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) 1998,1999,2000,2001,2002,2003,2004,2005,2006,2007,
   4 ;;   2008,2009,2010,2011,2012,2013,2014,2015 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    ?裘 ?老 ?毛 ?毳 ?尸 ?尺 ?尾 ?履 ?舟 ?方 ; 310
  99    ?儿 ?兄 ?兂 ?皃 ?𠑹 ?先 ?秃 ?見 ?覞 ?欠 ; 320
 100    ?㱃 ?㳄 ?旡 ?頁 ?𦣻 ?面 ?丏 ?首 ?𥄉 ?須 ; 330
 101    ?彡 ?彣 ?文 ?髟 ?后 ?司 ?卮 ?卩 ?印 ?色 ; 340
 102    ?𠨍 ?辟 ?勹 ?包 ?茍 ?鬼 ?甶 ?厶 ?嵬 ?山 ; 350
 103    ?屾 ?屵 ?广 ?厂 ?丸 ?危 ?石 ?長 ?勿 ?冄 ; 360
 104    ?而 ?豕 ?㣇 ?彑 ?豚 ?豸 ?𤉡 ?易 ?象 ?馬 ; 370
 105    ?𢊁 ?鹿 ?麤 ?㲋 ?兔 ?萈 ?犬 ?㹜 ?鼠 ?能 ; 380
 106    ?熊 ?火 ?炎 ?黑 ?囪 ?焱 ?炙 ?赤 ?大 ?亦 ; 390
 107    ])
 108
 109 (defun shuowen-radical (number)
 110   (aref shuowen-radicals (1- number)))
 111
 112 (defvar char-db-file-coding-system 'utf-8-mcs-er)
 113
 114 (defvar char-db-ignored-attributes '(ideographic-products))
 115
 116 (defvar char-db-coded-charset-priority-list
 117   '(ascii
 118     control-1
 119     latin-iso8859-1
 120     latin-iso8859-2
 121     latin-iso8859-3
 122     latin-iso8859-4
 123     latin-iso8859-9
 124     latin-jisx0201
 125     cyrillic-iso8859-5
 126     greek-iso8859-7
 127     thai-tis620
 128     ;; =mj
 129     ;; =adobe-japan1-0
 130     ;; =adobe-japan1-1
 131     ;; =adobe-japan1-2
 132     ;; =adobe-japan1-3
 133     ;; =adobe-japan1-4
 134     ;; =adobe-japan1-5
 135     ;; =adobe-japan1-6
 136     =jis-x0208
 137     =jis-x0208@1978
 138     =jis-x0208@1983
 139     japanese-jisx0212
 140     chinese-gb2312
 141     =jis-x0208@1990
 142     chinese-cns11643-1
 143     chinese-cns11643-2
 144     chinese-cns11643-3
 145     chinese-cns11643-4
 146     chinese-cns11643-5
 147     chinese-cns11643-6
 148     chinese-cns11643-7
 149     =jis-x0213-1
 150     =jis-x0213-1@2000
 151     =jis-x0213-1@2004
 152     =jis-x0213-2
 153     korean-ksc5601
 154     chinese-isoir165
 155     katakana-jisx0201
 156     hebrew-iso8859-8
 157     chinese-gb12345
 158     latin-viscii
 159     ethiopic-ucs
 160     =big5-cdp
 161     =hanyo-denshi/ja
 162     =hanyo-denshi/jb
 163     =hanyo-denshi/jc
 164     =hanyo-denshi/jd
 165     =hanyo-denshi/ft
 166     =hanyo-denshi/ia
 167     =hanyo-denshi/ib
 168     =hanyo-denshi/hg
 169     =hanyo-denshi/jt
 170     =hanyo-denshi/ks
 171     ;; ==mj
 172     ;; ==adobe-japan1-0
 173     ;; ==adobe-japan1-1
 174     ;; ==adobe-japan1-2
 175     ;; ==adobe-japan1-3
 176     ;; ==adobe-japan1-4
 177     ;; ==adobe-japan1-5
 178     ;; ==adobe-japan1-6
 179     ==jis-x0208
 180     ==jis-x0213-1
 181     ==jis-x0213-2
 182     ==hanyo-denshi/ja
 183     ==hanyo-denshi/jb
 184     ==hanyo-denshi/jc
 185     ==hanyo-denshi/ft
 186     ==hanyo-denshi/ib
 187     ==hanyo-denshi/hg
 188     ==hanyo-denshi/jt
 189     ==hanyo-denshi/ks
 190     =gt
 191     =gt-k
 192     =daikanwa
 193     =daikanwa@rev2
 194     =daikanwa@rev1
 195     =daikanwa/+p
 196     ==daikanwa
 197     =cbeta
 198     ideograph-hanziku-1
 199     ideograph-hanziku-2
 200     ideograph-hanziku-3
 201     ideograph-hanziku-4
 202     ideograph-hanziku-5
 203     ideograph-hanziku-6
 204     ideograph-hanziku-7
 205     ideograph-hanziku-8
 206     ideograph-hanziku-9
 207     ideograph-hanziku-10
 208     ideograph-hanziku-11
 209     ideograph-hanziku-12
 210     ;; =>>>adobe-japan1-0
 211     ;; =>>>adobe-japan1-1
 212     ;; =>>>adobe-japan1-2
 213     ;; =>>>adobe-japan1-3
 214     ;; =>>>adobe-japan1-4
 215     ;; =>>>adobe-japan1-5
 216     ;; =>>>adobe-japan1-6
 217     ;; =>>>jis-x0208
 218     ;; =>>>jis-x0213-1
 219     ;; =>>>jis-x0213-2
 220     ;; =>>>hanyo-denshi/ja
 221     ;; =>>>hanyo-denshi/jb
 222     ;; =>>>hanyo-denshi/jc
 223     ;; =>>>hanyo-denshi/ft
 224     ;; =>>>hanyo-denshi/ib
 225     ;; =>>>hanyo-denshi/hg
 226     ;; =>>>hanyo-denshi/jt
 227     ;; =>>>hanyo-denshi/ks
 228     ;; =>>>gt
 229     =>>adobe-japan1-0
 230     =>>adobe-japan1-1
 231     =>>adobe-japan1-2
 232     =>>adobe-japan1-3
 233     =>>adobe-japan1-4
 234     =>>adobe-japan1-5
 235     =>>adobe-japan1-6
 236     =>>jis-x0208
 237     =>>jis-x0213-1
 238     =>>jis-x0213-1@2000
 239     =>>jis-x0213-1@2004
 240     =>>jis-x0213-2
 241     =>>jis-x0208@1978
 242     =>>hanyo-denshi/ft
 243     =>>hanyo-denshi/jt
 244     =>>hanyo-denshi/ks
 245     =>>gt
 246     =>>daikanwa
 247     =>>cbeta
 248     =+>jis-x0208
 249     =+>jis-x0213-1
 250     =+>jis-x0213-2
 251     =+>adobe-japan1-0
 252     =+>adobe-japan1-1
 253     =+>adobe-japan1-2
 254     =+>adobe-japan1-3
 255     =+>adobe-japan1-4
 256     =+>adobe-japan1-5
 257     =+>adobe-japan1-6
 258     =+>jis-x0208@1978
 259     =>jis-x0208
 260     =>jis-x0208@1997
 261     =>jis-x0213-1
 262     =>jis-x0213-1@2000
 263     =>jis-x0213-1@2004
 264     =>jis-x0213-2
 265     ==>ucs@bucs
 266     =>iwds-1
 267     =>ucs@hanyo-denshi
 268     =>ucs@iso
 269     =>ucs@unicode
 270     =>ucs@jis
 271     =>ucs@cns
 272     =>ucs@ks
 273     =+>ucs@iso
 274     =+>ucs@unicode
 275     =+>ucs@jis
 276     =+>ucs@jis/1990
 277     =+>ucs@cns
 278     =+>ucs@ks
 279     =>>ucs@iso
 280     =>>ucs@unicode
 281     =>>ucs@jis
 282     =>>ucs@cns
 283     =>>>ucs@iso
 284     =>>>ucs@unicode
 285     ==ucs@iso
 286     ==ucs@unicode
 287     ;; ==ucs@cns
 288     ==gb2312
 289     ==ks-x1001
 290     ==cns11643-1
 291     ==cns11643-2
 292     ==cns11643-3
 293     ==cns11643-4
 294     ==cns11643-5
 295     ==cns11643-6
 296     ==cns11643-7
 297     ==gt
 298     ==jis-x0208@1990
 299     ;; ==jis-x0208@1983
 300     ==jis-x0208@1978
 301     ==gt-k
 302     =ucs@iso
 303     =ucs@unicode
 304     =ucs@cns
 305     ==big5-cdp
 306     ==cbeta
 307     =>>big5-cdp
 308     =>>gt-k
 309     =+>gt
 310     =>gt
 311     =>big5-cdp
 312     =>daikanwa
 313     =>daikanwa/ho
 314     =>cns11643-5
 315     =>cns11643-7
 316     =big5
 317     =big5-eten
 318     =>gt-k
 319     =zinbun-oracle
 320     =>zinbun-oracle
 321     =ruimoku-v6
 322     =>>ruimoku-v6
 323     ==ruimoku-v6
 324     =jef-china3
 325     =>cbeta
 326     =shinjigen
 327     =ucs-var-001
 328     =ucs-var-002
 329     =ucs-itaiji-001
 330     =ucs-itaiji-002
 331     =ucs-itaiji-003
 332     =ucs-itaiji-008
 333     =big5-cdp-var-3
 334     =big5-cdp-var-5
 335     =>ucs@iwds-1))
 336
 337
 338 ;;; @ char-db formatters
 339 ;;;
 340
 341 (defun char-db-make-char-spec (char)
 342   (let (ret char-spec)
 343     (cond ((characterp char)
 344            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 345                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 346                   (setq char-spec (list (cons '=ucs ret)))
 347                   (cond ((setq ret (get-char-attribute char 'name))
 348                          (setq char-spec (cons (cons 'name ret) char-spec))
 349                          )
 350                         ((setq ret (get-char-attribute char 'name*))
 351                          (setq char-spec (cons (cons 'name* ret) char-spec))
 352                          ))
 353                   )
