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 ;;   2008,2009,2010,2011,2012,2013,2014,2015,2016,2017 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/ia
 189     ==hanyo-denshi/ib
 190     ==hanyo-denshi/hg
 191     ==hanyo-denshi/jt
 192     ==hanyo-denshi/ks
 193     ==hanyo-denshi/tk
 194     =gt
 195     =gt-k
 196     =daikanwa
 197     =daikanwa@rev2
 198     =daikanwa@rev1
 199     =daikanwa/+p
 200     ==daikanwa
 201     =cbeta
 202     ideograph-hanziku-1
 203     ideograph-hanziku-2
 204     ideograph-hanziku-3
 205     ideograph-hanziku-4
 206     ideograph-hanziku-5
 207     ideograph-hanziku-6
 208     ideograph-hanziku-7
 209     ideograph-hanziku-8
 210     ideograph-hanziku-9
 211     ideograph-hanziku-10
 212     ideograph-hanziku-11
 213     ideograph-hanziku-12
 214     ;; =>>>adobe-japan1-0
 215     ;; =>>>adobe-japan1-1
 216     ;; =>>>adobe-japan1-2
 217     ;; =>>>adobe-japan1-3
 218     ;; =>>>adobe-japan1-4
 219     ;; =>>>adobe-japan1-5
 220     ;; =>>>adobe-japan1-6
 221     ;; =>>>jis-x0208
 222     ;; =>>>jis-x0213-1
 223     ;; =>>>jis-x0213-2
 224     ;; =>>>hanyo-denshi/ja
 225     ;; =>>>hanyo-denshi/jb
 226     ;; =>>>hanyo-denshi/jc
 227     ;; =>>>hanyo-denshi/ft
 228     ;; =>>>hanyo-denshi/ib
 229     ;; =>>>hanyo-denshi/hg
 230     ;; =>>>hanyo-denshi/jt
 231     ;; =>>>hanyo-denshi/ks
 232     ;; =>>>gt
 233     =>>adobe-japan1-0
 234     =>>adobe-japan1-1
 235     =>>adobe-japan1-2
 236     =>>adobe-japan1-3
 237     =>>adobe-japan1-4
 238     =>>adobe-japan1-5
 239     =>>adobe-japan1-6
 240     =>>jis-x0208
 241     =>>jis-x0213-1
 242     =>>jis-x0213-1@2000
 243     =>>jis-x0213-1@2004
 244     =>>jis-x0213-2
 245     =>>jis-x0208@1978
 246     =>>hanyo-denshi/ft
 247     =>>hanyo-denshi/jt
 248     =>>hanyo-denshi/ks
 249     =>>gt
 250     =>>daikanwa
 251     =>>cbeta
 252     =+>jis-x0208
 253     =+>jis-x0213-1
 254     =+>jis-x0213-2
 255     =+>adobe-japan1-0
 256     =+>adobe-japan1-1
 257     =+>adobe-japan1-2
 258     =+>adobe-japan1-3
 259     =+>adobe-japan1-4
 260     =+>adobe-japan1-5
 261     =+>adobe-japan1-6
 262     =+>jis-x0208@1978
 263     =>jis-x0208
 264     =>jis-x0208@1997
 265     =>jis-x0213-1
 266     =>jis-x0213-1@2000
 267     =>jis-x0213-1@2004
 268     =>jis-x0213-2
 269     ==>ucs@bucs
 270     =>iwds-1
 271     =>ucs@hanyo-denshi
 272     =>ucs@iso
 273     =>ucs@unicode
 274     =>ucs@jis
 275     =>ucs@cns
 276     =>ucs@ks
 277     =+>ucs@iso
 278     =+>ucs@unicode
 279     =+>ucs@jis
 280     =+>ucs@jis/1990
 281     =+>ucs@cns
 282     =+>ucs@ks
 283     =>>ucs@iso
 284     =>>ucs@unicode
 285     =>>ucs@jis
 286     =>>ucs@cns
 287     =>>>ucs@iso
 288     =>>>ucs@unicode
 289     ==ucs@iso
 290     ==ucs@unicode
 291     ;; ==ucs@cns
 292     ==gb2312
 293     ==ks-x1001
 294     ==cns11643-1
 295     ==cns11643-2
 296     ==cns11643-3
 297     ==cns11643-4
 298     ==cns11643-5
 299     ==cns11643-6
 300     ==cns11643-7
 301     ==gt
 302     ==jis-x0208@1990
 303     ;; ==jis-x0208@1983
 304     ==jis-x0208@1978
 305     ==gt-k
 306     =ucs@iso
 307     =ucs@unicode
 308     =ucs@cns
 309     ==big5-cdp
 310     ==cbeta
 311     =>>big5-cdp
 312     =>>gt-k
 313     =+>gt
 314     =>gt
 315     =>mj
 316     =>big5-cdp
 317     =>daikanwa
 318     =>daikanwa/ho
 319     =>cns11643-5
 320     =>cns11643-7
 321     =big5
 322     =big5-eten
 323     =>gt-k
 324     =zinbun-oracle
 325     =>zinbun-oracle
 326     =ruimoku-v6
 327     =>>ruimoku-v6
 328     ==ruimoku-v6
 329     =jef-china3
 330     =>cbeta
 331     =shinjigen
 332     =ucs-var-001
 333     =ucs-var-002
 334     =ucs-var-003
 335     =ucs-var-004
 336     =ucs-var-006
 337     =ucs-var-008
 338     =ucs-var-010
 339     =ucs-itaiji-001
 340     =ucs-itaiji-002
 341     =ucs-itaiji-003
 342     =ucs-itaiji-004
 343     =ucs-itaiji-005
 344     =ucs-itaiji-006
 345     =ucs-itaiji-008
 346     =ucs-itaiji-009
 347     =>ucs-itaiji-001
 348     =>ucs-itaiji-002
 349     =>ucs-itaiji-005
 350     =>big5-cdp-itaiji-001
 351     =big5-cdp-itaiji-001
 352     =big5-cdp-itaiji-002
 353     =big5-cdp-itaiji-003
 354     =big5-cdp-var-003
 355     =big5-cdp-var-004
 356     =big5-cdp-var-005
 357     =>ucs@iwds-1
 358     =>ucs@component
 359     ==ucs@gb
 360     =ucs@JP/hanazono))
 361
 362
 363 ;;; @ char-db formatters
 364 ;;;
 365
 366 (defun char-db-make-char-spec (char)
 367   (let (ret char-spec)
 368     (cond ((characterp char)
 369            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 370                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 371                   (setq char-spec (list (cons '=ucs ret)))
 372                   (cond ((setq ret (get-char-attribute char 'name))
 373                          (setq char-spec (cons (cons 'name ret) char-spec))
 374                          )
 375                         ((setq ret (get-char-attribute char 'name*))
 376                          (setq char-spec (cons (cons 'name* ret) char-spec))
 377                          ))
 378                   )
 379                  ((encode-char char '=mj 'defined-only)
 380                   (setq char-spec nil)
 381                   (dolist (ccs (charset-list))
 382                     (if (and (or (eq ccs '=mj)
 383                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 384                                  (string-match "=ucs@" (symbol-name ccs))
 385                                  )
 386                              (setq ccs (charset-name ccs))
 387                              (null (assq ccs char-spec))
 388                              (setq ret (encode-char char ccs 'defined-only)))
 389                         (setq char-spec (cons (cons ccs ret) char-spec))))
 390                   )
 391                  ((encode-char char '==mj 'defined-only)
 392                   (setq char-spec nil)
 393                   (dolist (ccs (charset-list))
 394                     (if (and (or (eq ccs '==mj)
 395                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 396                                  (string-match "=ucs@" (symbol-name ccs))
 397                                  )
 398                              (setq ccs (charset-name ccs))
 399                              (null (assq ccs char-spec))
