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'. Hash tables
52 ;; `utf-subst-table-for-decode' and `utf-subst-table-for-encode' are
53 ;; used to support encoding and decoding of about a quarter of the CJK
54 ;; space between U+3400 and U+DFFF.
56 ;; UTF-8 is defined in RFC 3629. A sketch of the encoding is:
59 ;; value | 1st byte | 2nd byte | 3rd byte
60 ;; --------------------+-----------+-----------+----------
61 ;; 0000 0000 0xxx xxxx | 0xxx xxxx | |
62 ;; 0000 0yyy yyxx xxxx | 110y yyyy | 10xx xxxx |
63 ;; zzzz yyyy yyxx xxxx | 1110 zzzz | 10yy yyyy | 10xx xxxx
67 (defvar ucs-mule-to-mule-unicode
(make-char-table 'translation-table nil
)
68 "Char table mapping characters to latin-iso8859-1 or mule-unicode-*.
70 If `unify-8859-on-encoding-mode' is non-nil, this table populates the
71 translation-table named `utf-translation-table-for-encode'.")
73 (define-translation-table 'utf-translation-table-for-encode
)
76 ;; Map Cyrillic and Greek to iso-8859 charsets, which take half the
77 ;; space of mule-unicode. For Latin scripts this isn't very
78 ;; important. Hebrew and Arabic might go here too when there's proper
81 (defvar utf-fragmentation-table
(make-char-table 'translation-table nil
)
82 "Char-table normally mapping non-Latin mule-unicode-* chars to iso-8859-*.
84 If `utf-fragment-on-decoding' is non-nil, this table populates the
85 translation-table named `utf-translation-table-for-decode'")
87 (defvar utf-defragmentation-table
(make-char-table 'translation-table nil
)
88 "Char-table for reverse mapping of `utf-fragmentation-table'.
90 If `utf-fragment-on-decoding' is non-nil and
91 `unify-8859-on-encoding-mode' is nil, this table populates the
92 translation-table named `utf-translation-table-for-encode'")
94 (define-translation-table 'utf-translation-table-for-decode
)
97 (defvar ucs-mule-cjk-to-unicode
(make-hash-table :test
'eq
)
98 "Hash table mapping Emacs CJK character sets to Unicode code points.
100 If `utf-translate-cjk' is non-nil, this table populates the
101 translation-hash-table named `utf-subst-table-for-encode'.")
103 (define-translation-hash-table 'utf-subst-table-for-encode
104 ucs-mule-cjk-to-unicode
)
106 (defvar ucs-unicode-to-mule-cjk
(make-hash-table :test
'eq
)
107 "Hash table mapping Unicode code points to Emacs CJK character sets.
109 If `utf-translate-cjk' is non-nil, this table populates the
110 translation-hash-table named `utf-subst-table-for-decode'.")
112 (define-translation-hash-table 'utf-subst-table-for-decode
113 ucs-unicode-to-mule-cjk
)
117 (aset utf-fragmentation-table
(car pair
) (cdr pair
))
118 (aset utf-defragmentation-table
(cdr pair
) (car pair
)))
119 '((?
\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)
120 (?
\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)
121 (?
\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)
122 (?
\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)
123 (?
\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)
124 (?
\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)
125 (?
\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)
126 (?
\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)
127 (?
\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)
128 (?
\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)
129 (?
\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$,1'>\e(B . ?\e,Fn\e(B)
130 (?
\e$
,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)
131 (?
\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)
132 (?
\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)
133 (?
\e$
,1'N
\e(B . ?
\e,F~
\e(B)
135 (?
\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)
136 (?
\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)
137 (?
\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)
138 (?
\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)
139 (?
\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)
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(?
\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)
142 (?
\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)
143 (?
\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)
144 (?
\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)
145 (?
\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)
146 (?
\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)
147 (?
\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)
148 (?
\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)
149 (?
\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)
150 (?
\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)
151 (?
\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)
152 (?
\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)
153 (?
\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)))
156 (defcustom utf-fragment-on-decoding nil
157 "Whether or not to decode some chars in UTF-8/16 text into iso8859 charsets.
158 Setting this means that the relevant Cyrillic and Greek characters are
159 decoded into the iso8859 charsets rather than into
160 mule-unicode-0100-24ff. The iso8859 charsets take half as much space
161 in the buffer, but using them may affect how the buffer can be re-encoded
162 and may require a different input method to search for them, for instance.