 354                  ((encode-char char '=mj 'defined-only)
 355                   (setq char-spec nil)
 356                   (dolist (ccs (charset-list))
 357                     (if (and (or (eq ccs '=mj)
 358                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 359                                  (string-match "=ucs@" (symbol-name ccs))
 360                                  )
 361                              (setq ccs (charset-name ccs))
 362                              (null (assq ccs char-spec))
 363                              (setq ret (encode-char char ccs 'defined-only)))
 364                         (setq char-spec (cons (cons ccs ret) char-spec))))
 365                   )
 366                  ((encode-char char '==mj 'defined-only)
 367                   (setq char-spec nil)
 368                   (dolist (ccs (charset-list))
 369                     (if (and (or (eq ccs '==mj)
 370                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 371                                  (string-match "=ucs@" (symbol-name ccs))
 372                                  )
 373                              (setq ccs (charset-name ccs))
 374                              (null (assq ccs char-spec))
 375                              (setq ret (encode-char char ccs 'defined-only)))
 376                         (setq char-spec (cons (cons ccs ret) char-spec))))
 377                   )
 378                  ((encode-char char '=adobe-japan1 'defined-only)
 379                   (setq char-spec nil)
 380                   (dolist (ccs (charset-list))
 381                     (if (and (or (memq ccs
 382                                        '(=adobe-japan1-0
 383                                          =adobe-japan1-1
 384                                          =adobe-japan1-2
 385                                          =adobe-japan1-3
 386                                          =adobe-japan1-4
 387                                          =adobe-japan1-5
 388                                          =adobe-japan1-6
 389                                          ))
 390                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 391                                  (string-match "=ucs@" (symbol-name ccs))
 392                                  )
 393                              (setq ccs (charset-name ccs))
 394                              (null (assq ccs char-spec))
 395                              (setq ret (encode-char char ccs 'defined-only)))
 396                         (setq char-spec (cons (cons ccs ret) char-spec))))
 397                   )
 398                  ((encode-char char '==adobe-japan1 'defined-only)
 399                   (setq char-spec nil)
 400                   (dolist (ccs (charset-list))
 401                     (if (and (or (memq ccs
 402                                        '(==adobe-japan1-0
 403                                          ==adobe-japan1-1
 404                                          ==adobe-japan1-2
 405                                          ==adobe-japan1-3
 406                                          ==adobe-japan1-4
 407                                          ==adobe-japan1-5
 408                                          ==adobe-japan1-6
 409                                          ))
 410                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 411                                  (string-match "=ucs@" (symbol-name ccs))
 412                                  )
 413                              (setq ccs (charset-name ccs))
 414                              (null (assq ccs char-spec))
 415                              (setq ret (encode-char char ccs 'defined-only)))
 416                         (setq char-spec (cons (cons ccs ret) char-spec))))
 417                   )
 418                  ((setq ret
 419                         (catch 'tag
 420                           (let ((rest char-db-coded-charset-priority-list)
 421                                 ccs)
 422                             (while rest
 423                               (setq ccs (charset-name
 424                                          (find-charset (car rest))))
 425                               (if (setq ret
 426                                         (encode-char char ccs
 427                                                      'defined-only))
 428                                   (throw 'tag (cons ccs ret)))
 429                               (setq rest (cdr rest))))))
 430                   (setq char-spec (list ret))
 431                   (dolist (ccs (delq (car ret) (charset-list)))
 432                     (if (and (or (charset-iso-final-char ccs)
 433                                  (memq ccs
 434                                        '(=daikanwa
 435                                          =daikanwa@rev2
 436                                          ;; =gt-k
 437                                          =jis-x0208@1997
 438                                          ))
 439                                  (eq (charset-property ccs 'iso-ir) 177)
 440                                  ;; (string-match "=ucs@" (symbol-name ccs))
 441                                  )
 442                              (setq ccs (charset-name ccs))
 443                              (null (assq ccs char-spec))
 444                              (setq ret (encode-char char ccs 'defined-only)))
 445                         (setq char-spec (cons (cons ccs ret) char-spec))))
 446                   (if (null char-spec)
 447                       (setq char-spec (split-char char)))
 448                   (cond ((setq ret (get-char-attribute char 'name))
 449                          (setq char-spec (cons (cons 'name ret) char-spec))
 450                          )
 451                         ((setq ret (get-char-attribute char 'name*))
 452                          (setq char-spec (cons (cons 'name* ret) char-spec))
 453                          ))
 454                   )
 455                  ((setq ret (get-char-attribute
 456                              char 'ideographic-combination))
 457                   (setq char-spec
 458                         (cons (cons 'ideographic-combination ret)
 459                               char-spec))
 460                   ))
 461            char-spec)
 462           ((consp char)
 463            char))))
 464
 465 (defun char-db-insert-char-spec (char &optional readable column
 466                                       required-features)
 467   (unless column
 468     (setq column (current-column)))
 469   (let (char-spec temp-char)
 470     (setq char-spec (char-db-make-char-spec char))
 471     (unless (or (characterp char) ; char
 472                 (condition-case nil
 473                     (setq char (find-char char-spec))
 474                   (error nil)))
 475       ;; define temporary character
 476       ;;   Current implementation is dirty.