 400                              (setq ret (encode-char char ccs 'defined-only)))
 401                         (setq char-spec (cons (cons ccs ret) char-spec))))
 402                   )
 403                  ((encode-char char '=adobe-japan1 'defined-only)
 404                   (setq char-spec nil)
 405                   (dolist (ccs (charset-list))
 406                     (if (and (or (memq ccs
 407                                        '(=adobe-japan1-0
 408                                          =adobe-japan1-1
 409                                          =adobe-japan1-2
 410                                          =adobe-japan1-3
 411                                          =adobe-japan1-4
 412                                          =adobe-japan1-5
 413                                          =adobe-japan1-6
 414                                          ))
 415                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 416                                  (string-match "=ucs@" (symbol-name ccs))
 417                                  )
 418                              (setq ccs (charset-name ccs))
 419                              (null (assq ccs char-spec))
 420                              (setq ret (encode-char char ccs 'defined-only)))
 421                         (setq char-spec (cons (cons ccs ret) char-spec))))
 422                   )
 423                  ((encode-char char '==adobe-japan1 'defined-only)
 424                   (setq char-spec nil)
 425                   (dolist (ccs (charset-list))
 426                     (if (and (or (memq ccs
 427                                        '(==adobe-japan1-0
 428                                          ==adobe-japan1-1
 429                                          ==adobe-japan1-2
 430                                          ==adobe-japan1-3
 431                                          ==adobe-japan1-4
 432                                          ==adobe-japan1-5
 433                                          ==adobe-japan1-6
 434                                          ))
 435                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 436                                  (string-match "=ucs@" (symbol-name ccs))
 437                                  )
 438                              (setq ccs (charset-name ccs))
 439                              (null (assq ccs char-spec))
 440                              (setq ret (encode-char char ccs 'defined-only)))
 441                         (setq char-spec (cons (cons ccs ret) char-spec))))
 442                   )
 443                  ((setq ret
 444                         (catch 'tag
 445                           (let ((rest char-db-coded-charset-priority-list)
 446                                 ccs)
 447                             (while rest
 448                               (setq ccs (charset-name
 449                                          (find-charset (car rest))))
 450                               (if (setq ret
 451                                         (encode-char char ccs
 452                                                      'defined-only))
 453                                   (throw 'tag (cons ccs ret)))
 454                               (setq rest (cdr rest))))))
 455                   (setq char-spec (list ret))
 456                   (dolist (ccs (delq (car ret) (charset-list)))
 457                     (if (and (or (charset-iso-final-char ccs)
 458                                  (memq ccs
 459                                        '(=daikanwa
 460                                          =daikanwa@rev2
 461                                          ;; =gt-k
 462                                          =jis-x0208@1997
 463                                          ))
 464                                  (eq (charset-property ccs 'iso-ir) 177)
 465                                  ;; (string-match "=ucs@" (symbol-name ccs))
 466                                  )
 467                              (setq ccs (charset-name ccs))
 468                              (null (assq ccs char-spec))
 469                              (setq ret (encode-char char ccs 'defined-only)))
 470                         (setq char-spec (cons (cons ccs ret) char-spec))))
 471                   (if (null char-spec)
 472                       (setq char-spec (split-char char)))
 473                   (cond ((setq ret (get-char-attribute char 'name))
 474                          (setq char-spec (cons (cons 'name ret) char-spec))
 475                          )
 476                         ((setq ret (get-char-attribute char 'name*))
 477                          (setq char-spec (cons (cons 'name* ret) char-spec))
 478                          ))
 479                   )
 480                  ((setq ret (get-char-attribute
 481                              char 'ideographic-combination))
 482                   (setq char-spec
 483                         (cons (cons 'ideographic-combination ret)
 484                               char-spec))
 485                   ))
 486            char-spec)
 487           ((consp char)
 488            char))))
 489
 490 (defun char-db-insert-char-spec (char &optional readable column
 491                                       required-features)
 492   (unless column
 493     (setq column (current-column)))
 494   (let (char-spec temp-char)
 495     (setq char-spec (char-db-make-char-spec char))
 496     (unless (or (characterp char) ; char
 497                 (condition-case nil
 498                     (setq char (find-char char-spec))
 499                   (error nil)))
 500       ;; define temporary character
 501       ;;   Current implementation is dirty.
 502       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 503                                          char-spec)))
 504       (remove-char-attribute temp-char 'ideograph-daikanwa)
 505       (setq char temp-char))
 506     (insert-char-attributes char
 507                             readable
 508                             (union (mapcar #'car char-spec)
 509                                    required-features)
 510                             nil 'for-sub-node)
 511     (when temp-char
 512       ;; undefine temporary character
 513       ;;   Current implementation is dirty.
