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     =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     ==hanyo-denshi/ja
 184     ==hanyo-denshi/jb
 185     ==hanyo-denshi/jc
 186     ==hanyo-denshi/ft
 187     ==hanyo-denshi/ib
 188     ==hanyo-denshi/hg
 189     ==hanyo-denshi/jt
 190     ==hanyo-denshi/ks
 191     =gt
 192     =gt-k
 193     =daikanwa
 194     =daikanwa@rev2
 195     =daikanwa@rev1
 196     =daikanwa/+p
 197     ==daikanwa
 198     =cbeta
 199     ideograph-hanziku-1
 200     ideograph-hanziku-2
 201     ideograph-hanziku-3
 202     ideograph-hanziku-4
 203     ideograph-hanziku-5
 204     ideograph-hanziku-6
 205     ideograph-hanziku-7
 206     ideograph-hanziku-8
 207     ideograph-hanziku-9
 208     ideograph-hanziku-10
 209     ideograph-hanziku-11
 210     ideograph-hanziku-12
 211     ;; =>>>adobe-japan1-0
 212     ;; =>>>adobe-japan1-1
 213     ;; =>>>adobe-japan1-2
 214     ;; =>>>adobe-japan1-3
 215     ;; =>>>adobe-japan1-4
 216     ;; =>>>adobe-japan1-5
 217     ;; =>>>adobe-japan1-6
 218     ;; =>>>jis-x0208
 219     ;; =>>>jis-x0213-1
 220     ;; =>>>jis-x0213-2
 221     ;; =>>>hanyo-denshi/ja
 222     ;; =>>>hanyo-denshi/jb
 223     ;; =>>>hanyo-denshi/jc
 224     ;; =>>>hanyo-denshi/ft
 225     ;; =>>>hanyo-denshi/ib
 226     ;; =>>>hanyo-denshi/hg
 227     ;; =>>>hanyo-denshi/jt
 228     ;; =>>>hanyo-denshi/ks
 229     ;; =>>>gt
 230     =>>adobe-japan1-0
 231     =>>adobe-japan1-1
 232     =>>adobe-japan1-2
 233     =>>adobe-japan1-3
 234     =>>adobe-japan1-4
 235     =>>adobe-japan1-5
 236     =>>adobe-japan1-6
 237     =>>jis-x0208
 238     =>>jis-x0213-1
 239     =>>jis-x0213-1@2000
 240     =>>jis-x0213-1@2004
 241     =>>jis-x0213-2
 242     =>>jis-x0208@1978
 243     =>>hanyo-denshi/ft
 244     =>>hanyo-denshi/jt
 245     =>>hanyo-denshi/ks
 246     =>>gt
 247     =>>daikanwa
 248     =>>cbeta
 249     =+>jis-x0208
 250     =+>jis-x0213-1
 251     =+>jis-x0213-2
 252     =+>adobe-japan1-0
 253     =+>adobe-japan1-1
 254     =+>adobe-japan1-2
 255     =+>adobe-japan1-3
 256     =+>adobe-japan1-4
 257     =+>adobe-japan1-5
 258     =+>adobe-japan1-6
 259     =+>jis-x0208@1978
 260     =>jis-x0208
 261     =>jis-x0208@1997
 262     =>jis-x0213-1
 263     =>jis-x0213-1@2000
 264     =>jis-x0213-1@2004
 265     =>jis-x0213-2
 266     ==>ucs@bucs
 267     =>iwds-1
 268     =>ucs@hanyo-denshi
 269     =>ucs@iso
 270     =>ucs@unicode
 271     =>ucs@jis
 272     =>ucs@cns
 273     =>ucs@ks
 274     =+>ucs@iso
 275     =+>ucs@unicode
 276     =+>ucs@jis
 277     =+>ucs@jis/1990
 278     =+>ucs@cns
 279     =+>ucs@ks
 280     =>>ucs@iso
 281     =>>ucs@unicode
 282     =>>ucs@jis
 283     =>>ucs@cns
 284     =>>>ucs@iso
 285     =>>>ucs@unicode
 286     ==ucs@iso
 287     ==ucs@unicode
 288     ;; ==ucs@cns
 289     ==gb2312
 290     ==ks-x1001
 291     ==cns11643-1
 292     ==cns11643-2
 293     ==cns11643-3
 294     ==cns11643-4
 295     ==cns11643-5
 296     ==cns11643-6
 297     ==cns11643-7
 298     ==gt
 299     ==jis-x0208@1990
 300     ;; ==jis-x0208@1983
 301     ==jis-x0208@1978
 302     ==gt-k
 303     =ucs@iso
 304     =ucs@unicode
 305     =ucs@cns
 306     ==big5-cdp
 307     ==cbeta
 308     =>>big5-cdp
 309     =>>gt-k
 310     =+>gt
 311     =>gt
 312     =>big5-cdp
 313     =>daikanwa
 314     =>daikanwa/ho
 315     =>cns11643-5
 316     =>cns11643-7
 317     =big5
 318     =big5-eten
 319     =>gt-k
 320     =zinbun-oracle
 321     =>zinbun-oracle
 322     =ruimoku-v6
 323     =>>ruimoku-v6
 324     ==ruimoku-v6
 325     =jef-china3
 326     =>cbeta
 327     =shinjigen
 328     =ucs-var-001
 329     =ucs-var-002
 330     =ucs-var-003
 331     =ucs-itaiji-001
 332     =ucs-itaiji-002
 333     =ucs-itaiji-003
 334     =ucs-itaiji-004
 335     =ucs-itaiji-005
 336     =ucs-itaiji-008
 337     =big5-cdp-var-3
 338     =big5-cdp-var-5
 339     =>ucs@iwds-1))
 340
 341
 342 ;;; @ char-db formatters
 343 ;;;
 344
 345 (defun char-db-make-char-spec (char)
 346   (let (ret char-spec)
 347     (cond ((characterp char)
 348            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 349                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 350                   (setq char-spec (list (cons '=ucs ret)))
 351                   (cond ((setq ret (get-char-attribute char 'name))
 352                          (setq char-spec (cons (cons 'name ret) char-spec))
 353                          )
 354                         ((setq ret (get-char-attribute char 'name*))
 355                          (setq char-spec (cons (cons 'name* ret) char-spec))
 356                          ))
 357                   )
 358                  ((encode-char char '=mj 'defined-only)
 359                   (setq char-spec nil)
 360                   (dolist (ccs (charset-list))
 361                     (if (and (or (eq ccs '=mj)
 362                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 363                                  (string-match "=ucs@" (symbol-name ccs))
 364                                  )
 365                              (setq ccs (charset-name ccs))
 366                              (null (assq ccs char-spec))
 367                              (setq ret (encode-char char ccs 'defined-only)))
 368                         (setq char-spec (cons (cons ccs ret) char-spec))))
 369                   )
 370                  ((encode-char char '==mj 'defined-only)
 371                   (setq char-spec nil)
 372                   (dolist (ccs (charset-list))
 373                     (if (and (or (eq ccs '==mj)
 374                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 375                                  (string-match "=ucs@" (symbol-name ccs))
 376                                  )
 377                              (setq ccs (charset-name ccs))
 378                              (null (assq ccs char-spec))
 379                              (setq ret (encode-char char ccs 'defined-only)))
 380                         (setq char-spec (cons (cons ccs ret) char-spec))))
 381                   )
 382                  ((encode-char char '=adobe-japan1 'defined-only)
