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-010
 335     =ucs-itaiji-001
 336     =ucs-itaiji-002
 337     =ucs-itaiji-003
 338     =ucs-itaiji-004
 339     =ucs-itaiji-005
 340     =ucs-itaiji-006
 341     =ucs-itaiji-008
 342     =big5-cdp-var-003
 343     =big5-cdp-var-004
 344     =big5-cdp-var-005
 345     =>ucs@iwds-1))
 346
 347
 348 ;;; @ char-db formatters
 349 ;;;
 350
 351 (defun char-db-make-char-spec (char)
 352   (let (ret char-spec)
 353     (cond ((characterp char)
 354            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 355                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 356                   (setq char-spec (list (cons '=ucs ret)))
 357                   (cond ((setq ret (get-char-attribute char 'name))
 358                          (setq char-spec (cons (cons 'name ret) char-spec))
 359                          )
 360                         ((setq ret (get-char-attribute char 'name*))
 361                          (setq char-spec (cons (cons 'name* ret) char-spec))
 362                          ))
 363                   )
 364                  ((encode-char char '=mj 'defined-only)
 365                   (setq char-spec nil)
 366                   (dolist (ccs (charset-list))
 367                     (if (and (or (eq ccs '=mj)
 368                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 369                                  (string-match "=ucs@" (symbol-name ccs))
 370                                  )
 371                              (setq ccs (charset-name ccs))
 372                              (null (assq ccs char-spec))
 373                              (setq ret (encode-char char ccs 'defined-only)))
 374                         (setq char-spec (cons (cons ccs ret) char-spec))))
 375                   )
 376                  ((encode-char char '==mj 'defined-only)
 377                   (setq char-spec nil)
 378                   (dolist (ccs (charset-list))
 379                     (if (and (or (eq ccs '==mj)
 380                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 381                                  (string-match "=ucs@" (symbol-name ccs))
 382                                  )
 383                              (setq ccs (charset-name ccs))
 384                              (null (assq ccs char-spec))
 385                              (setq ret (encode-char char ccs 'defined-only)))
 386                         (setq char-spec (cons (cons ccs ret) char-spec))))
 387                   )
 388                  ((encode-char char '=adobe-japan1 'defined-only)
 389                   (setq char-spec nil)
 390                   (dolist (ccs (charset-list))
 391                     (if (and (or (memq ccs
 392                                        '(=adobe-japan1-0
 393                                          =adobe-japan1-1
 394                                          =adobe-japan1-2
 395                                          =adobe-japan1-3
 396                                          =adobe-japan1-4
 397                                          =adobe-japan1-5
 398                                          =adobe-japan1-6
 399                                          ))
 400                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 401                                  (string-match "=ucs@" (symbol-name ccs))
 402                                  )
 403                              (setq ccs (charset-name ccs))
 404                              (null (assq ccs char-spec))
 405                              (setq ret (encode-char char ccs 'defined-only)))
 406                         (setq char-spec (cons (cons ccs ret) char-spec))))
 407                   )
 408                  ((encode-char char '==adobe-japan1 'defined-only)
 409                   (setq char-spec nil)
 410                   (dolist (ccs (charset-list))
 411                     (if (and (or (memq ccs
 412                                        '(==adobe-japan1-0
 413                                          ==adobe-japan1-1
 414                                          ==adobe-japan1-2
 415                                          ==adobe-japan1-3
 416                                          ==adobe-japan1-4
 417                                          ==adobe-japan1-5
 418                                          ==adobe-japan1-6
 419                                          ))
 420                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 421                                  (string-match "=ucs@" (symbol-name ccs))
 422                                  )
 423                              (setq ccs (charset-name ccs))
 424                              (null (assq ccs char-spec))
 425                              (setq ret (encode-char char ccs 'defined-only)))
 426                         (setq char-spec (cons (cons ccs ret) char-spec))))
 427                   )
 428                  ((setq ret
 429                         (catch 'tag
 430                           (let ((rest char-db-coded-charset-priority-list)
 431                                 ccs)
 432                             (while rest
 433                               (setq ccs (charset-name
 434                                          (find-charset (car rest))))
 435                               (if (setq ret
 436                                         (encode-char char ccs
 437                                                      'defined-only))
 438                                   (throw 'tag (cons ccs ret)))
 439                               (setq rest (cdr rest))))))
 440                   (setq char-spec (list ret))
 441                   (dolist (ccs (delq (car ret) (charset-list)))
 442                     (if (and (or (charset-iso-final-char ccs)
 443                                  (memq ccs
 444                                        '(=daikanwa
 445                                          =daikanwa@rev2
 446                                          ;; =gt-k
 447                                          =jis-x0208@1997
 448                                          ))
 449                                  (eq (charset-property ccs 'iso-ir) 177)
 450                                  ;; (string-match "=ucs@" (symbol-name ccs))
 451                                  )
 452                              (setq ccs (charset-name ccs))
 453                              (null (assq ccs char-spec))
 454                              (setq ret (encode-char char ccs 'defined-only)))
 455                         (setq char-spec (cons (cons ccs ret) char-spec))))
 456                   (if (null char-spec)
 457                       (setq char-spec (split-char char)))
 458                   (cond ((setq ret (get-char-attribute char 'name))
 459                          (setq char-spec (cons (cons 'name ret) char-spec))
 460                          )
 461                         ((setq ret (get-char-attribute char 'name*))
 462                          (setq char-spec (cons (cons 'name* ret) char-spec))
 463                          ))
 464                   )
 465                  ((setq ret (get-char-attribute
 466                              char 'ideographic-combination))
 467                   (setq char-spec
 468                         (cons (cons 'ideographic-combination ret)
 469                               char-spec))
 470                   ))
 471            char-spec)
 472           ((consp char)
 473            char))))
 474
 475 (defun char-db-insert-char-spec (char &optional readable column
 476                                       required-features)
 477   (unless column
 478     (setq column (current-column)))
 479   (let (char-spec temp-char)
 480     (setq char-spec (char-db-make-char-spec char))
 481     (unless (or (characterp char) ; char
 482                 (condition-case nil
 483                     (setq char (find-char char-spec))
 484                   (error nil)))
 485       ;; define temporary character
 486       ;;   Current implementation is dirty.
