git.chise.org Git - chise/xemacs-chise.git-/blob - lisp/utf-2000/char-db-util.el

   1 ;;; char-db-util.el --- Character Database utility -*- coding: utf-8-er; -*-
   2
   3 ;; Copyright (C) 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006,
   4 ;;   2007, 2008, 2009, 2010 MORIOKA Tomohiko.
   5
   6 ;; Author: MORIOKA Tomohiko <tomo@kanji.zinbun.kyoto-u.ac.jp>
   7 ;; Keywords: CHISE, Character Database, ISO/IEC 10646, UCS, Unicode, MULE.
   8
   9 ;; This file is part of XEmacs CHISE.
  10
  11 ;; XEmacs CHISE is free software; you can redistribute it and/or
  12 ;; modify it under the terms of the GNU General Public License as
  13 ;; published by the Free Software Foundation; either version 2, or (at
  14 ;; your option) any later version.
  15
  16 ;; XEmacs CHISE is distributed in the hope that it will be useful,
  17 ;; but 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 CHISE; see the file COPYING.  If not, write to
  23 ;; the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  24 ;; Boston, MA 02111-1307, USA.
  25
  26 ;;; Code:
  27
  28 (require 'alist)
  29 (require 'chise-subr)
  30
  31 (defconst unidata-normative-category-alist
  32   '(("Lu" letter        uppercase)
  33     ("Ll" letter        lowercase)
  34     ("Lt" letter        titlecase)
  35     ("Mn" mark          non-spacing)
  36     ("Mc" mark          spacing-combining)
  37     ("Me" mark          enclosing)
  38     ("Nd" number        decimal-digit)
  39     ("Nl" number        letter)
  40     ("No" number        other)
  41     ("Zs" separator     space)
  42     ("Zl" separator     line)
  43     ("Zp" separator     paragraph)
  44     ("Cc" other         control)
  45     ("Cf" other         format)
  46     ("Cs" other         surrogate)
  47     ("Co" other         private-use)
  48     ("Cn" other         not-assigned)))
  49
  50 (defconst unidata-informative-category-alist
  51   '(("Lm" letter        modifier)
  52     ("Lo" letter        other)
  53     ("Pc" punctuation   connector)
  54     ("Pd" punctuation   dash)
  55     ("Ps" punctuation   open)
  56     ("Pe" punctuation   close)
  57     ("Pi" punctuation   initial-quote)
  58     ("Pf" punctuation   final-quote)
  59     ("Po" punctuation   other)
  60     ("Sm" symbol        math)
  61     ("Sc" symbol        currency)
  62     ("Sk" symbol        modifier)
  63     ("So" symbol        other)
  64     ))
  65
  66 (defconst ideographic-radicals
  67   (let ((v (make-vector 215 nil))
  68         (i 1))
  69     (while (< i 215)
  70       (aset v i (decode-char '=ucs (+ #x2EFF i)))
  71       (setq i (1+ i)))
  72     v))
  73
  74 ;;;###autoload
  75 (defun ideographic-radical (number)
  76   "Return character corresponding with Kangxi-radical number."
  77   (aref ideographic-radicals number))
  78
  79 (defconst shuowen-radicals
  80   [?一 ?上 ?示 ?三 ?王 ?玉 ?玨 ?气 ?士 ?丨 ; 010
  81    ?屮 ?艸 ?蓐 ?茻 ?小 ?八 ?釆 ?半 ?牛 ?犛 ; 020
  82    ?告 ?口 ?凵 ?吅 ?哭 ?走 ?止 ?癶 ?步 ?此 ; 030
  83    ?正 ?是 ?辵 ?彳 ?廴 ?㢟 ?行 ?齒 ?牙 ?足 ; 040
  84    ?疋 ?品 ?龠 ?冊 ?㗊 ?舌 ?干 ?谷 ?只 ?㕯 ; 050
  85    ?句 ?丩 ?古 ?十 ?卅 ?言 ?誩 ?音 ?䇂 ?丵 ; 060
  86    ?菐 ?𠬞 ?廾 ?共 ?異 ?舁 ?𦥑 ?䢅 ?爨 ?革 ; 070
  87    ?鬲 ?䰜 ?爪 ?𠃨 ?鬥 ?又 ?𠂇 ?㕜 ?支 ?𦘒 ; 080
  88    ?聿 ?畫 ?隶 ?臤 ?臣 ?殳 ?殺 ?𠘧 ?寸 ?皮 ; 090
  89    ?㼱 ?攴 ?敎 ?卜 ?用 ?爻 ?㸚 ?𥄎 ?目 ?䀠 ; 100
  90    ?眉 ?盾 ?自 ?白 ?鼻 ?皕 ?習 ?羽 ?隹 ?奞 ; 110
  91    ?萑 ?𦫳 ?苜 ?羊 ?羴 ?瞿 ?雔 ?雥 ?鳥 ?烏 ; 120
  92    ?𠦒 ?冓 ?幺 ?𢆶 ?叀 ?玄 ?予 ?放 ?𠬪 ?𣦼 ; 130
  93    ?歺 ?死 ?冎 ?骨 ?肉 ?筋 ?刀 ?刃 ?㓞 ?丰 ; 140
  94    ?耒 ?𧢲 ?竹 ?箕 ?丌 ?左 ?工 ?㠭 ?巫 ?甘 ; 150
  95    ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?旨 ?喜 ?壴 ; 160
  96    ?鼓 ?豈 ?豆 ?豊 ?豐 ?䖒 ?虍 ?虎 ?虤 ?皿 ; 170
  97    ?𠙴 ?去 ?血 ?丶 ?丹 ?青 ?井 ?皀 ?鬯 ?食 ; 180
  98    ?亼 ?會 ?倉 ?入 ?缶 ?矢 ?高 ?冂 ?𩫏 ?京 ; 190
  99    ?亯 ?𣆪 ?畗 ?㐭 ?嗇 ?來 ?麥 ?夊 ?舛 ?䑞 ; 200
 100    ?韋 ?弟 ?夂 ?久 ?桀 ?木 ?東 ?林 ?才 ?叒 ; 210
 101    ?之 ?帀 ?出 ?𣎵 ?生 ?乇 ?𠂹 ?𠌶 ?華 ?𥝌 ; 220
 102    ?稽 ?巢 ?桼 ?束 ?㯻 ?囗 ?員 ?貝 ?邑 ?𨛜 ; 230
 103    ?日 ?旦 ?倝 ?㫃 ?冥 ?晶 ?月 ?有 ?明 ?囧 ; 240
 104    ?夕 ?多 ?毌 ?𢎘 ?𣐺 ?卣 ?齊 ?朿 ?片 ?鼎 ; 250
 105    ?克 ?彔 ?禾 ?秝 ?黍 ?香 ?米 ?毇 ?臼 ?凶 ; 260
 106    ?𣎳 ?林 ?麻 ?尗 ?耑 ?韭 ?瓜 ?瓠 ?宀 ?宮 ; 270
 107    ?呂 ?穴 ?㝱 ?𤕫 ?冖 ?𠔼 ?冃 ?㒳 ?网 ?襾 ; 280
 108    ?巾 ?巿 ?帛 ?白 ?㡀 ?黹 ?人 ?𠤎 ?匕 ?从 ; 290
 109    ])
 110
 111 (defun shuowen-radical (number)
 112   (aref shuowen-radicals (1- number)))
 113
 114 (defvar char-db-file-coding-system 'utf-8-mcs-er)
 115
 116 (defvar char-db-feature-domains
 117   '(ucs ucs/compat daikanwa cns gt jis jis/alt jis/a jis/b
 118         jis-x0212 jis-x0213 cdp shinjigen misc unknown))
 119
 120 (defvar char-db-ignored-attributes '(ideographic-products))
 121
 122 (defvar char-db-coded-charset-priority-list
 123   '(ascii
 124     control-1
 125     latin-iso8859-1
 126     latin-iso8859-2
 127     latin-iso8859-3
 128     latin-iso8859-4
 129     latin-iso8859-9
 130     latin-jisx0201
 131     cyrillic-iso8859-5
 132     greek-iso8859-7
 133     thai-tis620
 134     =jis-x0208
 135     =jis-x0208@1978
 136     =jis-x0208@1983
 137     japanese-jisx0212
 138     chinese-gb2312
 139     =jis-x0208@1990
 140     chinese-cns11643-1
 141     chinese-cns11643-2
 142     chinese-cns11643-3
 143     chinese-cns11643-4
 144     chinese-cns11643-5
