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) 1997,1999,2002,2003,2004,2005,2006,2008,2009 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  (let ((conf
 128         '(charset-g0 ascii
 129           charset-g2 t ;; unspecified but can be used later.
 130           seven t
 131           short t
 132           mnemonic "ISO7/SS"
 133           eol-type nil)))
 134    (if (featurep 'utf-2000)
 135        (list* 'ccs-priority-list
 136               '(ascii
 137                 =jis-x0208@1983 =jis-x0208@1978
 138                 latin-jisx0201)
 139               conf)
 140      conf)))
 141
 142 (make-coding-system
 143  'iso-2022-7bit 'iso2022
 144  "ISO 2022 based 7-bit encoding using only G0"
 145  '(charset-g0 ascii
 146    seven t
 147    short t
 148    mnemonic "ISO7"))
 149
 150 ;; compatibility for old XEmacsen
 151 (copy-coding-system 'iso-2022-7bit 'iso-2022-7)
 152
 153 (make-coding-system
 154  'iso-2022-8 'iso2022
 155  "ISO-2022 eight-bit coding system.  No single-shift or locking-shift."
 156  '(charset-g0 ascii
 157    charset-g1 latin-iso8859-1
 158    short t
 159    mnemonic "ISO8"
 160    ))
 161
 162 (make-coding-system
 163  'escape-quoted 'iso2022
 164  "ISO-2022 eight-bit coding system with escape quoting; used for .ELC files."
 165  '(charset-g0 ascii
 166    charset-g1 latin-iso8859-1
 167    eol-type lf
 168    escape-quoted t
 169    mnemonic "ESC/Quot"
 170    ))
 171
 172 (make-coding-system
 173  'iso-2022-lock 'iso2022
 174  "ISO-2022 coding system using Locking-Shift for 96-charset."
 175  '(charset-g0 ascii
 176    charset-g1 t ;; unspecified but can be used later.
 177    seven t
 178    lock-shift t
 179    mnemonic "ISO7/Lock"
 180    ))
 181
 182 (when (featurep 'utf-2000)
 183   (setq coded-charset-entity-reference-alist
 184         '(((ideograph-gt        . isolated)     "I-GT-" 5 d)
 185           ((=ruimoku-v6         . isolated)   "I-RUI6-" 4 X)
 186           ((chinese-big5-cdp    . isolated)    "I-CDP-" 4 X)
 187           ((ideograph-gt-k      . isolated)    "I-GT-K" 5 d)
 188           ((ideograph-cbeta     . isolated)      "I-CB" 5 d)
 189           ((ideograph-hanziku-1 . isolated)  "I-HZK01-" 4 X)
 190           ((ideograph-hanziku-2 . isolated)  "I-HZK02-" 4 X)
 191           ((ideograph-hanziku-3 . isolated)  "I-HZK03-" 4 X)
 192           ((ideograph-hanziku-4 . isolated)  "I-HZK04-" 4 X)
 193           ((ideograph-hanziku-5 . isolated)  "I-HZK05-" 4 X)
 194           ((ideograph-hanziku-6 . isolated)  "I-HZK06-" 4 X)
 195           ((ideograph-hanziku-7 . isolated)  "I-HZK07-" 4 X)
 196           ((ideograph-hanziku-8 . isolated)  "I-HZK08-" 4 X)
 197           ((ideograph-hanziku-9 . isolated)  "I-HZK09-" 4 X)
 198           ((ideograph-hanziku-10 . isolated) "I-HZK10-" 4 X)
 199           ((ideograph-hanziku-11 . isolated) "I-HZK11-" 4 X)
 200           ((ideograph-hanziku-12 . isolated) "I-HZK12-" 4 X)
 201           ((japanese-jisx0208-1990 . isolated) "I-J90-" 4 X)
 202           ((japanese-jisx0208   . isolated)    "I-J83-" 4 X)
 203           ((japanese-jisx0213-1 . isolated)    "I-JX1-" 4 X)
 204           ((japanese-jisx0213-2 . isolated)    "I-JX2-" 4 X)
 205           ((=jis-x0213-1@2004   . isolated)    "I-JX3-" 4 X)
