git.chise.org Git - chise/xemacs-chise.git.1/blob - lisp/mule/mule-coding.el

   1 ;;; mule-coding.el --- Coding-system functions for Mule.
   2
   3 ;; Copyright (C) 1995 Electrotechnical Laboratory, JAPAN.
   4 ;; Licensed to the Free Software Foundation.
   5 ;; Copyright (C) 1995 Amdahl Corporation.
   6 ;; Copyright (C) 1995 Sun Microsystems.
   7 ;; Copyright (C) 1997,1999,2002,2003,2004 MORIOKA Tomohiko
   8
   9 ;; This file is part of XEmacs.
  10
  11 ;; XEmacs is free software; you can redistribute it and/or modify it
  12 ;; 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 ;; XEmacs is distributed in the hope that it will be useful, but
  17 ;; WITHOUT ANY WARRANTY; without even the implied warranty of
  18 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  19 ;; General Public License for more details.
  20
  21 ;; You should have received a copy of the GNU General Public License
  22 ;; along with XEmacs; 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.
  25
  26 ;;; Commentary:
  27
  28 ;;; split off of mule.el and mostly moved to coding.el
  29
  30 ;;; Code:
  31
  32 (defun coding-system-force-on-output (coding-system register)
  33   "Return the 'force-on-output property of CODING-SYSTEM for the specified REGISTER."
  34   (check-type register integer)
  35   (coding-system-property
  36    coding-system
  37    (case register
  38      (0 'force-g0-on-output)
  39      (1 'force-g1-on-output)
  40      (2 'force-g2-on-output)
  41      (3 'force-g3-on-output)
  42      (t (signal 'args-out-of-range (list register 0 3))))))
  43
  44 (defun coding-system-short (coding-system)
  45   "Return the 'short property of CODING-SYSTEM."
  46   (coding-system-property coding-system 'short))
  47
  48 (defun coding-system-no-ascii-eol (coding-system)
  49   "Return the 'no-ascii-eol property of CODING-SYSTEM."
  50   (coding-system-property coding-system 'no-ascii-eol))
  51
  52 (defun coding-system-no-ascii-cntl (coding-system)
  53   "Return the 'no-ascii-cntl property of CODING-SYSTEM."
  54   (coding-system-property coding-system 'no-ascii-cntl))
  55
  56 (defun coding-system-seven (coding-system)
  57   "Return the 'seven property of CODING-SYSTEM."
  58   (coding-system-property coding-system 'seven))
  59
  60 (defun coding-system-lock-shift (coding-system)
  61   "Return the 'lock-shift property of CODING-SYSTEM."
  62   (coding-system-property coding-system 'lock-shift))
  63
  64 ;;(defun coding-system-use-japanese-jisx0201-roman (coding-system)
  65 ;;  "Return the 'use-japanese-jisx0201-roman property of CODING-SYSTEM."
  66 ;;  (coding-system-property coding-system 'use-japanese-jisx0201-roman))
  67
  68 ;;(defun coding-system-use-japanese-jisx0208-1978 (coding-system)
  69 ;;  "Return the 'use-japanese-jisx0208-1978 property of CODING-SYSTEM."
  70 ;;  (coding-system-property coding-system 'use-japanese-jisx0208-2978))
  71
  72 (defun coding-system-no-iso6429 (coding-system)
  73   "Return the 'no-iso6429 property of CODING-SYSTEM."
  74   (coding-system-property coding-system 'no-iso6429))
  75
  76 (defun coding-system-ccl-encode (coding-system)
  77   "Return the CCL 'encode property of CODING-SYSTEM."
  78   (coding-system-property coding-system 'encode))
  79
  80 (defun coding-system-ccl-decode (coding-system)
  81   "Return the CCL 'decode property of CODING-SYSTEM."
  82   (coding-system-property coding-system 'decode))
  83
  84 \f
  85 ;;;; Definitions of predefined coding systems
  86
  87 (make-coding-system
  88  'ctext 'iso2022
  89  "Coding-system used in X as Compound Text Encoding."
