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 ;;   2008,2009,2010,2011,2012,2013,2014,2015,2016 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     =hanyo-denshi/tk
 172     ;; ==mj
 173     ;; ==adobe-japan1-0
 174     ;; ==adobe-japan1-1
 175     ;; ==adobe-japan1-2
 176     ;; ==adobe-japan1-3
 177     ;; ==adobe-japan1-4
 178     ;; ==adobe-japan1-5
 179     ;; ==adobe-japan1-6
 180     ==jis-x0208
 181     ==jis-x0213-1
 182     ==jis-x0213-2
 183     ==jis-x0212
 184     ==hanyo-denshi/ja
 185     ==hanyo-denshi/jb
 186     ==hanyo-denshi/jc
 187     ==hanyo-denshi/ft
 188     ==hanyo-denshi/ib
 189     ==hanyo-denshi/hg
 190     ==hanyo-denshi/jt
 191     ==hanyo-denshi/ks
 192     =gt
 193     =gt-k
 194     =daikanwa
 195     =daikanwa@rev2
 196     =daikanwa@rev1
 197     =daikanwa/+p
 198     ==daikanwa
 199     =cbeta
 200     ideograph-hanziku-1
 201     ideograph-hanziku-2
 202     ideograph-hanziku-3
 203     ideograph-hanziku-4
 204     ideograph-hanziku-5
 205     ideograph-hanziku-6
 206     ideograph-hanziku-7
 207     ideograph-hanziku-8
 208     ideograph-hanziku-9
 209     ideograph-hanziku-10
 210     ideograph-hanziku-11
 211     ideograph-hanziku-12
 212     ;; =>>>adobe-japan1-0
 213     ;; =>>>adobe-japan1-1
 214     ;; =>>>adobe-japan1-2
 215     ;; =>>>adobe-japan1-3
 216     ;; =>>>adobe-japan1-4
 217     ;; =>>>adobe-japan1-5
 218     ;; =>>>adobe-japan1-6
 219     ;; =>>>jis-x0208
 220     ;; =>>>jis-x0213-1
 221     ;; =>>>jis-x0213-2
 222     ;; =>>>hanyo-denshi/ja
 223     ;; =>>>hanyo-denshi/jb
 224     ;; =>>>hanyo-denshi/jc
 225     ;; =>>>hanyo-denshi/ft
 226     ;; =>>>hanyo-denshi/ib
 227     ;; =>>>hanyo-denshi/hg
 228     ;; =>>>hanyo-denshi/jt
 229     ;; =>>>hanyo-denshi/ks
 230     ;; =>>>gt
 231     =>>adobe-japan1-0
 232     =>>adobe-japan1-1
 233     =>>adobe-japan1-2
 234     =>>adobe-japan1-3
 235     =>>adobe-japan1-4
 236     =>>adobe-japan1-5
 237     =>>adobe-japan1-6
 238     =>>jis-x0208
 239     =>>jis-x0213-1
 240     =>>jis-x0213-1@2000
 241     =>>jis-x0213-1@2004
 242     =>>jis-x0213-2
 243     =>>jis-x0208@1978
 244     =>>hanyo-denshi/ft
 245     =>>hanyo-denshi/jt
 246     =>>hanyo-denshi/ks
 247     =>>gt
 248     =>>daikanwa
 249     =>>cbeta
 250     =+>jis-x0208
 251     =+>jis-x0213-1
 252     =+>jis-x0213-2
 253     =+>adobe-japan1-0
 254     =+>adobe-japan1-1
 255     =+>adobe-japan1-2
 256     =+>adobe-japan1-3
 257     =+>adobe-japan1-4
 258     =+>adobe-japan1-5
 259     =+>adobe-japan1-6
 260     =+>jis-x0208@1978
 261     =>jis-x0208
 262     =>jis-x0208@1997
 263     =>jis-x0213-1
 264     =>jis-x0213-1@2000
 265     =>jis-x0213-1@2004
 266     =>jis-x0213-2
 267     ==>ucs@bucs
 268     =>iwds-1
 269     =>ucs@hanyo-denshi
 270     =>ucs@iso
 271     =>ucs@unicode
 272     =>ucs@jis
 273     =>ucs@cns
 274     =>ucs@ks
 275     =+>ucs@iso
 276     =+>ucs@unicode
 277     =+>ucs@jis
 278     =+>ucs@jis/1990
 279     =+>ucs@cns
 280     =+>ucs@ks
 281     =>>ucs@iso
 282     =>>ucs@unicode
 283     =>>ucs@jis
 284     =>>ucs@cns
 285     =>>>ucs@iso
 286     =>>>ucs@unicode
 287     ==ucs@iso
 288     ==ucs@unicode
 289     ;; ==ucs@cns
 290     ==gb2312
 291     ==ks-x1001
 292     ==cns11643-1
 293     ==cns11643-2
 294     ==cns11643-3
 295     ==cns11643-4
 296     ==cns11643-5
 297     ==cns11643-6
 298     ==cns11643-7
 299     ==gt
 300     ==jis-x0208@1990
 301     ;; ==jis-x0208@1983
 302     ==jis-x0208@1978
 303     ==gt-k
 304     =ucs@iso
 305     =ucs@unicode
 306     =ucs@cns
 307     ==big5-cdp
 308     ==cbeta
 309     =>>big5-cdp
 310     =>>gt-k
 311     =+>gt
 312     =>gt
 313     =>mj
 314     =>big5-cdp
 315     =>daikanwa
 316     =>daikanwa/ho
 317     =>cns11643-5
 318     =>cns11643-7
 319     =big5
 320     =big5-eten
 321     =>gt-k
 322     =zinbun-oracle
 323     =>zinbun-oracle
 324     =ruimoku-v6
 325     =>>ruimoku-v6
 326     ==ruimoku-v6
 327     =jef-china3
 328     =>cbeta
 329     =shinjigen
 330     =ucs-var-001
 331     =ucs-var-002
 332     =ucs-var-003
 333     =ucs-var-004
 334     =ucs-var-006
 335     =ucs-var-010
 336     =ucs-itaiji-001
 337     =ucs-itaiji-002
 338     =ucs-itaiji-003
 339     =ucs-itaiji-004
 340     =ucs-itaiji-005
 341     =ucs-itaiji-006
 342     =ucs-itaiji-008
 343     =ucs-itaiji-009
 344     =big5-cdp-var-003
 345     =big5-cdp-var-004
 346     =big5-cdp-var-005
 347     =>ucs@iwds-1
 348     ==ucs@gb
 349     =ucs@JP/hanazono))
 350
 351
 352 ;;; @ char-db formatters
 353 ;;;
 354
 355 (defun char-db-make-char-spec (char)
 356   (let (ret char-spec)
 357     (cond ((characterp char)
 358            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 359                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 360                   (setq char-spec (list (cons '=ucs ret)))
 361                   (cond ((setq ret (get-char-attribute char 'name))
 362                          (setq char-spec (cons (cons 'name ret) char-spec))
 363                          )
 364                         ((setq ret (get-char-attribute char 'name*))
 365                          (setq char-spec (cons (cons 'name* ret) char-spec))
 366                          ))
 367                   )
 368                  ((encode-char char '=mj 'defined-only)
 369                   (setq char-spec nil)
 370                   (dolist (ccs (charset-list))
 371                     (if (and (or (eq ccs '=mj)
 372                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 373                                  (string-match "=ucs@" (symbol-name ccs))
 374                                  )
 375                              (setq ccs (charset-name ccs))
 376                              (null (assq ccs char-spec))
 377                              (setq ret (encode-char char ccs 'defined-only)))
 378                         (setq char-spec (cons (cons ccs ret) char-spec))))
 379                   )
 380                  ((encode-char char '==mj 'defined-only)
 381                   (setq char-spec nil)
 382                   (dolist (ccs (charset-list))
 383                     (if (and (or (eq ccs '==mj)
 384                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 385                                  (string-match "=ucs@" (symbol-name ccs))
 386                                  )
 387                              (setq ccs (charset-name ccs))