 487       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 488                                          char-spec)))
 489       (remove-char-attribute temp-char 'ideograph-daikanwa)
 490       (setq char temp-char))
 491     (insert-char-attributes char
 492                             readable
 493                             (union (mapcar #'car char-spec)
 494                                    required-features)
 495                             nil 'for-sub-node)
 496     (when temp-char
 497       ;; undefine temporary character
 498       ;;   Current implementation is dirty.
 499       (setq char-spec (char-attribute-alist temp-char))
 500       (while char-spec
 501         (remove-char-attribute temp-char (car (car char-spec)))
 502         (setq char-spec (cdr char-spec))))))
 503
 504 (defun char-db-insert-alist (alist &optional readable column)
 505   (unless column
 506     (setq column (current-column)))
 507   (let ((line-breaking
 508          (concat "\n" (make-string (1+ column) ?\ )))
 509         name value
 510         ret al ; cal
 511         key
 512         lbs cell rest separator)
 513     (insert "(")
 514     (while alist
 515       (setq name (car (car alist))
 516             value (cdr (car alist)))
 517       (cond ((eq name 'char)
 518              (insert "(char . ")
 519              (if (setq ret (condition-case nil
 520                                (find-char value)
 521                              (error nil)))
 522                  (progn
 523                    (setq al nil
 524                          ;; cal nil
 525                          )
 526                    (while value
 527                      (setq key (car (car value)))
 528                      ;; (if (find-charset key)
 529                      ;;     (setq cal (cons key cal))
 530                      (setq al (cons key al))
 531                      ;; )
 532                      (setq value (cdr value)))
 533                    (insert-char-attributes ret
 534                                            readable
 535                                            (or al 'none) ; cal
 536                                            nil 'for-sub-node))
 537                (insert (prin1-to-string value)))
 538              (insert ")")
 539              (insert line-breaking))
 540             ((consp value)
 541              (insert (format "(%-18s " name))
 542              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 543              (while (consp value)
 544                (setq cell (car value))
 545                (if (and (consp cell)
 546                         (consp (car cell))
 547                         (setq ret (condition-case nil
 548                                       (find-char cell)
 549                                     (error nil)))
 550                         )
 551                    (progn
 552                      (setq rest cell
 553                            al nil
 554                            ;; cal nil
 555                            )
 556                      (while rest
 557                        (setq key (car (car rest)))
 558                        ;; (if (find-charset key)
 559                        ;;     (setq cal (cons key cal))
 560                        (setq al (cons key al))
 561                        ;; )
 562                        (setq rest (cdr rest)))
 563                      (if separator
 564                          (insert lbs))
 565                      (insert-char-attributes ret
 566                                              readable
 567                                              al ; cal
 568                                              nil 'for-sub-node)
 569                      (setq separator lbs))
 570                  (if separator
 571                      (insert separator))
 572                  (insert (prin1-to-string cell))
 573                  (setq separator " "))
 574                (setq value (cdr value)))
 575              (insert ")")
 576              (insert line-breaking))
 577             (t
 578              (insert (format "(%-18s . %S)%s"
 579                              name value
 580                              line-breaking))))
 581       (setq alist (cdr alist))))
 582   (insert ")"))
 583
 584 (defun char-db-insert-char-reference (plist &optional readable column)
 585   (unless column
 586     (setq column (current-column)))
 587   (let ((line-breaking
 588          (concat "\n" (make-string (1+ column) ?\ )))
 589         (separator "")
 590         name value)
 591     (insert "(")
 592     (while plist
 593       (setq name (pop plist))
 594       (setq value (pop plist))
 595       (cond ((eq name :char)
 596              (insert separator)
 597              (insert ":char\t")
 598              (cond ((numberp value)
 599                     (setq value (decode-char '=ucs value)))
 600                    ;; ((consp value)
 601                    ;;  (setq value (or (find-char value)
 602                    ;;                  value)))
 603                    )
 604              (char-db-insert-char-spec value readable)
 605              (insert line-breaking)
 606              (setq separator ""))
 607             ((eq name :radical)
 608              (insert (format "%s%s\t%d ; %c%s"
 609                              separator
 610                              name value
 611                              (ideographic-radical value)
 612                              line-breaking))
 613              (setq separator ""))
 614             (t
 615              (insert (format "%s%s\t%S" separator name value))
 616              (setq separator line-breaking)))
 617       ))
 618   (insert ")"))
 619
 620 (defun char-db-decode-isolated-char (ccs code-point)
 621   (let (ret)
 622     (setq ret
 623           (cond ((eq ccs 'arabic-iso8859-6)
 624                  (decode-char ccs code-point))
 625                 ;; ((eq ccs '=gt)
 626                 ;;  (decode-builtin-char '==gt code-point))
 627                 ((and (memq ccs '(=gt-pj-1
 628                                   =gt-pj-2
 629                                   =gt-pj-3
 630                                   =gt-pj-4
 631                                   =gt-pj-5
 632                                   =gt-pj-6
 633                                   =gt-pj-7
 634                                   =gt-pj-8
 635                                   =gt-pj-9
 636                                   =gt-pj-10
 637                                   =gt-pj-11))
 638                       (setq ret (decode-char ccs code-point))
 639                       (setq ret (encode-char ret '=gt 'defined-only)))
 640                  (decode-builtin-char '=gt ret))
 641                 (t
 642                  (decode-builtin-char ccs code-point))))
 643     (cond ((null ret)
 644            (or (decode-char ccs code-point)
 645                (define-char (list (cons ccs code-point)))))
 646           ((and (<= 0 (char-int ret))
 647                 (<= (char-int ret) #x1F))
 648            (decode-char '=ucs (+ #x2400 (char-int ret))))
 649           ((= (char-int ret) #x7F)
 650            ?\u2421)
 651           (t ret))))
 652
 653 (defvar char-db-convert-obsolete-format t)
 654
 655 (defun char-db-insert-ccs-feature (name value line-breaking)
 656   (cond
 657    ((integerp value)
 658     (insert