163 See `unify-8859-on-decoding-mode' and `unify-8859-on-encoding-mode'
164 for mechanisms to make this largely transparent.
166 Setting this variable outside customize has no effect."
170 (define-translation-table 'utf-translation-table-for-decode
171 utf-fragmentation-table
)
172 ;; Even if unify-8859-on-encoding-mode is off, make
173 ;; mule-utf-* encode characters in
174 ;; utf-fragmentation-table.
175 (unless (eq (get 'utf-translation-table-for-encode
177 ucs-mule-to-mule-unicode
)
178 (define-translation-table 'utf-translation-table-for-encode
179 utf-defragmentation-table
)))
180 (define-translation-table 'utf-translation-table-for-decode
)
181 ;; When unify-8859-on-encoding-mode is off, be sure to make
182 ;; mule-utf-* disabled for characters in
183 ;; utf-fragmentation-table.
184 (unless (eq (get 'utf-translation-table-for-encode
186 ucs-mule-to-mule-unicode
)
187 (define-translation-table 'utf-translation-table-for-encode
)))
193 (define-minor-mode utf-translate-cjk-mode
194 "Whether the UTF based coding systems should decode/encode CJK characters.
195 Enabling this loads tables which allow the coding systems mule-utf-8,
196 mule-utf-16le and mule-utf-16be to encode characters in the charsets
197 `korean-ksc5601', `chinese-gb2312', `chinese-big5-1',
198 `chinese-big5-2', `japanese-jisx0208' and `japanese-jisx0212', and to
199 decode the corresponding unicodes into such characters.
201 Where the charsets overlap, the one preferred for decoding is chosen
202 according to the language environment in effect when this option is
203 turned on: ksc5601 for Korean, gb2312 for Chinese-GB, big5 for
204 Chinese-Big5 and jisx for other environments.
206 The tables are large (over 40000 entries), so this option is not the
207 default. Also, installing them may be rather slow."
211 :set-after
'(current-language-environment)
214 (if utf-translate-cjk-mode
215 ;; Fixme: Allow the use of the CJK charsets to be
216 ;; customized by reordering and possible omission.
218 ;; Redefine them with realistic initial sizes and a
219 ;; smallish rehash size to avoid wasting significant
220 ;; space after they're built.
221 (setq ucs-mule-cjk-to-unicode
222 (make-hash-table :test
'eq
:size
43000 :rehash-size
1000)
223 ucs-unicode-to-mule-cjk
224 (make-hash-table :test
'eq
:size
21500 :rehash-size
1000))
225 ;; Load the files explicitly, to avoid having to keep
226 ;; around the large tables they contain (as well as the
227 ;; ones which get built).
229 ((string= "Korean" current-language-environment
)
232 (load "subst-gb2312")
234 ((string= "Chinese-BIG5" current-language-environment
)
237 (load "subst-gb2312")
239 ((string= "Chinese-GB" current-language-environment
)
243 (load "subst-gb2312"))
246 (load "subst-gb2312")
248 (load "subst-jis"))) ; jis covers as much as big5, gb2312
249 (define-translation-hash-table 'utf-subst-table-for-decode
250 ucs-unicode-to-mule-cjk
)
251 (define-translation-hash-table 'utf-subst-table-for-encode
252 ucs-mule-cjk-to-unicode
)
253 (set-char-table-extra-slot (get 'utf-translation-table-for-encode
255 1 ucs-mule-cjk-to-unicode
))
256 (define-translation-hash-table 'utf-subst-table-for-decode
257 (make-hash-table :test
'eq
))
258 (define-translation-hash-table 'utf-subst-table-for-encode
259 (make-hash-table :test
'eq
))
260 (set-char-table-extra-slot (get 'utf-translation-table-for-encode
264 (define-ccl-program ccl-decode-mule-utf-8
266 ;; charset | bytes in utf-8 | bytes in emacs
267 ;; -----------------------+----------------+---------------
269 ;; -----------------------+----------------+---------------
270 ;; eight-bit-control | 2 | 2
271 ;; eight-bit-graphic | 2 | 1
272 ;; latin-iso8859-1 | 2 | 2
273 ;; -----------------------+----------------+---------------
274 ;; mule-unicode-0100-24ff | 2 | 4
276 ;; -----------------------+----------------+---------------
277 ;; mule-unicode-0100-24ff | 3 | 4
279 ;; mule-unicode-2500-33ff | 3 | 4
280 ;; mule-unicode-e000-ffff | 3 | 4
282 ;; Thus magnification factor is two.