 145     chinese-cns11643-6
 146     chinese-cns11643-7
 147     =jis-x0213-1
 148     =jis-x0213-1@2000
 149     =jis-x0213-1@2004
 150     =jis-x0213-2
 151     korean-ksc5601
 152     chinese-isoir165
 153     katakana-jisx0201
 154     hebrew-iso8859-8
 155     chinese-gb12345
 156     latin-viscii
 157     ethiopic-ucs
 158     =big5-cdp
 159     =gt
 160     =>>gt
 161     =>gt
 162     ideograph-daikanwa-2
 163     ideograph-daikanwa
 164     =cbeta
 165     ideograph-hanziku-1
 166     ideograph-hanziku-2
 167     ideograph-hanziku-3
 168     ideograph-hanziku-4
 169     ideograph-hanziku-5
 170     ideograph-hanziku-6
 171     ideograph-hanziku-7
 172     ideograph-hanziku-8
 173     ideograph-hanziku-9
 174     ideograph-hanziku-10
 175     ideograph-hanziku-11
 176     ideograph-hanziku-12
 177     =gt-k
 178     =ucs@iso
 179     =ucs@unicode
 180     =>>ucs@unicode
 181     =>>jis-x0208
 182     =>>jis-x0213-1
 183     =>>jis-x0213-1@2000
 184     =>>jis-x0213-1@2004
 185     =>>jis-x0213-2
 186     =>>jis-x0208@1978
 187     =>jis-x0208
 188     =>jis-x0208@1997
 189     =>jis-x0213-1
 190     =>jis-x0213-1@2000
 191     =>jis-x0213-1@2004
 192     =>jis-x0213-2
 193     =big5
 194     =big5-eten
 195     =zinbun-oracle
 196     =>zinbun-oracle
 197     =ruimoku-v6
 198     =jef-china3
 199     =shinjigen))
 200
 201
 202 ;;; @ char-db formatters
 203 ;;;
 204
 205 (defun char-db-make-char-spec (char)
 206   (let (ret char-spec)
 207     (cond ((characterp char)
 208            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 209                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 210                   (setq char-spec (list (cons '=ucs ret)))
 211                   (cond ((setq ret (get-char-attribute char 'name))
 212                          (setq char-spec (cons (cons 'name ret) char-spec))
 213                          )
 214                         ((setq ret (get-char-attribute char 'name*))
 215                          (setq char-spec (cons (cons 'name* ret) char-spec))
 216                          ))
 217                   )
 218                  ((setq ret
 219                         (catch 'tag
 220                           (let ((rest char-db-coded-charset-priority-list)
 221                                 ccs)
 222                             (while rest
 223                               (setq ccs (charset-name
 224                                          (find-charset (car rest))))
 225                               (if (setq ret
 226                                         (encode-char char ccs
 227                                                      'defined-only))
 228                                   (throw 'tag (cons ccs ret)))
 229                               (setq rest (cdr rest))))))
 230                   (setq char-spec (list ret))
 231                   (dolist (ccs (delq (car ret) (charset-list)))
 232                     (if (and (or (charset-iso-final-char ccs)
 233                                  (memq ccs
 234                                        '(=daikanwa
 235                                          =daikanwa@rev2
 236                                          ;; =gt-k
 237                                          =jis-x0208@1997
 238                                          ))
 239                                  (string-match "=ucs@" (symbol-name ccs)))
 240                              (setq ccs (charset-name ccs))
 241                              (null (assq ccs char-spec))
 242                              (setq ret (encode-char char ccs 'defined-only)))
 243                         (setq char-spec (cons (cons ccs ret) char-spec))))
 244                   (if (null char-spec)
 245                       (setq char-spec (split-char char)))
 246                   (cond ((setq ret (get-char-attribute char 'name))
 247                          (setq char-spec (cons (cons 'name ret) char-spec))
 248                          )
 249                         ((setq ret (get-char-attribute char 'name*))
 250                          (setq char-spec (cons (cons 'name* ret) char-spec))
 251                          ))
 252                   )
 253                  ((setq ret (get-char-attribute
 254                              char 'ideographic-combination))
 255                   (setq char-spec
 256                         (cons (cons 'ideographic-combination ret)
 257                               char-spec))
 258                   ))
 259            char-spec)
 260           ((consp char)
 261            char))))
 262
 263 (defun char-db-insert-char-spec (char &optional readable column
 264                                       required-features)
 265   (unless column
 266     (setq column (current-column)))
 267   (let (char-spec temp-char)
 268     (setq char-spec (char-db-make-char-spec char))
 269     (unless (or (characterp char) ; char
 270                 (condition-case nil
 271                     (setq char (find-char char-spec))
 272                   (error nil)))
 273       ;; define temporary character
 274       ;;   Current implementation is dirty.
