git.chise.org Git - chise/xemacs-chise.git.1/blob - lisp/utf-2000/char-db-util.el

   1 ;;; char-db-util.el --- Character Database utility -*- coding: utf-8-er; -*-
   2
   3 ;; Copyright (C) 1998,1999,2000,2001,2002,2003,2004,2005,2006,2007,
   4 ;;   2008,2009,2010,2011,2012,2013,2014,2015,2016 MORIOKA Tomohiko.
   5
   6 ;; Author: MORIOKA Tomohiko <tomo@kanji.zinbun.kyoto-u.ac.jp>
   7 ;; Keywords: CHISE, Character Database, ISO/IEC 10646, UCS, Unicode, MULE.
   8
   9 ;; This file is part of XEmacs CHISE.
  10
  11 ;; XEmacs CHISE is free software; you can redistribute it and/or
  12 ;; modify it under the terms of the GNU General Public License as
  13 ;; published by the Free Software Foundation; either version 2, or (at
  14 ;; your option) any later version.
  15
  16 ;; XEmacs CHISE is distributed in the hope that it will be useful,
  17 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  18 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  19 ;; General Public License for more details.
  20
  21 ;; You should have received a copy of the GNU General Public License
  22 ;; along with XEmacs CHISE; see the file COPYING.  If not, write to
  23 ;; the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  24 ;; Boston, MA 02111-1307, USA.
  25
  26 ;;; Code:
  27
  28 (require 'chise-subr)
  29 (require 'ideograph-subr)
  30
  31 (defconst unidata-normative-category-alist
  32   '(("Lu" letter        uppercase)
  33     ("Ll" letter        lowercase)
  34     ("Lt" letter        titlecase)
  35     ("Mn" mark          non-spacing)
  36     ("Mc" mark          spacing-combining)
  37     ("Me" mark          enclosing)
  38     ("Nd" number        decimal-digit)
  39     ("Nl" number        letter)
  40     ("No" number        other)
  41     ("Zs" separator     space)
  42     ("Zl" separator     line)
  43     ("Zp" separator     paragraph)
  44     ("Cc" other         control)
  45     ("Cf" other         format)
  46     ("Cs" other         surrogate)
  47     ("Co" other         private-use)
  48     ("Cn" other         not-assigned)))
  49
  50 (defconst unidata-informative-category-alist
  51   '(("Lm" letter        modifier)
  52     ("Lo" letter        other)
  53     ("Pc" punctuation   connector)
  54     ("Pd" punctuation   dash)
  55     ("Ps" punctuation   open)
  56     ("Pe" punctuation   close)
  57     ("Pi" punctuation   initial-quote)
  58     ("Pf" punctuation   final-quote)
  59     ("Po" punctuation   other)
  60     ("Sm" symbol        math)
  61     ("Sc" symbol        currency)
  62     ("Sk" symbol        modifier)
  63     ("So" symbol        other)
  64     ))
  65
  66 (defconst shuowen-radicals
  67   [?一 ?上 ?示 ?三 ?王 ?玉 ?玨 ?气 ?士 ?丨 ; 010
  68    ?屮 ?艸 ?蓐 ?茻 ?小 ?八 ?釆 ?半 ?牛 ?犛 ; 020
  69    ?告 ?口 ?凵 ?吅 ?哭 ?走 ?止 ?癶 ?步 ?此 ; 030
  70    ?正 ?是 ?辵 ?彳 ?廴 ?㢟 ?行 ?齒 ?牙 ?足 ; 040
  71    ?疋 ?品 ?龠 ?冊 ?㗊 ?舌 ?干 ?谷 ?只 ?㕯 ; 050
  72    ?句 ?丩 ?古 ?十 ?卅 ?言 ?誩 ?音 ?䇂 ?丵 ; 060
  73    ?菐 ?𠬞 ?𠬜 ?共 ?異 ?舁 ?𦥑 ?䢅 ?爨 ?革 ; 070
  74    ?鬲 ?䰜 ?爪 ?𠃨 ?鬥 ?又 ?𠂇 ?㕜 ?支 ?𦘒 ; 080
  75    ?聿 ?畫 ?隶 ?臤 ?臣 ?殳 ?殺 ?𠘧 ?寸 ?皮 ; 090
  76    ?㼱 ?攴 ?敎 ?卜 ?用 ?爻 ?㸚 ?𥄎 ?目 ?䀠 ; 100
  77    ?眉 ?盾 ?自 ?白 ?鼻 ?皕 ?習 ?羽 ?隹 ?奞 ; 110
  78    ?萑 ?𦫳 ?苜 ?羊 ?羴 ?瞿 ?雔 ?雥 ?鳥 ?烏 ; 120
  79    ?𠦒 ?冓 ?幺 ?𢆶 ?叀 ?玄 ?予 ?放 ?𠬪 ?𣦼 ; 130
  80    ?歺 ?死 ?冎 ?骨 ?肉 ?筋 ?刀 ?刃 ?㓞 ?丰 ; 140
  81    ?耒 ?𧢲 ?竹 ?箕 ?丌 ?左 ?工 ?㠭 ?巫 ?甘 ; 150
  82    ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?旨 ?喜 ?壴 ; 160
  83 ;  ?旨 ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?喜 ?壴 ; 160
  84    ?鼓 ?豈 ?豆 ?豊 ?豐 ?䖒 ?虍 ?虎 ?虤 ?皿 ; 170
  85    ?𠙴 ?去 ?血 ?丶 ?丹 ?青 ?井 ?皀 ?鬯 ?食 ; 180
  86    ?亼 ?會 ?倉 ?入 ?缶 ?矢 ?高 ?冂 ?𩫏 ?京 ; 190
  87    ?亯 ?𣆪 ?畗 ?㐭 ?嗇 ?來 ?麥 ?夊 ?舛 ?䑞 ; 200
  88    ?韋 ?弟 ?夂 ?久 ?桀 ?木 ?東 ?林 ?才 ?叒 ; 210
  89    ?之 ?帀 ?出 ?𣎵 ?生 ?乇 ?𠂹 ?𠌶 ?華 ?𥝌 ; 220
  90    ?稽 ?巢 ?桼 ?束 ?㯻 ?囗 ?員 ?貝 ?邑 ?𨛜 ; 230
  91    ?日 ?旦 ?倝 ?㫃 ?冥 ?晶 ?月 ?有 ?明 ?囧 ; 240
  92    ?夕 ?多 ?毌 ?𢎘 ?𣐺 ?卣 ?齊 ?朿 ?片 ?鼎 ; 250
  93    ?克 ?彔 ?禾 ?秝 ?黍 ?香 ?米 ?毇 ?臼 ?凶 ; 260
  94    ?𣎳 ?林 ?麻 ?尗 ?耑 ?韭 ?瓜 ?瓠 ?宀 ?宮 ; 270
  95    ?呂 ?穴 ?㝱 ?𤕫 ?冖 ?𠔼 ?冃 ?㒳 ?网 ?襾 ; 280
  96    ?巾 ?巿 ?帛 ?白 ?㡀 ?黹 ?人 ?𠤎 ?匕 ?从 ; 290
  97    ?比 ?北 ?丘 ?㐺 ?𡈼 ?重 ?臥 ?身 ?㐆 ?衣 ; 300
  98    ?裘 ?老 ?毛 ?毳 ?尸 ?尺 ?尾 ?履 ?舟 ?方 ; 310
  99    ?儿 ?兄 ?兂 ?皃 ?𠑹 ?先 ?秃 ?見 ?覞 ?欠 ; 320
 100    ?㱃 ?㳄 ?旡 ?頁 ?𦣻 ?面 ?丏 ?首 ?𥄉 ?須 ; 330
 101    ?彡 ?彣 ?文 ?髟 ?后 ?司 ?卮 ?卩 ?印 ?色 ; 340
 102    ?𠨍 ?辟 ?勹 ?包 ?茍 ?鬼 ?甶 ?厶 ?嵬 ?山 ; 350
 103    ?屾 ?屵 ?广 ?厂 ?丸 ?危 ?石 ?長 ?勿 ?冄 ; 360
 104    ?而 ?豕 ?㣇 ?彑 ?豚 ?豸 ?𤉡 ?易 ?象 ?馬 ; 370
 105    ?𢊁 ?鹿 ?麤 ?㲋 ?兔 ?萈 ?犬 ?㹜 ?鼠 ?能 ; 380
 106    ?熊 ?火 ?炎 ?黑 ?囪 ?焱 ?炙 ?赤 ?大 ?亦 ; 390
 107    ])
 108
 109 (defun shuowen-radical (number)
 110   (aref shuowen-radicals (1- number)))
 111
 112 (defvar char-db-file-coding-system 'utf-8-mcs-er)
 113
 114 (defvar char-db-ignored-attributes '(ideographic-products))
 115
 116 (defvar char-db-coded-charset-priority-list
 117   '(ascii
 118     control-1
 119     latin-iso8859-1
 120     latin-iso8859-2
 121     latin-iso8859-3
 122     latin-iso8859-4
 123     latin-iso8859-9
 124     latin-jisx0201
 125     cyrillic-iso8859-5
 126     greek-iso8859-7
 127     thai-tis620
 128     ;; =mj
 129     ;; =adobe-japan1-0
 130     ;; =adobe-japan1-1
 131     ;; =adobe-japan1-2
 132     ;; =adobe-japan1-3
 133     ;; =adobe-japan1-4
 134     ;; =adobe-japan1-5
 135     ;; =adobe-japan1-6
 136     =jis-x0208
 137     =jis-x0208@1978
 138     =jis-x0208@1983
 139     japanese-jisx0212
 140     chinese-gb2312
 141     =jis-x0208@1990
 142     chinese-cns11643-1
 143     chinese-cns11643-2
 144     chinese-cns11643-3
 145     chinese-cns11643-4
 146     chinese-cns11643-5
 147     chinese-cns11643-6
 148     chinese-cns11643-7
 149     =jis-x0213-1
 150     =jis-x0213-1@2000
 151     =jis-x0213-1@2004
 152     =jis-x0213-2
 153     korean-ksc5601
 154     chinese-isoir165
 155     katakana-jisx0201
 156     hebrew-iso8859-8
 157     chinese-gb12345
 158     latin-viscii
 159     ethiopic-ucs
 160     =big5-cdp
 161     =hanyo-denshi/ja
 162     =hanyo-denshi/jb
 163     =hanyo-denshi/jc
 164     =hanyo-denshi/jd
 165     =hanyo-denshi/ft
 166     =hanyo-denshi/ia
 167     =hanyo-denshi/ib
 168     =hanyo-denshi/hg
 169     =hanyo-denshi/jt
 170     =hanyo-denshi/ks
 171     =hanyo-denshi/tk
 172     ;; ==mj
 173     ;; ==adobe-japan1-0
 174     ;; ==adobe-japan1-1
 175     ;; ==adobe-japan1-2
 176     ;; ==adobe-japan1-3
 177     ;; ==adobe-japan1-4