 514       (setq char-spec (char-attribute-alist temp-char))
 515       (while char-spec
 516         (remove-char-attribute temp-char (car (car char-spec)))
 517         (setq char-spec (cdr char-spec))))))
 518
 519 (defun char-db-insert-alist (alist &optional readable column)
 520   (unless column
 521     (setq column (current-column)))
 522   (let ((line-breaking
 523          (concat "\n" (make-string (1+ column) ?\ )))
 524         name value
 525         ret al ; cal
 526         key
 527         lbs cell rest separator)
 528     (insert "(")
 529     (while alist
 530       (setq name (car (car alist))
 531             value (cdr (car alist)))
 532       (cond ((eq name 'char)
 533              (insert "(char . ")
 534              (if (setq ret (condition-case nil
 535                                (find-char value)
 536                              (error nil)))
 537                  (progn
 538                    (setq al nil
 539                          ;; cal nil
 540                          )
 541                    (while value
 542                      (setq key (car (car value)))
 543                      ;; (if (find-charset key)
 544                      ;;     (setq cal (cons key cal))
 545                      (setq al (cons key al))
 546                      ;; )
 547                      (setq value (cdr value)))
 548                    (insert-char-attributes ret
 549                                            readable
 550                                            (or al 'none) ; cal
 551                                            nil 'for-sub-node))
 552                (insert (prin1-to-string value)))
 553              (insert ")")
 554              (insert line-breaking))
 555             ((consp value)
 556              (insert (format "(%-18s " name))
 557              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 558              (while (consp value)
 559                (setq cell (car value))
 560                (if (and (consp cell)
 561                         (consp (car cell))
 562                         (setq ret (condition-case nil
 563                                       (find-char cell)
 564                                     (error nil)))
 565                         )
 566                    (progn
 567                      (setq rest cell
 568                            al nil
 569                            ;; cal nil
 570                            )
 571                      (while rest
 572                        (setq key (car (car rest)))
 573                        ;; (if (find-charset key)
 574                        ;;     (setq cal (cons key cal))
 575                        (setq al (cons key al))
 576                        ;; )
 577                        (setq rest (cdr rest)))
 578                      (if separator
 579                          (insert lbs))
 580                      (insert-char-attributes ret
 581                                              readable
 582                                              al ; cal
 583                                              nil 'for-sub-node)
 584                      (setq separator lbs))
 585                  (if separator
 586                      (insert separator))
 587                  (insert (prin1-to-string cell))
 588                  (setq separator " "))
 589                (setq value (cdr value)))
 590              (insert ")")
 591              (insert line-breaking))
 592             (t
 593              (insert (format "(%-18s . %S)%s"
 594                              name value
 595                              line-breaking))))
 596       (setq alist (cdr alist))))
 597   (insert ")"))
 598
 599 (defun char-db-insert-char-reference (plist &optional readable column)
 600   (unless column
 601     (setq column (current-column)))
 602   (let ((line-breaking
 603          (concat "\n" (make-string (1+ column) ?\ )))
 604         (separator "")
 605         name value)
 606     (insert "(")
 607     (while plist
 608       (setq name (pop plist))
 609       (setq value (pop plist))
 610       (cond ((eq name :char)
 611              (insert separator)
 612              (insert ":char\t")
 613              (cond ((numberp value)
 614                     (setq value (decode-char '=ucs value)))
 615                    ;; ((consp value)
 616                    ;;  (setq value (or (find-char value)
 617                    ;;                  value)))
 618                    )
 619              (char-db-insert-char-spec value readable)
 620              (insert line-breaking)
 621              (setq separator ""))
 622             ((eq name :radical)
 623              (insert (format "%s%s\t%d ; %c%s"
 624                              separator
 625                              name value
 626                              (ideographic-radical value)
 627                              line-breaking))
 628              (setq separator ""))
 629             (t
 630              (insert (format "%s%s\t%S" separator name value))
 631              (setq separator line-breaking)))
 632       ))
 633   (insert ")"))
 634
 635 (defun char-db-decode-isolated-char (ccs code-point)
 636   (let (ret)
 637     (setq ret
 638           (cond ((eq ccs 'arabic-iso8859-6)
 639                  (decode-char ccs code-point))
 640                 ;; ((eq ccs '=gt)
 641                 ;;  (decode-builtin-char '==gt code-point))
 642                 ((and (memq ccs '(=gt-pj-1
 643                                   =gt-pj-2
 644                                   =gt-pj-3
 645                                   =gt-pj-4
 646                                   =gt-pj-5
 647                                   =gt-pj-6
 648                                   =gt-pj-7
 649                                   =gt-pj-8
 650                                   =gt-pj-9
 651                                   =gt-pj-10
 652                                   =gt-pj-11))
 653                       (setq ret (decode-char ccs code-point))
 654                       (setq ret (encode-char ret '=gt 'defined-only)))
 655                  (decode-builtin-char '=gt ret))
 656                 (t
 657                  (decode-builtin-char ccs code-point))))
 658     (cond ((null ret)
 659            (or (decode-char ccs code-point)
 660                (define-char (list (cons ccs code-point)))))
 661           ((and (<= 0 (char-int ret))
 662                 (<= (char-int ret) #x1F))
 663            (decode-char '=ucs (+ #x2400 (char-int ret))))
 664           ((= (char-int ret) #x7F)
 665            ?\u2421)
 666           (t ret))))
 667
 668 (defvar char-db-convert-obsolete-format t)
 669
 670 (defun char-db-insert-ccs-feature (name value line-breaking)
 671   (cond
 672    ((integerp value)
 673     (insert
 674      (format
 675       (cond
 676        ((memq name '(=>iwds-1
 677                      ==shinjigen
 678                      =shinjigen
 679                      =shinjigen@1ed ==shinjigen@1ed
 680                      =shinjigen@rev ==shinjigen@rev
 681                      =shinjigen/+p@rev ==shinjigen/+p@rev
 682                      ===daikanwa/ho ==daikanwa/ho
 683                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 684         "(%-18s .  %04d)\t; %c")
 685        ((eq name '=shinjigen@1ed/24pr)
 686         "(%-18s . %04d)\t; %c")
 687        ((or
 688          (memq name
 689                '(===daikanwa
 690                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 691                  =daikanwa@rev1 =daikanwa@rev2
 692                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 693                  =>>daikanwa/+p
 694                  =daikanwa/+2p =>>daikanwa/+2p
 695                  =gt ==gt ===gt
 696                  =>>gt =+>gt =>gt
 697                  =gt-k ==gt-k ===gt-k
 698                  =>>gt-k =>gt-k
 699                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 700                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 701                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 702                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 703                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 704                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 705                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 706                  =>>adobe-japan1-0 =+>adobe-japan1-0
 707                  =>>adobe-japan1-1 =+>adobe-japan1-1