 275       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 276                                          char-spec)))
 277       (remove-char-attribute temp-char 'ideograph-daikanwa)
 278       (setq char temp-char))
 279     (insert-char-attributes char
 280                             readable
 281                             (union (mapcar #'car char-spec)
 282                                    required-features))
 283     (when temp-char
 284       ;; undefine temporary character
 285       ;;   Current implementation is dirty.
 286       (setq char-spec (char-attribute-alist temp-char))
 287       (while char-spec
 288         (remove-char-attribute temp-char (car (car char-spec)))
 289         (setq char-spec (cdr char-spec))))))
 290
 291 (defun char-db-insert-alist (alist &optional readable column)
 292   (unless column
 293     (setq column (current-column)))
 294   (let ((line-breaking
 295          (concat "\n" (make-string (1+ column) ?\ )))
 296         name value
 297         ret al ; cal
 298         key
 299         lbs cell rest separator)
 300     (insert "(")
 301     (while alist
 302       (setq name (car (car alist))
 303             value (cdr (car alist)))
 304       (cond ((eq name 'char)
 305              (insert "(char . ")
 306              (if (setq ret (condition-case nil
 307                                (find-char value)
 308                              (error nil)))
 309                  (progn
 310                    (setq al nil
 311                          ;; cal nil
 312                          )
 313                    (while value
 314                      (setq key (car (car value)))
 315                      ;; (if (find-charset key)
 316                      ;;     (setq cal (cons key cal))
 317                      (setq al (cons key al))
 318                      ;; )
 319                      (setq value (cdr value)))
 320                    (insert-char-attributes ret
 321                                            readable
 322                                            (or al 'none) ; cal
 323                                            ))
 324                (insert (prin1-to-string value)))
 325              (insert ")")
 326              (insert line-breaking))
 327             ((consp value)
 328              (insert (format "(%-18s " name))
 329              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 330              (while (consp value)
 331                (setq cell (car value))
 332                (if (and (consp cell)
 333                         (consp (car cell))
 334                         (setq ret (condition-case nil
 335                                       (find-char cell)
 336                                     (error nil)))
 337                         )
 338                    (progn
 339                      (setq rest cell
 340                            al nil
 341                            ;; cal nil
 342                            )
 343                      (while rest
 344                        (setq key (car (car rest)))
 345                        ;; (if (find-charset key)
 346                        ;;     (setq cal (cons key cal))
 347                        (setq al (cons key al))
 348                        ;; )
 349                        (setq rest (cdr rest)))
 350                      (if separator
 351                          (insert lbs))
 352                      (insert-char-attributes ret
 353                                              readable
 354                                              al ; cal
 355                                              )
 356                      (setq separator lbs))
 357                  (if separator
 358                      (insert separator))
 359                  (insert (prin1-to-string cell))
 360                  (setq separator " "))
 361                (setq value (cdr value)))
 362              (insert ")")
 363              (insert line-breaking))
 364             (t
 365              (insert (format "(%-18s . %S)%s"
 366                              name value
 367                              line-breaking))))
 368       (setq alist (cdr alist))))
 369   (insert ")"))
 370
 371 (defun char-db-insert-char-reference (plist &optional readable column)
 372   (unless column
 373     (setq column (current-column)))
 374   (let ((line-breaking
 375          (concat "\n" (make-string (1+ column) ?\ )))
 376         (separator "")
 377         name value)
 378     (insert "(")
 379     (while plist
 380       (setq name (pop plist))
 381       (setq value (pop plist))
 382       (cond ((eq name :char)
 383              (insert separator)
 384              (insert ":char\t")
 385              (cond ((numberp value)
 386                     (setq value (decode-char '=ucs value)))
 387                    ;; ((consp value)
 388                    ;;  (setq value (or (find-char value)
 389                    ;;                  value)))
 390                    )
 391              (char-db-insert-char-spec value readable)
 392              (insert line-breaking)
 393              (setq separator ""))
 394             ((eq name :radical)
 395              (insert (format "%s%s\t%d ; %c%s"
 396                              separator
 397                              name value
 398                              (ideographic-radical value)
 399                              line-breaking))
 400              (setq separator ""))
 401             (t
 402              (insert (format "%s%s\t%S" separator name value))
 403              (setq separator line-breaking)))
 404       ))
 405   (insert ")"))
 406
 407 (defun char-db-decode-isolated-char (ccs code-point)
 408   (let (ret)
 409     (setq ret
 410           (cond ((eq ccs 'arabic-iso8859-6)
 411                  (decode-char ccs code-point))
 412                 ((and (memq ccs '(=gt-pj-1
 413                                   =gt-pj-2
 414                                   =gt-pj-3
 415                                   =gt-pj-4
 416                                   =gt-pj-5
 417                                   =gt-pj-6
 418                                   =gt-pj-7
 419                                   =gt-pj-8
 420                                   =gt-pj-9
 421                                   =gt-pj-10
 422                                   =gt-pj-11))
 423                       (setq ret (decode-char ccs code-point))
 424                       (setq ret (encode-char ret '=gt 'defined-only)))
 425                  (decode-builtin-char '=gt ret))
 426                 (t
 427                  (decode-builtin-char ccs code-point))))
 428     (cond ((and (<= 0 (char-int ret))
 429                 (<= (char-int ret) #x1F))
 430            (decode-char '=ucs (+ #x2400 (char-int ret))))
 431           ((= (char-int ret) #x7F)
 432            ?\u2421)
 433           (t ret))))
 434
 435 (defvar char-db-convert-obsolete-format t)
 436
 437 (defun char-db-insert-ccs-feature (name value line-breaking)
 438   (insert
 439    (format
 440     (cond ((memq name '(=shinjigen
 441                         =shinjigen@1ed
 442                         =shinjigen@rev =shinjigen/+p@rev))
 443            "(%-18s .  %04d)\t; %c")
 444           ((eq name '=shinjigen@1ed/24pr)
 445            "(%-18s . %04d)\t; %c")
 446           ((or (memq name '(=daikanwa
 447                             =daikanwa@rev1 =daikanwa@rev2
 448                             =gt =>>gt =>gt =gt-k =cbeta
 449                             =zinbun-oracle =>zinbun-oracle))
 450                (string-match "^=adobe-" (symbol-name name)))
 451            "(%-18s . %05d)\t; %c")
 452           ((eq name 'mojikyo)
 453            "(%-18s . %06d)\t; %c")
 454           ((>= (charset-dimension name) 2)
 455            "(%-18s . #x%04X)\t; %c")
 456           (t
 457            "(%-18s . #x%02X)\t; %c"))
 458     name
 459     (if (= (charset-iso-graphic-plane name) 1)
 460         (logior value
