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, 2012, 2013 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 ?𠬜 ?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 ;  ?旨 ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?喜 ?壴 ; 160
  84    ?鼓 ?豈 ?豆 ?豊 ?豐 ?䖒 ?虍 ?虎 ?虤 ?皿 ; 170
  85    ?𠙴 ?去 ?血 ?丶 ?丹 ?青 ?井 ?皀 ?鬯 ?食 ; 180
  86    ?亼 ?會 ?倉 ?入 ?缶 ?矢 ?高 ?冂 ?𩫏 ?京 ; 190
  87    ?亯 ?𣆪 ?畗 ?㐭 ?嗇 ?來 ?麥 ?夊 ?舛 ?䑞 ; 200
  88    ?韋 ?弟 ?夂 ?久 ?桀 ?木 ?東 ?林 ?才 ?叒 ; 210
  89    ?之 ?帀 ?出 ?𣎵 ?生 ?乇 ?𠂹 ?𠌶 ?華 ?𥝌 ; 220
  90    ?稽 ?巢 ?桼 ?束 ?㯻 ?囗 ?員 ?貝 ?邑 ?𨛜 ; 230
  91    ?日 ?旦 ?倝 ?㫃 ?冥 ?晶 ?月 ?有 ?明 ?囧 ; 240
  92    ?夕 ?多 ?毌 ?𢎘 ?𣐺 ?卣 ?齊 ?朿 ?片 ?鼎 ; 250
  93    ?克 ?彔 ?禾 ?秝 ?黍 ?香 ?米 ?毇 ?臼 ?凶 ; 260
  94    ?𣎳 ?林 ?麻 ?尗 ?耑 ?韭 ?瓜 ?瓠 ?宀 ?宮 ; 270
  95    ?呂 ?穴 ?㝱 ?𤕫 ?冖 ?𠔼 ?冃 ?㒳 ?网 ?襾 ; 280
  96    ?巾 ?巿 ?帛 ?白 ?㡀 ?黹 ?人 ?𠤎 ?匕 ?从 ; 290
  97    ?比 ?北 ?丘 ?㐺 ?𡈼 ?重 ?臥 ?身 ?㐆 ?衣 ; 300
  98    ])
  99
 100 (defun shuowen-radical (number)
 101   (aref shuowen-radicals (1- number)))
 102
 103 (defvar char-db-file-coding-system 'utf-8-mcs-er)
 104
 105 (defvar char-db-ignored-attributes '(ideographic-products))
 106
 107 (defvar char-db-coded-charset-priority-list
 108   '(ascii
 109     control-1
 110     latin-iso8859-1
 111     latin-iso8859-2
 112     latin-iso8859-3
 113     latin-iso8859-4
 114     latin-iso8859-9
 115     latin-jisx0201
 116     cyrillic-iso8859-5
 117     greek-iso8859-7
 118     thai-tis620
 119     =adobe-japan1-0
 120     =adobe-japan1-1
 121     =adobe-japan1-2
 122     =adobe-japan1-3
 123     =adobe-japan1-4
 124     =adobe-japan1-5
 125     =adobe-japan1-6
 126     =jis-x0208
 127     =jis-x0208@1978
 128     =jis-x0208@1983
 129     japanese-jisx0212
 130     chinese-gb2312
 131     =jis-x0208@1990
 132     chinese-cns11643-1
 133     chinese-cns11643-2
 134     chinese-cns11643-3
 135     chinese-cns11643-4
 136     chinese-cns11643-5
 137     chinese-cns11643-6
 138     chinese-cns11643-7
 139     =jis-x0213-1
 140     =jis-x0213-1@2000
 141     =jis-x0213-1@2004
 142     =jis-x0213-2
 143     korean-ksc5601
 144     chinese-isoir165
 145     katakana-jisx0201
 146     hebrew-iso8859-8
 147     chinese-gb12345
 148     latin-viscii
 149     ethiopic-ucs
 150     =big5-cdp
 151     =hanyo-denshi/ja
 152     =hanyo-denshi/jb
 153     =hanyo-denshi/jc
 154     =hanyo-denshi/jd
 155     =hanyo-denshi/ft
 156     =hanyo-denshi/ia
 157     =hanyo-denshi/ib
 158     =hanyo-denshi/hg
 159     =hanyo-denshi/jt
 160     =hanyo-denshi/ks
 161     =gt
 162     =gt-k
 163     ==adobe-japan1-0
 164     ==adobe-japan1-1
 165     ==adobe-japan1-2
 166     ==adobe-japan1-3
 167     ==adobe-japan1-4
 168     ==adobe-japan1-5
 169     ==adobe-japan1-6
 170     ==jis-x0208
 171     ==jis-x0213-1
 172     ==jis-x0213-2
 173     ==hanyo-denshi/ja
 174     ==hanyo-denshi/jb
 175     ==hanyo-denshi/jc
 176     ==hanyo-denshi/ft
 177     ==hanyo-denshi/ib
 178     ==hanyo-denshi/hg
 179     ==hanyo-denshi/jt
 180     ==hanyo-denshi/ks
 181     =daikanwa
 182     =daikanwa@rev2
 183     =daikanwa@rev1
 184     ==daikanwa
 185     =cbeta
 186     ideograph-hanziku-1
 187     ideograph-hanziku-2
 188     ideograph-hanziku-3
 189     ideograph-hanziku-4
 190     ideograph-hanziku-5
 191     ideograph-hanziku-6
 192     ideograph-hanziku-7
 193     ideograph-hanziku-8
 194     ideograph-hanziku-9
 195     ideograph-hanziku-10
 196     ideograph-hanziku-11
 197     ideograph-hanziku-12
 198     ;; =>>>adobe-japan1-0
 199     ;; =>>>adobe-japan1-1
 200     ;; =>>>adobe-japan1-2
 201     ;; =>>>adobe-japan1-3
 202     ;; =>>>adobe-japan1-4
 203     ;; =>>>adobe-japan1-5
 204     ;; =>>>adobe-japan1-6
 205     ;; =>>>jis-x0208
 206     ;; =>>>jis-x0213-1
 207     ;; =>>>jis-x0213-2
 208     ;; =>>>hanyo-denshi/ja
 209     ;; =>>>hanyo-denshi/jb
 210     ;; =>>>hanyo-denshi/jc
 211     ;; =>>>hanyo-denshi/ft
 212     ;; =>>>hanyo-denshi/ib
 213     ;; =>>>hanyo-denshi/hg
 214     ;; =>>>hanyo-denshi/jt
 215     ;; =>>>hanyo-denshi/ks
 216     ;; =>>>gt
 217     =>>adobe-japan1-0
 218     =>>adobe-japan1-1
 219     =>>adobe-japan1-2
 220     =>>adobe-japan1-3
 221     =>>adobe-japan1-4
 222     =>>adobe-japan1-5
 223     =>>adobe-japan1-6
 224     =>>jis-x0208
 225     =>>jis-x0213-1
 226     =>>jis-x0213-1@2000
 227     =>>jis-x0213-1@2004
 228     =>>jis-x0213-2
 229     =>>jis-x0208@1978
 230     =>>hanyo-denshi/ft
 231     =>>hanyo-denshi/jt
 232     =>>hanyo-denshi/ks
 233     =>>gt
 234     =>>daikanwa
 235     =>>cbeta
 236     =+>jis-x0208
 237     =+>jis-x0213-1
 238     =+>jis-x0213-2
 239     =+>adobe-japan1-0
 240     =+>adobe-japan1-1
 241     =+>adobe-japan1-2
 242     =+>adobe-japan1-3
 243     =+>adobe-japan1-4
 244     =+>adobe-japan1-5
 245     =+>adobe-japan1-6
 246     =+>jis-x0208@1978
 247     =>jis-x0208
 248     =>jis-x0208@1997
 249     =>jis-x0213-1
 250     =>jis-x0213-1@2000
 251     =>jis-x0213-1@2004
 252     =>jis-x0213-2
 253     ==>ucs@bucs
 254     =>ucs@iso
 255     =>ucs@unicode
 256     =>ucs@jis
 257     =>ucs@cns
 258     =>ucs@ks
 259     =+>ucs@iso
 260     =+>ucs@unicode
 261     =+>ucs@jis
 262     =+>ucs@jis/1990
 263     =+>ucs@cns
 264     =+>ucs@ks
 265     =>>ucs@iso
 266     =>>ucs@unicode
 267     =>>ucs@jis
 268     =>>ucs@cns
 269     =>>>ucs@iso
 270     =>>>ucs@unicode
 271     ==ucs@iso
 272     ==ucs@unicode
 273     ==gb2312
 274     ==ks-x1001
 275     ==cns11643-1
 276     ==cns11643-2
 277     ==cns11643-3
 278     ==gt
 279     ==jis-x0208@1990
 280     ==gt-k
 281     =ucs@iso
 282     =ucs@unicode
 283     =ucs@cns
 284     =>>big5-cdp
 285     =>>gt-k
 286     =+>gt
 287     =>gt
 288     =>big5-cdp
 289     =>daikanwa
 290     =>daikanwa/ho
 291     =>cns11643-7
 292     =big5
 293     =big5-eten
 294     =>gt-k
 295     =zinbun-oracle
 296     =>zinbun-oracle
 297     =ruimoku-v6
 298     =>>ruimoku-v6
 299     =jef-china3
 300     =shinjigen
 301     =big5-cdp-var-3
 302     =big5-cdp-var-5))
 303
 304
 305 ;;; @ char-db formatters
 306 ;;;
 307
 308 (defun char-db-make-char-spec (char)
 309   (let (ret char-spec)
 310     (cond ((characterp char)
