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 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    ?旨 ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?喜 ?壴 ; 160
  83    ?鼓 ?豈 ?豆 ?豊 ?豐 ?䖒 ?虍 ?虎 ?虤 ?皿 ; 170
  84    ?𠙴 ?去 ?血 ?丶 ?丹 ?青 ?井 ?皀 ?鬯 ?食 ; 180
  85    ?亼 ?會 ?倉 ?入 ?缶 ?矢 ?高 ?冂 ?𩫏 ?京 ; 190
  86    ?亯 ?𣆪 ?畗 ?㐭 ?嗇 ?來 ?麥 ?夊 ?舛 ?䑞 ; 200
  87    ?韋 ?弟 ?夂 ?久 ?桀 ?木 ?東 ?林 ?才 ?叒 ; 210
  88    ?之 ?帀 ?出 ?𣎵 ?生 ?乇 ?𠂹 ?𠌶 ?華 ?𥝌 ; 220
  89    ?稽 ?巢 ?桼 ?束 ?㯻 ?囗 ?員 ?貝 ?邑 ?𨛜 ; 230
  90    ?日 ?旦 ?倝 ?㫃 ?冥 ?晶 ?月 ?有 ?明 ?囧 ; 240
  91    ?夕 ?多 ?毌 ?𢎘 ?𣐺 ?卣 ?齊 ?朿 ?片 ?鼎 ; 250
  92    ?克 ?彔 ?禾 ?秝 ?黍 ?香 ?米 ?毇 ?臼 ?凶 ; 260
  93    ?𣎳 ?林 ?麻 ?尗 ?耑 ?韭 ?瓜 ?瓠 ?宀 ?宮 ; 270
  94    ?呂 ?穴 ?㝱 ?𤕫 ?冖 ?𠔼 ?冃 ?㒳 ?网 ?襾 ; 280
  95    ?巾 ?巿 ?帛 ?白 ?㡀 ?黹 ?人 ?𠤎 ?匕 ?从 ; 290
  96    ])
  97
  98 (defun shuowen-radical (number)
  99   (aref shuowen-radicals (1- number)))
 100
 101 (defvar char-db-file-coding-system 'utf-8-mcs-er)
 102
 103 (defvar char-db-ignored-attributes '(ideographic-products))
 104
 105 (defvar char-db-coded-charset-priority-list
 106   '(ascii
 107     control-1
 108     latin-iso8859-1
 109     latin-iso8859-2
 110     latin-iso8859-3
 111     latin-iso8859-4
 112     latin-iso8859-9
 113     latin-jisx0201
 114     cyrillic-iso8859-5
 115     greek-iso8859-7
 116     thai-tis620
 117     =adobe-japan1-0
 118     =adobe-japan1-1
 119     =adobe-japan1-2
 120     =adobe-japan1-3
 121     =adobe-japan1-4
 122     =adobe-japan1-5
 123     =adobe-japan1-6
 124     =jis-x0208
 125     =jis-x0208@1978
 126     =jis-x0208@1983
 127     japanese-jisx0212
 128     chinese-gb2312
 129     =jis-x0208@1990
 130     chinese-cns11643-1
 131     chinese-cns11643-2
 132     chinese-cns11643-3
 133     chinese-cns11643-4
 134     chinese-cns11643-5
 135     chinese-cns11643-6
 136     chinese-cns11643-7
 137     =jis-x0213-1
 138     =jis-x0213-1@2000
 139     =jis-x0213-1@2004
 140     =jis-x0213-2
 141     korean-ksc5601
 142     chinese-isoir165
 143     katakana-jisx0201
 144     hebrew-iso8859-8
 145     chinese-gb12345
 146     latin-viscii
 147     ethiopic-ucs
 148     =big5-cdp
 149     =hanyo-denshi/ja
 150     =hanyo-denshi/jb
 151     =hanyo-denshi/jc
 152     =hanyo-denshi/jd
 153     =hanyo-denshi/ft
 154     =hanyo-denshi/ia
 155     =hanyo-denshi/ib
 156     =hanyo-denshi/hg
 157     =hanyo-denshi/jt
 158     =hanyo-denshi/ks
 159     =gt
 160     =gt-k
 161     =daikanwa
 162     =daikanwa@rev2
 163     =daikanwa@rev1
 164     =cbeta
 165     ideograph-hanziku-1
 166     ideograph-hanziku-2
 167     ideograph-hanziku-3
 168     ideograph-hanziku-4
 169     ideograph-hanziku-5
 170     ideograph-hanziku-6
 171     ideograph-hanziku-7
 172     ideograph-hanziku-8
 173     ideograph-hanziku-9
 174     ideograph-hanziku-10
 175     ideograph-hanziku-11
 176     ideograph-hanziku-12
 177     =>>>adobe-japan1-0
 178     =>>>adobe-japan1-1
 179     =>>>adobe-japan1-2
 180     =>>>adobe-japan1-3
 181     =>>>adobe-japan1-4
 182     =>>>adobe-japan1-5
 183     =>>>adobe-japan1-6
 184     =>>>jis-x0208
 185     =>>>jis-x0213-1
 186     =>>>jis-x0213-2
 187     =>>>hanyo-denshi/ja
 188     =>>>hanyo-denshi/jb
 189     =>>>hanyo-denshi/jc
 190     =>>>hanyo-denshi/ft
 191     =>>>hanyo-denshi/ib
 192     =>>>hanyo-denshi/hg
 193     =>>>hanyo-denshi/jt
 194     =>>>hanyo-denshi/ks
 195     =>>>gt
 196     =>>adobe-japan1-0
 197     =>>adobe-japan1-1
 198     =>>adobe-japan1-2
 199     =>>adobe-japan1-3
 200     =>>adobe-japan1-4
 201     =>>adobe-japan1-5
 202     =>>adobe-japan1-6
 203     =>>jis-x0208
 204     =>>jis-x0213-1
 205     =>>jis-x0213-1@2000
 206     =>>jis-x0213-1@2004
 207     =>>jis-x0213-2
 208     =>>jis-x0208@1978
 209     =>>hanyo-denshi/ft
 210     =>>hanyo-denshi/jt
 211     =>>hanyo-denshi/ks
 212     =>>gt
 213     =>>daikanwa
 214     =+>jis-x0208
 215     =+>jis-x0213-1
 216     =+>jis-x0213-2
 217     =+>jis-x0208@1978
 218     =+>adobe-japan1
 219     =>jis-x0208
 220     =>jis-x0208@1997
 221     =>jis-x0213-1
 222     =>jis-x0213-1@2000
 223     =>jis-x0213-1@2004
 224     =>jis-x0213-2
 225     ==>ucs@bucs
 226     =>ucs@iso
 227     =>ucs@unicode
 228     =>ucs@jis
 229     =>ucs@cns
 230     =>ucs@ks
 231     =+>ucs@iso
 232     =+>ucs@unicode
 233     =+>ucs@jis
 234     =+>ucs@jis/1990
 235     =+>ucs@cns
 236     =+>ucs@ks
 237     =>>ucs@iso
 238     =>>ucs@unicode
 239     =>>ucs@jis
 240     =>>ucs@cns
 241     =>>>ucs@iso
 242     =>>>ucs@unicode
 243     =ucs@iso
 244     =ucs@unicode
 245     =>>big5-cdp
 246     =>>gt-k
 247     =+>gt
 248     =>gt
 249     =>big5-cdp
 250     =>daikanwa
 251     =>daikanwa/ho
 252     =>cns11643-7
 253     =big5
 254     =big5-eten
 255     =>gt-k
 256     =zinbun-oracle
 257     =>zinbun-oracle
 258     =ruimoku-v6
 259     =>>ruimoku-v6
 260     =jef-china3
 261     =shinjigen))
 262
 263
 264 ;;; @ char-db formatters
 265 ;;;
 266
 267 (defun char-db-make-char-spec (char)
 268   (let (ret char-spec)
 269     (cond ((characterp char)
 270            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 271                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 272                   (setq char-spec (list (cons '=ucs ret)))
 273                   (cond ((setq ret (get-char-attribute char 'name))
