git.chise.org Git - chise/xemacs-chise.git-/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-var-003
 346     =big5-cdp-var-004
 347     =big5-cdp-var-005
 348     =>ucs@iwds-1
 349     ==ucs@gb
 350     =ucs@JP/hanazono))
 351
 352
 353 ;;; @ char-db formatters
 354 ;;;
 355
 356 (defun char-db-make-char-spec (char)
 357   (let (ret char-spec)
 358     (cond ((characterp char)
 359            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 360                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 361                   (setq char-spec (list (cons '=ucs ret)))
 362                   (cond ((setq ret (get-char-attribute char 'name))
 363                          (setq char-spec (cons (cons 'name ret) char-spec))
 364                          )
 365                         ((setq ret (get-char-attribute char 'name*))
 366                          (setq char-spec (cons (cons 'name* ret) char-spec))
 367                          ))
 368                   )
 369                  ((encode-char char '=mj 'defined-only)
 370                   (setq char-spec nil)
 371                   (dolist (ccs (charset-list))
 372                     (if (and (or (eq ccs '=mj)
 373                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 374                                  (string-match "=ucs@" (symbol-name ccs))
 375                                  )
 376                              (setq ccs (charset-name ccs))
 377                              (null (assq ccs char-spec))
 378                              (setq ret (encode-char char ccs 'defined-only)))
 379                         (setq char-spec (cons (cons ccs ret) char-spec))))
 380                   )
 381                  ((encode-char char '==mj 'defined-only)
 382                   (setq char-spec nil)
 383                   (dolist (ccs (charset-list))
 384                     (if (and (or (eq ccs '==mj)
 385                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 386                                  (string-match "=ucs@" (symbol-name ccs))
 387                                  )
 388                              (setq ccs (charset-name ccs))
 389                              (null (assq ccs char-spec))
 390                              (setq ret (encode-char char ccs 'defined-only)))
 391                         (setq char-spec (cons (cons ccs ret) char-spec))))
 392                   )
 393                  ((encode-char char '=adobe-japan1 'defined-only)
 394                   (setq char-spec nil)
 395                   (dolist (ccs (charset-list))
 396                     (if (and (or (memq ccs
 397                                        '(=adobe-japan1-0
 398                                          =adobe-japan1-1
 399                                          =adobe-japan1-2
 400                                          =adobe-japan1-3
 401                                          =adobe-japan1-4
 402                                          =adobe-japan1-5
 403                                          =adobe-japan1-6
 404                                          ))
 405                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 406                                  (string-match "=ucs@" (symbol-name ccs))
 407                                  )
 408                              (setq ccs (charset-name ccs))
 409                              (null (assq ccs char-spec))
 410                              (setq ret (encode-char char ccs 'defined-only)))
 411                         (setq char-spec (cons (cons ccs ret) char-spec))))
 412                   )
 413                  ((encode-char char '==adobe-japan1 'defined-only)
 414                   (setq char-spec nil)
 415                   (dolist (ccs (charset-list))
 416                     (if (and (or (memq ccs
 417                                        '(==adobe-japan1-0
 418                                          ==adobe-japan1-1
 419                                          ==adobe-japan1-2
 420                                          ==adobe-japan1-3
 421                                          ==adobe-japan1-4
 422                                          ==adobe-japan1-5
 423                                          ==adobe-japan1-6
 424                                          ))
 425                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 426                                  (string-match "=ucs@" (symbol-name ccs))
 427                                  )
 428                              (setq ccs (charset-name ccs))
 429                              (null (assq ccs char-spec))
 430                              (setq ret (encode-char char ccs 'defined-only)))
 431                         (setq char-spec (cons (cons ccs ret) char-spec))))
 432                   )
 433                  ((setq ret
 434                         (catch 'tag
 435                           (let ((rest char-db-coded-charset-priority-list)
 436                                 ccs)
 437                             (while rest
 438                               (setq ccs (charset-name
 439                                          (find-charset (car rest))))
 440                               (if (setq ret
 441                                         (encode-char char ccs
 442                                                      'defined-only))
 443                                   (throw 'tag (cons ccs ret)))
 444                               (setq rest (cdr rest))))))
 445                   (setq char-spec (list ret))
 446                   (dolist (ccs (delq (car ret) (charset-list)))
 447                     (if (and (or (charset-iso-final-char ccs)
 448                                  (memq ccs
 449                                        '(=daikanwa
 450                                          =daikanwa@rev2
 451                                          ;; =gt-k
 452                                          =jis-x0208@1997
 453                                          ))
 454                                  (eq (charset-property ccs 'iso-ir) 177)
 455                                  ;; (string-match "=ucs@" (symbol-name ccs))
 456                                  )
 457                              (setq ccs (charset-name ccs))
 458                              (null (assq ccs char-spec))
 459                              (setq ret (encode-char char ccs 'defined-only)))
 460                         (setq char-spec (cons (cons ccs ret) char-spec))))
 461                   (if (null char-spec)
 462                       (setq char-spec (split-char char)))
 463                   (cond ((setq ret (get-char-attribute char 'name))
 464                          (setq char-spec (cons (cons 'name ret) char-spec))
 465                          )
 466                         ((setq ret (get-char-attribute char 'name*))
 467                          (setq char-spec (cons (cons 'name* ret) char-spec))
 468                          ))
 469                   )
 470                  ((setq ret (get-char-attribute
 471                              char 'ideographic-combination))
 472                   (setq char-spec
 473                         (cons (cons 'ideographic-combination ret)
 474                               char-spec))
 475                   ))
 476            char-spec)
 477           ((consp char)
 478            char))))
 479
 480 (defun char-db-insert-char-spec (char &optional readable column
 481                                       required-features)
 482   (unless column
 483     (setq column (current-column)))
 484   (let (char-spec temp-char)
 485     (setq char-spec (char-db-make-char-spec char))
 486     (unless (or (characterp char) ; char
 487                 (condition-case nil
 488                     (setq char (find-char char-spec))
 489                   (error nil)))
 490       ;; define temporary character
 491       ;;   Current implementation is dirty.