 311            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 312                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 313                   (setq char-spec (list (cons '=ucs ret)))
 314                   (cond ((setq ret (get-char-attribute char 'name))
 315                          (setq char-spec (cons (cons 'name ret) char-spec))
 316                          )
 317                         ((setq ret (get-char-attribute char 'name*))
 318                          (setq char-spec (cons (cons 'name* ret) char-spec))
 319                          ))
 320                   )
 321                  ((encode-char char '=adobe-japan1 'defined-only)
 322                   (setq char-spec nil)
 323                   (dolist (ccs (charset-list))
 324                     (if (and (or (memq ccs
 325                                        '(=adobe-japan1-0
 326                                          =adobe-japan1-1
 327                                          =adobe-japan1-2
 328                                          =adobe-japan1-3
 329                                          =adobe-japan1-4
 330                                          =adobe-japan1-5
 331                                          =adobe-japan1-6
 332                                          ))
 333                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 334                                  (string-match "=ucs@" (symbol-name ccs))
 335                                  )
 336                              (setq ccs (charset-name ccs))
 337                              (null (assq ccs char-spec))
 338                              (setq ret (encode-char char ccs 'defined-only)))
 339                         (setq char-spec (cons (cons ccs ret) char-spec))))
 340                   )
 341                  ((encode-char char '==adobe-japan1 'defined-only)
 342                   (setq char-spec nil)
 343                   (dolist (ccs (charset-list))
 344                     (if (and (or (memq ccs
 345                                        '(==adobe-japan1-0
 346                                          ==adobe-japan1-1
 347                                          ==adobe-japan1-2
 348                                          ==adobe-japan1-3
 349                                          ==adobe-japan1-4
 350                                          ==adobe-japan1-5
 351                                          ==adobe-japan1-6
 352                                          ))
 353                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 354                                  (string-match "=ucs@" (symbol-name ccs))
 355                                  )
 356                              (setq ccs (charset-name ccs))
 357                              (null (assq ccs char-spec))
 358                              (setq ret (encode-char char ccs 'defined-only)))
 359                         (setq char-spec (cons (cons ccs ret) char-spec))))
 360                   )
 361                  ((setq ret
 362                         (catch 'tag
 363                           (let ((rest char-db-coded-charset-priority-list)
 364                                 ccs)
 365                             (while rest
 366                               (setq ccs (charset-name
 367                                          (find-charset (car rest))))
 368                               (if (setq ret
 369                                         (encode-char char ccs
 370                                                      'defined-only))
 371                                   (throw 'tag (cons ccs ret)))
 372                               (setq rest (cdr rest))))))
 373                   (setq char-spec (list ret))
 374                   (dolist (ccs (delq (car ret) (charset-list)))
 375                     (if (and (or (charset-iso-final-char ccs)
 376                                  (memq ccs
 377                                        '(=daikanwa
 378                                          =daikanwa@rev2
 379                                          ;; =gt-k
 380                                          =jis-x0208@1997
 381                                          ))
 382                                  (eq (charset-property ccs 'iso-ir) 177)
 383                                  ;; (string-match "=ucs@" (symbol-name ccs))
 384                                  )
 385                              (setq ccs (charset-name ccs))
 386                              (null (assq ccs char-spec))
 387                              (setq ret (encode-char char ccs 'defined-only)))
 388                         (setq char-spec (cons (cons ccs ret) char-spec))))
 389                   (if (null char-spec)
 390                       (setq char-spec (split-char char)))
 391                   (cond ((setq ret (get-char-attribute char 'name))
 392                          (setq char-spec (cons (cons 'name ret) char-spec))
 393                          )
 394                         ((setq ret (get-char-attribute char 'name*))
 395                          (setq char-spec (cons (cons 'name* ret) char-spec))
 396                          ))
 397                   )
 398                  ((setq ret (get-char-attribute
 399                              char 'ideographic-combination))
 400                   (setq char-spec
 401                         (cons (cons 'ideographic-combination ret)
 402                               char-spec))
 403                   ))
 404            char-spec)
 405           ((consp char)
 406            char))))
 407
 408 (defun char-db-insert-char-spec (char &optional readable column
 409                                       required-features)
 410   (unless column
 411     (setq column (current-column)))
 412   (let (char-spec temp-char)
 413     (setq char-spec (char-db-make-char-spec char))
 414     (unless (or (characterp char) ; char
 415                 (condition-case nil
 416                     (setq char (find-char char-spec))
 417                   (error nil)))
 418       ;; define temporary character
 419       ;;   Current implementation is dirty.
