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,
   4 ;;   2007, 2008, 2009, 2010, 2011, 2012, 2013, 2014 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
 196     =cbeta
 197     ideograph-hanziku-1
 198     ideograph-hanziku-2
 199     ideograph-hanziku-3
 200     ideograph-hanziku-4
 201     ideograph-hanziku-5
 202     ideograph-hanziku-6
 203     ideograph-hanziku-7
 204     ideograph-hanziku-8
 205     ideograph-hanziku-9
 206     ideograph-hanziku-10
 207     ideograph-hanziku-11
 208     ideograph-hanziku-12
 209     ;; =>>>adobe-japan1-0
 210     ;; =>>>adobe-japan1-1
 211     ;; =>>>adobe-japan1-2
 212     ;; =>>>adobe-japan1-3
 213     ;; =>>>adobe-japan1-4
 214     ;; =>>>adobe-japan1-5
 215     ;; =>>>adobe-japan1-6
 216     ;; =>>>jis-x0208
 217     ;; =>>>jis-x0213-1
 218     ;; =>>>jis-x0213-2
 219     ;; =>>>hanyo-denshi/ja
 220     ;; =>>>hanyo-denshi/jb
 221     ;; =>>>hanyo-denshi/jc
 222     ;; =>>>hanyo-denshi/ft
 223     ;; =>>>hanyo-denshi/ib
 224     ;; =>>>hanyo-denshi/hg
 225     ;; =>>>hanyo-denshi/jt
 226     ;; =>>>hanyo-denshi/ks
 227     ;; =>>>gt
 228     =>>adobe-japan1-0
 229     =>>adobe-japan1-1
 230     =>>adobe-japan1-2
 231     =>>adobe-japan1-3
 232     =>>adobe-japan1-4
 233     =>>adobe-japan1-5
 234     =>>adobe-japan1-6
 235     =>>jis-x0208
 236     =>>jis-x0213-1
 237     =>>jis-x0213-1@2000
 238     =>>jis-x0213-1@2004
 239     =>>jis-x0213-2
 240     =>>jis-x0208@1978
 241     =>>hanyo-denshi/ft
 242     =>>hanyo-denshi/jt
 243     =>>hanyo-denshi/ks
 244     =>>gt
 245     =>>daikanwa
 246     =>>cbeta
 247     =+>jis-x0208
 248     =+>jis-x0213-1
 249     =+>jis-x0213-2
 250     =+>adobe-japan1-0
 251     =+>adobe-japan1-1
 252     =+>adobe-japan1-2
 253     =+>adobe-japan1-3
 254     =+>adobe-japan1-4
 255     =+>adobe-japan1-5
 256     =+>adobe-japan1-6
 257     =+>jis-x0208@1978
 258     =>jis-x0208
 259     =>jis-x0208@1997
 260     =>jis-x0213-1
 261     =>jis-x0213-1@2000
 262     =>jis-x0213-1@2004
 263     =>jis-x0213-2
 264     ==>ucs@bucs
 265     =>ucs@hanyo-denshi
 266     =>ucs@iso
 267     =>ucs@unicode
 268     =>ucs@jis
 269     =>ucs@cns
 270     =>ucs@ks
 271     =+>ucs@iso
 272     =+>ucs@unicode
 273     =+>ucs@jis
 274     =+>ucs@jis/1990
 275     =+>ucs@cns
 276     =+>ucs@ks
 277     =>>ucs@iso
 278     =>>ucs@unicode
 279     =>>ucs@jis
 280     =>>ucs@cns
 281     =>>>ucs@iso
 282     =>>>ucs@unicode
 283     ==ucs@iso
 284     ==ucs@unicode
 285     ;; ==ucs@cns
 286     ==gb2312
 287     ==ks-x1001
 288     ==cns11643-1
 289     ==cns11643-2
 290     ==cns11643-3
 291     ==cns11643-4
 292     ==cns11643-5
 293     ==cns11643-6
 294     ==cns11643-7
 295     ==gt
 296     ==jis-x0208@1990
 297     ;; ==jis-x0208@1983
 298     ;; ==jis-x0208@1978
 299     ==gt-k
 300     =ucs@iso
 301     =ucs@unicode
 302     =ucs@cns
 303     ==big5-cdp
 304     ==cbeta
 305     =>>big5-cdp
 306     =>>gt-k
 307     =+>gt
 308     =>gt
 309     =>big5-cdp
 310     =>daikanwa
 311     =>daikanwa/ho
 312     =>cns11643-7
 313     =big5
 314     =big5-eten
 315     =>gt-k
 316     =zinbun-oracle
 317     =>zinbun-oracle
 318     =ruimoku-v6
 319     =>>ruimoku-v6
 320     ==ruimoku-v6
 321     =jef-china3
 322     =>cbeta
 323     =shinjigen
 324     =big5-cdp-var-3
 325     =big5-cdp-var-5))
 326
 327
 328 ;;; @ char-db formatters
 329 ;;;
 330
 331 (defun char-db-make-char-spec (char)
 332   (let (ret char-spec)
 333     (cond ((characterp char)
 334            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 335                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 336                   (setq char-spec (list (cons '=ucs ret)))
 337                   (cond ((setq ret (get-char-attribute char 'name))
 338                          (setq char-spec (cons (cons 'name ret) char-spec))
 339                          )
 340                         ((setq ret (get-char-attribute char 'name*))
 341                          (setq char-spec (cons (cons 'name* ret) char-spec))
 342                          ))
 343                   )
 344                  ((encode-char char '=mj 'defined-only)
 345                   (setq char-spec nil)
 346                   (dolist (ccs (charset-list))
 347                     (if (and (or (eq ccs '=mj)
 348                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 349                                  (string-match "=ucs@" (symbol-name ccs))
 350                                  )
 351                              (setq ccs (charset-name ccs))
 352                              (null (assq ccs char-spec))
 353                              (setq ret (encode-char char ccs 'defined-only)))
 354                         (setq char-spec (cons (cons ccs ret) char-spec))))
 355                   )
 356                  ((encode-char char '==mj 'defined-only)
 357                   (setq char-spec nil)
 358                   (dolist (ccs (charset-list))
 359                     (if (and (or (eq ccs '==mj)
 360                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 361                                  (string-match "=ucs@" (symbol-name ccs))
 362                                  )
 363                              (setq ccs (charset-name ccs))
 364                              (null (assq ccs char-spec))
 365                              (setq ret (encode-char char ccs 'defined-only)))
 366                         (setq char-spec (cons (cons ccs ret) char-spec))))
 367                   )
 368                  ((encode-char char '=adobe-japan1 'defined-only)
 369                   (setq char-spec nil)
 370                   (dolist (ccs (charset-list))
 371                     (if (and (or (memq ccs
 372                                        '(=adobe-japan1-0
 373                                          =adobe-japan1-1
