771
+ − 1 ;;; unicode.el --- Unicode support -*- coding: iso-2022-7bit; -*-
+ − 2
778
+ − 3 ;; Copyright (C) 2001, 2002 Ben Wing.
771
+ − 4
+ − 5 ;; Keywords: multilingual, Unicode
+ − 6
+ − 7 ;; This file is part of XEmacs.
+ − 8
+ − 9 ;; XEmacs is free software; you can redistribute it and/or modify it
+ − 10 ;; under the terms of the GNU General Public License as published by
+ − 11 ;; the Free Software Foundation; either version 2, or (at your option)
+ − 12 ;; any later version.
+ − 13
+ − 14 ;; XEmacs is distributed in the hope that it will be useful, but
+ − 15 ;; WITHOUT ANY WARRANTY; without even the implied warranty of
+ − 16 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ − 17 ;; General Public License for more details.
+ − 18
+ − 19 ;; You should have received a copy of the GNU General Public License
+ − 20 ;; along with XEmacs; see the file COPYING. If not, write to the Free
+ − 21 ;; Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA
+ − 22 ;; 02111-1307, USA.
+ − 23
+ − 24 ;;; Synched up with: Not in FSF.
+ − 25
+ − 26 ;;; Commentary:
+ − 27
+ − 28 ;; Lisp support for Unicode, e.g. initialize the translation tables.
+ − 29
+ − 30 ;;; Code:
+ − 31
778
+ − 32 ; ;; Subsets of Unicode.
+ − 33
+ − 34 ; (make-charset 'mule-unicode-2500-33ff
+ − 35 ; "Unicode characters of the range U+2500..U+33FF."
+ − 36 ; '(dimension
+ − 37 ; 2
+ − 38 ; registry "ISO10646-1"
+ − 39 ; chars 96
+ − 40 ; columns 1
+ − 41 ; direction l2r
+ − 42 ; final ?2
+ − 43 ; graphic 0
+ − 44 ; short-name "Unicode subset 2"
+ − 45 ; long-name "Unicode subset (U+2500..U+33FF)"
+ − 46 ; ))
+ − 47
+ − 48
+ − 49 ; (make-charset 'mule-unicode-e000-ffff
+ − 50 ; "Unicode characters of the range U+E000..U+FFFF."
+ − 51 ; '(dimension
+ − 52 ; 2
+ − 53 ; registry "ISO10646-1"
+ − 54 ; chars 96
+ − 55 ; columns 1
+ − 56 ; direction l2r
+ − 57 ; final ?3
+ − 58 ; graphic 0
+ − 59 ; short-name "Unicode subset 3"
+ − 60 ; long-name "Unicode subset (U+E000+FFFF)"
+ − 61 ; ))
+ − 62
+ − 63
+ − 64 ; (make-charset 'mule-unicode-0100-24ff
+ − 65 ; "Unicode characters of the range U+0100..U+24FF."
+ − 66 ; '(dimension
+ − 67 ; 2
+ − 68 ; registry "ISO10646-1"
+ − 69 ; chars 96
+ − 70 ; columns 1
+ − 71 ; direction l2r
+ − 72 ; final ?1
+ − 73 ; graphic 0
+ − 74 ; short-name "Unicode subset"
+ − 75 ; long-name "Unicode subset (U+0100..U+24FF)"
+ − 76 ; ))
+ − 77
+ − 78
771
+ − 79 ;; NOTE: This takes only a fraction of a second on my Pentium III
+ − 80 ;; 700Mhz even with a totally optimization-disabled XEmacs.
+ − 81 (defun load-unicode-tables ()
+ − 82 "Initialize the Unicode translation tables for all standard charsets."
780
+ − 83 (let ((parse-args
+ − 84 '(("unicode/unicode-consortium"
877
+ − 85 ;; Due to the braindamaged way Mule treats the ASCII and Control-1
+ − 86 ;; charsets' types, trying to load them results in out-of-range
+ − 87 ;; warnings at unicode.c:1439. They're no-ops anyway, they're
+ − 88 ;; hardwired in unicode.c (unicode_to_ichar, ichar_to_unicode).
+ − 89 ;; ("8859-1.TXT" ascii #x00 #x7F #x0)
+ − 90 ;; ("8859-1.TXT" control-1 #x80 #x9F #x-80)
+ − 91 ;; The 8859-1.TXT G1 assignments are half no-ops, hardwired in
+ − 92 ;; unicode.c ichar_to_unicode, but not in unicode_to_ichar.
