1 ;;; utf-16.el --- UTF-16 encoding/decoding
3 ;; Copyright (C) 2001, 2002, 2003, 2004 Free Software Foundation, Inc.
4 ;; Copyright (C) 2002, 2003, 2004
5 ;; National Institute of Advanced Industrial Science and Technology (AIST)
6 ;; Registration Number H14PRO021
8 ;; Author: Dave Love <fx@gnu.org>
9 ;; Keywords: Unicode, UTF-16, i18n
11 ;; This file is part of GNU Emacs.
13 ;; GNU Emacs is free software; you can redistribute it and/or modify
14 ;; it under the terms of the GNU General Public License as published by
15 ;; the Free Software Foundation; either version 2, or (at your option)
18 ;; GNU Emacs is distributed in the hope that it will be useful,
19 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
20 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
21 ;; GNU General Public License for more details.
23 ;; You should have received a copy of the GNU General Public License
24 ;; along with GNU Emacs; see the file COPYING. If not, write to the
25 ;; Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
26 ;; Boston, MA 02110-1301, USA.
30 ;; Support for UTF-16, which is a two-byte encoding (modulo
31 ;; surrogates) of Unicode, defined in RFC 2781. It is written either
32 ;; in little or big endian order and either with or without the
33 ;; leading BOM (a two-byte signature which identifies their byte sex).
35 ;; We provide these base coding systems.
38 ;; mule-utf-16le little no
39 ;; mule-utf-16be big no
40 ;; mule-utf-16le-with-signature little yes
41 ;; mule-utf-16be-with-signature big yes
42 ;; mule-utf-16 both yes
44 ;; Note that un-decodable sequences aren't (yet?) preserved as raw
45 ;; bytes, as they are with utf-8, so reading and writing as utf-16 can
50 ;; We end up with trivially different -le and -be versions of most
51 ;; things below, sometimes with commonality abstracted into a let
52 ;; binding for maintenance convenience.
54 ;; Needed in macro expansion, so can't be let-bound. Zapped after use.
56 (defconst utf-16-decode-ucs
57 ;; If r5 is negative, r1 is a Unicode chacter code. Otherise, r5 is
58 ;; the first of a surrogate pair and r1 is the second of the pair.
59 ;; Output is charset ID in r0, code point in r1. R0 may be set to
60 ;; -1 in which case a caller should not write out r1.
63 (if ((r1 >= #xE000) | r0)
64 ;; Invalid second code of surrogate pair.
66 (call ccl-mule-utf-untrans))
68 (r1 += (((r5 - #xD800) << 10) + #x10000))))
71 (r0 = ,(charset-id 'ascii))
72 ((lookup-integer utf-subst-table-for-decode r1 r3)
73 (if r7 ; got a translation
76 (r0 = ,(charset-id 'eight-bit-control))
78 ((r0 = ,(charset-id 'latin-iso8859-1))
81 ((r0 = ,(charset-id 'mule-unicode-0100-24ff))
83 (r2 = (((r1 / 96) + 32) << 7))
87 ((r0 = ,(charset-id 'mule-unicode-2500-33ff))
89 (r2 = (((r1 / 96) + 32) << 7))
93 ;; We can't have this character.
95 (call ccl-mule-utf-untrans)
99 ;; The first code of a surrogate pair.
103 ;; The second code of a surrogate pair, invalid.
105 (call ccl-mule-utf-untrans)
109 ((r0 = ,(charset-id 'mule-unicode-e000-ffff))
111 (r2 = (((r1 / 96) + 32) << 7))
114 ;; We can't have this character.
116 (call ccl-mule-utf-untrans)
118 (r0 = -1)))))))))))))))
120 (defconst utf-16le-decode-loop
128 ((translate-character utf-translation-table-for-decode r0 r1)
129 (write-multibyte-character r0 r1)))
132 (defconst utf-16be-decode-loop
140 ((translate-character utf-translation-table-for-decode r0 r1)
141 (write-multibyte-character r0 r1)))
146 (define-ccl-program ccl-decode-mule-utf-16le
147 `(2 ; 2 bytes -> 1 to 4 bytes
148 ,utf-16le-decode-loop
151 (call ccl-mule-utf-untrans)))
155 (r0 = ,(charset-id 'eight-bit-control))
156 (r0 = ,(charset-id 'eight-bit-graphic)))
157 (write-multibyte-character r0 r3)))))
158 "Decode UTF-16LE (little endian without signature bytes).
159 Basic decoding is done into the charsets ascii, latin-iso8859-1 and
160 mule-unicode-*. Un-representable Unicode characters are decoded as
161 U+fffd. The result is run through the translation-table named
162 `utf-translation-table-for-decode'.")