 374                                          =adobe-japan1-2
 375                                          =adobe-japan1-3
 376                                          =adobe-japan1-4
 377                                          =adobe-japan1-5
 378                                          =adobe-japan1-6
 379                                          ))
 380                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 381                                  (string-match "=ucs@" (symbol-name ccs))
 382                                  )
 383                              (setq ccs (charset-name ccs))
 384                              (null (assq ccs char-spec))
 385                              (setq ret (encode-char char ccs 'defined-only)))
 386                         (setq char-spec (cons (cons ccs ret) char-spec))))
 387                   )
 388                  ((encode-char char '==adobe-japan1 'defined-only)
 389                   (setq char-spec nil)
 390                   (dolist (ccs (charset-list))
 391                     (if (and (or (memq ccs
 392                                        '(==adobe-japan1-0
 393                                          ==adobe-japan1-1
 394                                          ==adobe-japan1-2
 395                                          ==adobe-japan1-3
 396                                          ==adobe-japan1-4
 397                                          ==adobe-japan1-5
 398                                          ==adobe-japan1-6
 399                                          ))
 400                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 401                                  (string-match "=ucs@" (symbol-name ccs))
 402                                  )
 403                              (setq ccs (charset-name ccs))
 404                              (null (assq ccs char-spec))
 405                              (setq ret (encode-char char ccs 'defined-only)))
 406                         (setq char-spec (cons (cons ccs ret) char-spec))))
 407                   )
 408                  ((setq ret
 409                         (catch 'tag
 410                           (let ((rest char-db-coded-charset-priority-list)
 411                                 ccs)
 412                             (while rest
 413                               (setq ccs (charset-name
 414                                          (find-charset (car rest))))
 415                               (if (setq ret
 416                                         (encode-char char ccs
 417                                                      'defined-only))
 418                                   (throw 'tag (cons ccs ret)))
 419                               (setq rest (cdr rest))))))
 420                   (setq char-spec (list ret))
 421                   (dolist (ccs (delq (car ret) (charset-list)))
 422                     (if (and (or (charset-iso-final-char ccs)
 423                                  (memq ccs
 424                                        '(=daikanwa
 425                                          =daikanwa@rev2
 426                                          ;; =gt-k
 427                                          =jis-x0208@1997
 428                                          ))
 429                                  (eq (charset-property ccs 'iso-ir) 177)
 430                                  ;; (string-match "=ucs@" (symbol-name ccs))
 431                                  )
 432                              (setq ccs (charset-name ccs))
 433                              (null (assq ccs char-spec))
 434                              (setq ret (encode-char char ccs 'defined-only)))
 435                         (setq char-spec (cons (cons ccs ret) char-spec))))
 436                   (if (null char-spec)
 437                       (setq char-spec (split-char char)))
 438                   (cond ((setq ret (get-char-attribute char 'name))
 439                          (setq char-spec (cons (cons 'name ret) char-spec))
 440                          )
 441                         ((setq ret (get-char-attribute char 'name*))
 442                          (setq char-spec (cons (cons 'name* ret) char-spec))
 443                          ))
 444                   )
 445                  ((setq ret (get-char-attribute
 446                              char 'ideographic-combination))
 447                   (setq char-spec
 448                         (cons (cons 'ideographic-combination ret)
 449                               char-spec))
 450                   ))
 451            char-spec)
 452           ((consp char)
 453            char))))
 454
 455 (defun char-db-insert-char-spec (char &optional readable column
 456                                       required-features)
 457   (unless column
 458     (setq column (current-column)))
 459   (let (char-spec temp-char)
 460     (setq char-spec (char-db-make-char-spec char))
 461     (unless (or (characterp char) ; char
 462                 (condition-case nil
 463                     (setq char (find-char char-spec))
 464                   (error nil)))
 465       ;; define temporary character
 466       ;;   Current implementation is dirty.
 467       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 468                                          char-spec)))
 469       (remove-char-attribute temp-char 'ideograph-daikanwa)
 470       (setq char temp-char))
 471     (insert-char-attributes char
 472                             readable
 473                             (union (mapcar #'car char-spec)
 474                                    required-features)
 475                             nil 'for-sub-node)
 476     (when temp-char
 477       ;; undefine temporary character
 478       ;;   Current implementation is dirty.