 461                 (cond ((= (charset-dimension name) 1)
 462                        #x80)
 463                       ((= (charset-dimension name) 2)
 464                        #x8080)
 465                       ((= (charset-dimension name) 3)
 466                        #x808080)
 467                       (t 0)))
 468       value)
 469     (char-db-decode-isolated-char name value)))
 470   (if (and (= (charset-chars name) 94)
 471            (= (charset-dimension name) 2))
 472       (insert (format " [%02d-%02d]"
 473                       (- (lsh value -8) 32)
 474                       (- (logand value 255) 32))))
 475   (insert line-breaking))
 476
 477 (defun char-db-insert-relation-feature (char name value line-breaking
 478                                              ccss readable)
 479   (insert (format "(%-18s%s " name line-breaking))
 480   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 481         separator cell sources required-features
 482         ret)
 483     (while (consp value)
 484       (setq cell (car value))
 485       (if (integerp cell)
 486           (setq cell (decode-char '=ucs cell)))
 487       (cond
 488        ((eq name '->subsumptive)
 489         (when (or (not (some (lambda (atr)
 490                                (get-char-attribute cell atr))
 491                              char-db-ignored-attributes))
 492                   (some (lambda (ccs)
 493                           (encode-char cell ccs 'defined-only))
 494                         ccss))
 495           (if separator
 496               (insert lbs))
 497           (let ((char-db-ignored-attributes
 498                  (cons '<-subsumptive
 499                        char-db-ignored-attributes)))
 500             (insert-char-attributes cell readable))
 501           (setq separator lbs))
 502         )
 503        ((characterp cell)
 504         (setq sources
 505               (get-char-attribute
 506                char (intern (format "%s*sources" name))))
 507         (setq required-features nil)
 508         (dolist (source sources)
 509           (cond
 510            ((memq source '(JP
 511                            JP/Jouyou
 512                            shinjigen shinjigen@1ed shinjigen@rev))
 513             (setq required-features
 514                   (union required-features
 515                          '(=jis-x0208
 516                            =jis-x0208@1990
 517                            =jis-x0213-1-2000
 518                            =jis-x0213-2-2000
 519                            =jis-x0212
 520                            =jis-x0208@1983
 521                            =jis-x0208@1978
 522                            =shinjigen))))
 523            ((eq source 'CN)
 524             (setq required-features
 525                   (union required-features
 526                          '(=gb2312
 527                            =gb12345
 528                            =iso-ir165)))))
 529           (cond
 530            ((find-charset (setq ret (intern (format "=%s" source))))
 531             (setq required-features
 532                   (cons ret required-features)))
 533            (t (setq required-features
 534                     (cons source required-features)))))
 535         (cond ((string-match "@JP" (symbol-name name))
 536                (setq required-features
 537                      (union required-features
 538                             '(=jis-x0208
 539                               =jis-x0208@1990
 540                               =jis-x0213-1-2000
 541                               =jis-x0213-2-2000
 542                               =jis-x0212
 543                               =jis-x0208@1983
 544                               =jis-x0208@1978))))
 545               ((string-match "@CN" (symbol-name name))
 546                (setq required-features
 547                      (union required-features
 548                             '(=gb2312
 549                               =gb12345
 550                               =iso-ir165)))))
 551         (if separator
 552             (insert lbs))
 553         (if readable
 554             (insert (format "%S" cell))
 555           (char-db-insert-char-spec cell readable
 556                                     nil
 557                                     required-features))
 558         (setq separator lbs))
 559        ((consp cell)
 560         (if separator
 561             (insert lbs))
 562         (if (consp (car cell))
 563             (char-db-insert-char-spec cell readable)
 564           (char-db-insert-char-reference cell readable))
 565         (setq separator lbs))
 566        (t
 567         (if separator
 568             (insert separator))
 569         (insert (prin1-to-string cell))
 570         (setq separator " ")))
 571       (setq value (cdr value)))
 572     (insert ")")
 573     (insert line-breaking)))
 574
 575 (defun insert-char-attributes (char &optional readable attributes column)
 576   (unless column
 577     (setq column (current-column)))
 578   (let (name value ; has-long-ccs-name
 579         rest
 580         radical strokes
 581         (line-breaking
 582          (concat "\n" (make-string (1+ column) ?\ )))
 583         lbs cell separator ret
 584         key al cal
 585         dest-ccss ; sources required-features
 586         ccss)
 587     (let (atr-d)
 588       (setq attributes
 589             (sort (if attributes
 590                       (if (consp attributes)
 591                           (progn
 592                             (dolist (name attributes)
 593                               (unless (memq name char-db-ignored-attributes)
 594                                 (if (find-charset name)
 595                                     (push name ccss))
 596                                 (push name atr-d)))
 597                             atr-d))
 598                     (dolist (name (char-attribute-list))
 599                       (unless (memq name char-db-ignored-attributes)
 600                         (if (find-charset name)
 601                             (push name ccss))
 602                         (push name atr-d)))
 603                     atr-d)
 604                   #'char-attribute-name<)))
 605     (insert "(")
 606     (when (memq '<-subsumptive attributes)
 607       (when readable
 608         (when (setq value (get-char-attribute char '<-subsumptive))
 609           (char-db-insert-relation-feature char '<-subsumptive value
 610                                            line-breaking
 611                                            ccss readable)))
 612       (setq attributes (delq '<-subsumptive attributes)))
 613     (when (and (memq '<-denotational attributes)
 614                (setq value (get-char-attribute char '<-denotational)))
 615       (char-db-insert-relation-feature char '<-denotational value
 616                                        line-breaking
 617                                        ccss readable)
 618       (setq attributes (delq '<-denotational attributes)))
 619     (when (and (memq 'name attributes)
 620                (setq value (get-char-attribute char 'name)))
 621       (insert (format
 622                (if (> (+ (current-column) (length value)) 48)
 623                    "(name . %S)%s"
 624                  "(name               . %S)%s")
 625                value line-breaking))
 626       (setq attributes (delq 'name attributes))
 627       )
 628     (when (and (memq 'name* attributes)
 629                (setq value (get-char-attribute char 'name*)))
 630       (insert (format
 631                (if (> (+ (current-column) (length value)) 48)
 632                    "(name* . %S)%s"
 633                  "(name*              . %S)%s")
 634                value line-breaking))
 635       (setq attributes (delq 'name* attributes))
 636       )
 637     (when (and (memq 'script attributes)
 638                (setq value (get-char-attribute char 'script)))
