4010 796 table. This is not applicable to type 4 (CCL-based) coding systems.
0bca779a 797
a2852e45 798 o translation-table-for-encode
0bca779a 799
521d4010
DL
800 The value is a translation table to be applied on encoding. This is
801 not applicable to type 4 (CCL-based) coding systems.
0bca779a 802
a2852e45 803 o safe-chars
0bca779a 804
a2852e45
KH
805 The value is a char table. If a character has non-nil value in it,
806 the character is safely supported by the coding system. This
807 overrides the specification of safe-charsets.
808
809 o safe-charsets
521d4010 810
a2852e45
KH
811 The value is a list of charsets safely supported by the coding
812 system. The value t means that all charsets Emacs handles are
813 supported. Even if some charset is not in this list, it doesn't
521d4010
DL
814 mean that the charset can't be encoded in the coding system;
815 it just means that some other receiver of text encoded
a2852e45 816 in the coding system won't be able to handle that charset.
0bca779a 817
a2852e45 818 o mime-charset
0bca779a 819
d9f6dfe6 820 The value is a symbol whose name is the `MIME-charset' parameter of
a2852e45 821 the coding system.
0bca779a 822
3b6cc699
DL
823 o mime-text-unsuitable
824
825 A non-nil value means the `mime-charset' property names a charset
826 which is unsuitable for the top-level media type \"text\".
827
a2852e45 828 o valid-codes (meaningful only for a coding system based on CCL)
0bca779a 829
a2852e45
KH
830 The value is a list to indicate valid byte ranges of the encoded
831 file. Each element of the list is an integer or a cons of integer.
832 In the former case, the integer value is a valid byte code. In the
6fb7766c 833 latter case, the integers specify the range of valid byte codes.
a2852e45 834
1e7f6327
KH
835 o composition (meaningful only when TYPE is 0 or 2)
836
0ea0b6d2
DL
837 If the value is non-nil, the coding system preserves composition
838 information.
1e7f6327 839
d9e3229d
KH
840These properties are set in PLIST, a property list. This function
841also sets properties `coding-category' and `alias-coding-systems'
842automatically.
4ed46869 843
1b46a680
KH
844EOL-TYPE specifies the EOL type of the coding-system in one of the
845following formats:
846
847 o symbol (unix, dos, or mac)
848
849 The symbol `unix' means Unix-like EOL (LF), `dos' means
850 DOS-like EOL (CRLF), and `mac' means MAC-like EOL (CR).
851
852 o number (0, 1, or 2)
853
854 The number 0, 1, and 2 mean UNIX, DOS, and MAC-like EOL
855 respectively.
856
857 o vector of coding-systems of length 3
858
859 The EOL type is detected automatically for the coding system.
860 And, according to the detected EOL type, one of the coding
861 systems in the vector is selected. Elements of the vector
4d3a9228 862 corresponds to Unix-like EOL, DOS-like EOL, and Mac-like EOL
1b46a680
KH
863 in this order.
864
3bb1accb
KH
865Kludgy features for backward compatibility:
866
8671. If TYPE is 4 and car or cdr of FLAGS is a vector, the vector is
868treated as a compiled CCL code.
869
8702. If PROPERTIES is just a list of character sets, the list is set as
871a value of `safe-charsets' in PLIST."
50c29104
KH
872
873 ;; For compatiblity with XEmacs, we check the type of TYPE. If it
c3d0ee51
EZ
874 ;; is a symbol, perhaps, this function is called with XEmacs-style
875 ;; arguments. Here, try to transform that kind of arguments to
50c29104
KH
876 ;; Emacs style.
877 (if (symbolp type)
878 (let ((args (transform-make-coding-system-args coding-system type
879 mnemonic doc-string)))
880 (setq coding-system (car args)
1053cc93 881 type (nth 1 args)
50c29104
KH
882 mnemonic (nth 2 args)
883 doc-string (nth 3 args)
884 flags (nth 4 args)
885 properties (nth 5 args)
886 eol-type (nth 6 args))))
887
0269ddfb 888 ;; Set a value of `coding-system' property.
6e9722b0 889 (let ((coding-spec (make-vector 5 nil))
d9e3229d
KH
890 (no-initial-designation t)
891 (no-alternative-designation t)
e6cddb1a 892 (accept-latin-extra-code nil)
6e9722b0 893 coding-category)
cd9c3177 894 (if (or (not (integerp type)) (< type 0) (> type 5))
0269ddfb 895 (error "TYPE argument must be 0..5"))
8057896b 896 (if (or (not (integerp mnemonic)) (<= mnemonic ? ) (> mnemonic 127))
521d4010 897 (error "MNEMONIC argument must be an ASCII printable character"))
0269ddfb
KH
898 (aset coding-spec coding-spec-type-idx type)
899 (aset coding-spec coding-spec-mnemonic-idx mnemonic)
900 (aset coding-spec coding-spec-doc-string-idx
e127a722 901 (purecopy (if (stringp doc-string) doc-string "")))
6e9722b0
KH
902 (cond ((= type 0)
903 (setq coding-category 'coding-category-emacs-mule))
904 ((= type 1)
905 (setq coding-category 'coding-category-sjis))
906 ((= type 2) ; ISO2022
4ed46869 907 (let ((i 0)
6e9722b0 908 (vec (make-vector 32 nil))
05134257
KH
909 (g1-designation nil)
910 (fl flags))
4ed46869 911 (while (< i 4)
05134257 912 (let ((charset (car fl)))
6e9722b0
KH
913 (if (and no-initial-designation
914 (> i 0)
915 (or (charsetp charset)
916 (and (consp charset)
917 (charsetp (car charset)))))
918 (setq no-initial-designation nil))
919 (if (charsetp charset)
920 (if (= i 1) (setq g1-designation charset))
921 (if (consp charset)
922 (let ((tail charset)
923 elt)
924 (while tail
925 (setq elt (car tail))
d9e3229d
KH
926 (if (eq elt t)
927 (setq no-alternative-designation nil)
928 (if (and elt (not (charsetp elt)))
929 (error "Invalid charset: %s" elt)))
6e9722b0
KH
930 (setq tail (cdr tail)))
931 (setq g1-designation (car charset)))
d9e3229d
KH
932 (if charset
933 (if (eq charset t)
934 (setq no-alternative-designation nil)
935 (error "Invalid charset: %s" charset)))))
4ed46869 936 (aset vec i charset))
05134257
KH
937 (setq fl (cdr fl) i (1+ i)))
938 (while (and (< i 32) fl)
939 (aset vec i (car fl))
e6cddb1a
KH
940 (if (and (= i 16) ; ACCEPT-LATIN-EXTRA-CODE
941 (car fl))
942 (setq accept-latin-extra-code t))
05134257 943 (setq fl (cdr fl) i (1+ i)))
6e9722b0 944 (aset coding-spec 4 vec)
6e9722b0
KH
945 (setq coding-category
946 (if (aref vec 8) ; Use locking-shift.
dc64cd19
KH
947 (or (and (aref vec 7) 'coding-category-iso-7-else)
948 'coding-category-iso-8-else)
6e9722b0
KH
949 (if (aref vec 7) ; 7-bit only.
950 (if (aref vec 9) ; Use single-shift.
dc64cd19 951 'coding-category-iso-7-else
d9e3229d
KH
952 (if no-alternative-designation
953 'coding-category-iso-7-tight
954 'coding-category-iso-7))
3ad911d8
KH
955 (if (or no-initial-designation
956 (not no-alternative-designation))
dc64cd19 957 'coding-category-iso-8-else
6e9722b0
KH
958 (if (and (charsetp g1-designation)
959 (= (charset-dimension g1-designation) 2))
960 'coding-category-iso-8-2
961 'coding-category-iso-8-1)))))))
962 ((= type 3)
963 (setq coding-category 'coding-category-big5))
964 ((= type 4) ; private
c76b5c99 965 (setq coding-category 'coding-category-ccl)
3bb1accb
KH
966 (if (not (consp flags))
967 (error "Invalid FLAGS argument for TYPE 4 (CCL)")
968 (let ((decoder (check-ccl-program
969 (car flags)
970 (intern (format "%s-decoder" coding-system))))
971 (encoder (check-ccl-program
972 (cdr flags)
973 (intern (format "%s-encoder" coding-system)))))
974 (if (and decoder encoder)
975 (aset coding-spec 4 (cons decoder encoder))
976 (error "Invalid FLAGS argument for TYPE 4 (CCL)")))))
cd9c3177
KH
977 (t ; i.e. (= type 5)
978 (setq coding-category 'coding-category-raw-text)))
0269ddfb
KH
979
980 (let ((plist (list 'coding-category coding-category
d9e3229d 981 'alias-coding-systems (list coding-system))))
0269ddfb 982 (if no-initial-designation
d9e3229d
KH
983 (plist-put plist 'no-initial-designation t))
984 (if (and properties
985 (or (eq properties t)
986 (not (consp (car properties)))))
987 ;; In the old version, the arg PROPERTIES is a list to be
988 ;; set in PLIST as a value of property `safe-charsets'.
c11a8f77
KH
989 (setq properties (list (cons 'safe-charsets properties))))
990 ;; In the current version PROPERTIES is a property list.