 479       (setq char-spec (char-attribute-alist temp-char))
 480       (while char-spec
 481         (remove-char-attribute temp-char (car (car char-spec)))
 482         (setq char-spec (cdr char-spec))))))
 483
 484 (defun char-db-insert-alist (alist &optional readable column)
 485   (unless column
 486     (setq column (current-column)))
 487   (let ((line-breaking
 488          (concat "\n" (make-string (1+ column) ?\ )))
 489         name value
 490         ret al ; cal
 491         key
 492         lbs cell rest separator)
 493     (insert "(")
 494     (while alist
 495       (setq name (car (car alist))
 496             value (cdr (car alist)))
 497       (cond ((eq name 'char)
 498              (insert "(char . ")
 499              (if (setq ret (condition-case nil
 500                                (find-char value)
 501                              (error nil)))
 502                  (progn
 503                    (setq al nil
 504                          ;; cal nil
 505                          )
 506                    (while value
 507                      (setq key (car (car value)))
 508                      ;; (if (find-charset key)
 509                      ;;     (setq cal (cons key cal))
 510                      (setq al (cons key al))
 511                      ;; )
 512                      (setq value (cdr value)))
 513                    (insert-char-attributes ret
 514                                            readable
 515                                            (or al 'none) ; cal
 516                                            nil 'for-sub-node))
 517                (insert (prin1-to-string value)))
 518              (insert ")")
 519              (insert line-breaking))
 520             ((consp value)
 521              (insert (format "(%-18s " name))
 522              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 523              (while (consp value)
 524                (setq cell (car value))
 525                (if (and (consp cell)
 526                         (consp (car cell))
 527                         (setq ret (condition-case nil
 528                                       (find-char cell)
 529                                     (error nil)))
 530                         )
 531                    (progn
 532                      (setq rest cell
 533                            al nil
 534                            ;; cal nil
 535                            )
 536                      (while rest
 537                        (setq key (car (car rest)))
 538                        ;; (if (find-charset key)
 539                        ;;     (setq cal (cons key cal))
 540                        (setq al (cons key al))
 541                        ;; )
 542                        (setq rest (cdr rest)))
 543                      (if separator
 544                          (insert lbs))
 545                      (insert-char-attributes ret
 546                                              readable
 547                                              al ; cal
 548                                              nil 'for-sub-node)
 549                      (setq separator lbs))
 550                  (if separator
 551                      (insert separator))
 552                  (insert (prin1-to-string cell))
 553                  (setq separator " "))
 554                (setq value (cdr value)))
 555              (insert ")")
 556              (insert line-breaking))
 557             (t
 558              (insert (format "(%-18s . %S)%s"
 559                              name value
 560                              line-breaking))))
 561       (setq alist (cdr alist))))
 562   (insert ")"))
 563
 564 (defun char-db-insert-char-reference (plist &optional readable column)
 565   (unless column
 566     (setq column (current-column)))
 567   (let ((line-breaking
 568          (concat "\n" (make-string (1+ column) ?\ )))
 569         (separator "")
 570         name value)
 571     (insert "(")
 572     (while plist
 573       (setq name (pop plist))
 574       (setq value (pop plist))
 575       (cond ((eq name :char)
 576              (insert separator)
 577              (insert ":char\t")
 578              (cond ((numberp value)
 579                     (setq value (decode-char '=ucs value)))
 580                    ;; ((consp value)
 581                    ;;  (setq value (or (find-char value)
 582                    ;;                  value)))
 583                    )
 584              (char-db-insert-char-spec value readable)
 585              (insert line-breaking)
 586              (setq separator ""))
 587             ((eq name :radical)
 588              (insert (format "%s%s\t%d ; %c%s"
 589                              separator
 590                              name value
 591                              (ideographic-radical value)
 592                              line-breaking))
 593              (setq separator ""))
 594             (t
 595              (insert (format "%s%s\t%S" separator name value))
 596              (setq separator line-breaking)))
 597       ))
 598   (insert ")"))
 599
 600 (defun char-db-decode-isolated-char (ccs code-point)
 601   (let (ret)
 602     (setq ret
 603           (cond ((eq ccs 'arabic-iso8859-6)
 604                  (decode-char ccs code-point))
 605                 ;; ((eq ccs '=gt)
 606                 ;;  (decode-builtin-char '==gt code-point))
 607                 ((and (memq ccs '(=gt-pj-1
 608                                   =gt-pj-2
 609                                   =gt-pj-3
 610                                   =gt-pj-4
 611                                   =gt-pj-5
 612                                   =gt-pj-6
 613                                   =gt-pj-7
 614                                   =gt-pj-8
 615                                   =gt-pj-9
 616                                   =gt-pj-10
 617                                   =gt-pj-11))
 618                       (setq ret (decode-char ccs code-point))
 619                       (setq ret (encode-char ret '=gt 'defined-only)))
 620                  (decode-builtin-char '=gt ret))
 621                 (t
 622                  (decode-builtin-char ccs code-point))))
 623     (cond ((and (<= 0 (char-int ret))
 624                 (<= (char-int ret) #x1F))
 625            (decode-char '=ucs (+ #x2400 (char-int ret))))
 626           ((= (char-int ret) #x7F)
 627            ?\u2421)
 628           (t ret))))
 629
 630 (defvar char-db-convert-obsolete-format t)
 631
 632 (defun char-db-insert-ccs-feature (name value line-breaking)
 633   (cond
 634    ((integerp value)
 635     (insert
 636      (format
 637       (cond
 638        ((memq name '(==shinjigen
 639                      =shinjigen
 640                      =shinjigen@1ed ==shinjigen@1ed
 641                      =shinjigen@rev ==shinjigen@rev
 642                      =shinjigen/+p@rev ==shinjigen/+p@rev
 643                      ===daikanwa/ho ==daikanwa/ho
 644                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 645         "(%-18s .  %04d)\t; %c")
 646        ((eq name '=shinjigen@1ed/24pr)
 647         "(%-18s . %04d)\t; %c")
 648        ((or (memq name '(===daikanwa
 649                          ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 650                          =daikanwa@rev1 =daikanwa@rev2
 651                          =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 652                          =>>daikanwa/+p
 653                          =daikanwa/+2p =>>daikanwa/+2p
 654                          =gt ==gt ===gt
 655                          =>>gt =+>gt =>gt
 656                          =gt-k ==gt-k ===gt-k
 657                          =>>gt-k =>gt-k
 658                          =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 659                          =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 660                          =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 661                          =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 662                          =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 663                          =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 664                          =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 665                          =>>adobe-japan1-0 =+>adobe-japan1-0