 639       (insert (format "(script\t\t%s)%s"
 640                       (mapconcat (function prin1-to-string)
 641                                  value " ")
 642                       line-breaking))
 643       (setq attributes (delq 'script attributes))
 644       )
 645     (dolist (name '(=>ucs =>ucs*))
 646       (when (and (memq name attributes)
 647                  (setq value (get-char-attribute char name)))
 648         (insert (format "(%-18s . #x%04X)\t; %c%s"
 649                         name value (decode-char '=ucs value)
 650                         line-breaking))
 651         (setq attributes (delq name attributes))))
 652     (dolist (name '(=>ucs@gb =>ucs@cns =>ucs@jis =>ucs@ks =>ucs@big5))
 653       (when (and (memq name attributes)
 654                  (setq value (get-char-attribute char name)))
 655         (insert (format "(%-18s . #x%04X)\t; %c%s"
 656                         name value
 657                         (decode-char (intern
 658                                       (concat "="
 659                                               (substring
 660                                                (symbol-name name) 2)))
 661                                      value)
 662                         line-breaking))
 663         (setq attributes (delq name attributes))
 664         ))
 665     (dolist (name '(=>daikanwa))
 666       (when (and (memq name attributes)
 667                  (setq value (get-char-attribute char name)))
 668         (insert
 669          (if (integerp value)
 670              (format "(%-18s . %05d)\t; %c%s"
 671                      name value (decode-char '=daikanwa value)
 672                      line-breaking)
 673            (format "(%-18s %s)\t; %c%s"
 674                    name
 675                    (mapconcat (function prin1-to-string)
 676                               value " ")
 677                    (char-representative-of-daikanwa char)
 678                    line-breaking)))
 679         (setq attributes (delq name attributes))))
 680     (when (and (memq 'general-category attributes)
 681                (setq value (get-char-attribute char 'general-category)))
 682       (insert (format
 683                "(general-category\t%s) ; %s%s"
 684                (mapconcat (lambda (cell)
 685                             (format "%S" cell))
 686                           value " ")
 687                (cond ((rassoc value unidata-normative-category-alist)
 688                       "Normative Category")
 689                      ((rassoc value unidata-informative-category-alist)
 690                       "Informative Category")
 691                      (t
 692                       "Unknown Category"))
 693                line-breaking))
 694       (setq attributes (delq 'general-category attributes))
 695       )
 696     (when (and (memq 'bidi-category attributes)
 697                (setq value (get-char-attribute char 'bidi-category)))
 698       (insert (format "(bidi-category\t. %S)%s"
 699                       value
 700                       line-breaking))
 701       (setq attributes (delq 'bidi-category attributes))
 702       )
 703     (unless (or (not (memq 'mirrored attributes))
 704                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
 705                     'empty))
 706       (insert (format "(mirrored\t\t. %S)%s"
 707                       value
 708                       line-breaking))
 709       (setq attributes (delq 'mirrored attributes))
 710       )
 711     (cond
 712      ((and (memq 'decimal-digit-value attributes)
 713            (setq value (get-char-attribute char 'decimal-digit-value)))
 714       (insert (format "(decimal-digit-value . %S)%s"
 715                       value
 716                       line-breaking))
 717       (setq attributes (delq 'decimal-digit-value attributes))
 718       (when (and (memq 'digit-value attributes)
 719                  (setq value (get-char-attribute char 'digit-value)))
 720         (insert (format "(digit-value\t . %S)%s"
 721                         value
 722                         line-breaking))
 723         (setq attributes (delq 'digit-value attributes))
 724         )
 725       (when (and (memq 'numeric-value attributes)
 726                  (setq value (get-char-attribute char 'numeric-value)))
 727         (insert (format "(numeric-value\t . %S)%s"
 728                         value
 729                         line-breaking))
 730         (setq attributes (delq 'numeric-value attributes))
 731         )
 732       )
 733      (t
 734       (when (and (memq 'digit-value attributes)
 735                  (setq value (get-char-attribute char 'digit-value)))
 736         (insert (format "(digit-value\t. %S)%s"
 737                         value
 738                         line-breaking))
 739         (setq attributes (delq 'digit-value attributes))
 740         )
 741       (when (and (memq 'numeric-value attributes)
 742                  (setq value (get-char-attribute char 'numeric-value)))
 743         (insert (format "(numeric-value\t. %S)%s"
 744                         value
 745                         line-breaking))
 746         (setq attributes (delq 'numeric-value attributes))
 747         )))
 748     (when (and (memq 'iso-10646-comment attributes)
 749                (setq value (get-char-attribute char 'iso-10646-comment)))
 750       (insert (format "(iso-10646-comment\t. %S)%s"
 751                       value
 752                       line-breaking))
 753       (setq attributes (delq 'iso-10646-comment attributes))
 754       )
 755     (when (and (memq 'morohashi-daikanwa attributes)
 756                (setq value (get-char-attribute char 'morohashi-daikanwa)))
 757       (insert (format "(morohashi-daikanwa\t%s)%s"
 758                       (mapconcat (function prin1-to-string) value " ")
 759                       line-breaking))
 760       (setq attributes (delq 'morohashi-daikanwa attributes))
 761       )
 762     (setq radical nil
 763           strokes nil)
 764     (when (and (memq 'ideographic-radical attributes)
 765                (setq value (get-char-attribute char 'ideographic-radical)))
 766       (setq radical value)
 767       (insert (format "(ideographic-radical . %S)\t; %c%s"
 768                       radical
 769                       (ideographic-radical radical)
 770                       line-breaking))
 771       (setq attributes (delq 'ideographic-radical attributes))
 772       )
 773     (when (and (memq 'shuowen-radical attributes)
 774                (setq value (get-char-attribute char 'shuowen-radical)))
 775       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
 776                       value
 777                       (shuowen-radical value)
 778                       line-breaking))
 779       (setq attributes (delq 'shuowen-radical attributes))
 780       )
 781     (let (key)
 782       (dolist (domain
 783                (append
 784                 char-db-feature-domains
 785                 (let (dest domain)
 786                   (dolist (feature (char-attribute-list))
 787                     (setq feature (symbol-name feature))
 788                     (when (string-match
 789                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
 790                            feature)
 791                       (setq domain (intern (match-string 2 feature)))
 792                      (unless (memq domain dest)
 793                        (setq dest (cons domain dest)))))
 794                   (sort dest #'string<))))
 795         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
 796         (when (and (memq key attributes)
 797                    (setq value (get-char-attribute char key)))
 798           (setq radical value)
 799           (insert (format "(%s . %S)\t; %c%s"
 800                           key
 801                           radical