 388                              (null (assq ccs char-spec))
 389                              (setq ret (encode-char char ccs 'defined-only)))
 390                         (setq char-spec (cons (cons ccs ret) char-spec))))
 391                   )
 392                  ((encode-char char '=adobe-japan1 'defined-only)
 393                   (setq char-spec nil)
 394                   (dolist (ccs (charset-list))
 395                     (if (and (or (memq ccs
 396                                        '(=adobe-japan1-0
 397                                          =adobe-japan1-1
 398                                          =adobe-japan1-2
 399                                          =adobe-japan1-3
 400                                          =adobe-japan1-4
 401                                          =adobe-japan1-5
 402                                          =adobe-japan1-6
 403                                          ))
 404                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 405                                  (string-match "=ucs@" (symbol-name ccs))
 406                                  )
 407                              (setq ccs (charset-name ccs))
 408                              (null (assq ccs char-spec))
 409                              (setq ret (encode-char char ccs 'defined-only)))
 410                         (setq char-spec (cons (cons ccs ret) char-spec))))
 411                   )
 412                  ((encode-char char '==adobe-japan1 'defined-only)
 413                   (setq char-spec nil)
 414                   (dolist (ccs (charset-list))
 415                     (if (and (or (memq ccs
 416                                        '(==adobe-japan1-0
 417                                          ==adobe-japan1-1
 418                                          ==adobe-japan1-2
 419                                          ==adobe-japan1-3
 420                                          ==adobe-japan1-4
 421                                          ==adobe-japan1-5
 422                                          ==adobe-japan1-6
 423                                          ))
 424                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 425                                  (string-match "=ucs@" (symbol-name ccs))
 426                                  )
 427                              (setq ccs (charset-name ccs))
 428                              (null (assq ccs char-spec))
 429                              (setq ret (encode-char char ccs 'defined-only)))
 430                         (setq char-spec (cons (cons ccs ret) char-spec))))
 431                   )
 432                  ((setq ret
 433                         (catch 'tag
 434                           (let ((rest char-db-coded-charset-priority-list)
 435                                 ccs)
 436                             (while rest
 437                               (setq ccs (charset-name
 438                                          (find-charset (car rest))))
 439                               (if (setq ret
 440                                         (encode-char char ccs
 441                                                      'defined-only))
 442                                   (throw 'tag (cons ccs ret)))
 443                               (setq rest (cdr rest))))))
 444                   (setq char-spec (list ret))
 445                   (dolist (ccs (delq (car ret) (charset-list)))
 446                     (if (and (or (charset-iso-final-char ccs)
 447                                  (memq ccs
 448                                        '(=daikanwa
 449                                          =daikanwa@rev2
 450                                          ;; =gt-k
 451                                          =jis-x0208@1997
 452                                          ))
 453                                  (eq (charset-property ccs 'iso-ir) 177)
 454                                  ;; (string-match "=ucs@" (symbol-name ccs))
 455                                  )
 456                              (setq ccs (charset-name ccs))
 457                              (null (assq ccs char-spec))
 458                              (setq ret (encode-char char ccs 'defined-only)))
 459                         (setq char-spec (cons (cons ccs ret) char-spec))))
 460                   (if (null char-spec)
 461                       (setq char-spec (split-char char)))
 462                   (cond ((setq ret (get-char-attribute char 'name))
 463                          (setq char-spec (cons (cons 'name ret) char-spec))
 464                          )
 465                         ((setq ret (get-char-attribute char 'name*))
 466                          (setq char-spec (cons (cons 'name* ret) char-spec))
 467                          ))
 468                   )
 469                  ((setq ret (get-char-attribute
 470                              char 'ideographic-combination))
 471                   (setq char-spec
 472                         (cons (cons 'ideographic-combination ret)
 473                               char-spec))
 474                   ))
 475            char-spec)
 476           ((consp char)
 477            char))))
 478
 479 (defun char-db-insert-char-spec (char &optional readable column
 480                                       required-features)
 481   (unless column
 482     (setq column (current-column)))
 483   (let (char-spec temp-char)
 484     (setq char-spec (char-db-make-char-spec char))
 485     (unless (or (characterp char) ; char
 486                 (condition-case nil
 487                     (setq char (find-char char-spec))
 488                   (error nil)))
 489       ;; define temporary character
 490       ;;   Current implementation is dirty.
 491       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 492                                          char-spec)))
 493       (remove-char-attribute temp-char 'ideograph-daikanwa)
 494       (setq char temp-char))
 495     (insert-char-attributes char
 496                             readable
 497                             (union (mapcar #'car char-spec)
 498                                    required-features)
 499                             nil 'for-sub-node)
 500     (when temp-char
 501       ;; undefine temporary character
 502       ;;   Current implementation is dirty.