  90  '(charset-g0 ascii
  91    charset-g1 latin-iso8859-1
  92    eol-type nil
  93    mnemonic "CText"))
  94
  95 ;;; iso-8859-1 and ctext are aliases.
  96
  97 ;; (copy-coding-system 'ctext 'iso-8859-1)
  98 (make-coding-system
  99  'iso-8859-1 'no-conversion
 100  "Coding-system used in X as Compound Text Encoding."
 101  '(eol-type nil mnemonic "Noconv"))
 102
 103 (make-coding-system
 104  'iso-2022-8bit-ss2 'iso2022
 105  "ISO-2022 coding system using SS2 for 96-charset in 8-bit code."
 106  '(charset-g0 ascii
 107    charset-g1 latin-iso8859-1
 108    charset-g2 t ;; unspecified but can be used later.
 109    short t
 110    mnemonic "ISO8/SS"
 111    ))
 112
 113 (make-coding-system
 114  'iso-2022-7bit-ss2 'iso2022
 115  "ISO-2022 coding system using SS2 for 96-charset in 7-bit code."
 116  '(charset-g0 ascii
 117    charset-g2 t ;; unspecified but can be used later.
 118    seven t
 119    short t
 120    mnemonic "ISO7/SS"
 121    eol-type nil))
 122
 123 ;; (copy-coding-system 'iso-2022-7bit-ss2 'iso-2022-jp-2)
 124 (make-coding-system
 125  'iso-2022-jp-2 'iso2022
 126  "ISO-2022 coding system using SS2 for 96-charset in 7-bit code."
 127  '(charset-g0 ascii
 128    charset-g2 t ;; unspecified but can be used later.
 129    seven t
 130    short t
 131    mnemonic "ISO7/SS"
 132    eol-type nil))
 133
 134 (make-coding-system
 135  'iso-2022-7bit 'iso2022
 136  "ISO 2022 based 7-bit encoding using only G0"
 137  '(charset-g0 ascii
 138    seven t
 139    short t
 140    mnemonic "ISO7"))
 141
 142 ;; compatibility for old XEmacsen
 143 (copy-coding-system 'iso-2022-7bit 'iso-2022-7)
 144
 145 (make-coding-system
 146  'iso-2022-8 'iso2022
 147  "ISO-2022 eight-bit coding system.  No single-shift or locking-shift."
 148  '(charset-g0 ascii
 149    charset-g1 latin-iso8859-1
 150    short t
 151    mnemonic "ISO8"
 152    ))
 153
 154 (make-coding-system
 155  'escape-quoted 'iso2022
 156  "ISO-2022 eight-bit coding system with escape quoting; used for .ELC files."
 157  '(charset-g0 ascii
 158    charset-g1 latin-iso8859-1
 159    eol-type lf
 160    escape-quoted t
 161    mnemonic "ESC/Quot"
 162    ))
 163
 164 (make-coding-system
 165  'iso-2022-lock 'iso2022
 166  "ISO-2022 coding system using Locking-Shift for 96-charset."
 167  '(charset-g0 ascii
 168    charset-g1 t ;; unspecified but can be used later.
