1 ;;; utf-8.el --- UTF-8 decoding/encoding support -*- coding: iso-2022-7bit -*-
3 ;; Copyright (C) 2001 Electrotechnical Laboratory, JAPAN.
4 ;; Licensed to the Free Software Foundation.
5 ;; Copyright (C) 2001, 2002 Free Software Foundation, Inc.
7 ;; Author: TAKAHASHI Naoto <ntakahas@m17n.org>
9 ;; Keywords: multilingual, Unicode, UTF-8, 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., 59 Temple Place - Suite 330,
26 ;; Boston, MA 02111-1307, USA.
30 ;; The coding-system `mule-utf-8' basically supports encoding/decoding
31 ;; of the following character sets to and from UTF-8:
36 ;; mule-unicode-0100-24ff
37 ;; mule-unicode-2500-33ff
38 ;; mule-unicode-e000-ffff
40 ;; On decoding, Unicode characters that do not fit into the above
41 ;; character sets are handled as `eight-bit-control' or
42 ;; `eight-bit-graphic' characters to retain the information about the
43 ;; original byte sequence and text properties record the corresponding
46 ;; Fixme: note that reading and writing invalid utf-8 may not be
47 ;; idempotent -- to represent the bytes to fix that needs a new charset.
49 ;; Characters from other character sets can be encoded with mule-utf-8
50 ;; by populating the translation table
51 ;; `utf-translation-table-for-encode' and registering the translation
52 ;; with `register-char-codings'. Hash tables
53 ;; `utf-subst-table-for-decode' and `utf-subst-table-for-encode' are
54 ;; used to support encoding and decoding of about a quarter of the CJK
55 ;; space between U+3400 and U+DFFF.
57 ;; UTF-8 is defined in RFC 2279. A sketch of the encoding is:
60 ;; value | 1st byte | 2nd byte | 3rd byte
61 ;; --------------------+-----------+-----------+----------
62 ;; 0000 0000 0xxx xxxx | 0xxx xxxx | |
63 ;; 0000 0yyy yyxx xxxx | 110y yyyy | 10xx xxxx |
64 ;; zzzz yyyy yyxx xxxx | 1110 zzzz | 10yy yyyy | 10xx xxxx
68 (defvar ucs-mule-to-mule-unicode (make-char-table 'translation-table nil)
69 "Char table mapping characters to latin-iso8859-1 or mule-unicode-*.
71 If `unify-8859-on-encoding-mode' is non-nil, this table populates the
72 translation-table named `utf-translation-table-for-encode'.")
74 (define-translation-table 'utf-translation-table-for-encode)
77 ;; Map Cyrillic and Greek to iso-8859 charsets, which take half the
78 ;; space of mule-unicode. For Latin scripts this isn't very
79 ;; important. Hebrew and Arabic might go here too when there's proper
82 (defvar utf-fragmentation-table (make-char-table 'translation-table nil)
83 "Char-table normally mapping non-Latin mule-unicode-* chars to iso-8859-*.
85 If `utf-fragment-on-decoding' is non-nil, this table populates the
86 translation-table named `utf-translation-table-for-decode'")
88 (defvar utf-defragmentation-table (make-char-table 'translation-table nil)
89 "Char-table for reverse mapping of `utf-fragmentation-table'.
91 If `utf-fragment-on-decoding' is non-nil and
92 `unify-8859-on-encoding-mode' is nil, this table populates the
93 translation-table named `utf-translation-table-for-encode'")
95 (define-translation-table 'utf-translation-table-for-decode)
98 (defvar ucs-mule-cjk-to-unicode (make-hash-table :test 'eq :size 43000
100 "Hash table mapping Emacs CJK character sets to Unicode code points.
102 If `utf-translate-cjk' is non-nil, this table populates the
103 translation-hash-table named `utf-subst-table-for-encode'.")
105 (define-translation-hash-table 'utf-subst-table-for-encode
106 (make-hash-table :test 'eq :size 43000 :rehash-size 1000))
108 (defvar ucs-unicode-to-mule-cjk (make-hash-table :test 'eq :size 43000
110 "Hash table mapping Unicode code points to Emacs CJK character sets.
112 If `utf-translate-cjk' is non-nil, this table populates the
113 translation-hash-table named `utf-subst-table-for-decode'.")
115 (define-translation-hash-table 'utf-subst-table-for-decode
116 (make-hash-table :test 'eq :size 21500 :rehash-size 200))
120 (aset utf-fragmentation-table (car pair) (cdr pair))
121 (aset utf-defragmentation-table (cdr pair) (car pair)))
122 '((?
\e$,1&d
\e(B . ?
\e,F4
\e(B) (?
\e$,1&e
\e(B . ?
\e,F5
\e(B) (?
\e$,1&f
\e(B . ?
\e,F6
\e(B) (?