164 (define-ccl-program ccl-decode-mule-utf-16be
165 `(2 ; 2 bytes -> 1 to 4 bytes
166 ,utf-16be-decode-loop
169 (call ccl-mule-utf-untrans)))
172 (call ccl-mule-utf-untrans)))))
173 "Decode UTF-16BE (big endian without signature bytes).
174 Basic decoding is done into the charsets ascii, latin-iso8859-1 and
175 mule-unicode-*. Un-representable Unicode characters are
176 decoded as U+fffd. The result is run through the translation-table of
177 name `utf-translation-table-for-decode'.")
179 (define-ccl-program ccl-decode-mule-utf-16le-with-signature
183 ,@utf-16le-decode-loop)
186 (call ccl-mule-utf-untrans))))
187 "Like ccl-decode-utf-16le but skip the first 2-byte BOM.")
189 (define-ccl-program ccl-decode-mule-utf-16be-with-signature
193 ,@utf-16be-decode-loop)
196 (call ccl-mule-utf-untrans))))
197 "Like ccl-decode-utf-16be but skip the first 2-byte BOM.")
199 (define-ccl-program ccl-decode-mule-utf-16
206 ;; R1 is a BOM for little endian. We keep this character as
207 ;; is temporarily. It is removed by post-read-conversion
210 (write-multibyte-character r0 r1)
211 ,@utf-16le-decode-loop)
213 ;; R1 is a BOM for big endian, but we can't keep that
214 ;; character in the output because it can't be
215 ;; distinguished with the normal U+FEFF. So, we keep
219 (write-multibyte-character r0 r1))
220 ;; R1 is a normal Unicode character.
223 ((translate-character utf-translation-table-for-decode r0 r1)
224 (write-multibyte-character r0 r1)))))
225 ,@utf-16be-decode-loop)))
228 (call ccl-mule-utf-untrans))))
229 "Like ccl-decode-utf-16be/le but check the first BOM.")
231 (makunbound 'utf-16-decode-ucs) ; done with it
232 (makunbound 'utf-16le-decode-loop)
233 (makunbound 'utf-16be-decode-loop)
235 ;; UTF-16 decoder generates an UTF-8 sequence represented by a
236 ;; sequence eight-bit-control/graphic chars for an invalid byte (the
237 ;; last byte of an odd length source) and an untranslatable character
238 ;; (including an invalid surrogate-pair code-point).
240 ;; This CCL parses that sequence (the first byte is already in r1),
241 ;; and if the sequence represents an untranslatable character, it sets
242 ;; r1 to the original invalid code or untranslated Unicode character
243 ;; code, sets r2 to -1 (to prevent r2 and r3 are written), set2 r5 to
244 ;; -1 (to tell the caller that there's no pre-read character).
246 ;; If the sequence represents an invalid byte, it sets r1 to -1, r2 to
247 ;; the byte, sets r3 and r5 to -1.
249 ;; Otherwise, don't change r1, set r2 and r3 to already read
250 ;; eight-bit-control/graphic characters (if any), set r5 and r6 to the
251 ;; last character that invalidates the UTF-8 form.
253 ;; Note: For UTF-8 validation, we only check if a character is
254 ;; eight-bit-control/graphic or not. It may result in incorrect
255 ;; handling of random binary data, but such a data can't be encoded by
256 ;; UTF-16 anyway. At least, UTF-16 decoder doesn't generate such a
257 ;; sequence even if a source contains invalid byte-sequence.
259 (define-ccl-program ccl-mule-utf-16-encode-untrans
262 ;; Read the 2nd byte.
263 (read-multibyte-character r5 r6)
264 (r0 = (r5 != ,(charset-id 'eight-bit-control)))
265 (if ((r5 != ,(charset-id 'eight-bit-graphic)) & r0)
269 (end))) ; invalid UTF-8
274 ;; 2-byte UTF-8, i.e. originally an invalid byte.
276 (r2 |= ((r1 & #x1F) << 6))
281 ;; Read the 3rd byte.
282 (read-multibyte-character r5 r6)
283 (r0 = (r5 != ,(charset-id 'eight-bit-control)))
284 (if ((r5 != ,(charset-id 'eight-bit-graphic)) & r0)
285 ((end))) ; invalid UTF-8
287 (if (r1 < #xF0) ; valid 3-byte UTF-8
288 ((r1 = ((r1 & #x0F) << 12))
289 (r1 |= ((r2 & #x3F) << 6))
296 ;; Read the 4th byte.