 420       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 421                                          char-spec)))
 422       (remove-char-attribute temp-char 'ideograph-daikanwa)
 423       (setq char temp-char))
 424     (insert-char-attributes char
 425                             readable
 426                             (union (mapcar #'car char-spec)
 427                                    required-features)
 428                             nil 'for-sub-node)
 429     (when temp-char
 430       ;; undefine temporary character
 431       ;;   Current implementation is dirty.
 432       (setq char-spec (char-attribute-alist temp-char))
 433       (while char-spec
 434         (remove-char-attribute temp-char (car (car char-spec)))
 435         (setq char-spec (cdr char-spec))))))
 436
 437 (defun char-db-insert-alist (alist &optional readable column)
 438   (unless column
 439     (setq column (current-column)))
 440   (let ((line-breaking
 441          (concat "\n" (make-string (1+ column) ?\ )))
 442         name value
 443         ret al ; cal
 444         key
 445         lbs cell rest separator)
 446     (insert "(")
 447     (while alist
 448       (setq name (car (car alist))
 449             value (cdr (car alist)))
 450       (cond ((eq name 'char)
 451              (insert "(char . ")
 452              (if (setq ret (condition-case nil
 453                                (find-char value)
 454                              (error nil)))
 455                  (progn
 456                    (setq al nil
 457                          ;; cal nil
 458                          )
 459                    (while value
 460                      (setq key (car (car value)))
 461                      ;; (if (find-charset key)
 462                      ;;     (setq cal (cons key cal))
 463                      (setq al (cons key al))
 464                      ;; )
 465                      (setq value (cdr value)))
 466                    (insert-char-attributes ret
 467                                            readable
 468                                            (or al 'none) ; cal
 469                                            nil 'for-sub-node))
 470                (insert (prin1-to-string value)))
 471              (insert ")")
 472              (insert line-breaking))
 473             ((consp value)
 474              (insert (format "(%-18s " name))
 475              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 476              (while (consp value)
 477                (setq cell (car value))
 478                (if (and (consp cell)
 479                         (consp (car cell))
 480                         (setq ret (condition-case nil
 481                                       (find-char cell)
 482                                     (error nil)))
 483                         )
 484                    (progn
 485                      (setq rest cell
 486                            al nil
 487                            ;; cal nil
 488                            )
 489                      (while rest
 490                        (setq key (car (car rest)))
 491                        ;; (if (find-charset key)
 492                        ;;     (setq cal (cons key cal))
 493                        (setq al (cons key al))
 494                        ;; )
 495                        (setq rest (cdr rest)))
 496                      (if separator
 497                          (insert lbs))
 498                      (insert-char-attributes ret
 499                                              readable
 500                                              al ; cal
 501                                              nil 'for-sub-node)
 502                      (setq separator lbs))
 503                  (if separator
 504                      (insert separator))
 505                  (insert (prin1-to-string cell))
 506                  (setq separator " "))
 507                (setq value (cdr value)))
 508              (insert ")")
 509              (insert line-breaking))
 510             (t
 511              (insert (format "(%-18s . %S)%s"
 512                              name value
 513                              line-breaking))))
 514       (setq alist (cdr alist))))
 515   (insert ")"))
 516
 517 (defun char-db-insert-char-reference (plist &optional readable column)
 518   (unless column
 519     (setq column (current-column)))
 520   (let ((line-breaking
 521          (concat "\n" (make-string (1+ column) ?\ )))
 522         (separator "")
 523         name value)
 524     (insert "(")
 525     (while plist
 526       (setq name (pop plist))
 527       (setq value (pop plist))
 528       (cond ((eq name :char)
 529              (insert separator)
 530              (insert ":char\t")
 531              (cond ((numberp value)
 532                     (setq value (decode-char '=ucs value)))
 533                    ;; ((consp value)
 534                    ;;  (setq value (or (find-char value)
 535                    ;;                  value)))
 536                    )
 537              (char-db-insert-char-spec value readable)
 538              (insert line-breaking)
 539              (setq separator ""))
 540             ((eq name :radical)
 541              (insert (format "%s%s\t%d ; %c%s"
 542                              separator
 543                              name value
 544                              (ideographic-radical value)
 545                              line-breaking))
 546              (setq separator ""))
 547             (t
 548              (insert (format "%s%s\t%S" separator name value))
 549              (setq separator line-breaking)))
 550       ))
 551   (insert ")"))
 552
 553 (defun char-db-decode-isolated-char (ccs code-point)
 554   (let (ret)
 555     (setq ret
 556           (cond ((eq ccs 'arabic-iso8859-6)
 557                  (decode-char ccs code-point))
 558                 ;; ((eq ccs '=gt)
 559                 ;;  (decode-builtin-char '==gt code-point))
 560                 ((and (memq ccs '(=gt-pj-1
 561                                   =gt-pj-2
 562                                   =gt-pj-3
 563                                   =gt-pj-4
 564                                   =gt-pj-5
 565                                   =gt-pj-6
 566                                   =gt-pj-7
 567                                   =gt-pj-8
 568                                   =gt-pj-9
 569                                   =gt-pj-10
 570                                   =gt-pj-11))
 571                       (setq ret (decode-char ccs code-point))
 572                       (setq ret (encode-char ret '=gt 'defined-only)))
 573                  (decode-builtin-char '=gt ret))
 574                 (t
 575                  (decode-builtin-char ccs code-point))))
 576     (cond ((and (<= 0 (char-int ret))
 577                 (<= (char-int ret) #x1F))
 578            (decode-char '=ucs (+ #x2400 (char-int ret))))
 579           ((= (char-int ret) #x7F)
 580            ?\u2421)
 581           (t ret))))
 582
 583 (defvar char-db-convert-obsolete-format t)
 584
 585 (defun char-db-insert-ccs-feature (name value line-breaking)
 586   (cond
 587    ((integerp value)
 588     (insert
 589      (format
 590       (cond
 591        ((memq name '(==shinjigen
 592                      =shinjigen
 593                      =shinjigen@1ed
 594                      =shinjigen@rev ==shinjigen@rev
 595                      =shinjigen/+p@rev
 596                      ==daikanwa/ho
 597                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 598         "(%-18s .  %04d)\t; %c")
 599        ((eq name '=shinjigen@1ed/24pr)
 600         "(%-18s . %04d)\t; %c")
 601        ((or (memq name '(==daikanwa
 602                          =daikanwa =>>daikanwa =>daikanwa
 603                          =daikanwa@rev1 =daikanwa@rev2
 604                          =daikanwa/+p ==daikanwa/+p =>>daikanwa/+p
 605                          =daikanwa/+2p =>>daikanwa/+2p
 606                          =gt ==gt ; =>>>gt
 607                          =>>gt =+>gt =>gt
 608                          =gt-k ==gt-k =>>gt-k =>gt-k
 609                          =adobe-japan1-0 ==adobe-japan1-0 ; =>>>adobe-japan1-0
 610                          =adobe-japan1-1 ==adobe-japan1-1 ; =>>>adobe-japan1-1