285 ((r5 = ,(charset-id 'eight-bit-control
))
286 (r6 = ,(charset-id 'eight-bit-graphic
))
291 ;; 1byte encoding, i.e., ascii
294 (if (r0 < #xc0
) ; continuation byte (invalid here)
296 (write-multibyte-character r5 r0
)
297 (write-multibyte-character r6 r0
)))
298 ;; 2 byte encoding 00000yyyyyxxxxxx = 110yyyyy 10xxxxxx
303 (if ((r1 & #b11000000
) != #b10000000
)
304 ;; Invalid 2-byte sequence
306 (write-multibyte-character r5 r0
)
307 (write-multibyte-character r6 r0
))
311 (write-multibyte-character r5 r1
)
312 (write-multibyte-character r6 r1
))))
314 ((r3 = r0
) ; save in case of overlong sequence
320 ;; Now r1 holds scalar value
322 (if (r1 < 128) ; `overlong sequence'
324 (write-multibyte-character r5 r3
)
325 (write-multibyte-character r6 r3
))
329 (write-multibyte-character r5 r2
)
330 (write-multibyte-character r6 r2
))))
334 ((write-multibyte-character r5 r1
))
338 ((r0 = ,(charset-id 'latin-iso8859-1
))
340 (write-multibyte-character r0 r1
))
342 ;; mule-unicode-0100-24ff (< 0800)
343 ((r0 = ,(charset-id 'mule-unicode-0100-24ff
))
345 (r2 = (((r1 / 96) + 32) << 7))
349 utf-translation-table-for-decode r0 r1
)
350 (write-multibyte-character r0 r1
))))))))
353 ;; zzzzyyyyyyxxxxxx = 1110zzzz 10yyyyyy 10xxxxxx
359 ;; This is set to 1 if the encoding is invalid.
362 (r3 = (r1 & #b11000000
))
363 (r3 |
= ((r2 >> 2) & #b00110000
))
364 (if (r3 != #b10100000
)
366 ((r3 = ((r0 & #x0f
) << 12))
367 (r3 += ((r1 & #x3f
) << 6))
373 ;; Invalid 3-byte sequence
375 (write-multibyte-character r5 r0
)
376 (write-multibyte-character r6 r0
))
380 (write-multibyte-character r5 r1
)
381 (write-multibyte-character r6 r1
)))
385 (write-multibyte-character r5 r2
)
386 (write-multibyte-character r6 r2
))))
388 ;; mule-unicode-0100-24ff (>= 0800)
390 ((r0 = ,(charset-id 'mule-unicode-0100-24ff
))
394 (r1 += ((r3 + 32) << 7))
396 utf-translation-table-for-decode r0 r1
)
397 (write-multibyte-character r0 r1
))
399 ;; mule-unicode-2500-33ff
401 ((r4 = r3
) ; don't zap r3
402 (lookup-integer utf-subst-table-for-decode r4 r5
)
405 ((write-multibyte-character r4 r5
)
406 ;; Zapped through register starvation.
407 (r5 = ,(charset-id 'eight-bit-control
)))
408 ((r0 = ,(charset-id 'mule-unicode-2500-33ff
))
412 (r1 += ((r3 + 32) << 7))
413 (write-multibyte-character r0 r1
))))
416 ;; Try to convert to CJK chars, else keep
417 ;; them as eight-bit-{control|graphic}.
419 ((r4 = r3
) ; don't zap r3
420 (lookup-integer utf-subst-table-for-decode r4 r5
)
423 ((write-multibyte-character r4 r5
)
424 ;; Zapped through register starvation.
425 (r5 = ,(charset-id 'eight-bit-control
)))
426 ;; #xe0 <= r0 < #xf0, so r0 is eight-bit-graphic
428 (write-multibyte-character r3 r0
)
431 (write-multibyte-character r3 r1
)
435 (write-multibyte-character r3 r2
))))
437 ;; Surrogates, U+D800 .. U+DFFF
440 (write-multibyte-character r3 r0
) ; eight-bit-graphic
443 (write-multibyte-character r3 r1
)
447 (write-multibyte-character r3 r2
))
449 ;; mule-unicode-e000-ffff
450 ;; Fixme: fffe and ffff are invalid.
451 ((r4 = r3
) ; don't zap r3
452 (lookup-integer utf-subst-table-for-decode r4 r5
)
455 ((write-multibyte-character r4 r5
)
456 ;; Zapped through register starvation.