 178     ;; ==adobe-japan1-5
 179     ;; ==adobe-japan1-6
 180     ==jis-x0208
 181     ==jis-x0213-1
 182     ==jis-x0213-2
 183     ==jis-x0212
 184     ==hanyo-denshi/ja
 185     ==hanyo-denshi/jb
 186     ==hanyo-denshi/jc
 187     ==hanyo-denshi/ft
 188     ==hanyo-denshi/ib
 189     ==hanyo-denshi/hg
 190     ==hanyo-denshi/jt
 191     ==hanyo-denshi/ks
 192     =gt
 193     =gt-k
 194     =daikanwa
 195     =daikanwa@rev2
 196     =daikanwa@rev1
 197     =daikanwa/+p
 198     ==daikanwa
 199     =cbeta
 200     ideograph-hanziku-1
 201     ideograph-hanziku-2
 202     ideograph-hanziku-3
 203     ideograph-hanziku-4
 204     ideograph-hanziku-5
 205     ideograph-hanziku-6
 206     ideograph-hanziku-7
 207     ideograph-hanziku-8
 208     ideograph-hanziku-9
 209     ideograph-hanziku-10
 210     ideograph-hanziku-11
 211     ideograph-hanziku-12
 212     ;; =>>>adobe-japan1-0
 213     ;; =>>>adobe-japan1-1
 214     ;; =>>>adobe-japan1-2
 215     ;; =>>>adobe-japan1-3
 216     ;; =>>>adobe-japan1-4
 217     ;; =>>>adobe-japan1-5
 218     ;; =>>>adobe-japan1-6
 219     ;; =>>>jis-x0208
 220     ;; =>>>jis-x0213-1
 221     ;; =>>>jis-x0213-2
 222     ;; =>>>hanyo-denshi/ja
 223     ;; =>>>hanyo-denshi/jb
 224     ;; =>>>hanyo-denshi/jc
 225     ;; =>>>hanyo-denshi/ft
 226     ;; =>>>hanyo-denshi/ib
 227     ;; =>>>hanyo-denshi/hg
 228     ;; =>>>hanyo-denshi/jt
 229     ;; =>>>hanyo-denshi/ks
 230     ;; =>>>gt
 231     =>>adobe-japan1-0
 232     =>>adobe-japan1-1
 233     =>>adobe-japan1-2
 234     =>>adobe-japan1-3
 235     =>>adobe-japan1-4
 236     =>>adobe-japan1-5
 237     =>>adobe-japan1-6
 238     =>>jis-x0208
 239     =>>jis-x0213-1
 240     =>>jis-x0213-1@2000
 241     =>>jis-x0213-1@2004
 242     =>>jis-x0213-2
 243     =>>jis-x0208@1978
 244     =>>hanyo-denshi/ft
 245     =>>hanyo-denshi/jt
 246     =>>hanyo-denshi/ks
 247     =>>gt
 248     =>>daikanwa
 249     =>>cbeta
 250     =+>jis-x0208
 251     =+>jis-x0213-1
 252     =+>jis-x0213-2
 253     =+>adobe-japan1-0
 254     =+>adobe-japan1-1
 255     =+>adobe-japan1-2
 256     =+>adobe-japan1-3
 257     =+>adobe-japan1-4
 258     =+>adobe-japan1-5
 259     =+>adobe-japan1-6
 260     =+>jis-x0208@1978
 261     =>jis-x0208
 262     =>jis-x0208@1997
 263     =>jis-x0213-1
 264     =>jis-x0213-1@2000
 265     =>jis-x0213-1@2004
 266     =>jis-x0213-2
 267     ==>ucs@bucs
 268     =>iwds-1
 269     =>ucs@hanyo-denshi
 270     =>ucs@iso
 271     =>ucs@unicode
 272     =>ucs@jis
 273     =>ucs@cns
 274     =>ucs@ks
 275     =+>ucs@iso
 276     =+>ucs@unicode
 277     =+>ucs@jis
 278     =+>ucs@jis/1990
 279     =+>ucs@cns
 280     =+>ucs@ks
 281     =>>ucs@iso
 282     =>>ucs@unicode
 283     =>>ucs@jis
 284     =>>ucs@cns
 285     =>>>ucs@iso
 286     =>>>ucs@unicode
 287     ==ucs@iso
 288     ==ucs@unicode
 289     ;; ==ucs@cns
 290     ==gb2312
 291     ==ks-x1001
 292     ==cns11643-1
 293     ==cns11643-2
 294     ==cns11643-3
 295     ==cns11643-4
 296     ==cns11643-5
 297     ==cns11643-6
 298     ==cns11643-7
 299     ==gt
 300     ==jis-x0208@1990
 301     ;; ==jis-x0208@1983
 302     ==jis-x0208@1978
 303     ==gt-k
 304     =ucs@iso
 305     =ucs@unicode
 306     =ucs@cns
 307     ==big5-cdp
 308     ==cbeta
 309     =>>big5-cdp
 310     =>>gt-k
 311     =+>gt
 312     =>gt
 313     =>mj
 314     =>big5-cdp
 315     =>daikanwa
 316     =>daikanwa/ho
 317     =>cns11643-5
 318     =>cns11643-7
 319     =big5
 320     =big5-eten
 321     =>gt-k
 322     =zinbun-oracle
 323     =>zinbun-oracle
 324     =ruimoku-v6
 325     =>>ruimoku-v6
 326     ==ruimoku-v6
 327     =jef-china3
 328     =>cbeta
 329     =shinjigen
 330     =ucs-var-001
 331     =ucs-var-002
 332     =ucs-var-003
 333     =ucs-var-004
 334     =ucs-var-006
 335     =ucs-var-008
 336     =ucs-var-010
 337     =ucs-itaiji-001
 338     =ucs-itaiji-002
 339     =ucs-itaiji-003
 340     =ucs-itaiji-004
 341     =ucs-itaiji-005
 342     =ucs-itaiji-006
 343     =ucs-itaiji-008
 344     =ucs-itaiji-009
 345     =>big5-cdp-itaiji-001
 346     =big5-cdp-itaiji-001
 347     =big5-cdp-itaiji-002
 348     =big5-cdp-itaiji-003
 349     =big5-cdp-var-003
 350     =big5-cdp-var-004
 351     =big5-cdp-var-005
 352     =>ucs@iwds-1
 353     ==ucs@gb
 354     =ucs@JP/hanazono))
 355
 356
 357 ;;; @ char-db formatters
 358 ;;;
 359
 360 (defun char-db-make-char-spec (char)
 361   (let (ret char-spec)
 362     (cond ((characterp char)
 363            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 364                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 365                   (setq char-spec (list (cons '=ucs ret)))
 366                   (cond ((setq ret (get-char-attribute char 'name))
 367                          (setq char-spec (cons (cons 'name ret) char-spec))
 368                          )
 369                         ((setq ret (get-char-attribute char 'name*))
 370                          (setq char-spec (cons (cons 'name* ret) char-spec))
 371                          ))
 372                   )
 373                  ((encode-char char '=mj 'defined-only)
 374                   (setq char-spec nil)
 375                   (dolist (ccs (charset-list))
 376                     (if (and (or (eq ccs '=mj)
 377                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 378                                  (string-match "=ucs@" (symbol-name ccs))
 379                                  )
 380                              (setq ccs (charset-name ccs))
 381                              (null (assq ccs char-spec))
 382                              (setq ret (encode-char char ccs 'defined-only)))
 383                         (setq char-spec (cons (cons ccs ret) char-spec))))
 384                   )
 385                  ((encode-char char '==mj 'defined-only)
 386                   (setq char-spec nil)
 387                   (dolist (ccs (charset-list))
 388                     (if (and (or (eq ccs '==mj)
 389                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 390                                  (string-match "=ucs@" (symbol-name ccs))
 391                                  )
 392                              (setq ccs (charset-name ccs))
 393                              (null (assq ccs char-spec))