 274                          (setq char-spec (cons (cons 'name ret) char-spec))
 275                          )
 276                         ((setq ret (get-char-attribute char 'name*))
 277                          (setq char-spec (cons (cons 'name* ret) char-spec))
 278                          ))
 279                   )
 280                  ((encode-char char '=adobe-japan1 'defined-only)
 281                   (setq char-spec nil)
 282                   (dolist (ccs (charset-list))
 283                     (if (and (or (memq ccs
 284                                        '(=adobe-japan1-0
 285                                          =adobe-japan1-1
 286                                          =adobe-japan1-2
 287                                          =adobe-japan1-3
 288                                          =adobe-japan1-4
 289                                          =adobe-japan1-5
 290                                          =adobe-japan1-6
 291                                          ))
 292                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 293                                  (string-match "=ucs@" (symbol-name ccs))
 294                                  )
 295                              (setq ccs (charset-name ccs))
 296                              (null (assq ccs char-spec))
 297                              (setq ret (encode-char char ccs 'defined-only)))
 298                         (setq char-spec (cons (cons ccs ret) char-spec))))
 299                   )
 300                  ((setq ret
 301                         (catch 'tag
 302                           (let ((rest char-db-coded-charset-priority-list)
 303                                 ccs)
 304                             (while rest
 305                               (setq ccs (charset-name
 306                                          (find-charset (car rest))))
 307                               (if (setq ret
 308                                         (encode-char char ccs
 309                                                      'defined-only))
 310                                   (throw 'tag (cons ccs ret)))
 311                               (setq rest (cdr rest))))))
 312                   (setq char-spec (list ret))
 313                   (dolist (ccs (delq (car ret) (charset-list)))
 314                     (if (and (or (charset-iso-final-char ccs)
 315                                  (memq ccs
 316                                        '(=daikanwa
 317                                          =daikanwa@rev2
 318                                          ;; =gt-k
 319                                          =jis-x0208@1997
 320                                          ))
 321                                  (eq (charset-property ccs 'iso-ir) 177)
 322                                  ;; (string-match "=ucs@" (symbol-name ccs))
 323                                  )
 324                              (setq ccs (charset-name ccs))
 325                              (null (assq ccs char-spec))
 326                              (setq ret (encode-char char ccs 'defined-only)))
 327                         (setq char-spec (cons (cons ccs ret) char-spec))))
 328                   (if (null char-spec)
 329                       (setq char-spec (split-char char)))
 330                   (cond ((setq ret (get-char-attribute char 'name))
 331                          (setq char-spec (cons (cons 'name ret) char-spec))
 332                          )
 333                         ((setq ret (get-char-attribute char 'name*))
 334                          (setq char-spec (cons (cons 'name* ret) char-spec))
 335                          ))
 336                   )
 337                  ((setq ret (get-char-attribute
 338                              char 'ideographic-combination))
 339                   (setq char-spec
 340                         (cons (cons 'ideographic-combination ret)
 341                               char-spec))
 342                   ))
 343            char-spec)
 344           ((consp char)
 345            char))))
 346
 347 (defun char-db-insert-char-spec (char &optional readable column
 348                                       required-features)
 349   (unless column
 350     (setq column (current-column)))
 351   (let (char-spec temp-char)
 352     (setq char-spec (char-db-make-char-spec char))
 353     (unless (or (characterp char) ; char
 354                 (condition-case nil
 355                     (setq char (find-char char-spec))
 356                   (error nil)))
 357       ;; define temporary character
 358       ;;   Current implementation is dirty.
 359       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 360                                          char-spec)))
 361       (remove-char-attribute temp-char 'ideograph-daikanwa)
 362       (setq char temp-char))
 363     (insert-char-attributes char
 364                             readable
 365                             (union (mapcar #'car char-spec)
 366                                    required-features)
 367                             nil 'for-sub-node)
 368     (when temp-char
 369       ;; undefine temporary character
 370       ;;   Current implementation is dirty.