 492       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 493                                          char-spec)))
 494       (remove-char-attribute temp-char 'ideograph-daikanwa)
 495       (setq char temp-char))
 496     (insert-char-attributes char
 497                             readable
 498                             (union (mapcar #'car char-spec)
 499                                    required-features)
 500                             nil 'for-sub-node)
 501     (when temp-char
 502       ;; undefine temporary character
 503       ;;   Current implementation is dirty.
 504       (setq char-spec (char-attribute-alist temp-char))
 505       (while char-spec
 506         (remove-char-attribute temp-char (car (car char-spec)))
 507         (setq char-spec (cdr char-spec))))))
 508
 509 (defun char-db-insert-alist (alist &optional readable column)
 510   (unless column
 511     (setq column (current-column)))
 512   (let ((line-breaking
 513          (concat "\n" (make-string (1+ column) ?\ )))
 514         name value
 515         ret al ; cal
 516         key
 517         lbs cell rest separator)
 518     (insert "(")
 519     (while alist
 520       (setq name (car (car alist))
 521             value (cdr (car alist)))
 522       (cond ((eq name 'char)
 523              (insert "(char . ")
 524              (if (setq ret (condition-case nil
 525                                (find-char value)
 526                              (error nil)))
 527                  (progn
 528                    (setq al nil
 529                          ;; cal nil
 530                          )
 531                    (while value
 532                      (setq key (car (car value)))
 533                      ;; (if (find-charset key)
 534                      ;;     (setq cal (cons key cal))
 535                      (setq al (cons key al))
 536                      ;; )
 537                      (setq value (cdr value)))
 538                    (insert-char-attributes ret
 539                                            readable
 540                                            (or al 'none) ; cal
 541                                            nil 'for-sub-node))
 542                (insert (prin1-to-string value)))
 543              (insert ")")
 544              (insert line-breaking))
 545             ((consp value)
 546              (insert (format "(%-18s " name))
 547              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 548              (while (consp value)
 549                (setq cell (car value))
 550                (if (and (consp cell)
 551                         (consp (car cell))
 552                         (setq ret (condition-case nil
 553                                       (find-char cell)
 554                                     (error nil)))
 555                         )
 556                    (progn
 557                      (setq rest cell
 558                            al nil
 559                            ;; cal nil
 560                            )
 561                      (while rest
 562                        (setq key (car (car rest)))
 563                        ;; (if (find-charset key)
 564                        ;;     (setq cal (cons key cal))
 565                        (setq al (cons key al))
 566                        ;; )
 567                        (setq rest (cdr rest)))
 568                      (if separator
 569                          (insert lbs))
 570                      (insert-char-attributes ret
 571                                              readable
 572                                              al ; cal
 573                                              nil 'for-sub-node)
 574                      (setq separator lbs))
 575                  (if separator
 576                      (insert separator))
 577                  (insert (prin1-to-string cell))
 578                  (setq separator " "))
 579                (setq value (cdr value)))
 580              (insert ")")
 581              (insert line-breaking))
 582             (t
 583              (insert (format "(%-18s . %S)%s"
 584                              name value
 585                              line-breaking))))
 586       (setq alist (cdr alist))))
 587   (insert ")"))
 588
 589 (defun char-db-insert-char-reference (plist &optional readable column)
 590   (unless column
 591     (setq column (current-column)))
 592   (let ((line-breaking
 593          (concat "\n" (make-string (1+ column) ?\ )))
 594         (separator "")
 595         name value)
 596     (insert "(")
 597     (while plist
 598       (setq name (pop plist))
 599       (setq value (pop plist))
 600       (cond ((eq name :char)
 601              (insert separator)
 602              (insert ":char\t")
 603              (cond ((numberp value)
 604                     (setq value (decode-char '=ucs value)))
 605                    ;; ((consp value)
 606                    ;;  (setq value (or (find-char value)
 607                    ;;                  value)))
 608                    )
 609              (char-db-insert-char-spec value readable)
 610              (insert line-breaking)
 611              (setq separator ""))
 612             ((eq name :radical)
 613              (insert (format "%s%s\t%d ; %c%s"
 614                              separator
 615                              name value
 616                              (ideographic-radical value)
 617                              line-breaking))
 618              (setq separator ""))
 619             (t
 620              (insert (format "%s%s\t%S" separator name value))
 621              (setq separator line-breaking)))
 622       ))
 623   (insert ")"))
 624
 625 (defun char-db-decode-isolated-char (ccs code-point)
 626   (let (ret)
 627     (setq ret
 628           (cond ((eq ccs 'arabic-iso8859-6)
 629                  (decode-char ccs code-point))
 630                 ;; ((eq ccs '=gt)
 631                 ;;  (decode-builtin-char '==gt code-point))
 632                 ((and (memq ccs '(=gt-pj-1
 633                                   =gt-pj-2
 634                                   =gt-pj-3
 635                                   =gt-pj-4
 636                                   =gt-pj-5
 637                                   =gt-pj-6
 638                                   =gt-pj-7
 639                                   =gt-pj-8
 640                                   =gt-pj-9
 641                                   =gt-pj-10
 642                                   =gt-pj-11))
 643                       (setq ret (decode-char ccs code-point))
 644                       (setq ret (encode-char ret '=gt 'defined-only)))
 645                  (decode-builtin-char '=gt ret))
 646                 (t
 647                  (decode-builtin-char ccs code-point))))
 648     (cond ((null ret)
 649            (or (decode-char ccs code-point)
 650                (define-char (list (cons ccs code-point)))))
 651           ((and (<= 0 (char-int ret))
 652                 (<= (char-int ret) #x1F))
 653            (decode-char '=ucs (+ #x2400 (char-int ret))))
 654           ((= (char-int ret) #x7F)
 655            ?\u2421)
 656           (t ret))))
 657
 658 (defvar char-db-convert-obsolete-format t)
 659
 660 (defun char-db-insert-ccs-feature (name value line-breaking)
 661   (cond
 662    ((integerp value)
 663     (insert
 664      (format
 665       (cond