 503       (setq char-spec (char-attribute-alist temp-char))
 504       (while char-spec
 505         (remove-char-attribute temp-char (car (car char-spec)))
 506         (setq char-spec (cdr char-spec))))))
 507
 508 (defun char-db-insert-alist (alist &optional readable column)
 509   (unless column
 510     (setq column (current-column)))
 511   (let ((line-breaking
 512          (concat "\n" (make-string (1+ column) ?\ )))
 513         name value
 514         ret al ; cal
 515         key
 516         lbs cell rest separator)
 517     (insert "(")
 518     (while alist
 519       (setq name (car (car alist))
 520             value (cdr (car alist)))
 521       (cond ((eq name 'char)
 522              (insert "(char . ")
 523              (if (setq ret (condition-case nil
 524                                (find-char value)
 525                              (error nil)))
 526                  (progn
 527                    (setq al nil
 528                          ;; cal nil
 529                          )
 530                    (while value
 531                      (setq key (car (car value)))
 532                      ;; (if (find-charset key)
 533                      ;;     (setq cal (cons key cal))
 534                      (setq al (cons key al))
 535                      ;; )
 536                      (setq value (cdr value)))
 537                    (insert-char-attributes ret
 538                                            readable
 539                                            (or al 'none) ; cal
 540                                            nil 'for-sub-node))
 541                (insert (prin1-to-string value)))
 542              (insert ")")
 543              (insert line-breaking))
 544             ((consp value)
 545              (insert (format "(%-18s " name))
 546              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 547              (while (consp value)
 548                (setq cell (car value))
 549                (if (and (consp cell)
 550                         (consp (car cell))
 551                         (setq ret (condition-case nil
 552                                       (find-char cell)
 553                                     (error nil)))
 554                         )
 555                    (progn
 556                      (setq rest cell
 557                            al nil
 558                            ;; cal nil
 559                            )
 560                      (while rest
 561                        (setq key (car (car rest)))
 562                        ;; (if (find-charset key)
 563                        ;;     (setq cal (cons key cal))
 564                        (setq al (cons key al))
 565                        ;; )
 566                        (setq rest (cdr rest)))
 567                      (if separator
 568                          (insert lbs))
 569                      (insert-char-attributes ret
 570                                              readable
 571                                              al ; cal
 572                                              nil 'for-sub-node)
 573                      (setq separator lbs))
 574                  (if separator
 575                      (insert separator))
 576                  (insert (prin1-to-string cell))
 577                  (setq separator " "))
 578                (setq value (cdr value)))
 579              (insert ")")
 580              (insert line-breaking))
 581             (t
 582              (insert (format "(%-18s . %S)%s"
 583                              name value
 584                              line-breaking))))
 585       (setq alist (cdr alist))))
 586   (insert ")"))
 587
 588 (defun char-db-insert-char-reference (plist &optional readable column)
 589   (unless column
 590     (setq column (current-column)))
 591   (let ((line-breaking
 592          (concat "\n" (make-string (1+ column) ?\ )))
 593         (separator "")
 594         name value)
 595     (insert "(")
 596     (while plist
 597       (setq name (pop plist))
 598       (setq value (pop plist))
 599       (cond ((eq name :char)
 600              (insert separator)
 601              (insert ":char\t")
 602              (cond ((numberp value)
 603                     (setq value (decode-char '=ucs value)))
 604                    ;; ((consp value)
 605                    ;;  (setq value (or (find-char value)
 606                    ;;                  value)))
 607                    )
 608              (char-db-insert-char-spec value readable)
 609              (insert line-breaking)
 610              (setq separator ""))
 611             ((eq name :radical)
 612              (insert (format "%s%s\t%d ; %c%s"
 613                              separator
 614                              name value
 615                              (ideographic-radical value)
 616                              line-breaking))
 617              (setq separator ""))
 618             (t
 619              (insert (format "%s%s\t%S" separator name value))
 620              (setq separator line-breaking)))
 621       ))
 622   (insert ")"))
 623
 624 (defun char-db-decode-isolated-char (ccs code-point)
 625   (let (ret)
 626     (setq ret
 627           (cond ((eq ccs 'arabic-iso8859-6)
 628                  (decode-char ccs code-point))
 629                 ;; ((eq ccs '=gt)
 630                 ;;  (decode-builtin-char '==gt code-point))
 631                 ((and (memq ccs '(=gt-pj-1
 632                                   =gt-pj-2
 633                                   =gt-pj-3
 634                                   =gt-pj-4
 635                                   =gt-pj-5
 636                                   =gt-pj-6
 637                                   =gt-pj-7
 638                                   =gt-pj-8
 639                                   =gt-pj-9
 640                                   =gt-pj-10
 641                                   =gt-pj-11))
 642                       (setq ret (decode-char ccs code-point))
 643                       (setq ret (encode-char ret '=gt 'defined-only)))
 644                  (decode-builtin-char '=gt ret))
 645                 (t
 646                  (decode-builtin-char ccs code-point))))
 647     (cond ((null ret)
 648            (or (decode-char ccs code-point)
 649                (define-char (list (cons ccs code-point)))))
 650           ((and (<= 0 (char-int ret))
 651                 (<= (char-int ret) #x1F))
 652            (decode-char '=ucs (+ #x2400 (char-int ret))))
 653           ((= (char-int ret) #x7F)
 654            ?\u2421)
 655           (t ret))))
 656
 657 (defvar char-db-convert-obsolete-format t)
 658
 659 (defun char-db-insert-ccs-feature (name value line-breaking)
 660   (cond
 661    ((integerp value)
 662     (insert
 663      (format
 664       (cond
 665        ((memq name '(=>iwds-1
 666                      ==shinjigen
 667                      =shinjigen
 668                      =shinjigen@1ed ==shinjigen@1ed
 669                      =shinjigen@rev ==shinjigen@rev
 670                      =shinjigen/+p@rev ==shinjigen/+p@rev
 671                      ===daikanwa/ho ==daikanwa/ho
 672                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 673         "(%-18s .  %04d)\t; %c")
 674        ((eq name '=shinjigen@1ed/24pr)
 675         "(%-18s . %04d)\t; %c")
 676        ((or
 677          (memq name
 678                '(===daikanwa
 679                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 680                  =daikanwa@rev1 =daikanwa@rev2
 681                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 682                  =>>daikanwa/+p
 683                  =daikanwa/+2p =>>daikanwa/+2p
 684                  =gt ==gt ===gt
 685                  =>>gt =+>gt =>gt
 686                  =gt-k ==gt-k ===gt-k
 687                  =>>gt-k =>gt-k
 688                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 689                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 690                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 691                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 692                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 693                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 694                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 695                  =>>adobe-japan1-0 =+>adobe-japan1-0
