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

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