457 (r5 = ,(charset-id 'eight-bit-control
)))
458 ((r0 = ,(charset-id 'mule-unicode-e000-ffff
))
462 (r1 += ((r3 + 32) << 7))
463 (write-multibyte-character r0 r1
)))))))))))
467 ;; keep those bytes as eight-bit-{control|graphic}
468 ;; Fixme: allow lookup in utf-subst-table-for-decode.
473 ;; r0 > #xf0, thus eight-bit-graphic
474 (write-multibyte-character r6 r0
)
476 (if (r1 < #x80
) ; invalid byte
478 (write-multibyte-character r5 r1
))
479 (write-multibyte-character r6 r1
))
481 (if (r2 < #x80
) ; invalid byte
483 (write-multibyte-character r5 r2
))
484 (write-multibyte-character r6 r2
))
486 (if (r3 < #x80
) ; invalid byte
488 (write-multibyte-character r5 r3
))
489 (write-multibyte-character r6 r3
))
490 (if (r0 >= #xf8
) ; 5- or 6-byte encoding
494 (if (r0 < #x80
) ; invalid byte
496 (write-multibyte-character r5 r0
))
497 (write-multibyte-character r6 r0
))
498 (if (r0 >= #xfc
) ; 6-byte
502 (if (r0 < #x80
) ; invalid byte
504 (write-multibyte-character r5 r0
))
505 (write-multibyte-character r6 r0
)))))))
506 ;; else invalid byte >= #xfe
507 (write-multibyte-character r6 r0
))))))
515 (write-multibyte-character r5 r0
)
516 ((write-multibyte-character r6 r0
))))
521 (write-multibyte-character r5 r1
)
522 ((write-multibyte-character r6 r1
))))
527 (write-multibyte-character r5 r2
)
528 ((write-multibyte-character r6 r2
))))
533 (write-multibyte-character r5 r3
)
534 ((write-multibyte-character r6 r3
))))))))))))
536 "CCL program to decode UTF-8.
537 Basic decoding is done into the charsets ascii, latin-iso8859-1 and
538 mule-unicode-*, but see also `utf-fragmentation-table' and
539 `ucs-mule-cjk-to-unicode'.
540 Encodings of un-representable Unicode characters are decoded asis into
541 eight-bit-control and eight-bit-graphic characters.")
543 (define-ccl-program ccl-encode-mule-utf-8
549 (read-multibyte-character r0 r1
)
550 (translate-character utf-translation-table-for-encode r0 r1
))
551 (;; We have already done read-multibyte-character.
556 (if (r0 == ,(charset-id 'ascii
))
559 (if (r0 == ,(charset-id 'latin-iso8859-1
))
561 ;; 0000 0yyy yyxx xxxx 110y yyyy 10xx xxxx
562 ;; 20 0000 0000 1010 0000 1100 0010 1010 0000
563 ;; 7f 0000 0000 1111 1111 1100 0011 1011 1111
564 ((r0 = (((r1 & #x40
) >> 6) |
#xc2
))
569 (if (r0 == ,(charset-id 'mule-unicode-0100-24ff
))
570 ((r0 = ((((r1 & #x3f80
) >> 7) -
32) * 96))
571 ;; #x3f80 == (0011 1111 1000 0000)b
573 (r1 += (r0 + 224)) ; 240 == -32 + #x0100
574 ;; now r1 holds scalar value
577 ((r0 = (((r1 & #x07c0
) >> 6) |
#xc0
))
578 ;; #x07c0 == (0000 0111 1100 0000)b
583 ((r0 = (((r1 & #xf000
) >> 12) |
#xe0
))
584 (r2 = ((r1 & #x3f
) |
#x80
))
590 (if (r0 == ,(charset-id 'mule-unicode-2500-33ff
))
591 ((r0 = ((((r1 & #x3f80
) >> 7) -
32) * 96))
593 (r1 += (r0 + 9440)) ; 9440 == -32 + #x2500
594 (r0 = (((r1 & #xf000
) >> 12) |
#xe0
))
595 (r2 = ((r1 & #x3f
) |
#x80
))
601 (if (r0 == ,(charset-id 'mule-unicode-e000-ffff
))
602 ((r0 = ((((r1 & #x3f80
) >> 7) -
32) * 96))
604 (r1 += (r0 + 57312)) ; 57312 == -32 + #xe000
605 (r0 = (((r1 & #xf000
) >> 12) |
#xe0
))
606 (r2 = ((r1 & #x3f
) |
#x80