 696                  =>>adobe-japan1-1 =+>adobe-japan1-1
 697                  =>>adobe-japan1-2 =+>adobe-japan1-2
 698                  =>>adobe-japan1-3 =+>adobe-japan1-3
 699                  =>>adobe-japan1-4 =+>adobe-japan1-4
 700                  =>>adobe-japan1-5 =+>adobe-japan1-5
 701                  =>>adobe-japan1-6 =+>adobe-japan1-6
 702                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 703                  =zinbun-oracle =>zinbun-oracle
 704                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 705                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 706                  ===hng-khi ===hng-khm ===hng-hok ===hng-kyd ===hng-sok
 707                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 708                  ===hng-sys ===hng-tsu ===hng-tzj
 709                  ===hng-hos ===hng-nak ===hng-jhk
 710                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 711                  ===hng-nkm ===hng-k24 ===hng-nkk
 712                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 713                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 714                  ===hng-okd ===hng-wad ===hng-kmi
 715                  ===hng-zkd ===hng-doh ===hng-jyu
 716                  ===hng-tzs ===hng-kss ===hng-kyo
 717                  ===hng-smk))
 718          ;; (string-match "^=adobe-" (symbol-name name))
 719          )
 720         "(%-18s . %05d)\t; %c")
 721        ((memq name '(=hanyo-denshi/ks
 722                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 723                      =>>hanyo-denshi/ks
 724                      =koseki ==koseki
 725                      =mj ==mj ===mj =>>mj =>mj
 726                      =zihai mojikyo))
 727         "(%-18s . %06d)\t; %c")
 728        ((memq name '(=hanyo-denshi/tk ==hanyo-denshi/tk))
 729         "(%-18s . %08d)\t; %c")
 730        ((>= (charset-dimension name) 2)
 731         "(%-18s . #x%04X)\t; %c")
 732        (t
 733         "(%-18s . #x%02X)\t; %c"))
 734       name
 735       (if (= (charset-iso-graphic-plane name) 1)
 736           (logior value
 737                   (cond ((= (charset-dimension name) 1)
 738                          #x80)
 739                         ((= (charset-dimension name) 2)
 740                          #x8080)
 741                         ((= (charset-dimension name) 3)
 742                          #x808080)
 743                         (t 0)))
 744         value)
 745       (char-db-decode-isolated-char name value)))
 746     (if (and (= (charset-chars name) 94)
 747              (= (charset-dimension name) 2))
 748         (insert (format " [%02d-%02d]"
 749                         (- (lsh value -8) 32)
 750                         (- (logand value 255) 32))))
 751     )
 752    (t
 753     (insert (format "(%-18s . %s)" name value))
 754     ))
 755   (insert line-breaking))
 756
 757 (defun char-db-insert-relation-feature (char name value line-breaking
 758                                              ccss readable)
 759   (insert (format "(%-18s%s " name line-breaking))
 760   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 761         separator cell sources required-features
 762         ret)
 763     (while (consp value)
 764       (setq cell (car value))
 765       (if (integerp cell)
 766           (setq cell (decode-char '=ucs cell)))
 767       (cond
 768        ((eq name '->subsumptive)
 769         (when (or (not (some (lambda (atr)
 770                                (get-char-attribute cell atr))
 771                              char-db-ignored-attributes))
 772                   (some (lambda (ccs)
 773                           (encode-char cell ccs 'defined-only))
 774                         ccss))
 775           (if separator
 776               (insert lbs))
 777           (let ((char-db-ignored-attributes
 778                  (cons '<-subsumptive
 779                        char-db-ignored-attributes)))
 780             (insert-char-attributes cell readable nil nil 'for-sub-node))
 781           (setq separator lbs))
 782         )
 783        ((characterp cell)
 784         (setq sources
 785               (get-char-attribute
 786                char (intern (format "%s*sources" name))))
 787         (setq required-features nil)
 788         (dolist (source sources)
 789           (cond
 790            ((memq source '(JP
 791                            JP/Jouyou
 792                            shinjigen shinjigen@1ed shinjigen@rev))
 793             (setq required-features
 794                   (union required-features
 795                          '(=jis-x0208
 796                            =jis-x0208@1990
 797                            =jis-x0213-1@2000
 798                            =jis-x0213-1@2004
 799                            =jis-x0213-2
 800                            =jis-x0212
 801                            =jis-x0208@1983
 802                            =jis-x0208@1978
 803                            =shinjigen
 804                            =shinjigen@1ed
 805                            =shinjigen@rev
 806                            =shinjigen/+p@rev))))
 807            ((eq source 'CN)
 808             (setq required-features
 809                   (union required-features
 810                          '(=gb2312
 811                            =gb12345
 812                            =iso-ir165)))))
 813           (cond
 814            ((find-charset (setq ret (intern (format "=%s" source))))
 815             (setq required-features
 816                   (cons ret required-features)))
 817            (t (setq required-features
 818                     (cons source required-features)))))
 819         (cond ((string-match "@JP" (symbol-name name))
 820                (setq required-features
 821                      (union required-features
 822                             '(=jis-x0208
 823                               =jis-x0208@1990
 824                               =jis-x0213-1-2000
 825                               =jis-x0213-2-2000
 826                               =jis-x0212
 827                               =jis-x0208@1983
 828                               =jis-x0208@1978))))
 829               ((string-match "@CN" (symbol-name name))
 830                (setq required-features
 831                      (union required-features
 832                             '(=gb2312
 833                               =gb12345
 834                               =iso-ir165)))))
 835         (if separator
 836             (insert lbs))
 837         (if readable
 838             (insert (format "%S" cell))
 839           (char-db-insert-char-spec cell readable
 840                                     nil
 841                                     required-features))
 842         (setq separator lbs))
 843        ((consp cell)
 844         (if separator
 845             (insert lbs))
 846         (if (consp (car cell))
 847             (char-db-insert-char-spec cell readable)
 848           (char-db-insert-char-reference cell readable))
 849         (setq separator lbs))
 850        (t
 851         (if separator
 852             (insert separator))
 853         (insert (prin1-to-string cell))
 854         (setq separator " ")))
 855       (setq value (cdr value)))
 856     (insert ")")
 857     (insert line-breaking)))
 858
 859 (defun insert-char-attributes (char &optional readable attributes column
 860                                     for-sub-node)
 861   (unless column
 862     (setq column (current-column)))
 863   (let (name value ; has-long-ccs-name
 864         rest
 865         radical strokes
 866         (line-breaking
 867          (concat "\n" (make-string (1+ column) ?\ )))
 868         lbs cell separator ret
 869         key al cal
 870         dest-ccss ; sources required-features
 871         ccss)
 872     (let (atr-d)
 873       (setq attributes
 874             (sort (if attributes
 875                       (if (consp attributes)
 876                           (progn
 877                             (dolist (name attributes)
 878                               (unless (memq name char-db-ignored-attributes)
 879                                 (if (find-charset name)
 880                                     (push name ccss))
 881                                 (push name atr-d)))
 882                             atr-d))
 883                     (dolist (name (char-attribute-list))