 371       (setq char-spec (char-attribute-alist temp-char))
 372       (while char-spec
 373         (remove-char-attribute temp-char (car (car char-spec)))
 374         (setq char-spec (cdr char-spec))))))
 375
 376 (defun char-db-insert-alist (alist &optional readable column)
 377   (unless column
 378     (setq column (current-column)))
 379   (let ((line-breaking
 380          (concat "\n" (make-string (1+ column) ?\ )))
 381         name value
 382         ret al ; cal
 383         key
 384         lbs cell rest separator)
 385     (insert "(")
 386     (while alist
 387       (setq name (car (car alist))
 388             value (cdr (car alist)))
 389       (cond ((eq name 'char)
 390              (insert "(char . ")
 391              (if (setq ret (condition-case nil
 392                                (find-char value)
 393                              (error nil)))
 394                  (progn
 395                    (setq al nil
 396                          ;; cal nil
 397                          )
 398                    (while value
 399                      (setq key (car (car value)))
 400                      ;; (if (find-charset key)
 401                      ;;     (setq cal (cons key cal))
 402                      (setq al (cons key al))
 403                      ;; )
 404                      (setq value (cdr value)))
 405                    (insert-char-attributes ret
 406                                            readable
 407                                            (or al 'none) ; cal
 408                                            nil 'for-sub-node))
 409                (insert (prin1-to-string value)))
 410              (insert ")")
 411              (insert line-breaking))
 412             ((consp value)
 413              (insert (format "(%-18s " name))
 414              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 415              (while (consp value)
 416                (setq cell (car value))
 417                (if (and (consp cell)
 418                         (consp (car cell))
 419                         (setq ret (condition-case nil
 420                                       (find-char cell)
 421                                     (error nil)))
 422                         )
 423                    (progn
 424                      (setq rest cell
 425                            al nil
 426                            ;; cal nil
 427                            )
 428                      (while rest
 429                        (setq key (car (car rest)))
 430                        ;; (if (find-charset key)
 431                        ;;     (setq cal (cons key cal))
 432                        (setq al (cons key al))
 433                        ;; )
 434                        (setq rest (cdr rest)))
 435                      (if separator
 436                          (insert lbs))
 437                      (insert-char-attributes ret
 438                                              readable
 439                                              al ; cal
 440                                              nil 'for-sub-node)
 441                      (setq separator lbs))
 442                  (if separator
 443                      (insert separator))
 444                  (insert (prin1-to-string cell))
 445                  (setq separator " "))
 446                (setq value (cdr value)))
 447              (insert ")")
 448              (insert line-breaking))
 449             (t
 450              (insert (format "(%-18s . %S)%s"
 451                              name value
 452                              line-breaking))))
 453       (setq alist (cdr alist))))
 454   (insert ")"))
 455
 456 (defun char-db-insert-char-reference (plist &optional readable column)
 457   (unless column
 458     (setq column (current-column)))
 459   (let ((line-breaking
 460          (concat "\n" (make-string (1+ column) ?\ )))
 461         (separator "")
 462         name value)
 463     (insert "(")
 464     (while plist
 465       (setq name (pop plist))
 466       (setq value (pop plist))
 467       (cond ((eq name :char)
 468              (insert separator)
 469              (insert ":char\t")
 470              (cond ((numberp value)
 471                     (setq value (decode-char '=ucs value)))
 472                    ;; ((consp value)
 473                    ;;  (setq value (or (find-char value)
 474                    ;;                  value)))
 475                    )
 476              (char-db-insert-char-spec value readable)
 477              (insert line-breaking)
 478              (setq separator ""))
 479             ((eq name :radical)
 480              (insert (format "%s%s\t%d ; %c%s"
 481                              separator
 482                              name value
 483                              (ideographic-radical value)
 484                              line-breaking))
 485              (setq separator ""))
 486             (t
 487              (insert (format "%s%s\t%S" separator name value))
 488              (setq separator line-breaking)))
 489       ))
 490   (insert ")"))
 491
 492 (defun char-db-decode-isolated-char (ccs code-point)
 493   (let (ret)
 494     (setq ret
 495           (cond ((eq ccs 'arabic-iso8859-6)
 496                  (decode-char ccs code-point))
 497                 ((and (memq ccs '(=gt-pj-1
 498                                   =gt-pj-2
 499                                   =gt-pj-3
 500                                   =gt-pj-4
 501                                   =gt-pj-5
 502                                   =gt-pj-6
 503                                   =gt-pj-7
 504                                   =gt-pj-8
 505                                   =gt-pj-9
 506                                   =gt-pj-10
 507                                   =gt-pj-11))
 508                       (setq ret (decode-char ccs code-point))
 509                       (setq ret (encode-char ret '=gt 'defined-only)))
 510                  (decode-builtin-char '=gt ret))
 511                 (t
 512                  (decode-builtin-char ccs code-point))))
 513     (cond ((and (<= 0 (char-int ret))
 514                 (<= (char-int ret) #x1F))
 515            (decode-char '=ucs (+ #x2400 (char-int ret))))
 516           ((= (char-int ret) #x7F)
 517            ?\u2421)
 518           (t ret))))
 519
 520 (defvar char-db-convert-obsolete-format t)
 521
 522 (defun char-db-insert-ccs-feature (name value line-breaking)
 523   (cond
 524    ((integerp value)
 525     (insert
 526      (format
 527       (cond
 528        ((memq name '(=shinjigen
 529                      =shinjigen@1ed
 530                      =shinjigen@rev =shinjigen/+p@rev
 531                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 532         "(%-18s .  %04d)\t; %c")
 533        ((eq name '=shinjigen@1ed/24pr)
 534         "(%-18s . %04d)\t; %c")
 535        ((or (memq name '(=daikanwa =>>daikanwa =>daikanwa
 536                          =daikanwa@rev1 =daikanwa@rev2
 537                          =daikanwa/+p =>>daikanwa/+p
 538                          =daikanwa/+2p =>>daikanwa/+2p
 539                          =gt =>>>gt =>>gt =+>gt =>gt
 540                          =gt-k =>>gt-k =>gt-k
 541                          =adobe-japan1-0 =>>>adobe-japan1-0
 542                          =adobe-japan1-1 =>>>adobe-japan1-1
 543                          =adobe-japan1-2 =>>>adobe-japan1-2
 544                          =adobe-japan1-3 =>>>adobe-japan1-3
 545                          =adobe-japan1-4 =>>>adobe-japan1-4
 546                          =adobe-japan1-5 =>>>adobe-japan1-5
 547                          =adobe-japan1-6 =>>>adobe-japan1-6
