git.chise.org Git - chise/xemacs-chise.git-/blob - lisp/utf-2000/read-maps.el

   1 ;;; read-maps.el --- Read mapping-tables.
   2
   3 ;; Copyright (C) 2002 MORIOKA Tomohiko
   4
   5 ;; Author: MORIOKA Tomohiko <tomo@kanji.zinbun.kyoto-u.ac.jp>
   6 ;; Keywords: UTF-2000, UCS-4, character, CCS, multiscript, multilingual
   7
   8 ;; This file is part of XEmacs UTF-2000.
   9
  10 ;; XEmacs UTF-2000 is free software; you can redistribute it and/or
  11 ;; modify it under the terms of the GNU General Public License as
  12 ;; published by the Free Software Foundation; either version 2, or (at
  13 ;; your option) any later version.
  14
  15 ;; XEmacs UTF-2000 is distributed in the hope that it will be useful,
  16 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  17 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  18 ;; General Public License for more details.
  19
  20 ;; You should have received a copy of the GNU General Public License
  21 ;; along with XEmacs UTF-2000; see the file COPYING.  If not, write to
  22 ;; the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  23 ;; Boston, MA 02111-1307, USA.
  24
  25 ;;; Code:
  26
  27 (defun mapping-table-read-file (filename)
  28   (interactive "fMapping table : ")
  29   (with-temp-buffer
  30     (buffer-disable-undo)
  31     (insert-file-contents filename)
  32     (goto-char (point-min))
  33     (let (line ccs code ucs ucs-pat ucs-ccs ucs-code chr)
  34       (while (not (eobp))
  35         (cond ((looking-at "^J90-\\([0-9A-F][0-9A-F][0-9A-F][0-9A-F]\\)")
  36                (setq ccs 'japanese-jisx0208-1990
  37                      code (string-to-int (match-string 1) 16)
  38                      ucs-pat "\tJU[+-]\\([0-9A-F][0-9A-F][0-9A-F][0-9A-F]+\\)"
  39                      ucs-ccs 'ucs-jis)
  40                (goto-char (match-end 0))
  41                )
  42               ((looking-at "^JSP-\\([0-9A-F][0-9A-F][0-9A-F][0-9A-F]\\)")
  43                (setq ccs 'japanese-jisx0212
  44                      code (string-to-int (match-string 1) 16)
  45                      ucs-pat "\tJU[+-]\\([0-9A-F][0-9A-F][0-9A-F][0-9A-F]+\\)"
  46                      ucs-ccs 'ucs-jis)
  47                (goto-char (match-end 0))
  48                )
  49               ((looking-at "^JX1-\\([0-9A-F][0-9A-F][0-9A-F][0-9A-F]\\)")
  50                (setq ccs 'japanese-jisx0213-1
  51                      code (string-to-int (match-string 1) 16)
  52                      ucs-pat "\tJU[+-]\\([0-9A-F][0-9A-F][0-9A-F][0-9A-F]+\\)"
  53                      ucs-ccs 'ucs-jis)
  54                (goto-char (match-end 0))
  55                )
  56               ((looking-at "^JX2-\\([0-9A-F][0-9A-F][0-9A-F][0-9A-F]\\)")
  57                (setq ccs 'japanese-jisx0213-2
  58                      code (string-to-int (match-string 1) 16)
  59                      ucs-pat "\tJU[+-]\\([0-9A-F][0-9A-F][0-9A-F][0-9A-F]+\\)"
  60                      ucs-ccs 'ucs-jis)
  61                (goto-char (match-end 0))
  62                )
  63               ((looking-at "^C3-\\([0-9A-F][0-9A-F][0-9A-F][0-9A-F]\\)")
  64                (setq ccs 'chinese-cns11643-3
  65                      code (string-to-int (match-string 1) 16)
  66                      ucs-pat "\tCU[+-]\\([0-9A-F][0-9A-F][0-9A-F][0-9A-F]+\\)"
  67                      ucs-ccs 'ucs-cns)
  68                (goto-char (match-end 0))
  69                )
  70               ((looking-at "^C4-\\([0-9A-F][0-9A-F][0-9A-F][0-9A-F]\\)")
  71                (setq ccs 'chinese-cns11643-4
  72                      code (string-to-int (match-string 1) 16)
  73                      ucs-pat "\tCU[+-]\\([0-9A-F][0-9A-F][0-9A-F][0-9A-F]+\\)"
  74                      ucs-ccs 'ucs-cns)
  75                (goto-char (match-end 0))
  76                )
  77               ((looking-at "^B-\\([0-9A-F][0-9A-F][0-9A-F][0-9A-F]\\)")
  78                (setq ccs 'chinese-big5
  79                      code (string-to-int (match-string 1) 16)
  80                      ucs-pat "\tBU[+-]\\([0-9A-F][0-9A-F][0-9A-F][0-9A-F]+\\)"
  81                      ucs-ccs 'ucs-big5)
  82                (goto-char (match-end 0))
  83                )
  84               (t
  85                (setq ccs nil
  86                      code nil
  87                      ucs-pat nil
  88                      ucs-ccs nil)
  89                ))
  90         (setq ucs-code
  91               (if (looking-at ucs-pat)
  92                   (prog1
  93                       (string-to-int (match-string 1) 16)
  94                     (goto-char (match-end 0)))))
  95         (setq ucs
  96               (if (looking-at
  97                    "[ \t]*U[+-]\\([0-9A-F][0-9A-F][0-9A-F][0-9A-F]+\\)")
  98                   (string-to-int (match-string 1) 16)))
  99         (when (setq chr (decode-char ccs code))
 100           (unless (eq (get-char-attribute chr ccs) code)
 101             (put-char-attribute chr ccs code))
 102           (when (and ucs-code
 103                      (not (eq (or (get-char-attribute chr ucs-ccs)
 104                                   (get-char-attribute chr 'ucs)
 105                                   (get-char-attribute chr '=>ucs)
 106                                   (get-char-attribute chr '->ucs))
 107                               ucs-code)))
 108             (put-char-attribute chr ucs-ccs ucs-code))
 109           (when (and ucs
 110                      (not (eq (or (get-char-attribute chr 'ucs)
 111                                   (get-char-attribute chr '=>ucs)
 112                                   (get-char-attribute chr '->ucs))
 113                               ucs)))
 114             (put-char-attribute chr (if ucs-code
 115                                         '=>ucs
 116                                       ucs-ccs) ucs)))
 117         (forward-line)))))
 118
 119 (dolist (file '("J90-to-UCS.txt" "JSP-to-UCS.txt"
 120                 "JX1-to-UCS.txt" "JX2-to-UCS.txt"
 121                 ;; "C3-to-UCS.txt" ; "C4-to-UCS.txt"
 122                 "B-to-UCS.txt"))
 123   (mapping-table-read-file (expand-file-name file "../etc/char-data/")))
 124
 125 ;;; read-maps.el ends here