 884                       (unless (memq name char-db-ignored-attributes)
 885                         (if (find-charset name)
 886                             (push name ccss))
 887                         (push name atr-d)))
 888                     atr-d)
 889                   #'char-attribute-name<)))
 890     (insert "(")
 891     (when (memq '<-subsumptive attributes)
 892       (when (or readable (not for-sub-node))
 893         (when (setq value (get-char-attribute char '<-subsumptive))
 894           (char-db-insert-relation-feature char '<-subsumptive value
 895                                            line-breaking
 896                                            ccss readable)))
 897       (setq attributes (delq '<-subsumptive attributes)))
 898     (when (and (memq '<-denotational attributes)
 899                (setq value (get-char-attribute char '<-denotational)))
 900       (char-db-insert-relation-feature char '<-denotational value
 901                                        line-breaking
 902                                        ccss readable)
 903       (setq attributes (delq '<-denotational attributes)))
 904     (when (and (memq '<-denotational@component attributes)
 905                (setq value (get-char-attribute char '<-denotational@component)))
 906       (char-db-insert-relation-feature char '<-denotational@component value
 907                                        line-breaking
 908                                        ccss readable)
 909       (setq attributes (delq '<-denotational@component attributes)))
 910     (when (and (memq 'name attributes)
 911                (setq value (get-char-attribute char 'name)))
 912       (insert (format
 913                (if (> (+ (current-column) (length value)) 48)
 914                    "(name . %S)%s"
 915                  "(name               . %S)%s")
 916                value line-breaking))
 917       (setq attributes (delq 'name attributes))
 918       )
 919     (when (and (memq 'name* attributes)
 920                (setq value (get-char-attribute char 'name*)))
 921       (insert (format
 922                (if (> (+ (current-column) (length value)) 48)
 923                    "(name* . %S)%s"
 924                  "(name*              . %S)%s")
 925                value line-breaking))
 926       (setq attributes (delq 'name* attributes))
 927       )
 928     (when (and (memq 'script attributes)
 929                (setq value (get-char-attribute char 'script)))
 930       (insert (format "(script\t\t%s)%s"
 931                       (mapconcat (function prin1-to-string)
 932                                  value " ")
 933                       line-breaking))
 934       (setq attributes (delq 'script attributes))
 935       )
 936     (dolist (name '(=>ucs =>ucs*))
 937       (when (and (memq name attributes)
 938                  (setq value (get-char-attribute char name)))
 939         (insert (format "(%-18s . #x%04X)\t; %c%s"
 940                         name value (decode-char '=ucs value)
 941                         line-breaking))
 942         (setq attributes (delq name attributes))))
 943     (dolist (name '(=>ucs@gb =>ucs@big5))
 944       (when (and (memq name attributes)
 945                  (setq value (get-char-attribute char name)))
 946         (insert (format "(%-18s . #x%04X)\t; %c%s"
 947                         name value
 948                         (decode-char (intern
 949                                       (concat "="
 950                                               (substring
 951                                                (symbol-name name) 2)))
 952                                      value)
 953                         line-breaking))
 954         (setq attributes (delq name attributes))
 955         ))
 956     ;; (dolist (name '(=>daikanwa))
 957     ;;   (when (and (memq name attributes)
 958     ;;              (setq value (get-char-attribute char name)))
 959     ;;     (insert
 960     ;;      (if (integerp value)
 961     ;;          (format "(%-18s . %05d)\t; %c%s"
 962     ;;                  name value (decode-char '=daikanwa value)
 963     ;;                  line-breaking)
 964     ;;        (format "(%-18s %s)\t; %c%s"
 965     ;;                name
 966     ;;                (mapconcat (function prin1-to-string)
 967     ;;                           value " ")
 968     ;;                (char-representative-of-daikanwa char)
 969     ;;                line-breaking)))
 970     ;;     (setq attributes (delq name attributes))))
 971     (when (and (memq 'general-category attributes)
 972                (setq value (get-char-attribute char 'general-category)))
 973       (insert (format
 974                "(general-category\t%s) ; %s%s"
 975                (mapconcat (lambda (cell)
 976                             (format "%S" cell))
 977                           value " ")
 978                (cond ((rassoc value unidata-normative-category-alist)
 979                       "Normative Category")
 980                      ((rassoc value unidata-informative-category-alist)
 981                       "Informative Category")
 982                      (t
 983                       "Unknown Category"))
 984                line-breaking))
 985       (setq attributes (delq 'general-category attributes))
 986       )
 987     (when (and (memq 'bidi-category attributes)
 988                (setq value (get-char-attribute char 'bidi-category)))
 989       (insert (format "(bidi-category\t. %S)%s"
 990                       value
 991                       line-breaking))
 992       (setq attributes (delq 'bidi-category attributes))
 993       )
 994     (unless (or (not (memq 'mirrored attributes))
 995                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
 996                     'empty))
 997       (insert (format "(mirrored\t\t. %S)%s"
 998                       value
 999                       line-breaking))
1000       (setq attributes (delq 'mirrored attributes))
1001       )
1002     (cond
1003      ((and (memq 'decimal-digit-value attributes)
1004            (setq value (get-char-attribute char 'decimal-digit-value)))
1005       (insert (format "(decimal-digit-value . %S)%s"
1006                       value
1007                       line-breaking))
1008       (setq attributes (delq 'decimal-digit-value attributes))
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       )
1024      (t
1025       (when (and (memq 'digit-value attributes)
1026                  (setq value (get-char-attribute char 'digit-value)))
1027         (insert (format "(digit-value\t. %S)%s"
1028                         value
1029                         line-breaking))
1030         (setq attributes (delq 'digit-value attributes))
1031         )
1032       (when (and (memq 'numeric-value attributes)
1033                  (setq value (get-char-attribute char 'numeric-value)))
1034         (insert (format "(numeric-value\t. %S)%s"
1035                         value
1036                         line-breaking))
1037         (setq attributes (delq 'numeric-value attributes))
1038         )))
1039     (when (and (memq 'iso-10646-comment attributes)
1040                (setq value (get-char-attribute char 'iso-10646-comment)))
1041       (insert (format "(iso-10646-comment\t. %S)%s"
1042                       value
1043                       line-breaking))
1044       (setq attributes (delq 'iso-10646-comment attributes))
1045       )
1046     (when (and (memq 'morohashi-daikanwa attributes)
1047                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1048       (insert (format "(morohashi-daikanwa\t%s)%s"
1049                       (mapconcat (function prin1-to-string) value " ")
1050                       line-breaking))
1051       (setq attributes (delq 'morohashi-daikanwa attributes))
1052       )
1053     (setq radical nil
1054           strokes nil)
1055     (when (and (memq 'ideographic-radical attributes)
1056                (setq value (get-char-attribute char 'ideographic-radical)))
1057       (setq radical value)
1058       (insert (format "(ideographic-radical . %S)\t; %c%s"
1059                       radical
1060                       (ideographic-radical radical)
1061                       line-breaking))
1062       (setq attributes (delq 'ideographic-radical attributes))
1063       )
1064     (when (and (memq 'shuowen-radical attributes)
1065                (setq value (get-char-attribute char 'shuowen-radical)))