 548                          =>>adobe-japan1-0 =+>adobe-japan1-0
 549                          =>>adobe-japan1-1 =+>adobe-japan1-1
 550                          =>>adobe-japan1-2 =+>adobe-japan1-2
 551                          =>>adobe-japan1-3 =+>adobe-japan1-3
 552                          =>>adobe-japan1-4 =+>adobe-japan1-4
 553                          =>>adobe-japan1-5 =+>adobe-japan1-5
 554                          =>>adobe-japan1-6 =+>adobe-japan1-6
 555                          =cbeta =>>cbeta
 556                          =zinbun-oracle =>zinbun-oracle))
 557             ;; (string-match "^=adobe-" (symbol-name name))
 558             )
 559         "(%-18s . %05d)\t; %c")
 560        ((memq name '(=hanyo-denshi/ks =>>>hanyo-denshi/ks =>>hanyo-denshi/ks mojikyo))
 561         "(%-18s . %06d)\t; %c")
 562        ((>= (charset-dimension name) 2)
 563         "(%-18s . #x%04X)\t; %c")
 564        (t
 565         "(%-18s . #x%02X)\t; %c"))
 566       name
 567       (if (= (charset-iso-graphic-plane name) 1)
 568           (logior value
 569                   (cond ((= (charset-dimension name) 1)
 570                          #x80)
 571                         ((= (charset-dimension name) 2)
 572                          #x8080)
 573                         ((= (charset-dimension name) 3)
 574                          #x808080)
 575                         (t 0)))
 576         value)
 577       (char-db-decode-isolated-char name value)))
 578     (if (and (= (charset-chars name) 94)
 579              (= (charset-dimension name) 2))
 580         (insert (format " [%02d-%02d]"
 581                         (- (lsh value -8) 32)
 582                         (- (logand value 255) 32))))
 583     )
 584    (t
 585     (insert (format "(%-18s . %s)" name value))
 586     ))
 587   (insert line-breaking))
 588
 589 (defun char-db-insert-relation-feature (char name value line-breaking
 590                                              ccss readable)
 591   (insert (format "(%-18s%s " name line-breaking))
 592   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 593         separator cell sources required-features
 594         ret)
 595     (while (consp value)
 596       (setq cell (car value))
 597       (if (integerp cell)
 598           (setq cell (decode-char '=ucs cell)))
 599       (cond
 600        ((eq name '->subsumptive)
 601         (when (or (not (some (lambda (atr)
 602                                (get-char-attribute cell atr))
 603                              char-db-ignored-attributes))
 604                   (some (lambda (ccs)
 605                           (encode-char cell ccs 'defined-only))
 606                         ccss))
 607           (if separator
 608               (insert lbs))
 609           (let ((char-db-ignored-attributes
 610                  (cons '<-subsumptive
 611                        char-db-ignored-attributes)))
 612             (insert-char-attributes cell readable nil nil 'for-sub-node))
 613           (setq separator lbs))
 614         )
 615        ((characterp cell)
 616         (setq sources
 617               (get-char-attribute
 618                char (intern (format "%s*sources" name))))
 619         (setq required-features nil)
 620         (dolist (source sources)
 621           (cond
 622            ((memq source '(JP
 623                            JP/Jouyou
 624                            shinjigen shinjigen@1ed shinjigen@rev))
 625             (setq required-features
 626                   (union required-features
 627                          '(=jis-x0208
 628                            =jis-x0208@1990
 629                            =jis-x0213-1@2000
 630                            =jis-x0213-1@2004
 631                            =jis-x0213-2
 632                            =jis-x0212
 633                            =jis-x0208@1983
 634                            =jis-x0208@1978
 635                            =shinjigen))))
 636            ((eq source 'CN)
 637             (setq required-features
 638                   (union required-features
 639                          '(=gb2312
 640                            =gb12345
 641                            =iso-ir165)))))
 642           (cond
 643            ((find-charset (setq ret (intern (format "=%s" source))))
 644             (setq required-features
 645                   (cons ret required-features)))
 646            (t (setq required-features
 647                     (cons source required-features)))))
 648         (cond ((string-match "@JP" (symbol-name name))
 649                (setq required-features
 650                      (union required-features
 651                             '(=jis-x0208
 652                               =jis-x0208@1990
 653                               =jis-x0213-1-2000
 654                               =jis-x0213-2-2000
 655                               =jis-x0212
 656                               =jis-x0208@1983
 657                               =jis-x0208@1978))))
 658               ((string-match "@CN" (symbol-name name))
 659                (setq required-features
 660                      (union required-features
 661                             '(=gb2312
 662                               =gb12345
 663                               =iso-ir165)))))
 664         (if separator
 665             (insert lbs))
 666         (if readable
 667             (insert (format "%S" cell))
 668           (char-db-insert-char-spec cell readable
 669                                     nil
 670                                     required-features))
 671         (setq separator lbs))
 672        ((consp cell)
 673         (if separator
 674             (insert lbs))
 675         (if (consp (car cell))
 676             (char-db-insert-char-spec cell readable)
 677           (char-db-insert-char-reference cell readable))
 678         (setq separator lbs))
 679        (t
 680         (if separator
 681             (insert separator))
 682         (insert (prin1-to-string cell))
 683         (setq separator " ")))
 684       (setq value (cdr value)))
 685     (insert ")")
 686     (insert line-breaking)))
 687
 688 (defun insert-char-attributes (char &optional readable attributes column
 689                                     for-sub-node)
 690   (unless column
 691     (setq column (current-column)))
 692   (let (name value ; has-long-ccs-name
 693         rest
 694         radical strokes
 695         (line-breaking
 696          (concat "\n" (make-string (1+ column) ?\ )))
 697         lbs cell separator ret
 698         key al cal
 699         dest-ccss ; sources required-features
 700         ccss)
 701     (let (atr-d)
 702       (setq attributes
 703             (sort (if attributes
 704                       (if (consp attributes)
 705                           (progn
 706                             (dolist (name attributes)
 707                               (unless (memq name char-db-ignored-attributes)
 708                                 (if (find-charset name)
 709                                     (push name ccss))
 710                                 (push name atr-d)))
 711                             atr-d))
 712                     (dolist (name (char-attribute-list))
 713                       (unless (memq name char-db-ignored-attributes)
 714                         (if (find-charset name)
 715                             (push name ccss))
 716                         (push name atr-d)))
 717                     atr-d)
 718                   #'char-attribute-name<)))
 719     (insert "(")
 720     (when (memq '<-subsumptive attributes)
 721       (when (or readable (not for-sub-node))
 722         (when (setq value (get-char-attribute char '<-subsumptive))