 666        ((memq name '(=>iwds-1
 667                      ==shinjigen
 668                      =shinjigen
 669                      =shinjigen@1ed ==shinjigen@1ed
 670                      =shinjigen@rev ==shinjigen@rev
 671                      =shinjigen/+p@rev ==shinjigen/+p@rev
 672                      ===daikanwa/ho ==daikanwa/ho
 673                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 674         "(%-18s .  %04d)\t; %c")
 675        ((eq name '=shinjigen@1ed/24pr)
 676         "(%-18s . %04d)\t; %c")
 677        ((or
 678          (memq name
 679                '(===daikanwa
 680                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 681                  =daikanwa@rev1 =daikanwa@rev2
 682                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 683                  =>>daikanwa/+p
 684                  =daikanwa/+2p =>>daikanwa/+2p
 685                  =gt ==gt ===gt
 686                  =>>gt =+>gt =>gt
 687                  =gt-k ==gt-k ===gt-k
 688                  =>>gt-k =>gt-k
 689                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 690                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 691                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 692                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 693                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 694                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 695                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 696                  =>>adobe-japan1-0 =+>adobe-japan1-0
 697                  =>>adobe-japan1-1 =+>adobe-japan1-1
 698                  =>>adobe-japan1-2 =+>adobe-japan1-2
 699                  =>>adobe-japan1-3 =+>adobe-japan1-3
 700                  =>>adobe-japan1-4 =+>adobe-japan1-4
 701                  =>>adobe-japan1-5 =+>adobe-japan1-5
 702                  =>>adobe-japan1-6 =+>adobe-japan1-6
 703                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 704                  =zinbun-oracle =>zinbun-oracle
 705                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 706                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 707                  ===hng-khi ===hng-khm ===hng-hok ===hng-kyd ===hng-sok
 708                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 709                  ===hng-sys ===hng-tsu ===hng-tzj
 710                  ===hng-hos ===hng-nak ===hng-jhk
 711                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 712                  ===hng-nkm ===hng-k24 ===hng-nkk
 713                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 714                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 715                  ===hng-okd ===hng-wad ===hng-kmi
 716                  ===hng-zkd ===hng-doh ===hng-jyu
 717                  ===hng-tzs ===hng-kss ===hng-kyo
 718                  ===hng-smk))
 719          ;; (string-match "^=adobe-" (symbol-name name))
 720          )
 721         "(%-18s . %05d)\t; %c")
 722        ((memq name '(=hanyo-denshi/ks
 723                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 724                      =>>hanyo-denshi/ks
 725                      =koseki ==koseki
 726                      =mj ==mj ===mj =>>mj =>mj
 727                      =zihai mojikyo))
 728         "(%-18s . %06d)\t; %c")
 729        ((memq name '(=hanyo-denshi/tk ==hanyo-denshi/tk))
 730         "(%-18s . %08d)\t; %c")
 731        ((>= (charset-dimension name) 2)
 732         "(%-18s . #x%04X)\t; %c")
 733        (t
 734         "(%-18s . #x%02X)\t; %c"))
 735       name
 736       (if (= (charset-iso-graphic-plane name) 1)
 737           (logior value
 738                   (cond ((= (charset-dimension name) 1)
 739                          #x80)
 740                         ((= (charset-dimension name) 2)
 741                          #x8080)
 742                         ((= (charset-dimension name) 3)
 743                          #x808080)
 744                         (t 0)))
 745         value)
 746       (char-db-decode-isolated-char name value)))
 747     (if (and (= (charset-chars name) 94)
 748              (= (charset-dimension name) 2))
 749         (insert (format " [%02d-%02d]"
 750                         (- (lsh value -8) 32)
 751                         (- (logand value 255) 32))))
 752     )
 753    (t
 754     (insert (format "(%-18s . %s)" name value))
 755     ))
 756   (insert line-breaking))
 757
 758 (defun char-db-insert-relation-feature (char name value line-breaking
 759                                              ccss readable)
 760   (insert (format "(%-18s%s " name line-breaking))
 761   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 762         separator cell sources required-features
 763         ret)
 764     (if (characterp value)
 765         (setq value (list value)))
 766     (while (consp value)
 767       (setq cell (car value))
 768       (if (integerp cell)
 769           (setq cell (decode-char '=ucs cell)))
 770       (cond
 771        ((eq name '->subsumptive)
 772         (when (or (not (some (lambda (atr)
 773                                (get-char-attribute cell atr))
 774                              char-db-ignored-attributes))
 775                   (some (lambda (ccs)
 776                           (encode-char cell ccs 'defined-only))
 777                         ccss))
 778           (if separator
 779               (insert lbs))
 780           (let ((char-db-ignored-attributes
 781                  (cons '<-subsumptive
 782                        char-db-ignored-attributes)))
 783             (insert-char-attributes cell readable nil nil 'for-sub-node))
 784           (setq separator lbs))
 785         )
 786        ((characterp cell)
 787         (setq sources
 788               (get-char-attribute
 789                char (intern (format "%s*sources" name))))
 790         (setq required-features nil)
 791         (dolist (source sources)
 792           (cond
 793            ((memq source '(JP
 794                            JP/Jouyou
 795                            shinjigen shinjigen@1ed shinjigen@rev))
 796             (setq required-features
 797                   (union required-features
 798                          '(=jis-x0208
 799                            =jis-x0208@1990
 800                            =jis-x0213-1@2000
 801                            =jis-x0213-1@2004
 802                            =jis-x0213-2
 803                            =jis-x0212
 804                            =jis-x0208@1983
 805                            =jis-x0208@1978
 806                            =shinjigen
 807                            =shinjigen@1ed
 808                            =shinjigen@rev
 809                            =shinjigen/+p@rev))))
 810            ((eq source 'CN)
 811             (setq required-features
 812                   (union required-features
 813                          '(=gb2312
 814                            =gb12345
 815                            =iso-ir165)))))
 816           (cond
 817            ((find-charset (setq ret (intern (format "=%s" source))))
 818             (setq required-features
 819                   (cons ret required-features)))
 820            (t (setq required-features
 821                     (cons source required-features)))))
 822         (cond ((string-match "@JP" (symbol-name name))