 394                              (setq ret (encode-char char ccs 'defined-only)))
 395                         (setq char-spec (cons (cons ccs ret) char-spec))))
 396                   )
 397                  ((encode-char char '=adobe-japan1 'defined-only)
 398                   (setq char-spec nil)
 399                   (dolist (ccs (charset-list))
 400                     (if (and (or (memq ccs
 401                                        '(=adobe-japan1-0
 402                                          =adobe-japan1-1
 403                                          =adobe-japan1-2
 404                                          =adobe-japan1-3
 405                                          =adobe-japan1-4
 406                                          =adobe-japan1-5
 407                                          =adobe-japan1-6
 408                                          ))
 409                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 410                                  (string-match "=ucs@" (symbol-name ccs))
 411                                  )
 412                              (setq ccs (charset-name ccs))
 413                              (null (assq ccs char-spec))
 414                              (setq ret (encode-char char ccs 'defined-only)))
 415                         (setq char-spec (cons (cons ccs ret) char-spec))))
 416                   )
 417                  ((encode-char char '==adobe-japan1 'defined-only)
 418                   (setq char-spec nil)
 419                   (dolist (ccs (charset-list))
 420                     (if (and (or (memq ccs
 421                                        '(==adobe-japan1-0
 422                                          ==adobe-japan1-1
 423                                          ==adobe-japan1-2
 424                                          ==adobe-japan1-3
 425                                          ==adobe-japan1-4
 426                                          ==adobe-japan1-5
 427                                          ==adobe-japan1-6
 428                                          ))
 429                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 430                                  (string-match "=ucs@" (symbol-name ccs))
 431                                  )
 432                              (setq ccs (charset-name ccs))
 433                              (null (assq ccs char-spec))
 434                              (setq ret (encode-char char ccs 'defined-only)))
 435                         (setq char-spec (cons (cons ccs ret) char-spec))))
 436                   )
 437                  ((setq ret
 438                         (catch 'tag
 439                           (let ((rest char-db-coded-charset-priority-list)
 440                                 ccs)
 441                             (while rest
 442                               (setq ccs (charset-name
 443                                          (find-charset (car rest))))
 444                               (if (setq ret
 445                                         (encode-char char ccs
 446                                                      'defined-only))
 447                                   (throw 'tag (cons ccs ret)))
 448                               (setq rest (cdr rest))))))
 449                   (setq char-spec (list ret))
 450                   (dolist (ccs (delq (car ret) (charset-list)))
 451                     (if (and (or (charset-iso-final-char ccs)
 452                                  (memq ccs
 453                                        '(=daikanwa
 454                                          =daikanwa@rev2
 455                                          ;; =gt-k
 456                                          =jis-x0208@1997
 457                                          ))
 458                                  (eq (charset-property ccs 'iso-ir) 177)
 459                                  ;; (string-match "=ucs@" (symbol-name ccs))
 460                                  )
 461                              (setq ccs (charset-name ccs))
 462                              (null (assq ccs char-spec))
 463                              (setq ret (encode-char char ccs 'defined-only)))
 464                         (setq char-spec (cons (cons ccs ret) char-spec))))
 465                   (if (null char-spec)
 466                       (setq char-spec (split-char char)))
 467                   (cond ((setq ret (get-char-attribute char 'name))
 468                          (setq char-spec (cons (cons 'name ret) char-spec))
 469                          )
 470                         ((setq ret (get-char-attribute char 'name*))
 471                          (setq char-spec (cons (cons 'name* ret) char-spec))
 472                          ))
 473                   )
 474                  ((setq ret (get-char-attribute
 475                              char 'ideographic-combination))
 476                   (setq char-spec
 477                         (cons (cons 'ideographic-combination ret)
 478                               char-spec))
 479                   ))
 480            char-spec)
 481           ((consp char)
 482            char))))
 483
 484 (defun char-db-insert-char-spec (char &optional readable column
 485                                       required-features)
 486   (unless column
 487     (setq column (current-column)))
 488   (let (char-spec temp-char)
 489     (setq char-spec (char-db-make-char-spec char))
 490     (unless (or (characterp char) ; char
 491                 (condition-case nil
 492                     (setq char (find-char char-spec))
 493                   (error nil)))
 494       ;; define temporary character
 495       ;;   Current implementation is dirty.
 496       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 497                                          char-spec)))
 498       (remove-char-attribute temp-char 'ideograph-daikanwa)
 499       (setq char temp-char))
 500     (insert-char-attributes char
 501                             readable
 502                             (union (mapcar #'car char-spec)
 503                                    required-features)
 504                             nil 'for-sub-node)
 505     (when temp-char
 506       ;; undefine temporary character
 507       ;;   Current implementation is dirty.