 723           (char-db-insert-relation-feature char '<-subsumptive value
 724                                            line-breaking
 725                                            ccss readable)))
 726       (setq attributes (delq '<-subsumptive attributes)))
 727     (when (and (memq '<-denotational attributes)
 728                (setq value (get-char-attribute char '<-denotational)))
 729       (char-db-insert-relation-feature char '<-denotational value
 730                                        line-breaking
 731                                        ccss readable)
 732       (setq attributes (delq '<-denotational attributes)))
 733     (when (and (memq 'name attributes)
 734                (setq value (get-char-attribute char 'name)))
 735       (insert (format
 736                (if (> (+ (current-column) (length value)) 48)
 737                    "(name . %S)%s"
 738                  "(name               . %S)%s")
 739                value line-breaking))
 740       (setq attributes (delq 'name attributes))
 741       )
 742     (when (and (memq 'name* attributes)
 743                (setq value (get-char-attribute char 'name*)))
 744       (insert (format
 745                (if (> (+ (current-column) (length value)) 48)
 746                    "(name* . %S)%s"
 747                  "(name*              . %S)%s")
 748                value line-breaking))
 749       (setq attributes (delq 'name* attributes))
 750       )
 751     (when (and (memq 'script attributes)
 752                (setq value (get-char-attribute char 'script)))
 753       (insert (format "(script\t\t%s)%s"
 754                       (mapconcat (function prin1-to-string)
 755                                  value " ")
 756                       line-breaking))
 757       (setq attributes (delq 'script attributes))
 758       )
 759     (dolist (name '(=>ucs =>ucs*))
 760       (when (and (memq name attributes)
 761                  (setq value (get-char-attribute char name)))
 762         (insert (format "(%-18s . #x%04X)\t; %c%s"
 763                         name value (decode-char '=ucs value)
 764                         line-breaking))
 765         (setq attributes (delq name attributes))))
 766     (dolist (name '(=>ucs@gb =>ucs@big5))
 767       (when (and (memq name attributes)
 768                  (setq value (get-char-attribute char name)))
 769         (insert (format "(%-18s . #x%04X)\t; %c%s"
 770                         name value
 771                         (decode-char (intern
 772                                       (concat "="
 773                                               (substring
 774                                                (symbol-name name) 2)))
 775                                      value)
 776                         line-breaking))
 777         (setq attributes (delq name attributes))
 778         ))
 779     ;; (dolist (name '(=>daikanwa))
 780     ;;   (when (and (memq name attributes)
 781     ;;              (setq value (get-char-attribute char name)))
 782     ;;     (insert
 783     ;;      (if (integerp value)
 784     ;;          (format "(%-18s . %05d)\t; %c%s"
 785     ;;                  name value (decode-char '=daikanwa value)
 786     ;;                  line-breaking)
 787     ;;        (format "(%-18s %s)\t; %c%s"
 788     ;;                name
 789     ;;                (mapconcat (function prin1-to-string)
 790     ;;                           value " ")
 791     ;;                (char-representative-of-daikanwa char)
 792     ;;                line-breaking)))
 793     ;;     (setq attributes (delq name attributes))))
 794     (when (and (memq 'general-category attributes)
 795                (setq value (get-char-attribute char 'general-category)))
 796       (insert (format
 797                "(general-category\t%s) ; %s%s"
 798                (mapconcat (lambda (cell)
 799                             (format "%S" cell))
 800                           value " ")
 801                (cond ((rassoc value unidata-normative-category-alist)
 802                       "Normative Category")
 803                      ((rassoc value unidata-informative-category-alist)
 804                       "Informative Category")
 805                      (t
 806                       "Unknown Category"))
 807                line-breaking))
 808       (setq attributes (delq 'general-category attributes))
 809       )
 810     (when (and (memq 'bidi-category attributes)
 811                (setq value (get-char-attribute char 'bidi-category)))
 812       (insert (format "(bidi-category\t. %S)%s"
 813                       value
 814                       line-breaking))
 815       (setq attributes (delq 'bidi-category attributes))
 816       )
 817     (unless (or (not (memq 'mirrored attributes))
 818                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
 819                     'empty))
 820       (insert (format "(mirrored\t\t. %S)%s"
 821                       value
 822                       line-breaking))
 823       (setq attributes (delq 'mirrored attributes))
 824       )
 825     (cond
 826      ((and (memq 'decimal-digit-value attributes)
 827            (setq value (get-char-attribute char 'decimal-digit-value)))
 828       (insert (format "(decimal-digit-value . %S)%s"
 829                       value
 830                       line-breaking))
 831       (setq attributes (delq 'decimal-digit-value attributes))
 832       (when (and (memq 'digit-value attributes)
 833                  (setq value (get-char-attribute char 'digit-value)))
 834         (insert (format "(digit-value\t . %S)%s"
 835                         value
 836                         line-breaking))
 837         (setq attributes (delq 'digit-value attributes))
 838         )
 839       (when (and (memq 'numeric-value attributes)
 840                  (setq value (get-char-attribute char 'numeric-value)))
 841         (insert (format "(numeric-value\t . %S)%s"
 842                         value
 843                         line-breaking))
 844         (setq attributes (delq 'numeric-value attributes))
 845         )
 846       )
 847      (t
 848       (when (and (memq 'digit-value attributes)
 849                  (setq value (get-char-attribute char 'digit-value)))
 850         (insert (format "(digit-value\t. %S)%s"
 851                         value
 852                         line-breaking))
 853         (setq attributes (delq 'digit-value attributes))
 854         )
 855       (when (and (memq 'numeric-value attributes)
 856                  (setq value (get-char-attribute char 'numeric-value)))
 857         (insert (format "(numeric-value\t. %S)%s"
 858                         value
 859                         line-breaking))
 860         (setq attributes (delq 'numeric-value attributes))
 861         )))
 862     (when (and (memq 'iso-10646-comment attributes)
 863                (setq value (get-char-attribute char 'iso-10646-comment)))
 864       (insert (format "(iso-10646-comment\t. %S)%s"
 865                       value
 866                       line-breaking))
 867       (setq attributes (delq 'iso-10646-comment attributes))
 868       )
 869     (when (and (memq 'morohashi-daikanwa attributes)
 870                (setq value (get-char-attribute char 'morohashi-daikanwa)))
 871       (insert (format "(morohashi-daikanwa\t%s)%s"
 872                       (mapconcat (function prin1-to-string) value " ")
 873                       line-breaking))
 874       (setq attributes (delq 'morohashi-daikanwa attributes))
 875       )
 876     (setq radical nil
 877           strokes nil)
 878     (when (and (memq 'ideographic-radical attributes)
 879                (setq value (get-char-attribute char 'ideographic-radical)))