991 ;; Reflect it into PLIST one by one while handling safe-chars
992 ;; specially.
993 (let ((safe-charsets (cdr (assq 'safe-charsets properties)))
994 (safe-chars (cdr (assq 'safe-chars properties)))
995 (l properties)
996 prop val)
997 ;; If only safe-charsets is specified, make a char-table from
998 ;; it, and store that char-table as the value of `safe-chars'.
999 (if (and (not safe-chars) safe-charsets)
1000 (let (charset)
1001 (if (eq safe-charsets t)
1002 (setq safe-chars t)
1003 (setq safe-chars (make-char-table 'safe-chars))
1004 (while safe-charsets
1005 (setq charset (car safe-charsets)
1006 safe-charsets (cdr safe-charsets))
1007 (cond ((eq charset 'ascii)) ; just ignore
1008 ((eq charset 'eight-bit-control)
1009 (let ((i 128))
1010 (while (< i 160)
1011 (aset safe-chars i t)
1012 (setq i (1+ i)))))
1013 ((eq charset 'eight-bit-graphic)
1014 (let ((i 160))
1015 (while (< i 256)
1016 (aset safe-chars i t)
1017 (setq i (1+ i)))))
1018 (t
e6cddb1a
KH
1019 (aset safe-chars (make-char charset) t))))
1020 (if accept-latin-extra-code
1021 (let ((i 128))
1022 (while (< i 160)
1023 (if (aref latin-extra-code-table i)
1024 (aset safe-chars i t))
1025 (setq i (1+ i))))))
c11a8f77
KH
1026 (setq l (cons (cons 'safe-chars safe-chars) l))))
1027 (while l
1028 (setq prop (car (car l)) val (cdr (car l)) l (cdr l))
1029 (if (eq prop 'safe-chars)
1030 (progn
cc57cc54
KH
1031 (if (and (symbolp val)
1032 (get val 'translation-table))
1033 (setq safe-chars (get val 'translation-table)))
cc57cc54 1034 (setq val safe-chars)))
c11a8f77 1035 (plist-put plist prop val)))
1b46a680
KH
1036 ;; The property `coding-category' may have been set differently
1037 ;; through PROPERTIES.
1038 (setq coding-category (plist-get plist 'coding-category))
0269ddfb 1039 (aset coding-spec coding-spec-plist-idx plist))
6e9722b0 1040 (put coding-system 'coding-system coding-spec)
c95cf5ac 1041 (put coding-system 'coding-system-define-form nil)
6e9722b0
KH
1042 (put coding-category 'coding-systems
1043 (cons coding-system (get coding-category 'coding-systems))))
4ed46869 1044
1b46a680 1045 ;; Next, set a value of `eol-type' property.
af873cd1 1046 (if (not eol-type)
1b46a680
KH
1047 ;; If EOL-TYPE is nil, set a vector of subsidiary coding
1048 ;; systems, each corresponds to a coding system for the detected
1049 ;; EOL format.
1050 (setq eol-type (make-subsidiary-coding-system coding-system)))
1051 (setq eol-type
1052 (cond ((or (eq eol-type 'unix) (null eol-type))
1053 0)
1054 ((eq eol-type 'dos)
1055 1)
1056 ((eq eol-type 'mac)
1057 2)
1058 ((or (and (vectorp eol-type)
1059 (= (length eol-type) 3))
1060 (and (numberp eol-type)
1061 (and (>= eol-type 0)
1062 (<= eol-type 2))))
1063 eol-type)
1064 (t
1065 (error "Invalid EOL-TYPE spec:%S" eol-type))))
1066 (put coding-system 'eol-type eol-type)
0269ddfb 1067
620956ca
KH
1068 (define-coding-system-internal coding-system)
1069
0269ddfb
KH
1070 ;; At last, register CODING-SYSTEM in `coding-system-list' and
1071 ;; `coding-system-alist'.
857ea15c 1072 (add-to-coding-system-list coding-system)
0269ddfb 1073 (setq coding-system-alist (cons (list (symbol-name coding-system))
d9e3229d 1074 coding-system-alist))
05134257
KH
1075
1076 ;; For a coding system of cateogory iso-8-1 and iso-8-2, create
1077 ;; XXX-with-esc variants.
1078 (let ((coding-category (coding-system-category coding-system)))
1079 (if (or (eq coding-category 'coding-category-iso-8-1)
1080 (eq coding-category 'coding-category-iso-8-2))
1081 (let ((esc (intern (concat (symbol-name coding-system) "-with-esc")))
c11a8f77
KH
1082 (doc (format "Same as %s but can handle any charsets by ISO's escape sequences." coding-system))
1083 (safe-charsets (assq 'safe-charsets properties))
1084 (mime-charset (assq 'mime-charset properties)))
1085 (if safe-charsets
1086 (setcdr safe-charsets t)
1087 (setq properties (cons (cons 'safe-charsets t) properties)))
1088 (if mime-charset
1089 (setcdr mime-charset nil))
05134257
KH
1090 (make-coding-system esc type mnemonic doc
1091 (if (listp (car flags))
1092 (cons (append (car flags) '(t)) (cdr flags))
1093 (cons (list (car flags) t) (cdr flags)))
c11a8f77 1094 properties))))
05134257 1095
d9e3229d 1096 coding-system)
8057896b 1097
5470d5b6
KH
1098(put 'safe-chars 'char-table-extra-slots 0)
1099
a42763dc 1100(defun define-coding-system-alias (alias coding-system)
358d28fb 1101 "Define ALIAS as an alias for coding system CODING-SYSTEM."
0269ddfb 1102 (put alias 'coding-system (coding-system-spec coding-system))
c95cf5ac 1103 (put alias 'coding-system-define-form nil)
857ea15c 1104 (add-to-coding-system-list alias)
0269ddfb
KH
1105 (setq coding-system-alist (cons (list (symbol-name alias))
1106 coding-system-alist))
1107 (let ((eol-type (coding-system-eol-type coding-system)))
1108 (if (vectorp eol-type)
708fc465
KH
1109 (progn
1110 (nconc (coding-system-get alias 'alias-coding-systems) (list alias))
1111 (put alias 'eol-type (make-subsidiary-coding-system alias)))
0269ddfb 1112 (put alias 'eol-type eol-type))))
4ed46869 1113
bbdea948
RS
1114(defun merge-coding-systems (first second)
1115 "Fill in any unspecified aspects of coding system FIRST from SECOND.
1116Return the resulting coding system."
1117 (let ((base (coding-system-base second))
1118 (eol (coding-system-eol-type second)))
1119 ;; If FIRST doesn't specify text conversion, merge with that of SECOND.
1120 (if (eq (coding-system-base first) 'undecided)
1121 (setq first (coding-system-change-text-conversion first base)))
1122 ;; If FIRST doesn't specify eol conversion, merge with that of SECOND.
1123 (if (and (vectorp (coding-system-eol-type first))
1124 (numberp eol) (>= eol 0) (<= eol 2))
1125 (setq first (coding-system-change-eol-conversion
1126 first eol)))
1127 first))
1128
2b96ab33
KH
1129(defun autoload-coding-system (symbol form)
1130 "Define SYMBOL as a coding-system that is defined on demand.
1131
1132FROM is a form to evaluate to define the coding-system."
28849ffc
KH
1133 (put symbol 'coding-system-define-form form)
1134 (setq coding-system-alist (cons (list (symbol-name symbol))
1135 coding-system-alist)))
2b96ab33 1136
14b3fa07 1137(defun set-buffer-file-coding-system (coding-system &optional force nomodify)
358d28fb
RS
1138 "Set the file coding-system of the current buffer to CODING-SYSTEM.
1139This means that when you save the buffer, it will be converted
1140according to CODING-SYSTEM. For a list of possible values of CODING-SYSTEM,
1141use \\[list-coding-systems].
1142
bbdea948
RS
1143If CODING-SYSTEM leaves the text conversion unspecified, or if it
1144leaves the end-of-line conversion unspecified, FORCE controls what to
1145do. If FORCE is nil, get the unspecified aspect (or aspects) from the
1146buffer's previous `buffer-file-coding-system' value (if it is
3b6cc699 1147specified there). Otherwise, leave it unspecified.
aeef8f07
KH
1148
1149This marks the buffer modified so that the succeeding \\[save-buffer]
1150surely saves the buffer with CODING-SYSTEM. From a program, if you
14b3fa07
RS
1151don't want to mark the buffer modified, specify t for NOMODIFY.