 169    seven t
 170    lock-shift t
 171    mnemonic "ISO7/Lock"
 172    ))
 173
 174 (when (featurep 'utf-2000)
 175   (setq coded-charset-entity-reference-alist
 176         '(((ideograph-gt . isolated)            "I-GT-" 5 d)
 177           ( ideograph-gt                          "GT-" 5 d)
 178           ((=ruimoku-v6 . isolated)           "I-RUI6-" 4 X)
 179           ( =ruimoku-v6                         "RUI6-" 4 X)
 180           ((chinese-big5-cdp . isolated)       "I-CDP-" 4 X)
 181           ( chinese-big5-cdp                     "CDP-" 4 X)
 182           ((ideograph-daikanwa . isolated)       "I-M-" 5 d)
 183           ( ideograph-daikanwa                     "M-" 5 d)
 184           ((ideograph-gt-k . isolated)         "I-GT-K" 5 d)
 185           ( ideograph-gt-k                       "GT-K" 5 d)
 186           ((ideograph-cbeta . isolated)          "I-CB" 5 d)
 187           ( ideograph-cbeta                        "CB" 5 d)
 188           ((ideograph-hanziku-1 . isolated)  "I-HZK01-" 4 X)
 189           ( ideograph-hanziku-1                "HZK01-" 4 X)
 190           ((ideograph-hanziku-2 . isolated)  "I-HZK02-" 4 X)
 191           ( ideograph-hanziku-2                "HZK02-" 4 X)
 192           ((ideograph-hanziku-3 . isolated)  "I-HZK03-" 4 X)
 193           ( ideograph-hanziku-3                "HZK03-" 4 X)
 194           ((ideograph-hanziku-4 . isolated)  "I-HZK04-" 4 X)
 195           ( ideograph-hanziku-4                "HZK04-" 4 X)
 196           ((ideograph-hanziku-5 . isolated)  "I-HZK05-" 4 X)
 197           ( ideograph-hanziku-5                "HZK05-" 4 X)
 198           ((ideograph-hanziku-6 . isolated)  "I-HZK06-" 4 X)
 199           ( ideograph-hanziku-6                "HZK06-" 4 X)
 200           ((ideograph-hanziku-7 . isolated)  "I-HZK07-" 4 X)
 201           ( ideograph-hanziku-7                "HZK07-" 4 X)
 202           ((ideograph-hanziku-8 . isolated)  "I-HZK08-" 4 X)
 203           ( ideograph-hanziku-8                "HZK08-" 4 X)
 204           ((ideograph-hanziku-9 . isolated)  "I-HZK09-" 4 X)
 205           ( ideograph-hanziku-9                "HZK09-" 4 X)
 206           ((ideograph-hanziku-10 . isolated) "I-HZK10-" 4 X)
 207           ( ideograph-hanziku-10               "HZK10-" 4 X)
 208           ((ideograph-hanziku-11 . isolated) "I-HZK11-" 4 X)
 209           ( ideograph-hanziku-11               "HZK11-" 4 X)
 210           ((ideograph-hanziku-12 . isolated) "I-HZK12-" 4 X)
 211           ( ideograph-hanziku-12               "HZK12-" 4 X)
 212           ((ideograph-hanziku-1 . isolated)   "I-HZK1-" 4 X)
 213           ( ideograph-hanziku-1                 "HZK1-" 4 X)
 214           ((ideograph-hanziku-2 . isolated)   "I-HZK2-" 4 X)
 215           ( ideograph-hanziku-2                 "HZK2-" 4 X)
 216           ((japanese-jisx0208-1990 . isolated) "I-J90-" 4 X)
 217           ( japanese-jisx0208-1990               "J90-" 4 X)
 218           ((japanese-jisx0208 . isolated)      "I-J83-" 4 X)
 219           ( japanese-jisx0208                    "J83-" 4 X)
 220           ((japanese-jisx0213-1 . isolated)    "I-JX1-" 4 X)
 221           ( japanese-jisx0213-1                  "JX1-" 4 X)
 222           ((japanese-jisx0213-2 . isolated)    "I-JX2-" 4 X)
 223           ( japanese-jisx0213-2                  "JX2-" 4 X)
 224           ((japanese-jisx0212 . isolated)      "I-JSP-" 4 X)
 225           ( japanese-jisx0212                    "JSP-" 4 X)