\e$,1&h
\e(B . ?
\e,F8
\e(B) (?
\e$,1&i
\e(B . ?
\e,F9
\e(B)
123 (?
\e$,1&j
\e(B . ?
\e,F:
\e(B) (?
\e$,1&l
\e(B . ?
\e,F<
\e(B) (?
\e$,1&n
\e(B . ?
\e,F>
\e(B) (?
\e$,1&o
\e(B . ?
\e,F?
\e(B) (?
\e$,1&p
\e(B . ?
\e,F@
\e(B)
124 (?
\e$,1&q
\e(B . ?
\e,FA
\e(B) (?
\e$,1&r
\e(B . ?
\e,FB
\e(B) (?
\e$,1&s
\e(B . ?
\e,FC
\e(B) (?
\e$,1&t
\e(B . ?
\e,FD
\e(B) (?
\e$,1&u
\e(B . ?
\e,FE
\e(B)
125 (?
\e$,1&v
\e(B . ?
\e,FF
\e(B) (?
\e$,1&w
\e(B . ?
\e,FG
\e(B) (?
\e$,1&x
\e(B . ?
\e,FH
\e(B) (?
\e$,1&y
\e(B . ?
\e,FI
\e(B) (?
\e$,1&z
\e(B . ?
\e,FJ
\e(B)
126 (?
\e$,1&{
\e(B . ?
\e,FK
\e(B) (?
\e$,1&|
\e(B . ?
\e,FL
\e(B) (?
\e$,1&}
\e(B . ?
\e,FM
\e(B) (?
\e$,1&~
\e(B . ?
\e,FN
\e(B) (?
\e$,1&
\7f\e(B . ?
\e,FO
\e(B)
127 (?
\e$,1'
\e(B . ?
\e,FP
\e(B) (?
\e$,1'!
\e(B . ?
\e,FQ
\e(B) (?
\e$,1'#
\e(B . ?
\e,FS
\e(B) (?
\e$,1'$
\e(B . ?
\e,FT
\e(B) (?
\e$,1'%
\e(B . ?
\e,FU
\e(B)
128 (?
\e$,1'&
\e(B . ?
\e,FV
\e(B) (?
\e$,1''
\e(B . ?
\e,FW
\e(B) (?
\e$,1'(
\e(B . ?
\e,FX
\e(B) (?
\e$,1')
\e(B . ?
\e,FY
\e(B) (?
\e$,1'*
\e(B . ?
\e,FZ
\e(B)
129 (?
\e$,1'+
\e(B . ?
\e,F[
\e(B) (?
\e$,1',
\e(B . ?
\e,F\
\e(B) (?
\e$,1'-
\e(B . ?
\e,F]
\e(B) (?
\e$,1'.
\e(B . ?
\e,F^
\e(B) (?
\e$,1'/
\e(B . ?
\e,F_
\e(B)
130 (?
\e$,1'0
\e(B . ?
\e,F`
\e(B) (?
\e$,1'1
\e(B . ?
\e,Fa
\e(B) (?
\e$,1'2
\e(B . ?
\e,Fb
\e(B) (?
\e$,1'3
\e(B . ?
\e,Fc
\e(B) (?
\e$,1'4
\e(B . ?
\e,Fd
\e(B)
131 (?
\e$,1'5
\e(B . ?
\e,Fe
\e(B) (?
\e$,1'6
\e(B . ?
\e,Ff
\e(B) (?
\e$,1'7
\e(B . ?
\e,Fg
\e(B) (?
\e$,1'8
\e(B . ?
\e,Fh
\e(B) (?
\e$,1'9
\e(B . ?
\e,Fi
\e(B)
132 (?
\e$,1':
\e(B . ?
\e,Fj
\e(B) (?
\e$,1';
\e(B . ?
\e,Fk
\e(B) (?
\e$,1'<
\e(B . ?
\e,Fl
\e(B) (?
\e$,1'=
\e(B . ?
\e,Fm
\e(B) (?
\e$,1'>
\e(B . ?
\e,Fn
\e(B)
133 (?
\e$,1'?
\e(B . ?
\e,Fo
\e(B) (?
\e$,1'@
\e(B . ?
\e,Fp
\e(B) (?
\e$,1'A
\e(B . ?
\e,Fq
\e(B) (?
\e$,1'B
\e(B . ?
\e,Fr
\e(B) (?
\e$,1'C
\e(B . ?
\e,Fs
\e(B)
134 (?
\e$,1'D
\e(B . ?
\e,Ft
\e(B) (?
\e$,1'E
\e(B . ?
\e,Fu
\e(B) (?
\e$,1'F
\e(B . ?
\e,Fv
\e(B) (?
\e$,1'G
\e(B . ?
\e,Fw
\e(B) (?
\e$,1'H
\e(B . ?
\e,Fx
\e(B)
135 (?