 477       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 478                                          char-spec)))
 479       (remove-char-attribute temp-char 'ideograph-daikanwa)
 480       (setq char temp-char))
 481     (insert-char-attributes char
 482                             readable
 483                             (union (mapcar #'car char-spec)
 484                                    required-features)
 485                             nil 'for-sub-node)
 486     (when temp-char
 487       ;; undefine temporary character
 488       ;;   Current implementation is dirty.
 489       (setq char-spec (char-attribute-alist temp-char))
 490       (while char-spec
 491         (remove-char-attribute temp-char (car (car char-spec)))
 492         (setq char-spec (cdr char-spec))))))
 493
 494 (defun char-db-insert-alist (alist &optional readable column)
 495   (unless column
 496     (setq column (current-column)))
 497   (let ((line-breaking
 498          (concat "\n" (make-string (1+ column) ?\ )))
 499         name value
 500         ret al ; cal
 501         key
 502         lbs cell rest separator)
 503     (insert "(")
 504     (while alist
 505       (setq name (car (car alist))
 506             value (cdr (car alist)))
 507       (cond ((eq name 'char)
 508              (insert "(char . ")
 509              (if (setq ret (condition-case nil
 510                                (find-char value)
 511                              (error nil)))
 512                  (progn
 513                    (setq al nil
 514                          ;; cal nil
 515                          )
 516                    (while value
 517                      (setq key (car (car value)))
 518                      ;; (if (find-charset key)
 519                      ;;     (setq cal (cons key cal))
 520                      (setq al (cons key al))
 521                      ;; )
 522                      (setq value (cdr value)))
 523                    (insert-char-attributes ret
 524                                            readable
 525                                            (or al 'none) ; cal
 526                                            nil 'for-sub-node))
 527                (insert (prin1-to-string value)))
 528              (insert ")")
 529              (insert line-breaking))
 530             ((consp value)
 531              (insert (format "(%-18s " name))
 532              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 533              (while (consp value)
 534                (setq cell (car value))
 535                (if (and (consp cell)
 536                         (consp (car cell))
 537                         (setq ret (condition-case nil
 538                                       (find-char cell)
 539                                     (error nil)))
 540                         )
 541                    (progn
 542                      (setq rest cell
 543                            al nil
 544                            ;; cal nil
 545                            )
 546                      (while rest
 547                        (setq key (car (car rest)))
 548                        ;; (if (find-charset key)
 549                        ;;     (setq cal (cons key cal))
 550                        (setq al (cons key al))
 551                        ;; )
 552                        (setq rest (cdr rest)))
 553                      (if separator
 554                          (insert lbs))
 555                      (insert-char-attributes ret
 556                                              readable
 557                                              al ; cal
 558                                              nil 'for-sub-node)
 559                      (setq separator lbs))
 560                  (if separator
 561                      (insert separator))
 562                  (insert (prin1-to-string cell))
 563                  (setq separator " "))
 564                (setq value (cdr value)))
 565              (insert ")")
 566              (insert line-breaking))
 567             (t
 568              (insert (format "(%-18s . %S)%s"
 569                              name value
 570                              line-breaking))))
 571       (setq alist (cdr alist))))
 572   (insert ")"))
 573
 574 (defun char-db-insert-char-reference (plist &optional readable column)
 575   (unless column
 576     (setq column (current-column)))
 577   (let ((line-breaking
 578          (concat "\n" (make-string (1+ column) ?\ )))
 579         (separator "")
 580         name value)
 581     (insert "(")
 582     (while plist
 583       (setq name (pop plist))
 584       (setq value (pop plist))
 585       (cond ((eq name :char)
 586              (insert separator)
 587              (insert ":char\t")
 588              (cond ((numberp value)
 589                     (setq value (decode-char '=ucs value)))
 590                    ;; ((consp value)
 591                    ;;  (setq value (or (find-char value)
 592                    ;;                  value)))
 593                    )
 594              (char-db-insert-char-spec value readable)
 595              (insert line-breaking)
 596              (setq separator ""))
 597             ((eq name :radical)
 598              (insert (format "%s%s\t%d ; %c%s"
 599                              separator
 600                              name value
 601                              (ideographic-radical value)
 602                              line-breaking))
 603              (setq separator ""))
 604             (t
 605              (insert (format "%s%s\t%S" separator name value))
 606              (setq separator line-breaking)))
 607       ))
 608   (insert ")"))
 609
 610 (defun char-db-decode-isolated-char (ccs code-point)
 611   (let (ret)
 612     (setq ret
 613           (cond ((eq ccs 'arabic-iso8859-6)
 614                  (decode-char ccs code-point))
 615                 ;; ((eq ccs '=gt)
 616                 ;;  (decode-builtin-char '==gt code-point))
 617                 ((and (memq ccs '(=gt-pj-1
 618                                   =gt-pj-2
 619                                   =gt-pj-3
 620                                   =gt-pj-4
 621                                   =gt-pj-5
 622                                   =gt-pj-6
 623                                   =gt-pj-7
 624                                   =gt-pj-8
 625                                   =gt-pj-9
 626                                   =gt-pj-10
 627                                   =gt-pj-11))
 628                       (setq ret (decode-char ccs code-point))
 629                       (setq ret (encode-char ret '=gt 'defined-only)))
 630                  (decode-builtin-char '=gt ret))
 631                 (t
 632                  (decode-builtin-char ccs code-point))))
 633     (cond ((null ret)
 634            (or (decode-char ccs code-point)
 635                (define-char (list (cons ccs code-point)))))
 636           ((and (<= 0 (char-int ret))
 637                 (<= (char-int ret) #x1F))
 638            (decode-char '=ucs (+ #x2400 (char-int ret))))
 639           ((= (char-int ret) #x7F)
 640            ?\u2421)
 641           (t ret))))
 642
 643 (defvar char-db-convert-obsolete-format t)
 644
 645 (defun char-db-insert-ccs-feature (name value line-breaking)
 646   (cond
 647    ((integerp value)
 648     (insert
 649      (format
 650       (cond
 651        ((memq name '(=>iwds-1
 652                      ==shinjigen
 653                      =shinjigen
 654                      =shinjigen@1ed ==shinjigen@1ed
 655                      =shinjigen@rev ==shinjigen@rev
 656                      =shinjigen/+p@rev ==shinjigen/+p@rev
 657                      ===daikanwa/ho ==daikanwa/ho
 658                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 659         "(%-18s .  %04d)\t; %c")
 660        ((eq name '=shinjigen@1ed/24pr)
 661         "(%-18s . %04d)\t; %c")
 662        ((or
 663          (memq name
 664                '(===daikanwa
 665                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 666                  =daikanwa@rev1 =daikanwa@rev2
 667                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 668                  =>>daikanwa/+p
 669                  =daikanwa/+2p =>>daikanwa/+2p
 670                  =gt ==gt ===gt
 671                  =>>gt =+>gt =>gt
 672                  =gt-k ==gt-k ===gt-k
 673                  =>>gt-k =>gt-k