 708                  =>>adobe-japan1-2 =+>adobe-japan1-2
 709                  =>>adobe-japan1-3 =+>adobe-japan1-3
 710                  =>>adobe-japan1-4 =+>adobe-japan1-4
 711                  =>>adobe-japan1-5 =+>adobe-japan1-5
 712                  =>>adobe-japan1-6 =+>adobe-japan1-6
 713                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 714                  =zinbun-oracle =>zinbun-oracle
 715                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 716                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 717                  ===hng-khi ===hng-khm ===hng-hok ===hng-kyd ===hng-sok
 718                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 719                  ===hng-sys ===hng-tsu ===hng-tzj
 720                  ===hng-hos ===hng-nak ===hng-jhk
 721                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 722                  ===hng-nkm ===hng-k24 ===hng-nkk
 723                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 724                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 725                  ===hng-okd ===hng-wad ===hng-kmi
 726                  ===hng-zkd ===hng-doh ===hng-jyu
 727                  ===hng-tzs ===hng-kss ===hng-kyo
 728                  ===hng-smk))
 729          ;; (string-match "^=adobe-" (symbol-name name))
 730          )
 731         "(%-18s . %05d)\t; %c")
 732        ((memq name '(=hanyo-denshi/ks
 733                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 734                      =>>hanyo-denshi/ks
 735                      =koseki ==koseki
 736                      =mj ==mj ===mj =>>mj =>mj
 737                      =zihai mojikyo))
 738         "(%-18s . %06d)\t; %c")
 739        ((memq name '(=hanyo-denshi/tk ==hanyo-denshi/tk))
 740         "(%-18s . %08d)\t; %c")
 741        ((>= (charset-dimension name) 2)
 742         "(%-18s . #x%04X)\t; %c")
 743        (t
 744         "(%-18s . #x%02X)\t; %c"))
 745       name
 746       (if (= (charset-iso-graphic-plane name) 1)
 747           (logior value
 748                   (cond ((= (charset-dimension name) 1)
 749                          #x80)
 750                         ((= (charset-dimension name) 2)
 751                          #x8080)
 752                         ((= (charset-dimension name) 3)
 753                          #x808080)
 754                         (t 0)))
 755         value)
 756       (char-db-decode-isolated-char name value)))
 757     (if (and (= (charset-chars name) 94)
 758              (= (charset-dimension name) 2))
 759         (insert (format " [%02d-%02d]"
 760                         (- (lsh value -8) 32)
 761                         (- (logand value 255) 32))))
 762     )
 763    (t
 764     (insert (format "(%-18s . %s)" name value))
 765     ))
 766   (insert line-breaking))
 767
 768 (defun char-db-insert-relation-feature (char name value line-breaking
 769                                              ccss readable)
 770   (insert (format "(%-18s%s " name line-breaking))
 771   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 772         separator cell sources required-features
 773         ret)
 774     (while (consp value)
 775       (setq cell (car value))
 776       (if (integerp cell)
 777           (setq cell (decode-char '=ucs cell)))
 778       (cond
 779        ((eq name '->subsumptive)
 780         (when (or (not (some (lambda (atr)
 781                                (get-char-attribute cell atr))
 782                              char-db-ignored-attributes))
 783                   (some (lambda (ccs)
 784                           (encode-char cell ccs 'defined-only))
 785                         ccss))
 786           (if separator
 787               (insert lbs))
 788           (let ((char-db-ignored-attributes
 789                  (cons '<-subsumptive
 790                        char-db-ignored-attributes)))
 791             (insert-char-attributes cell readable nil nil 'for-sub-node))
 792           (setq separator lbs))
 793         )
 794        ((characterp cell)
 795         (setq sources
 796               (get-char-attribute
 797                char (intern (format "%s*sources" name))))
 798         (setq required-features nil)
 799         (dolist (source sources)
 800           (cond
 801            ((memq source '(JP
 802                            JP/Jouyou
 803                            shinjigen shinjigen@1ed shinjigen@rev))
 804             (setq required-features
 805                   (union required-features
 806                          '(=jis-x0208
 807                            =jis-x0208@1990
 808                            =jis-x0213-1@2000
 809                            =jis-x0213-1@2004
 810                            =jis-x0213-2
 811                            =jis-x0212
 812                            =jis-x0208@1983
 813                            =jis-x0208@1978
 814                            =shinjigen
 815                            =shinjigen@1ed
 816                            =shinjigen@rev
 817                            =shinjigen/+p@rev))))
 818            ((eq source 'CN)
 819             (setq required-features
 820                   (union required-features
 821                          '(=gb2312
 822                            =gb12345
 823                            =iso-ir165)))))
 824           (cond
 825            ((find-charset (setq ret (intern (format "=%s" source))))
 826             (setq required-features
 827                   (cons ret required-features)))
 828            (t (setq required-features
 829                     (cons source required-features)))))
 830         (cond ((string-match "@JP" (symbol-name name))
 831                (setq required-features
 832                      (union required-features
 833                             '(=jis-x0208
 834                               =jis-x0208@1990
 835                               =jis-x0213-1-2000
 836                               =jis-x0213-2-2000
 837                               =jis-x0212
 838                               =jis-x0208@1983
 839                               =jis-x0208@1978))))
 840               ((string-match "@CN" (symbol-name name))
 841                (setq required-features
 842                      (union required-features
 843                             '(=gb2312
 844                               =gb12345
 845                               =iso-ir165)))))
 846         (if separator
 847             (insert lbs))
 848         (if readable
 849             (insert (format "%S" cell))
 850           (char-db-insert-char-spec cell readable
 851                                     nil
 852                                     required-features))
 853         (setq separator lbs))
 854        ((consp cell)
 855         (if separator
 856             (insert lbs))
 857         (if (consp (car cell))
 858             (char-db-insert-char-spec cell readable)
 859           (char-db-insert-char-reference cell readable))
 860         (setq separator lbs))
 861        (t
 862         (if separator
 863             (insert separator))
 864         (insert (prin1-to-string cell))
 865         (setq separator " ")))
 866       (setq value (cdr value)))
 867     (insert ")")
 868     (insert line-breaking)))
 869
 870 (defun insert-char-attributes (char &optional readable attributes column
 871                                     for-sub-node)
 872   (unless column
 873     (setq column (current-column)))
 874   (let (name value ; has-long-ccs-name
 875         rest
 876         radical strokes
 877         (line-breaking
 878          (concat "\n" (make-string (1+ column) ?\ )))
 879         lbs cell separator ret
 880         key al cal
 881         dest-ccss ; sources required-features
 882         ccss)
 883     (let (atr-d)
 884       (setq attributes
 885             (sort (if attributes
 886                       (if (consp attributes)
 887                           (progn
 888                             (dolist (name attributes)
 889                               (unless (memq name char-db-ignored-attributes)
 890                                 (if (find-charset name)
 891                                     (push name ccss))
 892                                 (push name atr-d)))
 893                             atr-d))
 894                     (dolist (name (char-attribute-list))
 895                       (unless (memq name char-db-ignored-attributes)
 896                         (if (find-charset name)