 666                          =>>adobe-japan1-1 =+>adobe-japan1-1
 667                          =>>adobe-japan1-2 =+>adobe-japan1-2
 668                          =>>adobe-japan1-3 =+>adobe-japan1-3
 669                          =>>adobe-japan1-4 =+>adobe-japan1-4
 670                          =>>adobe-japan1-5 =+>adobe-japan1-5
 671                          =>>adobe-japan1-6 =+>adobe-japan1-6
 672                          =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 673                          =zinbun-oracle =>zinbun-oracle))
 674             ;; (string-match "^=adobe-" (symbol-name name))
 675             )
 676         "(%-18s . %05d)\t; %c")
 677        ((memq name '(=hanyo-denshi/ks
 678                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 679                      =>>hanyo-denshi/ks
 680                      =koseki
 681                      =mj ==mj ===mj =>>mj
 682                      =zihai mojikyo))
 683         "(%-18s . %06d)\t; %c")
 684        ((>= (charset-dimension name) 2)
 685         "(%-18s . #x%04X)\t; %c")
 686        (t
 687         "(%-18s . #x%02X)\t; %c"))
 688       name
 689       (if (= (charset-iso-graphic-plane name) 1)
 690           (logior value
 691                   (cond ((= (charset-dimension name) 1)
 692                          #x80)
 693                         ((= (charset-dimension name) 2)
 694                          #x8080)
 695                         ((= (charset-dimension name) 3)
 696                          #x808080)
 697                         (t 0)))
 698         value)
 699       (char-db-decode-isolated-char name value)))
 700     (if (and (= (charset-chars name) 94)
 701              (= (charset-dimension name) 2))
 702         (insert (format " [%02d-%02d]"
 703                         (- (lsh value -8) 32)
 704                         (- (logand value 255) 32))))
 705     )
 706    (t
 707     (insert (format "(%-18s . %s)" name value))
 708     ))
 709   (insert line-breaking))
 710
 711 (defun char-db-insert-relation-feature (char name value line-breaking
 712                                              ccss readable)
 713   (insert (format "(%-18s%s " name line-breaking))
 714   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 715         separator cell sources required-features
 716         ret)
 717     (while (consp value)
 718       (setq cell (car value))
 719       (if (integerp cell)
 720           (setq cell (decode-char '=ucs cell)))
 721       (cond
 722        ((eq name '->subsumptive)
 723         (when (or (not (some (lambda (atr)
 724                                (get-char-attribute cell atr))
 725                              char-db-ignored-attributes))
 726                   (some (lambda (ccs)
 727                           (encode-char cell ccs 'defined-only))
 728                         ccss))
 729           (if separator
 730               (insert lbs))
 731           (let ((char-db-ignored-attributes
 732                  (cons '<-subsumptive
 733                        char-db-ignored-attributes)))
 734             (insert-char-attributes cell readable nil nil 'for-sub-node))
 735           (setq separator lbs))
 736         )
 737        ((characterp cell)
 738         (setq sources
 739               (get-char-attribute
 740                char (intern (format "%s*sources" name))))
 741         (setq required-features nil)
 742         (dolist (source sources)
 743           (cond
 744            ((memq source '(JP
 745                            JP/Jouyou
 746                            shinjigen shinjigen@1ed shinjigen@rev))
 747             (setq required-features
 748                   (union required-features
 749                          '(=jis-x0208
 750                            =jis-x0208@1990
 751                            =jis-x0213-1@2000
 752                            =jis-x0213-1@2004
 753                            =jis-x0213-2
 754                            =jis-x0212
 755                            =jis-x0208@1983
 756                            =jis-x0208@1978
 757                            =shinjigen
 758                            =shinjigen@1ed
 759                            =shinjigen@rev
 760                            =shinjigen/+p@rev))))
 761            ((eq source 'CN)
 762             (setq required-features
 763                   (union required-features
 764                          '(=gb2312
 765                            =gb12345
 766                            =iso-ir165)))))
 767           (cond
 768            ((find-charset (setq ret (intern (format "=%s" source))))
 769             (setq required-features
 770                   (cons ret required-features)))
 771            (t (setq required-features
 772                     (cons source required-features)))))
 773         (cond ((string-match "@JP" (symbol-name name))
 774                (setq required-features
 775                      (union required-features
 776                             '(=jis-x0208
 777                               =jis-x0208@1990
 778                               =jis-x0213-1-2000
 779                               =jis-x0213-2-2000
 780                               =jis-x0212
 781                               =jis-x0208@1983
 782                               =jis-x0208@1978))))
 783               ((string-match "@CN" (symbol-name name))
 784                (setq required-features
 785                      (union required-features
 786                             '(=gb2312
 787                               =gb12345
 788                               =iso-ir165)))))
 789         (if separator
 790             (insert lbs))
 791         (if readable
 792             (insert (format "%S" cell))
 793           (char-db-insert-char-spec cell readable
 794                                     nil
 795                                     required-features))
 796         (setq separator lbs))
 797        ((consp cell)
 798         (if separator
 799             (insert lbs))
 800         (if (consp (car cell))
 801             (char-db-insert-char-spec cell readable)
 802           (char-db-insert-char-reference cell readable))
 803         (setq separator lbs))
 804        (t
 805         (if separator
 806             (insert separator))
 807         (insert (prin1-to-string cell))
 808         (setq separator " ")))
 809       (setq value (cdr value)))
 810     (insert ")")
 811     (insert line-breaking)))
 812
 813 (defun insert-char-attributes (char &optional readable attributes column
 814                                     for-sub-node)
 815   (unless column
 816     (setq column (current-column)))
 817   (let (name value ; has-long-ccs-name
 818         rest
 819         radical strokes
 820         (line-breaking
 821          (concat "\n" (make-string (1+ column) ?\ )))
 822         lbs cell separator ret
 823         key al cal
 824         dest-ccss ; sources required-features
 825         ccss)
 826     (let (atr-d)
 827       (setq attributes
 828             (sort (if attributes
 829                       (if (consp attributes)
 830                           (progn
 831                             (dolist (name attributes)
 832                               (unless (memq name char-db-ignored-attributes)
 833                                 (if (find-charset name)
 834                                     (push name ccss))
 835                                 (push name atr-d)))
 836                             atr-d))
 837                     (dolist (name (char-attribute-list))
 838                       (unless (memq name char-db-ignored-attributes)
 839                         (if (find-charset name)
 840                             (push name ccss))
 841                         (push name atr-d)))
 842                     atr-d)
 843                   #'char-attribute-name<)))
 844     (insert "(")
 845     (when (memq '<-subsumptive attributes)
 846       (when (or readable (not for-sub-node))
 847         (when (setq value (get-char-attribute char '<-subsumptive))
 848           (char-db-insert-relation-feature char '<-subsumptive value
 849                                            line-breaking