 674                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 675                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 676                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 677                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 678                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 679                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 680                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 681                  =>>adobe-japan1-0 =+>adobe-japan1-0
 682                  =>>adobe-japan1-1 =+>adobe-japan1-1
 683                  =>>adobe-japan1-2 =+>adobe-japan1-2
 684                  =>>adobe-japan1-3 =+>adobe-japan1-3
 685                  =>>adobe-japan1-4 =+>adobe-japan1-4
 686                  =>>adobe-japan1-5 =+>adobe-japan1-5
 687                  =>>adobe-japan1-6 =+>adobe-japan1-6
 688                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 689                  =zinbun-oracle =>zinbun-oracle
 690                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 691                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 692                  ===hng-khi ===hng-khm ===hng-hok ===hng-kyd ===hng-sok
 693                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 694                  ===hng-sys ===hng-tsu ===hng-tzj
 695                  ===hng-hos ===hng-nak ===hng-jhk
 696                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 697                  ===hng-nkm ===hng-k24 ===hng-nkk
 698                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 699                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 700                  ===hng-okd ===hng-wad ===hng-kmi
 701                  ===hng-zkd ===hng-doh ===hng-jyu
 702                  ===hng-tzs ===hng-kss ===hng-kyo
 703                  ===hng-smk))
 704          ;; (string-match "^=adobe-" (symbol-name name))
 705          )
 706         "(%-18s . %05d)\t; %c")
 707        ((memq name '(=hanyo-denshi/ks
 708                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 709                      =>>hanyo-denshi/ks
 710                      =koseki ==koseki
 711                      =mj ==mj ===mj =>>mj
 712                      =zihai mojikyo))
 713         "(%-18s . %06d)\t; %c")
 714        ((>= (charset-dimension name) 2)
 715         "(%-18s . #x%04X)\t; %c")
 716        (t
 717         "(%-18s . #x%02X)\t; %c"))
 718       name
 719       (if (= (charset-iso-graphic-plane name) 1)
 720           (logior value
 721                   (cond ((= (charset-dimension name) 1)
 722                          #x80)
 723                         ((= (charset-dimension name) 2)
 724                          #x8080)
 725                         ((= (charset-dimension name) 3)
 726                          #x808080)
 727                         (t 0)))
 728         value)
 729       (char-db-decode-isolated-char name value)))
 730     (if (and (= (charset-chars name) 94)
 731              (= (charset-dimension name) 2))
 732         (insert (format " [%02d-%02d]"
 733                         (- (lsh value -8) 32)
 734                         (- (logand value 255) 32))))
 735     )
 736    (t
 737     (insert (format "(%-18s . %s)" name value))
 738     ))
 739   (insert line-breaking))
 740
 741 (defun char-db-insert-relation-feature (char name value line-breaking
 742                                              ccss readable)
 743   (insert (format "(%-18s%s " name line-breaking))
 744   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 745         separator cell sources required-features
 746         ret)
 747     (while (consp value)
 748       (setq cell (car value))
 749       (if (integerp cell)
 750           (setq cell (decode-char '=ucs cell)))
 751       (cond
 752        ((eq name '->subsumptive)
 753         (when (or (not (some (lambda (atr)
 754                                (get-char-attribute cell atr))
 755                              char-db-ignored-attributes))
 756                   (some (lambda (ccs)
 757                           (encode-char cell ccs 'defined-only))
 758                         ccss))
 759           (if separator
 760               (insert lbs))
 761           (let ((char-db-ignored-attributes
 762                  (cons '<-subsumptive
 763                        char-db-ignored-attributes)))
 764             (insert-char-attributes cell readable nil nil 'for-sub-node))
 765           (setq separator lbs))
 766         )
 767        ((characterp cell)
 768         (setq sources
 769               (get-char-attribute
 770                char (intern (format "%s*sources" name))))
 771         (setq required-features nil)
 772         (dolist (source sources)
 773           (cond
 774            ((memq source '(JP
 775                            JP/Jouyou
 776                            shinjigen shinjigen@1ed shinjigen@rev))
 777             (setq required-features
 778                   (union required-features
 779                          '(=jis-x0208
 780                            =jis-x0208@1990
 781                            =jis-x0213-1@2000
 782                            =jis-x0213-1@2004
 783                            =jis-x0213-2
 784                            =jis-x0212
 785                            =jis-x0208@1983
 786                            =jis-x0208@1978
 787                            =shinjigen
 788                            =shinjigen@1ed
 789                            =shinjigen@rev
 790                            =shinjigen/+p@rev))))
 791            ((eq source 'CN)
 792             (setq required-features
 793                   (union required-features
 794                          '(=gb2312
 795                            =gb12345
 796                            =iso-ir165)))))
 797           (cond
 798            ((find-charset (setq ret (intern (format "=%s" source))))
 799             (setq required-features
 800                   (cons ret required-features)))
 801            (t (setq required-features
 802                     (cons source required-features)))))
 803         (cond ((string-match "@JP" (symbol-name name))
 804                (setq required-features
 805                      (union required-features
 806                             '(=jis-x0208
 807                               =jis-x0208@1990
 808                               =jis-x0213-1-2000
 809                               =jis-x0213-2-2000
 810                               =jis-x0212
 811                               =jis-x0208@1983
 812                               =jis-x0208@1978))))
 813               ((string-match "@CN" (symbol-name name))
 814                (setq required-features
 815                      (union required-features
 816                             '(=gb2312
 817                               =gb12345
 818                               =iso-ir165)))))
 819         (if separator
 820             (insert lbs))
 821         (if readable
 822             (insert (format "%S" cell))
 823           (char-db-insert-char-spec cell readable
 824                                     nil
 825                                     required-features))
 826         (setq separator lbs))
 827        ((consp cell)
 828         (if separator
 829             (insert lbs))
 830         (if (consp (car cell))
 831             (char-db-insert-char-spec cell readable)
 832           (char-db-insert-char-reference cell readable))
 833         (setq separator lbs))
 834        (t
 835         (if separator
 836             (insert separator))
 837         (insert (prin1-to-string cell))
 838         (setq separator " ")))
 839       (setq value (cdr value)))
 840     (insert ")")
 841     (insert line-breaking)))
 842
 843 (defun insert-char-attributes (char &optional readable attributes column
 844                                     for-sub-node)
 845   (unless column
 846     (setq column (current-column)))
 847   (let (name value ; has-long-ccs-name
 848         rest
 849         radical strokes
 850         (line-breaking
 851          (concat "\n" (make-string (1+ column) ?\ )))
 852         lbs cell separator ret
 853         key al cal
 854         dest-ccss ; sources required-features
 855         ccss)
 856     (let (atr-d)
 857       (setq attributes
 858             (sort (if attributes
 859                       (if (consp attributes)
 860                           (progn
 861                             (dolist (name attributes)
 862                               (unless (memq name char-db-ignored-attributes)
 863                                 (if (find-charset name)