 383                   (setq char-spec nil)
 384                   (dolist (ccs (charset-list))
 385                     (if (and (or (memq ccs
 386                                        '(=adobe-japan1-0
 387                                          =adobe-japan1-1
 388                                          =adobe-japan1-2
 389                                          =adobe-japan1-3
 390                                          =adobe-japan1-4
 391                                          =adobe-japan1-5
 392                                          =adobe-japan1-6
 393                                          ))
 394                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 395                                  (string-match "=ucs@" (symbol-name ccs))
 396                                  )
 397                              (setq ccs (charset-name ccs))
 398                              (null (assq ccs char-spec))
 399                              (setq ret (encode-char char ccs 'defined-only)))
 400                         (setq char-spec (cons (cons ccs ret) char-spec))))
 401                   )
 402                  ((encode-char char '==adobe-japan1 'defined-only)
 403                   (setq char-spec nil)
 404                   (dolist (ccs (charset-list))
 405                     (if (and (or (memq ccs
 406                                        '(==adobe-japan1-0
 407                                          ==adobe-japan1-1
 408                                          ==adobe-japan1-2
 409                                          ==adobe-japan1-3
 410                                          ==adobe-japan1-4
 411                                          ==adobe-japan1-5
 412                                          ==adobe-japan1-6
 413                                          ))
 414                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 415                                  (string-match "=ucs@" (symbol-name ccs))
 416                                  )
 417                              (setq ccs (charset-name ccs))
 418                              (null (assq ccs char-spec))
 419                              (setq ret (encode-char char ccs 'defined-only)))
 420                         (setq char-spec (cons (cons ccs ret) char-spec))))
 421                   )
 422                  ((setq ret
 423                         (catch 'tag
 424                           (let ((rest char-db-coded-charset-priority-list)
 425                                 ccs)
 426                             (while rest
 427                               (setq ccs (charset-name
 428                                          (find-charset (car rest))))
 429                               (if (setq ret
 430                                         (encode-char char ccs
 431                                                      'defined-only))
 432                                   (throw 'tag (cons ccs ret)))
 433                               (setq rest (cdr rest))))))
 434                   (setq char-spec (list ret))
 435                   (dolist (ccs (delq (car ret) (charset-list)))
 436                     (if (and (or (charset-iso-final-char ccs)
 437                                  (memq ccs
 438                                        '(=daikanwa
 439                                          =daikanwa@rev2
 440                                          ;; =gt-k
 441                                          =jis-x0208@1997
 442                                          ))
 443                                  (eq (charset-property ccs 'iso-ir) 177)
 444                                  ;; (string-match "=ucs@" (symbol-name ccs))
 445                                  )
 446                              (setq ccs (charset-name ccs))
 447                              (null (assq ccs char-spec))
 448                              (setq ret (encode-char char ccs 'defined-only)))
 449                         (setq char-spec (cons (cons ccs ret) char-spec))))
 450                   (if (null char-spec)
 451                       (setq char-spec (split-char char)))
 452                   (cond ((setq ret (get-char-attribute char 'name))
 453                          (setq char-spec (cons (cons 'name ret) char-spec))
 454                          )
 455                         ((setq ret (get-char-attribute char 'name*))
 456                          (setq char-spec (cons (cons 'name* ret) char-spec))
 457                          ))
 458                   )
 459                  ((setq ret (get-char-attribute
 460                              char 'ideographic-combination))
 461                   (setq char-spec
 462                         (cons (cons 'ideographic-combination ret)
 463                               char-spec))
 464                   ))
 465            char-spec)
 466           ((consp char)
 467            char))))
 468
 469 (defun char-db-insert-char-spec (char &optional readable column
 470                                       required-features)
 471   (unless column
 472     (setq column (current-column)))
 473   (let (char-spec temp-char)
 474     (setq char-spec (char-db-make-char-spec char))
 475     (unless (or (characterp char) ; char
 476                 (condition-case nil
 477                     (setq char (find-char char-spec))
 478                   (error nil)))
 479       ;; define temporary character
 480       ;;   Current implementation is dirty.
 481       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 482                                          char-spec)))
 483       (remove-char-attribute temp-char 'ideograph-daikanwa)
 484       (setq char temp-char))
 485     (insert-char-attributes char
 486                             readable
 487                             (union (mapcar #'car char-spec)
 488                                    required-features)
 489                             nil 'for-sub-node)
 490     (when temp-char
 491       ;; undefine temporary character
 492       ;;   Current implementation is dirty.