 850                                            ccss readable)))
 851       (setq attributes (delq '<-subsumptive attributes)))
 852     (when (and (memq '<-denotational attributes)
 853                (setq value (get-char-attribute char '<-denotational)))
 854       (char-db-insert-relation-feature char '<-denotational value
 855                                        line-breaking
 856                                        ccss readable)
 857       (setq attributes (delq '<-denotational attributes)))
 858     (when (and (memq 'name attributes)
 859                (setq value (get-char-attribute char 'name)))
 860       (insert (format
 861                (if (> (+ (current-column) (length value)) 48)
 862                    "(name . %S)%s"
 863                  "(name               . %S)%s")
 864                value line-breaking))
 865       (setq attributes (delq 'name attributes))
 866       )
 867     (when (and (memq 'name* attributes)
 868                (setq value (get-char-attribute char 'name*)))
 869       (insert (format
 870                (if (> (+ (current-column) (length value)) 48)
 871                    "(name* . %S)%s"
 872                  "(name*              . %S)%s")
 873                value line-breaking))
 874       (setq attributes (delq 'name* attributes))
 875       )
 876     (when (and (memq 'script attributes)
 877                (setq value (get-char-attribute char 'script)))
 878       (insert (format "(script\t\t%s)%s"
 879                       (mapconcat (function prin1-to-string)
 880                                  value " ")
 881                       line-breaking))
 882       (setq attributes (delq 'script attributes))
 883       )
 884     (dolist (name '(=>ucs =>ucs*))
 885       (when (and (memq name attributes)
 886                  (setq value (get-char-attribute char name)))
 887         (insert (format "(%-18s . #x%04X)\t; %c%s"
 888                         name value (decode-char '=ucs value)
 889                         line-breaking))
 890         (setq attributes (delq name attributes))))
 891     (dolist (name '(=>ucs@gb =>ucs@big5))
 892       (when (and (memq name attributes)
 893                  (setq value (get-char-attribute char name)))
 894         (insert (format "(%-18s . #x%04X)\t; %c%s"
 895                         name value
 896                         (decode-char (intern
 897                                       (concat "="
 898                                               (substring
 899                                                (symbol-name name) 2)))
 900                                      value)
 901                         line-breaking))
 902         (setq attributes (delq name attributes))
 903         ))
 904     ;; (dolist (name '(=>daikanwa))
 905     ;;   (when (and (memq name attributes)
 906     ;;              (setq value (get-char-attribute char name)))
 907     ;;     (insert
 908     ;;      (if (integerp value)
 909     ;;          (format "(%-18s . %05d)\t; %c%s"
 910     ;;                  name value (decode-char '=daikanwa value)
 911     ;;                  line-breaking)
 912     ;;        (format "(%-18s %s)\t; %c%s"
 913     ;;                name
 914     ;;                (mapconcat (function prin1-to-string)
 915     ;;                           value " ")
 916     ;;                (char-representative-of-daikanwa char)
 917     ;;                line-breaking)))
 918     ;;     (setq attributes (delq name attributes))))
 919     (when (and (memq 'general-category attributes)
 920                (setq value (get-char-attribute char 'general-category)))
 921       (insert (format
 922                "(general-category\t%s) ; %s%s"
 923                (mapconcat (lambda (cell)
 924                             (format "%S" cell))
 925                           value " ")
 926                (cond ((rassoc value unidata-normative-category-alist)
 927                       "Normative Category")
 928                      ((rassoc value unidata-informative-category-alist)
 929                       "Informative Category")
 930                      (t
 931                       "Unknown Category"))
 932                line-breaking))
 933       (setq attributes (delq 'general-category attributes))
 934       )
 935     (when (and (memq 'bidi-category attributes)
 936                (setq value (get-char-attribute char 'bidi-category)))
 937       (insert (format "(bidi-category\t. %S)%s"
 938                       value
 939                       line-breaking))
 940       (setq attributes (delq 'bidi-category attributes))
 941       )
 942     (unless (or (not (memq 'mirrored attributes))
 943                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
 944                     'empty))
 945       (insert (format "(mirrored\t\t. %S)%s"
 946                       value
 947                       line-breaking))
 948       (setq attributes (delq 'mirrored attributes))
 949       )
 950     (cond
 951      ((and (memq 'decimal-digit-value attributes)
 952            (setq value (get-char-attribute char 'decimal-digit-value)))
 953       (insert (format "(decimal-digit-value . %S)%s"
 954                       value
 955                       line-breaking))
 956       (setq attributes (delq 'decimal-digit-value attributes))
 957       (when (and (memq 'digit-value attributes)
 958                  (setq value (get-char-attribute char 'digit-value)))
 959         (insert (format "(digit-value\t . %S)%s"
 960                         value
 961                         line-breaking))
 962         (setq attributes (delq 'digit-value attributes))
 963         )
 964       (when (and (memq 'numeric-value attributes)
 965                  (setq value (get-char-attribute char 'numeric-value)))
 966         (insert (format "(numeric-value\t . %S)%s"
 967                         value
 968                         line-breaking))
 969         (setq attributes (delq 'numeric-value attributes))
 970         )
 971       )
 972      (t
 973       (when (and (memq 'digit-value attributes)
 974                  (setq value (get-char-attribute char 'digit-value)))
 975         (insert (format "(digit-value\t. %S)%s"
 976                         value
 977                         line-breaking))
 978         (setq attributes (delq 'digit-value attributes))
 979         )
 980       (when (and (memq 'numeric-value attributes)
 981                  (setq value (get-char-attribute char 'numeric-value)))
 982         (insert (format "(numeric-value\t. %S)%s"
 983                         value
 984                         line-breaking))
 985         (setq attributes (delq 'numeric-value attributes))
 986         )))
 987     (when (and (memq 'iso-10646-comment attributes)
 988                (setq value (get-char-attribute char 'iso-10646-comment)))
 989       (insert (format "(iso-10646-comment\t. %S)%s"
 990                       value
 991                       line-breaking))
 992       (setq attributes (delq 'iso-10646-comment attributes))
 993       )
 994     (when (and (memq 'morohashi-daikanwa attributes)
 995                (setq value (get-char-attribute char 'morohashi-daikanwa)))
 996       (insert (format "(morohashi-daikanwa\t%s)%s"
 997                       (mapconcat (function prin1-to-string) value " ")
 998                       line-breaking))
 999       (setq attributes (delq 'morohashi-daikanwa attributes))
1000       )
1001     (setq radical nil
1002           strokes nil)
1003     (when (and (memq 'ideographic-radical attributes)
1004                (setq value (get-char-attribute char 'ideographic-radical)))
1005       (setq radical value)
1006       (insert (format "(ideographic-radical . %S)\t; %c%s"
1007                       radical
1008                       (ideographic-radical radical)
1009                       line-breaking))
1010       (setq attributes (delq 'ideographic-radical attributes))
1011       )
1012     (when (and (memq 'shuowen-radical attributes)
1013                (setq value (get-char-attribute char 'shuowen-radical)))
1014       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1015                       value
1016                       (shuowen-radical value)
1017                       line-breaking))