 508       (setq char-spec (char-attribute-alist temp-char))
 509       (while char-spec
 510         (remove-char-attribute temp-char (car (car char-spec)))
 511         (setq char-spec (cdr char-spec))))))
 512
 513 (defun char-db-insert-alist (alist &optional readable column)
 514   (unless column
 515     (setq column (current-column)))
 516   (let ((line-breaking
 517          (concat "\n" (make-string (1+ column) ?\ )))
 518         name value
 519         ret al ; cal
 520         key
 521         lbs cell rest separator)
 522     (insert "(")
 523     (while alist
 524       (setq name (car (car alist))
 525             value (cdr (car alist)))
 526       (cond ((eq name 'char)
 527              (insert "(char . ")
 528              (if (setq ret (condition-case nil
 529                                (find-char value)
 530                              (error nil)))
 531                  (progn
 532                    (setq al nil
 533                          ;; cal nil
 534                          )
 535                    (while value
 536                      (setq key (car (car value)))
 537                      ;; (if (find-charset key)
 538                      ;;     (setq cal (cons key cal))
 539                      (setq al (cons key al))
 540                      ;; )
 541                      (setq value (cdr value)))
 542                    (insert-char-attributes ret
 543                                            readable
 544                                            (or al 'none) ; cal
 545                                            nil 'for-sub-node))
 546                (insert (prin1-to-string value)))
 547              (insert ")")
 548              (insert line-breaking))
 549             ((consp value)
 550              (insert (format "(%-18s " name))
 551              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 552              (while (consp value)
 553                (setq cell (car value))
 554                (if (and (consp cell)
 555                         (consp (car cell))
 556                         (setq ret (condition-case nil
 557                                       (find-char cell)
 558                                     (error nil)))
 559                         )
 560                    (progn
 561                      (setq rest cell
 562                            al nil
 563                            ;; cal nil
 564                            )
 565                      (while rest
 566                        (setq key (car (car rest)))
 567                        ;; (if (find-charset key)
 568                        ;;     (setq cal (cons key cal))
 569                        (setq al (cons key al))
 570                        ;; )
 571                        (setq rest (cdr rest)))
 572                      (if separator
 573                          (insert lbs))
 574                      (insert-char-attributes ret
 575                                              readable
 576                                              al ; cal
 577                                              nil 'for-sub-node)
 578                      (setq separator lbs))
 579                  (if separator
 580                      (insert separator))
 581                  (insert (prin1-to-string cell))
 582                  (setq separator " "))
 583                (setq value (cdr value)))
 584              (insert ")")
 585              (insert line-breaking))
 586             (t
 587              (insert (format "(%-18s . %S)%s"
 588                              name value
 589                              line-breaking))))
 590       (setq alist (cdr alist))))
 591   (insert ")"))
 592
 593 (defun char-db-insert-char-reference (plist &optional readable column)
 594   (unless column
 595     (setq column (current-column)))
 596   (let ((line-breaking
 597          (concat "\n" (make-string (1+ column) ?\ )))
 598         (separator "")
 599         name value)
 600     (insert "(")
 601     (while plist
 602       (setq name (pop plist))
 603       (setq value (pop plist))
 604       (cond ((eq name :char)
 605              (insert separator)
 606              (insert ":char\t")
 607              (cond ((numberp value)
 608                     (setq value (decode-char '=ucs value)))
 609                    ;; ((consp value)
 610                    ;;  (setq value (or (find-char value)
 611                    ;;                  value)))
 612                    )
 613              (char-db-insert-char-spec value readable)
 614              (insert line-breaking)
 615              (setq separator ""))
 616             ((eq name :radical)
 617              (insert (format "%s%s\t%d ; %c%s"
 618                              separator
 619                              name value
 620                              (ideographic-radical value)
 621                              line-breaking))
 622              (setq separator ""))
 623             (t
 624              (insert (format "%s%s\t%S" separator name value))
 625              (setq separator line-breaking)))
 626       ))
 627   (insert ")"))
 628
 629 (defun char-db-decode-isolated-char (ccs code-point)
 630   (let (ret)
 631     (setq ret
 632           (cond ((eq ccs 'arabic-iso8859-6)
 633                  (decode-char ccs code-point))
 634                 ;; ((eq ccs '=gt)
 635                 ;;  (decode-builtin-char '==gt code-point))
 636                 ((and (memq ccs '(=gt-pj-1
 637                                   =gt-pj-2
 638                                   =gt-pj-3
 639                                   =gt-pj-4
 640                                   =gt-pj-5
 641                                   =gt-pj-6
 642                                   =gt-pj-7
 643                                   =gt-pj-8
 644                                   =gt-pj-9
 645                                   =gt-pj-10
 646                                   =gt-pj-11))
 647                       (setq ret (decode-char ccs code-point))
 648                       (setq ret (encode-char ret '=gt 'defined-only)))
 649                  (decode-builtin-char '=gt ret))
 650                 (t
 651                  (decode-builtin-char ccs code-point))))
 652     (cond ((null ret)
 653            (or (decode-char ccs code-point)
 654                (define-char (list (cons ccs code-point)))))
 655           ((and (<= 0 (char-int ret))
 656                 (<= (char-int ret) #x1F))
 657            (decode-char '=ucs (+ #x2400 (char-int ret))))
 658           ((= (char-int ret) #x7F)
 659            ?\u2421)
 660           (t ret))))
 661
 662 (defvar char-db-convert-obsolete-format t)
 663
 664 (defun char-db-insert-ccs-feature (name value line-breaking)
 665   (cond
 666    ((integerp value)
 667     (insert
 668      (format
 669       (cond
 670        ((memq name '(=>iwds-1
 671                      ==shinjigen
 672                      =shinjigen
 673                      =shinjigen@1ed ==shinjigen@1ed
 674                      =shinjigen@rev ==shinjigen@rev
 675                      =shinjigen/+p@rev ==shinjigen/+p@rev
 676                      ===daikanwa/ho ==daikanwa/ho
 677                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 678         "(%-18s .  %04d)\t; %c")
 679        ((eq name '=shinjigen@1ed/24pr)
 680         "(%-18s . %04d)\t; %c")
 681        ((or
 682          (memq name
 683                '(===daikanwa
 684                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 685                  =daikanwa@rev1 =daikanwa@rev2
 686                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 687                  =>>daikanwa/+p
 688                  =daikanwa/+2p =>>daikanwa/+2p
 689                  =gt ==gt ===gt
 690                  =>>gt =+>gt =>gt
 691                  =gt-k ==gt-k ===gt-k
 692                  =>>gt-k =>gt-k
 693                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 694                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 695                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 696                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 697                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 698                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 699                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 700                  =>>adobe-japan1-0 =+>adobe-japan1-0
 701                  =>>adobe-japan1-1 =+>adobe-japan1-1
 702                  =>>adobe-japan1-2 =+>adobe-japan1-2
 703                  =>>adobe-japan1-3 =+>adobe-japan1-3
 704                  =>>adobe-japan1-4 =+>adobe-japan1-4
 705                  =>>adobe-japan1-5 =+>adobe-japan1-5
 706                  =>>adobe-japan1-6 =+>adobe-japan1-6
 707                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 708                  =zinbun-oracle =>zinbun-oracle
 709                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 710                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 711                  ===hng-khi ===hng-khm ===hng-hok ===hng-kyd ===hng-sok
 712                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 713                  ===hng-sys ===hng-tsu ===hng-tzj
 714                  ===hng-hos ===hng-nak ===hng-jhk
 715                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 716                  ===hng-nkm ===hng-k24 ===hng-nkk
 717                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 718                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 719                  ===hng-okd ===hng-wad ===hng-kmi
 720                  ===hng-zkd ===hng-doh ===hng-jyu
 721                  ===hng-tzs ===hng-kss ===hng-kyo
 722                  ===hng-smk))
 723          ;; (string-match "^=adobe-" (symbol-name name))
 724          )
 725         "(%-18s . %05d)\t; %c")
 726        ((memq name '(=hanyo-denshi/ks
 727                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 728                      =>>hanyo-denshi/ks
 729                      =koseki ==koseki
 730                      =mj ==mj ===mj =>>mj =>mj