 864                                     (push name ccss))
 865                                 (push name atr-d)))
 866                             atr-d))
 867                     (dolist (name (char-attribute-list))
 868                       (unless (memq name char-db-ignored-attributes)
 869                         (if (find-charset name)
 870                             (push name ccss))
 871                         (push name atr-d)))
 872                     atr-d)
 873                   #'char-attribute-name<)))
 874     (insert "(")
 875     (when (memq '<-subsumptive attributes)
 876       (when (or readable (not for-sub-node))
 877         (when (setq value (get-char-attribute char '<-subsumptive))
 878           (char-db-insert-relation-feature char '<-subsumptive value
 879                                            line-breaking
 880                                            ccss readable)))
 881       (setq attributes (delq '<-subsumptive attributes)))
 882     (when (and (memq '<-denotational attributes)
 883                (setq value (get-char-attribute char '<-denotational)))
 884       (char-db-insert-relation-feature char '<-denotational value
 885                                        line-breaking
 886                                        ccss readable)
 887       (setq attributes (delq '<-denotational attributes)))
 888     (when (and (memq '<-denotational@component attributes)
 889                (setq value (get-char-attribute char '<-denotational@component)))
 890       (char-db-insert-relation-feature char '<-denotational@component value
 891                                        line-breaking
 892                                        ccss readable)
 893       (setq attributes (delq '<-denotational@component attributes)))
 894     (when (and (memq 'name attributes)
 895                (setq value (get-char-attribute char 'name)))
 896       (insert (format
 897                (if (> (+ (current-column) (length value)) 48)
 898                    "(name . %S)%s"
 899                  "(name               . %S)%s")
 900                value line-breaking))
 901       (setq attributes (delq 'name attributes))
 902       )
 903     (when (and (memq 'name* attributes)
 904                (setq value (get-char-attribute char 'name*)))
 905       (insert (format
 906                (if (> (+ (current-column) (length value)) 48)
 907                    "(name* . %S)%s"
 908                  "(name*              . %S)%s")
 909                value line-breaking))
 910       (setq attributes (delq 'name* attributes))
 911       )
 912     (when (and (memq 'script attributes)
 913                (setq value (get-char-attribute char 'script)))
 914       (insert (format "(script\t\t%s)%s"
 915                       (mapconcat (function prin1-to-string)
 916                                  value " ")
 917                       line-breaking))
 918       (setq attributes (delq 'script attributes))
 919       )
 920     (dolist (name '(=>ucs =>ucs*))
 921       (when (and (memq name attributes)
 922                  (setq value (get-char-attribute char name)))
 923         (insert (format "(%-18s . #x%04X)\t; %c%s"
 924                         name value (decode-char '=ucs value)
 925                         line-breaking))
 926         (setq attributes (delq name attributes))))
 927     (dolist (name '(=>ucs@gb =>ucs@big5))
 928       (when (and (memq name attributes)
 929                  (setq value (get-char-attribute char name)))
 930         (insert (format "(%-18s . #x%04X)\t; %c%s"
 931                         name value
 932                         (decode-char (intern
 933                                       (concat "="
 934                                               (substring
 935                                                (symbol-name name) 2)))
 936                                      value)
 937                         line-breaking))
 938         (setq attributes (delq name attributes))
 939         ))
 940     ;; (dolist (name '(=>daikanwa))
 941     ;;   (when (and (memq name attributes)
 942     ;;              (setq value (get-char-attribute char name)))
 943     ;;     (insert
 944     ;;      (if (integerp value)
 945     ;;          (format "(%-18s . %05d)\t; %c%s"
 946     ;;                  name value (decode-char '=daikanwa value)
 947     ;;                  line-breaking)
 948     ;;        (format "(%-18s %s)\t; %c%s"
 949     ;;                name
 950     ;;                (mapconcat (function prin1-to-string)
 951     ;;                           value " ")
 952     ;;                (char-representative-of-daikanwa char)
 953     ;;                line-breaking)))
 954     ;;     (setq attributes (delq name attributes))))
 955     (when (and (memq 'general-category attributes)
 956                (setq value (get-char-attribute char 'general-category)))
 957       (insert (format
 958                "(general-category\t%s) ; %s%s"
 959                (mapconcat (lambda (cell)
 960                             (format "%S" cell))
 961                           value " ")
 962                (cond ((rassoc value unidata-normative-category-alist)
 963                       "Normative Category")
 964                      ((rassoc value unidata-informative-category-alist)
 965                       "Informative Category")
 966                      (t
 967                       "Unknown Category"))
 968                line-breaking))
 969       (setq attributes (delq 'general-category attributes))
 970       )
 971     (when (and (memq 'bidi-category attributes)
 972                (setq value (get-char-attribute char 'bidi-category)))
 973       (insert (format "(bidi-category\t. %S)%s"
 974                       value
 975                       line-breaking))
 976       (setq attributes (delq 'bidi-category attributes))
 977       )
 978     (unless (or (not (memq 'mirrored attributes))
 979                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
 980                     'empty))
 981       (insert (format "(mirrored\t\t. %S)%s"
 982                       value
 983                       line-breaking))
 984       (setq attributes (delq 'mirrored attributes))
 985       )
 986     (cond
 987      ((and (memq 'decimal-digit-value attributes)
 988            (setq value (get-char-attribute char 'decimal-digit-value)))
 989       (insert (format "(decimal-digit-value . %S)%s"
 990                       value
 991                       line-breaking))
 992       (setq attributes (delq 'decimal-digit-value attributes))
 993       (when (and (memq 'digit-value attributes)
 994                  (setq value (get-char-attribute char 'digit-value)))
 995         (insert (format "(digit-value\t . %S)%s"
 996                         value
 997                         line-breaking))
 998         (setq attributes (delq 'digit-value attributes))
 999         )
1000       (when (and (memq 'numeric-value attributes)
1001                  (setq value (get-char-attribute char 'numeric-value)))
1002         (insert (format "(numeric-value\t . %S)%s"
1003                         value
1004                         line-breaking))
1005         (setq attributes (delq 'numeric-value attributes))
1006         )
1007       )
1008      (t
1009       (when (and (memq 'digit-value attributes)
1010                  (setq value (get-char-attribute char 'digit-value)))
1011         (insert (format "(digit-value\t. %S)%s"
1012                         value
1013                         line-breaking))
1014         (setq attributes (delq 'digit-value attributes))
1015         )
1016       (when (and (memq 'numeric-value attributes)
1017                  (setq value (get-char-attribute char 'numeric-value)))
1018         (insert (format "(numeric-value\t. %S)%s"
1019                         value
1020                         line-breaking))
1021         (setq attributes (delq 'numeric-value attributes))
1022         )))
1023     (when (and (memq 'iso-10646-comment attributes)
1024                (setq value (get-char-attribute char 'iso-10646-comment)))
1025       (insert (format "(iso-10646-comment\t. %S)%s"
1026                       value
1027                       line-breaking))
1028       (setq attributes (delq 'iso-10646-comment attributes))
1029       )
1030     (when (and (memq 'morohashi-daikanwa attributes)
1031                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1032       (insert (format "(morohashi-daikanwa\t%s)%s"
1033                       (mapconcat (function prin1-to-string) value " ")
1034                       line-breaking))