\e$,1'I
\e(B . ?
\e,Fy
\e(B) (?
\e$,1'J
\e(B . ?
\e,Fz
\e(B) (?
\e$,1'K
\e(B . ?
\e,F{
\e(B) (?
\e$,1'L
\e(B . ?
\e,F|
\e(B) (?
\e$,1'M
\e(B . ?
\e,F}
\e(B)
136 (?
\e$,1'N
\e(B . ?
\e,F~
\e(B)
138 (?
\e$,1(!
\e(B . ?
\e,L!
\e(B) (?
\e$,1("
\e(B . ?
\e,L"
\e(B) (?
\e$,1(#
\e(B . ?
\e,L#
\e(B) (?
\e$,1($
\e(B . ?
\e,L$
\e(B)
139 (?
\e$,1(%
\e(B . ?
\e,L%
\e(B) (?
\e$,1(&
\e(B . ?
\e,L&
\e(B) (?
\e$,1('
\e(B . ?
\e,L'
\e(B) (?
\e$,1((
\e(B . ?
\e,L(
\e(B) (?
\e$,1()
\e(B . ?
\e,L)
\e(B)
140 (?
\e$,1(*
\e(B . ?
\e,L*
\e(B) (?
\e$,1(+
\e(B . ?
\e,L+
\e(B) (?
\e$,1(,
\e(B . ?
\e,L,
\e(B) (?
\e$,1(.
\e(B . ?
\e,L.
\e(B) (?
\e$,1(/
\e(B . ?
\e,L/
\e(B)
141 (?
\e$,1(0
\e(B . ?
\e,L0
\e(B) (?
\e$,1(1
\e(B . ?
\e,L1
\e(B) (?
\e$,1(2
\e(B . ?
\e,L2
\e(B) (?
\e$,1(3
\e(B . ?
\e,L3
\e(B) (?
\e$,1(4
\e(B . ?
\e,L4
\e(B)
142 (?
\e$,1(5
\e(B . ?
\e,L5
\e(B) (?
\e$,1(6
\e(B . ?
\e,L6
\e(B) (?
\e$,1(7
\e(B . ?
\e,L7
\e(B) (?
\e$,1(8
\e(B . ?
\e,L8
\e(B) (?
\e$,1(9
\e(B . ?
\e,L9
\e(B)
143 (?
\e$,1(:
\e(B . ?
\e,L:
\e(B) (?
\e$,1(;
\e(B . ?
\e,L;
\e(B) (?
\e$,1(<
\e(B . ?
\e,L<
\e(B) (?
\e$,1(=
\e(B . ?
\e,L=
\e(B) (?
\e$,1(>
\e(B . ?
\e,L>
\e(B)
144 (?
\e$,1(?
\e(B . ?
\e,L?
\e(B) (?
\e$,1(@
\e(B . ?
\e,L@
\e(B) (?
\e$,1(A
\e(B . ?
\e,LA
\e(B) (?
\e$,1(B
\e(B . ?
\e,LB
\e(B) (?
\e$,1(C
\e(B . ?
\e,LC
\e(B)
145 (?
\e$,1(D
\e(B . ?
\e,LD
\e(B) (?
\e$,1(E
\e(B . ?
\e,LE
\e(B) (?
\e$,1(F
\e(B . ?
\e,LF
\e(B) (?
\e$,1(G
\e(B . ?
\e,LG
\e(B) (?
\e$,1(H
\e(B . ?
\e,LH
\e(B)
146 (?
\e$,1(I
\e(B . ?
\e,LI
\e(B) (?
\e$,1(J
\e(B . ?
\e,LJ
\e(B) (?
\e$,1(K
\e(B . ?
\e,LK
\e(B) (?
\e$,1(L
\e(B . ?
\e,LL
\e(B) (?
\e$,1(M
\e(B . ?
\e,LM
\e(B)
147 (?
\e$,1(N
\e(B . ?
\e,LN
\e(B) (?
\e$,1(O
\e(B . ?
\e,LO
\e(B) (?
\e$,1(P
\e(B . ?
\e,LP
\e(B) (?
\e$,1(Q
\e(B . ?
\e,LQ
\e(B) (?
\e$,1(R
\e(B . ?
\e,LR
\e(B)
148 (?
\e$,1(S
\e(B . ?
\e,LS
\e(B) (?
\e$,1(T
\e(B . ?
\e,LT
\e(B) (?
\e$,1(U
\e(B . ?
\e,LU
\e(B) (?
\e$,1(V
\e(B . ?
\e,LV
\e(B) (?
\e$,1(W
\e(B . ?
\e,LW
\e(B)
149 (?
\e$,1(X
\e(B . ?
\e,LX
\e(B) (?
\e$,1(Y
\e(B . ?
\e,LY
\e(B) (?
\e$,1(Z
\e(B . ?