 493       (setq char-spec (char-attribute-alist temp-char))
 494       (while char-spec
 495         (remove-char-attribute temp-char (car (car char-spec)))
 496         (setq char-spec (cdr char-spec))))))
 497
 498 (defun char-db-insert-alist (alist &optional readable column)
 499   (unless column
 500     (setq column (current-column)))
 501   (let ((line-breaking
 502          (concat "\n" (make-string (1+ column) ?\ )))
 503         name value
 504         ret al ; cal
 505         key
 506         lbs cell rest separator)
 507     (insert "(")
 508     (while alist
 509       (setq name (car (car alist))
 510             value (cdr (car alist)))
 511       (cond ((eq name 'char)
 512              (insert "(char . ")
 513              (if (setq ret (condition-case nil
 514                                (find-char value)
 515                              (error nil)))
 516                  (progn
 517                    (setq al nil
 518                          ;; cal nil
 519                          )
 520                    (while value
 521                      (setq key (car (car value)))
 522                      ;; (if (find-charset key)
 523                      ;;     (setq cal (cons key cal))
 524                      (setq al (cons key al))
 525                      ;; )
 526                      (setq value (cdr value)))
 527                    (insert-char-attributes ret
 528                                            readable
 529                                            (or al 'none) ; cal
 530                                            nil 'for-sub-node))
 531                (insert (prin1-to-string value)))
 532              (insert ")")
 533              (insert line-breaking))
 534             ((consp value)
 535              (insert (format "(%-18s " name))
 536              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 537              (while (consp value)
 538                (setq cell (car value))
 539                (if (and (consp cell)
 540                         (consp (car cell))
 541                         (setq ret (condition-case nil
 542                                       (find-char cell)
 543                                     (error nil)))
 544                         )
 545                    (progn
 546                      (setq rest cell
 547                            al nil
 548                            ;; cal nil
 549                            )
 550                      (while rest
 551                        (setq key (car (car rest)))
 552                        ;; (if (find-charset key)
 553                        ;;     (setq cal (cons key cal))
 554                        (setq al (cons key al))
 555                        ;; )
 556                        (setq rest (cdr rest)))
 557                      (if separator
 558                          (insert lbs))
 559                      (insert-char-attributes ret
 560                                              readable
 561                                              al ; cal
 562                                              nil 'for-sub-node)
 563                      (setq separator lbs))
 564                  (if separator
 565                      (insert separator))
 566                  (insert (prin1-to-string cell))
 567                  (setq separator " "))
 568                (setq value (cdr value)))
 569              (insert ")")
 570              (insert line-breaking))
 571             (t
 572              (insert (format "(%-18s . %S)%s"
 573                              name value
 574                              line-breaking))))
 575       (setq alist (cdr alist))))
 576   (insert ")"))
 577
 578 (defun char-db-insert-char-reference (plist &optional readable column)
 579   (unless column
 580     (setq column (current-column)))
 581   (let ((line-breaking
 582          (concat "\n" (make-string (1+ column) ?\ )))
 583         (separator "")
 584         name value)
 585     (insert "(")
 586     (while plist
 587       (setq name (pop plist))
 588       (setq value (pop plist))
 589       (cond ((eq name :char)
 590              (insert separator)
 591              (insert ":char\t")
 592              (cond ((numberp value)
 593                     (setq value (decode-char '=ucs value)))
 594                    ;; ((consp value)
 595                    ;;  (setq value (or (find-char value)
 596                    ;;                  value)))
 597                    )
 598              (char-db-insert-char-spec value readable)
 599              (insert line-breaking)
 600              (setq separator ""))
 601             ((eq name :radical)
 602              (insert (format "%s%s\t%d ; %c%s"
 603                              separator
 604                              name value
 605                              (ideographic-radical value)
 606                              line-breaking))
 607              (setq separator ""))
 608             (t
 609              (insert (format "%s%s\t%S" separator name value))
 610              (setq separator line-breaking)))
 611       ))
 612   (insert ")"))
 613
 614 (defun char-db-decode-isolated-char (ccs code-point)
 615   (let (ret)
 616     (setq ret
 617           (cond ((eq ccs 'arabic-iso8859-6)
 618                  (decode-char ccs code-point))
 619                 ;; ((eq ccs '=gt)
 620                 ;;  (decode-builtin-char '==gt code-point))
 621                 ((and (memq ccs '(=gt-pj-1
 622                                   =gt-pj-2
 623                                   =gt-pj-3
 624                                   =gt-pj-4
 625                                   =gt-pj-5
 626                                   =gt-pj-6
 627                                   =gt-pj-7
 628                                   =gt-pj-8
 629                                   =gt-pj-9
 630                                   =gt-pj-10
 631                                   =gt-pj-11))
 632                       (setq ret (decode-char ccs code-point))
 633                       (setq ret (encode-char ret '=gt 'defined-only)))
 634                  (decode-builtin-char '=gt ret))
 635                 (t
 636                  (decode-builtin-char ccs code-point))))
 637     (cond ((null ret)
 638            (or (decode-char ccs code-point)
 639                (define-char (list (cons ccs code-point)))))
 640           ((and (<= 0 (char-int ret))
 641                 (<= (char-int ret) #x1F))
 642            (decode-char '=ucs (+ #x2400 (char-int ret))))
 643           ((= (char-int ret) #x7F)
 644            ?\u2421)
 645           (t ret))))
 646
 647 (defvar char-db-convert-obsolete-format t)
 648
 649 (defun char-db-insert-ccs-feature (name value line-breaking)
 650   (cond
 651    ((integerp value)
 652     (insert
 653      (format
 654       (cond
 655        ((memq name '(=>iwds-1
 656                      ==shinjigen
 657                      =shinjigen
 658                      =shinjigen@1ed ==shinjigen@1ed
 659                      =shinjigen@rev ==shinjigen@rev
 660                      =shinjigen/+p@rev ==shinjigen/+p@rev
 661                      ===daikanwa/ho ==daikanwa/ho
 662                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 663         "(%-18s .  %04d)\t; %c")
 664        ((eq name '=shinjigen@1ed/24pr)
 665         "(%-18s . %04d)\t; %c")
 666        ((or
 667          (memq name
 668                '(===daikanwa
 669                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 670                  =daikanwa@rev1 =daikanwa@rev2
 671                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 672                  =>>daikanwa/+p
 673                  =daikanwa/+2p =>>daikanwa/+2p
 674                  =gt ==gt ===gt
 675                  =>>gt =+>gt =>gt
 676                  =gt-k ==gt-k ===gt-k
 677                  =>>gt-k =>gt-k
 678                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 679                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 680                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 681                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 682                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 683                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 684                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 685                  =>>adobe-japan1-0 =+>adobe-japan1-0
 686                  =>>adobe-japan1-1 =+>adobe-japan1-1