 731                      =zihai mojikyo))
 732         "(%-18s . %06d)\t; %c")
 733        ((memq name '(=hanyo-denshi/tk ==hanyo-denshi/tk))
 734         "(%-18s . %08d)\t; %c")
 735        ((>= (charset-dimension name) 2)
 736         "(%-18s . #x%04X)\t; %c")
 737        (t
 738         "(%-18s . #x%02X)\t; %c"))
 739       name
 740       (if (= (charset-iso-graphic-plane name) 1)
 741           (logior value
 742                   (cond ((= (charset-dimension name) 1)
 743                          #x80)
 744                         ((= (charset-dimension name) 2)
 745                          #x8080)
 746                         ((= (charset-dimension name) 3)
 747                          #x808080)
 748                         (t 0)))
 749         value)
 750       (char-db-decode-isolated-char name value)))
 751     (if (and (= (charset-chars name) 94)
 752              (= (charset-dimension name) 2))
 753         (insert (format " [%02d-%02d]"
 754                         (- (lsh value -8) 32)
 755                         (- (logand value 255) 32))))
 756     )
 757    (t
 758     (insert (format "(%-18s . %s)" name value))
 759     ))
 760   (insert line-breaking))
 761
 762 (defun char-db-insert-relation-feature (char name value line-breaking
 763                                              ccss readable)
 764   (insert (format "(%-18s%s " name line-breaking))
 765   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 766         separator cell sources required-features
 767         ret)
 768     (while (consp value)
 769       (setq cell (car value))
 770       (if (integerp cell)
 771           (setq cell (decode-char '=ucs cell)))
 772       (cond
 773        ((eq name '->subsumptive)
 774         (when (or (not (some (lambda (atr)
 775                                (get-char-attribute cell atr))
 776                              char-db-ignored-attributes))
 777                   (some (lambda (ccs)
 778                           (encode-char cell ccs 'defined-only))
 779                         ccss))
 780           (if separator
 781               (insert lbs))
 782           (let ((char-db-ignored-attributes
 783                  (cons '<-subsumptive
 784                        char-db-ignored-attributes)))
 785             (insert-char-attributes cell readable nil nil 'for-sub-node))
 786           (setq separator lbs))
 787         )
 788        ((characterp cell)
 789         (setq sources
 790               (get-char-attribute
 791                char (intern (format "%s*sources" name))))
 792         (setq required-features nil)
 793         (dolist (source sources)
 794           (cond
 795            ((memq source '(JP
 796                            JP/Jouyou
 797                            shinjigen shinjigen@1ed shinjigen@rev))
 798             (setq required-features
 799                   (union required-features
 800                          '(=jis-x0208
 801                            =jis-x0208@1990
 802                            =jis-x0213-1@2000
 803                            =jis-x0213-1@2004
 804                            =jis-x0213-2
 805                            =jis-x0212
 806                            =jis-x0208@1983
 807                            =jis-x0208@1978
 808                            =shinjigen
 809                            =shinjigen@1ed
 810                            =shinjigen@rev
 811                            =shinjigen/+p@rev))))
 812            ((eq source 'CN)
 813             (setq required-features
 814                   (union required-features
 815                          '(=gb2312
 816                            =gb12345
 817                            =iso-ir165)))))
 818           (cond
 819            ((find-charset (setq ret (intern (format "=%s" source))))
 820             (setq required-features
 821                   (cons ret required-features)))
 822            (t (setq required-features
 823                     (cons source required-features)))))
 824         (cond ((string-match "@JP" (symbol-name name))
 825                (setq required-features
 826                      (union required-features
 827                             '(=jis-x0208
 828                               =jis-x0208@1990
 829                               =jis-x0213-1-2000
 830                               =jis-x0213-2-2000
 831                               =jis-x0212
 832                               =jis-x0208@1983
 833                               =jis-x0208@1978))))
 834               ((string-match "@CN" (symbol-name name))
 835                (setq required-features
 836                      (union required-features
 837                             '(=gb2312
 838                               =gb12345
 839                               =iso-ir165)))))
 840         (if separator
 841             (insert lbs))
 842         (if readable
 843             (insert (format "%S" cell))
 844           (char-db-insert-char-spec cell readable
 845                                     nil
 846                                     required-features))
 847         (setq separator lbs))
 848        ((consp cell)
 849         (if separator
 850             (insert lbs))
 851         (if (consp (car cell))
 852             (char-db-insert-char-spec cell readable)
 853           (char-db-insert-char-reference cell readable))
 854         (setq separator lbs))
 855        (t
 856         (if separator
 857             (insert separator))
 858         (insert (prin1-to-string cell))
 859         (setq separator " ")))
 860       (setq value (cdr value)))
 861     (insert ")")
 862     (insert line-breaking)))
 863
 864 (defun insert-char-attributes (char &optional readable attributes column
 865                                     for-sub-node)
 866   (unless column
 867     (setq column (current-column)))
 868   (let (name value ; has-long-ccs-name
 869         rest
 870         radical strokes
 871         (line-breaking
 872          (concat "\n" (make-string (1+ column) ?\ )))
 873         lbs cell separator ret
 874         key al cal
 875         dest-ccss ; sources required-features
 876         ccss)
 877     (let (atr-d)
 878       (setq attributes
 879             (sort (if attributes
 880                       (if (consp attributes)
 881                           (progn
 882                             (dolist (name attributes)
 883                               (unless (memq name char-db-ignored-attributes)
 884                                 (if (find-charset name)
 885                                     (push name ccss))
 886                                 (push name atr-d)))
 887                             atr-d))
 888                     (dolist (name (char-attribute-list))
 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                   #'char-attribute-name<)))
 895     (insert "(")
 896     (when (memq '<-subsumptive attributes)
 897       (when (or readable (not for-sub-node))
 898         (when (setq value (get-char-attribute char '<-subsumptive))
 899           (char-db-insert-relation-feature char '<-subsumptive value
 900                                            line-breaking
 901                                            ccss readable)))
 902       (setq attributes (delq '<-subsumptive attributes)))
 903     (when (and (memq '<-denotational attributes)
 904                (setq value (get-char-attribute char '<-denotational)))
 905       (char-db-insert-relation-feature char '<-denotational value
 906                                        line-breaking
 907                                        ccss readable)
 908       (setq attributes (delq '<-denotational attributes)))
 909     (when (and (memq '<-denotational@component attributes)
 910                (setq value (get-char-attribute char '<-denotational@component)))
 911       (char-db-insert-relation-feature char '<-denotational@component value
 912                                        line-breaking
 913                                        ccss readable)
 914       (setq attributes (delq '<-denotational@component attributes)))
 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 'name* attributes)
 925                (setq value (get-char-attribute char 'name*)))
 926       (insert (format
 927                (if (> (+ (current-column) (length value)) 48)
 928                    "(name* . %S)%s"
 929                  "(name*              . %S)%s")
 930                value line-breaking))
 931       (setq attributes (delq 'name* attributes))
 932       )
 933     (when (and (memq 'script attributes)
 934                (setq value (get-char-attribute char 'script)))
 935       (insert (format "(script\t\t%s)%s"
 936                       (mapconcat (function prin1-to-string)
 937                                  value " ")
 938                       line-breaking))
 939       (setq attributes (delq 'script attributes))
 940       )
 941     (dolist (name '(=>ucs =>ucs*))
 942       (when (and (memq name attributes)
 943                  (setq value (get-char-attribute char name)))
 944         (insert (format "(%-18s . #x%04X)\t; %c%s"
 945                         name value (decode-char '=ucs value)
 946                         line-breaking))
 947         (setq attributes (delq name attributes))))
 948     (dolist (name '(=>ucs@gb =>ucs@big5))
 949       (when (and (memq name attributes)
 950                  (setq value (get-char-attribute char name)))
 951         (insert (format "(%-18s . #x%04X)\t; %c%s"
 952                         name value
 953                         (decode-char (intern
 954                                       (concat "="
 955                                               (substring
 956                                                (symbol-name name) 2)))
 957                                      value)
 958                         line-breaking))
 959         (setq attributes (delq name attributes))
 960         ))
 961     ;; (dolist (name '(=>daikanwa))
 962     ;;   (when (and (memq name attributes)
 963     ;;              (setq value (get-char-attribute char name)))
 964     ;;     (insert