 823                (setq required-features
 824                      (union required-features
 825                             '(=jis-x0208
 826                               =jis-x0208@1990
 827                               =jis-x0213-1-2000
 828                               =jis-x0213-2-2000
 829                               =jis-x0212
 830                               =jis-x0208@1983
 831                               =jis-x0208@1978))))
 832               ((string-match "@CN" (symbol-name name))
 833                (setq required-features
 834                      (union required-features
 835                             '(=gb2312
 836                               =gb12345
 837                               =iso-ir165)))))
 838         (if separator
 839             (insert lbs))
 840         (if readable
 841             (insert (format "%S" cell))
 842           (char-db-insert-char-spec cell readable
 843                                     nil
 844                                     required-features))
 845         (setq separator lbs))
 846        ((consp cell)
 847         (if separator
 848             (insert lbs))
 849         (if (consp (car cell))
 850             (char-db-insert-char-spec cell readable)
 851           (char-db-insert-char-reference cell readable))
 852         (setq separator lbs))
 853        (t
 854         (if separator
 855             (insert separator))
 856         (insert (prin1-to-string cell))
 857         (setq separator " ")))
 858       (setq value (cdr value)))
 859     (insert ")")
 860     (insert line-breaking)))
 861
 862 (defun insert-char-attributes (char &optional readable attributes column
 863                                     for-sub-node)
 864   (unless column
 865     (setq column (current-column)))
 866   (let (name value ; has-long-ccs-name
 867         rest
 868         radical strokes
 869         (line-breaking
 870          (concat "\n" (make-string (1+ column) ?\ )))
 871         lbs cell separator ret
 872         key al cal
 873         dest-ccss ; sources required-features
 874         ccss)
 875     (let (atr-d)
 876       (setq attributes
 877             (sort (if attributes
 878                       (if (consp attributes)
 879                           (progn
 880                             (dolist (name attributes)
 881                               (unless (memq name char-db-ignored-attributes)
 882                                 (if (find-charset name)
 883                                     (push name ccss))
 884                                 (push name atr-d)))
 885                             atr-d))
 886                     (dolist (name (char-attribute-list))
 887                       (unless (memq name char-db-ignored-attributes)
 888                         (if (find-charset name)
 889                             (push name ccss))
 890                         (push name atr-d)))
 891                     atr-d)
 892                   #'char-attribute-name<)))
 893     (insert "(")
 894     (when (memq '<-subsumptive attributes)
 895       (when (or readable (not for-sub-node))
 896         (when (setq value (get-char-attribute char '<-subsumptive))
 897           (char-db-insert-relation-feature char '<-subsumptive value
 898                                            line-breaking
 899                                            ccss readable)))
 900       (setq attributes (delq '<-subsumptive attributes)))
 901     (when (and (memq '<-denotational attributes)
 902                (setq value (get-char-attribute char '<-denotational)))
 903       (char-db-insert-relation-feature char '<-denotational value
 904                                        line-breaking
 905                                        ccss readable)
 906       (setq attributes (delq '<-denotational attributes)))
 907     (when (and (memq '<-denotational@component attributes)
 908                (setq value (get-char-attribute char '<-denotational@component)))
 909       (char-db-insert-relation-feature char '<-denotational@component value
 910                                        line-breaking
 911                                        ccss readable)
 912       (setq attributes (delq '<-denotational@component attributes)))
 913     (when (and (memq 'name attributes)
 914                (setq value (get-char-attribute char 'name)))
 915       (insert (format
 916                (if (> (+ (current-column) (length value)) 48)
 917                    "(name . %S)%s"
 918                  "(name               . %S)%s")
 919                value line-breaking))
 920       (setq attributes (delq 'name attributes))
 921       )
 922     (when (and (memq 'name* attributes)
 923                (setq value (get-char-attribute char 'name*)))
 924       (insert (format
 925                (if (> (+ (current-column) (length value)) 48)
 926                    "(name* . %S)%s"
 927                  "(name*              . %S)%s")
 928                value line-breaking))
 929       (setq attributes (delq 'name* attributes))
 930       )
 931     (when (and (memq 'script attributes)
 932                (setq value (get-char-attribute char 'script)))
 933       (insert (format "(script\t\t%s)%s"
 934                       (mapconcat (function prin1-to-string)
 935                                  value " ")
 936                       line-breaking))
 937       (setq attributes (delq 'script attributes))
 938       )
 939     (dolist (name '(=>ucs =>ucs*))
 940       (when (and (memq name attributes)
 941                  (setq value (get-char-attribute char name)))
 942         (insert (format "(%-18s . #x%04X)\t; %c%s"
 943                         name value (decode-char '=ucs value)
 944                         line-breaking))
 945         (setq attributes (delq name attributes))))
 946     (dolist (name '(=>ucs@gb =>ucs@big5))
 947       (when (and (memq name attributes)
 948                  (setq value (get-char-attribute char name)))
 949         (insert (format "(%-18s . #x%04X)\t; %c%s"
 950                         name value
 951                         (decode-char (intern
 952                                       (concat "="
 953                                               (substring
 954                                                (symbol-name name) 2)))
 955                                      value)
 956                         line-breaking))
 957         (setq attributes (delq name attributes))
 958         ))
 959     ;; (dolist (name '(=>daikanwa))
 960     ;;   (when (and (memq name attributes)
 961     ;;              (setq value (get-char-attribute char name)))
 962     ;;     (insert
 963     ;;      (if (integerp value)
 964     ;;          (format "(%-18s . %05d)\t; %c%s"
 965     ;;                  name value (decode-char '=daikanwa value)
 966     ;;                  line-breaking)
 967     ;;        (format "(%-18s %s)\t; %c%s"
 968     ;;                name
 969     ;;                (mapconcat (function prin1-to-string)
 970     ;;                           value " ")
 971     ;;                (char-representative-of-daikanwa char)
 972     ;;                line-breaking)))
 973     ;;     (setq attributes (delq name attributes))))
 974     (when (and (memq 'general-category attributes)
 975                (setq value (get-char-attribute char 'general-category)))
 976       (insert (format
 977                "(general-category\t%s) ; %s%s"
