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           ((chinese-gb2312 . isolated)          "I-G0-" 4 X)
 245           ( chinese-gb2312                        "G0-" 4 X)
 246           ((=iso-ir165 . isolated)             "I-EGB-" 4 X)
 247           ( =iso-ir165                           "EGB-" 4 X)
 248           ((latin-iso8859-1 . isolated)     "I-LATIN1-" 2 X)
 249           ( latin-iso8859-1                   "LATIN1-" 2 X)
 250           ((latin-iso8859-2 . isolated)     "I-LATIN2-" 2 X)
 251           ( latin-iso8859-2                   "LATIN2-" 2 X)
 252           ((latin-iso8859-3 . isolated)     "I-LATIN3-" 2 X)
 253           ( latin-iso8859-3                   "LATIN3-" 2 X)
 254           ((latin-iso8859-4 . isolated)     "I-LATIN4-" 2 X)
 255           ( latin-iso8859-4                   "LATIN4-" 2 X)
 256           ((cyrillic-iso8859-5 . isolated) "I-CYRILLIC-" 2 X)
 257           ( cyrillic-iso8859-5              "CYRILLIC-" 2 X)
 258           ((greek-iso8859-7 . isolated)      "I-GREEK-" 2 X)
 259           ( greek-iso8859-7                    "GREEK-" 2 X)
 260           ((hebrew-iso8859-8 . isolated)    "I-HEBREW-" 2 X)
 261           ( hebrew-iso8859-8                  "HEBREW-" 2 X)
 262           ((latin-iso8859-9 . isolated)     "I-LATIN5-" 2 X)
 263           ( latin-iso8859-9                   "LATIN5-" 2 X)
 264           ((latin-jisx0201 . isolated)      "I-LATINJ-" 2 X)
 265           ( latin-jisx0201                    "LATINJ-" 2 X)
 266           ((katakana-jisx0201 . isolated) "I-KATAKANA-" 2 X)
 267           ( katakana-jisx0201               "KATAKANA-" 2 X)
 268           ;; ((latin-viscii . isolated)        "I-VISCII-" 2 X)
 269           ;; ( latin-viscii                      "VISCII-" 2 X)
 270           ((latin-tcvn5712 . isolated)      "I-VSCII2-" 2 X)
 271           ( latin-tcvn5712                    "VSCII2-" 2 X)
 272           ((thai-tis620 . isolated)           "I-THAI-" 2 X)
 273           ( thai-tis620                         "THAI-" 2 X)
 274           ((lao . isolated)               "I-MULE-LAO-" 2 X)
 275           ( lao                             "MULE-LAO-" 2 X)
 276           ( ethiopic                      "MULE-ETHIO-" 4 X)
 277           ((arabic-1-column . isolated)  "I-MULE-ARB1-" 2 X)
 278           ( arabic-1-column                "MULE-ARB1-" 2 X)
 279           ((arabic-2-column . isolated)  "I-MULE-ARB2-" 2 X)
 280           ( arabic-2-column                "MULE-ARB2-" 2 X)
 281           ((arabic-digit . isolated)     "I-MULE-ARBD-" 2 X)
 282           ( arabic-digit                   "MULE-ARBD-" 2 X)
 283           ((ipa . isolated)               "I-MULE-IPA-" 2 X)
 284           ( ipa                             "MULE-IPA-" 2 X)
 285           ((china3-jef . isolated)             "I-JC3-" 4 X)
 286           ( china3-jef                           "JC3-" 4 X)
 287           ((chinese-big5 . isolated)             "I-B-" 4 X)
 288           ( chinese-big5                           "B-" 4 X)
 289           ( chinese-big5                          "C0-" 4 X)
 290           ( =ucs@unicode                          "UU+" 4 X)
 291           ( =ucs@unicode                          "UU-" 8 X)
 292           ( =ucs@iso                               "U+" 4 X)
 293           ( =ucs@iso                               "U-" 8 X)
 294           ( =ucs@gb                               "GU+" 4 X)
 295           ( =ucs@gb                               "GU-" 8 X)
 296           ( =ucs@jis                              "JU+" 4 X)
 297           ( =ucs@jis                              "JU-" 8 X)
 298           ( =ucs@cns                              "CU+" 4 X)
 299           ( =ucs@cns                              "CU-" 8 X)
 300           ( =ucs@ks                               "KU+" 4 X)
 301           ( =ucs@ks                               "KU-" 8 X)
 302           ( =jis-x0208@1997                      "J97-" 4 X)
 303           ))
 304
 305   (make-coding-system
 306    'utf-8-mcs-er 'utf-8
 307    "Coding-system of UTF-8 with entity-reference."
 308    '(mnemonic "MTF8r" use-entity-reference t))
 309
 310   (make-coding-system
 311    'utf-8-gb 'utf-8
 312    "Coding-system of UTF-8 using GB mapping."
 313    '(mnemonic "UTF8G"
 314               charset-g0 ucs-gb
 315               charset-g1 =>ucs-gb
 316               charset-g2 =>ucs))
 317
 318   (make-coding-system
 319    'utf-8-gb-er 'utf-8
 320    "Coding-system of UTF-8 using GB mapping with entity-reference."
 321    '(mnemonic "UTF8Gr"
 322               charset-g0 ucs-gb
 323               charset-g1 =>ucs-gb
 324               charset-g2 =>ucs
 325               use-entity-reference t))
 326
 327   (make-coding-system
 328    'utf-8-cns 'utf-8
 329    "Coding-system of UTF-8 using CNS mapping."
