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, 2011 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@usual
 173     =>jis-x0213-2
 174     ==>ucs@bucs
 175     =>ucs@iso
 176     =>ucs@unicode
 177     =>ucs@jis
 178     =>ucs@JP
 179     =>ucs@cns
 180     =>ucs@ks
 181     =>>ucs@unicode
 182     =>>ucs@jis
 183     =>>ucs@cns
 184     =ucs@iso
 185     =ucs@unicode
 186     =>>big5-cdp
 187     =>>gt-k
 188     =>gt
 189     =>big5-cdp
 190     =>daikanwa
 191     =big5
 192     =big5-eten
 193     =zinbun-oracle
 194     =>zinbun-oracle
 195     =ruimoku-v6
 196     =>>ruimoku-v6
 197     =jef-china3
 198     =shinjigen))
 199
 200
 201 ;;; @ char-db formatters
 202 ;;;
 203
 204 (defun char-db-make-char-spec (char)
 205   (let (ret char-spec)
 206     (cond ((characterp char)
 207            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 208                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 209                   (setq char-spec (list (cons '=ucs ret)))
 210                   (cond ((setq ret (get-char-attribute char 'name))
 211                          (setq char-spec (cons (cons 'name ret) char-spec))
 212                          )
 213                         ((setq ret (get-char-attribute char 'name*))
 214                          (setq char-spec (cons (cons 'name* ret) char-spec))
 215                          ))
 216                   )
 217                  ((setq ret
 218                         (catch 'tag
 219                           (let ((rest char-db-coded-charset-priority-list)
 220                                 ccs)
 221                             (while rest
 222                               (setq ccs (charset-name
 223                                          (find-charset (car rest))))
 224                               (if (setq ret
 225                                         (encode-char char ccs
 226                                                      'defined-only))
 227                                   (throw 'tag (cons ccs ret)))
 228                               (setq rest (cdr rest))))))
 229                   (setq char-spec (list ret))
 230                   (dolist (ccs (delq (car ret) (charset-list)))
 231                     (if (and (or (charset-iso-final-char ccs)
 232                                  (memq ccs
 233                                        '(=daikanwa
 234                                          =daikanwa@rev2
 235                                          ;; =gt-k
 236                                          =jis-x0208@1997
 237                                          ))
 238                                  (string-match "=ucs@" (symbol-name ccs)))
 239                              (setq ccs (charset-name ccs))
 240                              (null (assq ccs char-spec))
 241                              (setq ret (encode-char char ccs 'defined-only)))
 242                         (setq char-spec (cons (cons ccs ret) char-spec))))
 243                   (if (null char-spec)
 244                       (setq char-spec (split-char char)))
 245                   (cond ((setq ret (get-char-attribute char 'name))
 246                          (setq char-spec (cons (cons 'name ret) char-spec))
 247                          )
 248                         ((setq ret (get-char-attribute char 'name*))
 249                          (setq char-spec (cons (cons 'name* ret) char-spec))
 250                          ))
 251                   )
 252                  ((setq ret (get-char-attribute
 253                              char 'ideographic-combination))
 254                   (setq char-spec
 255                         (cons (cons 'ideographic-combination ret)
 256                               char-spec))
 257                   ))
 258            char-spec)
 259           ((consp char)
 260            char))))
 261
 262 (defun char-db-insert-char-spec (char &optional readable column
 263                                       required-features)
 264   (unless column
 265     (setq column (current-column)))
 266   (let (char-spec temp-char)
 267     (setq char-spec (char-db-make-char-spec char))
 268     (unless (or (characterp char) ; char
 269                 (condition-case nil
 270                     (setq char (find-char char-spec))
 271                   (error nil)))
 272       ;; define temporary character
 273       ;;   Current implementation is dirty.
 274       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 275                                          char-spec)))
 276       (remove-char-attribute temp-char 'ideograph-daikanwa)
 277       (setq char temp-char))
 278     (insert-char-attributes char
 279                             readable
 280                             (union (mapcar #'car char-spec)
 281                                    required-features))
 282     (when temp-char
 283       ;; undefine temporary character
 284       ;;   Current implementation is dirty.