1152If you know exactly what coding system you want to use,
1153just set the variable `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))
de5ffead
RS
1164 (unless nomodify
1165 (set-buffer-modified-p t))
4ed46869
KH
1166 (force-mode-line-update))
1167
bbdea948
RS
1168(defun revert-buffer-with-coding-system (coding-system &optional force)
1169 "Visit the current buffer's file again using coding system CODING-SYSTEM.
1170For a list of possible values of CODING-SYSTEM, use \\[list-coding-systems].
1171
1172If CODING-SYSTEM leaves the text conversion unspecified, or if it
1173leaves the end-of-line conversion unspecified, FORCE controls what to
1174do. If FORCE is nil, get the unspecified aspect (or aspects) from the
1175buffer's previous `buffer-file-coding-system' value (if it is
1176specified there). Otherwise, determine it from the file contents as
1177usual for visiting a file."
1178 (interactive "zCoding system for visited file (default, nil): \nP")
1179 (check-coding-system coding-system)
1180 (if (and coding-system buffer-file-coding-system (null force))
1181 (setq coding-system
1182 (merge-coding-systems coding-system buffer-file-coding-system)))
1183 (let ((coding-system-for-read coding-system))
1184 (revert-buffer)))
1185
701414e3
KH
1186(defun set-file-name-coding-system (coding-system)
1187 "Set coding system for decoding and encoding file names to CODING-SYSTEM.
1188It actually just set the variable `file-name-coding-system' (which
1189see) to CODING-SYSTEM."
1190 (interactive "zCoding system for file names (default, nil): ")
1191 (check-coding-system coding-system)
1192 (setq file-name-coding-system coding-system))
1193
358d28fb
RS
1194(defvar default-terminal-coding-system nil
1195 "Default value for the terminal coding system.
1196This is normally set according to the selected language environment.
1197See also the command `set-terminal-coding-system'.")
1198
df100398
KH
1199(defun set-terminal-coding-system (coding-system)
1200 "Set coding system of your terminal to CODING-SYSTEM.
358d28fb
RS
1201All text output to the terminal will be encoded
1202with the specified coding system.
1203For a list of possible values of CODING-SYSTEM, use \\[list-coding-systems].
1204The default is determined by the selected language environment
1205or by the previous use of this command."
1206 (interactive
2e02a76f
RS
1207 (list (let ((default (if (and (not (terminal-coding-system))
1208 default-terminal-coding-system)
1209 default-terminal-coding-system)))
1210 (read-coding-system
1211 (format "Coding system for terminal display (default, %s): "
1212 default)
1213 default))))
358d28fb
RS
1214 (if (and (not coding-system)
1215 (not (terminal-coding-system)))
1216 (setq coding-system default-terminal-coding-system))
1217 (if coding-system
521d4010 1218 (setq default-terminal-coding-system coding-system))
df100398
KH
1219 (set-terminal-coding-system-internal coding-system)
1220 (redraw-frame (selected-frame)))
1221
358d28fb
RS
1222(defvar default-keyboard-coding-system nil
1223 "Default value of the keyboard coding system.
1224This is normally set according to the selected language environment.
1225See also the command `set-keyboard-coding-system'.")
1226
df100398 1227(defun set-keyboard-coding-system (coding-system)
358d28fb
RS
1228 "Set coding system for keyboard input to CODING-SYSTEM.
1229In addition, this command enables Encoded-kbd minor mode.
6d34f495
DL
1230\(If CODING-SYSTEM is nil, Encoded-kbd mode is turned off -- see
1231`encoded-kbd-mode'.)
358d28fb
RS
1232For a list of possible values of CODING-SYSTEM, use \\[list-coding-systems].
1233The default is determined by the selected language environment
1234or by the previous use of this command."
1235 (interactive
2e02a76f
RS
1236 (list (let ((default (if (and (not (keyboard-coding-system))
1237 default-keyboard-coding-system)
1238 default-keyboard-coding-system)))
1239 (read-coding-system
1240 (format "Coding system for keyboard input (default, %s): "
1241 default)
1242 default))))
358d28fb
RS
1243 (if (and (not coding-system)
1244 (not (keyboard-coding-system)))
1245 (setq coding-system default-keyboard-coding-system))
1246 (if coding-system
1247 (setq default-keyboard-coding-system coding-system))
df100398 1248 (set-keyboard-coding-system-internal coding-system)
b23bad0b 1249 (setq keyboard-coding-system coding-system)
df100398
KH
1250 (encoded-kbd-mode (if coding-system 1 0)))
1251
6d34f495
DL
1252(defcustom keyboard-coding-system nil
1253 "Specify coding system for keyboard input.
1254If you set this on a terminal which can't distinguish Meta keys from
12558-bit characters, you will have to use ESC to type Meta characters.
1256See Info node `Specify Coding' and Info node `Single-Byte Character Support'.
1257
237e5993
DL
1258On non-windowing terminals, this is set from the locale by default.
1259
6d34f495 1260Setting this variable directly does not take effect;
c4131d96 1261use either \\[customize] or \\[set-keyboard-coding-system]."
6d34f495
DL
1262 :type '(coding-system :tag "Coding system")
1263 :link '(info-link "(emacs)Specify Coding")
1264 :link '(info-link "(emacs)Single-Byte Character Support")
1265 :set (lambda (symbol value)
1266 ;; Don't load encoded-kbd-mode unnecessarily.
1267 (if (or value (boundp 'encoded-kbd-mode))
1268 (set-keyboard-coding-system value)
1269 (set-default 'keyboard-coding-system nil))) ; must initialize
bf247b6e 1270 :version "22.1"
6d34f495
DL
1271 :group 'keyboard
1272 :group 'mule)
1273
df100398 1274(defun set-buffer-process-coding-system (decoding encoding)
358d28fb 1275 "Set coding systems for the process associated with the current buffer.
df100398 1276DECODING is the coding system to be used to decode input from the process,
358d28fb
RS
1277ENCODING is the coding system to be used to encode output to the process.
1278
1279For a list of possible values of CODING-SYSTEM, use \\[list-coding-systems]."
4ed46869 1280 (interactive
83911021 1281 "zCoding-system for output from the process: \nzCoding-system for input to the process: ")
4ed46869
KH
1282 (let ((proc (get-buffer-process (current-buffer))))
1283 (if (null proc)
521d4010 1284 (error "No process")
df100398
KH
1285 (check-coding-system decoding)
1286 (check-coding-system encoding)
1287 (set-process-coding-system proc decoding encoding)))
4ed46869
KH
1288 (force-mode-line-update))
1289
d0b99881
RS
1290(defalias 'set-clipboard-coding-system 'set-selection-coding-system)
1291
14915c37 1292(defun set-selection-coding-system (coding-system)
8c52d564 1293 "Make CODING-SYSTEM used for communicating with other X clients.
b25eef20
KH
1294When sending or receiving text via cut_buffer, selection, and clipboard,
1295the text is encoded or decoded by CODING-SYSTEM."
a03b3ce1 1296 (interactive "zCoding system for X selection: ")
b25eef20 1297 (check-coding-system coding-system)
14915c37 1298 (setq selection-coding-system coding-system))
b25eef20 1299
e8dd0160 1300;; Coding system lastly specified by the command
a03b3ce1
KH
1301;; set-next-selection-coding-system.
1302(defvar last-next-selection-coding-system nil)
1303
1304(defun set-next-selection-coding-system (coding-system)
1305 "Make CODING-SYSTEM used for the next communication with other X clients.
1306This setting is effective for the next communication only."
1307 (interactive
1308 (list (read-coding-system
1309 (if last-next-selection-coding-system
1310 (format "Coding system for the next X selection (default, %S): "
1311 last-next-selection-coding-system)
1312 "Coding system for the next X selection: ")
1313 last-next-selection-coding-system)))
1314 (if coding-system
1315 (setq last-next-selection-coding-system coding-system)
1316 (setq coding-system last-next-selection-coding-system))
1317 (check-coding-system coding-system)
1318
1319 (setq next-selection-coding-system coding-system))
1320
4ed46869 1321(defun set-coding-priority (arg)
521d4010
DL
1322 "Set priority of coding categories according to ARG.
1323ARG is a list of coding categories ordered by priority."
d9e3229d
KH
1324 (let ((l arg)
1325 (current-list (copy-sequence coding-category-list)))
e8dd0160 1326 ;; Check the validity of ARG while deleting coding categories in
d9e3229d
KH
1327 ;; ARG from CURRENT-LIST. We assume that CODING-CATEGORY-LIST
1328 ;; contains all coding categories.