 965     ;;      (if (integerp value)
 966     ;;          (format "(%-18s . %05d)\t; %c%s"
 967     ;;                  name value (decode-char '=daikanwa value)
 968     ;;                  line-breaking)
 969     ;;        (format "(%-18s %s)\t; %c%s"
 970     ;;                name
 971     ;;                (mapconcat (function prin1-to-string)
 972     ;;                           value " ")
 973     ;;                (char-representative-of-daikanwa char)
 974     ;;                line-breaking)))
 975     ;;     (setq attributes (delq name attributes))))
 976     (when (and (memq 'general-category attributes)
 977                (setq value (get-char-attribute char 'general-category)))
 978       (insert (format
 979                "(general-category\t%s) ; %s%s"
 980                (mapconcat (lambda (cell)
 981                             (format "%S" cell))
 982                           value " ")
 983                (cond ((rassoc value unidata-normative-category-alist)
 984                       "Normative Category")
 985                      ((rassoc value unidata-informative-category-alist)
 986                       "Informative Category")
 987                      (t
 988                       "Unknown Category"))
 989                line-breaking))
 990       (setq attributes (delq 'general-category attributes))
 991       )
 992     (when (and (memq 'bidi-category attributes)
 993                (setq value (get-char-attribute char 'bidi-category)))
 994       (insert (format "(bidi-category\t. %S)%s"
 995                       value
 996                       line-breaking))
 997       (setq attributes (delq 'bidi-category attributes))
 998       )
 999     (unless (or (not (memq 'mirrored attributes))
1000                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
1001                     'empty))
1002       (insert (format "(mirrored\t\t. %S)%s"
1003                       value
1004                       line-breaking))
1005       (setq attributes (delq 'mirrored attributes))
1006       )
1007     (cond
1008      ((and (memq 'decimal-digit-value attributes)
1009            (setq value (get-char-attribute char 'decimal-digit-value)))
1010       (insert (format "(decimal-digit-value . %S)%s"
1011                       value
1012                       line-breaking))
1013       (setq attributes (delq 'decimal-digit-value attributes))
1014       (when (and (memq 'digit-value attributes)
1015                  (setq value (get-char-attribute char 'digit-value)))
1016         (insert (format "(digit-value\t . %S)%s"
1017                         value
1018                         line-breaking))
1019         (setq attributes (delq 'digit-value attributes))
1020         )
1021       (when (and (memq 'numeric-value attributes)
1022                  (setq value (get-char-attribute char 'numeric-value)))
1023         (insert (format "(numeric-value\t . %S)%s"
1024                         value
1025                         line-breaking))
1026         (setq attributes (delq 'numeric-value attributes))
1027         )
1028       )
1029      (t
1030       (when (and (memq 'digit-value attributes)
1031                  (setq value (get-char-attribute char 'digit-value)))
1032         (insert (format "(digit-value\t. %S)%s"
1033                         value
1034                         line-breaking))
1035         (setq attributes (delq 'digit-value attributes))
1036         )
1037       (when (and (memq 'numeric-value attributes)
1038                  (setq value (get-char-attribute char 'numeric-value)))
1039         (insert (format "(numeric-value\t. %S)%s"
1040                         value
1041                         line-breaking))
1042         (setq attributes (delq 'numeric-value attributes))
1043         )))
1044     (when (and (memq 'iso-10646-comment attributes)
1045                (setq value (get-char-attribute char 'iso-10646-comment)))
1046       (insert (format "(iso-10646-comment\t. %S)%s"
1047                       value
1048                       line-breaking))
1049       (setq attributes (delq 'iso-10646-comment attributes))
1050       )
1051     (when (and (memq 'morohashi-daikanwa attributes)
1052                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1053       (insert (format "(morohashi-daikanwa\t%s)%s"
1054                       (mapconcat (function prin1-to-string) value " ")
1055                       line-breaking))
1056       (setq attributes (delq 'morohashi-daikanwa attributes))
1057       )
1058     (setq radical nil
1059           strokes nil)
1060     (when (and (memq 'ideographic-radical attributes)
1061                (setq value (get-char-attribute char 'ideographic-radical)))
1062       (setq radical value)
1063       (insert (format "(ideographic-radical . %S)\t; %c%s"
1064                       radical
1065                       (ideographic-radical radical)
1066                       line-breaking))
1067       (setq attributes (delq 'ideographic-radical attributes))
1068       )
1069     (when (and (memq 'shuowen-radical attributes)
1070                (setq value (get-char-attribute char 'shuowen-radical)))
1071       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1072                       value
1073                       (shuowen-radical value)
1074                       line-breaking))
1075       (setq attributes (delq 'shuowen-radical attributes))
1076       )
1077     (let (key)
1078       (dolist (domain
1079                (append
1080                 char-db-feature-domains
1081                 (let (dest domain)
1082                   (dolist (feature (char-attribute-list))
1083                     (setq feature (symbol-name feature))
1084                     (when (string-match
1085                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1086                            feature)
1087                       (setq domain (intern (match-string 2 feature)))
1088                      (unless (memq domain dest)
1089                        (setq dest (cons domain dest)))))
1090                   (sort dest #'string<))))
1091         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1092         (when (and (memq key attributes)
1093                    (setq value (get-char-attribute char key)))
1094           (setq radical value)
1095           (insert (format "(%s . %S)\t; %c%s"
1096                           key
1097                           radical
1098                           (ideographic-radical radical)
1099                           line-breaking))
1100           (setq attributes (delq key attributes))
1101           )
1102         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1103         (when (and (memq key attributes)
1104                    (setq value (get-char-attribute char key)))
1105           (setq strokes value)
1106           (insert (format "(%s . %S)%s"
1107                           key
1108                           strokes
1109                           line-breaking))
1110           (setq attributes (delq key attributes))
1111           )
1112         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1113         (when (and (memq key attributes)
1114                    (setq value (get-char-attribute char key)))
1115           (insert (format "(%s       . %S)%s"
1116                           key
1117                           value
1118                           line-breaking))
1119           (setq attributes (delq key attributes))
1120           )
1121         (dolist (feature '(ideographic-radical
1122                            ideographic-strokes
1123                            total-strokes))
1124           (setq key (intern (format "%s@%s*sources" feature domain)))
1125           (when (and (memq key attributes)
1126                      (setq value (get-char-attribute char key)))
1127             (insert (format "(%s%s" key line-breaking))
1128             (dolist (cell value)
1129               (insert (format " %s" cell)))
1130             (insert ")")
1131             (insert line-breaking)
1132             (setq attributes (delq key attributes))
1133             ))
1134         ))
1135     (when (and (memq 'ideographic-strokes attributes)
1136                (setq value (get-char-attribute char 'ideographic-strokes)))
1137       (setq strokes value)
1138       (insert (format "(ideographic-strokes . %S)%s"
1139                       strokes
1140                       line-breaking))
1141       (setq attributes (delq 'ideographic-strokes attributes))
1142       )
1143     (when (and (memq 'kangxi-radical attributes)
1144                (setq value (get-char-attribute char 'kangxi-radical)))
1145       (unless (eq value radical)
1146         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1147                         value
1148                         (ideographic-radical value)
1149                         line-breaking))
1150         (or radical
1151             (setq radical value)))
1152       (setq attributes (delq 'kangxi-radical attributes))
1153       )
1154     (when (and (memq 'kangxi-strokes attributes)
1155                (setq value (get-char-attribute char 'kangxi-strokes)))
1156       (unless (eq value strokes)
1157         (insert (format "(kangxi-strokes\t . %S)%s"
1158                         value
1159                         line-breaking))
1160         (or strokes
1161             (setq strokes value)))
1162       (setq attributes (delq 'kangxi-strokes attributes))
1163       )
1164     ;; (when (and (memq 'japanese-radical attributes)
1165     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1166     ;;   (unless (eq value radical)
1167     ;;     (insert (format "(japanese-radical\t . %S)\t; %c%s"
1168     ;;                     value
1169     ;;                     (ideographic-radical value)
1170     ;;                     line-breaking))
1171     ;;     (or radical
1172     ;;         (setq radical value)))
1173     ;;   (setq attributes (delq 'japanese-radical attributes))
1174     ;;   )
1175     (when (and (memq 'japanese-strokes attributes)
1176                (setq value (get-char-attribute char 'japanese-strokes)))
1177       (unless (eq value strokes)