 285       (setq char-spec (char-attribute-alist temp-char))
 286       (while char-spec
 287         (remove-char-attribute temp-char (car (car char-spec)))
 288         (setq char-spec (cdr char-spec))))))
 289
 290 (defun char-db-insert-alist (alist &optional readable column)
 291   (unless column
 292     (setq column (current-column)))
 293   (let ((line-breaking
 294          (concat "\n" (make-string (1+ column) ?\ )))
 295         name value
 296         ret al ; cal
 297         key
 298         lbs cell rest separator)
 299     (insert "(")
 300     (while alist
 301       (setq name (car (car alist))
 302             value (cdr (car alist)))
 303       (cond ((eq name 'char)
 304              (insert "(char . ")
 305              (if (setq ret (condition-case nil
 306                                (find-char value)
 307                              (error nil)))
 308                  (progn
 309                    (setq al nil
 310                          ;; cal nil
 311                          )
 312                    (while value
 313                      (setq key (car (car value)))
 314                      ;; (if (find-charset key)
 315                      ;;     (setq cal (cons key cal))
 316                      (setq al (cons key al))
 317                      ;; )
 318                      (setq value (cdr value)))
 319                    (insert-char-attributes ret
 320                                            readable
 321                                            (or al 'none) ; cal
 322                                            ))
 323                (insert (prin1-to-string value)))
 324              (insert ")")
 325              (insert line-breaking))
 326             ((consp value)
 327              (insert (format "(%-18s " name))
 328              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 329              (while (consp value)
 330                (setq cell (car value))
 331                (if (and (consp cell)
 332                         (consp (car cell))
 333                         (setq ret (condition-case nil
 334                                       (find-char cell)
 335                                     (error nil)))
 336                         )
 337                    (progn
 338                      (setq rest cell
 339                            al nil
 340                            ;; cal nil
 341                            )
 342                      (while rest
 343                        (setq key (car (car rest)))
 344                        ;; (if (find-charset key)
 345                        ;;     (setq cal (cons key cal))
 346                        (setq al (cons key al))
 347                        ;; )
 348                        (setq rest (cdr rest)))
 349                      (if separator
 350                          (insert lbs))
 351                      (insert-char-attributes ret
 352                                              readable
 353                                              al ; cal
 354                                              )
 355                      (setq separator lbs))
 356                  (if separator
 357                      (insert separator))
 358                  (insert (prin1-to-string cell))
 359                  (setq separator " "))
 360                (setq value (cdr value)))
 361              (insert ")")
 362              (insert line-breaking))
 363             (t
 364              (insert (format "(%-18s . %S)%s"
 365                              name value
 366                              line-breaking))))
 367       (setq alist (cdr alist))))
 368   (insert ")"))
 369
 370 (defun char-db-insert-char-reference (plist &optional readable column)
 371   (unless column
 372     (setq column (current-column)))
 373   (let ((line-breaking
 374          (concat "\n" (make-string (1+ column) ?\ )))
 375         (separator "")
 376         name value)
 377     (insert "(")
 378     (while plist
 379       (setq name (pop plist))
 380       (setq value (pop plist))
 381       (cond ((eq name :char)
 382              (insert separator)
 383              (insert ":char\t")
 384              (cond ((numberp value)
 385                     (setq value (decode-char '=ucs value)))
 386                    ;; ((consp value)
 387                    ;;  (setq value (or (find-char value)
 388                    ;;                  value)))
 389                    )
 390              (char-db-insert-char-spec value readable)
 391              (insert line-breaking)
 392              (setq separator ""))
 393             ((eq name :radical)
 394              (insert (format "%s%s\t%d ; %c%s"
 395                              separator
 396                              name value
 397                              (ideographic-radical value)
 398                              line-breaking))
 399              (setq separator ""))
 400             (t
 401              (insert (format "%s%s\t%S" separator name value))
 402              (setq separator line-breaking)))
 403       ))
 404   (insert ")"))
 405
 406 (defun char-db-decode-isolated-char (ccs code-point)
 407   (let (ret)
 408     (setq ret
 409           (cond ((eq ccs 'arabic-iso8859-6)
 410                  (decode-char ccs code-point))
 411                 ((and (memq ccs '(=gt-pj-1
 412                                   =gt-pj-2
 413                                   =gt-pj-3
 414                                   =gt-pj-4
 415                                   =gt-pj-5
 416                                   =gt-pj-6
 417                                   =gt-pj-7
 418                                   =gt-pj-8
 419                                   =gt-pj-9
 420                                   =gt-pj-10
 421                                   =gt-pj-11))
 422                       (setq ret (decode-char ccs code-point))
 423                       (setq ret (encode-char ret '=gt 'defined-only)))
 424                  (decode-builtin-char '=gt ret))
 425                 (t
 426                  (decode-builtin-char ccs code-point))))
 427     (cond ((and (<= 0 (char-int ret))
 428                 (<= (char-int ret) #x1F))
 429            (decode-char '=ucs (+ #x2400 (char-int ret))))
 430           ((= (char-int ret) #x7F)
 431            ?\u2421)
 432           (t ret))))
 433
 434 (defvar char-db-convert-obsolete-format t)
 435
 436 (defun char-db-insert-ccs-feature (name value line-breaking)
 437   (insert
 438    (format
 439     (cond ((memq name '(=shinjigen
 440                         =shinjigen@1ed
 441                         =shinjigen@rev =shinjigen/+p@rev
 442                         =daikanwa/ho))
 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                             =daikanwa/+p =daikanwa/+2p
 449                             =>>daikanwa =>daikanwa
 450                             =gt =>>gt =>gt =gt-k =>>gt-k =cbeta
 451                             =zinbun-oracle =>zinbun-oracle))
 452                (string-match "^=adobe-" (symbol-name name)))
 453            "(%-18s . %05d)\t; %c")