))
612 (if (r0 == ,(charset-id 'eight-bit-control
))
614 ;; 0000 0yyy yyxx xxxx 110y yyyy 10xx xxxx
615 ;; 80 0000 0000 1000 0000 1100 0010 1000 0000
616 ;; 9f 0000 0000 1001 1111 1100 0010 1001 1111
620 (if (r0 == ,(charset-id 'eight-bit-graphic
))
622 ;; 0000 0yyy yyxx xxxx 110y yyyy 10xx xxxx
623 ;; a0 0000 0000 1010 0000 1100 0010 1010 0000
624 ;; ff 0000 0000 1111 1111 1101 1111 1011 1111
627 (read-multibyte-character r0 r1
)
628 (if (r0 != ,(charset-id 'eight-bit-graphic
))
629 (if (r0 != ,(charset-id 'eight-bit-control
))
633 ((read-multibyte-character r0 r2
)
634 (if (r0 != ,(charset-id 'eight-bit-graphic
))
635 (if (r0 != ,(charset-id 'eight-bit-control
))
645 ((lookup-character utf-subst-table-for-encode r0 r1
)
646 (if r7
; lookup succeeded
647 ((r1 = (((r0 & #xf000
) >> 12) |
#xe0
))
648 (r2 = ((r0 & #x3f
) |
#x80
))
653 ;; Unsupported character.
654 ;; Output U+FFFD, which is `ef bf bd' in UTF-8.
657 (write #xbd
)))))))))))
665 "CCL program to encode into UTF-8.")
668 (define-ccl-program ccl-untranslated-to-ucs
670 (if (r0 < #xf0
) ; 3-byte encoding, as above
672 (r3 = (r1 & #b11000000
))
673 (r3 |
= ((r2 >> 2) & #b00110000
))
674 (if (r3 != #b10100000
)
676 ((r3 = ((r0 & #x0f
) << 12))
677 (r3 += ((r1 & #x3f
) << 6))
684 (if (r0 < #xf8
) ; 4-byte (Mule-UCS recipe)
694 ((r1 = ((r1 & #x3F
) << 12))
695 (r2 = ((r2 & #x3F
) << 6))
697 (r0 = (((((r0 & #x07
) << 18) | r1
) | r2
) | r3
)))))))))
699 "Decode 3- or 4-byte sequences in r0, r1, r2 [,r3] to unicodes in r0.
700 r0 == 0 for invalid sequence.")
702 (defvar utf-8-ccl-regs
(make-vector 8 0))
704 (defsubst utf-8-untranslated-to-ucs
()
705 "Return the UCS code for an untranslated sequence of raw bytes t point.
706 Only for 3- or 4-byte sequences."
707 (aset utf-8-ccl-regs
0 (or (char-after) 0))
708 (aset utf-8-ccl-regs
1 (or (char-after (1+ (point))) 0))
709 (aset utf-8-ccl-regs
2 (or (char-after (+ 2 (point))) 0))
710 (aset utf-8-ccl-regs
3 (or (char-after (+ 3 (point))) 0))
711 (ccl-execute 'ccl-untranslated-to-ucs utf-8-ccl-regs
)
712 (aref utf-8-ccl-regs
0))
714 (defun utf-8-help-echo (window object position
)
715 (format "Untranslated Unicode U+%04X"
716 (get-char-property position
'untranslated-utf-8 object
)))
718 ;; We compose the untranslatable sequences into a single character.
719 ;; This is infelicitous for editing, because there's currently no
720 ;; mechanism for treating compositions as atomic, but is OK for
721 ;; display. They are composed to U+FFFD with help-echo which
722 ;; indicates the unicodes they represent. This function GCs too much.
723 (defsubst utf-8-compose
()
724 "Put a suitable composition on an untranslatable sequence.
725 Return the sequence's length."
726 (let* ((u (utf-8-untranslated-to-ucs))
732 (put-text-property (point) (min (point-max) (+ l
(point)))
733 'untranslated-utf-8 u
)
734 (put-text-property (point) (min (point-max) (+ l
(point)))
735 'help-echo
'utf-8-help-echo
)
736 (compose-region (point) (+ l
(point)) ?
\e$
,3u=\e(B)
739 (defcustom utf-8-compose-scripts nil
740 "*Non-nil means compose various scripts on decoding utf-8 text."