 226           ((japanese-jisx0208-1978 . isolated) "I-J78-" 4 X)
 227           ( japanese-jisx0208-1978               "J78-" 4 X)
 228           ((chinese-cns11643-1 . isolated)      "I-C1-" 4 X)
 229           ( chinese-cns11643-1                    "C1-" 4 X)
 230           ((chinese-cns11643-2 . isolated)      "I-C2-" 4 X)
 231           ( chinese-cns11643-2                    "C2-" 4 X)
 232           ((chinese-cns11643-3 . isolated)      "I-C3-" 4 X)
 233           ( chinese-cns11643-3                    "C3-" 4 X)
 234           ((chinese-cns11643-4 . isolated)      "I-C4-" 4 X)
 235           ( chinese-cns11643-4                    "C4-" 4 X)
 236           ((chinese-cns11643-5 . isolated)      "I-C5-" 4 X)
 237           ( chinese-cns11643-5                    "C5-" 4 X)
 238           ((chinese-cns11643-6 . isolated)      "I-C6-" 4 X)
 239           ( chinese-cns11643-6                    "C6-" 4 X)
 240           ((chinese-cns11643-7 . isolated)      "I-C7-" 4 X)
 241           ( chinese-cns11643-7                    "C7-" 4 X)
 242           ((korean-ksc5601 . isolated)          "I-K0-" 4 X)
 243           ( korean-ksc5601                        "K0-" 4 X)
 244           ( china3-jef                           "JC3-" 4 X)
 245           ( chinese-big5                           "B-" 4 X)
 246           ( chinese-big5                          "C0-" 4 X)
 247           ))
 248
 249   (make-coding-system
 250    'utf-8-mcs-er 'utf-8
 251    "Coding-system of UTF-8 with entity-reference."
 252    '(mnemonic "MTF8r" use-entity-reference t))
 253
 254   (make-coding-system
 255    'utf-8-gb 'utf-8
 256    "Coding-system of UTF-8 using GB mapping."
 257    '(mnemonic "UTF8G"
 258               charset-g0 ucs-gb
 259               charset-g1 =>ucs-gb
 260               charset-g2 =>ucs))
 261
 262   (make-coding-system
 263    'utf-8-gb-er 'utf-8
 264    "Coding-system of UTF-8 using GB mapping with entity-reference."
 265    '(mnemonic "UTF8Gr"
 266               charset-g0 ucs-gb
 267               charset-g1 =>ucs-gb
 268               charset-g2 =>ucs
 269               use-entity-reference t))
 270
 271   (make-coding-system
 272    'utf-8-cns 'utf-8
 273    "Coding-system of UTF-8 using CNS mapping."
 274    '(mnemonic "UTF8C"
 275               charset-g0 ucs-cns
 276               charset-g1 =>ucs-cns
 277               charset-g2 =>ucs))
 278
 279   (make-coding-system
 280    'utf-8-cns-er 'utf-8
 281    "Coding-system of UTF-8 using CNS mapping with entity-reference."
 282    '(mnemonic "UTF8Cr"
 283               charset-g0 ucs-cns
 284               charset-g1 =>ucs-cns
 285               charset-g2 =>ucs
 286               use-entity-reference t))
 287
 288   (make-coding-system
 289    'utf-8-big5 'utf-8
 290    "Coding-system of UTF-8 using Big5 mapping."
 291    '(mnemonic "UTF8B"
 292               charset-g0 ucs-big5
 293               charset-g1 =>ucs-big5
 294               charset-g2 =>ucs))
 295
 296   (make-coding-system
 297    'utf-8-big5-er 'utf-8
 298    "Coding-system of UTF-8 using Big5 mapping with entity-reference."
 299    '(mnemonic "UTF8Br"
 300               charset-g0 ucs-big5
 301               charset-g1 =>ucs-big5
 302               charset-g2 =>ucs
 303               use-entity-reference t))
 304
 305   (make-coding-system
 306    'utf-8-jis 'utf-8
 307    "Coding-system of UTF-8 using JIS mapping."