 611                          =adobe-japan1-2 ==adobe-japan1-2 ; =>>>adobe-japan1-2
 612                          =adobe-japan1-3 ==adobe-japan1-3 ; =>>>adobe-japan1-3
 613                          =adobe-japan1-4 ==adobe-japan1-4 ; =>>>adobe-japan1-4
 614                          =adobe-japan1-5 ==adobe-japan1-5 ; =>>>adobe-japan1-5
 615                          =adobe-japan1-6 ==adobe-japan1-6 ; =>>>adobe-japan1-6
 616                          =>>adobe-japan1-0 =+>adobe-japan1-0
 617                          =>>adobe-japan1-1 =+>adobe-japan1-1
 618                          =>>adobe-japan1-2 =+>adobe-japan1-2
 619                          =>>adobe-japan1-3 =+>adobe-japan1-3
 620                          =>>adobe-japan1-4 =+>adobe-japan1-4
 621                          =>>adobe-japan1-5 =+>adobe-japan1-5
 622                          =>>adobe-japan1-6 =+>adobe-japan1-6
 623                          =cbeta =>>cbeta
 624                          =zinbun-oracle =>zinbun-oracle))
 625             ;; (string-match "^=adobe-" (symbol-name name))
 626             )
 627         "(%-18s . %05d)\t; %c")
 628        ((memq name '(=hanyo-denshi/ks
 629                      ==hanyo-denshi/ks ; =>>>hanyo-denshi/ks
 630                      =>>hanyo-denshi/ks
 631                      =zihai mojikyo))
 632         "(%-18s . %06d)\t; %c")
 633        ((>= (charset-dimension name) 2)
 634         "(%-18s . #x%04X)\t; %c")
 635        (t
 636         "(%-18s . #x%02X)\t; %c"))
 637       name
 638       (if (= (charset-iso-graphic-plane name) 1)
 639           (logior value
 640                   (cond ((= (charset-dimension name) 1)
 641                          #x80)
 642                         ((= (charset-dimension name) 2)
 643                          #x8080)
 644                         ((= (charset-dimension name) 3)
 645                          #x808080)
 646                         (t 0)))
 647         value)
 648       (char-db-decode-isolated-char name value)))
 649     (if (and (= (charset-chars name) 94)
 650              (= (charset-dimension name) 2))
 651         (insert (format " [%02d-%02d]"
 652                         (- (lsh value -8) 32)
 653                         (- (logand value 255) 32))))
 654     )
 655    (t
 656     (insert (format "(%-18s . %s)" name value))
 657     ))
 658   (insert line-breaking))
 659
 660 (defun char-db-insert-relation-feature (char name value line-breaking
 661                                              ccss readable)
 662   (insert (format "(%-18s%s " name line-breaking))
 663   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 664         separator cell sources required-features
 665         ret)
 666     (while (consp value)
 667       (setq cell (car value))
 668       (if (integerp cell)
 669           (setq cell (decode-char '=ucs cell)))
 670       (cond
 671        ((eq name '->subsumptive)
 672         (when (or (not (some (lambda (atr)
 673                                (get-char-attribute cell atr))
 674                              char-db-ignored-attributes))
 675                   (some (lambda (ccs)
 676                           (encode-char cell ccs 'defined-only))
 677                         ccss))
 678           (if separator
 679               (insert lbs))
 680           (let ((char-db-ignored-attributes
 681                  (cons '<-subsumptive
 682                        char-db-ignored-attributes)))
 683             (insert-char-attributes cell readable nil nil 'for-sub-node))
 684           (setq separator lbs))
 685         )
 686        ((characterp cell)
 687         (setq sources
 688               (get-char-attribute
 689                char (intern (format "%s*sources" name))))
 690         (setq required-features nil)
 691         (dolist (source sources)
 692           (cond
 693            ((memq source '(JP
 694                            JP/Jouyou
 695                            shinjigen shinjigen@1ed shinjigen@rev))
 696             (setq required-features
 697                   (union required-features
 698                          '(=jis-x0208
 699                            =jis-x0208@1990
 700                            =jis-x0213-1@2000
 701                            =jis-x0213-1@2004
 702                            =jis-x0213-2
 703                            =jis-x0212
 704                            =jis-x0208@1983
 705                            =jis-x0208@1978
 706                            =shinjigen
 707                            =shinjigen@1ed
 708                            =shinjigen@rev
 709                            =shinjigen/+p@rev))))
 710            ((eq source 'CN)
 711             (setq required-features
 712                   (union required-features
 713                          '(=gb2312
 714                            =gb12345
 715                            =iso-ir165)))))
 716           (cond
 717            ((find-charset (setq ret (intern (format "=%s" source))))
 718             (setq required-features
 719                   (cons ret required-features)))
 720            (t (setq required-features
 721                     (cons source required-features)))))
 722         (cond ((string-match "@JP" (symbol-name name))
 723                (setq required-features
 724                      (union required-features
 725                             '(=jis-x0208
 726                               =jis-x0208@1990
 727                               =jis-x0213-1-2000
 728                               =jis-x0213-2-2000
 729                               =jis-x0212
 730                               =jis-x0208@1983
 731                               =jis-x0208@1978))))
 732               ((string-match "@CN" (symbol-name name))
 733                (setq required-features
 734                      (union required-features
 735                             '(=gb2312
 736                               =gb12345
 737                               =iso-ir165)))))
 738         (if separator
 739             (insert lbs))
 740         (if readable
 741             (insert (format "%S" cell))
 742           (char-db-insert-char-spec cell readable
 743                                     nil
 744                                     required-features))
 745         (setq separator lbs))
 746        ((consp cell)
 747         (if separator
 748             (insert lbs))
 749         (if (consp (car cell))
 750             (char-db-insert-char-spec cell readable)
 751           (char-db-insert-char-reference cell readable))
 752         (setq separator lbs))
 753        (t
 754         (if separator
 755             (insert separator))
 756         (insert (prin1-to-string cell))
 757         (setq separator " ")))
 758       (setq value (cdr value)))
 759     (insert ")")
 760     (insert line-breaking)))
 761
 762 (defun insert-char-attributes (char &optional readable attributes column
 763                                     for-sub-node)
 764   (unless column
 765     (setq column (current-column)))
 766   (let (name value ; has-long-ccs-name
 767         rest
 768         radical strokes
 769         (line-breaking
 770          (concat "\n" (make-string (1+ column) ?\ )))
 771         lbs cell separator ret
 772         key al cal
 773         dest-ccss ; sources required-features
 774         ccss)
 775     (let (atr-d)
 776       (setq attributes
 777             (sort (if attributes
 778                       (if (consp attributes)
 779                           (progn
 780                             (dolist (name attributes)
 781                               (unless (memq name char-db-ignored-attributes)
 782                                 (if (find-charset name)
 783                                     (push name ccss))
 784                                 (push name atr-d)))
 785                             atr-d))
 786                     (dolist (name (char-attribute-list))
 787                       (unless (memq name char-db-ignored-attributes)
 788                         (if (find-charset name)
 789                             (push name ccss))
 790                         (push name atr-d)))
 791                     atr-d)
 792                   #'char-attribute-name<)))
 793     (insert "(")
 794     (when (memq '<-subsumptive attributes)
 795       (when (or readable (not for-sub-node))