1066       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1067                       value
1068                       (shuowen-radical value)
1069                       line-breaking))
1070       (setq attributes (delq 'shuowen-radical attributes))
1071       )
1072     (let (key)
1073       (dolist (domain
1074                (append
1075                 char-db-feature-domains
1076                 (let (dest domain)
1077                   (dolist (feature (char-attribute-list))
1078                     (setq feature (symbol-name feature))
1079                     (when (string-match
1080                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1081                            feature)
1082                       (setq domain (intern (match-string 2 feature)))
1083                      (unless (memq domain dest)
1084                        (setq dest (cons domain dest)))))
1085                   (sort dest #'string<))))
1086         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1087         (when (and (memq key attributes)
1088                    (setq value (get-char-attribute char key)))
1089           (setq radical value)
1090           (insert (format "(%s . %S)\t; %c%s"
1091                           key
1092                           radical
1093                           (ideographic-radical radical)
1094                           line-breaking))
1095           (setq attributes (delq key attributes))
1096           )
1097         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1098         (when (and (memq key attributes)
1099                    (setq value (get-char-attribute char key)))
1100           (setq strokes value)
1101           (insert (format "(%s . %S)%s"
1102                           key
1103                           strokes
1104                           line-breaking))
1105           (setq attributes (delq key attributes))
1106           )
1107         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1108         (when (and (memq key attributes)
1109                    (setq value (get-char-attribute char key)))
1110           (insert (format "(%s       . %S)%s"
1111                           key
1112                           value
1113                           line-breaking))
1114           (setq attributes (delq key attributes))
1115           )
1116         (dolist (feature '(ideographic-radical
1117                            ideographic-strokes
1118                            total-strokes))
1119           (setq key (intern (format "%s@%s*sources" feature domain)))
1120           (when (and (memq key attributes)
1121                      (setq value (get-char-attribute char key)))
1122             (insert (format "(%s%s" key line-breaking))
1123             (dolist (cell value)
1124               (insert (format " %s" cell)))
1125             (insert ")")
1126             (insert line-breaking)
1127             (setq attributes (delq key attributes))
1128             ))
1129         ))
1130     (when (and (memq 'ideographic-strokes attributes)
1131                (setq value (get-char-attribute char 'ideographic-strokes)))
1132       (setq strokes value)
1133       (insert (format "(ideographic-strokes . %S)%s"
1134                       strokes
1135                       line-breaking))
1136       (setq attributes (delq 'ideographic-strokes attributes))
1137       )
1138     (when (and (memq 'kangxi-radical attributes)
1139                (setq value (get-char-attribute char 'kangxi-radical)))
1140       (unless (eq value radical)
1141         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1142                         value
1143                         (ideographic-radical value)
1144                         line-breaking))
1145         (or radical
1146             (setq radical value)))
1147       (setq attributes (delq 'kangxi-radical attributes))
1148       )
1149     (when (and (memq 'kangxi-strokes attributes)
1150                (setq value (get-char-attribute char 'kangxi-strokes)))
1151       (unless (eq value strokes)
1152         (insert (format "(kangxi-strokes\t . %S)%s"
1153                         value
1154                         line-breaking))
1155         (or strokes
1156             (setq strokes value)))
1157       (setq attributes (delq 'kangxi-strokes attributes))
1158       )
1159     ;; (when (and (memq 'japanese-radical attributes)
1160     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1161     ;;   (unless (eq value radical)
1162     ;;     (insert (format "(japanese-radical\t . %S)\t; %c%s"
1163     ;;                     value
1164     ;;                     (ideographic-radical value)
1165     ;;                     line-breaking))
1166     ;;     (or radical
1167     ;;         (setq radical value)))
1168     ;;   (setq attributes (delq 'japanese-radical attributes))
1169     ;;   )
1170     (when (and (memq 'japanese-strokes attributes)
1171                (setq value (get-char-attribute char 'japanese-strokes)))
1172       (unless (eq value strokes)
1173         (insert (format "(japanese-strokes\t . %S)%s"
1174                         value
1175                         line-breaking))
1176         (or strokes
1177             (setq strokes value)))
1178       (setq attributes (delq 'japanese-strokes attributes))
1179       )
1180     (when (and (memq 'cns-radical attributes)
1181                (setq value (get-char-attribute char 'cns-radical)))
1182       (insert (format "(cns-radical\t . %S)\t; %c%s"
1183                       value
1184                       (ideographic-radical value)
1185                       line-breaking))
1186       (setq attributes (delq 'cns-radical attributes))
1187       )
1188     (when (and (memq 'cns-strokes attributes)
1189                (setq value (get-char-attribute char 'cns-strokes)))
1190       (unless (eq value strokes)
1191         (insert (format "(cns-strokes\t . %S)%s"
1192                         value
1193                         line-breaking))
1194         (or strokes
1195             (setq strokes value)))
1196       (setq attributes (delq 'cns-strokes attributes))
1197       )
1198     ;; (when (and (memq 'shinjigen-1-radical attributes)
1199     ;;            (setq value (get-char-attribute char 'shinjigen-1-radical)))
1200     ;;   (unless (eq value radical)
1201     ;;     (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
1202     ;;                     value
1203     ;;                     (ideographic-radical value)
1204     ;;                     line-breaking))
1205     ;;     (or radical
1206     ;;         (setq radical value)))
1207     ;;   (setq attributes (delq 'shinjigen-1-radical attributes))
1208     ;;   )
1209     ;; (when (and (memq 'ideographic- attributes)
1210     ;;            (setq value (get-char-attribute char 'ideographic-)))
1211     ;;   (insert "(ideographic-       ")
1212     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1213     ;;         separator nil)
1214     ;;   (while (consp value)
1215     ;;     (setq cell (car value))
1216     ;;     (if (integerp cell)
1217     ;;         (setq cell (decode-char '=ucs cell)))
1218     ;;     (cond ((characterp cell)
1219     ;;            (if separator
1220     ;;                (insert lbs))
1221     ;;            (if readable
1222     ;;                (insert (format "%S" cell))
1223     ;;              (char-db-insert-char-spec cell readable))
1224     ;;            (setq separator lbs))
1225     ;;           ((consp cell)
1226     ;;            (if separator
1227     ;;                (insert lbs))
1228     ;;            (if (consp (car cell))
1229     ;;                (char-db-insert-char-spec cell readable)
1230     ;;              (char-db-insert-char-reference cell readable))
1231     ;;            (setq separator lbs))
1232     ;;           (t
1233     ;;            (if separator
1234     ;;                (insert separator))
1235     ;;            (insert (prin1-to-string cell))
1236     ;;            (setq separator " ")))
1237     ;;     (setq value (cdr value)))
1238     ;;   (insert ")")
1239     ;;   (insert line-breaking)
1240     ;;   (setq attributes (delq 'ideographic- attributes)))
1241     (when (and (memq 'total-strokes attributes)
1242                (setq value (get-char-attribute char 'total-strokes)))
1243       (insert (format "(total-strokes       . %S)%s"
1244                       value
1245                       line-breaking))