 330    '(mnemonic "UTF8C"
 331               charset-g0 ucs-cns
 332               charset-g1 =>ucs-cns
 333               charset-g2 =>ucs))
 334
 335   (make-coding-system
 336    'utf-8-cns-er 'utf-8
 337    "Coding-system of UTF-8 using CNS mapping with entity-reference."
 338    '(mnemonic "UTF8Cr"
 339               charset-g0 ucs-cns
 340               charset-g1 =>ucs-cns
 341               charset-g2 =>ucs
 342               use-entity-reference t))
 343
 344   (make-coding-system
 345    'utf-8-big5 'utf-8
 346    "Coding-system of UTF-8 using Big5 mapping."
 347    '(mnemonic "UTF8B"
 348               charset-g0 ucs-big5
 349               charset-g1 =>ucs-big5
 350               charset-g2 =>ucs))
 351
 352   (make-coding-system
 353    'utf-8-big5-er 'utf-8
 354    "Coding-system of UTF-8 using Big5 mapping with entity-reference."
 355    '(mnemonic "UTF8Br"
 356               charset-g0 ucs-big5
 357               charset-g1 =>ucs-big5
 358               charset-g2 =>ucs
 359               use-entity-reference t))
 360
 361   (make-coding-system
 362    'utf-8-jis 'utf-8
 363    "Coding-system of UTF-8 using JIS mapping."
 364    '(mnemonic "UTF8J"
 365               charset-g0 =ucs@jis-2000
 366               charset-g1 =>ucs-jis
 367               charset-g2 =>ucs))
 368
 369   (make-coding-system
 370    'utf-8-jis-er 'utf-8
 371    "Coding-system of UTF-8 using JIS mapping with entity-reference."
 372    '(mnemonic "UTF8Jr"
 373               charset-g0 =ucs@jis-2000
 374               charset-g1 =>ucs-jis
 375               charset-g2 =>ucs
 376               use-entity-reference t))
 377
 378   (make-coding-system
 379    'utf-8-jp 'utf-8
 380    "Coding-system of UTF-8 for common glyphs used in Japan."
 381    '(mnemonic "UTF8J"
 382               charset-g0 =ucs@jp
 383               charset-g1 =>ucs-jis
 384               charset-g2 =>ucs))
 385
 386   (make-coding-system
 387    'utf-8-jp-er 'utf-8
 388    "Coding-system of UTF-8 using =ucs@jp mapping with entity-reference."
 389    '(mnemonic "UTF8Jr"
 390               charset-g0 =ucs@jp
 391               charset-g1 =>ucs-jis
 392               charset-g2 =>ucs
 393               use-entity-reference t))
 394
 395   (make-coding-system
 396    'utf-8-ks 'utf-8
 397    "Coding-system of UTF-8 using KS mapping."
 398    '(mnemonic "UTF8K"
 399               charset-g0 ucs-ks
 400               charset-g1 =>ucs-ks
 401               charset-g2 =>ucs))
 402
 403   (make-coding-system
 404    'utf-8-ks-er 'utf-8
 405    "Coding-system of UTF-8 using KS mapping with entity-reference."
 406    '(mnemonic "UTF8Kr"
 407               charset-g0 ucs-ks
 408               charset-g1 =>ucs-ks
 409               charset-g2 =>ucs
 410               use-entity-reference t))
 411
 412   (define-coding-system-alias 'utf-8 'utf-8-mcs)
 413   (define-coding-system-alias 'utf-8-er 'utf-8-mcs-er)
 414   )
 415
 416 (make-coding-system
 417  'euc-jisx0213 'iso2022
 418  "Coding-system of Japanese EUC based on JIS X 0213."
 419  '(charset-g0 ascii
 420    charset-g1 japanese-jisx0213-1
 421    charset-g2 katakana-jisx0201
 422    charset-g3 japanese-jisx0213-2
 423    short t
 424    mnemonic "Ja/EUC0213"
 425    ))
 426
 427 ;; initialize the coding categories to something semi-reasonable
 428 ;; so that the remaining Lisp files can contain extended characters.
 429 ;; (They will be in ISO-7 format)
 430 ;; #### This list needs to be synched with the ones in mule-cmds.el.
 431
 432 (if (featurep 'utf-2000)
 433     (set-coding-priority-list '(iso-7
 434                                 no-conversion
 435                                 utf-8
 436                                 iso-8-1
 437                                 iso-8-2
 438                                 iso-8-designate
 439                                 iso-lock-shift
 440                                 shift-jis
 441                                 big5
 442                                 ucs-4))
 443   (set-coding-priority-list '(iso-7
 444                               no-conversion
 445                               ;; utf-8
 446                               iso-8-1
 447                               iso-8-2
 448                               iso-8-designate
 449                               iso-lock-shift
 450                               shift-jis
 451                               big5
 452                               ;; ucs-4
 453                               )))
 454
 455 (set-coding-category-system 'iso-7 'iso-2022-7)
 456 (set-coding-category-system 'iso-8-designate 'ctext)
 457 (set-coding-category-system 'iso-8-1 'ctext)
 458 (set-coding-category-system 'iso-lock-shift 'iso-2022-lock)
 459 (set-coding-category-system 'no-conversion 'no-conversion)
 460
 461 (setq-default buffer-file-coding-system 'iso-2022-8)
 462
 463 ;;; mule-coding.el ends here