 796         (when (setq value (get-char-attribute char '<-subsumptive))
 797           (char-db-insert-relation-feature char '<-subsumptive value
 798                                            line-breaking
 799                                            ccss readable)))
 800       (setq attributes (delq '<-subsumptive attributes)))
 801     (when (and (memq '<-denotational attributes)
 802                (setq value (get-char-attribute char '<-denotational)))
 803       (char-db-insert-relation-feature char '<-denotational value
 804                                        line-breaking
 805                                        ccss readable)
 806       (setq attributes (delq '<-denotational attributes)))
 807     (when (and (memq 'name attributes)
 808                (setq value (get-char-attribute char 'name)))
 809       (insert (format
 810                (if (> (+ (current-column) (length value)) 48)
 811                    "(name . %S)%s"
 812                  "(name               . %S)%s")
 813                value line-breaking))
 814       (setq attributes (delq 'name attributes))
 815       )
 816     (when (and (memq 'name* attributes)
 817                (setq value (get-char-attribute char 'name*)))
 818       (insert (format
 819                (if (> (+ (current-column) (length value)) 48)
 820                    "(name* . %S)%s"
 821                  "(name*              . %S)%s")
 822                value line-breaking))
 823       (setq attributes (delq 'name* attributes))
 824       )
 825     (when (and (memq 'script attributes)
 826                (setq value (get-char-attribute char 'script)))
 827       (insert (format "(script\t\t%s)%s"
 828                       (mapconcat (function prin1-to-string)
 829                                  value " ")
 830                       line-breaking))
 831       (setq attributes (delq 'script attributes))
 832       )
 833     (dolist (name '(=>ucs =>ucs*))
 834       (when (and (memq name attributes)
 835                  (setq value (get-char-attribute char name)))
 836         (insert (format "(%-18s . #x%04X)\t; %c%s"
 837                         name value (decode-char '=ucs value)
 838                         line-breaking))
 839         (setq attributes (delq name attributes))))
 840     (dolist (name '(=>ucs@gb =>ucs@big5))
 841       (when (and (memq name attributes)
 842                  (setq value (get-char-attribute char name)))
 843         (insert (format "(%-18s . #x%04X)\t; %c%s"
 844                         name value
 845                         (decode-char (intern
 846                                       (concat "="
 847                                               (substring
 848                                                (symbol-name name) 2)))
 849                                      value)
 850                         line-breaking))
 851         (setq attributes (delq name attributes))
 852         ))
 853     ;; (dolist (name '(=>daikanwa))
 854     ;;   (when (and (memq name attributes)
 855     ;;              (setq value (get-char-attribute char name)))
 856     ;;     (insert
 857     ;;      (if (integerp value)
 858     ;;          (format "(%-18s . %05d)\t; %c%s"
 859     ;;                  name value (decode-char '=daikanwa value)
 860     ;;                  line-breaking)
 861     ;;        (format "(%-18s %s)\t; %c%s"
 862     ;;                name
 863     ;;                (mapconcat (function prin1-to-string)
 864     ;;                           value " ")
 865     ;;                (char-representative-of-daikanwa char)
 866     ;;                line-breaking)))
 867     ;;     (setq attributes (delq name attributes))))
 868     (when (and (memq 'general-category attributes)
 869                (setq value (get-char-attribute char 'general-category)))
 870       (insert (format
 871                "(general-category\t%s) ; %s%s"
 872                (mapconcat (lambda (cell)
 873                             (format "%S" cell))
 874                           value " ")
 875                (cond ((rassoc value unidata-normative-category-alist)
 876                       "Normative Category")
 877                      ((rassoc value unidata-informative-category-alist)
 878                       "Informative Category")
 879                      (t
 880                       "Unknown Category"))
 881                line-breaking))
 882       (setq attributes (delq 'general-category attributes))
 883       )
 884     (when (and (memq 'bidi-category attributes)
 885                (setq value (get-char-attribute char 'bidi-category)))
 886       (insert (format "(bidi-category\t. %S)%s"
 887                       value
 888                       line-breaking))
 889       (setq attributes (delq 'bidi-category attributes))
 890       )
 891     (unless (or (not (memq 'mirrored attributes))
 892                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
 893                     'empty))
 894       (insert (format "(mirrored\t\t. %S)%s"
 895                       value
 896                       line-breaking))
 897       (setq attributes (delq 'mirrored attributes))
 898       )
 899     (cond
 900      ((and (memq 'decimal-digit-value attributes)
 901            (setq value (get-char-attribute char 'decimal-digit-value)))
 902       (insert (format "(decimal-digit-value . %S)%s"
 903                       value
 904                       line-breaking))
 905       (setq attributes (delq 'decimal-digit-value attributes))
 906       (when (and (memq 'digit-value attributes)
 907                  (setq value (get-char-attribute char 'digit-value)))
 908         (insert (format "(digit-value\t . %S)%s"
 909                         value
 910                         line-breaking))
 911         (setq attributes (delq 'digit-value attributes))
 912         )
 913       (when (and (memq 'numeric-value attributes)
 914                  (setq value (get-char-attribute char 'numeric-value)))
 915         (insert (format "(numeric-value\t . %S)%s"
 916                         value
 917                         line-breaking))
 918         (setq attributes (delq 'numeric-value attributes))
 919         )
 920       )
 921      (t
 922       (when (and (memq 'digit-value attributes)
 923                  (setq value (get-char-attribute char 'digit-value)))
 924         (insert (format "(digit-value\t. %S)%s"
 925                         value
 926                         line-breaking))
 927         (setq attributes (delq 'digit-value attributes))
 928         )
 929       (when (and (memq 'numeric-value attributes)
 930                  (setq value (get-char-attribute char 'numeric-value)))
 931         (insert (format "(numeric-value\t. %S)%s"
 932                         value
 933                         line-breaking))
 934         (setq attributes (delq 'numeric-value attributes))
 935         )))
 936     (when (and (memq 'iso-10646-comment attributes)
 937                (setq value (get-char-attribute char 'iso-10646-comment)))
 938       (insert (format "(iso-10646-comment\t. %S)%s"
 939                       value
 940                       line-breaking))
 941       (setq attributes (delq 'iso-10646-comment attributes))
 942       )
 943     (when (and (memq 'morohashi-daikanwa attributes)
 944                (setq value (get-char-attribute char 'morohashi-daikanwa)))
 945       (insert (format "(morohashi-daikanwa\t%s)%s"
 946                       (mapconcat (function prin1-to-string) value " ")
 947                       line-breaking))
 948       (setq attributes (delq 'morohashi-daikanwa attributes))
 949       )
 950     (setq radical nil
 951           strokes nil)
 952     (when (and (memq 'ideographic-radical attributes)
 953                (setq value (get-char-attribute char 'ideographic-radical)))
 954       (setq radical value)
 955       (insert (format "(ideographic-radical . %S)\t; %c%s"
 956                       radical
 957                       (ideographic-radical radical)
 958                       line-breaking))
 959       (setq attributes (delq 'ideographic-radical attributes))
 960       )
 961     (when (and (memq 'shuowen-radical attributes)
 962                (setq value (get-char-attribute char 'shuowen-radical)))
 963       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