\e,LZ
\e(B) (?
\e$,1([
\e(B . ?
\e,L[
\e(B) (?
\e$,1(\
\e(B . ?
\e,L\
\e(B)
150 (?
\e$,1(]
\e(B . ?
\e,L]
\e(B) (?
\e$,1(^
\e(B . ?
\e,L^
\e(B) (?
\e$,1(_
\e(B . ?
\e,L_
\e(B) (?
\e$,1(`
\e(B . ?
\e,L`
\e(B) (?
\e$,1(a
\e(B . ?
\e,La
\e(B)
151 (?
\e$,1(b
\e(B . ?
\e,Lb
\e(B) (?
\e$,1(c
\e(B . ?
\e,Lc
\e(B) (?
\e$,1(d
\e(B . ?
\e,Ld
\e(B) (?
\e$,1(e
\e(B . ?
\e,Le
\e(B) (?
\e$,1(f
\e(B . ?
\e,Lf
\e(B)
152 (?
\e$,1(g
\e(B . ?
\e,Lg
\e(B) (?
\e$,1(h
\e(B . ?
\e,Lh
\e(B) (?
\e$,1(i
\e(B . ?
\e,Li
\e(B) (?
\e$,1(j
\e(B . ?
\e,Lj
\e(B) (?
\e$,1(k
\e(B . ?
\e,Lk
\e(B)
153 (?
\e$,1(l
\e(B . ?
\e,Ll
\e(B) (?
\e$,1(m
\e(B . ?
\e,Lm
\e(B) (?
\e$,1(n
\e(B . ?
\e,Ln
\e(B) (?
\e$,1(o
\e(B . ?
\e,Lo
\e(B) (?
\e$,1(q
\e(B . ?
\e,Lq
\e(B)
154 (?
\e$,1(r
\e(B . ?
\e,Lr
\e(B) (?
\e$,1(s
\e(B . ?
\e,Ls
\e(B) (?
\e$,1(t
\e(B . ?
\e,Lt
\e(B) (?
\e$,1(u
\e(B . ?
\e,Lu
\e(B) (?
\e$,1(v
\e(B . ?
\e,Lv
\e(B)
155 (?
\e$,1(w
\e(B . ?
\e,Lw
\e(B) (?
\e$,1(x
\e(B . ?
\e,Lx
\e(B) (?
\e$,1(y
\e(B . ?
\e,Ly
\e(B) (?
\e$,1(z
\e(B . ?
\e,Lz
\e(B) (?
\e$,1({
\e(B . ?
\e,L{
\e(B)
156 (?
\e$,1(|
\e(B . ?
\e,L|
\e(B) (?
\e$,1(~
\e(B . ?
\e,L~
\e(B) (?
\e$,1(
\7f\e(B . ?
\e,L
\7f\e(B)))
159 (defcustom utf-fragment-on-decoding nil
160 "Whether or not to decode some chars in UTF-8/16 text into iso8859 charsets.
161 Setting this means that the relevant Cyrillic and Greek characters are
162 decoded into the iso8859 charsets rather than into
163 mule-unicode-0100-24ff. The iso8859 charsets take half as much space
164 in the buffer, but using them may affect how the buffer can be re-encoded
165 and may require a different input method to search for them, for instance.
166 See `unify-8859-on-decoding-mode' and `unify-8859-on-encoding-mode'
167 for mechanisms to make this largely transparent.
169 Setting this variable outside customize has no effect."
173 (define-translation-table 'utf-translation-table-for-decode
174 utf-fragmentation-table)
175 ;; Even if unify-8859-on-encoding-mode is off, make
176 ;; mule-utf-* encode characters in
177 ;; utf-fragmentation-table.
178 (unless (eq (get 'utf-translation-table-for-encode
180 ucs-mule-to-mule-unicode)
181 (define-translation-table 'utf-translation-table-for-encode
182 utf-defragmentation-table)
183 (dolist (coding '(mule-utf-8 mule-utf-16-be mule-utf-16-le))
184 (register-char-codings coding utf-defragmentation-table))))
185 (define-translation-table 'utf-translation-table-for-decode)
186 ;; When unify-8859-on-encoding-mode is off, be sure to make
187 ;; mule-utf-* disabled for characters in
188 ;; utf-fragmentation-table.
189 (unless (eq (get 'utf-translation-table-for-encode
191 ucs-mule-to-mule-unicode)
192 (define-translation-table 'utf-translation-table-for-encode)
195 (if (and (>= key 128) val)
196 (aset char-coding-system-table key
198 (delq 'mule-utf-16-le
199 (delq 'mule-utf-16-be
200 (aref char-coding-system-table
202 utf-defragmentation-table)))
208 (defcustom utf-translate-cjk nil
209 "Whether the UTF based coding systems should decode/encode CJK characters.