 687                  =>>adobe-japan1-2 =+>adobe-japan1-2
 688                  =>>adobe-japan1-3 =+>adobe-japan1-3
 689                  =>>adobe-japan1-4 =+>adobe-japan1-4
 690                  =>>adobe-japan1-5 =+>adobe-japan1-5
 691                  =>>adobe-japan1-6 =+>adobe-japan1-6
 692                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 693                  =zinbun-oracle =>zinbun-oracle
 694                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 695                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 696                  ===hng-khi ===hng-khm ===hng-hok ===hng-kyd ===hng-sok
 697                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 698                  ===hng-sys ===hng-tsu ===hng-tzj
 699                  ===hng-hos ===hng-nak ===hng-jhk
 700                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 701                  ===hng-nkm ===hng-k24 ===hng-nkk
 702                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 703                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 704                  ===hng-okd ===hng-wad ===hng-kmi
 705                  ===hng-zkd ===hng-doh ===hng-jyu
 706                  ===hng-tzs ===hng-kss ===hng-kyo
 707                  ===hng-smk))
 708          ;; (string-match "^=adobe-" (symbol-name name))
 709          )
 710         "(%-18s . %05d)\t; %c")
 711        ((memq name '(=hanyo-denshi/ks
 712                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 713                      =>>hanyo-denshi/ks
 714                      =koseki ==koseki
 715                      =mj ==mj ===mj =>>mj
 716                      =zihai mojikyo))
 717         "(%-18s . %06d)\t; %c")
 718        ((memq name '(=hanyo-denshi/tk))
 719         "(%-18s . %08d)\t; %c")
 720        ((>= (charset-dimension name) 2)
 721         "(%-18s . #x%04X)\t; %c")
 722        (t
 723         "(%-18s . #x%02X)\t; %c"))
 724       name
 725       (if (= (charset-iso-graphic-plane name) 1)
 726           (logior value
 727                   (cond ((= (charset-dimension name) 1)
 728                          #x80)
 729                         ((= (charset-dimension name) 2)
 730                          #x8080)
 731                         ((= (charset-dimension name) 3)
 732                          #x808080)
 733                         (t 0)))
 734         value)
 735       (char-db-decode-isolated-char name value)))
 736     (if (and (= (charset-chars name) 94)
 737              (= (charset-dimension name) 2))
 738         (insert (format " [%02d-%02d]"
 739                         (- (lsh value -8) 32)
 740                         (- (logand value 255) 32))))
 741     )
 742    (t
 743     (insert (format "(%-18s . %s)" name value))
 744     ))
 745   (insert line-breaking))
 746
 747 (defun char-db-insert-relation-feature (char name value line-breaking
 748                                              ccss readable)
 749   (insert (format "(%-18s%s " name line-breaking))
 750   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 751         separator cell sources required-features
 752         ret)
 753     (while (consp value)
 754       (setq cell (car value))
 755       (if (integerp cell)
 756           (setq cell (decode-char '=ucs cell)))
 757       (cond
 758        ((eq name '->subsumptive)
 759         (when (or (not (some (lambda (atr)
 760                                (get-char-attribute cell atr))
 761                              char-db-ignored-attributes))
 762                   (some (lambda (ccs)
 763                           (encode-char cell ccs 'defined-only))
 764                         ccss))
 765           (if separator
 766               (insert lbs))
 767           (let ((char-db-ignored-attributes
 768                  (cons '<-subsumptive
 769                        char-db-ignored-attributes)))
 770             (insert-char-attributes cell readable nil nil 'for-sub-node))
 771           (setq separator lbs))
 772         )
 773        ((characterp cell)
 774         (setq sources
 775               (get-char-attribute
 776                char (intern (format "%s*sources" name))))
 777         (setq required-features nil)
 778         (dolist (source sources)
 779           (cond
 780            ((memq source '(JP
 781                            JP/Jouyou
 782                            shinjigen shinjigen@1ed shinjigen@rev))
 783             (setq required-features
 784                   (union required-features
 785                          '(=jis-x0208
 786                            =jis-x0208@1990
 787                            =jis-x0213-1@2000
 788                            =jis-x0213-1@2004
 789                            =jis-x0213-2
 790                            =jis-x0212
 791                            =jis-x0208@1983
 792                            =jis-x0208@1978
 793                            =shinjigen
 794                            =shinjigen@1ed
 795                            =shinjigen@rev
 796                            =shinjigen/+p@rev))))
 797            ((eq source 'CN)
 798             (setq required-features
 799                   (union required-features
 800                          '(=gb2312
 801                            =gb12345
 802                            =iso-ir165)))))
 803           (cond
 804            ((find-charset (setq ret (intern (format "=%s" source))))
 805             (setq required-features
 806                   (cons ret required-features)))
 807            (t (setq required-features
 808                     (cons source required-features)))))
 809         (cond ((string-match "@JP" (symbol-name name))
 810                (setq required-features
 811                      (union required-features
 812                             '(=jis-x0208
 813                               =jis-x0208@1990
 814                               =jis-x0213-1-2000
 815                               =jis-x0213-2-2000
 816                               =jis-x0212
 817                               =jis-x0208@1983
 818                               =jis-x0208@1978))))
 819               ((string-match "@CN" (symbol-name name))
 820                (setq required-features
 821                      (union required-features
 822                             '(=gb2312
 823                               =gb12345
 824                               =iso-ir165)))))
 825         (if separator
 826             (insert lbs))
 827         (if readable
 828             (insert (format "%S" cell))
 829           (char-db-insert-char-spec cell readable
 830                                     nil
 831                                     required-features))
 832         (setq separator lbs))
 833        ((consp cell)
 834         (if separator
 835             (insert lbs))
 836         (if (consp (car cell))
 837             (char-db-insert-char-spec cell readable)
 838           (char-db-insert-char-reference cell readable))
 839         (setq separator lbs))
 840        (t
 841         (if separator
 842             (insert separator))
 843         (insert (prin1-to-string cell))
 844         (setq separator " ")))
 845       (setq value (cdr value)))
 846     (insert ")")
 847     (insert line-breaking)))
 848
 849 (defun insert-char-attributes (char &optional readable attributes column
 850                                     for-sub-node)
 851   (unless column
 852     (setq column (current-column)))
 853   (let (name value ; has-long-ccs-name
 854         rest
 855         radical strokes
 856         (line-breaking
 857          (concat "\n" (make-string (1+ column) ?\ )))
 858         lbs cell separator ret
 859         key al cal
 860         dest-ccss ; sources required-features
 861         ccss)
 862     (let (atr-d)
 863       (setq attributes
 864             (sort (if attributes
 865                       (if (consp attributes)
 866                           (progn
 867                             (dolist (name attributes)
 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                     (dolist (name (char-attribute-list))
 874                       (unless (memq name char-db-ignored-attributes)
 875                         (if (find-charset name)
 876                             (push name ccss))
 877                         (push name atr-d)))
 878                     atr-d)
 879                   #'char-attribute-name<)))