 964                       value
 965                       (shuowen-radical value)
 966                       line-breaking))
 967       (setq attributes (delq 'shuowen-radical attributes))
 968       )
 969     (let (key)
 970       (dolist (domain
 971                (append
 972                 char-db-feature-domains
 973                 (let (dest domain)
 974                   (dolist (feature (char-attribute-list))
 975                     (setq feature (symbol-name feature))
 976                     (when (string-match
 977                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
 978                            feature)
 979                       (setq domain (intern (match-string 2 feature)))
 980                      (unless (memq domain dest)
 981                        (setq dest (cons domain dest)))))
 982                   (sort dest #'string<))))
 983         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
 984         (when (and (memq key attributes)
 985                    (setq value (get-char-attribute char key)))
 986           (setq radical value)
 987           (insert (format "(%s . %S)\t; %c%s"
 988                           key
 989                           radical
 990                           (ideographic-radical radical)
 991                           line-breaking))
 992           (setq attributes (delq key attributes))
 993           )
 994         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
 995         (when (and (memq key attributes)
 996                    (setq value (get-char-attribute char key)))
 997           (setq strokes value)
 998           (insert (format "(%s . %S)%s"
 999                           key
1000                           strokes
1001                           line-breaking))
1002           (setq attributes (delq key attributes))
1003           )
1004         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1005         (when (and (memq key attributes)
1006                    (setq value (get-char-attribute char key)))
1007           (insert (format "(%s       . %S)%s"
1008                           key
1009                           value
1010                           line-breaking))
1011           (setq attributes (delq key attributes))
1012           )
1013         (dolist (feature '(ideographic-radical
1014                            ideographic-strokes
1015                            total-strokes))
1016           (setq key (intern (format "%s@%s*sources" feature domain)))
1017           (when (and (memq key attributes)
1018                      (setq value (get-char-attribute char key)))
1019             (insert (format "(%s%s" key line-breaking))
1020             (dolist (cell value)
1021               (insert (format " %s" cell)))
1022             (insert ")")
1023             (insert line-breaking)
1024             (setq attributes (delq key attributes))
1025             ))
1026         ))
1027     (when (and (memq 'ideographic-strokes attributes)
1028                (setq value (get-char-attribute char 'ideographic-strokes)))
1029       (setq strokes value)
1030       (insert (format "(ideographic-strokes . %S)%s"
1031                       strokes
1032                       line-breaking))
1033       (setq attributes (delq 'ideographic-strokes attributes))
1034       )
1035     (when (and (memq 'kangxi-radical attributes)
1036                (setq value (get-char-attribute char 'kangxi-radical)))
1037       (unless (eq value radical)
1038         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1039                         value
1040                         (ideographic-radical value)
1041                         line-breaking))
1042         (or radical
1043             (setq radical value)))
1044       (setq attributes (delq 'kangxi-radical attributes))
1045       )
1046     (when (and (memq 'kangxi-strokes attributes)
1047                (setq value (get-char-attribute char 'kangxi-strokes)))
1048       (unless (eq value strokes)
1049         (insert (format "(kangxi-strokes\t . %S)%s"
1050                         value
1051                         line-breaking))
1052         (or strokes
1053             (setq strokes value)))
1054       (setq attributes (delq 'kangxi-strokes attributes))
1055       )
1056     (when (and (memq 'japanese-radical attributes)
1057                (setq value (get-char-attribute char 'japanese-radical)))
1058       (unless (eq value radical)
1059         (insert (format "(japanese-radical\t . %S)\t; %c%s"
1060                         value
1061                         (ideographic-radical value)
1062                         line-breaking))
1063         (or radical
1064             (setq radical value)))
1065       (setq attributes (delq 'japanese-radical attributes))
1066       )
1067     (when (and (memq 'japanese-strokes attributes)
1068                (setq value (get-char-attribute char 'japanese-strokes)))
1069       (unless (eq value strokes)
1070         (insert (format "(japanese-strokes\t . %S)%s"
1071                         value
1072                         line-breaking))
1073         (or strokes
1074             (setq strokes value)))
1075       (setq attributes (delq 'japanese-strokes attributes))
1076       )
1077     (when (and (memq 'cns-radical attributes)
1078                (setq value (get-char-attribute char 'cns-radical)))
1079       (insert (format "(cns-radical\t . %S)\t; %c%s"
1080                       value
1081                       (ideographic-radical value)
1082                       line-breaking))
1083       (setq attributes (delq 'cns-radical attributes))
1084       )
1085     (when (and (memq 'cns-strokes attributes)
1086                (setq value (get-char-attribute char 'cns-strokes)))
1087       (unless (eq value strokes)
1088         (insert (format "(cns-strokes\t . %S)%s"
1089                         value
1090                         line-breaking))
1091         (or strokes
1092             (setq strokes value)))
1093       (setq attributes (delq 'cns-strokes attributes))
1094       )
1095     ;; (when (and (memq 'shinjigen-1-radical attributes)
1096     ;;            (setq value (get-char-attribute char 'shinjigen-1-radical)))
1097     ;;   (unless (eq value radical)
1098     ;;     (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
1099     ;;                     value
1100     ;;                     (ideographic-radical value)
1101     ;;                     line-breaking))
1102     ;;     (or radical
1103     ;;         (setq radical value)))
1104     ;;   (setq attributes (delq 'shinjigen-1-radical attributes))
1105     ;;   )
1106     (when (and (memq 'ideographic- attributes)
1107                (setq value (get-char-attribute char 'ideographic-)))
1108       (insert "(ideographic-       ")
1109       (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1110             separator nil)
1111       (while (consp value)
1112         (setq cell (car value))
1113         (if (integerp cell)
1114             (setq cell (decode-char '=ucs cell)))
1115         (cond ((characterp cell)
1116                (if separator
1117                    (insert lbs))
1118                (if readable
1119                    (insert (format "%S" cell))
1120                  (char-db-insert-char-spec cell readable))
1121                (setq separator lbs))
1122               ((consp cell)
1123                (if separator
1124                    (insert lbs))
1125                (if (consp (car cell))
1126                    (char-db-insert-char-spec cell readable)
1127                  (char-db-insert-char-reference cell readable))
1128                (setq separator lbs))
1129               (t
1130                (if separator
1131                    (insert separator))
1132                (insert (prin1-to-string cell))
1133                (setq separator " ")))
1134         (setq value (cdr value)))
1135       (insert ")")
1136       (insert line-breaking)
1137       (setq attributes (delq 'ideographic- attributes)))
1138     (when (and (memq 'total-strokes attributes)
1139                (setq value (get-char-attribute char 'total-strokes)))
1140       (insert (format "(total-strokes       . %S)%s"
1141                       value
1142                       line-breaking))
1143       (setq attributes (delq 'total-strokes attributes))
1144       )