 802                           (ideographic-radical radical)
 803                           line-breaking))
 804           (setq attributes (delq key attributes))
 805           )
 806         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
 807         (when (and (memq key attributes)
 808                    (setq value (get-char-attribute char key)))
 809           (setq strokes value)
 810           (insert (format "(%s . %S)%s"
 811                           key
 812                           strokes
 813                           line-breaking))
 814           (setq attributes (delq key attributes))
 815           )
 816         (setq key (intern (format "%s@%s" 'total-strokes domain)))
 817         (when (and (memq key attributes)
 818                    (setq value (get-char-attribute char key)))
 819           (insert (format "(%s       . %S)%s"
 820                           key
 821                           value
 822                           line-breaking))
 823           (setq attributes (delq key attributes))
 824           )
 825         (dolist (feature '(ideographic-radical
 826                            ideographic-strokes
 827                            total-strokes))
 828           (setq key (intern (format "%s@%s*sources" feature domain)))
 829           (when (and (memq key attributes)
 830                      (setq value (get-char-attribute char key)))
 831             (insert (format "(%s%s" key line-breaking))
 832             (dolist (cell value)
 833               (insert (format " %s" cell)))
 834             (insert ")")
 835             (insert line-breaking)
 836             (setq attributes (delq key attributes))
 837             ))
 838         ))
 839     (when (and (memq 'ideographic-strokes attributes)
 840                (setq value (get-char-attribute char 'ideographic-strokes)))
 841       (setq strokes value)
 842       (insert (format "(ideographic-strokes . %S)%s"
 843                       strokes
 844                       line-breaking))
 845       (setq attributes (delq 'ideographic-strokes attributes))
 846       )
 847     (when (and (memq 'kangxi-radical attributes)
 848                (setq value (get-char-attribute char 'kangxi-radical)))
 849       (unless (eq value radical)
 850         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
 851                         value
 852                         (ideographic-radical value)
 853                         line-breaking))
 854         (or radical
 855             (setq radical value)))
 856       (setq attributes (delq 'kangxi-radical attributes))
 857       )
 858     (when (and (memq 'kangxi-strokes attributes)
 859                (setq value (get-char-attribute char 'kangxi-strokes)))
 860       (unless (eq value strokes)
 861         (insert (format "(kangxi-strokes\t . %S)%s"
 862                         value
 863                         line-breaking))
 864         (or strokes
 865             (setq strokes value)))
 866       (setq attributes (delq 'kangxi-strokes attributes))
 867       )
 868     (when (and (memq 'japanese-radical attributes)
 869                (setq value (get-char-attribute char 'japanese-radical)))
 870       (unless (eq value radical)
 871         (insert (format "(japanese-radical\t . %S)\t; %c%s"
 872                         value
 873                         (ideographic-radical value)
 874                         line-breaking))
 875         (or radical
 876             (setq radical value)))
 877       (setq attributes (delq 'japanese-radical attributes))
 878       )
 879     (when (and (memq 'japanese-strokes attributes)
 880                (setq value (get-char-attribute char 'japanese-strokes)))
 881       (unless (eq value strokes)
 882         (insert (format "(japanese-strokes\t . %S)%s"
 883                         value
 884                         line-breaking))
 885         (or strokes
 886             (setq strokes value)))
 887       (setq attributes (delq 'japanese-strokes attributes))
 888       )
 889     (when (and (memq 'cns-radical attributes)
 890                (setq value (get-char-attribute char 'cns-radical)))
 891       (insert (format "(cns-radical\t . %S)\t; %c%s"
 892                       value
 893                       (ideographic-radical value)
 894                       line-breaking))
 895       (setq attributes (delq 'cns-radical attributes))
 896       )
 897     (when (and (memq 'cns-strokes attributes)
 898                (setq value (get-char-attribute char 'cns-strokes)))
 899       (unless (eq value strokes)
 900         (insert (format "(cns-strokes\t . %S)%s"
 901                         value
 902                         line-breaking))
 903         (or strokes
 904             (setq strokes value)))
 905       (setq attributes (delq 'cns-strokes attributes))
 906       )
 907     (when (and (memq 'shinjigen-1-radical attributes)
 908                (setq value (get-char-attribute char 'shinjigen-1-radical)))
 909       (unless (eq value radical)
 910         (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
 911                         value
 912                         (ideographic-radical value)
 913                         line-breaking))
 914         (or radical
 915             (setq radical value)))
 916       (setq attributes (delq 'shinjigen-1-radical attributes))
 917       )
 918     (when (and (memq 'ideographic- attributes)
 919                (setq value (get-char-attribute char 'ideographic-)))
 920       (insert "(ideographic-       ")
 921       (setq lbs (concat "\n" (make-string (current-column) ?\ ))
 922             separator nil)
 923       (while (consp value)
 924         (setq cell (car value))
 925         (if (integerp cell)
 926             (setq cell (decode-char '=ucs cell)))
 927         (cond ((characterp cell)
 928                (if separator
 929                    (insert lbs))
 930                (if readable
 931                    (insert (format "%S" cell))
 932                  (char-db-insert-char-spec cell readable))
 933                (setq separator lbs))
 934               ((consp cell)
 935                (if separator
 936                    (insert lbs))
 937                (if (consp (car cell))
 938                    (char-db-insert-char-spec cell readable)
 939                  (char-db-insert-char-reference cell readable))
 940                (setq separator lbs))
 941               (t
 942                (if separator
 943                    (insert separator))
 944                (insert (prin1-to-string cell))
 945                (setq separator " ")))
 946         (setq value (cdr value)))
 947       (insert ")")
 948       (insert line-breaking)
 949       (setq attributes (delq 'ideographic- attributes)))
 950     (when (and (memq 'total-strokes attributes)
 951                (setq value (get-char-attribute char 'total-strokes)))
 952       (insert (format "(total-strokes       . %S)%s"
 953                       value
 954                       line-breaking))
 955       (setq attributes (delq 'total-strokes attributes))
 956       )
 957     (when (and (memq '->ideograph attributes)
 958                (setq value (get-char-attribute char '->ideograph)))
 959       (insert (format "(->ideograph\t%s)%s"
 960                       (mapconcat (lambda (code)
 961                                    (cond ((symbolp code)
 962                                           (symbol-name code))
 963                                          ((integerp code)
 964                                           (format "#x%04X" code))
 965                                          (t
 966                                           (format "%s %S"
 967                                                   line-breaking code))))
 968                                  value " ")
 969                       line-breaking))
 970       (setq attributes (delq '->ideograph attributes))
 971       )
 972     ;; (when (and (memq '->decomposition attributes)