 897                             (push name ccss))
 898                         (push name atr-d)))
 899                     atr-d)
 900                   #'char-attribute-name<)))
 901     (insert "(")
 902     (when (memq '<-subsumptive attributes)
 903       (when (or readable (not for-sub-node))
 904         (when (setq value (get-char-attribute char '<-subsumptive))
 905           (char-db-insert-relation-feature char '<-subsumptive value
 906                                            line-breaking
 907                                            ccss readable)))
 908       (setq attributes (delq '<-subsumptive attributes)))
 909     (when (and (memq '<-denotational attributes)
 910                (setq value (get-char-attribute char '<-denotational)))
 911       (char-db-insert-relation-feature char '<-denotational value
 912                                        line-breaking
 913                                        ccss readable)
 914       (setq attributes (delq '<-denotational attributes)))
 915     (when (and (memq '<-denotational@component attributes)
 916                (setq value (get-char-attribute char '<-denotational@component)))
 917       (char-db-insert-relation-feature char '<-denotational@component value
 918                                        line-breaking
 919                                        ccss readable)
 920       (setq attributes (delq '<-denotational@component attributes)))
 921     (when (and (memq 'name attributes)
 922                (setq value (get-char-attribute char 'name)))
 923       (insert (format
 924                (if (> (+ (current-column) (length value)) 48)
 925                    "(name . %S)%s"
 926                  "(name               . %S)%s")
 927                value line-breaking))
 928       (setq attributes (delq 'name attributes))
 929       )
 930     (when (and (memq 'name* attributes)
 931                (setq value (get-char-attribute char 'name*)))
 932       (insert (format
 933                (if (> (+ (current-column) (length value)) 48)
 934                    "(name* . %S)%s"
 935                  "(name*              . %S)%s")
 936                value line-breaking))
 937       (setq attributes (delq 'name* attributes))
 938       )
 939     (when (and (memq 'script attributes)
 940                (setq value (get-char-attribute char 'script)))
 941       (insert (format "(script\t\t%s)%s"
 942                       (mapconcat (function prin1-to-string)
 943                                  value " ")
 944                       line-breaking))
 945       (setq attributes (delq 'script attributes))
 946       )
 947     (dolist (name '(=>ucs =>ucs*))
 948       (when (and (memq name attributes)
 949                  (setq value (get-char-attribute char name)))
 950         (insert (format "(%-18s . #x%04X)\t; %c%s"
 951                         name value (decode-char '=ucs value)
 952                         line-breaking))
 953         (setq attributes (delq name attributes))))
 954     (dolist (name '(=>ucs@gb =>ucs@big5))
 955       (when (and (memq name attributes)
 956                  (setq value (get-char-attribute char name)))
 957         (insert (format "(%-18s . #x%04X)\t; %c%s"
 958                         name value
 959                         (decode-char (intern
 960                                       (concat "="
 961                                               (substring
 962                                                (symbol-name name) 2)))
 963                                      value)
 964                         line-breaking))
 965         (setq attributes (delq name attributes))
 966         ))
 967     ;; (dolist (name '(=>daikanwa))
 968     ;;   (when (and (memq name attributes)
 969     ;;              (setq value (get-char-attribute char name)))
 970     ;;     (insert
 971     ;;      (if (integerp value)
 972     ;;          (format "(%-18s . %05d)\t; %c%s"
 973     ;;                  name value (decode-char '=daikanwa value)
 974     ;;                  line-breaking)
 975     ;;        (format "(%-18s %s)\t; %c%s"
 976     ;;                name
 977     ;;                (mapconcat (function prin1-to-string)
 978     ;;                           value " ")
 979     ;;                (char-representative-of-daikanwa char)
 980     ;;                line-breaking)))
 981     ;;     (setq attributes (delq name attributes))))
 982     (when (and (memq 'general-category attributes)
 983                (setq value (get-char-attribute char 'general-category)))
 984       (insert (format
 985                "(general-category\t%s) ; %s%s"
 986                (mapconcat (lambda (cell)
 987                             (format "%S" cell))
 988                           value " ")
 989                (cond ((rassoc value unidata-normative-category-alist)
 990                       "Normative Category")
 991                      ((rassoc value unidata-informative-category-alist)
 992                       "Informative Category")
 993                      (t
 994                       "Unknown Category"))
 995                line-breaking))
 996       (setq attributes (delq 'general-category attributes))
 997       )
 998     (when (and (memq 'bidi-category attributes)
 999                (setq value (get-char-attribute char 'bidi-category)))
1000       (insert (format "(bidi-category\t. %S)%s"
1001                       value
1002                       line-breaking))
1003       (setq attributes (delq 'bidi-category attributes))
1004       )
1005     (unless (or (not (memq 'mirrored attributes))
1006                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
1007                     'empty))
1008       (insert (format "(mirrored\t\t. %S)%s"
1009                       value
1010                       line-breaking))
1011       (setq attributes (delq 'mirrored attributes))
1012       )
1013     (cond
1014      ((and (memq 'decimal-digit-value attributes)
1015            (setq value (get-char-attribute char 'decimal-digit-value)))
1016       (insert (format "(decimal-digit-value . %S)%s"
1017                       value
1018                       line-breaking))
1019       (setq attributes (delq 'decimal-digit-value attributes))
1020       (when (and (memq 'digit-value attributes)
1021                  (setq value (get-char-attribute char 'digit-value)))
1022         (insert (format "(digit-value\t . %S)%s"
1023                         value
1024                         line-breaking))
1025         (setq attributes (delq 'digit-value attributes))
1026         )
1027       (when (and (memq 'numeric-value attributes)
1028                  (setq value (get-char-attribute char 'numeric-value)))
1029         (insert (format "(numeric-value\t . %S)%s"
1030                         value
1031                         line-breaking))
1032         (setq attributes (delq 'numeric-value attributes))
1033         )
1034       )
1035      (t
1036       (when (and (memq 'digit-value attributes)
1037                  (setq value (get-char-attribute char 'digit-value)))
1038         (insert (format "(digit-value\t. %S)%s"
1039                         value
1040                         line-breaking))
1041         (setq attributes (delq 'digit-value attributes))
1042         )
1043       (when (and (memq 'numeric-value attributes)
1044                  (setq value (get-char-attribute char 'numeric-value)))
1045         (insert (format "(numeric-value\t. %S)%s"
1046                         value
1047                         line-breaking))
1048         (setq attributes (delq 'numeric-value attributes))
1049         )))
1050     (when (and (memq 'iso-10646-comment attributes)
1051                (setq value (get-char-attribute char 'iso-10646-comment)))
1052       (insert (format "(iso-10646-comment\t. %S)%s"
1053                       value
1054                       line-breaking))
1055       (setq attributes (delq 'iso-10646-comment attributes))
1056       )
1057     (when (and (memq 'morohashi-daikanwa attributes)
1058                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1059       (insert (format "(morohashi-daikanwa\t%s)%s"
1060                       (mapconcat (function prin1-to-string) value " ")
1061                       line-breaking))
1062       (setq attributes (delq 'morohashi-daikanwa attributes))
1063       )
1064     (setq radical nil
1065           strokes nil)
1066     (when (and (memq 'ideographic-radical attributes)
1067                (setq value (get-char-attribute char 'ideographic-radical)))