1035       (setq attributes (delq 'morohashi-daikanwa attributes))
1036       )
1037     (setq radical nil
1038           strokes nil)
1039     (when (and (memq 'ideographic-radical attributes)
1040                (setq value (get-char-attribute char 'ideographic-radical)))
1041       (setq radical value)
1042       (insert (format "(ideographic-radical . %S)\t; %c%s"
1043                       radical
1044                       (ideographic-radical radical)
1045                       line-breaking))
1046       (setq attributes (delq 'ideographic-radical attributes))
1047       )
1048     (when (and (memq 'shuowen-radical attributes)
1049                (setq value (get-char-attribute char 'shuowen-radical)))
1050       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1051                       value
1052                       (shuowen-radical value)
1053                       line-breaking))
1054       (setq attributes (delq 'shuowen-radical attributes))
1055       )
1056     (let (key)
1057       (dolist (domain
1058                (append
1059                 char-db-feature-domains
1060                 (let (dest domain)
1061                   (dolist (feature (char-attribute-list))
1062                     (setq feature (symbol-name feature))
1063                     (when (string-match
1064                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1065                            feature)
1066                       (setq domain (intern (match-string 2 feature)))
1067                      (unless (memq domain dest)
1068                        (setq dest (cons domain dest)))))
1069                   (sort dest #'string<))))
1070         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1071         (when (and (memq key attributes)
1072                    (setq value (get-char-attribute char key)))
1073           (setq radical value)
1074           (insert (format "(%s . %S)\t; %c%s"
1075                           key
1076                           radical
1077                           (ideographic-radical radical)
1078                           line-breaking))
1079           (setq attributes (delq key attributes))
1080           )
1081         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1082         (when (and (memq key attributes)
1083                    (setq value (get-char-attribute char key)))
1084           (setq strokes value)
1085           (insert (format "(%s . %S)%s"
1086                           key
1087                           strokes
1088                           line-breaking))
1089           (setq attributes (delq key attributes))
1090           )
1091         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1092         (when (and (memq key attributes)
1093                    (setq value (get-char-attribute char key)))
1094           (insert (format "(%s       . %S)%s"
1095                           key
1096                           value
1097                           line-breaking))
1098           (setq attributes (delq key attributes))
1099           )
1100         (dolist (feature '(ideographic-radical
1101                            ideographic-strokes
1102                            total-strokes))
1103           (setq key (intern (format "%s@%s*sources" feature domain)))
1104           (when (and (memq key attributes)
1105                      (setq value (get-char-attribute char key)))
1106             (insert (format "(%s%s" key line-breaking))
1107             (dolist (cell value)
1108               (insert (format " %s" cell)))
1109             (insert ")")
1110             (insert line-breaking)
1111             (setq attributes (delq key attributes))
1112             ))
1113         ))
1114     (when (and (memq 'ideographic-strokes attributes)
1115                (setq value (get-char-attribute char 'ideographic-strokes)))
1116       (setq strokes value)
1117       (insert (format "(ideographic-strokes . %S)%s"
1118                       strokes
1119                       line-breaking))
1120       (setq attributes (delq 'ideographic-strokes attributes))
1121       )
1122     (when (and (memq 'kangxi-radical attributes)
1123                (setq value (get-char-attribute char 'kangxi-radical)))
1124       (unless (eq value radical)
1125         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1126                         value
1127                         (ideographic-radical value)
1128                         line-breaking))
1129         (or radical
1130             (setq radical value)))
1131       (setq attributes (delq 'kangxi-radical attributes))
1132       )
1133     (when (and (memq 'kangxi-strokes attributes)
1134                (setq value (get-char-attribute char 'kangxi-strokes)))
1135       (unless (eq value strokes)
1136         (insert (format "(kangxi-strokes\t . %S)%s"
1137                         value
1138                         line-breaking))
1139         (or strokes
1140             (setq strokes value)))
1141       (setq attributes (delq 'kangxi-strokes attributes))
1142       )
1143     (when (and (memq 'japanese-radical attributes)
1144                (setq value (get-char-attribute char 'japanese-radical)))
1145       (unless (eq value radical)
1146         (insert (format "(japanese-radical\t . %S)\t; %c%s"
1147                         value
1148                         (ideographic-radical value)
1149                         line-breaking))
1150         (or radical
1151             (setq radical value)))
1152       (setq attributes (delq 'japanese-radical attributes))
1153       )
1154     (when (and (memq 'japanese-strokes attributes)
1155                (setq value (get-char-attribute char 'japanese-strokes)))
1156       (unless (eq value strokes)
1157         (insert (format "(japanese-strokes\t . %S)%s"
1158                         value
1159                         line-breaking))
1160         (or strokes
1161             (setq strokes value)))
1162       (setq attributes (delq 'japanese-strokes attributes))
1163       )
1164     (when (and (memq 'cns-radical attributes)
1165                (setq value (get-char-attribute char 'cns-radical)))
1166       (insert (format "(cns-radical\t . %S)\t; %c%s"
1167                       value
1168                       (ideographic-radical value)
1169                       line-breaking))
1170       (setq attributes (delq 'cns-radical attributes))
1171       )
1172     (when (and (memq 'cns-strokes attributes)
1173                (setq value (get-char-attribute char 'cns-strokes)))
1174       (unless (eq value strokes)
1175         (insert (format "(cns-strokes\t . %S)%s"
1176                         value
1177                         line-breaking))
1178         (or strokes
1179             (setq strokes value)))
1180       (setq attributes (delq 'cns-strokes attributes))
1181       )
1182     ;; (when (and (memq 'shinjigen-1-radical attributes)
1183     ;;            (setq value (get-char-attribute char 'shinjigen-1-radical)))
1184     ;;   (unless (eq value radical)
1185     ;;     (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
1186     ;;                     value
1187     ;;                     (ideographic-radical value)
1188     ;;                     line-breaking))
1189     ;;     (or radical
1190     ;;         (setq radical value)))
1191     ;;   (setq attributes (delq 'shinjigen-1-radical attributes))
1192     ;;   )
1193     (when (and (memq 'ideographic- attributes)
1194                (setq value (get-char-attribute char 'ideographic-)))
1195       (insert "(ideographic-       ")
1196       (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1197             separator nil)
1198       (while (consp value)
1199         (setq cell (car value))
1200         (if (integerp cell)
1201             (setq cell (decode-char '=ucs cell)))
1202         (cond ((characterp cell)
1203                (if separator
1204                    (insert lbs))
1205                (if readable
1206                    (insert (format "%S" cell))
1207                  (char-db-insert-char-spec cell readable))
1208                (setq separator lbs))
1209               ((consp cell)
1210                (if separator
1211                    (insert lbs))
1212                (if (consp (car cell))
1213                    (char-db-insert-char-spec cell readable)
1214                  (char-db-insert-char-reference cell readable))
1215                (setq separator lbs))
1216               (t
1217                (if separator
1218                    (insert separator))
1219                (insert (prin1-to-string cell))