297 (read-multibyte-character r5 r6)
298 (r0 = (r5 != ,(charset-id 'eight-bit-control)))
299 (if ((r5 != ,(charset-id 'eight-bit-graphic)) & r0)
300 (end)) ; livalid UTF-8
302 ;; valid 4-byte UTF-8
303 (r1 = ((r1 & #x07) << 18))
304 (r1 |= ((r2 & #x3F) << 12))
305 (r1 |= ((r3 & #x3F) << 6))
319 (defconst utf-16-decode-to-ucs
320 ;; Read a character and set r1 to the corresponding Unicode code.
321 ;; If r5 is not negative, it means that we have already read a
322 ;; character into r5 and r6.
323 ;; If an invalid eight-bit-control/graphic sequence is found, r2 and
324 ;; r3 may contain a byte to written out, r5 and r6 may contain a
325 ;; pre-read character. Usually they are set to -1.
327 (read-multibyte-character r0 r1)
331 (lookup-character utf-subst-table-for-encode r0 r1)
335 ((translate-character utf-translation-table-for-encode r0 r1)
336 (if (r0 == ,(charset-id 'ascii))
338 (if (r0 == ,(charset-id 'latin-iso8859-1))
340 (if (r0 == ,(charset-id 'eight-bit-control))
342 (if (r0 == ,(charset-id 'eight-bit-graphic))
343 (call ccl-mule-utf-16-encode-untrans)
344 ((r2 = ((r1 & #x7f) - 32))
345 (r3 = ((((r1 >> 7) - 32) * 96) + r2))
348 (if (r0 == ,(charset-id 'mule-unicode-0100-24ff))
350 (if (r0 == ,(charset-id 'mule-unicode-2500-33ff))
352 (if (r0 == ,(charset-id 'mule-unicode-e000-ffff))
354 (r1 = #xfffd)))))))))))))
356 (defconst utf-16le-encode-loop
359 ,@utf-16-decode-to-ucs
362 (r0 = ((r1 >> 10) + #xD800))
365 (r1 = ((r1 & #x3FF) + #xDC00))))
375 (defconst utf-16be-encode-loop
378 ,@utf-16-decode-to-ucs
381 (r0 = ((r1 >> 10) + #xD800))
384 (r1 = ((r1 & #x3FF) + #xDC00))))
396 (define-ccl-program ccl-encode-mule-utf-16le
398 ,utf-16le-encode-loop)
399 "Encode to UTF-16LE (little endian without signature).
400 Characters from the charsets ascii, eight-bit-control,
401 eight-bit-graphic, latin-iso8859-1 and mule-unicode-* are encoded
402 after translation through the translation-table of name
403 `utf-translation-table-for-encode'.
404 Others are encoded as U+FFFD.")
406 (define-ccl-program ccl-encode-mule-utf-16be
408 ,utf-16be-encode-loop)
409 "Encode to UTF-16BE (big endian without signature).
410 Characters from the charsets ascii, eight-bit-control,
411 eight-bit-graphic, latin-iso8859-1 and mule-unicode-* are encoded
412 after translation through the translation-table named
413 `utf-translation-table-for-encode'.
414 Others are encoded as U+FFFD.")
416 (define-ccl-program ccl-encode-mule-utf-16le-with-signature
420 ,@utf-16le-encode-loop))
421 "Encode to UTF-16 (little endian with signature).
422 Characters from the charsets ascii, eight-bit-control,
423 eight-bit-graphic, latin-iso8859-1 and mule-unicode-* are encoded
424 after translation through the translation-table of name
425 `utf-translation-table-for-encode'.
426 Others are encoded as U+FFFD.")
428 (define-ccl-program ccl-encode-mule-utf-16be-with-signature
432 ,@utf-16be-encode-loop))
433 "Encode to UTF-16 (big endian with signature).
434 Characters from the charsets ascii, eight-bit-control,
435 eight-bit-graphic, latin-iso8859-1 and mule-unicode-* are encoded
436 after translation through the translation-table named
437 `utf-translation-table-for-encode'.