 659      (format
 660       (cond
 661        ((memq name '(=>iwds-1
 662                      ==shinjigen
 663                      =shinjigen
 664                      =shinjigen@1ed ==shinjigen@1ed
 665                      =shinjigen@rev ==shinjigen@rev
 666                      =shinjigen/+p@rev ==shinjigen/+p@rev
 667                      ===daikanwa/ho ==daikanwa/ho
 668                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 669         "(%-18s .  %04d)\t; %c")
 670        ((eq name '=shinjigen@1ed/24pr)
 671         "(%-18s . %04d)\t; %c")
 672        ((or
 673          (memq name
 674                '(===daikanwa
 675                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 676                  =daikanwa@rev1 =daikanwa@rev2
 677                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 678                  =>>daikanwa/+p
 679                  =daikanwa/+2p =>>daikanwa/+2p
 680                  =gt ==gt ===gt
 681                  =>>gt =+>gt =>gt
 682                  =gt-k ==gt-k ===gt-k
 683                  =>>gt-k =>gt-k
 684                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 685                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 686                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 687                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 688                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 689                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 690                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 691                  =>>adobe-japan1-0 =+>adobe-japan1-0
 692                  =>>adobe-japan1-1 =+>adobe-japan1-1
 693                  =>>adobe-japan1-2 =+>adobe-japan1-2
 694                  =>>adobe-japan1-3 =+>adobe-japan1-3
 695                  =>>adobe-japan1-4 =+>adobe-japan1-4
 696                  =>>adobe-japan1-5 =+>adobe-japan1-5
 697                  =>>adobe-japan1-6 =+>adobe-japan1-6
 698                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 699                  =zinbun-oracle =>zinbun-oracle
 700                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 701                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 702                  ===hng-khi ===hng-khm ===hng-hok ===hng-kyd ===hng-sok
 703                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 704                  ===hng-sys ===hng-tsu ===hng-tzj
 705                  ===hng-hos ===hng-nak ===hng-jhk
 706                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 707                  ===hng-nkm ===hng-k24 ===hng-nkk
 708                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 709                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 710                  ===hng-okd ===hng-wad ===hng-kmi
 711                  ===hng-zkd ===hng-doh ===hng-jyu
 712                  ===hng-tzs ===hng-kss ===hng-kyo
 713                  ===hng-smk))
 714          ;; (string-match "^=adobe-" (symbol-name name))
 715          )
 716         "(%-18s . %05d)\t; %c")
 717        ((memq name '(=hanyo-denshi/ks
 718                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 719                      =>>hanyo-denshi/ks
 720                      =koseki ==koseki
 721                      =mj ==mj ===mj =>>mj =>mj
 722                      =zihai mojikyo))
 723         "(%-18s . %06d)\t; %c")
 724        ((memq name '(=hanyo-denshi/tk ==hanyo-denshi/tk))
 725         "(%-18s . %08d)\t; %c")
 726        ((>= (charset-dimension name) 2)
 727         "(%-18s . #x%04X)\t; %c")
 728        (t
 729         "(%-18s . #x%02X)\t; %c"))
 730       name
 731       (if (= (charset-iso-graphic-plane name) 1)
 732           (logior value
 733                   (cond ((= (charset-dimension name) 1)
 734                          #x80)
 735                         ((= (charset-dimension name) 2)
 736                          #x8080)
 737                         ((= (charset-dimension name) 3)
 738                          #x808080)
 739                         (t 0)))
 740         value)
 741       (char-db-decode-isolated-char name value)))
 742     (if (and (= (charset-chars name) 94)
 743              (= (charset-dimension name) 2))
 744         (insert (format " [%02d-%02d]"
 745                         (- (lsh value -8) 32)
 746                         (- (logand value 255) 32))))
 747     )
 748    (t
 749     (insert (format "(%-18s . %s)" name value))
 750     ))
 751   (insert line-breaking))
 752
 753 (defun char-db-insert-relation-feature (char name value line-breaking
 754                                              ccss readable)
 755   (insert (format "(%-18s%s " name line-breaking))
 756   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 757         separator cell sources required-features
 758         ret)
 759     (while (consp value)
 760       (setq cell (car value))
 761       (if (integerp cell)
 762           (setq cell (decode-char '=ucs cell)))
 763       (cond
 764        ((eq name '->subsumptive)
 765         (when (or (not (some (lambda (atr)
 766                                (get-char-attribute cell atr))
 767                              char-db-ignored-attributes))
 768                   (some (lambda (ccs)
 769                           (encode-char cell ccs 'defined-only))
 770                         ccss))
 771           (if separator
 772               (insert lbs))
 773           (let ((char-db-ignored-attributes
 774                  (cons '<-subsumptive
 775                        char-db-ignored-attributes)))
 776             (insert-char-attributes cell readable nil nil 'for-sub-node))
 777           (setq separator lbs))
 778         )
 779        ((characterp cell)
 780         (setq sources
 781               (get-char-attribute
 782                char (intern (format "%s*sources" name))))
 783         (setq required-features nil)
 784         (dolist (source sources)
 785           (cond
 786            ((memq source '(JP
 787                            JP/Jouyou
 788                            shinjigen shinjigen@1ed shinjigen@rev))
 789             (setq required-features
 790                   (union required-features
 791                          '(=jis-x0208
 792                            =jis-x0208@1990
 793                            =jis-x0213-1@2000
 794                            =jis-x0213-1@2004
 795                            =jis-x0213-2
 796                            =jis-x0212
 797                            =jis-x0208@1983
 798                            =jis-x0208@1978
 799                            =shinjigen
 800                            =shinjigen@1ed
 801                            =shinjigen@rev
 802                            =shinjigen/+p@rev))))
 803            ((eq source 'CN)
 804             (setq required-features
 805                   (union required-features
 806                          '(=gb2312
 807                            =gb12345
 808                            =iso-ir165)))))
 809           (cond
 810            ((find-charset (setq ret (intern (format "=%s" source))))
 811             (setq required-features
 812                   (cons ret required-features)))
 813            (t (setq required-features
 814                     (cons source required-features)))))