 978                (mapconcat (lambda (cell)
 979                             (format "%S" cell))
 980                           value " ")
 981                (cond ((rassoc value unidata-normative-category-alist)
 982                       "Normative Category")
 983                      ((rassoc value unidata-informative-category-alist)
 984                       "Informative Category")
 985                      (t
 986                       "Unknown Category"))
 987                line-breaking))
 988       (setq attributes (delq 'general-category attributes))
 989       )
 990     (when (and (memq 'bidi-category attributes)
 991                (setq value (get-char-attribute char 'bidi-category)))
 992       (insert (format "(bidi-category\t. %S)%s"
 993                       value
 994                       line-breaking))
 995       (setq attributes (delq 'bidi-category attributes))
 996       )
 997     (unless (or (not (memq 'mirrored attributes))
 998                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
 999                     'empty))
1000       (insert (format "(mirrored\t\t. %S)%s"
1001                       value
1002                       line-breaking))
1003       (setq attributes (delq 'mirrored attributes))
1004       )
1005     (cond
1006      ((and (memq 'decimal-digit-value attributes)
1007            (setq value (get-char-attribute char 'decimal-digit-value)))
1008       (insert (format "(decimal-digit-value . %S)%s"
1009                       value
1010                       line-breaking))
1011       (setq attributes (delq 'decimal-digit-value attributes))
1012       (when (and (memq 'digit-value attributes)
1013                  (setq value (get-char-attribute char 'digit-value)))
1014         (insert (format "(digit-value\t . %S)%s"
1015                         value
1016                         line-breaking))
1017         (setq attributes (delq 'digit-value attributes))
1018         )
1019       (when (and (memq 'numeric-value attributes)
1020                  (setq value (get-char-attribute char 'numeric-value)))
1021         (insert (format "(numeric-value\t . %S)%s"
1022                         value
1023                         line-breaking))
1024         (setq attributes (delq 'numeric-value attributes))
1025         )
1026       )
1027      (t
1028       (when (and (memq 'digit-value attributes)
1029                  (setq value (get-char-attribute char 'digit-value)))
1030         (insert (format "(digit-value\t. %S)%s"
1031                         value
1032                         line-breaking))
1033         (setq attributes (delq 'digit-value attributes))
1034         )
1035       (when (and (memq 'numeric-value attributes)
1036                  (setq value (get-char-attribute char 'numeric-value)))
1037         (insert (format "(numeric-value\t. %S)%s"
1038                         value
1039                         line-breaking))
1040         (setq attributes (delq 'numeric-value attributes))
1041         )))
1042     (when (and (memq 'iso-10646-comment attributes)
1043                (setq value (get-char-attribute char 'iso-10646-comment)))
1044       (insert (format "(iso-10646-comment\t. %S)%s"
1045                       value
1046                       line-breaking))
1047       (setq attributes (delq 'iso-10646-comment attributes))
1048       )
1049     (when (and (memq 'morohashi-daikanwa attributes)
1050                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1051       (insert (format "(morohashi-daikanwa\t%s)%s"
1052                       (mapconcat (function prin1-to-string) value " ")
1053                       line-breaking))
1054       (setq attributes (delq 'morohashi-daikanwa attributes))
1055       )
1056     (setq radical nil
1057           strokes nil)
1058     (when (and (memq 'ideographic-radical attributes)
1059                (setq value (get-char-attribute char 'ideographic-radical)))
1060       (setq radical value)
1061       (insert (format "(ideographic-radical . %S)\t; %c%s"
1062                       radical
1063                       (ideographic-radical radical)
1064                       line-breaking))
1065       (setq attributes (delq 'ideographic-radical attributes))
1066       )
1067     (when (and (memq 'shuowen-radical attributes)
1068                (setq value (get-char-attribute char 'shuowen-radical)))
1069       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1070                       value
1071                       (shuowen-radical value)
1072                       line-breaking))
1073       (setq attributes (delq 'shuowen-radical attributes))
1074       )
1075     (let (key)
1076       (dolist (domain
1077                (append
1078                 char-db-feature-domains
1079                 (let (dest domain)
1080                   (dolist (feature (char-attribute-list))
1081                     (setq feature (symbol-name feature))
1082                     (when (string-match
1083                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1084                            feature)
1085                       (setq domain (intern (match-string 2 feature)))
1086                      (unless (memq domain dest)
1087                        (setq dest (cons domain dest)))))
1088                   (sort dest #'string<))))
1089         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1090         (when (and (memq key attributes)
1091                    (setq value (get-char-attribute char key)))
1092           (setq radical value)
1093           (insert (format "(%s . %S)\t; %c%s"
1094                           key
1095                           radical
1096                           (ideographic-radical radical)
1097                           line-breaking))
1098           (setq attributes (delq key attributes))
1099           )
1100         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1101         (when (and (memq key attributes)
1102                    (setq value (get-char-attribute char key)))
1103           (setq strokes value)
1104           (insert (format "(%s . %S)%s"
1105                           key
1106                           strokes
1107                           line-breaking))
1108           (setq attributes (delq key attributes))
1109           )
1110         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1111         (when (and (memq key attributes)
1112                    (setq value (get-char-attribute char key)))
1113           (insert (format "(%s       . %S)%s"
1114                           key
1115                           value
1116                           line-breaking))
1117           (setq attributes (delq key attributes))
1118           )
1119         (dolist (feature '(ideographic-radical
1120                            ideographic-strokes
1121                            total-strokes))
1122           (setq key (intern (format "%s@%s*sources" feature domain)))
1123           (when (and (memq key attributes)
1124                      (setq value (get-char-attribute char key)))
1125             (insert (format "(%s%s" key line-breaking))
1126             (dolist (cell value)
1127               (insert (format " %s" cell)))
1128             (insert ")")
1129             (insert line-breaking)
1130             (setq attributes (delq key attributes))
1131             ))
1132         ))
1133     (when (and (memq 'ideographic-strokes attributes)