438 Others are encoded as U+FFFD.")
440 (makunbound 'utf-16-decode-to-ucs)
441 (makunbound 'utf-16le-encode-loop)
442 (makunbound 'utf-16be-encode-loop)
444 (defun mule-utf-16-post-read-conversion (length)
446 (setq length (utf-8-post-read-conversion length))
447 (let ((char (following-char)))
448 (cond ((= char (decode-char 'ucs #xFFFE))
450 (setq last-coding-system-used
451 (coding-system-change-text-conversion
452 last-coding-system-used
453 'mule-utf-16le-with-signature))
454 (setq length (1- length)))
455 ((= char (decode-char 'ucs #xFFFF))
457 (setq last-coding-system-used
458 (coding-system-change-text-conversion
459 last-coding-system-used
460 'mule-utf-16be-with-signature))
461 (setq length (1- length)))
463 (setq last-coding-system-used 'mule-utf-16be)))))
468 It supports Unicode characters of these ranges:
469 U+0000..U+33FF, U+E000..U+FFFF.
470 They correspond to these Emacs character sets:
471 ascii, latin-iso8859-1, mule-unicode-0100-24ff,
472 mule-unicode-2500-33ff, mule-unicode-e000-ffff
474 On decoding (e.g. reading a file), Unicode characters not in the above
475 ranges are decoded as U+FFFD, effectively corrupting the data
476 if they are re-encoded.
478 On encoding (e.g. writing a file), Emacs characters not belonging to
479 any of the character sets listed above are encoded into the byte
480 sequence representing U+FFFD (REPLACEMENT CHARACTER).")
481 (props `((safe-charsets
486 mule-unicode-0100-24ff
487 mule-unicode-2500-33ff
488 mule-unicode-e000-ffff
489 ,@(if utf-translate-cjk-mode
490 utf-translate-cjk-charsets))
491 (valid-codes (0 . 255))
492 (mime-text-unsuitable . t)
493 (pre-write-conversion . utf-8-pre-write-conversion)
494 (dependency unify-8859-on-encoding-mode
495 unify-8859-on-decoding-mode
496 utf-fragment-on-decoding
497 utf-translate-cjk-mode))))
500 ?u ; Mule-UCS uses ?U, but code-pages uses that for koi8-u.
502 "UTF-16LE encoding for Emacs-supported Unicode characters."
504 '(ccl-decode-mule-utf-16le . ccl-encode-mule-utf-16le)
506 (post-read-conversion . utf-8-post-read-conversion)
507 (ascii-incompatible . t)
508 (mime-charset . utf-16le)))
513 "UTF-16BE encoding for Emacs-supported Unicode characters."
516 '(ccl-decode-mule-utf-16be . ccl-encode-mule-utf-16be)
518 (post-read-conversion . utf-8-post-read-conversion)
519 (ascii-incompatible . t)
520 (mime-charset . utf-16be)))
523 'mule-utf-16le-with-signature 4 ?u
525 "Little endian UTF-16 (with BOM) for Emacs-supported Unicode characters."
528 '(ccl-decode-mule-utf-16le-with-signature
529 . ccl-encode-mule-utf-16le-with-signature)
531 (post-read-conversion . utf-8-post-read-conversion)
532 (coding-category . coding-category-utf-16-le)
533 (ascii-incompatible . t)
534 (mime-charset . utf-16)))
537 'mule-utf-16be-with-signature 4 ?u
539 "Big endian UTF-16 (with BOM) for Emacs-supported Unicode characters."
542 '(ccl-decode-mule-utf-16be-with-signature
543 . ccl-encode-mule-utf-16be-with-signature)
545 (post-read-conversion . utf-8-post-read-conversion)
546 (coding-category . coding-category-utf-16-be)
547 (ascii-incompatible . t)
548 (mime-charset . utf-16)))
553 "UTF-16 (with or without BOM) for Emacs-supported Unicode characters."
556 '(ccl-decode-mule-utf-16 . ccl-encode-mule-utf-16be-with-signature)
558 (post-read-conversion . mule-utf-16-post-read-conversion)
559 (coding-category . coding-category-utf-16-be)
560 (ascii-incompatible . t)
561 (mime-charset . utf-16)))
564 (define-coding-system-alias 'utf-16le 'mule-utf-16le)
565 (define-coding-system-alias 'utf-16be 'mule-utf-16be)
566 (define-coding-system-alias 'utf-16le-with-signature
567 'mule-utf-16le-with-signature)
568 (define-coding-system-alias 'utf-16be-with-signature
569 'mule-utf-16be-with-signature)
570 (define-coding-system-alias 'utf-16 'mule-utf-16)
572 ;; For backward compatibility.
573 (define-coding-system-alias 'mule-utf-16-le 'mule-utf-16le-with-signature)
574 (define-coding-system-alias 'utf-16-le 'mule-utf-16le-with-signature)
575 (define-coding-system-alias 'mule-utf-16-be 'mule-utf-16be-with-signature)
576 (define-coding-system-alias 'utf-16-be 'mule-utf-16be-with-signature)
578 ;;; arch-tag: 85455d46-d9c9-466d-a6f3-c3582a7367c4
579 ;;; utf-16.el ends here