 815         (cond ((string-match "@JP" (symbol-name name))
 816                (setq required-features
 817                      (union required-features
 818                             '(=jis-x0208
 819                               =jis-x0208@1990
 820                               =jis-x0213-1-2000
 821                               =jis-x0213-2-2000
 822                               =jis-x0212
 823                               =jis-x0208@1983
 824                               =jis-x0208@1978))))
 825               ((string-match "@CN" (symbol-name name))
 826                (setq required-features
 827                      (union required-features
 828                             '(=gb2312
 829                               =gb12345
 830                               =iso-ir165)))))
 831         (if separator
 832             (insert lbs))
 833         (if readable
 834             (insert (format "%S" cell))
 835           (char-db-insert-char-spec cell readable
 836                                     nil
 837                                     required-features))
 838         (setq separator lbs))
 839        ((consp cell)
 840         (if separator
 841             (insert lbs))
 842         (if (consp (car cell))
 843             (char-db-insert-char-spec cell readable)
 844           (char-db-insert-char-reference cell readable))
 845         (setq separator lbs))
 846        (t
 847         (if separator
 848             (insert separator))
 849         (insert (prin1-to-string cell))
 850         (setq separator " ")))
 851       (setq value (cdr value)))
 852     (insert ")")
 853     (insert line-breaking)))
 854
 855 (defun insert-char-attributes (char &optional readable attributes column
 856                                     for-sub-node)
 857   (unless column
 858     (setq column (current-column)))
 859   (let (name value ; has-long-ccs-name
 860         rest
 861         radical strokes
 862         (line-breaking
 863          (concat "\n" (make-string (1+ column) ?\ )))
 864         lbs cell separator ret
 865         key al cal
 866         dest-ccss ; sources required-features
 867         ccss)
 868     (let (atr-d)
 869       (setq attributes
 870             (sort (if attributes
 871                       (if (consp attributes)
 872                           (progn
 873                             (dolist (name attributes)
 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                     (dolist (name (char-attribute-list))
 880                       (unless (memq name char-db-ignored-attributes)
 881                         (if (find-charset name)
 882                             (push name ccss))
 883                         (push name atr-d)))
 884                     atr-d)
 885                   #'char-attribute-name<)))
 886     (insert "(")
 887     (when (memq '<-subsumptive attributes)
 888       (when (or readable (not for-sub-node))
 889         (when (setq value (get-char-attribute char '<-subsumptive))
 890           (char-db-insert-relation-feature char '<-subsumptive value
 891                                            line-breaking
 892                                            ccss readable)))
 893       (setq attributes (delq '<-subsumptive attributes)))
 894     (when (and (memq '<-denotational attributes)
 895                (setq value (get-char-attribute char '<-denotational)))
 896       (char-db-insert-relation-feature char '<-denotational value
 897                                        line-breaking
 898                                        ccss readable)
 899       (setq attributes (delq '<-denotational attributes)))
 900     (when (and (memq '<-denotational@component attributes)
 901                (setq value (get-char-attribute char '<-denotational@component)))
 902       (char-db-insert-relation-feature char '<-denotational@component value
 903                                        line-breaking
 904                                        ccss readable)
 905       (setq attributes (delq '<-denotational@component attributes)))
 906     (when (and (memq 'name attributes)
 907                (setq value (get-char-attribute char 'name)))
 908       (insert (format
 909                (if (> (+ (current-column) (length value)) 48)
 910                    "(name . %S)%s"
 911                  "(name               . %S)%s")
 912                value line-breaking))
 913       (setq attributes (delq 'name attributes))
 914       )
 915     (when (and (memq 'name* attributes)
 916                (setq value (get-char-attribute char 'name*)))
 917       (insert (format
 918                (if (> (+ (current-column) (length value)) 48)
 919                    "(name* . %S)%s"
 920                  "(name*              . %S)%s")
 921                value line-breaking))
 922       (setq attributes (delq 'name* attributes))
 923       )
 924     (when (and (memq 'script attributes)
 925                (setq value (get-char-attribute char 'script)))
 926       (insert (format "(script\t\t%s)%s"
 927                       (mapconcat (function prin1-to-string)
 928                                  value " ")
 929                       line-breaking))
 930       (setq attributes (delq 'script attributes))
 931       )
 932     (dolist (name '(=>ucs =>ucs*))
 933       (when (and (memq name attributes)
 934                  (setq value (get-char-attribute char name)))
 935         (insert (format "(%-18s . #x%04X)\t; %c%s"
 936                         name value (decode-char '=ucs value)
 937                         line-breaking))
 938         (setq attributes (delq name attributes))))
 939     (dolist (name '(=>ucs@gb =>ucs@big5))
 940       (when (and (memq name attributes)
 941                  (setq value (get-char-attribute char name)))
 942         (insert (format "(%-18s . #x%04X)\t; %c%s"
 943                         name value
 944                         (decode-char (intern
 945                                       (concat "="
 946                                               (substring
 947                                                (symbol-name name) 2)))
 948                                      value)
 949                         line-breaking))
 950         (setq attributes (delq name attributes))
 951         ))
 952     ;; (dolist (name '(=>daikanwa))
 953     ;;   (when (and (memq name attributes)
 954     ;;              (setq value (get-char-attribute char name)))
 955     ;;     (insert
 956     ;;      (if (integerp value)
 957     ;;          (format "(%-18s . %05d)\t; %c%s"
 958     ;;                  name value (decode-char '=daikanwa value)
 959     ;;                  line-breaking)
 960     ;;        (format "(%-18s %s)\t; %c%s"
 961     ;;                name
 962     ;;                (mapconcat (function prin1-to-string)
 963     ;;                           value " ")
 964     ;;                (char-representative-of-daikanwa char)
 965     ;;                line-breaking)))
 966     ;;     (setq attributes (delq name attributes))))
 967     (when (and (memq 'general-category attributes)
 968                (setq value (get-char-attribute char 'general-category)))
 969       (insert (format
 970                "(general-category\t%s) ; %s%s"
 971                (mapconcat (lambda (cell)
 972                             (format "%S" cell))
 973                           value " ")