 308    '(mnemonic "UTF8J"
 309               charset-g0 =ucs@jis-2000
 310               charset-g1 =>ucs-jis
 311               charset-g2 =>ucs))
 312
 313   (make-coding-system
 314    'utf-8-jis-er 'utf-8
 315    "Coding-system of UTF-8 using JIS mapping with entity-reference."
 316    '(mnemonic "UTF8Jr"
 317               charset-g0 =ucs@jis-2000
 318               charset-g1 =>ucs-jis
 319               charset-g2 =>ucs
 320               use-entity-reference t))
 321
 322   (make-coding-system
 323    'utf-8-jp 'utf-8
 324    "Coding-system of UTF-8 for common glyphs used in Japan."
 325    '(mnemonic "UTF8J"
 326               charset-g0 =ucs@jp
 327               charset-g1 =>ucs-jis
 328               charset-g2 =>ucs))
 329
 330   (make-coding-system
 331    'utf-8-jp-er 'utf-8
 332    "Coding-system of UTF-8 using =ucs@jp mapping with entity-reference."
 333    '(mnemonic "UTF8Jr"
 334               charset-g0 =ucs@jp
 335               charset-g1 =>ucs-jis
 336               charset-g2 =>ucs
 337               use-entity-reference t))
 338
 339   (make-coding-system
 340    'utf-8-ks 'utf-8
 341    "Coding-system of UTF-8 using KS mapping."
 342    '(mnemonic "UTF8K"
 343               charset-g0 ucs-ks
 344               charset-g1 =>ucs-ks
 345               charset-g2 =>ucs))
 346
 347   (make-coding-system
 348    'utf-8-ks-er 'utf-8
 349    "Coding-system of UTF-8 using KS mapping with entity-reference."
 350    '(mnemonic "UTF8Kr"
 351               charset-g0 ucs-ks
 352               charset-g1 =>ucs-ks
 353               charset-g2 =>ucs
 354               use-entity-reference t))
 355
 356   (define-coding-system-alias 'utf-8 'utf-8-mcs)
 357   (define-coding-system-alias 'utf-8-er 'utf-8-mcs-er)
 358   )
 359
 360 (make-coding-system
 361  'euc-jisx0213 'iso2022
 362  "Coding-system of Japanese EUC based on JIS X 0213."
 363  '(charset-g0 ascii
 364    charset-g1 japanese-jisx0213-1
 365    charset-g2 katakana-jisx0201
 366    charset-g3 japanese-jisx0213-2
 367    short t
 368    mnemonic "Ja/EUC0213"
 369    ))
 370
 371 ;; initialize the coding categories to something semi-reasonable
 372 ;; so that the remaining Lisp files can contain extended characters.
 373 ;; (They will be in ISO-7 format)
 374 ;; #### This list needs to be synched with the ones in mule-cmds.el.
 375
 376 (if (featurep 'utf-2000)
 377     (set-coding-priority-list '(iso-7
 378                                 no-conversion
 379                                 utf-8
 380                                 iso-8-1
 381                                 iso-8-2
 382                                 iso-8-designate
 383                                 iso-lock-shift
 384                                 shift-jis
 385                                 big5
 386                                 ucs-4))
 387   (set-coding-priority-list '(iso-7
 388                               no-conversion
 389                               ;; utf-8
 390                               iso-8-1
 391                               iso-8-2
 392                               iso-8-designate
 393                               iso-lock-shift
 394                               shift-jis
 395                               big5
 396                               ;; ucs-4
 397                               )))
 398
 399 (set-coding-category-system 'iso-7 'iso-2022-7)
 400 (set-coding-category-system 'iso-8-designate 'ctext)
 401 (set-coding-category-system 'iso-8-1 'ctext)
 402 (set-coding-category-system 'iso-lock-shift 'iso-2022-lock)
 403 (set-coding-category-system 'no-conversion 'no-conversion)
 404
 405 (setq-default buffer-file-coding-system 'iso-2022-8)
 406
 407 ;;; mule-coding.el ends here