 880       (setq radical value)
 881       (insert (format "(ideographic-radical . %S)\t; %c%s"
 882                       radical
 883                       (ideographic-radical radical)
 884                       line-breaking))
 885       (setq attributes (delq 'ideographic-radical attributes))
 886       )
 887     (when (and (memq 'shuowen-radical attributes)
 888                (setq value (get-char-attribute char 'shuowen-radical)))
 889       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
 890                       value
 891                       (shuowen-radical value)
 892                       line-breaking))
 893       (setq attributes (delq 'shuowen-radical attributes))
 894       )
 895     (let (key)
 896       (dolist (domain
 897                (append
 898                 char-db-feature-domains
 899                 (let (dest domain)
 900                   (dolist (feature (char-attribute-list))
 901                     (setq feature (symbol-name feature))
 902                     (when (string-match
 903                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
 904                            feature)
 905                       (setq domain (intern (match-string 2 feature)))
 906                      (unless (memq domain dest)
 907                        (setq dest (cons domain dest)))))
 908                   (sort dest #'string<))))
 909         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
 910         (when (and (memq key attributes)
 911                    (setq value (get-char-attribute char key)))
 912           (setq radical value)
 913           (insert (format "(%s . %S)\t; %c%s"
 914                           key
 915                           radical
 916                           (ideographic-radical radical)
 917                           line-breaking))
 918           (setq attributes (delq key attributes))
 919           )
 920         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
 921         (when (and (memq key attributes)
 922                    (setq value (get-char-attribute char key)))
 923           (setq strokes value)
 924           (insert (format "(%s . %S)%s"
 925                           key
 926                           strokes
 927                           line-breaking))
 928           (setq attributes (delq key attributes))
 929           )
 930         (setq key (intern (format "%s@%s" 'total-strokes domain)))
 931         (when (and (memq key attributes)
 932                    (setq value (get-char-attribute char key)))
 933           (insert (format "(%s       . %S)%s"
 934                           key
 935                           value
 936                           line-breaking))
 937           (setq attributes (delq key attributes))
 938           )
 939         (dolist (feature '(ideographic-radical
 940                            ideographic-strokes
 941                            total-strokes))
 942           (setq key (intern (format "%s@%s*sources" feature domain)))
 943           (when (and (memq key attributes)
 944                      (setq value (get-char-attribute char key)))
 945             (insert (format "(%s%s" key line-breaking))
 946             (dolist (cell value)
 947               (insert (format " %s" cell)))
 948             (insert ")")
 949             (insert line-breaking)
 950             (setq attributes (delq key attributes))
 951             ))
 952         ))
 953     (when (and (memq 'ideographic-strokes attributes)
 954                (setq value (get-char-attribute char 'ideographic-strokes)))
 955       (setq strokes value)
 956       (insert (format "(ideographic-strokes . %S)%s"
 957                       strokes
 958                       line-breaking))
 959       (setq attributes (delq 'ideographic-strokes attributes))
 960       )
 961     (when (and (memq 'kangxi-radical attributes)
 962                (setq value (get-char-attribute char 'kangxi-radical)))
 963       (unless (eq value radical)
 964         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
 965                         value
 966                         (ideographic-radical value)
 967                         line-breaking))
 968         (or radical
 969             (setq radical value)))
 970       (setq attributes (delq 'kangxi-radical attributes))
 971       )
 972     (when (and (memq 'kangxi-strokes attributes)
 973                (setq value (get-char-attribute char 'kangxi-strokes)))
 974       (unless (eq value strokes)
 975         (insert (format "(kangxi-strokes\t . %S)%s"
 976                         value
 977                         line-breaking))
 978         (or strokes
 979             (setq strokes value)))
 980       (setq attributes (delq 'kangxi-strokes attributes))
 981       )
 982     (when (and (memq 'japanese-radical attributes)
 983                (setq value (get-char-attribute char 'japanese-radical)))
 984       (unless (eq value radical)
 985         (insert (format "(japanese-radical\t . %S)\t; %c%s"
 986                         value
 987                         (ideographic-radical value)
 988                         line-breaking))
 989         (or radical
 990             (setq radical value)))
 991       (setq attributes (delq 'japanese-radical attributes))
 992       )
 993     (when (and (memq 'japanese-strokes attributes)
 994                (setq value (get-char-attribute char 'japanese-strokes)))
 995       (unless (eq value strokes)
 996         (insert (format "(japanese-strokes\t . %S)%s"
 997                         value
 998                         line-breaking))
 999         (or strokes
1000             (setq strokes value)))
1001       (setq attributes (delq 'japanese-strokes attributes))
1002       )
1003     (when (and (memq 'cns-radical attributes)
1004                (setq value (get-char-attribute char 'cns-radical)))
1005       (insert (format "(cns-radical\t . %S)\t; %c%s"
1006                       value
1007                       (ideographic-radical value)
1008                       line-breaking))
1009       (setq attributes (delq 'cns-radical attributes))
1010       )
1011     (when (and (memq 'cns-strokes attributes)
1012                (setq value (get-char-attribute char 'cns-strokes)))
1013       (unless (eq value strokes)
1014         (insert (format "(cns-strokes\t . %S)%s"
1015                         value
1016                         line-breaking))
1017         (or strokes
1018             (setq strokes value)))
1019       (setq attributes (delq 'cns-strokes attributes))
1020       )
1021     (when (and (memq 'shinjigen-1-radical attributes)
1022                (setq value (get-char-attribute char 'shinjigen-1-radical)))
1023       (unless (eq value radical)
1024         (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
1025                         value
1026                         (ideographic-radical value)
1027                         line-breaking))
1028         (or radical
1029             (setq radical value)))
1030       (setq attributes (delq 'shinjigen-1-radical attributes))
1031       )
1032     (when (and (memq 'ideographic- attributes)
1033                (setq value (get-char-attribute char 'ideographic-)))
1034       (insert "(ideographic-       ")
1035       (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1036             separator nil)
1037       (while (consp value)
1038         (setq cell (car value))
1039         (if (integerp cell)
1040             (setq cell (decode-char '=ucs cell)))
1041         (cond ((characterp cell)
1042                (if separator
1043                    (insert lbs))
1044                (if readable
1045                    (insert (format "%S" cell))