 454           ((eq name 'mojikyo)
 455            "(%-18s . %06d)\t; %c")
 456           ((>= (charset-dimension name) 2)
 457            "(%-18s . #x%04X)\t; %c")
 458           (t
 459            "(%-18s . #x%02X)\t; %c"))
 460     name
 461     (if (= (charset-iso-graphic-plane name) 1)
 462         (logior value
 463                 (cond ((= (charset-dimension name) 1)
 464                        #x80)
 465                       ((= (charset-dimension name) 2)
 466                        #x8080)
 467                       ((= (charset-dimension name) 3)
 468                        #x808080)
 469                       (t 0)))
 470       value)
 471     (char-db-decode-isolated-char name value)))
 472   (if (and (= (charset-chars name) 94)
 473            (= (charset-dimension name) 2))
 474       (insert (format " [%02d-%02d]"
 475                       (- (lsh value -8) 32)
 476                       (- (logand value 255) 32))))
 477   (insert line-breaking))
 478
 479 (defun char-db-insert-relation-feature (char name value line-breaking
 480                                              ccss readable)
 481   (insert (format "(%-18s%s " name line-breaking))
 482   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 483         separator cell sources required-features
 484         ret)
 485     (while (consp value)
 486       (setq cell (car value))
 487       (if (integerp cell)
 488           (setq cell (decode-char '=ucs cell)))
 489       (cond
 490        ((eq name '->subsumptive)
 491         (when (or (not (some (lambda (atr)
 492                                (get-char-attribute cell atr))
 493                              char-db-ignored-attributes))
 494                   (some (lambda (ccs)
 495                           (encode-char cell ccs 'defined-only))
 496                         ccss))
 497           (if separator
 498               (insert lbs))
 499           (let ((char-db-ignored-attributes
 500                  (cons '<-subsumptive
 501                        char-db-ignored-attributes)))
 502             (insert-char-attributes cell readable))
 503           (setq separator lbs))
 504         )
 505        ((characterp cell)
 506         (setq sources
 507               (get-char-attribute
 508                char (intern (format "%s*sources" name))))
 509         (setq required-features nil)
 510         (dolist (source sources)
 511           (cond
 512            ((memq source '(JP
 513                            JP/Jouyou
 514                            shinjigen shinjigen@1ed shinjigen@rev))
 515             (setq required-features
 516                   (union required-features
 517                          '(=jis-x0208
 518                            =jis-x0208@1990
 519                            =jis-x0213-1@2000
 520                            =jis-x0213-1@2004
 521                            =jis-x0213-2
 522                            =jis-x0212
 523                            =jis-x0208@1983
 524                            =jis-x0208@1978
 525                            =shinjigen))))
 526            ((eq source 'CN)
 527             (setq required-features
 528                   (union required-features
 529                          '(=gb2312
 530                            =gb12345
 531                            =iso-ir165)))))
 532           (cond
 533            ((find-charset (setq ret (intern (format "=%s" source))))
 534             (setq required-features
 535                   (cons ret required-features)))
 536            (t (setq required-features
 537                     (cons source required-features)))))
 538         (cond ((string-match "@JP" (symbol-name name))
 539                (setq required-features
 540                      (union required-features
 541                             '(=jis-x0208
 542                               =jis-x0208@1990
 543                               =jis-x0213-1-2000
 544                               =jis-x0213-2-2000
 545                               =jis-x0212
 546                               =jis-x0208@1983
 547                               =jis-x0208@1978))))
 548               ((string-match "@CN" (symbol-name name))
 549                (setq required-features
 550                      (union required-features
 551                             '(=gb2312
 552                               =gb12345
 553                               =iso-ir165)))))
 554         (if separator
 555             (insert lbs))
 556         (if readable
 557             (insert (format "%S" cell))
 558           (char-db-insert-char-spec cell readable
 559                                     nil
 560                                     required-features))
 561         (setq separator lbs))
 562        ((consp cell)
 563         (if separator
 564             (insert lbs))
 565         (if (consp (car cell))
 566             (char-db-insert-char-spec cell readable)
 567           (char-db-insert-char-reference cell readable))
 568         (setq separator lbs))
 569        (t
 570         (if separator
 571             (insert separator))
 572         (insert (prin1-to-string cell))
 573         (setq separator " ")))
 574       (setq value (cdr value)))
 575     (insert ")")
 576     (insert line-breaking)))
 577
 578 (defun insert-char-attributes (char &optional readable attributes column)
 579   (unless column
 580     (setq column (current-column)))
 581   (let (name value ; has-long-ccs-name
 582         rest
 583         radical strokes
 584         (line-breaking
 585          (concat "\n" (make-string (1+ column) ?\ )))
 586         lbs cell separator ret
 587         key al cal
 588         dest-ccss ; sources required-features
 589         ccss)
 590     (let (atr-d)
 591       (setq attributes
 592             (sort (if attributes
 593                       (if (consp attributes)
 594                           (progn
 595                             (dolist (name attributes)
 596                               (unless (memq name char-db-ignored-attributes)
 597                                 (if (find-charset name)
 598                                     (push name ccss))
 599                                 (push name atr-d)))
 600                             atr-d))
 601                     (dolist (name (char-attribute-list))
 602                       (unless (memq name char-db-ignored-attributes)
 603                         (if (find-charset name)
 604                             (push name ccss))
 605                         (push name atr-d)))
 606                     atr-d)
 607                   #'char-attribute-name<)))
 608     (insert "(")
 609     (when (memq '<-subsumptive attributes)
 610       (when readable
 611         (when (setq value (get-char-attribute char '<-subsumptive))
 612           (char-db-insert-relation-feature char '<-subsumptive value
 613                                            line-breaking
 614                                            ccss readable)))
 615       (setq attributes (delq '<-subsumptive attributes)))
 616     (when (and (memq '<-denotational attributes)