1068       (setq radical value)
1069       (insert (format "(ideographic-radical . %S)\t; %c%s"
1070                       radical
1071                       (ideographic-radical radical)
1072                       line-breaking))
1073       (setq attributes (delq 'ideographic-radical attributes))
1074       )
1075     (when (and (memq 'shuowen-radical attributes)
1076                (setq value (get-char-attribute char 'shuowen-radical)))
1077       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1078                       value
1079                       (shuowen-radical value)
1080                       line-breaking))
1081       (setq attributes (delq 'shuowen-radical attributes))
1082       )
1083     (let (key)
1084       (dolist (domain
1085                (append
1086                 char-db-feature-domains
1087                 (let (dest domain)
1088                   (dolist (feature (char-attribute-list))
1089                     (setq feature (symbol-name feature))
1090                     (when (string-match
1091                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1092                            feature)
1093                       (setq domain (intern (match-string 2 feature)))
1094                      (unless (memq domain dest)
1095                        (setq dest (cons domain dest)))))
1096                   (sort dest #'string<))))
1097         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1098         (when (and (memq key attributes)
1099                    (setq value (get-char-attribute char key)))
1100           (setq radical value)
1101           (insert (format "(%s . %S)\t; %c%s"
1102                           key
1103                           radical
1104                           (ideographic-radical radical)
1105                           line-breaking))
1106           (setq attributes (delq key attributes))
1107           )
1108         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1109         (when (and (memq key attributes)
1110                    (setq value (get-char-attribute char key)))
1111           (setq strokes value)
1112           (insert (format "(%s . %S)%s"
1113                           key
1114                           strokes
1115                           line-breaking))
1116           (setq attributes (delq key attributes))
1117           )
1118         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1119         (when (and (memq key attributes)
1120                    (setq value (get-char-attribute char key)))
1121           (insert (format "(%s       . %S)%s"
1122                           key
1123                           value
1124                           line-breaking))
1125           (setq attributes (delq key attributes))
1126           )
1127         (dolist (feature '(ideographic-radical
1128                            ideographic-strokes
1129                            total-strokes))
1130           (setq key (intern (format "%s@%s*sources" feature domain)))
1131           (when (and (memq key attributes)
1132                      (setq value (get-char-attribute char key)))
1133             (insert (format "(%s%s" key line-breaking))
1134             (dolist (cell value)
1135               (insert (format " %s" cell)))
1136             (insert ")")
1137             (insert line-breaking)
1138             (setq attributes (delq key attributes))
1139             ))
1140         ))
1141     (when (and (memq 'ideographic-strokes attributes)
1142                (setq value (get-char-attribute char 'ideographic-strokes)))
1143       (setq strokes value)
1144       (insert (format "(ideographic-strokes . %S)%s"
1145                       strokes
1146                       line-breaking))
1147       (setq attributes (delq 'ideographic-strokes attributes))
1148       )
1149     (when (and (memq 'kangxi-radical attributes)
1150                (setq value (get-char-attribute char 'kangxi-radical)))
1151       (unless (eq value radical)
1152         (insert (format "(kangxi-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 'kangxi-radical attributes))
1159       )
1160     (when (and (memq 'kangxi-strokes attributes)
1161                (setq value (get-char-attribute char 'kangxi-strokes)))
1162       (unless (eq value strokes)
1163         (insert (format "(kangxi-strokes\t . %S)%s"
1164                         value
1165                         line-breaking))
1166         (or strokes
1167             (setq strokes value)))
1168       (setq attributes (delq 'kangxi-strokes attributes))
1169       )
1170     ;; (when (and (memq 'japanese-radical attributes)
1171     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1172     ;;   (unless (eq value radical)
1173     ;;     (insert (format "(japanese-radical\t . %S)\t; %c%s"
1174     ;;                     value
1175     ;;                     (ideographic-radical value)
1176     ;;                     line-breaking))
1177     ;;     (or radical
1178     ;;         (setq radical value)))
1179     ;;   (setq attributes (delq 'japanese-radical attributes))
1180     ;;   )
1181     (when (and (memq 'japanese-strokes attributes)
1182                (setq value (get-char-attribute char 'japanese-strokes)))
1183       (unless (eq value strokes)
1184         (insert (format "(japanese-strokes\t . %S)%s"
1185                         value
1186                         line-breaking))
1187         (or strokes
1188             (setq strokes value)))
1189       (setq attributes (delq 'japanese-strokes attributes))
1190       )
1191     (when (and (memq 'cns-radical attributes)
1192                (setq value (get-char-attribute char 'cns-radical)))
1193       (insert (format "(cns-radical\t . %S)\t; %c%s"
1194                       value
1195                       (ideographic-radical value)
1196                       line-breaking))
1197       (setq attributes (delq 'cns-radical attributes))
1198       )
1199     (when (and (memq 'cns-strokes attributes)
1200                (setq value (get-char-attribute char 'cns-strokes)))
1201       (unless (eq value strokes)
1202         (insert (format "(cns-strokes\t . %S)%s"
1203                         value
1204                         line-breaking))
1205         (or strokes
1206             (setq strokes value)))
1207       (setq attributes (delq 'cns-strokes attributes))
1208       )
1209     ;; (when (and (memq 'shinjigen-1-radical attributes)
1210     ;;            (setq value (get-char-attribute char 'shinjigen-1-radical)))
1211     ;;   (unless (eq value radical)
1212     ;;     (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
1213     ;;                     value
1214     ;;                     (ideographic-radical value)
1215     ;;                     line-breaking))
1216     ;;     (or radical
1217     ;;         (setq radical value)))
1218     ;;   (setq attributes (delq 'shinjigen-1-radical attributes))
1219     ;;   )
1220     ;; (when (and (memq 'ideographic- attributes)
1221     ;;            (setq value (get-char-attribute char 'ideographic-)))
1222     ;;   (insert "(ideographic-       ")
1223     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1224     ;;         separator nil)
1225     ;;   (while (consp value)
1226     ;;     (setq cell (car value))
1227     ;;     (if (integerp cell)
1228     ;;         (setq cell (decode-char '=ucs cell)))
1229     ;;     (cond ((characterp cell)
1230     ;;            (if separator
1231     ;;                (insert lbs))
1232     ;;            (if readable
1233     ;;                (insert (format "%S" cell))
1234     ;;              (char-db-insert-char-spec cell readable))
1235     ;;            (setq separator lbs))
1236     ;;           ((consp cell)
1237     ;;            (if separator
1238     ;;                (insert lbs))
1239     ;;            (if (consp (car cell))
1240     ;;                (char-db-insert-char-spec cell readable)
1241     ;;              (char-db-insert-char-reference cell readable))
1242     ;;            (setq separator lbs))
1243     ;;           (t
1244     ;;            (if separator
1245     ;;                (insert separator))
1246     ;;            (insert (prin1-to-string cell))
1247     ;;            (setq separator " ")))
1248     ;;     (setq value (cdr value)))
1249     ;;   (insert ")")
1250     ;;   (insert line-breaking)
1251     ;;   (setq attributes (delq 'ideographic- attributes)))
1252     (when (and (memq 'total-strokes attributes)
1253                (setq value (get-char-attribute char 'total-strokes)))