 880     (insert "(")
 881     (when (memq '<-subsumptive attributes)
 882       (when (or readable (not for-sub-node))
 883         (when (setq value (get-char-attribute char '<-subsumptive))
 884           (char-db-insert-relation-feature char '<-subsumptive value
 885                                            line-breaking
 886                                            ccss readable)))
 887       (setq attributes (delq '<-subsumptive attributes)))
 888     (when (and (memq '<-denotational attributes)
 889                (setq value (get-char-attribute char '<-denotational)))
 890       (char-db-insert-relation-feature char '<-denotational value
 891                                        line-breaking
 892                                        ccss readable)
 893       (setq attributes (delq '<-denotational attributes)))
 894     (when (and (memq '<-denotational@component attributes)
 895                (setq value (get-char-attribute char '<-denotational@component)))
 896       (char-db-insert-relation-feature char '<-denotational@component value
 897                                        line-breaking
 898                                        ccss readable)
 899       (setq attributes (delq '<-denotational@component attributes)))
 900     (when (and (memq 'name attributes)
 901                (setq value (get-char-attribute char 'name)))
 902       (insert (format
 903                (if (> (+ (current-column) (length value)) 48)
 904                    "(name . %S)%s"
 905                  "(name               . %S)%s")
 906                value line-breaking))
 907       (setq attributes (delq 'name attributes))
 908       )
 909     (when (and (memq 'name* attributes)
 910                (setq value (get-char-attribute char 'name*)))
 911       (insert (format
 912                (if (> (+ (current-column) (length value)) 48)
 913                    "(name* . %S)%s"
 914                  "(name*              . %S)%s")
 915                value line-breaking))
 916       (setq attributes (delq 'name* attributes))
 917       )
 918     (when (and (memq 'script attributes)
 919                (setq value (get-char-attribute char 'script)))
 920       (insert (format "(script\t\t%s)%s"
 921                       (mapconcat (function prin1-to-string)
 922                                  value " ")
 923                       line-breaking))
 924       (setq attributes (delq 'script attributes))
 925       )
 926     (dolist (name '(=>ucs =>ucs*))
 927       (when (and (memq name attributes)
 928                  (setq value (get-char-attribute char name)))
 929         (insert (format "(%-18s . #x%04X)\t; %c%s"
 930                         name value (decode-char '=ucs value)
 931                         line-breaking))
 932         (setq attributes (delq name attributes))))
 933     (dolist (name '(=>ucs@gb =>ucs@big5))
 934       (when (and (memq name attributes)
 935                  (setq value (get-char-attribute char name)))
 936         (insert (format "(%-18s . #x%04X)\t; %c%s"
 937                         name value
 938                         (decode-char (intern
 939                                       (concat "="
 940                                               (substring
 941                                                (symbol-name name) 2)))
 942                                      value)
 943                         line-breaking))
 944         (setq attributes (delq name attributes))
 945         ))
 946     ;; (dolist (name '(=>daikanwa))
 947     ;;   (when (and (memq name attributes)
 948     ;;              (setq value (get-char-attribute char name)))
 949     ;;     (insert
 950     ;;      (if (integerp value)
 951     ;;          (format "(%-18s . %05d)\t; %c%s"
 952     ;;                  name value (decode-char '=daikanwa value)
 953     ;;                  line-breaking)
 954     ;;        (format "(%-18s %s)\t; %c%s"
 955     ;;                name
 956     ;;                (mapconcat (function prin1-to-string)
 957     ;;                           value " ")
 958     ;;                (char-representative-of-daikanwa char)
 959     ;;                line-breaking)))
 960     ;;     (setq attributes (delq name attributes))))
 961     (when (and (memq 'general-category attributes)
 962                (setq value (get-char-attribute char 'general-category)))
 963       (insert (format
 964                "(general-category\t%s) ; %s%s"
 965                (mapconcat (lambda (cell)
 966                             (format "%S" cell))
 967                           value " ")
 968                (cond ((rassoc value unidata-normative-category-alist)
 969                       "Normative Category")
 970                      ((rassoc value unidata-informative-category-alist)
 971                       "Informative Category")
 972                      (t
 973                       "Unknown Category"))
 974                line-breaking))
 975       (setq attributes (delq 'general-category attributes))
 976       )
 977     (when (and (memq 'bidi-category attributes)
 978                (setq value (get-char-attribute char 'bidi-category)))
 979       (insert (format "(bidi-category\t. %S)%s"
 980                       value
 981                       line-breaking))
 982       (setq attributes (delq 'bidi-category attributes))
 983       )
 984     (unless (or (not (memq 'mirrored attributes))
 985                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
 986                     'empty))
 987       (insert (format "(mirrored\t\t. %S)%s"
 988                       value
 989                       line-breaking))
 990       (setq attributes (delq 'mirrored attributes))
 991       )
 992     (cond
 993      ((and (memq 'decimal-digit-value attributes)
 994            (setq value (get-char-attribute char 'decimal-digit-value)))
 995       (insert (format "(decimal-digit-value . %S)%s"
 996                       value
 997                       line-breaking))
 998       (setq attributes (delq 'decimal-digit-value attributes))
 999       (when (and (memq 'digit-value attributes)
1000                  (setq value (get-char-attribute char 'digit-value)))
1001         (insert (format "(digit-value\t . %S)%s"
1002                         value
1003                         line-breaking))
1004         (setq attributes (delq 'digit-value attributes))
1005         )
1006       (when (and (memq 'numeric-value attributes)
1007                  (setq value (get-char-attribute char 'numeric-value)))
1008         (insert (format "(numeric-value\t . %S)%s"
1009                         value
1010                         line-breaking))
1011         (setq attributes (delq 'numeric-value attributes))
1012         )
1013       )
1014      (t
1015       (when (and (memq 'digit-value attributes)
1016                  (setq value (get-char-attribute char 'digit-value)))
1017         (insert (format "(digit-value\t. %S)%s"
1018                         value
1019                         line-breaking))
1020         (setq attributes (delq 'digit-value attributes))
1021         )
1022       (when (and (memq 'numeric-value attributes)
1023                  (setq value (get-char-attribute char 'numeric-value)))
1024         (insert (format "(numeric-value\t. %S)%s"
1025                         value
1026                         line-breaking))
1027         (setq attributes (delq 'numeric-value attributes))
1028         )))
1029     (when (and (memq 'iso-10646-comment attributes)
1030                (setq value (get-char-attribute char 'iso-10646-comment)))
1031       (insert (format "(iso-10646-comment\t. %S)%s"
1032                       value
1033                       line-breaking))
1034       (setq attributes (delq 'iso-10646-comment attributes))
1035       )
1036     (when (and (memq 'morohashi-daikanwa attributes)
1037                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1038       (insert (format "(morohashi-daikanwa\t%s)%s"
1039                       (mapconcat (function prin1-to-string) value " ")
1040                       line-breaking))
1041       (setq attributes (delq 'morohashi-daikanwa attributes))
1042       )
1043     (setq radical nil
1044           strokes nil)
1045     (when (and (memq 'ideographic-radical attributes)
1046                (setq value (get-char-attribute char 'ideographic-radical)))
1047       (setq radical value)
1048       (insert (format "(ideographic-radical . %S)\t; %c%s"