 617                (setq value (get-char-attribute char '<-denotational)))
 618       (char-db-insert-relation-feature char '<-denotational value
 619                                        line-breaking
 620                                        ccss readable)
 621       (setq attributes (delq '<-denotational attributes)))
 622     (when (and (memq 'name attributes)
 623                (setq value (get-char-attribute char 'name)))
 624       (insert (format
 625                (if (> (+ (current-column) (length value)) 48)
 626                    "(name . %S)%s"
 627                  "(name               . %S)%s")
 628                value line-breaking))
 629       (setq attributes (delq 'name attributes))
 630       )
 631     (when (and (memq 'name* attributes)
 632                (setq value (get-char-attribute char 'name*)))
 633       (insert (format
 634                (if (> (+ (current-column) (length value)) 48)
 635                    "(name* . %S)%s"
 636                  "(name*              . %S)%s")
 637                value line-breaking))
 638       (setq attributes (delq 'name* attributes))
 639       )
 640     (when (and (memq 'script attributes)
 641                (setq value (get-char-attribute char 'script)))
 642       (insert (format "(script\t\t%s)%s"
 643                       (mapconcat (function prin1-to-string)
 644                                  value " ")
 645                       line-breaking))
 646       (setq attributes (delq 'script attributes))
 647       )
 648     (dolist (name '(=>ucs =>ucs*))
 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 (decode-char '=ucs value)
 653                         line-breaking))
 654         (setq attributes (delq name attributes))))
 655     (dolist (name '(=>ucs@gb =>ucs@big5))
 656       (when (and (memq name attributes)
 657                  (setq value (get-char-attribute char name)))
 658         (insert (format "(%-18s . #x%04X)\t; %c%s"
 659                         name value
 660                         (decode-char (intern
 661                                       (concat "="
 662                                               (substring
 663                                                (symbol-name name) 2)))
 664                                      value)
 665                         line-breaking))
 666         (setq attributes (delq name attributes))
 667         ))
 668     ;; (dolist (name '(=>daikanwa))
 669     ;;   (when (and (memq name attributes)
 670     ;;              (setq value (get-char-attribute char name)))
 671     ;;     (insert
 672     ;;      (if (integerp value)
 673     ;;          (format "(%-18s . %05d)\t; %c%s"
 674     ;;                  name value (decode-char '=daikanwa value)
 675     ;;                  line-breaking)
 676     ;;        (format "(%-18s %s)\t; %c%s"
 677     ;;                name
 678     ;;                (mapconcat (function prin1-to-string)
 679     ;;                           value " ")
 680     ;;                (char-representative-of-daikanwa char)
 681     ;;                line-breaking)))
 682     ;;     (setq attributes (delq name attributes))))
 683     (when (and (memq 'general-category attributes)
 684                (setq value (get-char-attribute char 'general-category)))
 685       (insert (format
 686                "(general-category\t%s) ; %s%s"
 687                (mapconcat (lambda (cell)
 688                             (format "%S" cell))
 689                           value " ")
 690                (cond ((rassoc value unidata-normative-category-alist)
 691                       "Normative Category")
 692                      ((rassoc value unidata-informative-category-alist)
 693                       "Informative Category")
 694                      (t
 695                       "Unknown Category"))
 696                line-breaking))
 697       (setq attributes (delq 'general-category attributes))
 698       )
 699     (when (and (memq 'bidi-category attributes)
 700                (setq value (get-char-attribute char 'bidi-category)))
 701       (insert (format "(bidi-category\t. %S)%s"
 702                       value
 703                       line-breaking))
 704       (setq attributes (delq 'bidi-category attributes))
 705       )
 706     (unless (or (not (memq 'mirrored attributes))
 707                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
 708                     'empty))
 709       (insert (format "(mirrored\t\t. %S)%s"
 710                       value
 711                       line-breaking))
 712       (setq attributes (delq 'mirrored attributes))
 713       )
 714     (cond
 715      ((and (memq 'decimal-digit-value attributes)
 716            (setq value (get-char-attribute char 'decimal-digit-value)))
 717       (insert (format "(decimal-digit-value . %S)%s"
 718                       value
 719                       line-breaking))
 720       (setq attributes (delq 'decimal-digit-value attributes))
 721       (when (and (memq 'digit-value attributes)
 722                  (setq value (get-char-attribute char 'digit-value)))
 723         (insert (format "(digit-value\t . %S)%s"
 724                         value
 725                         line-breaking))
 726         (setq attributes (delq 'digit-value attributes))
 727         )
 728       (when (and (memq 'numeric-value attributes)
 729                  (setq value (get-char-attribute char 'numeric-value)))
 730         (insert (format "(numeric-value\t . %S)%s"
 731                         value
 732                         line-breaking))
 733         (setq attributes (delq 'numeric-value attributes))
 734         )
 735       )
 736      (t
 737       (when (and (memq 'digit-value attributes)
 738                  (setq value (get-char-attribute char 'digit-value)))
 739         (insert (format "(digit-value\t. %S)%s"
 740                         value
 741                         line-breaking))
 742         (setq attributes (delq 'digit-value attributes))
 743         )
 744       (when (and (memq 'numeric-value attributes)
 745                  (setq value (get-char-attribute char 'numeric-value)))
 746         (insert (format "(numeric-value\t. %S)%s"
 747                         value
 748                         line-breaking))
 749         (setq attributes (delq 'numeric-value attributes))
 750         )))
 751     (when (and (memq 'iso-10646-comment attributes)
 752                (setq value (get-char-attribute char 'iso-10646-comment)))
 753       (insert (format "(iso-10646-comment\t. %S)%s"
 754                       value
 755                       line-breaking))
 756       (setq attributes (delq 'iso-10646-comment attributes))
 757       )
 758     (when (and (memq 'morohashi-daikanwa attributes)
 759                (setq value (get-char-attribute char 'morohashi-daikanwa)))
 760       (insert (format "(morohashi-daikanwa\t%s)%s"
 761                       (mapconcat (function prin1-to-string) value " ")
 762                       line-breaking))
 763       (setq attributes (delq 'morohashi-daikanwa attributes))
