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