1329 (while l
1330 (if (or (null (get (car l) 'coding-category-index))
1331 (null (memq (car l) current-list)))
1332 (error "Invalid or duplicated element in argument: %s" arg))
1333 (setq current-list (delq (car l) current-list))
1334 (setq l (cdr l)))
4ed46869 1335 ;; Update `coding-category-list' and return it.
2feaf204
KH
1336 (setq coding-category-list (append arg current-list))
1337 (set-coding-priority-internal)))
4ed46869 1338
835cbadb
EZ
1339;;; X selections
1340
cc926903 1341(defvar ctext-non-standard-encodings-alist
6c805023
KH
1342 '(("big5-0" big5 2 (chinese-big5-1 chinese-big5-2))
1343 ("ISO8859-14" iso-8859-14 1 latin-iso8859-14)
1344 ("ISO8859-15" iso-8859-15 1 latin-iso8859-15))
1345 "Alist of non-standard encoding names vs the corresponding usages in CTEXT.
1346
1347It controls how extended segments of a compound text are handled
1348by the coding system `compound-text-with-extensions'.
1349
1350Each element has the form (ENCODING-NAME CODING-SYSTEM N-OCTET CHARSET).
1351
1352ENCODING-NAME is an encoding name of an \"extended segments\".
1353
1354CODING-SYSTEM is the coding-system to encode (or decode) the
1355characters into (or from) the extended segment.
1356
1357N-OCTET is the number of octets (bytes) that encodes a character
1358in the segment. It can be 0 (meaning the number of octets per
1359character is variable), 1, 2, 3, or 4.
1360
1361CHARSET is a charater set containing characters that are encoded
1362in the segment. It can be a list of character sets. It can also
1363be a char-table, in which case characters that have non-nil value
1364in the char-table are the target.
1365
1366On decoding CTEXT, all encoding names listed here are recognized.
1367
1368On encoding CTEXT, encoding names in the variable
1369`ctext-non-standard-encodings' (which see) and in the information
1370listed for the current language environment under the key
1371`ctext-non-standard-encodings' are used.")
1372
1373(defvar ctext-non-standard-encodings
1374 '("big5-0")
1375 "List of non-standard encoding names used in extended segments of CTEXT.
1376Each element must be one of the names listed in the variable
1377`ctext-non-standard-encodings-alist' (which see).")
cc926903
KH
1378
1379(defvar ctext-non-standard-encodings-regexp
1380 (string-to-multibyte
1381 (concat
1382 ;; For non-standard encodings.
1383 "\\(\e%/[0-4][\200-\377][\200-\377]\\([^\002]+\\)\002\\)"
1384 "\\|"
1385 ;; For UTF-8 encoding.
1386 "\\(\e%G[^\e]*\e%@\\)")))
835cbadb
EZ
1387
1388;; Functions to support "Non-Standard Character Set Encodings" defined
6c805023
KH
1389;; by the COMPOUND-TEXT spec. They also support "The UTF-8 encoding"
1390;; described in the section 7 of the documentation of COMPOUND-TEXT
1391;; distributed with XFree86.
5c88a01e 1392
835cbadb
EZ
1393(defun ctext-post-read-conversion (len)
1394 "Decode LEN characters encoded as Compound Text with Extended Segments."
835cbadb 1395 (save-match-data
cc926903
KH
1396 (save-restriction
1397 (let ((case-fold-search nil)
1398 (in-workbuf (string= (buffer-name) " *code-converting-work*"))
1399 last-coding-system-used
1400 pos bytes)
1401 (or in-workbuf
1402 (narrow-to-region (point) (+ (point) len)))
cc926903
KH
1403 (if in-workbuf
1404 (set-buffer-multibyte t))
1405 (while (re-search-forward ctext-non-standard-encodings-regexp
1406 nil 'move)
1407 (setq pos (match-beginning 0))
1408 (if (match-beginning 1)
1409 ;; ESC % / [0-4] M L --ENCODING-NAME-- \002 --BYTES--
1410 (let* ((M (char-after (+ pos 4)))
1411 (L (char-after (+ pos 5)))
1412 (encoding (match-string 2))
9857367f 1413 (encoding-info (assoc-string
6c805023 1414 encoding
c12bc1fb 1415 ctext-non-standard-encodings-alist t))
6c805023
KH
1416 (coding (if encoding-info
1417 (nth 1 encoding-info)
1418 (setq encoding (intern (downcase encoding)))
1419 (and (coding-system-p encoding)
1420 encoding))))
cc926903
KH
1421 (setq bytes (- (+ (* (- M 128) 128) (- L 128))
1422 (- (point) (+ pos 6))))
1423 (when coding
1424 (delete-region pos (point))
1425 (forward-char bytes)
1426 (decode-coding-region (- (point) bytes) (point) coding)))
1427 ;; ESC % G --UTF-8-BYTES-- ESC % @
6c805023
KH
1428 (delete-char -3)
1429 (delete-region pos (+ pos 3))
1430 (decode-coding-region pos (point) 'utf-8))))
cc926903
KH
1431 (goto-char (point-min))
1432 (- (point-max) (point)))))
835cbadb 1433
6c805023
KH
1434;; Return a char table of extended segment usage for each character.
1435;; Each value of the char table is nil, one of the elements of
1436;; `ctext-non-standard-encodings-alist', or the symbol `utf-8'.
1437
1438(defun ctext-non-standard-encodings-table ()
1439 (let ((table (make-char-table 'translation-table)))
1440 (aset table (make-char 'mule-unicode-0100-24ff) 'utf-8)
1441 (aset table (make-char 'mule-unicode-2500-33ff) 'utf-8)
1442 (aset table (make-char 'mule-unicode-e000-ffff) 'utf-8)
1443 (dolist (encoding (reverse
1444 (append
1445 (get-language-info current-language-environment
1446 'ctext-non-standard-encodings)
1447 ctext-non-standard-encodings)))
1448 (let* ((slot (assoc encoding ctext-non-standard-encodings-alist))
1449 (charset (nth 3 slot)))
1450 (if charset
1451 (cond ((charsetp charset)
1452 (aset table (make-char charset) slot))
1453 ((listp charset)
1454 (dolist (elt charset)
1455 (aset table (make-char elt) slot)))
1456 ((char-table-p charset)
9857367f 1457 (map-char-table #'(lambda (k v)
6c805023
KH
1458 (if (and v (> k 128)) (aset table k slot)))
1459 charset))))))
1460 table))
835cbadb
EZ
1461
1462(defun ctext-pre-write-conversion (from to)
5dde3c71
EZ
1463 "Encode characters between FROM and TO as Compound Text w/Extended Segments.
1464
1465If FROM is a string, or if the current buffer is not the one set up for us
cc926903 1466by encode-coding-string, generate a new temp buffer, insert the
5dde3c71 1467text, and convert it in the temporary buffer. Otherwise, convert in-place."
835cbadb 1468 (save-match-data
002dc239
KH
1469 (let ((workbuf (get-buffer-create " *code-conversion-work*")))
1470 ;; Setup a working buffer if necessary.
1471 (cond ((stringp from)
1472 (set-buffer workbuf)
1473 (erase-buffer)
cc926903 1474 (set-buffer-multibyte (multibyte-string-p from))
002dc239
KH
1475 (insert from))
1476 ((not (eq (current-buffer) workbuf))
1477 (let ((buf (current-buffer))
1478 (multibyte enable-multibyte-characters))
1479 (set-buffer workbuf)
1480 (erase-buffer)
1481 (set-buffer-multibyte multibyte)
1482 (insert-buffer-substring buf from to)))))
cc926903
KH
1483
1484 ;; Now we can encode the whole buffer.
6c805023 1485 (let ((encoding-table (ctext-non-standard-encodings-table))
cc926903 1486 last-coding-system-used
6c805023
KH
1487 last-pos last-encoding-info
1488 encoding-info end-pos)
1489 (goto-char (setq last-pos (point-min)))
1490 (setq end-pos (point-marker))
1491 (while (re-search-forward "[^\000-\177]+" nil t)
1492 ;; Found a sequence of non-ASCII characters.
1493 (setq last-pos (match-beginning 0)
1494 last-encoding-info (aref encoding-table (char-after last-pos)))
1495 (set-marker end-pos (match-end 0))
1496 (goto-char (1+ last-pos))
1497 (catch 'tag
1498 (while t
1499 (setq encoding-info
1500 (if (< (point) end-pos)
1501 (aref encoding-table (following-char))))
1502 (unless (eq last-encoding-info encoding-info)
1503 (cond ((consp last-encoding-info)
1504 ;; Encode the previous range using an extended
1505 ;; segment.