1220                (setq separator " ")))
1221         (setq value (cdr value)))
1222       (insert ")")
1223       (insert line-breaking)
1224       (setq attributes (delq 'ideographic- attributes)))
1225     (when (and (memq 'total-strokes attributes)
1226                (setq value (get-char-attribute char 'total-strokes)))
1227       (insert (format "(total-strokes       . %S)%s"
1228                       value
1229                       line-breaking))
1230       (setq attributes (delq 'total-strokes attributes))
1231       )
1232     (when (and (memq '->ideograph attributes)
1233                (setq value (get-char-attribute char '->ideograph)))
1234       (insert (format "(->ideograph\t%s)%s"
1235                       (mapconcat (lambda (code)
1236                                    (cond ((symbolp code)
1237                                           (symbol-name code))
1238                                          ((integerp code)
1239                                           (format "#x%04X" code))
1240                                          (t
1241                                           (format "%s %S"
1242                                                   line-breaking code))))
1243                                  value " ")
1244                       line-breaking))
1245       (setq attributes (delq '->ideograph attributes))
1246       )
1247     ;; (when (and (memq '->decomposition attributes)
1248     ;;            (setq value (get-char-attribute char '->decomposition)))
1249     ;;   (insert (format "(->decomposition\t%s)%s"
1250     ;;                   (mapconcat (lambda (code)
1251     ;;                                (cond ((symbolp code)
1252     ;;                                       (symbol-name code))
1253     ;;                                      ((characterp code)
1254     ;;                                       (if readable
1255     ;;                                           (format "%S" code)
1256     ;;                                         (format "#x%04X"
1257     ;;                                                 (char-int code))
1258     ;;                                         ))
1259     ;;                                      ((integerp code)
1260     ;;                                       (format "#x%04X" code))
1261     ;;                                      (t
1262     ;;                                       (format "%s%S" line-breaking code))))
1263     ;;                              value " ")
1264     ;;                   line-breaking))
1265     ;;   (setq attributes (delq '->decomposition attributes))
1266     ;;   )
1267     (if (equal (get-char-attribute char '->titlecase)
1268                (get-char-attribute char '->uppercase))
1269         (setq attributes (delq '->titlecase attributes)))
1270     (when (and (memq '->mojikyo attributes)
1271                (setq value (get-char-attribute char '->mojikyo)))
1272       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1273                       value (decode-char 'mojikyo value)
1274                       line-breaking))
1275       (setq attributes (delq '->mojikyo attributes))
1276       )
1277     (when (and (memq 'hanyu-dazidian-vol attributes)
1278                (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1279       (insert (format "(hanyu-dazidian-vol  . %d)%s"
1280                       value line-breaking))
1281       (setq attributes (delq 'hanyu-dazidian-vol attributes))
1282       )
1283     (when (and (memq 'hanyu-dazidian-page attributes)
1284                (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1285       (insert (format "(hanyu-dazidian-page . %d)%s"
1286                       value line-breaking))
1287       (setq attributes (delq 'hanyu-dazidian-page attributes))
1288       )
1289     (when (and (memq 'hanyu-dazidian-char attributes)
1290                (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1291       (insert (format "(hanyu-dazidian-char . %d)%s"
1292                       value line-breaking))
1293       (setq attributes (delq 'hanyu-dazidian-char attributes))
1294       )
1295     (unless readable
1296       (dolist (ignored '(composition
1297                          ->denotational <-subsumptive ->ucs-unified
1298                          ->ideographic-component-forms))
1299         (setq attributes (delq ignored attributes))))
1300     (while attributes
1301       (setq name (car attributes))
1302       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1303                   'value-is-empty)
1304         (cond ((setq ret (find-charset name))
1305                (setq name (charset-name ret))
1306                (when (not (memq name dest-ccss))
1307                  (setq dest-ccss (cons name dest-ccss))
1308                  (char-db-insert-ccs-feature name value line-breaking))
1309                )
1310               ((string-match "^=>ucs@" (symbol-name name))
1311                (insert (format "(%-18s . #x%04X)\t; %c%s"
1312                                name value (decode-char '=ucs value)
1313                                line-breaking))
1314                )
1315               ((eq name 'jisx0208-1978/4X)
1316                (insert (format "(%-18s . #x%04X)%s"
1317                                name value
1318                                line-breaking))
1319                )
1320               ((and
1321                 (not readable)
1322                 (not (eq name '->subsumptive))
1323                 (not (eq name '->uppercase))
1324                 (not (eq name '->lowercase))
1325                 (not (eq name '->titlecase))
1326                 (not (eq name '->canonical))
1327                 (not (eq name '->Bopomofo))
1328                 (not (eq name '->mistakable))
1329                 (not (eq name '->ideographic-variants))
1330                 (null (get-char-attribute
1331                        char (intern (format "%s*sources" name))))
1332                 (not (string-match "\\*sources$" (symbol-name name)))
1333                 (null (get-char-attribute
1334                        char (intern (format "%s*note" name))))
1335                 (not (string-match "\\*note$" (symbol-name name)))
1336                 (or (eq name '<-identical)
1337                     (eq name '<-uppercase)
1338                     (eq name '<-lowercase)
1339                     (eq name '<-titlecase)
1340                     (eq name '<-canonical)
1341                     (eq name '<-ideographic-variants)
1342                     ;; (eq name '<-synonyms)
1343                     (string-match "^<-synonyms" (symbol-name name))
1344                     (eq name '<-mistakable)
1345                     (when (string-match "^->" (symbol-name name))
1346                       (cond
1347                        ((string-match "^->fullwidth" (symbol-name name))
1348                         (not (and (consp value)
1349                                   (characterp (car value))
1350                                   (encode-char
1351                                    (car value) '=ucs 'defined-only)))
1352                         )
1353                        (t)))
1354                     ))
1355                )
1356               ((or (eq name 'ideographic-structure)
1357                    (eq name 'ideographic-combination)
1358                    (eq name 'ideographic-)
1359                    (eq name '=decomposition)
1360                    (char-feature-base-name= '=decomposition name)
1361                    (char-feature-base-name= '=>decomposition name)
1362                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1363                    ;;               (symbol-name name))
1364                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1365                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1366                                  (symbol-name name))
1367                    )
1368                (char-db-insert-relation-feature char name value
1369                                                 line-breaking
1370                                                 ccss readable))
1371               ((memq name '(ideograph=
1372                             original-ideograph-of
1373                             ancient-ideograph-of
1374                             vulgar-ideograph-of
1375                             wrong-ideograph-of
1376                             ;; simplified-ideograph-of
1377                             ideographic-variants
1378                             ;; ideographic-different-form-of
1379                             ))