210 Enabling this loads tables which allow the coding systems mule-utf-8,
211 mule-utf-16-le and mule-utf-16-be to encode characters in the charsets
212 `korean-ksc5601', `chinese-gb2312', `chinese-big5-1',
213 `chinese-big5-2', `japanese-jisx0208' and `japanese-jisx0212', and to
214 decode the corresponding unicodes into such characters.
216 Where the charsets overlap, the one preferred for decoding is chosen
217 according to the language environment in effect when this option is
218 turned on: ksc5601 for Korean, gb2312 for Chinese-GB, big5 for
219 Chinese-Big5 and jisx for other environments.
221 The tables are large (over 40000 entries), so this option is not the
222 default. Also, installing them may be rather slow."
226 ;; Load the files explicitly, to avoid having to keep
227 ;; around the large tables they contain (as well as the
228 ;; ones which get built).
230 ((string= "Korean" current-language-environment)
233 (load "subst-gb2312")
235 ((string= "Chinese-BIG5" current-language-environment)
238 (load "subst-gb2312")
240 ((string= "Chinese-GB" current-language-environment)
244 (load "subst-gb2312"))
247 (load "subst-gb2312")
249 (load "subst-jis"))) ; jis covers as much as big5, gb2312
250 (let ((table (make-char-table 'translation-table)))
251 (maphash (lambda (k v)
253 ucs-mule-cjk-to-unicode)
254 (register-char-codings 'mule-utf-8 table)
255 (register-char-codings 'mule-utf-16-le table)
256 (register-char-codings 'mule-utf-16-be table))
257 (define-translation-hash-table 'utf-subst-table-for-decode
258 ucs-unicode-to-mule-cjk)
259 (define-translation-hash-table 'utf-subst-table-for-encode
260 ucs-mule-cjk-to-unicode))
263 (if (gethash k ucs-mule-cjk-to-unicode)
264 (aset char-coding-system-table k
266 (delq 'mule-utf-16-le
267 (delq 'mule-utf-16-be v))))))
268 char-coding-system-table)
269 (define-translation-hash-table 'utf-subst-table-for-decode
270 (make-hash-table :test 'eq))
271 (define-translation-hash-table 'utf-subst-table-for-encode
272 (make-hash-table :test 'eq)))
276 :set-after '(current-language-environment)
279 (define-ccl-program ccl-decode-mule-utf-8
281 ;; charset | bytes in utf-8 | bytes in emacs
282 ;; -----------------------+----------------+---------------
284 ;; -----------------------+----------------+---------------
285 ;; eight-bit-control | 2 | 2
286 ;; eight-bit-graphic | 2 | 1
287 ;; latin-iso8859-1 | 2 | 2
288 ;; -----------------------+----------------+---------------
289 ;; mule-unicode-0100-24ff | 2 | 4
291 ;; -----------------------+----------------+---------------
292 ;; mule-unicode-0100-24ff | 3 | 4
294 ;; mule-unicode-2500-33ff | 3 | 4
295 ;; mule-unicode-e000-ffff | 3 | 4
297 ;; Thus magnification factor is two.
300 ((r5 = ,(charset-id 'eight-bit-control))
301 (r6 = ,(charset-id 'eight-bit-graphic))
305 ;; 1byte encoding, i.e., ascii
308 (if (r0 < #xc0) ; continuation byte (invalid here)
310 (write-multibyte-character r5 r0)
311 (write-multibyte-character r6 r0))
312 ;; 2 byte encoding 00000yyyyyxxxxxx = 110yyyyy 10xxxxxx
316 (if ((r1 & #b11000000) != #b10000000)
317 ;; Invalid 2-byte sequence
319 (write-multibyte-character r5 r0)
320 (write-multibyte-character r6 r0))
324 (write-multibyte-character r5 r1)
325 (write-multibyte-character r6 r1))))
327 ((r3 = r0) ; save in case of overlong sequence
333 ;; Now r1 holds scalar value
335 (if (r1 < 128) ; `overlong sequence'
337 (write-multibyte-character r5 r3)
338 (write-multibyte-character r6 r3))
342 (write-multibyte-character r5 r2)
343 (write-multibyte-character r6 r2))))
347 ((write-multibyte-character r5 r1))
351 ((r0 = ,(charset-id 'latin-iso8859-1))
353 (write-multibyte-character r0 r1))
355 ;; mule-unicode-0100-24ff (< 0800)
356 ((r0 = ,(charset-id 'mule-unicode-0100-24ff))
358 (r2 = (((r1 / 96) + 32) << 7))
362 utf-translation-table-for-decode r0 r1)
363 (write-multibyte-character r0 r1))))))))
366 ;; zzzzyyyyyyxxxxxx = 1110zzzz 10yyyyyy 10xxxxxx
370 ;; This is set to 1 if the encoding is invalid.