1178         (insert (format "(japanese-strokes\t . %S)%s"
1179                         value
1180                         line-breaking))
1181         (or strokes
1182             (setq strokes value)))
1183       (setq attributes (delq 'japanese-strokes attributes))
1184       )
1185     (when (and (memq 'cns-radical attributes)
1186                (setq value (get-char-attribute char 'cns-radical)))
1187       (insert (format "(cns-radical\t . %S)\t; %c%s"
1188                       value
1189                       (ideographic-radical value)
1190                       line-breaking))
1191       (setq attributes (delq 'cns-radical attributes))
1192       )
1193     (when (and (memq 'cns-strokes attributes)
1194                (setq value (get-char-attribute char 'cns-strokes)))
1195       (unless (eq value strokes)
1196         (insert (format "(cns-strokes\t . %S)%s"
1197                         value
1198                         line-breaking))
1199         (or strokes
1200             (setq strokes value)))
1201       (setq attributes (delq 'cns-strokes attributes))
1202       )
1203     ;; (when (and (memq 'shinjigen-1-radical attributes)
1204     ;;            (setq value (get-char-attribute char 'shinjigen-1-radical)))
1205     ;;   (unless (eq value radical)
1206     ;;     (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
1207     ;;                     value
1208     ;;                     (ideographic-radical value)
1209     ;;                     line-breaking))
1210     ;;     (or radical
1211     ;;         (setq radical value)))
1212     ;;   (setq attributes (delq 'shinjigen-1-radical attributes))
1213     ;;   )
1214     ;; (when (and (memq 'ideographic- attributes)
1215     ;;            (setq value (get-char-attribute char 'ideographic-)))
1216     ;;   (insert "(ideographic-       ")
1217     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1218     ;;         separator nil)
1219     ;;   (while (consp value)
1220     ;;     (setq cell (car value))
1221     ;;     (if (integerp cell)
1222     ;;         (setq cell (decode-char '=ucs cell)))
1223     ;;     (cond ((characterp cell)
1224     ;;            (if separator
1225     ;;                (insert lbs))
1226     ;;            (if readable
1227     ;;                (insert (format "%S" cell))
1228     ;;              (char-db-insert-char-spec cell readable))
1229     ;;            (setq separator lbs))
1230     ;;           ((consp cell)
1231     ;;            (if separator
1232     ;;                (insert lbs))
1233     ;;            (if (consp (car cell))
1234     ;;                (char-db-insert-char-spec cell readable)
1235     ;;              (char-db-insert-char-reference cell readable))
1236     ;;            (setq separator lbs))
1237     ;;           (t
1238     ;;            (if separator
1239     ;;                (insert separator))
1240     ;;            (insert (prin1-to-string cell))
1241     ;;            (setq separator " ")))
1242     ;;     (setq value (cdr value)))
1243     ;;   (insert ")")
1244     ;;   (insert line-breaking)
1245     ;;   (setq attributes (delq 'ideographic- attributes)))
1246     (when (and (memq 'total-strokes attributes)
1247                (setq value (get-char-attribute char 'total-strokes)))
1248       (insert (format "(total-strokes       . %S)%s"
1249                       value
1250                       line-breaking))
1251       (setq attributes (delq 'total-strokes attributes))
1252       )
1253     ;; (when (and (memq '->ideograph attributes)
1254     ;;            (setq value (get-char-attribute char '->ideograph)))
1255     ;;   (insert (format "(->ideograph\t%s)%s"
1256     ;;                   (mapconcat (lambda (code)
1257     ;;                                (cond ((symbolp code)
1258     ;;                                       (symbol-name code))
1259     ;;                                      ((integerp code)
1260     ;;                                       (format "#x%04X" code))
1261     ;;                                      (t
1262     ;;                                       (format "%s %S"
1263     ;;                                               line-breaking code))))
1264     ;;                              value " ")
1265     ;;                   line-breaking))
1266     ;;   (setq attributes (delq '->ideograph attributes))
1267     ;;   )
1268     ;; (when (and (memq '->decomposition attributes)
1269     ;;            (setq value (get-char-attribute char '->decomposition)))
1270     ;;   (insert (format "(->decomposition\t%s)%s"
1271     ;;                   (mapconcat (lambda (code)
1272     ;;                                (cond ((symbolp code)
1273     ;;                                       (symbol-name code))
1274     ;;                                      ((characterp code)
1275     ;;                                       (if readable
1276     ;;                                           (format "%S" code)
1277     ;;                                         (format "#x%04X"
1278     ;;                                                 (char-int code))
1279     ;;                                         ))
1280     ;;                                      ((integerp code)
1281     ;;                                       (format "#x%04X" code))
1282     ;;                                      (t
1283     ;;                                       (format "%s%S" line-breaking code))))
1284     ;;                              value " ")
1285     ;;                   line-breaking))
1286     ;;   (setq attributes (delq '->decomposition attributes))
1287     ;;   )
1288     (if (equal (get-char-attribute char '->titlecase)
1289                (get-char-attribute char '->uppercase))
1290         (setq attributes (delq '->titlecase attributes)))
1291     (when (and (memq '->mojikyo attributes)
1292                (setq value (get-char-attribute char '->mojikyo)))
1293       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1294                       value (decode-char 'mojikyo value)
1295                       line-breaking))
1296       (setq attributes (delq '->mojikyo attributes))
1297       )
1298     ;; (when (and (memq 'hanyu-dazidian-vol attributes)
1299     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1300     ;;   (insert (format "(hanyu-dazidian-vol  . %d)%s"
1301     ;;                   value line-breaking))
1302     ;;   (setq attributes (delq 'hanyu-dazidian-vol attributes))
1303     ;;   )
1304     ;; (when (and (memq 'hanyu-dazidian-page attributes)
1305     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1306     ;;   (insert (format "(hanyu-dazidian-page . %d)%s"
1307     ;;                   value line-breaking))
1308     ;;   (setq attributes (delq 'hanyu-dazidian-page attributes))
1309     ;;   )
1310     ;; (when (and (memq 'hanyu-dazidian-char attributes)
1311     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1312     ;;   (insert (format "(hanyu-dazidian-char . %d)%s"
1313     ;;                   value line-breaking))
1314     ;;   (setq attributes (delq 'hanyu-dazidian-char attributes))
1315     ;;   )
1316     (unless readable
1317       (dolist (ignored '(composition
1318                          ->denotational <-subsumptive ->ucs-unified
1319                          ->ideographic-component-forms))
1320         (setq attributes (delq ignored attributes))))
1321     (while attributes
1322       (setq name (car attributes))
1323       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1324                   'value-is-empty)
1325         (cond ((setq ret (find-charset name))
1326                (setq name (charset-name ret))
1327                (when (not (memq name dest-ccss))
1328                  (setq dest-ccss (cons name dest-ccss))
1329                  (char-db-insert-ccs-feature name value line-breaking))
1330                )
1331               ((string-match "^=>ucs@" (symbol-name name))
1332                (insert (format "(%-18s . #x%04X)\t; %c%s"
1333                                name value (decode-char '=ucs value)
1334                                line-breaking))
1335                )
1336               ((eq name 'jisx0208-1978/4X)
1337                (insert (format "(%-18s . #x%04X)%s"
1338                                name value
1339                                line-breaking))
1340                )
1341               ((and
1342                 (not readable)
1343                 (not (eq name '->subsumptive))
1344                 (not (eq name '->uppercase))
1345                 (not (eq name '->lowercase))
1346                 (not (eq name '->titlecase))
1347                 (not (eq name '->canonical))
1348                 (not (eq name '->Bopomofo))
1349                 (not (eq name '->mistakable))
1350                 (not (eq name '->ideographic-variants))
1351                 (null (get-char-attribute
1352                        char (intern (format "%s*sources" name))))
1353                 (not (string-match "\\*sources$" (symbol-name name)))
1354                 (null (get-char-attribute
1355                        char (intern (format "%s*note" name))))
1356                 (not (string-match "\\*note$" (symbol-name name)))
1357                 (or (eq name '<-identical)
1358                     (eq name '<-uppercase)
1359                     (eq name '<-lowercase)
1360                     (eq name '<-titlecase)
1361                     (eq name '<-canonical)
1362                     (eq name '<-ideographic-variants)
1363                     ;; (eq name '<-synonyms)
1364                     (string-match "^<-synonyms" (symbol-name name))
1365                     (eq name '<-mistakable)
1366                     (when (string-match "^->" (symbol-name name))
1367                       (cond
1368                        ((string-match "^->fullwidth" (symbol-name name))
1369                         (not (and (consp value)
1370                                   (characterp (car value))
1371                                   (encode-char
1372                                    (car value) '=ucs 'defined-only)))
1373                         )