 974                (cond ((rassoc value unidata-normative-category-alist)
 975                       "Normative Category")
 976                      ((rassoc value unidata-informative-category-alist)
 977                       "Informative Category")
 978                      (t
 979                       "Unknown Category"))
 980                line-breaking))
 981       (setq attributes (delq 'general-category attributes))
 982       )
 983     (when (and (memq 'bidi-category attributes)
 984                (setq value (get-char-attribute char 'bidi-category)))
 985       (insert (format "(bidi-category\t. %S)%s"
 986                       value
 987                       line-breaking))
 988       (setq attributes (delq 'bidi-category attributes))
 989       )
 990     (unless (or (not (memq 'mirrored attributes))
 991                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
 992                     'empty))
 993       (insert (format "(mirrored\t\t. %S)%s"
 994                       value
 995                       line-breaking))
 996       (setq attributes (delq 'mirrored attributes))
 997       )
 998     (cond
 999      ((and (memq 'decimal-digit-value attributes)
1000            (setq value (get-char-attribute char 'decimal-digit-value)))
1001       (insert (format "(decimal-digit-value . %S)%s"
1002                       value
1003                       line-breaking))
1004       (setq attributes (delq 'decimal-digit-value attributes))
1005       (when (and (memq 'digit-value attributes)
1006                  (setq value (get-char-attribute char 'digit-value)))
1007         (insert (format "(digit-value\t . %S)%s"
1008                         value
1009                         line-breaking))
1010         (setq attributes (delq 'digit-value attributes))
1011         )
1012       (when (and (memq 'numeric-value attributes)
1013                  (setq value (get-char-attribute char 'numeric-value)))
1014         (insert (format "(numeric-value\t . %S)%s"
1015                         value
1016                         line-breaking))
1017         (setq attributes (delq 'numeric-value attributes))
1018         )
1019       )
1020      (t
1021       (when (and (memq 'digit-value attributes)
1022                  (setq value (get-char-attribute char 'digit-value)))
1023         (insert (format "(digit-value\t. %S)%s"
1024                         value
1025                         line-breaking))
1026         (setq attributes (delq 'digit-value attributes))
1027         )
1028       (when (and (memq 'numeric-value attributes)
1029                  (setq value (get-char-attribute char 'numeric-value)))
1030         (insert (format "(numeric-value\t. %S)%s"
1031                         value
1032                         line-breaking))
1033         (setq attributes (delq 'numeric-value attributes))
1034         )))
1035     (when (and (memq 'iso-10646-comment attributes)
1036                (setq value (get-char-attribute char 'iso-10646-comment)))
1037       (insert (format "(iso-10646-comment\t. %S)%s"
1038                       value
1039                       line-breaking))
1040       (setq attributes (delq 'iso-10646-comment attributes))
1041       )
1042     (when (and (memq 'morohashi-daikanwa attributes)
1043                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1044       (insert (format "(morohashi-daikanwa\t%s)%s"
1045                       (mapconcat (function prin1-to-string) value " ")
1046                       line-breaking))
1047       (setq attributes (delq 'morohashi-daikanwa attributes))
1048       )
1049     (setq radical nil
1050           strokes nil)
1051     (when (and (memq 'ideographic-radical attributes)
1052                (setq value (get-char-attribute char 'ideographic-radical)))
1053       (setq radical value)
1054       (insert (format "(ideographic-radical . %S)\t; %c%s"
1055                       radical
1056                       (ideographic-radical radical)
1057                       line-breaking))
1058       (setq attributes (delq 'ideographic-radical attributes))
1059       )
1060     (when (and (memq 'shuowen-radical attributes)
1061                (setq value (get-char-attribute char 'shuowen-radical)))
1062       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1063                       value
1064                       (shuowen-radical value)
1065                       line-breaking))
1066       (setq attributes (delq 'shuowen-radical attributes))
1067       )
1068     (let (key)
1069       (dolist (domain
1070                (append
1071                 char-db-feature-domains
1072                 (let (dest domain)
1073                   (dolist (feature (char-attribute-list))
1074                     (setq feature (symbol-name feature))
1075                     (when (string-match
1076                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1077                            feature)
1078                       (setq domain (intern (match-string 2 feature)))
1079                      (unless (memq domain dest)
1080                        (setq dest (cons domain dest)))))
1081                   (sort dest #'string<))))
1082         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1083         (when (and (memq key attributes)
1084                    (setq value (get-char-attribute char key)))
1085           (setq radical value)
1086           (insert (format "(%s . %S)\t; %c%s"
1087                           key
1088                           radical
1089                           (ideographic-radical radical)
1090                           line-breaking))
1091           (setq attributes (delq key attributes))
1092           )
1093         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1094         (when (and (memq key attributes)
1095                    (setq value (get-char-attribute char key)))
1096           (setq strokes value)
1097           (insert (format "(%s . %S)%s"
1098                           key
1099                           strokes
1100                           line-breaking))
1101           (setq attributes (delq key attributes))
1102           )
1103         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1104         (when (and (memq key attributes)
1105                    (setq value (get-char-attribute char key)))
1106           (insert (format "(%s       . %S)%s"
1107                           key
1108                           value
1109                           line-breaking))
1110           (setq attributes (delq key attributes))
1111           )
1112         (dolist (feature '(ideographic-radical
1113                            ideographic-strokes
1114                            total-strokes))
1115           (setq key (intern (format "%s@%s*sources" feature domain)))
1116           (when (and (memq key attributes)
1117                      (setq value (get-char-attribute char key)))
1118             (insert (format "(%s%s" key line-breaking))
1119             (dolist (cell value)
1120               (insert (format " %s" cell)))
1121             (insert ")")
1122             (insert line-breaking)
1123             (setq attributes (delq key attributes))
1124             ))
1125         ))
1126     (when (and (memq 'ideographic-strokes attributes)
1127                (setq value (get-char-attribute char 'ideographic-strokes)))
1128       (setq strokes value)