 764       )
 765     (setq radical nil
 766           strokes nil)
 767     (when (and (memq 'ideographic-radical attributes)
 768                (setq value (get-char-attribute char 'ideographic-radical)))
 769       (setq radical value)
 770       (insert (format "(ideographic-radical . %S)\t; %c%s"
 771                       radical
 772                       (ideographic-radical radical)
 773                       line-breaking))
 774       (setq attributes (delq 'ideographic-radical attributes))
 775       )
 776     (when (and (memq 'shuowen-radical attributes)
 777                (setq value (get-char-attribute char 'shuowen-radical)))
 778       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
 779                       value
 780                       (shuowen-radical value)
 781                       line-breaking))
 782       (setq attributes (delq 'shuowen-radical attributes))
 783       )
 784     (let (key)
 785       (dolist (domain
 786                (append
 787                 char-db-feature-domains
 788                 (let (dest domain)
 789                   (dolist (feature (char-attribute-list))
 790                     (setq feature (symbol-name feature))
 791                     (when (string-match
 792                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
 793                            feature)
 794                       (setq domain (intern (match-string 2 feature)))
 795                      (unless (memq domain dest)
 796                        (setq dest (cons domain dest)))))
 797                   (sort dest #'string<))))
 798         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
 799         (when (and (memq key attributes)
 800                    (setq value (get-char-attribute char key)))
 801           (setq radical value)
 802           (insert (format "(%s . %S)\t; %c%s"
 803                           key
 804                           radical
 805                           (ideographic-radical radical)
 806                           line-breaking))
 807           (setq attributes (delq key attributes))
 808           )
 809         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
 810         (when (and (memq key attributes)
 811                    (setq value (get-char-attribute char key)))
 812           (setq strokes value)
 813           (insert (format "(%s . %S)%s"
 814                           key
 815                           strokes
 816                           line-breaking))
 817           (setq attributes (delq key attributes))
 818           )
 819         (setq key (intern (format "%s@%s" 'total-strokes domain)))
 820         (when (and (memq key attributes)
 821                    (setq value (get-char-attribute char key)))
 822           (insert (format "(%s       . %S)%s"
 823                           key
 824                           value
 825                           line-breaking))
 826           (setq attributes (delq key attributes))
 827           )
 828         (dolist (feature '(ideographic-radical
 829                            ideographic-strokes
 830                            total-strokes))
 831           (setq key (intern (format "%s@%s*sources" feature domain)))
 832           (when (and (memq key attributes)
 833                      (setq value (get-char-attribute char key)))
 834             (insert (format "(%s%s" key line-breaking))
 835             (dolist (cell value)
 836               (insert (format " %s" cell)))
 837             (insert ")")
 838             (insert line-breaking)
 839             (setq attributes (delq key attributes))
 840             ))
 841         ))
 842     (when (and (memq 'ideographic-strokes attributes)
 843                (setq value (get-char-attribute char 'ideographic-strokes)))
 844       (setq strokes value)
 845       (insert (format "(ideographic-strokes . %S)%s"
 846                       strokes
 847                       line-breaking))
 848       (setq attributes (delq 'ideographic-strokes attributes))
 849       )
 850     (when (and (memq 'kangxi-radical attributes)
 851                (setq value (get-char-attribute char 'kangxi-radical)))
 852       (unless (eq value radical)
 853         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
 854                         value
 855                         (ideographic-radical value)
 856                         line-breaking))
 857         (or radical
 858             (setq radical value)))
 859       (setq attributes (delq 'kangxi-radical attributes))
 860       )
 861     (when (and (memq 'kangxi-strokes attributes)
 862                (setq value (get-char-attribute char 'kangxi-strokes)))
 863       (unless (eq value strokes)
 864         (insert (format "(kangxi-strokes\t . %S)%s"
 865                         value
 866                         line-breaking))
 867         (or strokes
 868             (setq strokes value)))
 869       (setq attributes (delq 'kangxi-strokes attributes))
 870       )
 871     (when (and (memq 'japanese-radical attributes)
 872                (setq value (get-char-attribute char 'japanese-radical)))
 873       (unless (eq value radical)
 874         (insert (format "(japanese-radical\t . %S)\t; %c%s"
 875                         value
 876                         (ideographic-radical value)
 877                         line-breaking))
 878         (or radical
 879             (setq radical value)))
 880       (setq attributes (delq 'japanese-radical attributes))
 881       )
 882     (when (and (memq 'japanese-strokes attributes)
 883                (setq value (get-char-attribute char 'japanese-strokes)))
 884       (unless (eq value strokes)
 885         (insert (format "(japanese-strokes\t . %S)%s"
 886                         value
 887                         line-breaking))
 888         (or strokes
 889             (setq strokes value)))
 890       (setq attributes (delq 'japanese-strokes attributes))
 891       )
 892     (when (and (memq 'cns-radical attributes)
 893                (setq value (get-char-attribute char 'cns-radical)))
 894       (insert (format "(cns-radical\t . %S)\t; %c%s"
 895                       value
 896                       (ideographic-radical value)
 897                       line-breaking))
 898       (setq attributes (delq 'cns-radical attributes))
 899       )
 900     (when (and (memq 'cns-strokes attributes)
 901                (setq value (get-char-attribute char 'cns-strokes)))
 902       (unless (eq value strokes)
 903         (insert (format "(cns-strokes\t . %S)%s"
 904                         value
 905                         line-breaking))
 906         (or strokes
 907             (setq strokes value)))
 908       (setq attributes (delq 'cns-strokes attributes))
 909       )
 910     (when (and (memq 'shinjigen-1-radical attributes)
 911                (setq value (get-char-attribute char 'shinjigen-1-radical)))
 912       (unless (eq value radical)
 913         (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
 914                         value
 915                         (ideographic-radical value)