745 (defun utf-8-post-read-conversion (length)
746 "Compose untranslated utf-8 sequences into single characters.
747 Also compose particular scripts if `utf-8-compose-scripts' is non-nil."
749 ;; Can't do eval-when-compile to insert a multibyte constant
750 ;; version of the string in the loop, since it's always loaded as
751 ;; unibyte from a byte-compiled file.
752 (let ((range (string-as-multibyte "^\xe1-\xf7")))
753 (while (and (skip-chars-forward range
)
755 (forward-char (utf-8-compose)))))
756 ;; Fixme: Takahashi-san implies it may not work this easily. I
757 ;; asked why but didn't get a reply. -- fx
758 (when (and utf-8-compose-scripts
(> length
1))
759 ;; These currently have definitions which cover the relevant
760 ;; unicodes. We could avoid loading thai-util &c by checking
761 ;; whether the region contains any characters with the appropriate
762 ;; categories. There aren't yet Unicode-based rules for Tibetan.
763 (save-excursion (setq length
(diacritic-post-read-conversion length
)))
764 (save-excursion (setq length
(thai-post-read-conversion length
)))
765 (save-excursion (setq length
(lao-post-read-conversion length
)))
766 (save-excursion (setq length
(devanagari-post-read-conversion length
)))
767 (save-excursion (setq length
(malayalam-post-read-conversion length
)))
768 (save-excursion (setq length
(tamil-post-read-conversion length
))))
771 ;; ucs-tables is preloaded
772 ;; (defun utf-8-pre-write-conversion (beg end)
773 ;; "Semi-dummy pre-write function effectively to autoload ucs-tables."
774 ;; ;; Ensure translation-table is loaded.
775 ;; (require 'ucs-tables)
776 ;; ;; Don't do this again.
777 ;; (coding-system-put 'mule-utf-8 'pre-write-conversion nil)
782 "UTF-8 encoding for Emacs-supported Unicode characters.
783 It supports Unicode characters of these ranges:
784 U+0000..U+33FF, U+E000..U+FFFF.
785 They correspond to these Emacs character sets:
786 ascii, latin-iso8859-1, mule-unicode-0100-24ff,
787 mule-unicode-2500-33ff, mule-unicode-e000-ffff
789 On decoding (e.g. reading a file), Unicode characters not in the above
790 ranges are decoded into sequences of eight-bit-control and
791 eight-bit-graphic characters to preserve their byte sequences. The
792 byte sequence is preserved on i/o for valid utf-8, but not necessarily
795 On encoding (e.g. writing a file), Emacs characters not belonging to
796 any of the character sets listed above are encoded into the UTF-8 byte
797 sequence representing U+FFFD (REPLACEMENT CHARACTER)."
799 '(ccl-decode-mule-utf-8 . ccl-encode-mule-utf-8
)
805 mule-unicode-0100-24ff
806 mule-unicode-2500-33ff
807 mule-unicode-e000-ffff
)
808 (mime-charset . utf-8
)
809 (coding-category . coding-category-utf-8
)
810 (valid-codes (0 .
255))
811 ;; (pre-write-conversion . utf-8-pre-write-conversion)
812 (post-read-conversion . utf-8-post-read-conversion
)
813 (translation-table-for-encode . utf-translation-table-for-encode
)
814 (dependency unify-8859-on-encoding-mode
815 unify-8859-on-decoding-mode
816 utf-fragment-on-decoding
819 (define-coding-system-alias 'utf-8
'mule-utf-8
)
821 ;; I think this needs special private charsets defined for the
822 ;; untranslated sequences, if it's going to work well.
824 ;;; (defun utf-8-compose-function (pos to pattern &optional string)
825 ;;; (let* ((prop (get-char-property pos 'composition string))
826 ;;; (l (and prop (- (cadr prop) (car prop)))))
827 ;;; (cond ((and l (> l (- to pos)))
828 ;;; (delete-region pos to))
829 ;;; ((and (> (char-after pos) 224)
830 ;;; (< (char-after pos) 256)
831 ;;; (save-restriction
832 ;;; (narrow-to-region pos to)
833 ;;; (utf-8-compose)))
837 ;;; (aset composition-function-table
839 ;;; `((,(string-as-multibyte "[\200-\237\240-\377]")
840 ;;; . utf-8-compose-function))))
842 ;;; arch-tag: b08735b7-753b-4ae6-b754-0f3efe4515c5
843 ;;; utf-8.el ends here