 206           ((japanese-jisx0212   . isolated)    "I-JSP-" 4 X)
 207           ((japanese-jisx0208-1978 . isolated) "I-J78-" 4 X)
 208           ((chinese-gb2312      . isolated)     "I-G0-" 4 X)
 209           ((chinese-cns11643-1  . isolated)     "I-C1-" 4 X)
 210           ((chinese-cns11643-2  . isolated)     "I-C2-" 4 X)
 211           ((chinese-cns11643-3  . isolated)     "I-C3-" 4 X)
 212           ((chinese-cns11643-4  . isolated)     "I-C4-" 4 X)
 213           ((chinese-cns11643-5  . isolated)     "I-C5-" 4 X)
 214           ((chinese-cns11643-6  . isolated)     "I-C6-" 4 X)
 215           ((chinese-cns11643-7  . isolated)     "I-C7-" 4 X)
 216           ((korean-ksc5601      . isolated)     "I-K0-" 4 X)
 217           ((=iso-ir165          . isolated)    "I-EGB-" 4 X)
 218           ((ideograph-daikanwa  . isolated)      "I-M-" 5 d)
 219           ((chinese-big5        . isolated)      "I-B-" 4 X)
 220           ((latin-iso8859-1     . isolated) "I-LATIN1-" 2 X)
 221           ((latin-iso8859-2     . isolated) "I-LATIN2-" 2 X)
 222           ((latin-iso8859-3     . isolated) "I-LATIN3-" 2 X)
 223           ((latin-iso8859-4     . isolated) "I-LATIN4-" 2 X)
 224           ((cyrillic-iso8859-5  . isolated) "I-CYRILLIC-" 2 X)
 225           ((greek-iso8859-7     . isolated)  "I-GREEK-" 2 X)
 226           ((hebrew-iso8859-8    . isolated) "I-HEBREW-" 2 X)
 227           ((latin-iso8859-9     . isolated) "I-LATIN5-" 2 X)
 228           ((latin-jisx0201      . isolated) "I-LATINJ-" 2 X)
 229           ((katakana-jisx0201   . isolated) "I-KATAKANA-" 2 X)
 230           ((latin-tcvn5712      . isolated) "I-VSCII2-" 2 X)
 231           ;; ((latin-viscii     . isolated) "I-VISCII-" 2 X)
 232           ((latin-viscii-upper  . isolated) "I-MULE-VIET-U-" 2 X)
 233           ((latin-viscii-lower  . isolated) "I-MULE-VIET-L-" 2 X)
 234           ((thai-tis620         . isolated)   "I-THAI-" 2 X)
 235           ((lao                 . isolated) "I-MULE-LAO-" 2 X)
 236           ((arabic-1-column     . isolated) "I-MULE-ARB1-" 2 X)
 237           ((arabic-2-column     . isolated) "I-MULE-ARB2-" 2 X)
 238           ((arabic-digit        . isolated) "I-MULE-ARBD-" 2 X)
 239           ((ipa                 . isolated) "I-MULE-IPA-" 2 X)
 240           ((china3-jef          . isolated)    "I-JC3-" 4 X)
 241           ( ideograph-gt                          "GT-" 5 d)
 242           (<=>gt                                 "aGT-" 5 d)
 243           ( =zinbun-oracle                       "ZOB-" 4 d)
 244           ( =ruimoku-v6                         "RUI6-" 4 X)
 245           ( chinese-big5-cdp                     "CDP-" 4 X)
 246           ( ideograph-daikanwa                     "M-" 5 d)
 247           ( ideograph-gt-k                       "GT-K" 5 d)
 248           ( ideograph-cbeta                        "CB" 5 d)
 249           ( ideograph-hanziku-1                "HZK01-" 4 X)
 250           ( ideograph-hanziku-2                "HZK02-" 4 X)
 251           ( ideograph-hanziku-3                "HZK03-" 4 X)
 252           ( ideograph-hanziku-4                "HZK04-" 4 X)
 253           ( ideograph-hanziku-5                "HZK05-" 4 X)