1254       (insert (format "(total-strokes       . %S)%s"
1255                       value
1256                       line-breaking))
1257       (setq attributes (delq 'total-strokes attributes))
1258       )
1259     ;; (when (and (memq '->ideograph attributes)
1260     ;;            (setq value (get-char-attribute char '->ideograph)))
1261     ;;   (insert (format "(->ideograph\t%s)%s"
1262     ;;                   (mapconcat (lambda (code)
1263     ;;                                (cond ((symbolp code)
1264     ;;                                       (symbol-name code))
1265     ;;                                      ((integerp code)
1266     ;;                                       (format "#x%04X" code))
1267     ;;                                      (t
1268     ;;                                       (format "%s %S"
1269     ;;                                               line-breaking code))))
1270     ;;                              value " ")
1271     ;;                   line-breaking))
1272     ;;   (setq attributes (delq '->ideograph attributes))
1273     ;;   )
1274     ;; (when (and (memq '->decomposition attributes)
1275     ;;            (setq value (get-char-attribute char '->decomposition)))
1276     ;;   (insert (format "(->decomposition\t%s)%s"
1277     ;;                   (mapconcat (lambda (code)
1278     ;;                                (cond ((symbolp code)
1279     ;;                                       (symbol-name code))
1280     ;;                                      ((characterp code)
1281     ;;                                       (if readable
1282     ;;                                           (format "%S" code)
1283     ;;                                         (format "#x%04X"
1284     ;;                                                 (char-int code))
1285     ;;                                         ))
1286     ;;                                      ((integerp code)
1287     ;;                                       (format "#x%04X" code))
1288     ;;                                      (t
1289     ;;                                       (format "%s%S" line-breaking code))))
1290     ;;                              value " ")
1291     ;;                   line-breaking))
1292     ;;   (setq attributes (delq '->decomposition attributes))
1293     ;;   )
1294     (if (equal (get-char-attribute char '->titlecase)
1295                (get-char-attribute char '->uppercase))
1296         (setq attributes (delq '->titlecase attributes)))
1297     (when (and (memq '->mojikyo attributes)
1298                (setq value (get-char-attribute char '->mojikyo)))
1299       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1300                       value (decode-char 'mojikyo value)
1301                       line-breaking))
1302       (setq attributes (delq '->mojikyo attributes))
1303       )
1304     ;; (when (and (memq 'hanyu-dazidian-vol attributes)
1305     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1306     ;;   (insert (format "(hanyu-dazidian-vol  . %d)%s"
1307     ;;                   value line-breaking))
1308     ;;   (setq attributes (delq 'hanyu-dazidian-vol attributes))
1309     ;;   )
1310     ;; (when (and (memq 'hanyu-dazidian-page attributes)
1311     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1312     ;;   (insert (format "(hanyu-dazidian-page . %d)%s"
1313     ;;                   value line-breaking))
1314     ;;   (setq attributes (delq 'hanyu-dazidian-page attributes))
1315     ;;   )
1316     ;; (when (and (memq 'hanyu-dazidian-char attributes)
1317     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1318     ;;   (insert (format "(hanyu-dazidian-char . %d)%s"
1319     ;;                   value line-breaking))
1320     ;;   (setq attributes (delq 'hanyu-dazidian-char attributes))
1321     ;;   )
1322     (unless readable
1323       (dolist (ignored '(composition
1324                          ->denotational <-subsumptive ->ucs-unified
1325                          ->ideographic-component-forms))
1326         (setq attributes (delq ignored attributes))))
1327     (while attributes
1328       (setq name (car attributes))
1329       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1330                   'value-is-empty)
1331         (cond ((setq ret (find-charset name))
1332                (setq name (charset-name ret))
1333                (when (not (memq name dest-ccss))
1334                  (setq dest-ccss (cons name dest-ccss))
1335                  (char-db-insert-ccs-feature name value line-breaking))
1336                )
1337               ((string-match "^=>ucs@" (symbol-name name))
1338                (insert (format "(%-18s . #x%04X)\t; %c%s"
1339                                name value (decode-char '=ucs value)
1340                                line-breaking))
1341                )
1342               ((eq name 'jisx0208-1978/4X)
1343                (insert (format "(%-18s . #x%04X)%s"
1344                                name value
1345                                line-breaking))
1346                )
1347               ((and
1348                 (not readable)
1349                 (not (eq name '->subsumptive))
1350                 (not (eq name '->uppercase))
1351                 (not (eq name '->lowercase))
1352                 (not (eq name '->titlecase))
1353                 (not (eq name '->canonical))
1354                 (not (eq name '->Bopomofo))
1355                 (not (eq name '->mistakable))
1356                 (not (eq name '->ideographic-variants))
1357                 (null (get-char-attribute
1358                        char (intern (format "%s*sources" name))))
1359                 (not (string-match "\\*sources$" (symbol-name name)))
1360                 (null (get-char-attribute
1361                        char (intern (format "%s*note" name))))
1362                 (not (string-match "\\*note$" (symbol-name name)))
1363                 (or (eq name '<-identical)
1364                     (eq name '<-uppercase)
1365                     (eq name '<-lowercase)
1366                     (eq name '<-titlecase)
1367                     (eq name '<-canonical)
1368                     (eq name '<-ideographic-variants)
1369                     ;; (eq name '<-synonyms)
1370                     (string-match "^<-synonyms" (symbol-name name))
1371                     (eq name '<-mistakable)
1372                     (when (string-match "^->" (symbol-name name))
1373                       (cond
1374                        ((string-match "^->fullwidth" (symbol-name name))
1375                         (not (and (consp value)
1376                                   (characterp (car value))
1377                                   (encode-char
1378                                    (car value) '=ucs 'defined-only)))
1379                         )
1380                        (t)))
1381                     ))
1382                )
1383               ((or (eq name 'ideographic-structure)
1384                    (eq name 'ideographic-combination)
1385                    ;; (eq name 'ideographic-)
1386                    (eq name '=decomposition)
1387                    (char-feature-base-name= '=decomposition name)
1388                    (char-feature-base-name= '=>decomposition name)
1389                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1390                    ;;               (symbol-name name))
1391                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1392                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1393                                  (symbol-name name))
1394                    )
1395                (char-db-insert-relation-feature char name value
1396                                                 line-breaking
1397                                                 ccss readable))
1398               ;; ((memq name '(ideograph=
1399               ;;               original-ideograph-of
1400               ;;               ancient-ideograph-of
1401               ;;               vulgar-ideograph-of
1402               ;;               wrong-ideograph-of
1403               ;;               ;; simplified-ideograph-of
1404               ;;               ideographic-variants
1405               ;;               ;; ideographic-different-form-of
1406               ;;               ))
1407               ;;  (insert (format "(%-18s%s " name line-breaking))
1408               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1409               ;;        separator nil)