1018       (setq attributes (delq 'shuowen-radical attributes))
1019       )
1020     (let (key)
1021       (dolist (domain
1022                (append
1023                 char-db-feature-domains
1024                 (let (dest domain)
1025                   (dolist (feature (char-attribute-list))
1026                     (setq feature (symbol-name feature))
1027                     (when (string-match
1028                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1029                            feature)
1030                       (setq domain (intern (match-string 2 feature)))
1031                      (unless (memq domain dest)
1032                        (setq dest (cons domain dest)))))
1033                   (sort dest #'string<))))
1034         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1035         (when (and (memq key attributes)
1036                    (setq value (get-char-attribute char key)))
1037           (setq radical value)
1038           (insert (format "(%s . %S)\t; %c%s"
1039                           key
1040                           radical
1041                           (ideographic-radical radical)
1042                           line-breaking))
1043           (setq attributes (delq key attributes))
1044           )
1045         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1046         (when (and (memq key attributes)
1047                    (setq value (get-char-attribute char key)))
1048           (setq strokes value)
1049           (insert (format "(%s . %S)%s"
1050                           key
1051                           strokes
1052                           line-breaking))
1053           (setq attributes (delq key attributes))
1054           )
1055         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1056         (when (and (memq key attributes)
1057                    (setq value (get-char-attribute char key)))
1058           (insert (format "(%s       . %S)%s"
1059                           key
1060                           value
1061                           line-breaking))
1062           (setq attributes (delq key attributes))
1063           )
1064         (dolist (feature '(ideographic-radical
1065                            ideographic-strokes
1066                            total-strokes))
1067           (setq key (intern (format "%s@%s*sources" feature domain)))
1068           (when (and (memq key attributes)
1069                      (setq value (get-char-attribute char key)))
1070             (insert (format "(%s%s" key line-breaking))
1071             (dolist (cell value)
1072               (insert (format " %s" cell)))
1073             (insert ")")
1074             (insert line-breaking)
1075             (setq attributes (delq key attributes))
1076             ))
1077         ))
1078     (when (and (memq 'ideographic-strokes attributes)
1079                (setq value (get-char-attribute char 'ideographic-strokes)))
1080       (setq strokes value)
1081       (insert (format "(ideographic-strokes . %S)%s"
1082                       strokes
1083                       line-breaking))
1084       (setq attributes (delq 'ideographic-strokes attributes))
1085       )
1086     (when (and (memq 'kangxi-radical attributes)
1087                (setq value (get-char-attribute char 'kangxi-radical)))
1088       (unless (eq value radical)
1089         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1090                         value
1091                         (ideographic-radical value)
1092                         line-breaking))
1093         (or radical
1094             (setq radical value)))
1095       (setq attributes (delq 'kangxi-radical attributes))
1096       )
1097     (when (and (memq 'kangxi-strokes attributes)
1098                (setq value (get-char-attribute char 'kangxi-strokes)))
1099       (unless (eq value strokes)
1100         (insert (format "(kangxi-strokes\t . %S)%s"
1101                         value
1102                         line-breaking))
1103         (or strokes
1104             (setq strokes value)))
1105       (setq attributes (delq 'kangxi-strokes attributes))
1106       )
1107     (when (and (memq 'japanese-radical attributes)
1108                (setq value (get-char-attribute char 'japanese-radical)))
1109       (unless (eq value radical)
1110         (insert (format "(japanese-radical\t . %S)\t; %c%s"
1111                         value
1112                         (ideographic-radical value)
1113                         line-breaking))
1114         (or radical
1115             (setq radical value)))
1116       (setq attributes (delq 'japanese-radical attributes))
1117       )
1118     (when (and (memq 'japanese-strokes attributes)
1119                (setq value (get-char-attribute char 'japanese-strokes)))
1120       (unless (eq value strokes)
1121         (insert (format "(japanese-strokes\t . %S)%s"
1122                         value
1123                         line-breaking))
1124         (or strokes
1125             (setq strokes value)))
1126       (setq attributes (delq 'japanese-strokes attributes))
1127       )
1128     (when (and (memq 'cns-radical attributes)
1129                (setq value (get-char-attribute char 'cns-radical)))
1130       (insert (format "(cns-radical\t . %S)\t; %c%s"
1131                       value
1132                       (ideographic-radical value)
1133                       line-breaking))
1134       (setq attributes (delq 'cns-radical attributes))
1135       )
1136     (when (and (memq 'cns-strokes attributes)
1137                (setq value (get-char-attribute char 'cns-strokes)))
1138       (unless (eq value strokes)
1139         (insert (format "(cns-strokes\t . %S)%s"
1140                         value
1141                         line-breaking))
1142         (or strokes
1143             (setq strokes value)))
1144       (setq attributes (delq 'cns-strokes attributes))
1145       )
1146     ;; (when (and (memq 'shinjigen-1-radical attributes)
1147     ;;            (setq value (get-char-attribute char 'shinjigen-1-radical)))
1148     ;;   (unless (eq value radical)
1149     ;;     (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
1150     ;;                     value
1151     ;;                     (ideographic-radical value)
1152     ;;                     line-breaking))
1153     ;;     (or radical
1154     ;;         (setq radical value)))
1155     ;;   (setq attributes (delq 'shinjigen-1-radical attributes))
1156     ;;   )
1157     (when (and (memq 'ideographic- attributes)
1158                (setq value (get-char-attribute char 'ideographic-)))
1159       (insert "(ideographic-       ")
1160       (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1161             separator nil)
1162       (while (consp value)
1163         (setq cell (car value))
1164         (if (integerp cell)
1165             (setq cell (decode-char '=ucs cell)))
1166         (cond ((characterp cell)
1167                (if separator
1168                    (insert lbs))
1169                (if readable
1170                    (insert (format "%S" cell))
1171                  (char-db-insert-char-spec cell readable))
1172                (setq separator lbs))
1173               ((consp cell)
1174                (if separator
1175                    (insert lbs))
1176                (if (consp (car cell))
1177                    (char-db-insert-char-spec cell readable)
1178                  (char-db-insert-char-reference cell readable))
1179                (setq separator lbs))
1180               (t
1181                (if separator
1182                    (insert separator))
1183                (insert (prin1-to-string cell))
1184                (setq separator " ")))
1185         (setq value (cdr value)))
1186       (insert ")")
1187       (insert line-breaking)
1188       (setq attributes (delq 'ideographic- attributes)))
1189     (when (and (memq 'total-strokes attributes)
1190                (setq value (get-char-attribute char 'total-strokes)))
1191       (insert (format "(total-strokes       . %S)%s"
1192                       value
1193                       line-breaking))
1194       (setq attributes (delq 'total-strokes attributes))
1195       )
1196     (when (and (memq '->ideograph attributes)
1197                (setq value (get-char-attribute char '->ideograph)))