 254           ( ideograph-hanziku-6                "HZK06-" 4 X)
 255           ( ideograph-hanziku-7                "HZK07-" 4 X)
 256           ( ideograph-hanziku-8                "HZK08-" 4 X)
 257           ( ideograph-hanziku-9                "HZK09-" 4 X)
 258           ( ideograph-hanziku-10               "HZK10-" 4 X)
 259           ( ideograph-hanziku-11               "HZK11-" 4 X)
 260           ( ideograph-hanziku-12               "HZK12-" 4 X)
 261           ((ideograph-hanziku-1 . isolated)   "I-HZK1-" 4 X)
 262           ( ideograph-hanziku-1                 "HZK1-" 4 X)
 263           ((ideograph-hanziku-2 . isolated)   "I-HZK2-" 4 X)
 264           ( ideograph-hanziku-2                 "HZK2-" 4 X)
 265           ( japanese-jisx0208-1990               "J90-" 4 X)
 266           ( japanese-jisx0208                    "J83-" 4 X)
 267           ( japanese-jisx0213-1                  "JX1-" 4 X)
 268           ( japanese-jisx0213-2                  "JX2-" 4 X)
 269           ( =jis-x0213-1@2004                    "JX3-" 4 X)
 270           ( japanese-jisx0212                    "JSP-" 4 X)
 271           ( japanese-jisx0208-1978               "J78-" 4 X)
 272           ( chinese-cns11643-1                    "C1-" 4 X)
 273           ( chinese-cns11643-2                    "C2-" 4 X)
 274           ( chinese-cns11643-3                    "C3-" 4 X)
 275           ( chinese-cns11643-4                    "C4-" 4 X)
 276           ( chinese-cns11643-5                    "C5-" 4 X)
 277           ( chinese-cns11643-6                    "C6-" 4 X)
 278           ( chinese-cns11643-7                    "C7-" 4 X)
 279           ( korean-ksc5601                        "K0-" 4 X)
 280           ( chinese-gb2312                        "G0-" 4 X)
 281           ( =iso-ir165                           "EGB-" 4 X)
 282           ( latin-iso8859-1                   "LATIN1-" 2 X)
 283           ( latin-iso8859-2                   "LATIN2-" 2 X)
 284           ( latin-iso8859-3                   "LATIN3-" 2 X)
 285           ( latin-iso8859-4                   "LATIN4-" 2 X)
 286           ( cyrillic-iso8859-5              "CYRILLIC-" 2 X)
 287           ( greek-iso8859-7                    "GREEK-" 2 X)
 288           ( hebrew-iso8859-8                  "HEBREW-" 2 X)
 289           ( latin-iso8859-9                   "LATIN5-" 2 X)
 290           ( latin-jisx0201                    "LATINJ-" 2 X)
 291           ( katakana-jisx0201               "KATAKANA-" 2 X)
 292           ( latin-tcvn5712                    "VSCII2-" 2 X)
 293           ( latin-viscii                      "VISCII-" 2 X)
 294           ( latin-viscii-upper           "MULE-VIET-U-" 2 X)
 295           ( latin-viscii-lower           "MULE-VIET-L-" 2 X)
 296           ( thai-tis620                         "THAI-" 2 X)
 297           ( lao                             "MULE-LAO-" 2 X)
 298           ( ethiopic                      "MULE-ETHIO-" 4 X)
 299           ( arabic-1-column                "MULE-ARB1-" 2 X)
 300           ( arabic-2-column                "MULE-ARB2-" 2 X)
 301           ( arabic-digit                   "MULE-ARBD-" 2 X)
 302           ( ipa                             "MULE-IPA-" 2 X)
 303           ( china3-jef                           "JC3-" 4 X)