1134                (setq value (get-char-attribute char 'ideographic-strokes)))
1135       (setq strokes value)
1136       (insert (format "(ideographic-strokes . %S)%s"
1137                       strokes
1138                       line-breaking))
1139       (setq attributes (delq 'ideographic-strokes attributes))
1140       )
1141     (when (and (memq 'kangxi-radical attributes)
1142                (setq value (get-char-attribute char 'kangxi-radical)))
1143       (unless (eq value radical)
1144         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1145                         value
1146                         (ideographic-radical value)
1147                         line-breaking))
1148         (or radical
1149             (setq radical value)))
1150       (setq attributes (delq 'kangxi-radical attributes))
1151       )
1152     (when (and (memq 'kangxi-strokes attributes)
1153                (setq value (get-char-attribute char 'kangxi-strokes)))
1154       (unless (eq value strokes)
1155         (insert (format "(kangxi-strokes\t . %S)%s"
1156                         value
1157                         line-breaking))
1158         (or strokes
1159             (setq strokes value)))
1160       (setq attributes (delq 'kangxi-strokes attributes))
1161       )
1162     ;; (when (and (memq 'japanese-radical attributes)
1163     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1164     ;;   (unless (eq value radical)
1165     ;;     (insert (format "(japanese-radical\t . %S)\t; %c%s"
1166     ;;                     value
1167     ;;                     (ideographic-radical value)
1168     ;;                     line-breaking))
1169     ;;     (or radical
1170     ;;         (setq radical value)))
1171     ;;   (setq attributes (delq 'japanese-radical attributes))
1172     ;;   )
1173     (when (and (memq 'japanese-strokes attributes)
1174                (setq value (get-char-attribute char 'japanese-strokes)))
1175       (unless (eq value strokes)
1176         (insert (format "(japanese-strokes\t . %S)%s"
1177                         value
1178                         line-breaking))
1179         (or strokes
1180             (setq strokes value)))
1181       (setq attributes (delq 'japanese-strokes attributes))
1182       )
1183     (when (and (memq 'cns-radical attributes)
1184                (setq value (get-char-attribute char 'cns-radical)))
1185       (insert (format "(cns-radical\t . %S)\t; %c%s"
1186                       value
1187                       (ideographic-radical value)
1188                       line-breaking))
1189       (setq attributes (delq 'cns-radical attributes))
1190       )
1191     (when (and (memq 'cns-strokes attributes)
1192                (setq value (get-char-attribute char 'cns-strokes)))
1193       (unless (eq value strokes)
1194         (insert (format "(cns-strokes\t . %S)%s"
1195                         value
1196                         line-breaking))
1197         (or strokes
1198             (setq strokes value)))
1199       (setq attributes (delq 'cns-strokes attributes))
1200       )
1201     ;; (when (and (memq 'shinjigen-1-radical attributes)
1202     ;;            (setq value (get-char-attribute char 'shinjigen-1-radical)))
1203     ;;   (unless (eq value radical)
1204     ;;     (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
1205     ;;                     value
1206     ;;                     (ideographic-radical value)
1207     ;;                     line-breaking))
1208     ;;     (or radical
1209     ;;         (setq radical value)))
1210     ;;   (setq attributes (delq 'shinjigen-1-radical attributes))
1211     ;;   )
1212     ;; (when (and (memq 'ideographic- attributes)
1213     ;;            (setq value (get-char-attribute char 'ideographic-)))
1214     ;;   (insert "(ideographic-       ")
1215     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1216     ;;         separator nil)
1217     ;;   (while (consp value)
1218     ;;     (setq cell (car value))
1219     ;;     (if (integerp cell)
1220     ;;         (setq cell (decode-char '=ucs cell)))
1221     ;;     (cond ((characterp cell)
1222     ;;            (if separator
1223     ;;                (insert lbs))
1224     ;;            (if readable
1225     ;;                (insert (format "%S" cell))
1226     ;;              (char-db-insert-char-spec cell readable))
1227     ;;            (setq separator lbs))
1228     ;;           ((consp cell)
1229     ;;            (if separator
1230     ;;                (insert lbs))
1231     ;;            (if (consp (car cell))
1232     ;;                (char-db-insert-char-spec cell readable)
1233     ;;              (char-db-insert-char-reference cell readable))
1234     ;;            (setq separator lbs))
1235     ;;           (t
1236     ;;            (if separator
1237     ;;                (insert separator))
1238     ;;            (insert (prin1-to-string cell))
1239     ;;            (setq separator " ")))
1240     ;;     (setq value (cdr value)))
1241     ;;   (insert ")")
1242     ;;   (insert line-breaking)
1243     ;;   (setq attributes (delq 'ideographic- attributes)))
1244     (when (and (memq 'total-strokes attributes)
1245                (setq value (get-char-attribute char 'total-strokes)))
1246       (insert (format "(total-strokes       . %S)%s"
1247                       value
1248                       line-breaking))
1249       (setq attributes (delq 'total-strokes attributes))
1250       )
1251     ;; (when (and (memq '->ideograph attributes)
1252     ;;            (setq value (get-char-attribute char '->ideograph)))
1253     ;;   (insert (format "(->ideograph\t%s)%s"
1254     ;;                   (mapconcat (lambda (code)
1255     ;;                                (cond ((symbolp code)
1256     ;;                                       (symbol-name code))
1257     ;;                                      ((integerp code)
1258     ;;                                       (format "#x%04X" code))
1259     ;;                                      (t
1260     ;;                                       (format "%s %S"
1261     ;;                                               line-breaking code))))
1262     ;;                              value " ")
1263     ;;                   line-breaking))
1264     ;;   (setq attributes (delq '->ideograph attributes))
1265     ;;   )
1266     ;; (when (and (memq '->decomposition attributes)
1267     ;;            (setq value (get-char-attribute char '->decomposition)))
1268     ;;   (insert (format "(->decomposition\t%s)%s"
1269     ;;                   (mapconcat (lambda (code)
1270     ;;                                (cond ((symbolp code)
1271     ;;                                       (symbol-name code))
1272     ;;                                      ((characterp code)
1273     ;;                                       (if readable
1274     ;;                                           (format "%S" code)
1275     ;;                                         (format "#x%04X"
1276     ;;                                                 (char-int code))
1277     ;;                                         ))
1278     ;;                                      ((integerp code)
1279     ;;                                       (format "#x%04X" code))
1280     ;;                                      (t
1281     ;;                                       (format "%s%S" line-breaking code))))