1129       (insert (format "(ideographic-strokes . %S)%s"
1130                       strokes
1131                       line-breaking))
1132       (setq attributes (delq 'ideographic-strokes attributes))
1133       )
1134     (when (and (memq 'kangxi-radical attributes)
1135                (setq value (get-char-attribute char 'kangxi-radical)))
1136       (unless (eq value radical)
1137         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1138                         value
1139                         (ideographic-radical value)
1140                         line-breaking))
1141         (or radical
1142             (setq radical value)))
1143       (setq attributes (delq 'kangxi-radical attributes))
1144       )
1145     (when (and (memq 'kangxi-strokes attributes)
1146                (setq value (get-char-attribute char 'kangxi-strokes)))
1147       (unless (eq value strokes)
1148         (insert (format "(kangxi-strokes\t . %S)%s"
1149                         value
1150                         line-breaking))
1151         (or strokes
1152             (setq strokes value)))
1153       (setq attributes (delq 'kangxi-strokes attributes))
1154       )
1155     ;; (when (and (memq 'japanese-radical attributes)
1156     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1157     ;;   (unless (eq value radical)
1158     ;;     (insert (format "(japanese-radical\t . %S)\t; %c%s"
1159     ;;                     value
1160     ;;                     (ideographic-radical value)
1161     ;;                     line-breaking))
1162     ;;     (or radical
1163     ;;         (setq radical value)))
1164     ;;   (setq attributes (delq 'japanese-radical attributes))
1165     ;;   )
1166     (when (and (memq 'japanese-strokes attributes)
1167                (setq value (get-char-attribute char 'japanese-strokes)))
1168       (unless (eq value strokes)
1169         (insert (format "(japanese-strokes\t . %S)%s"
1170                         value
1171                         line-breaking))
1172         (or strokes
1173             (setq strokes value)))
1174       (setq attributes (delq 'japanese-strokes attributes))
1175       )
1176     (when (and (memq 'cns-radical attributes)
1177                (setq value (get-char-attribute char 'cns-radical)))
1178       (insert (format "(cns-radical\t . %S)\t; %c%s"
1179                       value
1180                       (ideographic-radical value)
1181                       line-breaking))
1182       (setq attributes (delq 'cns-radical attributes))
1183       )
1184     (when (and (memq 'cns-strokes attributes)
1185                (setq value (get-char-attribute char 'cns-strokes)))
1186       (unless (eq value strokes)
1187         (insert (format "(cns-strokes\t . %S)%s"
1188                         value
1189                         line-breaking))
1190         (or strokes
1191             (setq strokes value)))
1192       (setq attributes (delq 'cns-strokes attributes))
1193       )
1194     ;; (when (and (memq 'shinjigen-1-radical attributes)
1195     ;;            (setq value (get-char-attribute char 'shinjigen-1-radical)))
1196     ;;   (unless (eq value radical)
1197     ;;     (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
1198     ;;                     value
1199     ;;                     (ideographic-radical value)
1200     ;;                     line-breaking))
1201     ;;     (or radical
1202     ;;         (setq radical value)))
1203     ;;   (setq attributes (delq 'shinjigen-1-radical attributes))
1204     ;;   )
1205     ;; (when (and (memq 'ideographic- attributes)
1206     ;;            (setq value (get-char-attribute char 'ideographic-)))
1207     ;;   (insert "(ideographic-       ")
1208     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1209     ;;         separator nil)
1210     ;;   (while (consp value)
1211     ;;     (setq cell (car value))
1212     ;;     (if (integerp cell)
1213     ;;         (setq cell (decode-char '=ucs cell)))
1214     ;;     (cond ((characterp cell)
1215     ;;            (if separator
1216     ;;                (insert lbs))
1217     ;;            (if readable
1218     ;;                (insert (format "%S" cell))
1219     ;;              (char-db-insert-char-spec cell readable))
1220     ;;            (setq separator lbs))
1221     ;;           ((consp cell)
1222     ;;            (if separator
1223     ;;                (insert lbs))
1224     ;;            (if (consp (car cell))
1225     ;;                (char-db-insert-char-spec cell readable)
1226     ;;              (char-db-insert-char-reference cell readable))
1227     ;;            (setq separator lbs))
1228     ;;           (t
1229     ;;            (if separator
1230     ;;                (insert separator))
1231     ;;            (insert (prin1-to-string cell))
1232     ;;            (setq separator " ")))
1233     ;;     (setq value (cdr value)))
1234     ;;   (insert ")")
1235     ;;   (insert line-breaking)
1236     ;;   (setq attributes (delq 'ideographic- attributes)))
1237     (when (and (memq 'total-strokes attributes)
1238                (setq value (get-char-attribute char 'total-strokes)))
1239       (insert (format "(total-strokes       . %S)%s"
1240                       value
1241                       line-breaking))
1242       (setq attributes (delq 'total-strokes attributes))
1243       )
1244     ;; (when (and (memq '->ideograph attributes)
1245     ;;            (setq value (get-char-attribute char '->ideograph)))
1246     ;;   (insert (format "(->ideograph\t%s)%s"
1247     ;;                   (mapconcat (lambda (code)
1248     ;;                                (cond ((symbolp code)
1249     ;;                                       (symbol-name code))
1250     ;;                                      ((integerp code)
1251     ;;                                       (format "#x%04X" code))
1252     ;;                                      (t
1253     ;;                                       (format "%s %S"
1254     ;;                                               line-breaking code))))
1255     ;;                              value " ")
1256     ;;                   line-breaking))
1257     ;;   (setq attributes (delq '->ideograph attributes))
1258     ;;   )
1259     ;; (when (and (memq '->decomposition attributes)
1260     ;;            (setq value (get-char-attribute char '->decomposition)))
1261     ;;   (insert (format "(->decomposition\t%s)%s"
1262     ;;                   (mapconcat (lambda (code)
1263     ;;                                (cond ((symbolp code)
1264     ;;                                       (symbol-name code))
1265     ;;                                      ((characterp code)
1266     ;;                                       (if readable
1267     ;;                                           (format "%S" code)
1268     ;;                                         (format "#x%04X"
1269     ;;                                                 (char-int code))
1270     ;;                                         ))
1271     ;;                                      ((integerp code)
1272     ;;                                       (format "#x%04X" code))
1273     ;;                                      (t
1274     ;;                                       (format "%s%S" line-breaking code))))
1275     ;;                              value " ")
1276     ;;                   line-breaking))