1046                  (char-db-insert-char-spec cell readable))
1047                (setq separator lbs))
1048               ((consp cell)
1049                (if separator
1050                    (insert lbs))
1051                (if (consp (car cell))
1052                    (char-db-insert-char-spec cell readable)
1053                  (char-db-insert-char-reference cell readable))
1054                (setq separator lbs))
1055               (t
1056                (if separator
1057                    (insert separator))
1058                (insert (prin1-to-string cell))
1059                (setq separator " ")))
1060         (setq value (cdr value)))
1061       (insert ")")
1062       (insert line-breaking)
1063       (setq attributes (delq 'ideographic- attributes)))
1064     (when (and (memq 'total-strokes attributes)
1065                (setq value (get-char-attribute char 'total-strokes)))
1066       (insert (format "(total-strokes       . %S)%s"
1067                       value
1068                       line-breaking))
1069       (setq attributes (delq 'total-strokes attributes))
1070       )
1071     (when (and (memq '->ideograph attributes)
1072                (setq value (get-char-attribute char '->ideograph)))
1073       (insert (format "(->ideograph\t%s)%s"
1074                       (mapconcat (lambda (code)
1075                                    (cond ((symbolp code)
1076                                           (symbol-name code))
1077                                          ((integerp code)
1078                                           (format "#x%04X" code))
1079                                          (t
1080                                           (format "%s %S"
1081                                                   line-breaking code))))
1082                                  value " ")
1083                       line-breaking))
1084       (setq attributes (delq '->ideograph attributes))
1085       )
1086     ;; (when (and (memq '->decomposition attributes)
1087     ;;            (setq value (get-char-attribute char '->decomposition)))
1088     ;;   (insert (format "(->decomposition\t%s)%s"
1089     ;;                   (mapconcat (lambda (code)
1090     ;;                                (cond ((symbolp code)
1091     ;;                                       (symbol-name code))
1092     ;;                                      ((characterp code)
1093     ;;                                       (if readable
1094     ;;                                           (format "%S" code)
1095     ;;                                         (format "#x%04X"
1096     ;;                                                 (char-int code))
1097     ;;                                         ))
1098     ;;                                      ((integerp code)
1099     ;;                                       (format "#x%04X" code))
1100     ;;                                      (t
1101     ;;                                       (format "%s%S" line-breaking code))))
1102     ;;                              value " ")
1103     ;;                   line-breaking))
1104     ;;   (setq attributes (delq '->decomposition attributes))
1105     ;;   )
1106     (if (equal (get-char-attribute char '->titlecase)
1107                (get-char-attribute char '->uppercase))
1108         (setq attributes (delq '->titlecase attributes)))
1109     (when (and (memq '->mojikyo attributes)
1110                (setq value (get-char-attribute char '->mojikyo)))
1111       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1112                       value (decode-char 'mojikyo value)
1113                       line-breaking))
1114       (setq attributes (delq '->mojikyo attributes))
1115       )
1116     (when (and (memq 'hanyu-dazidian-vol attributes)
1117                (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1118       (insert (format "(hanyu-dazidian-vol  . %d)%s"
1119                       value line-breaking))
1120       (setq attributes (delq 'hanyu-dazidian-vol attributes))
1121       )
1122     (when (and (memq 'hanyu-dazidian-page attributes)
1123                (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1124       (insert (format "(hanyu-dazidian-page . %d)%s"
1125                       value line-breaking))
1126       (setq attributes (delq 'hanyu-dazidian-page attributes))
1127       )
1128     (when (and (memq 'hanyu-dazidian-char attributes)
1129                (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1130       (insert (format "(hanyu-dazidian-char . %d)%s"
1131                       value line-breaking))
1132       (setq attributes (delq 'hanyu-dazidian-char attributes))
1133       )
1134     (unless readable
1135       (dolist (ignored '(composition
1136                          ->denotational <-subsumptive ->ucs-unified
1137                          ->ideographic-component-forms))
1138         (setq attributes (delq ignored attributes))))
1139     (while attributes
1140       (setq name (car attributes))
1141       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1142                   'value-is-empty)
1143         (cond ((setq ret (find-charset name))
1144                (setq name (charset-name ret))
1145                (when (not (memq name dest-ccss))
1146                  (setq dest-ccss (cons name dest-ccss))
1147                  (char-db-insert-ccs-feature name value line-breaking))
1148                )
1149               ((string-match "^=>ucs@" (symbol-name name))
1150                (insert (format "(%-18s . #x%04X)\t; %c%s"
1151                                name value (decode-char '=ucs value)
1152                                line-breaking))
1153                )
1154               ((eq name 'jisx0208-1978/4X)
1155                (insert (format "(%-18s . #x%04X)%s"
1156                                name value
1157                                line-breaking))
1158                )
1159               ((and
1160                 (not readable)
1161                 (not (eq name '->subsumptive))
1162                 (not (eq name '->uppercase))
1163                 (not (eq name '->lowercase))
1164                 (not (eq name '->titlecase))
1165                 (not (eq name '->canonical))
1166                 (not (eq name '->Bopomofo))
1167                 (not (eq name '->mistakable))
1168                 (not (eq name '->ideographic-variants))
1169                 (null (get-char-attribute
1170                        char (intern (format "%s*sources" name))))
1171                 (not (string-match "\\*sources$" (symbol-name name)))
1172                 (null (get-char-attribute
1173                        char (intern (format "%s*note" name))))
1174                 (not (string-match "\\*note$" (symbol-name name)))
1175                 (or (eq name '<-identical)
1176                     (eq name '<-uppercase)
1177                     (eq name '<-lowercase)
1178                     (eq name '<-titlecase)
1179                     (eq name '<-canonical)
1180                     (eq name '<-ideographic-variants)
1181                     ;; (eq name '<-synonyms)
1182                     (string-match "^<-synonyms" (symbol-name name))
1183                     (eq name '<-mistakable)
1184                     (when (string-match "^->" (symbol-name name))
1185                       (cond
1186                        ((string-match "^->fullwidth" (symbol-name name))
1187                         (not (and (consp value)
1188                                   (characterp (car value))
1189                                   (encode-char
1190                                    (car value) '=ucs 'defined-only)))
1191                         )