1410               ;;  (while (consp value)
1411               ;;    (setq cell (car value))
1412               ;;    (if (and (consp cell)
1413               ;;             (consp (car cell)))
1414               ;;        (progn
1415               ;;          (if separator
1416               ;;              (insert lbs))
1417               ;;          (char-db-insert-alist cell readable)
1418               ;;          (setq separator lbs))
1419               ;;      (if separator
1420               ;;          (insert separator))
1421               ;;      (insert (prin1-to-string cell))
1422               ;;      (setq separator " "))
1423               ;;    (setq value (cdr value)))
1424               ;;  (insert ")")
1425               ;;  (insert line-breaking))
1426               ((consp value)
1427                (insert (format "(%-18s " name))
1428                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1429                      separator nil)
1430                (while (consp value)
1431                  (setq cell (car value))
1432                  (if (and (consp cell)
1433                           (consp (car cell))
1434                           (setq ret (condition-case nil
1435                                         (find-char cell)
1436                                       (error nil))))
1437                      (progn
1438                        (setq rest cell
1439                              al nil
1440                              cal nil)
1441                        (while rest
1442                          (setq key (car (car rest)))
1443                          (if (find-charset key)
1444                              (setq cal (cons key cal))
1445                            (setq al (cons key al)))
1446                          (setq rest (cdr rest)))
1447                        (if separator
1448                            (insert lbs))
1449                        (insert-char-attributes ret
1450                                                readable
1451                                                al ; cal
1452                                                nil 'for-sub-node)
1453                        (setq separator lbs))
1454                    (setq ret (prin1-to-string cell))
1455                    (if separator
1456                        (if (< (+ (current-column)
1457                                  (length ret)
1458                                  (length separator))
1459                               76)
1460                            (insert separator)
1461                          (insert lbs)))
1462                    (insert ret)
1463                    (setq separator " "))
1464                  (setq value (cdr value)))
1465                (insert ")")
1466                (insert line-breaking))
1467               (t
1468                (insert (format "(%-18s" name))
1469                (setq ret (prin1-to-string value))
1470                (unless (< (+ (current-column)
1471                              (length ret)
1472                              3)
1473                           76)
1474                  (insert line-breaking))
1475                (insert " . " ret ")" line-breaking)
1476                ;; (insert (format "(%-18s . %S)%s"
1477                ;;                 name value
1478                ;;                 line-breaking))
1479                )
1480               ))
1481       (setq attributes (cdr attributes)))
1482     (insert ")")))
1483
1484 (defun insert-char-data (char &optional readable
1485                               attributes)
1486   (save-restriction
1487     (narrow-to-region (point)(point))
1488     (insert "(define-char
1489   '")
1490     (insert-char-attributes char readable attributes)
1491     (insert ")\n")
1492     (goto-char (point-min))
1493     (while (re-search-forward "[ \t]+$" nil t)
1494       (replace-match ""))
1495     ;; from tabify.
1496     (goto-char (point-min))
1497     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1498       (let ((column (current-column))
1499             (indent-tabs-mode t))
1500         (delete-region (match-beginning 0) (point))
1501         (indent-to column)))
1502     (goto-char (point-max))
1503     ;; (tabify (point-min)(point-max))
1504     ))
1505
1506 (defun insert-char-data-with-variant (char &optional printable
1507                                            no-ucs-unified
1508                                            script excluded-script)
1509   (insert-char-data char printable)
1510   (let ((variants (char-variants char))
1511         rest
1512         variant vs ret)
1513     (setq variants (sort variants #'<))
1514     (setq rest variants)
1515     (setq variants (cons char variants))
1516     (while rest
1517       (setq variant (car rest))
1518       (unless (get-char-attribute variant '<-subsumptive)
1519         (if (and (or (null script)
1520                      (null (setq vs (get-char-attribute variant 'script)))
1521                      (memq script vs))
1522                  (or (null excluded-script)
1523                      (null (setq vs (get-char-attribute variant 'script)))
1524                      (not (memq excluded-script vs))))
1525             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1526               (insert-char-data variant printable)
1527               (if (setq ret (char-variants variant))
1528                   (while ret
1529                     (or (memq (car ret) variants)
1530                         ;; (get-char-attribute (car ret) '<-subsumptive)
1531                         (setq rest (nconc rest (list (car ret)))))
1532                     (setq ret (cdr ret)))))))
1533       (setq rest (cdr rest)))))
1534
1535 (defun insert-char-range-data (min max &optional script excluded-script)
1536   (let ((code min)
1537         char)
1538     (while (<= code max)
1539       (setq char (decode-char '=ucs code))
1540       (if (encode-char char '=ucs 'defined-only)
1541           (insert-char-data-with-variant char nil 'no-ucs-unified
1542                                          script excluded-script))
1543       (setq code (1+ code)))))
1544
1545 (defun write-char-range-data-to-file (min max file
1546                                           &optional script excluded-script)
1547   (let ((coding-system-for-write char-db-file-coding-system))
1548     (with-temp-buffer
1549       (insert (format ";; -*- coding: %s -*-\n"
1550                       char-db-file-coding-system))
1551       (insert-char-range-data min max script excluded-script)
1552       (write-region (point-min)(point-max) file))))
1553
1554 (defvar what-character-original-window-configuration)
1555
1556 ;;;###autoload
1557 (defun what-char-definition (char)
1558   (interactive (list (char-after)))
1559   (let ((buf (get-buffer-create "*Character Description*"))
1560         (the-buf (current-buffer))
1561         (win-conf (current-window-configuration)))
1562     (pop-to-buffer buf)
1563     (make-local-variable 'what-character-original-window-configuration)
1564     (setq what-character-original-window-configuration win-conf)
1565     (setq buffer-read-only nil)
1566     (erase-buffer)
1567     (condition-case err
1568         (progn
1569           (insert-char-data-with-variant char 'printable)
1570           (unless (char-attribute-alist char)
1571             (insert (format ";; = %c\n"
1572                             (let* ((rest (split-char char))
1573                                    (ccs (pop rest))
1574                                    (code (pop rest)))
1575                               (while rest
1576                                 (setq code (logior (lsh code 8)
1577                                                    (pop rest))))
1578                               (decode-char ccs code)))))
1579           ;; (char-db-update-comment)
1580           (set-buffer-modified-p nil)
1581           (view-mode the-buf (lambda (buf)
1582                                (set-window-configuration
1583                                 what-character-original-window-configuration)
1584                                ))
1585           (goto-char (point-min)))
1586       (error (progn
1587                (set-window-configuration
1588                 what-character-original-window-configuration)
1589                (signal (car err) (cdr err)))))))
1590
1591
1592 ;;; @ end
1593 ;;;
1594
1595 (provide 'char-db-util)
1596
1597 ;;; char-db-util.el ends here