1282     ;;                              value " ")
1283     ;;                   line-breaking))
1284     ;;   (setq attributes (delq '->decomposition attributes))
1285     ;;   )
1286     (if (equal (get-char-attribute char '->titlecase)
1287                (get-char-attribute char '->uppercase))
1288         (setq attributes (delq '->titlecase attributes)))
1289     (when (and (memq '->mojikyo attributes)
1290                (setq value (get-char-attribute char '->mojikyo)))
1291       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1292                       value (decode-char 'mojikyo value)
1293                       line-breaking))
1294       (setq attributes (delq '->mojikyo attributes))
1295       )
1296     ;; (when (and (memq 'hanyu-dazidian-vol attributes)
1297     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1298     ;;   (insert (format "(hanyu-dazidian-vol  . %d)%s"
1299     ;;                   value line-breaking))
1300     ;;   (setq attributes (delq 'hanyu-dazidian-vol attributes))
1301     ;;   )
1302     ;; (when (and (memq 'hanyu-dazidian-page attributes)
1303     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1304     ;;   (insert (format "(hanyu-dazidian-page . %d)%s"
1305     ;;                   value line-breaking))
1306     ;;   (setq attributes (delq 'hanyu-dazidian-page attributes))
1307     ;;   )
1308     ;; (when (and (memq 'hanyu-dazidian-char attributes)
1309     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1310     ;;   (insert (format "(hanyu-dazidian-char . %d)%s"
1311     ;;                   value line-breaking))
1312     ;;   (setq attributes (delq 'hanyu-dazidian-char attributes))
1313     ;;   )
1314     (unless readable
1315       (dolist (ignored '(composition
1316                          ->denotational <-subsumptive ->ucs-unified
1317                          ->ideographic-component-forms))
1318         (setq attributes (delq ignored attributes))))
1319     (while attributes
1320       (setq name (car attributes))
1321       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1322                   'value-is-empty)
1323         (cond ((setq ret (find-charset name))
1324                (setq name (charset-name ret))
1325                (when (not (memq name dest-ccss))
1326                  (setq dest-ccss (cons name dest-ccss))
1327                  (char-db-insert-ccs-feature name value line-breaking))
1328                )
1329               ((string-match "^=>ucs@" (symbol-name name))
1330                (insert (format "(%-18s . #x%04X)\t; %c%s"
1331                                name value (decode-char '=ucs value)
1332                                line-breaking))
1333                )
1334               ((eq name 'jisx0208-1978/4X)
1335                (insert (format "(%-18s . #x%04X)%s"
1336                                name value
1337                                line-breaking))
1338                )
1339               ((and
1340                 (not readable)
1341                 (not (eq name '->subsumptive))
1342                 (not (eq name '->uppercase))
1343                 (not (eq name '->lowercase))
1344                 (not (eq name '->titlecase))
1345                 (not (eq name '->canonical))
1346                 (not (eq name '->Bopomofo))
1347                 (not (eq name '->mistakable))
1348                 (not (eq name '->ideographic-variants))
1349                 (null (get-char-attribute
1350                        char (intern (format "%s*sources" name))))
1351                 (not (string-match "\\*sources$" (symbol-name name)))
1352                 (null (get-char-attribute
1353                        char (intern (format "%s*note" name))))
1354                 (not (string-match "\\*note$" (symbol-name name)))
1355                 (or (eq name '<-identical)
1356                     (eq name '<-uppercase)
1357                     (eq name '<-lowercase)
1358                     (eq name '<-titlecase)
1359                     (eq name '<-canonical)
1360                     (eq name '<-ideographic-variants)
1361                     ;; (eq name '<-synonyms)
1362                     (string-match "^<-synonyms" (symbol-name name))
1363                     (eq name '<-mistakable)
1364                     (when (string-match "^->" (symbol-name name))
1365                       (cond
1366                        ((string-match "^->fullwidth" (symbol-name name))
1367                         (not (and (consp value)
1368                                   (characterp (car value))
1369                                   (encode-char
1370                                    (car value) '=ucs 'defined-only)))
1371                         )
1372                        (t)))
1373                     ))
1374                )
1375               ((or (eq name 'ideographic-structure)
1376                    (eq name 'ideographic-combination)
1377                    ;; (eq name 'ideographic-)
1378                    (eq name '=decomposition)
1379                    (char-feature-base-name= '=decomposition name)
1380                    (char-feature-base-name= '=>decomposition name)
1381                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1382                    ;;               (symbol-name name))
1383                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1384                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1385                                  (symbol-name name))
1386                    )
1387                (char-db-insert-relation-feature char name value
1388                                                 line-breaking
1389                                                 ccss readable))
1390               ;; ((memq name '(ideograph=
1391               ;;               original-ideograph-of
1392               ;;               ancient-ideograph-of
1393               ;;               vulgar-ideograph-of
1394               ;;               wrong-ideograph-of
1395               ;;               ;; simplified-ideograph-of
1396               ;;               ideographic-variants
1397               ;;               ;; ideographic-different-form-of
1398               ;;               ))
1399               ;;  (insert (format "(%-18s%s " name line-breaking))
1400               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1401               ;;        separator nil)
1402               ;;  (while (consp value)
1403               ;;    (setq cell (car value))
1404               ;;    (if (and (consp cell)
1405               ;;             (consp (car cell)))
1406               ;;        (progn
1407               ;;          (if separator
1408               ;;              (insert lbs))
1409               ;;          (char-db-insert-alist cell readable)
1410               ;;          (setq separator lbs))
1411               ;;      (if separator
1412               ;;          (insert separator))
1413               ;;      (insert (prin1-to-string cell))
1414               ;;      (setq separator " "))
1415               ;;    (setq value (cdr value)))
1416               ;;  (insert ")")
1417               ;;  (insert line-breaking))
1418               ((consp value)
1419                (insert (format "(%-18s " name))
1420                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1421                      separator nil)
1422                (while (consp value)