 304           ( chinese-big5                           "B-" 4 X)
 305           ( chinese-big5                          "C0-" 4 X)
 306           ( =ucs@iso                               "U-" 8 X)
 307           ( =ucs@unicode                          "UU+" 4 X)
 308           ( =ucs@unicode                          "UU-" 8 X)
 309           ( =ucs@iso                               "U+" 4 X)
 310           ( =ucs@gb                               "GU+" 4 X)
 311           ( =ucs@gb                               "GU-" 8 X)
 312           ( =ucs@jis                              "JU+" 4 X)
 313           ( =ucs@jis                              "JU-" 8 X)
 314           ( =ucs@cns                              "CU+" 4 X)
 315           ( =ucs@cns                              "CU-" 8 X)
 316           ( =ucs@ks                               "KU+" 4 X)
 317           ( =ucs@ks                               "KU-" 8 X)
 318           ( =jis-x0208@1997                      "J97-" 4 X)
 319           ( =jis-x0208@1978/i1                 "J78i1-" 4 X)
 320           ( =shinjigen@rev                      "SJG2-" 4 d)
 321           ( =shinjigen@1ed                      "SJG1-" 4 d)
 322           ))
 323
 324   (make-coding-system
 325    'utf-8-mcs-er 'utf-8
 326    "Coding-system of UTF-8 with entity-reference."
 327    '(mnemonic "MTF8r" use-entity-reference t))
 328
 329   (make-coding-system
 330    'utf-8-gb 'utf-8
 331    "Coding-system of UTF-8 using GB mapping."
 332    '(mnemonic "UTF8G"
 333               charset-g0 ucs-gb
 334               charset-g1 =>ucs-gb
 335               charset-g2 =>ucs))
 336
 337   (make-coding-system
 338    'utf-8-gb-er 'utf-8
 339    "Coding-system of UTF-8 using GB mapping with entity-reference."
 340    '(mnemonic "UTF8Gr"
 341               charset-g0 ucs-gb
 342               charset-g1 =>ucs-gb
 343               charset-g2 =>ucs
 344               use-entity-reference t))
 345
 346   (make-coding-system
 347    'utf-8-cns 'utf-8
 348    "Coding-system of UTF-8 using CNS mapping."
 349    '(mnemonic "UTF8C"
 350               charset-g0 ucs-cns
 351               charset-g1 =>ucs-cns
 352               charset-g2 =>ucs))
 353
 354   (make-coding-system
 355    'utf-8-cns-er 'utf-8
 356    "Coding-system of UTF-8 using CNS mapping with entity-reference."
 357    '(mnemonic "UTF8Cr"
 358               charset-g0 ucs-cns
 359               charset-g1 =>ucs-cns
 360               charset-g2 =>ucs
 361               use-entity-reference t))
 362
 363   (make-coding-system
 364    'utf-8-big5 'utf-8
 365    "Coding-system of UTF-8 using Big5 mapping."
 366    '(mnemonic "UTF8B"
 367               charset-g0 ucs-big5
 368               charset-g1 =>ucs-big5
 369               charset-g2 =>ucs))
 370
 371   (make-coding-system
 372    'utf-8-big5-er 'utf-8
 373    "Coding-system of UTF-8 using Big5 mapping with entity-reference."
 374    '(mnemonic "UTF8Br"
 375               charset-g0 ucs-big5
 376               charset-g1 =>ucs-big5
 377               charset-g2 =>ucs
 378               use-entity-reference t))
 379
 380   (make-coding-system
 381    'utf-8-jis 'utf-8
 382    "Coding-system of UTF-8 using JIS mapping."