780
+ − 93 ("8859-1.TXT" latin-iso8859-1 #xA0 #xFF #x-80)
+ − 94 ;; "8859-10.TXT"
+ − 95 ;; "8859-13.TXT"
+ − 96 ("8859-14.TXT" latin-iso8859-14 #xA0 #xFF #x-80)
+ − 97 ("8859-15.TXT" latin-iso8859-15 #xA0 #xFF #x-80)
+ − 98 ("8859-2.TXT" latin-iso8859-2 #xA0 #xFF #x-80)
+ − 99 ("8859-3.TXT" latin-iso8859-3 #xA0 #xFF #x-80)
+ − 100 ("8859-4.TXT" latin-iso8859-4 #xA0 #xFF #x-80)
+ − 101 ("8859-5.TXT" cyrillic-iso8859-5 #xA0 #xFF #x-80)
+ − 102 ("8859-6.TXT" arabic-iso8859-6 #xA0 #xFF #x-80)
+ − 103 ("8859-7.TXT" greek-iso8859-7 #xA0 #xFF #x-80)
+ − 104 ("8859-8.TXT" hebrew-iso8859-8 #xA0 #xFF #x-80)
+ − 105 ("8859-9.TXT" latin-iso8859-9 #xA0 #xFF #x-80)
+ − 106 ;; charset for Big5 does not matter; specifying `big5' will
+ − 107 ;; automatically make the right thing happen
+ − 108 ("BIG5.TXT" chinese-big5-1 nil nil nil big5)
+ − 109 ("CNS11643.TXT" chinese-cns11643-1 #x10000 #x1FFFF #x-10000)
+ − 110 ("CNS11643.TXT" chinese-cns11643-2 #x20000 #x2FFFF #x-20000)
+ − 111 ;; "CP1250.TXT"
+ − 112 ;; "CP1251.TXT"
+ − 113 ;; "CP1252.TXT"
+ − 114 ;; "CP1253.TXT"
+ − 115 ;; "CP1254.TXT"
+ − 116 ;; "CP1255.TXT"
+ − 117 ;; "CP1256.TXT"
+ − 118 ;; "CP1257.TXT"
+ − 119 ;; "CP1258.TXT"
+ − 120 ;; "CP874.TXT"
+ − 121 ;; "CP932.TXT"
+ − 122 ;; "CP936.TXT"
+ − 123 ;; "CP949.TXT"
+ − 124 ;; "CP950.TXT"
+ − 125 ;; "GB12345.TXT"
+ − 126 ("GB2312.TXT" chinese-gb2312)
+ − 127 ;; "HANGUL.TXT"
+ − 128 ("JIS0201.TXT" latin-jisx0201 #x21 #x80)
+ − 129 ("JIS0201.TXT" katakana-jisx0201 #xA0 #xFF #x-80)
+ − 130 ("JIS0208.TXT" japanese-jisx0208 nil nil nil ignore-first-column)
+ − 131 ("JIS0212.TXT" japanese-jisx0212)
+ − 132 ;; "JOHAB.TXT"
+ − 133 ;; "KOI8-R.TXT"
+ − 134 ;; "KSC5601.TXT"
+ − 135 ;; note that KSC5601.TXT as currently distributed is NOT what
+ − 136 ;; it claims to be! see comments in KSX1001.TXT.