373 (r3 = (r1 & #b11000000))
374 (r3 |= ((r2 >> 2) & #b00110000))
375 (if (r3 != #b10100000)
377 ((r3 = ((r0 & #x0f) << 12))
378 (r3 += ((r1 & #x3f) << 6))
384 ;; Invalid 3-byte sequence
386 (write-multibyte-character r5 r0)
387 (write-multibyte-character r6 r0))
391 (write-multibyte-character r5 r1)
392 (write-multibyte-character r6 r1)))
396 (write-multibyte-character r5 r2)
397 (write-multibyte-character r6 r2))))
399 ;; mule-unicode-0100-24ff (>= 0800)
401 ((r0 = ,(charset-id 'mule-unicode-0100-24ff))
405 (r1 += ((r3 + 32) << 7))
407 utf-translation-table-for-decode r0 r1)
408 (write-multibyte-character r0 r1))
410 ;; mule-unicode-2500-33ff
412 ((r4 = r3) ; don't zap r3
413 (lookup-integer utf-subst-table-for-decode r4 r5)
416 ((write-multibyte-character r4 r5)
417 ;; Zapped through register starvation.
418 (r5 = ,(charset-id 'eight-bit-control)))
419 ((r0 = ,(charset-id 'mule-unicode-2500-33ff))
423 (r1 += ((r3 + 32) << 7))
424 (write-multibyte-character r0 r1))))
427 ;; Try to convert to CJK chars, else keep
428 ;; them as eight-bit-{control|graphic}.
430 ((r4 = r3) ; don't zap r3
431 (lookup-integer utf-subst-table-for-decode r4 r5)
434 ((write-multibyte-character r4 r5)
435 ;; Zapped through register starvation.
436 (r5 = ,(charset-id 'eight-bit-control)))
437 ;; #xe0 <= r0 < #xf0, so r0 is eight-bit-graphic
439 (write-multibyte-character r3 r0)
442 (write-multibyte-character r3 r1)
446 (write-multibyte-character r3 r2))))
448 ;; Surrogates, U+D800 .. U+DFFF
451 (write-multibyte-character r3 r0) ; eight-bit-graphic
454 (write-multibyte-character r3 r1)
458 (write-multibyte-character r3 r2))
460 ;; mule-unicode-e000-ffff
461 ;; Fixme: fffe and ffff are invalid.
462 ((r0 = ,(charset-id 'mule-unicode-e000-ffff))
466 (r1 += ((r3 + 32) << 7))
467 (write-multibyte-character r0 r1)))))))))
471 ;; keep those bytes as eight-bit-{control|graphic}
472 ;; Fixme: allow lookup in utf-subst-table-for-decode.
474 ;; r0 > #xf0, thus eight-bit-graphic
475 (write-multibyte-character r6 r0)
477 (if (r1 < #x80) ; invalid byte
479 (write-multibyte-character r5 r1))
480 (write-multibyte-character r6 r1))
482 (if (r2 < #x80) ; invalid byte
484 (write-multibyte-character r5 r2))
485 (write-multibyte-character r6 r2))
487 (if (r3 < #x80) ; invalid byte
489 (write-multibyte-character r5 r3))
490 (write-multibyte-character r6 r3))
491 (if (r0 >= #xf8) ; 5- or 6-byte encoding
494 (if (r1 < #x80) ; invalid byte
496 (write-multibyte-character r5 r1))
497 (write-multibyte-character r6 r1))
498 (if (r0 >= #xfc) ; 6-byte
501 (if (r1 < #x80) ; invalid byte
503 (write-multibyte-character r5 r1))
504 (write-multibyte-character r6 r1)))))))
505 ;; else invalid byte >= #xfe
506 (write-multibyte-character r6 r0))))))
509 "CCL program to decode UTF-8.
510 Basic decoding is done into the charsets ascii, latin-iso8859-1 and
511 mule-unicode-*, but see also `utf-fragmentation-table' and
512 `ucs-mule-cjk-to-unicode'.
513 Encodings of un-representable Unicode characters are decoded asis into
514 eight-bit-control and eight-bit-graphic characters.")
516 (define-ccl-program ccl-encode-mule-utf-8
522 (read-multibyte-character r0 r1)
523 (translate-character utf-translation-table-for-encode r0 r1))
524 (;; We have already done read-multibyte-character.