1145     (when (and (memq '->ideograph attributes)
1146                (setq value (get-char-attribute char '->ideograph)))
1147       (insert (format "(->ideograph\t%s)%s"
1148                       (mapconcat (lambda (code)
1149                                    (cond ((symbolp code)
1150                                           (symbol-name code))
1151                                          ((integerp code)
1152                                           (format "#x%04X" code))
1153                                          (t
1154                                           (format "%s %S"
1155                                                   line-breaking code))))
1156                                  value " ")
1157                       line-breaking))
1158       (setq attributes (delq '->ideograph attributes))
1159       )
1160     ;; (when (and (memq '->decomposition attributes)
1161     ;;            (setq value (get-char-attribute char '->decomposition)))
1162     ;;   (insert (format "(->decomposition\t%s)%s"
1163     ;;                   (mapconcat (lambda (code)
1164     ;;                                (cond ((symbolp code)
1165     ;;                                       (symbol-name code))
1166     ;;                                      ((characterp code)
1167     ;;                                       (if readable
1168     ;;                                           (format "%S" code)
1169     ;;                                         (format "#x%04X"
1170     ;;                                                 (char-int code))
1171     ;;                                         ))
1172     ;;                                      ((integerp code)
1173     ;;                                       (format "#x%04X" code))
1174     ;;                                      (t
1175     ;;                                       (format "%s%S" line-breaking code))))
1176     ;;                              value " ")
1177     ;;                   line-breaking))
1178     ;;   (setq attributes (delq '->decomposition attributes))
1179     ;;   )
1180     (if (equal (get-char-attribute char '->titlecase)
1181                (get-char-attribute char '->uppercase))
1182         (setq attributes (delq '->titlecase attributes)))
1183     (when (and (memq '->mojikyo attributes)
1184                (setq value (get-char-attribute char '->mojikyo)))
1185       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1186                       value (decode-char 'mojikyo value)
1187                       line-breaking))
1188       (setq attributes (delq '->mojikyo attributes))
1189       )
1190     (when (and (memq 'hanyu-dazidian-vol attributes)
1191                (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1192       (insert (format "(hanyu-dazidian-vol  . %d)%s"
1193                       value line-breaking))
1194       (setq attributes (delq 'hanyu-dazidian-vol attributes))
1195       )
1196     (when (and (memq 'hanyu-dazidian-page attributes)
1197                (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1198       (insert (format "(hanyu-dazidian-page . %d)%s"
1199                       value line-breaking))
1200       (setq attributes (delq 'hanyu-dazidian-page attributes))
1201       )
1202     (when (and (memq 'hanyu-dazidian-char attributes)
1203                (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1204       (insert (format "(hanyu-dazidian-char . %d)%s"
1205                       value line-breaking))
1206       (setq attributes (delq 'hanyu-dazidian-char attributes))
1207       )
1208     (unless readable
1209       (dolist (ignored '(composition
1210                          ->denotational <-subsumptive ->ucs-unified
1211                          ->ideographic-component-forms))
1212         (setq attributes (delq ignored attributes))))
1213     (while attributes
1214       (setq name (car attributes))
1215       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1216                   'value-is-empty)
1217         (cond ((setq ret (find-charset name))
1218                (setq name (charset-name ret))
1219                (when (not (memq name dest-ccss))
1220                  (setq dest-ccss (cons name dest-ccss))
1221                  (char-db-insert-ccs-feature name value line-breaking))
1222                )
1223               ((string-match "^=>ucs@" (symbol-name name))
1224                (insert (format "(%-18s . #x%04X)\t; %c%s"
1225                                name value (decode-char '=ucs value)
1226                                line-breaking))
1227                )
1228               ((eq name 'jisx0208-1978/4X)
1229                (insert (format "(%-18s . #x%04X)%s"
1230                                name value
1231                                line-breaking))
1232                )
1233               ((and
1234                 (not readable)
1235                 (not (eq name '->subsumptive))
1236                 (not (eq name '->uppercase))
1237                 (not (eq name '->lowercase))
1238                 (not (eq name '->titlecase))
1239                 (not (eq name '->canonical))
1240                 (not (eq name '->Bopomofo))
1241                 (not (eq name '->mistakable))
1242                 (not (eq name '->ideographic-variants))
1243                 (null (get-char-attribute
1244                        char (intern (format "%s*sources" name))))
1245                 (not (string-match "\\*sources$" (symbol-name name)))
1246                 (null (get-char-attribute
1247                        char (intern (format "%s*note" name))))
1248                 (not (string-match "\\*note$" (symbol-name name)))
1249                 (or (eq name '<-identical)
1250                     (eq name '<-uppercase)
1251                     (eq name '<-lowercase)
1252                     (eq name '<-titlecase)
1253                     (eq name '<-canonical)
1254                     (eq name '<-ideographic-variants)
1255                     ;; (eq name '<-synonyms)
1256                     (string-match "^<-synonyms" (symbol-name name))
1257                     (eq name '<-mistakable)
1258                     (when (string-match "^->" (symbol-name name))
1259                       (cond
1260                        ((string-match "^->fullwidth" (symbol-name name))
1261                         (not (and (consp value)
1262                                   (characterp (car value))
1263                                   (encode-char
1264                                    (car value) '=ucs 'defined-only)))
1265                         )
1266                        (t)))
1267                     ))
1268                )
1269               ((or (eq name 'ideographic-structure)
1270                    (eq name 'ideographic-combination)
1271                    (eq name 'ideographic-)
1272                    (eq name '=decomposition)
1273                    (char-feature-base-name= '=decomposition name)
1274                    (char-feature-base-name= '=>decomposition name)
1275                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1276                    ;;               (symbol-name name))
1277                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1278                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1279                                  (symbol-name name))
1280                    )
1281                (char-db-insert-relation-feature char name value
1282                                                 line-breaking
1283                                                 ccss readable))
1284               ((memq name '(ideograph=
1285                             original-ideograph-of
1286                             ancient-ideograph-of
1287                             vulgar-ideograph-of
1288                             wrong-ideograph-of
1289                             ;; simplified-ideograph-of
1290                             ideographic-variants
1291                             ;; ideographic-different-form-of
1292                             ))
1293                (insert (format "(%-18s%s " name line-breaking))
1294                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1295                      separator nil)
1296                (while (consp value)
1297                  (setq cell (car value))