 916                         line-breaking))
 917         (or radical
 918             (setq radical value)))
 919       (setq attributes (delq 'shinjigen-1-radical attributes))
 920       )
 921     (when (and (memq 'ideographic- attributes)
 922                (setq value (get-char-attribute char 'ideographic-)))
 923       (insert "(ideographic-       ")
 924       (setq lbs (concat "\n" (make-string (current-column) ?\ ))
 925             separator nil)
 926       (while (consp value)
 927         (setq cell (car value))
 928         (if (integerp cell)
 929             (setq cell (decode-char '=ucs cell)))
 930         (cond ((characterp cell)
 931                (if separator
 932                    (insert lbs))
 933                (if readable
 934                    (insert (format "%S" cell))
 935                  (char-db-insert-char-spec cell readable))
 936                (setq separator lbs))
 937               ((consp cell)
 938                (if separator
 939                    (insert lbs))
 940                (if (consp (car cell))
 941                    (char-db-insert-char-spec cell readable)
 942                  (char-db-insert-char-reference cell readable))
 943                (setq separator lbs))
 944               (t
 945                (if separator
 946                    (insert separator))
 947                (insert (prin1-to-string cell))
 948                (setq separator " ")))
 949         (setq value (cdr value)))
 950       (insert ")")
 951       (insert line-breaking)
 952       (setq attributes (delq 'ideographic- attributes)))
 953     (when (and (memq 'total-strokes attributes)
 954                (setq value (get-char-attribute char 'total-strokes)))
 955       (insert (format "(total-strokes       . %S)%s"
 956                       value
 957                       line-breaking))
 958       (setq attributes (delq 'total-strokes attributes))
 959       )
 960     (when (and (memq '->ideograph attributes)
 961                (setq value (get-char-attribute char '->ideograph)))
 962       (insert (format "(->ideograph\t%s)%s"
 963                       (mapconcat (lambda (code)
 964                                    (cond ((symbolp code)
 965                                           (symbol-name code))
 966                                          ((integerp code)
 967                                           (format "#x%04X" code))
 968                                          (t
 969                                           (format "%s %S"
 970                                                   line-breaking code))))
 971                                  value " ")
 972                       line-breaking))
 973       (setq attributes (delq '->ideograph attributes))
 974       )
 975     ;; (when (and (memq '->decomposition attributes)
 976     ;;            (setq value (get-char-attribute char '->decomposition)))
 977     ;;   (insert (format "(->decomposition\t%s)%s"
 978     ;;                   (mapconcat (lambda (code)
 979     ;;                                (cond ((symbolp code)
 980     ;;                                       (symbol-name code))
 981     ;;                                      ((characterp code)
 982     ;;                                       (if readable
 983     ;;                                           (format "%S" code)
 984     ;;                                         (format "#x%04X"
 985     ;;                                                 (char-int code))
 986     ;;                                         ))
 987     ;;                                      ((integerp code)
 988     ;;                                       (format "#x%04X" code))
 989     ;;                                      (t
 990     ;;                                       (format "%s%S" line-breaking code))))
 991     ;;                              value " ")
 992     ;;                   line-breaking))
 993     ;;   (setq attributes (delq '->decomposition attributes))
 994     ;;   )
 995     (if (equal (get-char-attribute char '->titlecase)
 996                (get-char-attribute char '->uppercase))
 997         (setq attributes (delq '->titlecase attributes)))
 998     (when (and (memq '->mojikyo attributes)
 999                (setq value (get-char-attribute char '->mojikyo)))
1000       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1001                       value (decode-char 'mojikyo value)
1002                       line-breaking))
1003       (setq attributes (delq '->mojikyo attributes))
1004       )
1005     (when (and (memq 'hanyu-dazidian-vol attributes)
1006                (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1007       (insert (format "(hanyu-dazidian-vol  . %d)%s"
1008                       value line-breaking))
1009       (setq attributes (delq 'hanyu-dazidian-vol attributes))
1010       )
1011     (when (and (memq 'hanyu-dazidian-page attributes)
1012                (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1013       (insert (format "(hanyu-dazidian-page . %d)%s"
1014                       value line-breaking))
1015       (setq attributes (delq 'hanyu-dazidian-page attributes))
1016       )
1017     (when (and (memq 'hanyu-dazidian-char attributes)
1018                (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1019       (insert (format "(hanyu-dazidian-char . %d)%s"
1020                       value line-breaking))
1021       (setq attributes (delq 'hanyu-dazidian-char attributes))
1022       )
1023     (unless readable
1024       (dolist (ignored '(composition
1025                          ->denotational <-subsumptive ->ucs-unified
1026                          ->ideographic-component-forms))
1027         (setq attributes (delq ignored attributes))))
1028     (while attributes
1029       (setq name (car attributes))
1030       (if (setq value (get-char-attribute char name))
1031           (cond ((setq ret (find-charset name))
1032                  (setq name (charset-name ret))
1033                  (if (and (not (memq name dest-ccss))
1034                           (prog1
1035                               (setq value (get-char-attribute char name))
1036                             (setq dest-ccss (cons name dest-ccss))))
1037                      (char-db-insert-ccs-feature name value line-breaking))
1038                  )
1039                 ((string-match "^=>ucs@" (symbol-name name))
1040                  (insert (format "(%-18s . #x%04X)\t; %c%s"
1041                                  name value (decode-char '=ucs value)
1042                                  line-breaking))
1043                  )
1044                 ((eq name 'jisx0208-1978/4X)
1045                  (insert (format "(%-18s . #x%04X)%s"
1046                                  name value
1047                                  line-breaking))
1048                  )
1049                 ((and
1050                   (not readable)
1051                   (not (eq name '->subsumptive))
1052                   (not (eq name '->uppercase))
1053                   (not (eq name '->lowercase))