529 (if (r0 == ,(charset-id 'ascii))
532 (if (r0 == ,(charset-id 'latin-iso8859-1))
534 ;; 0000 0yyy yyxx xxxx 110y yyyy 10xx xxxx
535 ;; 20 0000 0000 1010 0000 1100 0010 1010 0000
536 ;; 7f 0000 0000 1111 1111 1100 0011 1011 1111
537 ((r0 = (((r1 & #x40) >> 6) | #xc2))
542 (if (r0 == ,(charset-id 'mule-unicode-0100-24ff))
543 ((r0 = ((((r1 & #x3f80) >> 7) - 32) * 96))
544 ;; #x3f80 == (0011 1111 1000 0000)b
546 (r1 += (r0 + 224)) ; 240 == -32 + #x0100
547 ;; now r1 holds scalar value
550 ((r0 = (((r1 & #x07c0) >> 6) | #xc0))
551 ;; #x07c0 == (0000 0111 1100 0000)b
556 ((r0 = (((r1 & #xf000) >> 12) | #xe0))
557 (r2 = ((r1 & #x3f) | #x80))
563 (if (r0 == ,(charset-id 'mule-unicode-2500-33ff))
564 ((r0 = ((((r1 & #x3f80) >> 7) - 32) * 96))
566 (r1 += (r0 + 9440)) ; 9440 == -32 + #x2500
567 (r0 = (((r1 & #xf000) >> 12) | #xe0))
568 (r2 = ((r1 & #x3f) | #x80))
574 (if (r0 == ,(charset-id 'mule-unicode-e000-ffff))
575 ((r0 = ((((r1 & #x3f80) >> 7) - 32) * 96))
577 (r1 += (r0 + 57312)) ; 57312 == -32 + #xe000
578 (r0 = (((r1 & #xf000) >> 12) | #xe0))
579 (r2 = ((r1 & #x3f) | #x80))
585 (if (r0 == ,(charset-id 'eight-bit-control))
587 ;; 0000 0yyy yyxx xxxx 110y yyyy 10xx xxxx
588 ;; 80 0000 0000 1000 0000 1100 0010 1000 0000
589 ;; 9f 0000 0000 1001 1111 1100 0010 1001 1111
593 (if (r0 == ,(charset-id 'eight-bit-graphic))
595 ;; 0000 0yyy yyxx xxxx 110y yyyy 10xx xxxx
596 ;; a0 0000 0000 1010 0000 1100 0010 1010 0000
597 ;; ff 0000 0000 1111 1111 1101 1111 1011 1111
600 (read-multibyte-character r0 r1)
601 (if (r0 != ,(charset-id 'eight-bit-graphic))
602 (if (r0 != ,(charset-id 'eight-bit-control))
606 ((read-multibyte-character r0 r2)
607 (if (r0 != ,(charset-id 'eight-bit-graphic))
608 (if (r0 != ,(charset-id 'eight-bit-control))
618 ((lookup-character utf-subst-table-for-encode r0 r1)
619 (if r7 ; lookup succeeded
620 ((r1 = (((r0 & #xf000) >> 12) | #xe0))
621 (r2 = ((r0 & #x3f) | #x80))
626 ;; Unsupported character.
627 ;; Output U+FFFD, which is `ef bf bd' in UTF-8.
630 (write #xbd)))))))))))
638 "CCL program to encode into UTF-8.")
641 (define-ccl-program ccl-untranslated-to-ucs
643 (if (r0 < #xf0) ; 3-byte encoding, as above
645 (r3 = (r1 & #b11000000))
646 (r3 |= ((r2 >> 2) & #b00110000))
647 (if (r3 != #b10100000)
649 ((r3 = ((r0 & #x0f) << 12))
650 (r3 += ((r1 & #x3f) << 6))
657 (if (r0 < #xf8) ; 4-byte (Mule-UCS recipe)
667 ((r1 = ((r1 & #x3F) << 12))
668 (r2 = ((r2 & #x3F) << 6))
670 (r0 = (((((r0 & #x07) << 18) | r1) | r2) | r3)))))))))
672 "Decode 3- or 4-byte sequences in r0, r1, r2 [,r3] to unicodes in r0.
673 r0 == 0 for invalid sequence.")
675 (defvar utf-8-ccl-regs (make-vector 8 0))
677 (defsubst utf-8-untranslated-to-ucs ()
678 "Return the UCS code for an untranslated sequence of raw bytes t point.
679 Only for 3- or 4-byte sequences."
680 (aset utf-8-ccl-regs 0 (or (char-after) 0))
681 (aset utf-8-ccl-regs 1 (or (char-after (1+ (point))) 0))
682 (aset utf-8-ccl-regs 2 (or (char-after (+ 2 (point))) 0))
683 (aset utf-8-ccl-regs 3 (or (char-after (+ 3 (point))) 0))
684 (ccl-execute 'ccl-untranslated-to-ucs utf-8-ccl-regs)
685 (aref utf-8-ccl-regs 0))
687 (defun utf-8-help-echo (window object position)
688 (format "Untranslated Unicode U+%04X"
689 (get-char-property position 'untranslated-utf-8 object)))
691 ;; We compose the untranslatable sequences into a single character.
692 ;; This is infelicitous for editing, because there's currently no
693 ;; mechanism for treating compositions as atomic, but is OK for
694 ;; display. They are composed to U+FFFD with help-echo which
695 ;; indicates the unicodes they represent. This function GCs too much.