+ − 137 ("KSX1001.TXT" korean-ksc5601)
+ − 138 ;; "OLD5601.TXT"
+ − 139 ;; "SHIFTJIS.TXT"
+ − 140 )
+ − 141 ("unicode/mule-ucs"
+ − 142 ;; use these instead of the above ones once we support surrogates
+ − 143 ;;("chinese-cns11643-1.txt" chinese-cns11643-1)
+ − 144 ;;("chinese-cns11643-2.txt" chinese-cns11643-2)
+ − 145 ;;("chinese-cns11643-3.txt" chinese-cns11643-3)
+ − 146 ;;("chinese-cns11643-4.txt" chinese-cns11643-4)
+ − 147 ;;("chinese-cns11643-5.txt" chinese-cns11643-5)
+ − 148 ;;("chinese-cns11643-6.txt" chinese-cns11643-6)
+ − 149 ;;("chinese-cns11643-7.txt" chinese-cns11643-7)
+ − 150 ("chinese-sisheng.txt" chinese-sisheng)
+ − 151 ("ethiopic.txt" ethiopic)
+ − 152 ("indian-is13194.txt" indian-is13194)
+ − 153 ("ipa.txt" ipa)
+ − 154 ("thai-tis620.txt" thai-tis620)
+ − 155 ("tibetan.txt" tibetan)
+ − 156 ("vietnamese-viscii-lower.txt" vietnamese-viscii-lower)
+ − 157 ("vietnamese-viscii-upper.txt" vietnamese-viscii-upper)
+ − 158 )
+ − 159 ("unicode/other"
+ − 160 ("lao.txt" lao)
+ − 161 )
771
+ − 162 )))
780
+ − 163 (mapcar #'(lambda (tables)
+ − 164 (let ((undir
+ − 165 (expand-file-name (car tables) data-directory)))
+ − 166 (mapcar #'(lambda (args)
1318
+ − 167 (apply 'load-unicode-mapping-table
780
+ − 168 (expand-file-name (car args) undir)
+ − 169 (cdr args)))
+ − 170 (cdr tables))))
771
+ − 171 parse-args)))
+ − 172
+ − 173 (defun init-unicode-at-startup ()
+ − 174 (load-unicode-tables))
+ − 175
+ − 176 (make-coding-system
+ − 177 'utf-16 'unicode
+ − 178 "UTF-16"
+ − 179 '(mnemonic "UTF-16"
+ − 180 documentation
+ − 181 "UTF-16 Unicode encoding -- the standard (almost-) fixed-width
+ − 182 two-byte encoding, with surrogates. It will be fixed-width if all
+ − 183 characters are in the BMP (Basic Multilingual Plane -- first 65536
+ − 184 codepoints). Cannot represent characters with codepoints above
+ − 185 0x10FFFF (a little more than 1,000,000). Unicode and ISO guarantee
+ − 186 never to encode any characters outside this range -- all the rest are
+ − 187 for private, corporate or internal use."
+ − 188 type utf-16))
+ − 189
+ − 190 (make-coding-system
+ − 191 'utf-16-bom 'unicode
+ − 192 "UTF-16 w/BOM"
+ − 193 '(mnemonic "UTF16-BOM"
+ − 194 documentation
+ − 195 "UTF-16 Unicode encoding with byte order mark (BOM) at the beginning.
+ − 196 The BOM is Unicode character U+FEFF -- i.e. the first two bytes are
+ − 197 0xFE and 0xFF, respectively, or reversed in a little-endian
+ − 198 representation. It has been sanctioned by the Unicode Consortium for
+ − 199 use at the beginning of a Unicode stream as a marker of the byte order
+ − 200 of the stream, and commonly appears in Unicode files under Microsoft
+ − 201 Windows, where it also functions as a magic cookie identifying a
+ − 202 Unicode file. The character is called \"ZERO WIDTH NO-BREAK SPACE\"
+ − 203 and is suitable as a byte-order marker because:
+ − 204
+ − 205 -- it has no displayable representation
+ − 206 -- due to its semantics it never normally appears at the beginning
+ − 207 of a stream
+ − 208 -- its reverse U+FFFE is not a legal Unicode character
+ − 209 -- neither byte sequence is at all likely in any other standard
+ − 210 encoding, particularly at the beginning of a stream
+ − 211
+ − 212 This coding system will insert a BOM at the beginning of a stream when
+ − 213 writing and strip it off when reading."
+ − 214 type utf-16
+ − 215 need-bom t))
+ − 216
+ − 217 (make-coding-system
+ − 218 'utf-16-little-endian 'unicode
+ − 219 "UTF-16 Little Endian"
+ − 220 '(mnemonic "UTF16-LE"
+ − 221 documentation
+ − 222 "Little-endian version of UTF-16 Unicode encoding.
+ − 223 See `utf-16' coding system."
+ − 224 type utf-16
+ − 225 little-endian t))
+ − 226
+ − 227 (make-coding-system
+ − 228 'utf-16-little-endian-bom 'unicode
+ − 229 "UTF-16 Little Endian w/BOM"
+ − 230 '(mnemonic "MSW-Unicode"
+ − 231 documentation
+ − 232 "Little-endian version of UTF-16 Unicode encoding, with byte order mark.
+ − 233 Standard encoding for representing Unicode under MS Windows. See
+ − 234 `utf-16-bom' coding system."
+ − 235 type utf-16
+ − 236 little-endian t
+ − 237 need-bom t))
+ − 238
+ − 239 (make-coding-system
+ − 240 'ucs-4 'unicode
+ − 241 "UCS-4"
+ − 242 '(mnemonic "UCS4"
+ − 243 documentation
+ − 244 "UCS-4 Unicode encoding -- fully fixed-width four-byte encoding."