1423                  (setq cell (car value))
1424                  (if (and (consp cell)
1425                           (consp (car cell))
1426                           (setq ret (condition-case nil
1427                                         (find-char cell)
1428                                       (error nil))))
1429                      (progn
1430                        (setq rest cell
1431                              al nil
1432                              cal nil)
1433                        (while rest
1434                          (setq key (car (car rest)))
1435                          (if (find-charset key)
1436                              (setq cal (cons key cal))
1437                            (setq al (cons key al)))
1438                          (setq rest (cdr rest)))
1439                        (if separator
1440                            (insert lbs))
1441                        (insert-char-attributes ret
1442                                                readable
1443                                                al ; cal
1444                                                nil 'for-sub-node)
1445                        (setq separator lbs))
1446                    (setq ret (prin1-to-string cell))
1447                    (if separator
1448                        (if (< (+ (current-column)
1449                                  (length ret)
1450                                  (length separator))
1451                               76)
1452                            (insert separator)
1453                          (insert lbs)))
1454                    (insert ret)
1455                    (setq separator " "))
1456                  (setq value (cdr value)))
1457                (insert ")")
1458                (insert line-breaking))
1459               (t
1460                (insert (format "(%-18s" name))
1461                (setq ret (prin1-to-string value))
1462                (unless (< (+ (current-column)
1463                              (length ret)
1464                              3)
1465                           76)
1466                  (insert line-breaking))
1467                (insert " . " ret ")" line-breaking)
1468                ;; (insert (format "(%-18s . %S)%s"
1469                ;;                 name value
1470                ;;                 line-breaking))
1471                )
1472               ))
1473       (setq attributes (cdr attributes)))
1474     (insert ")")))
1475
1476 (defun insert-char-data (char &optional readable
1477                               attributes)
1478   (save-restriction
1479     (narrow-to-region (point)(point))
1480     (insert "(define-char
1481   '")
1482     (insert-char-attributes char readable attributes)
1483     (insert ")\n")
1484     (goto-char (point-min))
1485     (while (re-search-forward "[ \t]+$" nil t)
1486       (replace-match ""))
1487     ;; from tabify.
1488     (goto-char (point-min))
1489     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1490       (let ((column (current-column))
1491             (indent-tabs-mode t))
1492         (delete-region (match-beginning 0) (point))
1493         (indent-to column)))
1494     (goto-char (point-max))
1495     ;; (tabify (point-min)(point-max))
1496     ))
1497
1498 (defun insert-char-data-with-variant (char &optional printable
1499                                            no-ucs-unified
1500                                            script excluded-script)
1501   (insert-char-data char printable)
1502   (let ((variants (char-variants char))
1503         rest
1504         variant vs ret)
1505     (setq variants (sort variants #'<))
1506     (setq rest variants)
1507     (setq variants (cons char variants))
1508     (while rest
1509       (setq variant (car rest))
1510       (unless (get-char-attribute variant '<-subsumptive)
1511         (if (and (or (null script)
1512                      (null (setq vs (get-char-attribute variant 'script)))
1513                      (memq script vs))
1514                  (or (null excluded-script)
1515                      (null (setq vs (get-char-attribute variant 'script)))
1516                      (not (memq excluded-script vs))))
1517             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1518               (insert-char-data variant printable)
1519               (if (setq ret (char-variants variant))
1520                   (while ret
1521                     (or (memq (car ret) variants)
1522                         ;; (get-char-attribute (car ret) '<-subsumptive)
1523                         (setq rest (nconc rest (list (car ret)))))
1524                     (setq ret (cdr ret)))))))
1525       (setq rest (cdr rest)))))
1526
1527 (defun insert-char-range-data (min max &optional script excluded-script)
1528   (let ((code min)
1529         char)
1530     (while (<= code max)
1531       (setq char (decode-char '=ucs code))
1532       (if (encode-char char '=ucs 'defined-only)
1533           (insert-char-data-with-variant char nil 'no-ucs-unified
1534                                          script excluded-script))
1535       (setq code (1+ code)))))
1536
1537 (defun write-char-range-data-to-file (min max file
1538                                           &optional script excluded-script)
1539   (let ((coding-system-for-write char-db-file-coding-system))
1540     (with-temp-buffer
1541       (insert (format ";; -*- coding: %s -*-\n"
1542                       char-db-file-coding-system))
1543       (insert-char-range-data min max script excluded-script)
1544       (write-region (point-min)(point-max) file))))
1545
1546 (defvar what-character-original-window-configuration)
1547
1548 ;;;###autoload
1549 (defun what-char-definition (char)
1550   (interactive (list (char-after)))
1551   (let ((buf (get-buffer-create "*Character Description*"))
1552         (the-buf (current-buffer))
1553         (win-conf (current-window-configuration)))
1554     (pop-to-buffer buf)
1555     (make-local-variable 'what-character-original-window-configuration)
1556     (setq what-character-original-window-configuration win-conf)
1557     (setq buffer-read-only nil)
1558     (erase-buffer)
1559     (condition-case err
1560         (progn
1561           (insert-char-data-with-variant char 'printable)
1562           (unless (char-attribute-alist char)
1563             (insert (format ";; = %c\n"
1564                             (let* ((rest (split-char char))
1565                                    (ccs (pop rest))
1566                                    (code (pop rest)))
1567                               (while rest
1568                                 (setq code (logior (lsh code 8)
1569                                                    (pop rest))))
1570                               (decode-char ccs code)))))
1571           ;; (char-db-update-comment)
1572           (set-buffer-modified-p nil)
1573           (view-mode the-buf (lambda (buf)
1574                                (set-window-configuration
1575                                 what-character-original-window-configuration)
1576                                ))
1577           (goto-char (point-min)))
1578       (error (progn
1579                (set-window-configuration
1580                 what-character-original-window-configuration)
1581                (signal (car err) (cdr err)))))))
1582
1583
1584 ;;; @ end
1585 ;;;
1586
1587 (provide 'char-db-util)
1588
1589 ;;; char-db-util.el ends here