1049                       radical
1050                       (ideographic-radical radical)
1051                       line-breaking))
1052       (setq attributes (delq 'ideographic-radical attributes))
1053       )
1054     (when (and (memq 'shuowen-radical attributes)
1055                (setq value (get-char-attribute char 'shuowen-radical)))
1056       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1057                       value
1058                       (shuowen-radical value)
1059                       line-breaking))
1060       (setq attributes (delq 'shuowen-radical attributes))
1061       )
1062     (let (key)
1063       (dolist (domain
1064                (append
1065                 char-db-feature-domains
1066                 (let (dest domain)
1067                   (dolist (feature (char-attribute-list))
1068                     (setq feature (symbol-name feature))
1069                     (when (string-match
1070                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1071                            feature)
1072                       (setq domain (intern (match-string 2 feature)))
1073                      (unless (memq domain dest)
1074                        (setq dest (cons domain dest)))))
1075                   (sort dest #'string<))))
1076         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1077         (when (and (memq key attributes)
1078                    (setq value (get-char-attribute char key)))
1079           (setq radical value)
1080           (insert (format "(%s . %S)\t; %c%s"
1081                           key
1082                           radical
1083                           (ideographic-radical radical)
1084                           line-breaking))
1085           (setq attributes (delq key attributes))
1086           )
1087         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1088         (when (and (memq key attributes)
1089                    (setq value (get-char-attribute char key)))
1090           (setq strokes value)
1091           (insert (format "(%s . %S)%s"
1092                           key
1093                           strokes
1094                           line-breaking))
1095           (setq attributes (delq key attributes))
1096           )
1097         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1098         (when (and (memq key attributes)
1099                    (setq value (get-char-attribute char key)))
1100           (insert (format "(%s       . %S)%s"
1101                           key
1102                           value
1103                           line-breaking))
1104           (setq attributes (delq key attributes))
1105           )
1106         (dolist (feature '(ideographic-radical
1107                            ideographic-strokes
1108                            total-strokes))
1109           (setq key (intern (format "%s@%s*sources" feature domain)))
1110           (when (and (memq key attributes)
1111                      (setq value (get-char-attribute char key)))
1112             (insert (format "(%s%s" key line-breaking))
1113             (dolist (cell value)
1114               (insert (format " %s" cell)))
1115             (insert ")")
1116             (insert line-breaking)
1117             (setq attributes (delq key attributes))
1118             ))
1119         ))
1120     (when (and (memq 'ideographic-strokes attributes)
1121                (setq value (get-char-attribute char 'ideographic-strokes)))
1122       (setq strokes value)
1123       (insert (format "(ideographic-strokes . %S)%s"
1124                       strokes
1125                       line-breaking))
1126       (setq attributes (delq 'ideographic-strokes attributes))
1127       )
1128     (when (and (memq 'kangxi-radical attributes)
1129                (setq value (get-char-attribute char 'kangxi-radical)))
1130       (unless (eq value radical)
1131         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1132                         value
1133                         (ideographic-radical value)
1134                         line-breaking))
1135         (or radical
1136             (setq radical value)))
1137       (setq attributes (delq 'kangxi-radical attributes))
1138       )
1139     (when (and (memq 'kangxi-strokes attributes)
1140                (setq value (get-char-attribute char 'kangxi-strokes)))
1141       (unless (eq value strokes)
1142         (insert (format "(kangxi-strokes\t . %S)%s"
1143                         value
1144                         line-breaking))
1145         (or strokes
1146             (setq strokes value)))
1147       (setq attributes (delq 'kangxi-strokes attributes))
1148       )
1149     (when (and (memq 'japanese-radical attributes)
1150                (setq value (get-char-attribute char 'japanese-radical)))
1151       (unless (eq value radical)
1152         (insert (format "(japanese-radical\t . %S)\t; %c%s"
1153                         value
1154                         (ideographic-radical value)
1155                         line-breaking))
1156         (or radical
1157             (setq radical value)))
1158       (setq attributes (delq 'japanese-radical attributes))
1159       )
1160     (when (and (memq 'japanese-strokes attributes)
1161                (setq value (get-char-attribute char 'japanese-strokes)))
1162       (unless (eq value strokes)
1163         (insert (format "(japanese-strokes\t . %S)%s"
1164                         value
1165                         line-breaking))
1166         (or strokes
1167             (setq strokes value)))
1168       (setq attributes (delq 'japanese-strokes attributes))
1169       )
1170     (when (and (memq 'cns-radical attributes)
1171                (setq value (get-char-attribute char 'cns-radical)))
1172       (insert (format "(cns-radical\t . %S)\t; %c%s"
1173                       value
1174                       (ideographic-radical value)
1175                       line-breaking))
1176       (setq attributes (delq 'cns-radical attributes))
1177       )
1178     (when (and (memq 'cns-strokes attributes)
1179                (setq value (get-char-attribute char 'cns-strokes)))
1180       (unless (eq value strokes)
1181         (insert (format "(cns-strokes\t . %S)%s"
1182                         value
1183                         line-breaking))
1184         (or strokes
1185             (setq strokes value)))
1186       (setq attributes (delq 'cns-strokes attributes))
1187       )
1188     ;; (when (and (memq 'shinjigen-1-radical attributes)
1189     ;;            (setq value (get-char-attribute char 'shinjigen-1-radical)))
1190     ;;   (unless (eq value radical)
1191     ;;     (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
1192     ;;                     value
1193     ;;                     (ideographic-radical value)
1194     ;;                     line-breaking))
1195     ;;     (or radical
1196     ;;         (setq radical value)))
1197     ;;   (setq attributes (delq 'shinjigen-1-radical attributes))
1198     ;;   )
1199     (when (and (memq 'ideographic- attributes)
1200                (setq value (get-char-attribute char 'ideographic-)))
1201       (insert "(ideographic-       ")
1202       (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1203             separator nil)
1204       (while (consp value)
1205         (setq cell (car value))
1206         (if (integerp cell)
1207             (setq cell (decode-char '=ucs cell)))
1208         (cond ((characterp cell)
1209                (if separator
1210                    (insert lbs))
1211                (if readable
1212                    (insert (format "%S" cell))
1213                  (char-db-insert-char-spec cell readable))
1214                (setq separator lbs))
1215               ((consp cell)
1216                (if separator
1217                    (insert lbs))
1218                (if (consp (car cell))
1219                    (char-db-insert-char-spec cell readable)
1220                  (char-db-insert-char-reference cell readable))
1221                (setq separator lbs))
1222               (t
1223                (if separator
1224                    (insert separator))
1225                (insert (prin1-to-string cell))
1226                (setq separator " ")))
1227         (setq value (cdr value)))
1228       (insert ")")
1229       (insert line-breaking)
1230       (setq attributes (delq 'ideographic- attributes)))
1231     (when (and (memq 'total-strokes attributes)