1506 (let ((encoding-name (car last-encoding-info))
1507 (coding-system (nth 1 last-encoding-info))
1508 (noctets (nth 2 last-encoding-info))
1509 len)
1510 (encode-coding-region last-pos (point) coding-system)
1511 (setq len (+ (length encoding-name) 1
1512 (- (point) last-pos)))
1513 (save-excursion
1514 (goto-char last-pos)
9857367f 1515 (insert (string-to-multibyte
6c805023
KH
1516 (format "\e%%/%d%c%c%s\ 2"
1517 noctets
1518 (+ (/ len 128) 128)
1519 (+ (% len 128) 128)
1520 encoding-name))))))
1521 ((eq last-encoding-info 'utf-8)
1522 ;; Encode the previous range using UTF-8 encoding
1523 ;; extention.
1524 (encode-coding-region last-pos (point) 'mule-utf-8)
1525 (save-excursion
1526 (goto-char last-pos)
1527 (insert "\e%G"))
1528 (insert "\e%@")))
1529 (setq last-pos (point)
1530 last-encoding-info encoding-info))
1531 (if (< (point) end-pos)
1532 (forward-char 1)
1533 (throw 'tag nil)))))
1534 (set-marker end-pos nil)
cc926903 1535 (goto-char (point-min))))
5dde3c71 1536 ;; Must return nil, as build_annotations_2 expects that.
835cbadb
EZ
1537 nil)
1538
4ed46869
KH
1539;;; FILE I/O
1540
e76938e7 1541(defcustom auto-coding-alist
d8df8af1
EZ
1542 '(("\\.\\(arc\\|zip\\|lzh\\|zoo\\|[jew]ar\\|xpi\\)\\'" . no-conversion)
1543 ("\\.\\(ARC\\|ZIP\\|LZH\\|ZOO\\|[JEW]AR\\|XPI\\)\\'" . no-conversion)
e3668698 1544 ("\\.\\(sx[dmicw]\\|tar\\|tgz\\)\\'" . no-conversion)
6caef2da 1545 ("\\.\\(gz\\|Z\\|bz\\|bz2\\|gpg\\)\\'" . no-conversion)
ba40634b 1546 ("\\.\\(jpe?g\\|png\\|gif\\|tiff?\\|p[bpgn]m\\)\\'" . no-conversion)
45885400 1547 ("/#[^/]+#\\'" . emacs-mule))
835f49b8
KH
1548 "Alist of filename patterns vs corresponding coding systems.
1549Each element looks like (REGEXP . CODING-SYSTEM).
558b0c86 1550A file whose name matches REGEXP is decoded by CODING-SYSTEM on reading.
835f49b8 1551
7fed493a
RS
1552The settings in this alist take priority over `coding:' tags
1553in the file (see the function `set-auto-coding')
e76938e7
DL
1554and the contents of `file-coding-system-alist'."
1555 :group 'files
1556 :group 'mule
1557 :type '(repeat (cons (regexp :tag "File name regexp")
1558 (symbol :tag "Coding system"))))
835f49b8 1559
502522b2
GM
1560(defcustom auto-coding-regexp-alist
1561 '(("^BABYL OPTIONS:[ \t]*-\\*-[ \t]*rmail[ \t]*-\\*-" . no-conversion))
1562 "Alist of patterns vs corresponding coding systems.
1563Each element looks like (REGEXP . CODING-SYSTEM).
1564A file whose first bytes match REGEXP is decoded by CODING-SYSTEM on reading.
1565
1566The settings in this alist take priority over `coding:' tags
1567in the file (see the function `set-auto-coding')
1568and the contents of `file-coding-system-alist'."
1569 :group 'files
1570 :group 'mule
1571 :type '(repeat (cons (regexp :tag "Regexp")
1572 (symbol :tag "Coding system"))))
1573
d9f6dfe6 1574;; See the bottom of this file for built-in auto coding functions.
447404a3
CW
1575(defcustom auto-coding-functions '(sgml-xml-auto-coding-function
1576 sgml-html-meta-auto-coding-function)
d9f6dfe6
CW
1577 "A list of functions which attempt to determine a coding system.
1578
66643502
RS
1579Each function in this list should be written to operate on the
1580current buffer, but should not modify it in any way. The buffer
1581will contain undecoded text of parts of the file. Each function
1582should take one argument, SIZE, which says how many
1583characters (starting from point) it should look at.
1584
1585If one of these functions succeeds in determining a coding
1586system, it should return that coding system. Otherwise, it
1587should return nil.
1588
1589If a file has a `coding:' tag, that takes precedence over these
1590functions, so they won't be called at all."
d9f6dfe6
CW
1591 :group 'files
1592 :group 'mule
1593 :type '(repeat function))
1594
1c4cc63a
KH
1595(defvar set-auto-coding-for-load nil
1596 "Non-nil means look for `load-coding' property instead of `coding'.
1597This is used for loading and byte-compiling Emacs Lisp files.")
1598
8a592131
RS
1599(defun auto-coding-alist-lookup (filename)
1600 "Return the coding system specified by `auto-coding-alist' for FILENAME."
1601 (let ((alist auto-coding-alist)
c60ee5e7 1602 (case-fold-search (memq system-type '(vax-vms windows-nt ms-dos cygwin)))
8a592131
RS
1603 coding-system)
1604 (while (and alist (not coding-system))
1605 (if (string-match (car (car alist)) filename)
1606 (setq coding-system (cdr (car alist)))
1607 (setq alist (cdr alist))))
1608 coding-system))
1609
835f49b8
KH
1610(defun set-auto-coding (filename size)
1611 "Return coding system for a file FILENAME of which SIZE bytes follow point.
1c4cc63a
KH
1612These bytes should include at least the first 1k of the file
1613and the last 3k of the file, but the middle may be omitted.
63561304 1614
d21ba5e0
DL
1615The function checks FILENAME against the variable `auto-coding-alist'.
1616If FILENAME doesn't match any entries in the variable, it checks the
502522b2 1617contents of the current buffer following point against
447404a3 1618`auto-coding-regexp-alist'. If no match is found, it checks for a
502522b2 1619`coding:' tag in the first one or two lines following point. If no
d21ba5e0 1620`coding:' tag is found, it checks any local variables list in the last
447404a3 16213K bytes out of the SIZE bytes. Finally, if none of these methods
d21ba5e0
DL
1622succeed, it checks to see if any function in `auto-coding-functions'
1623gives a match.
63561304 1624
d21ba5e0
DL
1625The return value is the specified coding system, or nil if nothing is
1626specified.
87aba788 1627
ba74e833 1628The variable `set-auto-coding-function' (which see) is set to this
87aba788 1629function by default."
502522b2 1630 (or (auto-coding-alist-lookup filename)
447404a3
CW
1631 ;; Try using `auto-coding-regexp-alist'.
1632 (save-excursion
1633 (let ((alist auto-coding-regexp-alist)
1634 coding-system)
1635 (while (and alist (not coding-system))
1636 (let ((regexp (car (car alist))))
1637 (when (re-search-forward regexp (+ (point) size) t)
1638 (setq coding-system (cdr (car alist)))))
0bca779a 1639 (setq alist (cdr alist)))
447404a3 1640 coding-system))
502522b2
GM
1641 (let* ((case-fold-search t)
1642 (head-start (point))
1643 (head-end (+ head-start (min size 1024)))
1644 (tail-start (+ head-start (max (- size 3072) 0)))
1645 (tail-end (+ head-start size))
1646 coding-system head-found tail-found pos)
1647 ;; Try a short cut by searching for the string "coding:"
1648 ;; and for "unibyte:" at the head and tail of SIZE bytes.
1649 (setq head-found (or (search-forward "coding:" head-end t)
1650 (search-forward "unibyte:" head-end t)))
1651 (if (and head-found (> head-found tail-start))
1652 ;; Head and tail are overlapped.
1653 (setq tail-found head-found)
1654 (goto-char tail-start)
1655 (setq tail-found (or (search-forward "coding:" tail-end t)
1656 (search-forward "unibyte:" tail-end t))))
1657
1658 ;; At first check the head.
1659 (when head-found
1660 (goto-char head-start)
6b66d028
RS
1661 (setq head-end (set-auto-mode-1))
1662 (setq head-start (point))
1d8e9a7c 1663 (when (and head-end (< head-found head-end))
835f49b8 1664 (goto-char head-start)
502522b2
GM
1665 (when (and set-auto-coding-for-load
1666 (re-search-forward
6b66d028 1667 "\\(.*;\\)?[ \t]*unibyte:[ \t]*\\([^ ;]+\\)"
502522b2
GM
1668 head-end t))
1669 (setq coding-system 'raw-text))
1670 (when (and (not coding-system)
1671 (re-search-forward
6b66d028 1672 "\\(.*;\\)?[ \t]*coding:[ \t]*\\([^ ;]+\\)"
502522b2
GM
1673 head-end t))
1674 (setq coding-system (intern (match-string 2)))
1675 (or (coding-system-p coding-system)
1676 (setq coding-system nil)))))
1677
1678 ;; If no coding: tag in the head, check the tail.
c251286b
KH
1679 ;; Here we must pay attention to the case that the end-of-line
1680 ;; is just "\r" and we can't use "^" nor "$" in regexp.