 973     ;;            (setq value (get-char-attribute char '->decomposition)))
 974     ;;   (insert (format "(->decomposition\t%s)%s"
 975     ;;                   (mapconcat (lambda (code)
 976     ;;                                (cond ((symbolp code)
 977     ;;                                       (symbol-name code))
 978     ;;                                      ((characterp code)
 979     ;;                                       (if readable
 980     ;;                                           (format "%S" code)
 981     ;;                                         (format "#x%04X"
 982     ;;                                                 (char-int code))
 983     ;;                                         ))
 984     ;;                                      ((integerp code)
 985     ;;                                       (format "#x%04X" code))
 986     ;;                                      (t
 987     ;;                                       (format "%s%S" line-breaking code))))
 988     ;;                              value " ")
 989     ;;                   line-breaking))
 990     ;;   (setq attributes (delq '->decomposition attributes))
 991     ;;   )
 992     (if (equal (get-char-attribute char '->titlecase)
 993                (get-char-attribute char '->uppercase))
 994         (setq attributes (delq '->titlecase attributes)))
 995     (when (and (memq '->mojikyo attributes)
 996                (setq value (get-char-attribute char '->mojikyo)))
 997       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
 998                       value (decode-char 'mojikyo value)
 999                       line-breaking))
1000       (setq attributes (delq '->mojikyo attributes))
1001       )
1002     (when (and (memq 'hanyu-dazidian-vol attributes)
1003                (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1004       (insert (format "(hanyu-dazidian-vol  . %d)%s"
1005                       value line-breaking))
1006       (setq attributes (delq 'hanyu-dazidian-vol attributes))
1007       )
1008     (when (and (memq 'hanyu-dazidian-page attributes)
1009                (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1010       (insert (format "(hanyu-dazidian-page . %d)%s"
1011                       value line-breaking))
1012       (setq attributes (delq 'hanyu-dazidian-page attributes))
1013       )
1014     (when (and (memq 'hanyu-dazidian-char attributes)
1015                (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1016       (insert (format "(hanyu-dazidian-char . %d)%s"
1017                       value line-breaking))
1018       (setq attributes (delq 'hanyu-dazidian-char attributes))
1019       )
1020     (unless readable
1021       (dolist (ignored '(composition
1022                          ->denotational <-subsumptive ->ucs-unified
1023                          ->ideographic-component-forms))
1024         (setq attributes (delq ignored attributes))))
1025     (while attributes
1026       (setq name (car attributes))
1027       (if (setq value (get-char-attribute char name))
1028           (cond ((setq ret (find-charset name))
1029                  (setq name (charset-name ret))
1030                  (if (and (not (memq name dest-ccss))
1031                           (prog1
1032                               (setq value (get-char-attribute char name))
1033                             (setq dest-ccss (cons name dest-ccss))))
1034                      (char-db-insert-ccs-feature name value line-breaking))
1035                  )
1036                 ((string-match "^=>ucs@" (symbol-name name))
1037                  (insert (format "(%-18s . #x%04X)\t; %c%s"
1038                                  name value (decode-char '=ucs value)
1039                                  line-breaking))
1040                  )
1041                 ((eq name 'jisx0208-1978/4X)
1042                  (insert (format "(%-18s . #x%04X)%s"
1043                                  name value
1044                                  line-breaking))
1045                  )
1046                 ((and
1047                   (not readable)
1048                   (not (eq name '->subsumptive))
1049                   (not (eq name '->uppercase))
1050                   (not (eq name '->lowercase))
1051                   (not (eq name '->titlecase))
1052                   (not (eq name '->canonical))
1053                   (not (eq name '->Bopomofo))
1054                   (not (eq name '->mistakable))
1055                   (not (eq name '->ideographic-variants))
1056                   (null (get-char-attribute
1057                          char (intern (format "%s*sources" name))))
1058                   (not (string-match "\\*sources$" (symbol-name name)))
1059                   (null (get-char-attribute
1060                          char (intern (format "%s*note" name))))
1061                   (not (string-match "\\*note$" (symbol-name name)))
1062                   (or (eq name '<-identical)
1063                       (eq name '<-uppercase)
1064                       (eq name '<-lowercase)
1065                       (eq name '<-titlecase)
1066                       (eq name '<-canonical)
1067                       (eq name '<-ideographic-variants)
1068                       ;; (eq name '<-synonyms)
1069                       (string-match "^<-synonyms" (symbol-name name))
1070                       (eq name '<-mistakable)
1071                       (when (string-match "^->" (symbol-name name))
1072                         (cond
1073                          ((string-match "^->fullwidth" (symbol-name name))
1074                           (not (and (consp value)
1075                                     (characterp (car value))
1076                                     (encode-char
1077                                      (car value) '=ucs 'defined-only)))
1078                           )
1079                          (t)))
1080                       ))
1081                  )
1082                 ((or (eq name 'ideographic-structure)
1083                      (eq name 'ideographic-combination)
1084                      (eq name 'ideographic-)
1085                      (eq name '=decomposition)
1086                      (string-match "^=>decomposition" (symbol-name name))
1087                      (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1088                      (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1089                                    (symbol-name name))
1090                      )
1091                  (char-db-insert-relation-feature char name value
1092                                                   line-breaking
1093                                                   ccss readable))
1094                 ((memq name '(ideograph=
1095                               original-ideograph-of
1096                               ancient-ideograph-of
1097                               vulgar-ideograph-of
1098                               wrong-ideograph-of
1099                               ;; simplified-ideograph-of
1100                               ideographic-variants
1101                               ;; ideographic-different-form-of
1102                               ))
1103                  (insert (format "(%-18s%s " name line-breaking))
1104                  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1105                        separator nil)
1106                  (while (consp value)
1107                    (setq cell (car value))
1108                    (if (and (consp cell)
1109                             (consp (car cell)))
1110                        (progn
1111                          (if separator
1112                              (insert lbs))
1113                          (char-db-insert-alist cell readable)
1114                          (setq separator lbs))
1115                      (if separator
1116                          (insert separator))