+ − 245 type ucs-4))
+ − 246
+ − 247 (make-coding-system
+ − 248 'ucs-4-little-endian 'unicode
+ − 249 "UCS-4 Little Endian"
+ − 250 '(mnemonic "UCS4-LE"
+ − 251 documentation
+ − 252 "Little-endian version of UCS-4 Unicode encoding. See `ucs-4' coding system."
+ − 253 type ucs-4
+ − 254 little-endian t))
+ − 255
+ − 256 (make-coding-system
+ − 257 'utf-8 'unicode
+ − 258 "UTF-8"
+ − 259 '(mnemonic "UTF8"
+ − 260 documentation
+ − 261 "UTF-8 Unicode encoding -- ASCII-compatible 8-bit variable-width encoding
+ − 262 with the same principles as the Mule-internal encoding:
+ − 263
+ − 264 -- All ASCII characters (codepoints 0 through 127) are represented
+ − 265 by themselves (i.e. using one byte, with the same value as the
+ − 266 ASCII codepoint), and these bytes are disjoint from bytes
+ − 267 representing non-ASCII characters.
+ − 268
+ − 269 This means that any 8-bit clean application can safely process
+ − 270 UTF-8-encoded text as it were ASCII, with no corruption (e.g. a
+ − 271 '/' byte is always a slash character, never the second byte of
+ − 272 some other character, as with Big5, so a pathname encoded in
+ − 273 UTF-8 can safely be split up into components and reassembled
+ − 274 again using standard ASCII processes).
+ − 275
+ − 276 -- Leading bytes and non-leading bytes in the encoding of a
+ − 277 character are disjoint, so moving backwards is easy.
+ − 278
+ − 279 -- Given only the leading byte, you know how many following bytes
+ − 280 are present.
+ − 281 "
+ − 282 type utf-8))
+ − 283
985
+ − 284 (make-coding-system
+ − 285 'utf-8-bom 'unicode
+ − 286 "UTF-8 w/BOM"
+ − 287 '(mnemonic "MSW-UTF8"
+ − 288 documentation
+ − 289 "UTF-8 Unicode encoding, with byte order mark.
+ − 290 Standard encoding for representing UTF-8 under MS Windows."
+ − 291 type utf-8
+ − 292 little-endian t
+ − 293 need-bom t))
+ − 294
771
+ − 295 ;; #### UTF-7 is not yet implemented, and it's tricky to do. There's
+ − 296 ;; an implementation in appendix A.1 of the Unicode Standard, Version
+ − 297 ;; 2.0, but I don't know its licensing characteristics.
+ − 298
+ − 299 ; (make-coding-system
+ − 300 ; 'utf-7 'unicode
+ − 301 ; "UTF-7"
+ − 302 ; '(mnemonic "UTF7"
+ − 303 ; documentation
+ − 304 ; "UTF-7 Unicode encoding -- 7-bit-ASCII modal Internet-mail-compatible
+ − 305 ; encoding especially designed for headers, with the following
+ − 306 ; properties:
+ − 307
+ − 308 ; -- Only characters that are considered safe for passing through any mail
+ − 309 ; gateway without damage are used.
+ − 310
+ − 311 ; -- This is a modal encoding, with two states. The first, default
+ − 312 ; state encodes the most common Unicode characters (upper and
+ − 313 ; lowercase letters, digits, and 9 common punctuation marks) as
+ − 314 ; themselves, and the second state, entered using '+' and
+ − 315 ; terminated with '-' or any character disallowed in state 2,
+ − 316 ; encodes any Unicode characters by first converting to UTF-16,
+ − 317 ; most significant byte first, and then to a slightly modified
+ − 318 ; Base64 encoding. (Thus, UTF-7 has the same limitations on the
+ − 319 ; characters it can encode as UTF-16.)
+ − 320
+ − 321 ; -- The modified Base64 encoding deviates from standard Base64 in
+ − 322 ; that it omits the `=' pad character. This is eliminated so as to
+ − 323 ; avoid conflicts with the use of `=' as an escape in the
+ − 324 ; Quoted-Printable encoding and the related Q encoding for headers:
+ − 325 ; With this modification, non-whitespace chars in UTF-7 will be
+ − 326 ; represented in Quoted-Printable and in Q as-is, with no further
+ − 327 ; encoding.
+ − 328
+ − 329 ; For more information, see Appendix A.1 of The Unicode Standard 2.0, or
+ − 330 ; wherever it is in v3.0."
+ − 331 ; type utf-7))