502522b2
GM
1681 (when (and tail-found (not coding-system))
1682 (goto-char tail-start)
c251286b 1683 (re-search-forward "[\r\n]\^L" nil t)
502522b2 1684 (if (re-search-forward
9857367f 1685 "[\r\n]\\([^[\r\n]*\\)[ \t]*Local Variables:[ \t]*\\([^\r\n]*\\)[\r\n]"
c251286b
KH
1686 tail-end t)
1687 ;; The prefix is what comes before "local variables:" in its
1688 ;; line. The suffix is what comes after "local variables:"
502522b2
GM
1689 ;; in its line.
1690 (let* ((prefix (regexp-quote (match-string 1)))
1691 (suffix (regexp-quote (match-string 2)))
1692 (re-coding
1693 (concat
c251286b 1694 "[\r\n]" prefix
cfe98f50
GM
1695 ;; N.B. without the \n below, the regexp can
1696 ;; eat newlines.
c251286b
KH
1697 "[ \t]*coding[ \t]*:[ \t]*\\([^ \t\r\n]+\\)[ \t]*"
1698 suffix "[\r\n]"))
502522b2
GM
1699 (re-unibyte
1700 (concat
c251286b
KH
1701 "[\r\n]" prefix
1702 "[ \t]*unibyte[ \t]*:[ \t]*\\([^ \t\r\n]+\\)[ \t]*"
1703 suffix "[\r\n]"))
502522b2 1704 (re-end
9857367f 1705 (concat "[\r\n]" prefix "[ \t]*End *:[ \t]*" suffix
c251286b
KH
1706 "[\r\n]?"))
1707 (pos (1- (point))))
1708 (forward-char -1) ; skip back \r or \n.
502522b2
GM
1709 (re-search-forward re-end tail-end 'move)
1710 (setq tail-end (point))
1711 (goto-char pos)
1712 (when (and set-auto-coding-for-load
1713 (re-search-forward re-unibyte tail-end t))
1714 (setq coding-system 'raw-text))
1715 (when (and (not coding-system)
1716 (re-search-forward re-coding tail-end t))
1717 (setq coding-system (intern (match-string 1)))
1718 (or (coding-system-p coding-system)
1719 (setq coding-system nil))))))
447404a3
CW
1720 coding-system)
1721 ;; Finally, try all the `auto-coding-functions'.
1722 (let ((funcs auto-coding-functions)
1723 (coding-system nil))
1724 (while (and funcs (not coding-system))
1725 (setq coding-system (condition-case e
1726 (save-excursion
1727 (goto-char (point-min))
1728 (funcall (pop funcs) size))
1729 (error nil))))
502522b2 1730 coding-system)))
63561304
KH
1731
1732(setq set-auto-coding-function 'set-auto-coding)
87aba788 1733
0436cc1b
KH
1734;; This variable is set in these two cases:
1735;; (1) A file is read by a coding system specified explicitly.
1736;; after-insert-file-set-coding sets this value to
1737;; coding-system-for-read.
1738;; (2) A buffer is saved.
1739;; After writing, basic-save-buffer-1 sets this value to
1740;; last-coding-system-used.
1741;; This variable is used for decoding in revert-buffer.
e205f8c1 1742(defvar buffer-file-coding-system-explicit nil
0436cc1b
KH
1743 "The file coding system explicitly specified for the current buffer.
1744Internal use only.")
e205f8c1
KH
1745(make-variable-buffer-local 'buffer-file-coding-system-explicit)
1746(put 'buffer-file-coding-system-explicit 'permanent-local t)
0436cc1b
KH
1747
1748(defun after-insert-file-set-coding (inserted &optional visit)
872a0a6f
RS
1749 "Set `buffer-file-coding-system' of current buffer after text is inserted.
1750INSERTED is the number of characters that were inserted, as figured
1751in the situation before this function. Return the number of characters
1752inserted, as figured in the situation after. The two numbers can be
0436cc1b
KH
1753different if the buffer has become unibyte.
1754The optional second arg VISIT non-nil means that we are visiting a file."
1755 (if (and visit
1756 coding-system-for-read
1757 (not (eq coding-system-for-read 'auto-save-coding)))
e205f8c1 1758 (setq buffer-file-coding-system-explicit coding-system-for-read))
4ed46869
KH
1759 (if last-coding-system-used
1760 (let ((coding-system
1761 (find-new-buffer-file-coding-system last-coding-system-used))
1762 (modified-p (buffer-modified-p)))
0269ddfb 1763 (when coding-system
14b3fa07
RS
1764 ;; Tell set-buffer-file-coding-system not to mark the file
1765 ;; as modified; we just read it, and it's supposed to be unmodified.
1766 ;; Marking it modified would try to lock it, which would
1767 ;; check the modtime, and we don't want to do that again now.
1768 (set-buffer-file-coding-system coding-system t t)
d0c26c63
KH
1769 (if (and enable-multibyte-characters
1770 (or (eq coding-system 'no-conversion)
ccb77b4e 1771 (eq (coding-system-type coding-system) 5))
136e48e4
KH
1772 ;; If buffer was unmodified and the size is the
1773 ;; same as INSERTED, we must be visiting it.
1774 (not modified-p)
1775 (= (buffer-size) inserted))
ccb77b4e
RS
1776 ;; For coding systems no-conversion and raw-text...,
1777 ;; edit the buffer as unibyte.
14b3fa07
RS
1778 (let ((pos-marker (copy-marker (+ (point) inserted)))
1779 ;; Prevent locking.
1780 (buffer-file-name nil))
d0c26c63 1781 (set-buffer-multibyte nil)
872a0a6f 1782 (setq inserted (- pos-marker (point)))))
0269ddfb 1783 (set-buffer-modified-p modified-p))))
d0c26c63 1784 inserted)
4ed46869 1785
8057896b 1786;; The coding-spec and eol-type of coding-system returned is decided
4ed46869
KH
1787;; independently in the following order.
1788;; 1. That of buffer-file-coding-system locally bound.
1789;; 2. That of CODING.
1790
1791(defun find-new-buffer-file-coding-system (coding)
1792 "Return a coding system for a buffer when a file of CODING is inserted.
a73a8c89
KH
1793The local variable `buffer-file-coding-system' of the current buffer
1794is set to the returned value.
509064c5 1795Return nil if there's no need to set `buffer-file-coding-system'."
4ed46869 1796 (let (local-coding local-eol
b685f8d6 1797 found-coding found-eol
4ed46869
KH
1798 new-coding new-eol)
1799 (if (null coding)
1800 ;; Nothing found about coding.
1801 nil
1802
b685f8d6
RS
1803 ;; Get information of `buffer-file-coding-system' in LOCAL-EOL
1804 ;; and LOCAL-CODING.
1805 (setq local-eol (coding-system-eol-type buffer-file-coding-system))
1806 (if (null (numberp local-eol))
1807 ;; But eol-type is not yet set.
1808 (setq local-eol nil))
0269ddfb
KH
1809 (if (and buffer-file-coding-system
1810 (not (eq (coding-system-type buffer-file-coding-system) t)))
1811 ;; This is not `undecided'.
1812 (setq local-coding (coding-system-base buffer-file-coding-system)))
b685f8d6
RS
1813
1814 (if (and (local-variable-p 'buffer-file-coding-system)
1815 local-eol local-coding)
4ed46869
KH
1816 ;; The current buffer has already set full coding-system, we
1817 ;; had better not change it.
1818 nil
1819
8057896b 1820 (setq found-eol (coding-system-eol-type coding))
4ed46869 1821 (if (null (numberp found-eol))
be02cd54
EZ
1822 ;; But eol-type is not found.
1823 ;; If EOL conversions are inhibited, force unix eol-type.
1824 (setq found-eol (if inhibit-eol-conversion 0)))
c76b5c99
KH
1825 (if (eq (coding-system-type coding) t)
1826 (setq found-coding 'undecided)
1827 (setq found-coding (coding-system-base coding)))
1828
1829 (if (and (not found-eol) (eq found-coding 'undecided))
1830 ;; No valid coding information found.
1831 nil
1832
1833 ;; Some coding information (eol or text) found.
1834
1835 ;; The local setting takes precedence over the found one.
1836 (setq new-coding (if (local-variable-p 'buffer-file-coding-system)
1837 (or local-coding found-coding)
1838 (or found-coding local-coding)))
1839 (setq new-eol (if (local-variable-p 'buffer-file-coding-system)
1840 (or local-eol found-eol)
1841 (or found-eol local-eol)))
1842
1843 (let ((eol-type (coding-system-eol-type new-coding)))
1844 (if (and (numberp new-eol) (vectorp eol-type))
1845 (aref eol-type new-eol)
1846 new-coding)))))))
4ed46869 1847
fe831d33
GV
1848(defun modify-coding-system-alist (target-type regexp coding-system)
1849 "Modify one of look up tables for finding a coding system on I/O operation.