696 (defsubst utf-8-compose ()
697 "Put a suitable composition on an untranslatable sequence.
698 Return the sequence's length."
699 (let* ((u (utf-8-untranslated-to-ucs))
705 (put-text-property (point) (min (point-max) (+ l (point)))
706 'untranslated-utf-8 u)
707 (put-text-property (point) (min (point-max) (+ l (point)))
708 'help-echo 'utf-8-help-echo)
709 (compose-region (point) (+ l (point)) ?
\e$,3u=
\e(B)
712 (defcustom utf-8-compose-scripts nil
713 "*Non-nil means compose various scripts on decoding utf-8 text."
718 (defun utf-8-post-read-conversion (length)
719 "Compose untranslated utf-8 sequences into single characters.
720 Also compose particular scripts if `utf-8-compose-scripts' is non-nil."
722 ;; Can't do eval-when-compile to insert a multibyte constant
723 ;; version of the string in the loop, since it's always loaded as
724 ;; unibyte from a byte-compiled file.
725 (let ((range (string-as-multibyte "^\xe1-\xf7")))
726 (while (and (skip-chars-forward range)
728 (forward-char (utf-8-compose)))))
729 ;; Fixme: Takahashi-san implies it may not work this easily. I
730 ;; asked why but didn't get a reply. -- fx
731 (when (and utf-8-compose-scripts (> length 1))
732 ;; These currently have definitions which cover the relevant
733 ;; unicodes. We could avoid loading thai-util &c by checking
734 ;; whether the region contains any characters with the appropriate
735 ;; categories. There aren't yet Unicode-based rules for Tibetan.
736 (save-excursion (setq length (diacritic-post-read-conversion length)))
737 (save-excursion (setq length (thai-post-read-conversion length)))
738 (save-excursion (setq length (lao-post-read-conversion length)))
740 (setq length (in-is13194-devanagari-post-read-conversion length))))
743 ;; ucs-tables is preloaded
744 ;; (defun utf-8-pre-write-conversion (beg end)
745 ;; "Semi-dummy pre-write function effectively to autoload ucs-tables."
746 ;; ;; Ensure translation-table is loaded.
747 ;; (require 'ucs-tables)
748 ;; ;; Don't do this again.
749 ;; (coding-system-put 'mule-utf-8 'pre-write-conversion nil)
754 "UTF-8 encoding for Emacs-supported Unicode characters.
755 It supports Unicode characters of these ranges:
756 U+0000..U+33FF, U+E000..U+FFFF.
757 They correspond to these Emacs character sets:
758 ascii, latin-iso8859-1, mule-unicode-0100-24ff,
759 mule-unicode-2500-33ff, mule-unicode-e000-ffff
761 On decoding (e.g. reading a file), Unicode characters not in the above
762 ranges are decoded into sequences of eight-bit-control and
763 eight-bit-graphic characters to preserve their byte sequences. The
764 byte sequence is preserved on i/o for valid utf-8, but not necessarily
767 On encoding (e.g. writing a file), Emacs characters not belonging to
768 any of the character sets listed above are encoded into the UTF-8 byte
769 sequence representing U+FFFD (REPLACEMENT CHARACTER)."
771 '(ccl-decode-mule-utf-8 . ccl-encode-mule-utf-8)
777 mule-unicode-0100-24ff
778 mule-unicode-2500-33ff
779 mule-unicode-e000-ffff)
780 (mime-charset . utf-8)
781 (coding-category . coding-category-utf-8)
782 (valid-codes (0 . 255))
783 ;; (pre-write-conversion . utf-8-pre-write-conversion)
784 (post-read-conversion . utf-8-post-read-conversion)
785 (dependency unify-8859-on-encoding-mode
786 unify-8859-on-decoding-mode
787 utf-fragment-on-decoding
790 (define-coding-system-alias 'utf-8 'mule-utf-8)
792 ;; I think this needs special private charsets defined for the
793 ;; untranslated sequences, if it's going to work well.
795 ;;; (defun utf-8-compose-function (pos to pattern &optional string)
796 ;;; (let* ((prop (get-char-property pos 'composition string))
797 ;;; (l (and prop (- (cadr prop) (car prop)))))
798 ;;; (cond ((and l (> l (- to pos)))
799 ;;; (delete-region pos to))
800 ;;; ((and (> (char-after pos) 224)
801 ;;; (< (char-after pos) 256)
802 ;;; (save-restriction
803 ;;; (narrow-to-region pos to)
804 ;;; (utf-8-compose)))
808 ;;; (aset composition-function-table
810 ;;; `((,(string-as-multibyte "[\200-\237\240-\377]")
811 ;;; . utf-8-compose-function))))
813 ;;; utf-8.el ends here