1298                  (if (and (consp cell)
1299                           (consp (car cell)))
1300                      (progn
1301                        (if separator
1302                            (insert lbs))
1303                        (char-db-insert-alist cell readable)
1304                        (setq separator lbs))
1305                    (if separator
1306                        (insert separator))
1307                    (insert (prin1-to-string cell))
1308                    (setq separator " "))
1309                  (setq value (cdr value)))
1310                (insert ")")
1311                (insert line-breaking))
1312               ((consp value)
1313                (insert (format "(%-18s " name))
1314                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1315                      separator nil)
1316                (while (consp value)
1317                  (setq cell (car value))
1318                  (if (and (consp cell)
1319                           (consp (car cell))
1320                           (setq ret (condition-case nil
1321                                         (find-char cell)
1322                                       (error nil))))
1323                      (progn
1324                        (setq rest cell
1325                              al nil
1326                              cal nil)
1327                        (while rest
1328                          (setq key (car (car rest)))
1329                          (if (find-charset key)
1330                              (setq cal (cons key cal))
1331                            (setq al (cons key al)))
1332                          (setq rest (cdr rest)))
1333                        (if separator
1334                            (insert lbs))
1335                        (insert-char-attributes ret
1336                                                readable
1337                                                al ; cal
1338                                                nil 'for-sub-node)
1339                        (setq separator lbs))
1340                    (setq ret (prin1-to-string cell))
1341                    (if separator
1342                        (if (< (+ (current-column)
1343                                  (length ret)
1344                                  (length separator))
1345                               76)
1346                            (insert separator)
1347                          (insert lbs)))
1348                    (insert ret)
1349                    (setq separator " "))
1350                  (setq value (cdr value)))
1351                (insert ")")
1352                (insert line-breaking))
1353               (t
1354                (insert (format "(%-18s" name))
1355                (setq ret (prin1-to-string value))
1356                (unless (< (+ (current-column)
1357                              (length ret)
1358                              3)
1359                           76)
1360                  (insert line-breaking))
1361                (insert " . " ret ")" line-breaking)
1362                ;; (insert (format "(%-18s . %S)%s"
1363                ;;                 name value
1364                ;;                 line-breaking))
1365                )
1366               ))
1367       (setq attributes (cdr attributes)))
1368     (insert ")")))
1369
1370 (defun insert-char-data (char &optional readable
1371                               attributes)
1372   (save-restriction
1373     (narrow-to-region (point)(point))
1374     (insert "(define-char
1375   '")
1376     (insert-char-attributes char readable attributes)
1377     (insert ")\n")
1378     (goto-char (point-min))
1379     (while (re-search-forward "[ \t]+$" nil t)
1380       (replace-match ""))
1381     ;; from tabify.
1382     (goto-char (point-min))
1383     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1384       (let ((column (current-column))
1385             (indent-tabs-mode t))
1386         (delete-region (match-beginning 0) (point))
1387         (indent-to column)))
1388     (goto-char (point-max))
1389     ;; (tabify (point-min)(point-max))
1390     ))
1391
1392 (defun insert-char-data-with-variant (char &optional printable
1393                                            no-ucs-unified
1394                                            script excluded-script)
1395   (insert-char-data char printable)
1396   (let ((variants (char-variants char))
1397         rest
1398         variant vs ret)
1399     (setq variants (sort variants #'<))
1400     (setq rest variants)
1401     (setq variants (cons char variants))
1402     (while rest
1403       (setq variant (car rest))
1404       (unless (get-char-attribute variant '<-subsumptive)
1405         (if (and (or (null script)
1406                      (null (setq vs (get-char-attribute variant 'script)))
1407                      (memq script vs))
1408                  (or (null excluded-script)
1409                      (null (setq vs (get-char-attribute variant 'script)))
1410                      (not (memq excluded-script vs))))
1411             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1412               (insert-char-data variant printable)
1413               (if (setq ret (char-variants variant))
1414                   (while ret
1415                     (or (memq (car ret) variants)
1416                         ;; (get-char-attribute (car ret) '<-subsumptive)
1417                         (setq rest (nconc rest (list (car ret)))))
1418                     (setq ret (cdr ret)))))))
1419       (setq rest (cdr rest)))))
1420
1421 (defun insert-char-range-data (min max &optional script excluded-script)
1422   (let ((code min)
1423         char)
1424     (while (<= code max)
1425       (setq char (decode-char '=ucs code))
1426       (if (encode-char char '=ucs 'defined-only)
1427           (insert-char-data-with-variant char nil 'no-ucs-unified
1428                                          script excluded-script))
1429       (setq code (1+ code)))))
1430
1431 (defun write-char-range-data-to-file (min max file
1432                                           &optional script excluded-script)
1433   (let ((coding-system-for-write char-db-file-coding-system))
1434     (with-temp-buffer
1435       (insert (format ";; -*- coding: %s -*-\n"
1436                       char-db-file-coding-system))
1437       (insert-char-range-data min max script excluded-script)
1438       (write-region (point-min)(point-max) file))))
1439
1440 (defvar what-character-original-window-configuration)
1441
1442 ;;;###autoload
1443 (defun what-char-definition (char)
1444   (interactive (list (char-after)))
1445   (let ((buf (get-buffer-create "*Character Description*"))
1446         (the-buf (current-buffer))
1447         (win-conf (current-window-configuration)))
1448     (pop-to-buffer buf)
1449     (make-local-variable 'what-character-original-window-configuration)
1450     (setq what-character-original-window-configuration win-conf)
1451     (setq buffer-read-only nil)
1452     (erase-buffer)
1453     (condition-case err
1454         (progn
1455           (insert-char-data-with-variant char 'printable)
1456           (unless (char-attribute-alist char)
1457             (insert (format ";; = %c\n"
1458                             (let* ((rest (split-char char))
1459                                    (ccs (pop rest))
1460                                    (code (pop rest)))
1461                               (while rest
1462                                 (setq code (logior (lsh code 8)
1463                                                    (pop rest))))
1464                               (decode-char ccs code)))))
1465           ;; (char-db-update-comment)
1466           (set-buffer-modified-p nil)
1467           (view-mode the-buf (lambda (buf)
1468                                (set-window-configuration
1469                                 what-character-original-window-configuration)
1470                                ))
1471           (goto-char (point-min)))
1472       (error (progn
1473                (set-window-configuration
1474                 what-character-original-window-configuration)
1475                (signal (car err) (cdr err)))))))
1476
1477
1478 ;;; @ end
1479 ;;;
1480
1481 (provide 'char-db-util)
1482
1483 ;;; char-db-util.el ends here