1277     ;;   (setq attributes (delq '->decomposition attributes))
1278     ;;   )
1279     (if (equal (get-char-attribute char '->titlecase)
1280                (get-char-attribute char '->uppercase))
1281         (setq attributes (delq '->titlecase attributes)))
1282     (when (and (memq '->mojikyo attributes)
1283                (setq value (get-char-attribute char '->mojikyo)))
1284       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1285                       value (decode-char 'mojikyo value)
1286                       line-breaking))
1287       (setq attributes (delq '->mojikyo attributes))
1288       )
1289     (when (and (memq 'hanyu-dazidian-vol attributes)
1290                (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1291       (insert (format "(hanyu-dazidian-vol  . %d)%s"
1292                       value line-breaking))
1293       (setq attributes (delq 'hanyu-dazidian-vol attributes))
1294       )
1295     (when (and (memq 'hanyu-dazidian-page attributes)
1296                (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1297       (insert (format "(hanyu-dazidian-page . %d)%s"
1298                       value line-breaking))
1299       (setq attributes (delq 'hanyu-dazidian-page attributes))
1300       )
1301     (when (and (memq 'hanyu-dazidian-char attributes)
1302                (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1303       (insert (format "(hanyu-dazidian-char . %d)%s"
1304                       value line-breaking))
1305       (setq attributes (delq 'hanyu-dazidian-char attributes))
1306       )
1307     (unless readable
1308       (dolist (ignored '(composition
1309                          ->denotational <-subsumptive ->ucs-unified
1310                          ->ideographic-component-forms))
1311         (setq attributes (delq ignored attributes))))
1312     (while attributes
1313       (setq name (car attributes))
1314       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1315                   'value-is-empty)
1316         (cond ((setq ret (find-charset name))
1317                (setq name (charset-name ret))
1318                (when (not (memq name dest-ccss))
1319                  (setq dest-ccss (cons name dest-ccss))
1320                  (char-db-insert-ccs-feature name value line-breaking))
1321                )
1322               ((string-match "^=>ucs@" (symbol-name name))
1323                (insert (format "(%-18s . #x%04X)\t; %c%s"
1324                                name value (decode-char '=ucs value)
1325                                line-breaking))
1326                )
1327               ((eq name 'jisx0208-1978/4X)
1328                (insert (format "(%-18s . #x%04X)%s"
1329                                name value
1330                                line-breaking))
1331                )
1332               ((and
1333                 (not readable)
1334                 (not (eq name '->subsumptive))
1335                 (not (eq name '->uppercase))
1336                 (not (eq name '->lowercase))
1337                 (not (eq name '->titlecase))
1338                 (not (eq name '->canonical))
1339                 (not (eq name '->Bopomofo))
1340                 (not (eq name '->mistakable))
1341                 (not (eq name '->ideographic-variants))
1342                 (null (get-char-attribute
1343                        char (intern (format "%s*sources" name))))
1344                 (not (string-match "\\*sources$" (symbol-name name)))
1345                 (null (get-char-attribute
1346                        char (intern (format "%s*note" name))))
1347                 (not (string-match "\\*note$" (symbol-name name)))
1348                 (or (eq name '<-identical)
1349                     (eq name '<-uppercase)
1350                     (eq name '<-lowercase)
1351                     (eq name '<-titlecase)
1352                     (eq name '<-canonical)
1353                     (eq name '<-ideographic-variants)
1354                     ;; (eq name '<-synonyms)
1355                     (string-match "^<-synonyms" (symbol-name name))
1356                     (eq name '<-mistakable)
1357                     (when (string-match "^->" (symbol-name name))
1358                       (cond
1359                        ((string-match "^->fullwidth" (symbol-name name))
1360                         (not (and (consp value)
1361                                   (characterp (car value))
1362                                   (encode-char
1363                                    (car value) '=ucs 'defined-only)))
1364                         )
1365                        (t)))
1366                     ))
1367                )
1368               ((or (eq name 'ideographic-structure)
1369                    (eq name 'ideographic-combination)
1370                    ;; (eq name 'ideographic-)
1371                    (eq name '=decomposition)
1372                    (char-feature-base-name= '=decomposition name)
1373                    (char-feature-base-name= '=>decomposition name)
1374                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1375                    ;;               (symbol-name name))
1376                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1377                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1378                                  (symbol-name name))
1379                    )
1380                (char-db-insert-relation-feature char name value
1381                                                 line-breaking
1382                                                 ccss readable))
1383               ;; ((memq name '(ideograph=
1384               ;;               original-ideograph-of
1385               ;;               ancient-ideograph-of
1386               ;;               vulgar-ideograph-of
1387               ;;               wrong-ideograph-of
1388               ;;               ;; simplified-ideograph-of
1389               ;;               ideographic-variants
1390               ;;               ;; ideographic-different-form-of
1391               ;;               ))
1392               ;;  (insert (format "(%-18s%s " name line-breaking))
1393               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1394               ;;        separator nil)
1395               ;;  (while (consp value)
1396               ;;    (setq cell (car value))
1397               ;;    (if (and (consp cell)
1398               ;;             (consp (car cell)))
1399               ;;        (progn
1400               ;;          (if separator
1401               ;;              (insert lbs))
1402               ;;          (char-db-insert-alist cell readable)
1403               ;;          (setq separator lbs))
1404               ;;      (if separator
1405               ;;          (insert separator))
1406               ;;      (insert (prin1-to-string cell))
1407               ;;      (setq separator " "))
1408               ;;    (setq value (cdr value)))
1409               ;;  (insert ")")
1410               ;;  (insert line-breaking))
1411               ((consp value)
1412                (insert (format "(%-18s " name))
1413                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1414                      separator nil)
1415                (while (consp value)
1416                  (setq cell (car value))