1117                      (insert (prin1-to-string cell))
1118                      (setq separator " "))
1119                    (setq value (cdr value)))
1120                  (insert ")")
1121                  (insert line-breaking))
1122                 ((consp value)
1123                  (insert (format "(%-18s " name))
1124                  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1125                        separator nil)
1126                  (while (consp value)
1127                    (setq cell (car value))
1128                    (if (and (consp cell)
1129                             (consp (car cell))
1130                             (setq ret (condition-case nil
1131                                           (find-char cell)
1132                                         (error nil))))
1133                        (progn
1134                          (setq rest cell
1135                                al nil
1136                                cal nil)
1137                          (while rest
1138                            (setq key (car (car rest)))
1139                            (if (find-charset key)
1140                                (setq cal (cons key cal))
1141                              (setq al (cons key al)))
1142                            (setq rest (cdr rest)))
1143                          (if separator
1144                              (insert lbs))
1145                          (insert-char-attributes ret
1146                                                  readable
1147                                                  al cal)
1148                          (setq separator lbs))
1149                      (setq ret (prin1-to-string cell))
1150                      (if separator
1151                          (if (< (+ (current-column)
1152                                    (length ret)
1153                                    (length separator))
1154                                 76)
1155                              (insert separator)
1156                            (insert lbs)))
1157                      (insert ret)
1158                      (setq separator " "))
1159                    (setq value (cdr value)))
1160                  (insert ")")
1161                  (insert line-breaking))
1162                 (t
1163                  (insert (format "(%-18s" name))
1164                  (setq ret (prin1-to-string value))
1165                  (unless (< (+ (current-column)
1166                                (length ret)
1167                                3)
1168                             76)
1169                    (insert line-breaking))
1170                  (insert " . " ret ")" line-breaking)
1171                  ;; (insert (format "(%-18s . %S)%s"
1172                  ;;                 name value
1173                  ;;                 line-breaking))
1174                  )
1175                 ))
1176       (setq attributes (cdr attributes)))
1177     (insert ")")))
1178
1179 (defun insert-char-data (char &optional readable
1180                               attributes)
1181   (save-restriction
1182     (narrow-to-region (point)(point))
1183     (insert "(define-char
1184   '")
1185     (insert-char-attributes char readable attributes)
1186     (insert ")\n")
1187     (goto-char (point-min))
1188     (while (re-search-forward "[ \t]+$" nil t)
1189       (replace-match ""))
1190     ;; from tabify.
1191     (goto-char (point-min))
1192     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1193       (let ((column (current-column))
1194             (indent-tabs-mode t))
1195         (delete-region (match-beginning 0) (point))
1196         (indent-to column)))
1197     (goto-char (point-max))
1198     ;; (tabify (point-min)(point-max))
1199     ))
1200
1201 (defun insert-char-data-with-variant (char &optional printable
1202                                            no-ucs-unified
1203                                            script excluded-script)
1204   (insert-char-data char printable)
1205   (let ((variants (char-variants char))
1206         rest
1207         variant vs ret)
1208     (setq variants (sort variants #'<))
1209     (setq rest variants)
1210     (setq variants (cons char variants))
1211     (while rest
1212       (setq variant (car rest))
1213       (unless (get-char-attribute variant '<-subsumptive)
1214         (if (and (or (null script)
1215                      (null (setq vs (get-char-attribute variant 'script)))
1216                      (memq script vs))
1217                  (or (null excluded-script)
1218                      (null (setq vs (get-char-attribute variant 'script)))
1219                      (not (memq excluded-script vs))))
1220             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1221               (insert-char-data variant printable)
1222               (if (setq ret (char-variants variant))
1223                   (while ret
1224                     (or (memq (car ret) variants)
1225                         ;; (get-char-attribute (car ret) '<-subsumptive)
1226                         (setq rest (nconc rest (list (car ret)))))
1227                     (setq ret (cdr ret)))))))
1228       (setq rest (cdr rest)))))
1229
1230 (defun insert-char-range-data (min max &optional script excluded-script)
1231   (let ((code min)
1232         char)
1233     (while (<= code max)
1234       (setq char (decode-char '=ucs code))
1235       (if (encode-char char '=ucs 'defined-only)
1236           (insert-char-data-with-variant char nil 'no-ucs-unified
1237                                          script excluded-script))
1238       (setq code (1+ code)))))
1239
1240 (defun write-char-range-data-to-file (min max file
1241                                           &optional script excluded-script)
1242   (let ((coding-system-for-write char-db-file-coding-system))
1243     (with-temp-buffer
1244       (insert (format ";; -*- coding: %s -*-\n"
1245                       char-db-file-coding-system))
1246       (insert-char-range-data min max script excluded-script)
1247       (write-region (point-min)(point-max) file))))
1248
1249 (defvar what-character-original-window-configuration)
1250
1251 ;;;###autoload
1252 (defun what-char-definition (char)
1253   (interactive (list (char-after)))
1254   (let ((buf (get-buffer-create "*Character Description*"))
1255         (the-buf (current-buffer))
1256         (win-conf (current-window-configuration)))
1257     (pop-to-buffer buf)
1258     (make-local-variable 'what-character-original-window-configuration)
1259     (setq what-character-original-window-configuration win-conf)
1260     (setq buffer-read-only nil)
1261     (erase-buffer)
1262     (condition-case err
1263         (progn
1264           (insert-char-data-with-variant char 'printable)
1265           (unless (char-attribute-alist char)
1266             (insert (format ";; = %c\n"
1267                             (let* ((rest (split-char char))
1268                                    (ccs (pop rest))
1269                                    (code (pop rest)))
1270                               (while rest
1271                                 (setq code (logior (lsh code 8)
1272                                                    (pop rest))))
1273                               (decode-char ccs code)))))
1274           ;; (char-db-update-comment)
1275           (set-buffer-modified-p nil)
1276           (view-mode the-buf (lambda (buf)
1277                                (set-window-configuration
1278                                 what-character-original-window-configuration)
1279                                ))
1280           (goto-char (point-min)))
1281       (error (progn
1282                (set-window-configuration
1283                 what-character-original-window-configuration)
1284                (signal (car err) (cdr err)))))))
1285
1286
1287 ;;; @ end
1288 ;;;
1289
1290 (provide 'char-db-util)
1291
1292 ;;; char-db-util.el ends here