1246       (setq attributes (delq 'total-strokes attributes))
1247       )
1248     ;; (when (and (memq '->ideograph attributes)
1249     ;;            (setq value (get-char-attribute char '->ideograph)))
1250     ;;   (insert (format "(->ideograph\t%s)%s"
1251     ;;                   (mapconcat (lambda (code)
1252     ;;                                (cond ((symbolp code)
1253     ;;                                       (symbol-name code))
1254     ;;                                      ((integerp code)
1255     ;;                                       (format "#x%04X" code))
1256     ;;                                      (t
1257     ;;                                       (format "%s %S"
1258     ;;                                               line-breaking code))))
1259     ;;                              value " ")
1260     ;;                   line-breaking))
1261     ;;   (setq attributes (delq '->ideograph attributes))
1262     ;;   )
1263     ;; (when (and (memq '->decomposition attributes)
1264     ;;            (setq value (get-char-attribute char '->decomposition)))
1265     ;;   (insert (format "(->decomposition\t%s)%s"
1266     ;;                   (mapconcat (lambda (code)
1267     ;;                                (cond ((symbolp code)
1268     ;;                                       (symbol-name code))
1269     ;;                                      ((characterp code)
1270     ;;                                       (if readable
1271     ;;                                           (format "%S" code)
1272     ;;                                         (format "#x%04X"
1273     ;;                                                 (char-int code))
1274     ;;                                         ))
1275     ;;                                      ((integerp code)
1276     ;;                                       (format "#x%04X" code))
1277     ;;                                      (t
1278     ;;                                       (format "%s%S" line-breaking code))))
1279     ;;                              value " ")
1280     ;;                   line-breaking))
1281     ;;   (setq attributes (delq '->decomposition attributes))
1282     ;;   )
1283     (if (equal (get-char-attribute char '->titlecase)
1284                (get-char-attribute char '->uppercase))
1285         (setq attributes (delq '->titlecase attributes)))
1286     (when (and (memq '->mojikyo attributes)
1287                (setq value (get-char-attribute char '->mojikyo)))
1288       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1289                       value (decode-char 'mojikyo value)
1290                       line-breaking))
1291       (setq attributes (delq '->mojikyo attributes))
1292       )
1293     ;; (when (and (memq 'hanyu-dazidian-vol attributes)
1294     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1295     ;;   (insert (format "(hanyu-dazidian-vol  . %d)%s"
1296     ;;                   value line-breaking))
1297     ;;   (setq attributes (delq 'hanyu-dazidian-vol attributes))
1298     ;;   )
1299     ;; (when (and (memq 'hanyu-dazidian-page attributes)
1300     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1301     ;;   (insert (format "(hanyu-dazidian-page . %d)%s"
1302     ;;                   value line-breaking))
1303     ;;   (setq attributes (delq 'hanyu-dazidian-page attributes))
1304     ;;   )
1305     ;; (when (and (memq 'hanyu-dazidian-char attributes)
1306     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1307     ;;   (insert (format "(hanyu-dazidian-char . %d)%s"
1308     ;;                   value line-breaking))
1309     ;;   (setq attributes (delq 'hanyu-dazidian-char attributes))
1310     ;;   )
1311     (unless readable
1312       (dolist (ignored '(composition
1313                          ->denotational <-subsumptive ->ucs-unified
1314                          ->ideographic-component-forms))
1315         (setq attributes (delq ignored attributes))))
1316     (while attributes
1317       (setq name (car attributes))
1318       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1319                   'value-is-empty)
1320         (cond ((setq ret (find-charset name))
1321                (setq name (charset-name ret))
1322                (when (not (memq name dest-ccss))
1323                  (setq dest-ccss (cons name dest-ccss))
1324                  (char-db-insert-ccs-feature name value line-breaking))
1325                )
1326               ((string-match "^=>ucs@" (symbol-name name))
1327                (insert (format "(%-18s . #x%04X)\t; %c%s"
1328                                name value (decode-char '=ucs value)
1329                                line-breaking))
1330                )
1331               ((eq name 'jisx0208-1978/4X)
1332                (insert (format "(%-18s . #x%04X)%s"
1333                                name value
1334                                line-breaking))
1335                )
1336               ((and
1337                 (not readable)
1338                 (not (eq name '->subsumptive))
1339                 (not (eq name '->uppercase))
1340                 (not (eq name '->lowercase))
1341                 (not (eq name '->titlecase))
1342                 (not (eq name '->canonical))
1343                 (not (eq name '->Bopomofo))
1344                 (not (eq name '->mistakable))
1345                 (not (eq name '->ideographic-variants))
1346                 (null (get-char-attribute
1347                        char (intern (format "%s*sources" name))))
1348                 (not (string-match "\\*sources$" (symbol-name name)))
1349                 (null (get-char-attribute
1350                        char (intern (format "%s*note" name))))
1351                 (not (string-match "\\*note$" (symbol-name name)))
1352                 (or (eq name '<-identical)
1353                     (eq name '<-uppercase)
1354                     (eq name '<-lowercase)
1355                     (eq name '<-titlecase)
1356                     (eq name '<-canonical)
1357                     (eq name '<-ideographic-variants)
1358                     ;; (eq name '<-synonyms)
1359                     (string-match "^<-synonyms" (symbol-name name))
1360                     (eq name '<-mistakable)
1361                     (when (string-match "^->" (symbol-name name))
1362                       (cond
1363                        ((string-match "^->fullwidth" (symbol-name name))
1364                         (not (and (consp value)
1365                                   (characterp (car value))
1366                                   (encode-char
1367                                    (car value) '=ucs 'defined-only)))
1368                         )
1369                        (t)))
1370                     ))
1371                )
1372               ((or (eq name 'ideographic-structure)
1373                    (eq name 'ideographic-combination)
1374                    ;; (eq name 'ideographic-)
1375                    (eq name '=decomposition)
1376                    (char-feature-base-name= '=decomposition name)
1377                    (char-feature-base-name= '=>decomposition name)
1378                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1379                    ;;               (symbol-name name))
1380                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1381                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1382                                  (symbol-name name))
1383                    )
1384                (char-db-insert-relation-feature char name value
1385                                                 line-breaking
1386                                                 ccss readable))
1387               ;; ((memq name '(ideograph=
1388               ;;               original-ideograph-of
1389               ;;               ancient-ideograph-of
1390               ;;               vulgar-ideograph-of
1391               ;;               wrong-ideograph-of
1392               ;;               ;; simplified-ideograph-of
1393               ;;               ideographic-variants
1394               ;;               ;; ideographic-different-form-of
1395               ;;               ))
1396               ;;  (insert (format "(%-18s%s " name line-breaking))
1397               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1398               ;;        separator nil)