8c453b46
RS
1850There are three of such tables, `file-coding-system-alist',
1851`process-coding-system-alist', and `network-coding-system-alist'.
fe831d33
GV
1852
1853TARGET-TYPE specifies which of them to modify.
8c453b46
RS
1854If it is `file', it affects `file-coding-system-alist' (which see).
1855If it is `process', it affects `process-coding-system-alist' (which see).
e8dd0160 1856If it is `network', it affects `network-coding-system-alist' (which see).
fe831d33
GV
1857
1858REGEXP is a regular expression matching a target of I/O operation.
1859The target is a file name if TARGET-TYPE is `file', a program name if
1860TARGET-TYPE is `process', or a network service name or a port number
1861to connect to if TARGET-TYPE is `network'.
1862
1863CODING-SYSTEM is a coding system to perform code conversion on the I/O
8c453b46
RS
1864operation, or a cons cell (DECODING . ENCODING) specifying the coding systems
1865for decoding and encoding respectively,
1866or a function symbol which, when called, returns such a cons cell."
fe831d33
GV
1867 (or (memq target-type '(file process network))
1868 (error "Invalid target type: %s" target-type))
1869 (or (stringp regexp)
1870 (and (eq target-type 'network) (integerp regexp))
1871 (error "Invalid regular expression: %s" regexp))
1872 (if (symbolp coding-system)
1873 (if (not (fboundp coding-system))
1874 (progn
1875 (check-coding-system coding-system)
1876 (setq coding-system (cons coding-system coding-system))))
1877 (check-coding-system (car coding-system))
1878 (check-coding-system (cdr coding-system)))
1879 (cond ((eq target-type 'file)
1880 (let ((slot (assoc regexp file-coding-system-alist)))
1881 (if slot
1882 (setcdr slot coding-system)
1883 (setq file-coding-system-alist
1884 (cons (cons regexp coding-system)
1885 file-coding-system-alist)))))
1886 ((eq target-type 'process)
1887 (let ((slot (assoc regexp process-coding-system-alist)))
1888 (if slot
1889 (setcdr slot coding-system)
1890 (setq process-coding-system-alist
1891 (cons (cons regexp coding-system)
1892 process-coding-system-alist)))))
1893 (t
1894 (let ((slot (assoc regexp network-coding-system-alist)))
1895 (if slot
1896 (setcdr slot coding-system)
1897 (setq network-coding-system-alist
1898 (cons (cons regexp coding-system)
1899 network-coding-system-alist)))))))
1900
db046b7d
KH
1901(defun decode-coding-inserted-region (from to filename
1902 &optional visit beg end replace)
f29387e8 1903 "Decode the region between FROM and TO as if it is read from file FILENAME.
9c848353 1904The idea is that the text between FROM and TO was just inserted somehow.
f29387e8 1905Optional arguments VISIT, BEG, END, and REPLACE are the same as those
9c848353
RS
1906of the function `insert-file-contents'.
1907Part of the job of this function is setting `buffer-undo-list' appropriately."
f29387e8
KH
1908 (save-excursion
1909 (save-restriction
9c848353
RS
1910 (let ((coding coding-system-for-read)
1911 undo-list-saved)
1912 (if visit
1913 ;; Temporarily turn off undo recording, if we're decoding the
1914 ;; text of a visited file.
1915 (setq buffer-undo-list t)
1916 ;; Otherwise, if we can recognize the undo elt for the insertion,
1917 ;; remove it and get ready to replace it later.
1918 ;; In the mean time, turn off undo recording.
bf247b6e 1919 (let ((last (car-safe buffer-undo-list)))
9c848353
RS
1920 (if (and (consp last) (eql (car last) from) (eql (cdr last) to))
1921 (setq undo-list-saved (cdr buffer-undo-list)
1922 buffer-undo-list t))))
1923 (narrow-to-region from to)
1924 (goto-char (point-min))
f29387e8
KH
1925 (or coding
1926 (setq coding (funcall set-auto-coding-function
1927 filename (- (point-max) (point-min)))))
1928 (or coding
b69ff51b
KH
1929 (setq coding (car (find-operation-coding-system
1930 'insert-file-contents
1931 filename visit beg end replace))))
f29387e8
KH
1932 (if (coding-system-p coding)
1933 (or enable-multibyte-characters
1934 (setq coding
1935 (coding-system-change-text-conversion coding 'raw-text)))
1936 (setq coding nil))
1937 (if coding
b12e19b2 1938 (decode-coding-region (point-min) (point-max) coding)
9c848353
RS
1939 (setq last-coding-system-used coding))
1940 ;; If we're decoding the text of a visited file,
1941 ;; the undo list should start out empty.
1942 (if visit
1943 (setq buffer-undo-list nil)
1944 ;; If we decided to replace the undo entry for the insertion,
1945 ;; do so now.
1946 (if undo-list-saved
1947 (setq buffer-undo-list
1948 (cons (cons from (point-max)) undo-list-saved))))))))
f29387e8 1949
27a91cf7
KH
1950(defun recode-region (start end new-coding coding)
1951 "Re-decode the region (previously decoded by CODING) by NEW-CODING."
1952 (interactive
1953 (list (region-beginning) (region-end)
1954 (read-coding-system "Text was really in: ")
1955 (let ((coding (or buffer-file-coding-system last-coding-system-used)))
1956 (read-coding-system
1957 (concat "But was interpreted as"
1958 (if coding (format " (default %S): " coding) ": "))
1959 coding))))
1960 (or (and new-coding coding)
1961 (error "Coding system not specified"))
1962 ;; Check it before we encode the region.
1963 (check-coding-system new-coding)
1964 (save-restriction
1965 (narrow-to-region start end)
1966 (encode-coding-region (point-min) (point-max) coding)
1967 (decode-coding-region (point-min) (point-max) new-coding)))
1968
b25eef20 1969(defun make-translation-table (&rest args)
a284eea3 1970 "Make a translation table from arguments.
d38b07f9 1971A translation table is a char table intended for character
a284eea3
DL
1972translation in CCL programs.
1973
d38b07f9 1974Each argument is a list of elements of the form (FROM . TO), where FROM
a284eea3 1975is a character to be translated to TO.
13d5617d 1976
452fdb31 1977FROM can be a generic character (see `make-char'). In this case, TO is
d38b07f9 1978a generic character containing the same number of characters, or an
452fdb31 1979ordinary character. If FROM and TO are both generic characters, all
b25eef20 1980characters belonging to FROM are translated to characters belonging to TO
4e003d37
KH
1981without changing their position code(s).
1982
1983The arguments and forms in each argument are processed in the given
1984order, and if a previous form already translates TO to some other
1985character, say TO-ALT, FROM is also translated to TO-ALT."
f967223b 1986 (let ((table (make-char-table 'translation-table))
a73a8c89
KH
1987 revlist)
1988 (while args
1989 (let ((elts (car args)))
1990 (while elts
13d5617d
KH
1991 (let* ((from (car (car elts)))
1992 (from-i 0) ; degree of freedom of FROM
1993 (from-rev (nreverse (split-char from)))
1994 (to (cdr (car elts)))
1995 (to-i 0) ; degree of freedom of TO
1996 (to-rev (nreverse (split-char to))))
1997 ;; Check numbers of heading 0s in FROM-REV and TO-REV.
1998 (while (eq (car from-rev) 0)
1999 (setq from-i (1+ from-i) from-rev (cdr from-rev)))
2000 (while (eq (car to-rev) 0)
2001 (setq to-i (1+ to-i) to-rev (cdr to-rev)))
2002 (if (and (/= from-i to-i) (/= to-i 0))
2003 (error "Invalid character pair (%d . %d)" from to))
b25eef20
KH
2004 ;; If we have already translated TO to TO-ALT, FROM should
2005 ;; also be translated to TO-ALT. But, this is only if TO
2006 ;; is a generic character or TO-ALT is not a generic
13d5617d
KH
2007 ;; character.
2008 (let ((to-alt (aref table to)))
2009 (if (and to-alt
2010 (or (> to-i 0) (not (generic-char-p to-alt))))
2011 (setq to to-alt)))
2012 (if (> from-i 0)
2013 (set-char-table-default table from to)
2014 (aset table from to))
b25eef20
KH
2015 ;; If we have already translated some chars to FROM, they
2016 ;; should also be translated to TO.
a73a8c89
KH
2017 (let ((l (assq from revlist)))
2018 (if l
2019 (let ((ch (car l)))
2020 (setcar l to)
2021 (setq l (cdr l))
2022 (while l
2023 (aset table ch to)
2024 (setq l (cdr l)) ))))
2025 ;; Now update REVLIST.