1198       (insert (format "(->ideograph\t%s)%s"
1199                       (mapconcat (lambda (code)
1200                                    (cond ((symbolp code)
1201                                           (symbol-name code))
1202                                          ((integerp code)
1203                                           (format "#x%04X" code))
1204                                          (t
1205                                           (format "%s %S"
1206                                                   line-breaking code))))
1207                                  value " ")
1208                       line-breaking))
1209       (setq attributes (delq '->ideograph attributes))
1210       )
1211     ;; (when (and (memq '->decomposition attributes)
1212     ;;            (setq value (get-char-attribute char '->decomposition)))
1213     ;;   (insert (format "(->decomposition\t%s)%s"
1214     ;;                   (mapconcat (lambda (code)
1215     ;;                                (cond ((symbolp code)
1216     ;;                                       (symbol-name code))
1217     ;;                                      ((characterp code)
1218     ;;                                       (if readable
1219     ;;                                           (format "%S" code)
1220     ;;                                         (format "#x%04X"
1221     ;;                                                 (char-int code))
1222     ;;                                         ))
1223     ;;                                      ((integerp code)
1224     ;;                                       (format "#x%04X" code))
1225     ;;                                      (t
1226     ;;                                       (format "%s%S" line-breaking code))))
1227     ;;                              value " ")
1228     ;;                   line-breaking))
1229     ;;   (setq attributes (delq '->decomposition attributes))
1230     ;;   )
1231     (if (equal (get-char-attribute char '->titlecase)
1232                (get-char-attribute char '->uppercase))
1233         (setq attributes (delq '->titlecase attributes)))
1234     (when (and (memq '->mojikyo attributes)
1235                (setq value (get-char-attribute char '->mojikyo)))
1236       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1237                       value (decode-char 'mojikyo value)
1238                       line-breaking))
1239       (setq attributes (delq '->mojikyo attributes))
1240       )
1241     (when (and (memq 'hanyu-dazidian-vol attributes)
1242                (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1243       (insert (format "(hanyu-dazidian-vol  . %d)%s"
1244                       value line-breaking))
1245       (setq attributes (delq 'hanyu-dazidian-vol attributes))
1246       )
1247     (when (and (memq 'hanyu-dazidian-page attributes)
1248                (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1249       (insert (format "(hanyu-dazidian-page . %d)%s"
1250                       value line-breaking))
1251       (setq attributes (delq 'hanyu-dazidian-page attributes))
1252       )
1253     (when (and (memq 'hanyu-dazidian-char attributes)
1254                (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1255       (insert (format "(hanyu-dazidian-char . %d)%s"
1256                       value line-breaking))
1257       (setq attributes (delq 'hanyu-dazidian-char attributes))
1258       )
1259     (unless readable
1260       (dolist (ignored '(composition
1261                          ->denotational <-subsumptive ->ucs-unified
1262                          ->ideographic-component-forms))
1263         (setq attributes (delq ignored attributes))))
1264     (while attributes
1265       (setq name (car attributes))
1266       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1267                   'value-is-empty)
1268         (cond ((setq ret (find-charset name))
1269                (setq name (charset-name ret))
1270                (when (not (memq name dest-ccss))
1271                  (setq dest-ccss (cons name dest-ccss))
1272                  (char-db-insert-ccs-feature name value line-breaking))
1273                )
1274               ((string-match "^=>ucs@" (symbol-name name))
1275                (insert (format "(%-18s . #x%04X)\t; %c%s"
1276                                name value (decode-char '=ucs value)
1277                                line-breaking))
1278                )
1279               ((eq name 'jisx0208-1978/4X)
1280                (insert (format "(%-18s . #x%04X)%s"
1281                                name value
1282                                line-breaking))
1283                )
1284               ((and
1285                 (not readable)
1286                 (not (eq name '->subsumptive))
1287                 (not (eq name '->uppercase))
1288                 (not (eq name '->lowercase))
1289                 (not (eq name '->titlecase))
1290                 (not (eq name '->canonical))
1291                 (not (eq name '->Bopomofo))
1292                 (not (eq name '->mistakable))
1293                 (not (eq name '->ideographic-variants))
1294                 (null (get-char-attribute
1295                        char (intern (format "%s*sources" name))))
1296                 (not (string-match "\\*sources$" (symbol-name name)))
1297                 (null (get-char-attribute
1298                        char (intern (format "%s*note" name))))
1299                 (not (string-match "\\*note$" (symbol-name name)))
1300                 (or (eq name '<-identical)
1301                     (eq name '<-uppercase)
1302                     (eq name '<-lowercase)
1303                     (eq name '<-titlecase)
1304                     (eq name '<-canonical)
1305                     (eq name '<-ideographic-variants)
1306                     ;; (eq name '<-synonyms)
1307                     (string-match "^<-synonyms" (symbol-name name))
1308                     (eq name '<-mistakable)
1309                     (when (string-match "^->" (symbol-name name))
1310                       (cond
1311                        ((string-match "^->fullwidth" (symbol-name name))
1312                         (not (and (consp value)
1313                                   (characterp (car value))
1314                                   (encode-char
1315                                    (car value) '=ucs 'defined-only)))
1316                         )
1317                        (t)))
1318                     ))
1319                )
1320               ((or (eq name 'ideographic-structure)
1321                    (eq name 'ideographic-combination)
1322                    (eq name 'ideographic-)
1323                    (eq name '=decomposition)
1324                    (char-feature-base-name= '=decomposition name)
1325                    (char-feature-base-name= '=>decomposition name)
1326                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1327                    ;;               (symbol-name name))
1328                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1329                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1330                                  (symbol-name name))
1331                    )
1332                (char-db-insert-relation-feature char name value
1333                                                 line-breaking
1334                                                 ccss readable))
1335               ((memq name '(ideograph=
1336                             original-ideograph-of
1337                             ancient-ideograph-of
1338                             vulgar-ideograph-of
1339                             wrong-ideograph-of
1340                             ;; simplified-ideograph-of
1341                             ideographic-variants
1342                             ;; ideographic-different-form-of
1343                             ))
1344                (insert (format "(%-18s%s " name line-breaking))
1345                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1346                      separator nil)
1347                (while (consp value)
1348                  (setq cell (car value))
1349                  (if (and (consp cell)