1399               ;;  (while (consp value)
1400               ;;    (setq cell (car value))
1401               ;;    (if (and (consp cell)
1402               ;;             (consp (car cell)))
1403               ;;        (progn
1404               ;;          (if separator
1405               ;;              (insert lbs))
1406               ;;          (char-db-insert-alist cell readable)
1407               ;;          (setq separator lbs))
1408               ;;      (if separator
1409               ;;          (insert separator))
1410               ;;      (insert (prin1-to-string cell))
1411               ;;      (setq separator " "))
1412               ;;    (setq value (cdr value)))
1413               ;;  (insert ")")
1414               ;;  (insert line-breaking))
1415               ((consp value)
1416                (insert (format "(%-18s " name))
1417                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1418                      separator nil)
1419                (while (consp value)
1420                  (setq cell (car value))
1421                  (if (and (consp cell)
1422                           (consp (car cell))
1423                           (setq ret (condition-case nil
1424                                         (find-char cell)
1425                                       (error nil))))
1426                      (progn
1427                        (setq rest cell
1428                              al nil
1429                              cal nil)
1430                        (while rest
1431                          (setq key (car (car rest)))
1432                          (if (find-charset key)
1433                              (setq cal (cons key cal))
1434                            (setq al (cons key al)))
1435                          (setq rest (cdr rest)))
1436                        (if separator
1437                            (insert lbs))
1438                        (insert-char-attributes ret
1439                                                readable
1440                                                al ; cal
1441                                                nil 'for-sub-node)
1442                        (setq separator lbs))
1443                    (setq ret (prin1-to-string cell))
1444                    (if separator
1445                        (if (< (+ (current-column)
1446                                  (length ret)
1447                                  (length separator))
1448                               76)
1449                            (insert separator)
1450                          (insert lbs)))
1451                    (insert ret)
1452                    (setq separator " "))
1453                  (setq value (cdr value)))
1454                (insert ")")
1455                (insert line-breaking))
1456               (t
1457                (insert (format "(%-18s" name))
1458                (setq ret (prin1-to-string value))
1459                (unless (< (+ (current-column)
1460                              (length ret)
1461                              3)
1462                           76)
1463                  (insert line-breaking))
1464                (insert " . " ret ")" line-breaking)
1465                ;; (insert (format "(%-18s . %S)%s"
1466                ;;                 name value
1467                ;;                 line-breaking))
1468                )
1469               ))
1470       (setq attributes (cdr attributes)))
1471     (insert ")")))
1472
1473 (defun insert-char-data (char &optional readable
1474                               attributes)
1475   (save-restriction
1476     (narrow-to-region (point)(point))
1477     (insert "(define-char
1478   '")
1479     (insert-char-attributes char readable attributes)
1480     (insert ")\n")
1481     (goto-char (point-min))
1482     (while (re-search-forward "[ \t]+$" nil t)
1483       (replace-match ""))
1484     ;; from tabify.
1485     (goto-char (point-min))
1486     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1487       (let ((column (current-column))
1488             (indent-tabs-mode t))
1489         (delete-region (match-beginning 0) (point))
1490         (indent-to column)))
1491     (goto-char (point-max))
1492     ;; (tabify (point-min)(point-max))
1493     ))
1494
1495 (defun insert-char-data-with-variant (char &optional printable
1496                                            no-ucs-unified
1497                                            script excluded-script)
1498   (insert-char-data char printable)
1499   (let ((variants (char-variants char))
1500         rest
1501         variant vs ret)
1502     (setq variants (sort variants #'<))
1503     (setq rest variants)
1504     (setq variants (cons char variants))
1505     (while rest
1506       (setq variant (car rest))
1507       (unless (get-char-attribute variant '<-subsumptive)
1508         (if (and (or (null script)
1509                      (null (setq vs (get-char-attribute variant 'script)))
1510                      (memq script vs))
1511                  (or (null excluded-script)
1512                      (null (setq vs (get-char-attribute variant 'script)))
1513                      (not (memq excluded-script vs))))
1514             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1515               (insert-char-data variant printable)
1516               (if (setq ret (char-variants variant))
1517                   (while ret
1518                     (or (memq (car ret) variants)
1519                         ;; (get-char-attribute (car ret) '<-subsumptive)
1520                         (setq rest (nconc rest (list (car ret)))))
1521                     (setq ret (cdr ret)))))))
1522       (setq rest (cdr rest)))))
1523
1524 (defun insert-char-range-data (min max &optional script excluded-script)
1525   (let ((code min)
1526         char)
1527     (while (<= code max)
1528       (setq char (decode-char '=ucs code))
1529       (if (encode-char char '=ucs 'defined-only)
1530           (insert-char-data-with-variant char nil 'no-ucs-unified
1531                                          script excluded-script))
1532       (setq code (1+ code)))))
1533
1534 (defun write-char-range-data-to-file (min max file
1535                                           &optional script excluded-script)
1536   (let ((coding-system-for-write char-db-file-coding-system))
1537     (with-temp-buffer
1538       (insert (format ";; -*- coding: %s -*-\n"
1539                       char-db-file-coding-system))
1540       (insert-char-range-data min max script excluded-script)
1541       (write-region (point-min)(point-max) file))))
1542
1543 (defvar what-character-original-window-configuration)
1544
1545 ;;;###autoload
1546 (defun what-char-definition (char)
1547   (interactive (list (char-after)))
1548   (let ((buf (get-buffer-create "*Character Description*"))
1549         (the-buf (current-buffer))
1550         (win-conf (current-window-configuration)))
1551     (pop-to-buffer buf)
1552     (make-local-variable 'what-character-original-window-configuration)
1553     (setq what-character-original-window-configuration win-conf)
1554     (setq buffer-read-only nil)
1555     (erase-buffer)
1556     (condition-case err
1557         (progn
1558           (insert-char-data-with-variant char 'printable)
1559           (unless (char-attribute-alist char)
1560             (insert (format ";; = %c\n"
1561                             (let* ((rest (split-char char))
1562                                    (ccs (pop rest))
1563                                    (code (pop rest)))
1564                               (while rest
1565                                 (setq code (logior (lsh code 8)
1566                                                    (pop rest))))
1567                               (decode-char ccs code)))))
1568           ;; (char-db-update-comment)
1569           (set-buffer-modified-p nil)
1570           (view-mode the-buf (lambda (buf)
1571                                (set-window-configuration
1572                                 what-character-original-window-configuration)
1573                                ))
1574           (goto-char (point-min)))
1575       (error (progn
1576                (set-window-configuration
1577                 what-character-original-window-configuration)
1578                (signal (car err) (cdr err)))))))
1579
1580
1581 ;;; @ end
1582 ;;;
1583
1584 (provide 'char-db-util)
1585
1586 ;;; char-db-util.el ends here