1232                (setq value (get-char-attribute char 'total-strokes)))
1233       (insert (format "(total-strokes       . %S)%s"
1234                       value
1235                       line-breaking))
1236       (setq attributes (delq 'total-strokes attributes))
1237       )
1238     (when (and (memq '->ideograph attributes)
1239                (setq value (get-char-attribute char '->ideograph)))
1240       (insert (format "(->ideograph\t%s)%s"
1241                       (mapconcat (lambda (code)
1242                                    (cond ((symbolp code)
1243                                           (symbol-name code))
1244                                          ((integerp code)
1245                                           (format "#x%04X" code))
1246                                          (t
1247                                           (format "%s %S"
1248                                                   line-breaking code))))
1249                                  value " ")
1250                       line-breaking))
1251       (setq attributes (delq '->ideograph attributes))
1252       )
1253     ;; (when (and (memq '->decomposition attributes)
1254     ;;            (setq value (get-char-attribute char '->decomposition)))
1255     ;;   (insert (format "(->decomposition\t%s)%s"
1256     ;;                   (mapconcat (lambda (code)
1257     ;;                                (cond ((symbolp code)
1258     ;;                                       (symbol-name code))
1259     ;;                                      ((characterp code)
1260     ;;                                       (if readable
1261     ;;                                           (format "%S" code)
1262     ;;                                         (format "#x%04X"
1263     ;;                                                 (char-int code))
1264     ;;                                         ))
1265     ;;                                      ((integerp code)
1266     ;;                                       (format "#x%04X" code))
1267     ;;                                      (t
1268     ;;                                       (format "%s%S" line-breaking code))))
1269     ;;                              value " ")
1270     ;;                   line-breaking))
1271     ;;   (setq attributes (delq '->decomposition attributes))
1272     ;;   )
1273     (if (equal (get-char-attribute char '->titlecase)
1274                (get-char-attribute char '->uppercase))
1275         (setq attributes (delq '->titlecase attributes)))
1276     (when (and (memq '->mojikyo attributes)
1277                (setq value (get-char-attribute char '->mojikyo)))
1278       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1279                       value (decode-char 'mojikyo value)
1280                       line-breaking))
1281       (setq attributes (delq '->mojikyo attributes))
1282       )
1283     (when (and (memq 'hanyu-dazidian-vol attributes)
1284                (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1285       (insert (format "(hanyu-dazidian-vol  . %d)%s"
1286                       value line-breaking))
1287       (setq attributes (delq 'hanyu-dazidian-vol attributes))
1288       )
1289     (when (and (memq 'hanyu-dazidian-page attributes)
1290                (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1291       (insert (format "(hanyu-dazidian-page . %d)%s"
1292                       value line-breaking))
1293       (setq attributes (delq 'hanyu-dazidian-page attributes))
1294       )
1295     (when (and (memq 'hanyu-dazidian-char attributes)
1296                (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1297       (insert (format "(hanyu-dazidian-char . %d)%s"
1298                       value line-breaking))
1299       (setq attributes (delq 'hanyu-dazidian-char attributes))
1300       )
1301     (unless readable
1302       (dolist (ignored '(composition
1303                          ->denotational <-subsumptive ->ucs-unified
1304                          ->ideographic-component-forms))
1305         (setq attributes (delq ignored attributes))))
1306     (while attributes
1307       (setq name (car attributes))
1308       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1309                   'value-is-empty)
1310         (cond ((setq ret (find-charset name))
1311                (setq name (charset-name ret))
1312                (when (not (memq name dest-ccss))
1313                  (setq dest-ccss (cons name dest-ccss))
1314                  (char-db-insert-ccs-feature name value line-breaking))
1315                )
1316               ((string-match "^=>ucs@" (symbol-name name))
1317                (insert (format "(%-18s . #x%04X)\t; %c%s"
1318                                name value (decode-char '=ucs value)
1319                                line-breaking))
1320                )
1321               ((eq name 'jisx0208-1978/4X)
1322                (insert (format "(%-18s . #x%04X)%s"
1323                                name value
1324                                line-breaking))
1325                )
1326               ((and
1327                 (not readable)
1328                 (not (eq name '->subsumptive))
1329                 (not (eq name '->uppercase))
1330                 (not (eq name '->lowercase))
1331                 (not (eq name '->titlecase))
1332                 (not (eq name '->canonical))
1333                 (not (eq name '->Bopomofo))
1334                 (not (eq name '->mistakable))
1335                 (not (eq name '->ideographic-variants))
1336                 (null (get-char-attribute
1337                        char (intern (format "%s*sources" name))))
1338                 (not (string-match "\\*sources$" (symbol-name name)))
1339                 (null (get-char-attribute
1340                        char (intern (format "%s*note" name))))
1341                 (not (string-match "\\*note$" (symbol-name name)))
1342                 (or (eq name '<-identical)
1343                     (eq name '<-uppercase)
1344                     (eq name '<-lowercase)
1345                     (eq name '<-titlecase)
1346                     (eq name '<-canonical)
1347                     (eq name '<-ideographic-variants)
1348                     ;; (eq name '<-synonyms)
1349                     (string-match "^<-synonyms" (symbol-name name))
1350                     (eq name '<-mistakable)
1351                     (when (string-match "^->" (symbol-name name))
1352                       (cond
1353                        ((string-match "^->fullwidth" (symbol-name name))
1354                         (not (and (consp value)
1355                                   (characterp (car value))
1356                                   (encode-char
1357                                    (car value) '=ucs 'defined-only)))
1358                         )
1359                        (t)))
1360                     ))
1361                )
1362               ((or (eq name 'ideographic-structure)
1363                    (eq name 'ideographic-combination)
1364                    (eq name 'ideographic-)
1365                    (eq name '=decomposition)
1366                    (char-feature-base-name= '=decomposition name)
1367                    (char-feature-base-name= '=>decomposition name)
1368                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1369                    ;;               (symbol-name name))
1370                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1371                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1372                                  (symbol-name name))
1373                    )
1374                (char-db-insert-relation-feature char name value
1375                                                 line-breaking
1376                                                 ccss readable))
1377               ((memq name '(ideograph=
1378                             original-ideograph-of
1379                             ancient-ideograph-of
1380                             vulgar-ideograph-of
1381                             wrong-ideograph-of
1382                             ;; simplified-ideograph-of
1383                             ideographic-variants
1384                             ;; ideographic-different-form-of
1385                             ))
1386                (insert (format "(%-18s%s " name line-breaking))
1387                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1388                      separator nil)