1054                   (not (eq name '->titlecase))
1055                   (not (eq name '->canonical))
1056                   (not (eq name '->Bopomofo))
1057                   (not (eq name '->mistakable))
1058                   (not (eq name '->ideographic-variants))
1059                   (null (get-char-attribute
1060                          char (intern (format "%s*sources" name))))
1061                   (not (string-match "\\*sources$" (symbol-name name)))
1062                   (null (get-char-attribute
1063                          char (intern (format "%s*note" name))))
1064                   (not (string-match "\\*note$" (symbol-name name)))
1065                   (or (eq name '<-identical)
1066                       (eq name '<-uppercase)
1067                       (eq name '<-lowercase)
1068                       (eq name '<-titlecase)
1069                       (eq name '<-canonical)
1070                       (eq name '<-ideographic-variants)
1071                       ;; (eq name '<-synonyms)
1072                       (string-match "^<-synonyms" (symbol-name name))
1073                       (eq name '<-mistakable)
1074                       (when (string-match "^->" (symbol-name name))
1075                         (cond
1076                          ((string-match "^->fullwidth" (symbol-name name))
1077                           (not (and (consp value)
1078                                     (characterp (car value))
1079                                     (encode-char
1080                                      (car value) '=ucs 'defined-only)))
1081                           )
1082                          (t)))
1083                       ))
1084                  )
1085                 ((or (eq name 'ideographic-structure)
1086                      (eq name 'ideographic-combination)
1087                      (eq name 'ideographic-)
1088                      (eq name '=decomposition)
1089                      (string-match "^=>decomposition" (symbol-name name))
1090                      (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1091                      (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1092                                    (symbol-name name))
1093                      )
1094                  (char-db-insert-relation-feature char name value
1095                                                   line-breaking
1096                                                   ccss readable))
1097                 ((memq name '(ideograph=
1098                               original-ideograph-of
1099                               ancient-ideograph-of
1100                               vulgar-ideograph-of
1101                               wrong-ideograph-of
1102                               ;; simplified-ideograph-of
1103                               ideographic-variants
1104                               ;; ideographic-different-form-of
1105                               ))
1106                  (insert (format "(%-18s%s " name line-breaking))
1107                  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1108                        separator nil)
1109                  (while (consp value)
1110                    (setq cell (car value))
1111                    (if (and (consp cell)
1112                             (consp (car cell)))
1113                        (progn
1114                          (if separator
1115                              (insert lbs))
1116                          (char-db-insert-alist cell readable)
1117                          (setq separator lbs))
1118                      (if separator
1119                          (insert separator))
1120                      (insert (prin1-to-string cell))
1121                      (setq separator " "))
1122                    (setq value (cdr value)))
1123                  (insert ")")
1124                  (insert line-breaking))
1125                 ((consp value)
1126                  (insert (format "(%-18s " name))
1127                  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1128                        separator nil)
1129                  (while (consp value)
1130                    (setq cell (car value))
1131                    (if (and (consp cell)
1132                             (consp (car cell))
1133                             (setq ret (condition-case nil
1134                                           (find-char cell)
1135                                         (error nil))))
1136                        (progn
1137                          (setq rest cell
1138                                al nil
1139                                cal nil)
1140                          (while rest
1141                            (setq key (car (car rest)))
1142                            (if (find-charset key)
1143                                (setq cal (cons key cal))
1144                              (setq al (cons key al)))
1145                            (setq rest (cdr rest)))
1146                          (if separator
1147                              (insert lbs))
1148                          (insert-char-attributes ret
1149                                                  readable
1150                                                  al cal)
1151                          (setq separator lbs))
1152                      (setq ret (prin1-to-string cell))
1153                      (if separator
1154                          (if (< (+ (current-column)
1155                                    (length ret)
1156                                    (length separator))
1157                                 76)
1158                              (insert separator)
1159                            (insert lbs)))
1160                      (insert ret)
1161                      (setq separator " "))
1162                    (setq value (cdr value)))
1163                  (insert ")")
1164                  (insert line-breaking))
1165                 (t
1166                  (insert (format "(%-18s" name))
1167                  (setq ret (prin1-to-string value))
1168                  (unless (< (+ (current-column)
1169                                (length ret)
1170                                3)
1171                             76)
1172                    (insert line-breaking))
1173                  (insert " . " ret ")" line-breaking)
1174                  ;; (insert (format "(%-18s . %S)%s"
1175                  ;;                 name value
1176                  ;;                 line-breaking))
1177                  )
1178                 ))
1179       (setq attributes (cdr attributes)))
1180     (insert ")")))
1181
1182 (defun insert-char-data (char &optional readable
1183                               attributes)
1184   (save-restriction
1185     (narrow-to-region (point)(point))
1186     (insert "(define-char
1187   '")
1188     (insert-char-attributes char readable attributes)
1189     (insert ")\n")
1190     (goto-char (point-min))
1191     (while (re-search-forward "[ \t]+$" nil t)
1192       (replace-match ""))
1193     ;; from tabify.