2026 (let ((l (assq to revlist)))
2027 (if l
2028 (setcdr l (cons from (cdr l)))
2029 (setq revlist (cons (list to from) revlist)))))
2030 (setq elts (cdr elts))))
2031 (setq args (cdr args)))
2032 ;; Return TABLE just created.
2033 table))
2034
c76b5c99
KH
2035(defun make-translation-table-from-vector (vec)
2036 "Make translation table from decoding vector VEC.
9e3b6057
DL
2037VEC is an array of 256 elements to map unibyte codes to multibyte
2038characters. Elements may be nil for undefined code points.
c76b5c99
KH
2039See also the variable `nonascii-translation-table'."
2040 (let ((table (make-char-table 'translation-table))
2041 (rev-table (make-char-table 'translation-table))
c76b5c99 2042 ch)
9e3b6057 2043 (dotimes (i 256)
c76b5c99 2044 (setq ch (aref vec i))
9e3b6057
DL
2045 (when ch
2046 (aset table i ch)
2047 (if (>= ch 256)
2048 (aset rev-table ch i))))
c76b5c99
KH
2049 (set-char-table-extra-slot table 0 rev-table)
2050 table))
2051
f967223b 2052(defun define-translation-table (symbol &rest args)
a284eea3
DL
2053 "Define SYMBOL as the name of translation table made by ARGS.
2054This sets up information so that the table can be used for
2055translations in a CCL program.
b25eef20 2056
a284eea3
DL
2057If the first element of ARGS is a char-table whose purpose is
2058`translation-table', just define SYMBOL to name it. (Note that this
2059function does not bind SYMBOL.)
007c79c8 2060
a284eea3 2061Any other ARGS should be suitable as arguments of the function
007c79c8 2062`make-translation-table' (which see).
b25eef20 2063
452fdb31 2064This function sets properties `translation-table' and
521d4010
DL
2065`translation-table-id' of SYMBOL to the created table itself and the
2066identification number of the table respectively. It also registers
2067the table in `translation-table-vector'."
007c79c8
KH
2068 (let ((table (if (and (char-table-p (car args))
2069 (eq (char-table-subtype (car args))
2070 'translation-table))
2071 (car args)
2072 (apply 'make-translation-table args)))
f967223b 2073 (len (length translation-table-vector))
d9e3229d 2074 (id 0)
b25eef20 2075 (done nil))
f967223b 2076 (put symbol 'translation-table table)
b25eef20
KH
2077 (while (not done)
2078 (if (>= id len)
f967223b
KH
2079 (setq translation-table-vector
2080 (vconcat translation-table-vector (make-vector len nil))))
2081 (let ((slot (aref translation-table-vector id)))
b25eef20
KH
2082 (if (or (not slot)
2083 (eq (car slot) symbol))
2084 (progn
f967223b 2085 (aset translation-table-vector id (cons symbol table))
007c79c8
KH
2086 (setq done t))
2087 (setq id (1+ id)))))
f967223b 2088 (put symbol 'translation-table-id id)
d9e3229d
KH
2089 id))
2090
0e86dca1
KH
2091(defun translate-region (start end table)
2092 "From START to END, translate characters according to TABLE.
2093TABLE is a string or a char-table.
2094If TABLE is a string, the Nth character in it is the mapping
2095for the character with code N.
2096If TABLE is a char-table, the element for character N is the mapping
2097for the character with code N.
2098It returns the number of characters changed."
2099 (interactive
2100 (list (region-beginning)
2101 (region-end)
2102 (let (table l)
2103 (dotimes (i (length translation-table-vector))
2104 (if (consp (aref translation-table-vector i))
2105 (push (list (symbol-name
2106 (car (aref translation-table-vector i)))) l)))
2107 (if (not l)
2108 (error "No translation table defined"))
2109 (while (not table)
2110 (setq table (completing-read "Translation table: " l nil t)))
2111 (intern table))))
2112 (if (symbolp table)
2113 (let ((val (get table 'translation-table)))
2114 (or (char-table-p val)
2115 (error "Invalid translation table name: %s" table))
2116 (setq table val)))
2117 (translate-region-internal start end table))
2118
35554641
KH
2119(put 'with-category-table 'lisp-indent-function 1)
2120
ef6e365d
JPW
2121(defmacro with-category-table (table &rest body)
2122 "Evaluate BODY with category table of current buffer set to TABLE.
2123The category table of the current buffer is saved, BODY is evaluated,
2124then the saved table is restored, even in case of an abnormal exit.
2125Value is what BODY returns."
2126 (let ((old-table (make-symbol "old-table"))
2127 (old-buffer (make-symbol "old-buffer")))
2128 `(let ((,old-table (category-table))
2129 (,old-buffer (current-buffer)))
2130 (unwind-protect
2131 (progn
2132 (set-category-table ,table)
2133 ,@body)
2134 (save-current-buffer
2135 (set-buffer ,old-buffer)
2136 (set-category-table ,old-table))))))
35554641 2137
394e4eb0
DL
2138(defun define-translation-hash-table (symbol table)
2139 "Define SYMBOL as the name of the hash translation TABLE for use in CCL.
2140
2141Analogous to `define-translation-table', but updates
2142`translation-hash-table-vector' and the table is for use in the CCL
2143`lookup-integer' and `lookup-character' functions."
2144 (unless (and (symbolp symbol)
2145 (hash-table-p table))
2146 (error "Bad args to define-translation-hash-table"))
2147 (let ((len (length translation-hash-table-vector))
2148 (id 0)
2149 done)
2150 (put symbol 'translation-hash-table table)
2151 (while (not done)
2152 (if (>= id len)
2153 (setq translation-hash-table-vector
2154 (vconcat translation-hash-table-vector [nil])))
2155 (let ((slot (aref translation-hash-table-vector id)))
2156 (if (or (not slot)
2157 (eq (car slot) symbol))
2158 (progn
2159 (aset translation-hash-table-vector id (cons symbol table))
2160 (setq done t))
2161 (setq id (1+ id)))))
2162 (put symbol 'translation-hash-table-id id)
2163 id))
2164
69eba008
KH
2165;;; Initialize some variables.
2166
2167(put 'use-default-ascent 'char-table-extra-slots 0)
2168(setq use-default-ascent (make-char-table 'use-default-ascent))
d6d6d592
KH
2169(put 'ignore-relative-composition 'char-table-extra-slots 0)
2170(setq ignore-relative-composition
2171 (make-char-table 'ignore-relative-composition))
69eba008 2172
d9f6dfe6
CW
2173
2174;;; Built-in auto-coding-functions:
2175
2176(defun sgml-xml-auto-coding-function (size)
2177 "Determine whether the buffer is XML, and if so, its encoding.
2178This function is intended to be added to `auto-coding-functions'."
c069d3ac
SM
2179 (setq size (+ (point) size))
2180 (when (re-search-forward "\\`[[:space:]\n]*<\\?xml" size t)
d9f6dfe6
CW
2181 (let ((end (save-excursion
2182 ;; This is a hack.
2f4e9c47 2183 (re-search-forward "\"\\s-*\\?>" size t))))
d9f6dfe6
CW
2184 (when end
2185 (if (re-search-forward "encoding=\"\\(.+?\\)\"" end t)
447404a3
CW
2186 (let* ((match (match-string 1))
2187 (sym (intern (downcase match))))
2188 (if (coding-system-p sym)
2189 sym
2190 (message "Warning: unknown coding system \"%s\"" match)
2191 nil))
d9f6dfe6
CW
2192 'utf-8)))))
2193
447404a3
CW
2194(defun sgml-html-meta-auto-coding-function (size)
2195 "If the buffer has an HTML meta tag, use it to determine encoding.
2196This function is intended to be added to `auto-coding-functions'."
c069d3ac 2197 (setq size (min (+ (point) size)
447404a3
CW
2198 ;; Only search forward 10 lines
2199 (save-excursion
2200 (forward-line 10)
2201 (point))))
81668eb2 2202 (when (and (search-forward "<html" size t)
447404a3
CW
2203 (re-search-forward "<meta\\s-+http-equiv=\"content-type\"\\s-+content=\"text/\\sw+;\\s-*charset=\\(.+?\\)\"" size t))
2204 (let* ((match (match-string 1))
2205 (sym (intern (downcase match))))
2206 (if (coding-system-p sym)
2207 sym
2208 (message "Warning: unknown coding system \"%s\"" match)
2209 nil))))
0bca779a 2210
69eba008 2211;;;
4ed46869
KH
2212(provide 'mule)
2213
ab5796a9 2214;;; arch-tag: 9aebaa6e-0e8a-40a9-b857-cb5d04a39e7c
4ed46869 2215;;; mule.el ends here