 383    '(mnemonic "UTF8J"
 384               charset-g0 =ucs@jis-2000
 385               charset-g1 =>ucs-jis
 386               charset-g2 =>ucs))
 387
 388   (make-coding-system
 389    'utf-8-jis-er 'utf-8
 390    "Coding-system of UTF-8 using JIS mapping with entity-reference."
 391    '(mnemonic "UTF8Jr"
 392               charset-g0 =ucs@jis-2000
 393               charset-g1 =>ucs-jis
 394               charset-g2 =>ucs
 395               use-entity-reference t))
 396
 397   (make-coding-system
 398    'utf-8-jp 'utf-8
 399    "Coding-system of UTF-8 for common glyphs used in Japan."
 400    '(mnemonic "UTF8J"
 401               charset-g0 =ucs@jp
 402               charset-g1 =>ucs-jis
 403               charset-g2 =>ucs))
 404
 405   (make-coding-system
 406    'utf-8-jp-er 'utf-8
 407    "Coding-system of UTF-8 using =ucs@jp mapping with entity-reference."
 408    '(mnemonic "UTF8Jr"
 409               charset-g0 =ucs@jp
 410               charset-g1 =>ucs-jis
 411               charset-g2 =>ucs
 412               use-entity-reference t))
 413
 414   (make-coding-system
 415    'utf-8-ks 'utf-8
 416    "Coding-system of UTF-8 using KS mapping."
 417    '(mnemonic "UTF8K"
 418               charset-g0 ucs-ks
 419               charset-g1 =>ucs-ks
 420               charset-g2 =>ucs))
 421
 422   (make-coding-system
 423    'utf-8-ks-er 'utf-8
 424    "Coding-system of UTF-8 using KS mapping with entity-reference."
 425    '(mnemonic "UTF8Kr"
 426               charset-g0 ucs-ks
 427               charset-g1 =>ucs-ks
 428               charset-g2 =>ucs
 429               use-entity-reference t))
 430
 431   (define-coding-system-alias 'utf-8 'utf-8-mcs)
 432   (define-coding-system-alias 'utf-8-er 'utf-8-mcs-er)
 433   )
 434
 435 (make-coding-system
 436  'euc-jisx0213 'iso2022
 437  "Coding-system of Japanese EUC based on JIS X 0213."
 438  '(charset-g0 ascii
 439    charset-g1 japanese-jisx0213-1
 440    charset-g2 katakana-jisx0201
 441    charset-g3 japanese-jisx0213-2
 442    short t
 443    mnemonic "Ja/EUC0213"
 444    ))
 445
 446 ;; initialize the coding categories to something semi-reasonable
 447 ;; so that the remaining Lisp files can contain extended characters.
 448 ;; (They will be in ISO-7 format)
 449 ;; #### This list needs to be synched with the ones in mule-cmds.el.
 450
 451 (if (featurep 'utf-2000)
 452     (set-coding-priority-list '(iso-7
 453                                 no-conversion
 454                                 utf-8
 455                                 iso-8-1
 456                                 iso-8-2
 457                                 iso-8-designate
 458                                 iso-lock-shift
 459                                 shift-jis
 460                                 big5
 461                                 ucs-4))
 462   (set-coding-priority-list '(iso-7
 463                               no-conversion
 464                               ;; utf-8
 465                               iso-8-1
 466                               iso-8-2
 467                               iso-8-designate
 468                               iso-lock-shift
 469                               shift-jis
 470                               big5
 471                               ;; ucs-4
 472                               )))
 473
 474 (set-coding-category-system 'iso-7 'iso-2022-7)
 475 (set-coding-category-system 'iso-8-designate 'ctext)
 476 (set-coding-category-system 'iso-8-1 'ctext)
 477 (set-coding-category-system 'iso-lock-shift 'iso-2022-lock)
 478 (set-coding-category-system 'no-conversion 'no-conversion)
 479
 480 (setq-default buffer-file-coding-system 'iso-2022-8)
 481
 482 ;;; mule-coding.el ends here