1389                (while (consp value)
1390                  (setq cell (car value))
1391                  (if (and (consp cell)
1392                           (consp (car cell)))
1393                      (progn
1394                        (if separator
1395                            (insert lbs))
1396                        (char-db-insert-alist cell readable)
1397                        (setq separator lbs))
1398                    (if separator
1399                        (insert separator))
1400                    (insert (prin1-to-string cell))
1401                    (setq separator " "))
1402                  (setq value (cdr value)))
1403                (insert ")")
1404                (insert line-breaking))
1405               ((consp value)
1406                (insert (format "(%-18s " name))
1407                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1408                      separator nil)
1409                (while (consp value)
1410                  (setq cell (car value))
1411                  (if (and (consp cell)
1412                           (consp (car cell))
1413                           (setq ret (condition-case nil
1414                                         (find-char cell)
1415                                       (error nil))))
1416                      (progn
1417                        (setq rest cell
1418                              al nil
1419                              cal nil)
1420                        (while rest
1421                          (setq key (car (car rest)))
1422                          (if (find-charset key)
1423                              (setq cal (cons key cal))
1424                            (setq al (cons key al)))
1425                          (setq rest (cdr rest)))
1426                        (if separator
1427                            (insert lbs))
1428                        (insert-char-attributes ret
1429                                                readable
1430                                                al ; cal
1431                                                nil 'for-sub-node)
1432                        (setq separator lbs))
1433                    (setq ret (prin1-to-string cell))
1434                    (if separator
1435                        (if (< (+ (current-column)
1436                                  (length ret)
1437                                  (length separator))
1438                               76)
1439                            (insert separator)
1440                          (insert lbs)))
1441                    (insert ret)
1442                    (setq separator " "))
1443                  (setq value (cdr value)))
1444                (insert ")")
1445                (insert line-breaking))
1446               (t
1447                (insert (format "(%-18s" name))
1448                (setq ret (prin1-to-string value))
1449                (unless (< (+ (current-column)
1450                              (length ret)
1451                              3)
1452                           76)
1453                  (insert line-breaking))
1454                (insert " . " ret ")" line-breaking)
1455                ;; (insert (format "(%-18s . %S)%s"
1456                ;;                 name value
1457                ;;                 line-breaking))
1458                )
1459               ))
1460       (setq attributes (cdr attributes)))
1461     (insert ")")))
1462
1463 (defun insert-char-data (char &optional readable
1464                               attributes)
1465   (save-restriction
1466     (narrow-to-region (point)(point))
1467     (insert "(define-char
1468   '")
1469     (insert-char-attributes char readable attributes)
1470     (insert ")\n")
1471     (goto-char (point-min))
1472     (while (re-search-forward "[ \t]+$" nil t)
1473       (replace-match ""))
1474     ;; from tabify.
1475     (goto-char (point-min))
1476     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1477       (let ((column (current-column))
1478             (indent-tabs-mode t))
1479         (delete-region (match-beginning 0) (point))
1480         (indent-to column)))
1481     (goto-char (point-max))
1482     ;; (tabify (point-min)(point-max))
1483     ))
1484
1485 (defun insert-char-data-with-variant (char &optional printable
1486                                            no-ucs-unified
1487                                            script excluded-script)
1488   (insert-char-data char printable)
1489   (let ((variants (char-variants char))
1490         rest
1491         variant vs ret)
1492     (setq variants (sort variants #'<))
1493     (setq rest variants)
1494     (setq variants (cons char variants))
1495     (while rest
1496       (setq variant (car rest))
1497       (unless (get-char-attribute variant '<-subsumptive)
1498         (if (and (or (null script)
1499                      (null (setq vs (get-char-attribute variant 'script)))
1500                      (memq script vs))
1501                  (or (null excluded-script)
1502                      (null (setq vs (get-char-attribute variant 'script)))
1503                      (not (memq excluded-script vs))))
1504             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1505               (insert-char-data variant printable)
1506               (if (setq ret (char-variants variant))
1507                   (while ret
1508                     (or (memq (car ret) variants)
1509                         ;; (get-char-attribute (car ret) '<-subsumptive)
1510                         (setq rest (nconc rest (list (car ret)))))
1511                     (setq ret (cdr ret)))))))
1512       (setq rest (cdr rest)))))
1513
1514 (defun insert-char-range-data (min max &optional script excluded-script)
1515   (let ((code min)
1516         char)
1517     (while (<= code max)
1518       (setq char (decode-char '=ucs code))
1519       (if (encode-char char '=ucs 'defined-only)
1520           (insert-char-data-with-variant char nil 'no-ucs-unified
1521                                          script excluded-script))
1522       (setq code (1+ code)))))
1523
1524 (defun write-char-range-data-to-file (min max file
1525                                           &optional script excluded-script)
1526   (let ((coding-system-for-write char-db-file-coding-system))
1527     (with-temp-buffer
1528       (insert (format ";; -*- coding: %s -*-\n"
1529                       char-db-file-coding-system))
1530       (insert-char-range-data min max script excluded-script)
1531       (write-region (point-min)(point-max) file))))
1532
1533 (defvar what-character-original-window-configuration)
1534
1535 ;;;###autoload
1536 (defun what-char-definition (char)
1537   (interactive (list (char-after)))
1538   (let ((buf (get-buffer-create "*Character Description*"))
1539         (the-buf (current-buffer))
1540         (win-conf (current-window-configuration)))
1541     (pop-to-buffer buf)
1542     (make-local-variable 'what-character-original-window-configuration)
1543     (setq what-character-original-window-configuration win-conf)
1544     (setq buffer-read-only nil)
1545     (erase-buffer)
1546     (condition-case err
1547         (progn
1548           (insert-char-data-with-variant char 'printable)
1549           (unless (char-attribute-alist char)
1550             (insert (format ";; = %c\n"
1551                             (let* ((rest (split-char char))
1552                                    (ccs (pop rest))
1553                                    (code (pop rest)))
1554                               (while rest
1555                                 (setq code (logior (lsh code 8)
1556                                                    (pop rest))))
1557                               (decode-char ccs code)))))
1558           ;; (char-db-update-comment)
1559           (set-buffer-modified-p nil)
1560           (view-mode the-buf (lambda (buf)
1561                                (set-window-configuration
1562                                 what-character-original-window-configuration)
1563                                ))
1564           (goto-char (point-min)))
1565       (error (progn
1566                (set-window-configuration
1567                 what-character-original-window-configuration)
1568                (signal (car err) (cdr err)))))))
1569
1570
1571 ;;; @ end
1572 ;;;
1573
1574 (provide 'char-db-util)
1575
1576 ;;; char-db-util.el ends here