1194     (goto-char (point-min))
1195     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1196       (let ((column (current-column))
1197             (indent-tabs-mode t))
1198         (delete-region (match-beginning 0) (point))
1199         (indent-to column)))
1200     (goto-char (point-max))
1201     ;; (tabify (point-min)(point-max))
1202     ))
1203
1204 (defun insert-char-data-with-variant (char &optional printable
1205                                            no-ucs-unified
1206                                            script excluded-script)
1207   (insert-char-data char printable)
1208   (let ((variants (char-variants char))
1209         rest
1210         variant vs ret)
1211     (setq variants (sort variants #'<))
1212     (setq rest variants)
1213     (setq variants (cons char variants))
1214     (while rest
1215       (setq variant (car rest))
1216       (unless (get-char-attribute variant '<-subsumptive)
1217         (if (and (or (null script)
1218                      (null (setq vs (get-char-attribute variant 'script)))
1219                      (memq script vs))
1220                  (or (null excluded-script)
1221                      (null (setq vs (get-char-attribute variant 'script)))
1222                      (not (memq excluded-script vs))))
1223             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1224               (insert-char-data variant printable)
1225               (if (setq ret (char-variants variant))
1226                   (while ret
1227                     (or (memq (car ret) variants)
1228                         ;; (get-char-attribute (car ret) '<-subsumptive)
1229                         (setq rest (nconc rest (list (car ret)))))
1230                     (setq ret (cdr ret)))))))
1231       (setq rest (cdr rest)))))
1232
1233 (defun insert-char-range-data (min max &optional script excluded-script)
1234   (let ((code min)
1235         char)
1236     (while (<= code max)
1237       (setq char (decode-char '=ucs code))
1238       (if (encode-char char '=ucs 'defined-only)
1239           (insert-char-data-with-variant char nil 'no-ucs-unified
1240                                          script excluded-script))
1241       (setq code (1+ code)))))
1242
1243 (defun write-char-range-data-to-file (min max file
1244                                           &optional script excluded-script)
1245   (let ((coding-system-for-write char-db-file-coding-system))
1246     (with-temp-buffer
1247       (insert (format ";; -*- coding: %s -*-\n"
1248                       char-db-file-coding-system))
1249       (insert-char-range-data min max script excluded-script)
1250       (write-region (point-min)(point-max) file))))
1251
1252 (defvar what-character-original-window-configuration)
1253
1254 ;;;###autoload
1255 (defun what-char-definition (char)
1256   (interactive (list (char-after)))
1257   (let ((buf (get-buffer-create "*Character Description*"))
1258         (the-buf (current-buffer))
1259         (win-conf (current-window-configuration)))
1260     (pop-to-buffer buf)
1261     (make-local-variable 'what-character-original-window-configuration)
1262     (setq what-character-original-window-configuration win-conf)
1263     (setq buffer-read-only nil)
1264     (erase-buffer)
1265     (condition-case err
1266         (progn
1267           (insert-char-data-with-variant char 'printable)
1268           (unless (char-attribute-alist char)
1269             (insert (format ";; = %c\n"
1270                             (let* ((rest (split-char char))
1271                                    (ccs (pop rest))
1272                                    (code (pop rest)))
1273                               (while rest
1274                                 (setq code (logior (lsh code 8)
1275                                                    (pop rest))))
1276                               (decode-char ccs code)))))
1277           ;; (char-db-update-comment)
1278           (set-buffer-modified-p nil)
1279           (view-mode the-buf (lambda (buf)
1280                                (set-window-configuration
1281                                 what-character-original-window-configuration)
1282                                ))
1283           (goto-char (point-min)))
1284       (error (progn
1285                (set-window-configuration
1286                 what-character-original-window-configuration)
1287                (signal (car err) (cdr err)))))))
1288
1289
1290 ;;; @ end
1291 ;;;
1292
1293 (provide 'char-db-util)
1294
1295 ;;; char-db-util.el ends here