1380                (insert (format "(%-18s%s " name line-breaking))
1381                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1382                      separator nil)
1383                (while (consp value)
1384                  (setq cell (car value))
1385                  (if (and (consp cell)
1386                           (consp (car cell)))
1387                      (progn
1388                        (if separator
1389                            (insert lbs))
1390                        (char-db-insert-alist cell readable)
1391                        (setq separator lbs))
1392                    (if separator
1393                        (insert separator))
1394                    (insert (prin1-to-string cell))
1395                    (setq separator " "))
1396                  (setq value (cdr value)))
1397                (insert ")")
1398                (insert line-breaking))
1399               ((consp value)
1400                (insert (format "(%-18s " name))
1401                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1402                      separator nil)
1403                (while (consp value)
1404                  (setq cell (car value))
1405                  (if (and (consp cell)
1406                           (consp (car cell))
1407                           (setq ret (condition-case nil
1408                                         (find-char cell)
1409                                       (error nil))))
1410                      (progn
1411                        (setq rest cell
1412                              al nil
1413                              cal nil)
1414                        (while rest
1415                          (setq key (car (car rest)))
1416                          (if (find-charset key)
1417                              (setq cal (cons key cal))
1418                            (setq al (cons key al)))
1419                          (setq rest (cdr rest)))
1420                        (if separator
1421                            (insert lbs))
1422                        (insert-char-attributes ret
1423                                                readable
1424                                                al ; cal
1425                                                nil 'for-sub-node)
1426                        (setq separator lbs))
1427                    (setq ret (prin1-to-string cell))
1428                    (if separator
1429                        (if (< (+ (current-column)
1430                                  (length ret)
1431                                  (length separator))
1432                               76)
1433                            (insert separator)
1434                          (insert lbs)))
1435                    (insert ret)
1436                    (setq separator " "))
1437                  (setq value (cdr value)))
1438                (insert ")")
1439                (insert line-breaking))
1440               (t
1441                (insert (format "(%-18s" name))
1442                (setq ret (prin1-to-string value))
1443                (unless (< (+ (current-column)
1444                              (length ret)
1445                              3)
1446                           76)
1447                  (insert line-breaking))
1448                (insert " . " ret ")" line-breaking)
1449                ;; (insert (format "(%-18s . %S)%s"
1450                ;;                 name value
1451                ;;                 line-breaking))
1452                )
1453               ))
1454       (setq attributes (cdr attributes)))
1455     (insert ")")))
1456
1457 (defun insert-char-data (char &optional readable
1458                               attributes)
1459   (save-restriction
1460     (narrow-to-region (point)(point))
1461     (insert "(define-char
1462   '")
1463     (insert-char-attributes char readable attributes)
1464     (insert ")\n")
1465     (goto-char (point-min))
1466     (while (re-search-forward "[ \t]+$" nil t)
1467       (replace-match ""))
1468     ;; from tabify.
1469     (goto-char (point-min))
1470     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1471       (let ((column (current-column))
1472             (indent-tabs-mode t))
1473         (delete-region (match-beginning 0) (point))
1474         (indent-to column)))
1475     (goto-char (point-max))
1476     ;; (tabify (point-min)(point-max))
1477     ))
1478
1479 (defun insert-char-data-with-variant (char &optional printable
1480                                            no-ucs-unified
1481                                            script excluded-script)
1482   (insert-char-data char printable)
1483   (let ((variants (char-variants char))
1484         rest
1485         variant vs ret)
1486     (setq variants (sort variants #'<))
1487     (setq rest variants)
1488     (setq variants (cons char variants))
1489     (while rest
1490       (setq variant (car rest))
1491       (unless (get-char-attribute variant '<-subsumptive)
1492         (if (and (or (null script)
1493                      (null (setq vs (get-char-attribute variant 'script)))
1494                      (memq script vs))
1495                  (or (null excluded-script)
1496                      (null (setq vs (get-char-attribute variant 'script)))
1497                      (not (memq excluded-script vs))))
1498             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1499               (insert-char-data variant printable)
1500               (if (setq ret (char-variants variant))
1501                   (while ret
1502                     (or (memq (car ret) variants)
1503                         ;; (get-char-attribute (car ret) '<-subsumptive)
1504                         (setq rest (nconc rest (list (car ret)))))
1505                     (setq ret (cdr ret)))))))
1506       (setq rest (cdr rest)))))
1507
1508 (defun insert-char-range-data (min max &optional script excluded-script)
1509   (let ((code min)
1510         char)
1511     (while (<= code max)
1512       (setq char (decode-char '=ucs code))
1513       (if (encode-char char '=ucs 'defined-only)
1514           (insert-char-data-with-variant char nil 'no-ucs-unified
1515                                          script excluded-script))
1516       (setq code (1+ code)))))
1517
1518 (defun write-char-range-data-to-file (min max file
1519                                           &optional script excluded-script)
1520   (let ((coding-system-for-write char-db-file-coding-system))
1521     (with-temp-buffer
1522       (insert (format ";; -*- coding: %s -*-\n"
1523                       char-db-file-coding-system))
1524       (insert-char-range-data min max script excluded-script)
1525       (write-region (point-min)(point-max) file))))
1526
1527 (defvar what-character-original-window-configuration)
1528
1529 ;;;###autoload
1530 (defun what-char-definition (char)
1531   (interactive (list (char-after)))
1532   (let ((buf (get-buffer-create "*Character Description*"))
1533         (the-buf (current-buffer))
1534         (win-conf (current-window-configuration)))
1535     (pop-to-buffer buf)
1536     (make-local-variable 'what-character-original-window-configuration)
1537     (setq what-character-original-window-configuration win-conf)
1538     (setq buffer-read-only nil)
1539     (erase-buffer)
1540     (condition-case err
1541         (progn
1542           (insert-char-data-with-variant char 'printable)
1543           (unless (char-attribute-alist char)
1544             (insert (format ";; = %c\n"
1545                             (let* ((rest (split-char char))
1546                                    (ccs (pop rest))
1547                                    (code (pop rest)))
1548                               (while rest
1549                                 (setq code (logior (lsh code 8)
1550                                                    (pop rest))))
1551                               (decode-char ccs code)))))
1552           ;; (char-db-update-comment)
1553           (set-buffer-modified-p nil)
1554           (view-mode the-buf (lambda (buf)
1555                                (set-window-configuration
1556                                 what-character-original-window-configuration)
1557                                ))
1558           (goto-char (point-min)))
1559       (error (progn
1560                (set-window-configuration
1561                 what-character-original-window-configuration)
1562                (signal (car err) (cdr err)))))))
1563
1564
1565 ;;; @ end
1566 ;;;
1567
1568 (provide 'char-db-util)
1569
1570 ;;; char-db-util.el ends here