1417                  (if (and (consp cell)
1418                           (consp (car cell))
1419                           (setq ret (condition-case nil
1420                                         (find-char cell)
1421                                       (error nil))))
1422                      (progn
1423                        (setq rest cell
1424                              al nil
1425                              cal nil)
1426                        (while rest
1427                          (setq key (car (car rest)))
1428                          (if (find-charset key)
1429                              (setq cal (cons key cal))
1430                            (setq al (cons key al)))
1431                          (setq rest (cdr rest)))
1432                        (if separator
1433                            (insert lbs))
1434                        (insert-char-attributes ret
1435                                                readable
1436                                                al ; cal
1437                                                nil 'for-sub-node)
1438                        (setq separator lbs))
1439                    (setq ret (prin1-to-string cell))
1440                    (if separator
1441                        (if (< (+ (current-column)
1442                                  (length ret)
1443                                  (length separator))
1444                               76)
1445                            (insert separator)
1446                          (insert lbs)))
1447                    (insert ret)
1448                    (setq separator " "))
1449                  (setq value (cdr value)))
1450                (insert ")")
1451                (insert line-breaking))
1452               (t
1453                (insert (format "(%-18s" name))
1454                (setq ret (prin1-to-string value))
1455                (unless (< (+ (current-column)
1456                              (length ret)
1457                              3)
1458                           76)
1459                  (insert line-breaking))
1460                (insert " . " ret ")" line-breaking)
1461                ;; (insert (format "(%-18s . %S)%s"
1462                ;;                 name value
1463                ;;                 line-breaking))
1464                )
1465               ))
1466       (setq attributes (cdr attributes)))
1467     (insert ")")))
1468
1469 (defun insert-char-data (char &optional readable
1470                               attributes)
1471   (save-restriction
1472     (narrow-to-region (point)(point))
1473     (insert "(define-char
1474   '")
1475     (insert-char-attributes char readable attributes)
1476     (insert ")\n")
1477     (goto-char (point-min))
1478     (while (re-search-forward "[ \t]+$" nil t)
1479       (replace-match ""))
1480     ;; from tabify.
1481     (goto-char (point-min))
1482     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1483       (let ((column (current-column))
1484             (indent-tabs-mode t))
1485         (delete-region (match-beginning 0) (point))
1486         (indent-to column)))
1487     (goto-char (point-max))
1488     ;; (tabify (point-min)(point-max))
1489     ))
1490
1491 (defun insert-char-data-with-variant (char &optional printable
1492                                            no-ucs-unified
1493                                            script excluded-script)
1494   (insert-char-data char printable)
1495   (let ((variants (char-variants char))
1496         rest
1497         variant vs ret)
1498     (setq variants (sort variants #'<))
1499     (setq rest variants)
1500     (setq variants (cons char variants))
1501     (while rest
1502       (setq variant (car rest))
1503       (unless (get-char-attribute variant '<-subsumptive)
1504         (if (and (or (null script)
1505                      (null (setq vs (get-char-attribute variant 'script)))
1506                      (memq script vs))
1507                  (or (null excluded-script)
1508                      (null (setq vs (get-char-attribute variant 'script)))
1509                      (not (memq excluded-script vs))))
1510             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1511               (insert-char-data variant printable)
1512               (if (setq ret (char-variants variant))
1513                   (while ret
1514                     (or (memq (car ret) variants)
1515                         ;; (get-char-attribute (car ret) '<-subsumptive)
1516                         (setq rest (nconc rest (list (car ret)))))
1517                     (setq ret (cdr ret)))))))
1518       (setq rest (cdr rest)))))
1519
1520 (defun insert-char-range-data (min max &optional script excluded-script)
1521   (let ((code min)
1522         char)
1523     (while (<= code max)
1524       (setq char (decode-char '=ucs code))
1525       (if (encode-char char '=ucs 'defined-only)
1526           (insert-char-data-with-variant char nil 'no-ucs-unified
1527                                          script excluded-script))
1528       (setq code (1+ code)))))
1529
1530 (defun write-char-range-data-to-file (min max file
1531                                           &optional script excluded-script)
1532   (let ((coding-system-for-write char-db-file-coding-system))
1533     (with-temp-buffer
1534       (insert (format ";; -*- coding: %s -*-\n"
1535                       char-db-file-coding-system))
1536       (insert-char-range-data min max script excluded-script)
1537       (write-region (point-min)(point-max) file))))
1538
1539 (defvar what-character-original-window-configuration)
1540
1541 ;;;###autoload
1542 (defun what-char-definition (char)
1543   (interactive (list (char-after)))
1544   (let ((buf (get-buffer-create "*Character Description*"))
1545         (the-buf (current-buffer))
1546         (win-conf (current-window-configuration)))
1547     (pop-to-buffer buf)
1548     (make-local-variable 'what-character-original-window-configuration)
1549     (setq what-character-original-window-configuration win-conf)
1550     (setq buffer-read-only nil)
1551     (erase-buffer)
1552     (condition-case err
1553         (progn
1554           (insert-char-data-with-variant char 'printable)
1555           (unless (char-attribute-alist char)
1556             (insert (format ";; = %c\n"
1557                             (let* ((rest (split-char char))
1558                                    (ccs (pop rest))
1559                                    (code (pop rest)))
1560                               (while rest
1561                                 (setq code (logior (lsh code 8)
1562                                                    (pop rest))))
1563                               (decode-char ccs code)))))
1564           ;; (char-db-update-comment)
1565           (set-buffer-modified-p nil)
1566           (view-mode the-buf (lambda (buf)
1567                                (set-window-configuration
1568                                 what-character-original-window-configuration)
1569                                ))
1570           (goto-char (point-min)))
1571       (error (progn
1572                (set-window-configuration
1573                 what-character-original-window-configuration)
1574                (signal (car err) (cdr err)))))))
1575
1576
1577 ;;; @ end
1578 ;;;
1579
1580 (provide 'char-db-util)
1581
1582 ;;; char-db-util.el ends here