1374                        (t)))
1375                     ))
1376                )
1377               ((or (eq name 'ideographic-structure)
1378                    (eq name 'ideographic-combination)
1379                    ;; (eq name 'ideographic-)
1380                    (eq name '=decomposition)
1381                    (char-feature-base-name= '=decomposition name)
1382                    (char-feature-base-name= '=>decomposition name)
1383                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1384                    ;;               (symbol-name name))
1385                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1386                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1387                                  (symbol-name name))
1388                    )
1389                (char-db-insert-relation-feature char name value
1390                                                 line-breaking
1391                                                 ccss readable))
1392               ;; ((memq name '(ideograph=
1393               ;;               original-ideograph-of
1394               ;;               ancient-ideograph-of
1395               ;;               vulgar-ideograph-of
1396               ;;               wrong-ideograph-of
1397               ;;               ;; simplified-ideograph-of
1398               ;;               ideographic-variants
1399               ;;               ;; ideographic-different-form-of
1400               ;;               ))
1401               ;;  (insert (format "(%-18s%s " name line-breaking))
1402               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1403               ;;        separator nil)
1404               ;;  (while (consp value)
1405               ;;    (setq cell (car value))
1406               ;;    (if (and (consp cell)
1407               ;;             (consp (car cell)))
1408               ;;        (progn
1409               ;;          (if separator
1410               ;;              (insert lbs))
1411               ;;          (char-db-insert-alist cell readable)
1412               ;;          (setq separator lbs))
1413               ;;      (if separator
1414               ;;          (insert separator))
1415               ;;      (insert (prin1-to-string cell))
1416               ;;      (setq separator " "))
1417               ;;    (setq value (cdr value)))
1418               ;;  (insert ")")
1419               ;;  (insert line-breaking))
1420               ((consp value)
1421                (insert (format "(%-18s " name))
1422                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1423                      separator nil)
1424                (while (consp value)
1425                  (setq cell (car value))
1426                  (if (and (consp cell)
1427                           (consp (car cell))
1428                           (setq ret (condition-case nil
1429                                         (find-char cell)
1430                                       (error nil))))
1431                      (progn
1432                        (setq rest cell
1433                              al nil
1434                              cal nil)
1435                        (while rest
1436                          (setq key (car (car rest)))
1437                          (if (find-charset key)
1438                              (setq cal (cons key cal))
1439                            (setq al (cons key al)))
1440                          (setq rest (cdr rest)))
1441                        (if separator
1442                            (insert lbs))
1443                        (insert-char-attributes ret
1444                                                readable
1445                                                al ; cal
1446                                                nil 'for-sub-node)
1447                        (setq separator lbs))
1448                    (setq ret (prin1-to-string cell))
1449                    (if separator
1450                        (if (< (+ (current-column)
1451                                  (length ret)
1452                                  (length separator))
1453                               76)
1454                            (insert separator)
1455                          (insert lbs)))
1456                    (insert ret)
1457                    (setq separator " "))
1458                  (setq value (cdr value)))
1459                (insert ")")
1460                (insert line-breaking))
1461               (t
1462                (insert (format "(%-18s" name))
1463                (setq ret (prin1-to-string value))
1464                (unless (< (+ (current-column)
1465                              (length ret)
1466                              3)
1467                           76)
1468                  (insert line-breaking))
1469                (insert " . " ret ")" line-breaking)
1470                ;; (insert (format "(%-18s . %S)%s"
1471                ;;                 name value
1472                ;;                 line-breaking))
1473                )
1474               ))
1475       (setq attributes (cdr attributes)))
1476     (insert ")")))
1477
1478 (defun insert-char-data (char &optional readable
1479                               attributes)
1480   (save-restriction
1481     (narrow-to-region (point)(point))
1482     (insert "(define-char
1483   '")
1484     (insert-char-attributes char readable attributes)
1485     (insert ")\n")
1486     (goto-char (point-min))
1487     (while (re-search-forward "[ \t]+$" nil t)
1488       (replace-match ""))
1489     ;; from tabify.
1490     (goto-char (point-min))
1491     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1492       (let ((column (current-column))
1493             (indent-tabs-mode t))
1494         (delete-region (match-beginning 0) (point))
1495         (indent-to column)))
1496     (goto-char (point-max))
1497     ;; (tabify (point-min)(point-max))
1498     ))
1499
1500 (defun insert-char-data-with-variant (char &optional printable
1501                                            no-ucs-unified
1502                                            script excluded-script)
1503   (insert-char-data char printable)
1504   (let ((variants (char-variants char))
1505         rest
1506         variant vs ret)
1507     (setq variants (sort variants #'<))
1508     (setq rest variants)
1509     (setq variants (cons char variants))
1510     (while rest
1511       (setq variant (car rest))
1512       (unless (get-char-attribute variant '<-subsumptive)
1513         (if (and (or (null script)
1514                      (null (setq vs (get-char-attribute variant 'script)))
1515                      (memq script vs))
1516                  (or (null excluded-script)
1517                      (null (setq vs (get-char-attribute variant 'script)))
1518                      (not (memq excluded-script vs))))
1519             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1520               (insert-char-data variant printable)
1521               (if (setq ret (char-variants variant))
1522                   (while ret
1523                     (or (memq (car ret) variants)
1524                         ;; (get-char-attribute (car ret) '<-subsumptive)
1525                         (setq rest (nconc rest (list (car ret)))))
1526                     (setq ret (cdr ret)))))))
1527       (setq rest (cdr rest)))))
1528
1529 (defun insert-char-range-data (min max &optional script excluded-script)
1530   (let ((code min)
1531         char)
1532     (while (<= code max)
1533       (setq char (decode-char '=ucs code))
1534       (if (encode-char char '=ucs 'defined-only)
1535           (insert-char-data-with-variant char nil 'no-ucs-unified
1536                                          script excluded-script))
1537       (setq code (1+ code)))))
1538
1539 (defun write-char-range-data-to-file (min max file
1540                                           &optional script excluded-script)
1541   (let ((coding-system-for-write char-db-file-coding-system))
1542     (with-temp-buffer
1543       (insert (format ";; -*- coding: %s -*-\n"
1544                       char-db-file-coding-system))
1545       (insert-char-range-data min max script excluded-script)
1546       (write-region (point-min)(point-max) file))))
1547
1548 (defvar what-character-original-window-configuration)
1549
1550 ;;;###autoload
1551 (defun what-char-definition (char)
1552   (interactive (list (char-after)))
1553   (let ((buf (get-buffer-create "*Character Description*"))
1554         (the-buf (current-buffer))
1555         (win-conf (current-window-configuration)))
1556     (pop-to-buffer buf)
1557     (make-local-variable 'what-character-original-window-configuration)
1558     (setq what-character-original-window-configuration win-conf)
1559     (setq buffer-read-only nil)
1560     (erase-buffer)
1561     (condition-case err
1562         (progn
1563           (insert-char-data-with-variant char 'printable)
1564           (unless (char-attribute-alist char)
1565             (insert (format ";; = %c\n"
1566                             (let* ((rest (split-char char))
1567                                    (ccs (pop rest))
1568                                    (code (pop rest)))
1569                               (while rest
1570                                 (setq code (logior (lsh code 8)
1571                                                    (pop rest))))
1572                               (decode-char ccs code)))))
1573           ;; (char-db-update-comment)
1574           (set-buffer-modified-p nil)
1575           (view-mode the-buf (lambda (buf)
1576                                (set-window-configuration
1577                                 what-character-original-window-configuration)
1578                                ))
1579           (goto-char (point-min)))
1580       (error (progn
1581                (set-window-configuration
1582                 what-character-original-window-configuration)
1583                (signal (car err) (cdr err)))))))
1584
1585
1586 ;;; @ end
1587 ;;;
1588
1589 (provide 'char-db-util)
1590
1591 ;;; char-db-util.el ends here