1350                           (consp (car cell)))
1351                      (progn
1352                        (if separator
1353                            (insert lbs))
1354                        (char-db-insert-alist cell readable)
1355                        (setq separator lbs))
1356                    (if separator
1357                        (insert separator))
1358                    (insert (prin1-to-string cell))
1359                    (setq separator " "))
1360                  (setq value (cdr value)))
1361                (insert ")")
1362                (insert line-breaking))
1363               ((consp value)
1364                (insert (format "(%-18s " name))
1365                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1366                      separator nil)
1367                (while (consp value)
1368                  (setq cell (car value))
1369                  (if (and (consp cell)
1370                           (consp (car cell))
1371                           (setq ret (condition-case nil
1372                                         (find-char cell)
1373                                       (error nil))))
1374                      (progn
1375                        (setq rest cell
1376                              al nil
1377                              cal nil)
1378                        (while rest
1379                          (setq key (car (car rest)))
1380                          (if (find-charset key)
1381                              (setq cal (cons key cal))
1382                            (setq al (cons key al)))
1383                          (setq rest (cdr rest)))
1384                        (if separator
1385                            (insert lbs))
1386                        (insert-char-attributes ret
1387                                                readable
1388                                                al ; cal
1389                                                nil 'for-sub-node)
1390                        (setq separator lbs))
1391                    (setq ret (prin1-to-string cell))
1392                    (if separator
1393                        (if (< (+ (current-column)
1394                                  (length ret)
1395                                  (length separator))
1396                               76)
1397                            (insert separator)
1398                          (insert lbs)))
1399                    (insert ret)
1400                    (setq separator " "))
1401                  (setq value (cdr value)))
1402                (insert ")")
1403                (insert line-breaking))
1404               (t
1405                (insert (format "(%-18s" name))
1406                (setq ret (prin1-to-string value))
1407                (unless (< (+ (current-column)
1408                              (length ret)
1409                              3)
1410                           76)
1411                  (insert line-breaking))
1412                (insert " . " ret ")" line-breaking)
1413                ;; (insert (format "(%-18s . %S)%s"
1414                ;;                 name value
1415                ;;                 line-breaking))
1416                )
1417               ))
1418       (setq attributes (cdr attributes)))
1419     (insert ")")))
1420
1421 (defun insert-char-data (char &optional readable
1422                               attributes)
1423   (save-restriction
1424     (narrow-to-region (point)(point))
1425     (insert "(define-char
1426   '")
1427     (insert-char-attributes char readable attributes)
1428     (insert ")\n")
1429     (goto-char (point-min))
1430     (while (re-search-forward "[ \t]+$" nil t)
1431       (replace-match ""))
1432     ;; from tabify.
1433     (goto-char (point-min))
1434     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1435       (let ((column (current-column))
1436             (indent-tabs-mode t))
1437         (delete-region (match-beginning 0) (point))
1438         (indent-to column)))
1439     (goto-char (point-max))
1440     ;; (tabify (point-min)(point-max))
1441     ))
1442
1443 (defun insert-char-data-with-variant (char &optional printable
1444                                            no-ucs-unified
1445                                            script excluded-script)
1446   (insert-char-data char printable)
1447   (let ((variants (char-variants char))
1448         rest
1449         variant vs ret)
1450     (setq variants (sort variants #'<))
1451     (setq rest variants)
1452     (setq variants (cons char variants))
1453     (while rest
1454       (setq variant (car rest))
1455       (unless (get-char-attribute variant '<-subsumptive)
1456         (if (and (or (null script)
1457                      (null (setq vs (get-char-attribute variant 'script)))
1458                      (memq script vs))
1459                  (or (null excluded-script)
1460                      (null (setq vs (get-char-attribute variant 'script)))
1461                      (not (memq excluded-script vs))))
1462             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1463               (insert-char-data variant printable)
1464               (if (setq ret (char-variants variant))
1465                   (while ret
1466                     (or (memq (car ret) variants)
1467                         ;; (get-char-attribute (car ret) '<-subsumptive)
1468                         (setq rest (nconc rest (list (car ret)))))
1469                     (setq ret (cdr ret)))))))
1470       (setq rest (cdr rest)))))
1471
1472 (defun insert-char-range-data (min max &optional script excluded-script)
1473   (let ((code min)
1474         char)
1475     (while (<= code max)
1476       (setq char (decode-char '=ucs code))
1477       (if (encode-char char '=ucs 'defined-only)
1478           (insert-char-data-with-variant char nil 'no-ucs-unified
1479                                          script excluded-script))
1480       (setq code (1+ code)))))
1481
1482 (defun write-char-range-data-to-file (min max file
1483                                           &optional script excluded-script)
1484   (let ((coding-system-for-write char-db-file-coding-system))
1485     (with-temp-buffer
1486       (insert (format ";; -*- coding: %s -*-\n"
1487                       char-db-file-coding-system))
1488       (insert-char-range-data min max script excluded-script)
1489       (write-region (point-min)(point-max) file))))
1490
1491 (defvar what-character-original-window-configuration)
1492
1493 ;;;###autoload
1494 (defun what-char-definition (char)
1495   (interactive (list (char-after)))
1496   (let ((buf (get-buffer-create "*Character Description*"))
1497         (the-buf (current-buffer))
1498         (win-conf (current-window-configuration)))
1499     (pop-to-buffer buf)
1500     (make-local-variable 'what-character-original-window-configuration)
1501     (setq what-character-original-window-configuration win-conf)
1502     (setq buffer-read-only nil)
1503     (erase-buffer)
1504     (condition-case err
1505         (progn
1506           (insert-char-data-with-variant char 'printable)
1507           (unless (char-attribute-alist char)
1508             (insert (format ";; = %c\n"
1509                             (let* ((rest (split-char char))
1510                                    (ccs (pop rest))
1511                                    (code (pop rest)))
1512                               (while rest
1513                                 (setq code (logior (lsh code 8)
1514                                                    (pop rest))))
1515                               (decode-char ccs code)))))
1516           ;; (char-db-update-comment)
1517           (set-buffer-modified-p nil)
1518           (view-mode the-buf (lambda (buf)
1519                                (set-window-configuration
1520                                 what-character-original-window-configuration)
1521                                ))
1522           (goto-char (point-min)))
1523       (error (progn
1524                (set-window-configuration
1525                 what-character-original-window-configuration)
1526                (signal (car err) (cdr err)))))))
1527
1528
1529 ;;; @ end
1530 ;;;
1531
1532 (provide 'char-db-util)
1533
1534 ;;; char-db-util.el ends here