1192                        (t)))
1193                     ))
1194                )
1195               ((or (eq name 'ideographic-structure)
1196                    (eq name 'ideographic-combination)
1197                    (eq name 'ideographic-)
1198                    (eq name '=decomposition)
1199                    (char-feature-base-name= '=decomposition name)
1200                    (char-feature-base-name= '=>decomposition name)
1201                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1202                    ;;               (symbol-name name))
1203                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1204                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1205                                  (symbol-name name))
1206                    )
1207                (char-db-insert-relation-feature char name value
1208                                                 line-breaking
1209                                                 ccss readable))
1210               ((memq name '(ideograph=
1211                             original-ideograph-of
1212                             ancient-ideograph-of
1213                             vulgar-ideograph-of
1214                             wrong-ideograph-of
1215                             ;; simplified-ideograph-of
1216                             ideographic-variants
1217                             ;; ideographic-different-form-of
1218                             ))
1219                (insert (format "(%-18s%s " name line-breaking))
1220                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1221                      separator nil)
1222                (while (consp value)
1223                  (setq cell (car value))
1224                  (if (and (consp cell)
1225                           (consp (car cell)))
1226                      (progn
1227                        (if separator
1228                            (insert lbs))
1229                        (char-db-insert-alist cell readable)
1230                        (setq separator lbs))
1231                    (if separator
1232                        (insert separator))
1233                    (insert (prin1-to-string cell))
1234                    (setq separator " "))
1235                  (setq value (cdr value)))
1236                (insert ")")
1237                (insert line-breaking))
1238               ((consp value)
1239                (insert (format "(%-18s " name))
1240                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1241                      separator nil)
1242                (while (consp value)
1243                  (setq cell (car value))
1244                  (if (and (consp cell)
1245                           (consp (car cell))
1246                           (setq ret (condition-case nil
1247                                         (find-char cell)
1248                                       (error nil))))
1249                      (progn
1250                        (setq rest cell
1251                              al nil
1252                              cal nil)
1253                        (while rest
1254                          (setq key (car (car rest)))
1255                          (if (find-charset key)
1256                              (setq cal (cons key cal))
1257                            (setq al (cons key al)))
1258                          (setq rest (cdr rest)))
1259                        (if separator
1260                            (insert lbs))
1261                        (insert-char-attributes ret
1262                                                readable
1263                                                al ; cal
1264                                                nil 'for-sub-node)
1265                        (setq separator lbs))
1266                    (setq ret (prin1-to-string cell))
1267                    (if separator
1268                        (if (< (+ (current-column)
1269                                  (length ret)
1270                                  (length separator))
1271                               76)
1272                            (insert separator)
1273                          (insert lbs)))
1274                    (insert ret)
1275                    (setq separator " "))
1276                  (setq value (cdr value)))
1277                (insert ")")
1278                (insert line-breaking))
1279               (t
1280                (insert (format "(%-18s" name))
1281                (setq ret (prin1-to-string value))
1282                (unless (< (+ (current-column)
1283                              (length ret)
1284                              3)
1285                           76)
1286                  (insert line-breaking))
1287                (insert " . " ret ")" line-breaking)
1288                ;; (insert (format "(%-18s . %S)%s"
1289                ;;                 name value
1290                ;;                 line-breaking))
1291                )
1292               ))
1293       (setq attributes (cdr attributes)))
1294     (insert ")")))
1295
1296 (defun insert-char-data (char &optional readable
1297                               attributes)
1298   (save-restriction
1299     (narrow-to-region (point)(point))
1300     (insert "(define-char
1301   '")
1302     (insert-char-attributes char readable attributes)
1303     (insert ")\n")
1304     (goto-char (point-min))
1305     (while (re-search-forward "[ \t]+$" nil t)
1306       (replace-match ""))
1307     ;; from tabify.
1308     (goto-char (point-min))
1309     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1310       (let ((column (current-column))
1311             (indent-tabs-mode t))
1312         (delete-region (match-beginning 0) (point))
1313         (indent-to column)))
1314     (goto-char (point-max))
1315     ;; (tabify (point-min)(point-max))
1316     ))
1317
1318 (defun insert-char-data-with-variant (char &optional printable
1319                                            no-ucs-unified
1320                                            script excluded-script)
1321   (insert-char-data char printable)
1322   (let ((variants (char-variants char))
1323         rest
1324         variant vs ret)
1325     (setq variants (sort variants #'<))
1326     (setq rest variants)
1327     (setq variants (cons char variants))
1328     (while rest
1329       (setq variant (car rest))
1330       (unless (get-char-attribute variant '<-subsumptive)
1331         (if (and (or (null script)
1332                      (null (setq vs (get-char-attribute variant 'script)))
1333                      (memq script vs))
1334                  (or (null excluded-script)
1335                      (null (setq vs (get-char-attribute variant 'script)))
1336                      (not (memq excluded-script vs))))
1337             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1338               (insert-char-data variant printable)
1339               (if (setq ret (char-variants variant))
1340                   (while ret
1341                     (or (memq (car ret) variants)
1342                         ;; (get-char-attribute (car ret) '<-subsumptive)
1343                         (setq rest (nconc rest (list (car ret)))))
1344                     (setq ret (cdr ret)))))))
1345       (setq rest (cdr rest)))))
1346
1347 (defun insert-char-range-data (min max &optional script excluded-script)
1348   (let ((code min)
1349         char)
1350     (while (<= code max)
1351       (setq char (decode-char '=ucs code))
1352       (if (encode-char char '=ucs 'defined-only)
1353           (insert-char-data-with-variant char nil 'no-ucs-unified
1354                                          script excluded-script))
1355       (setq code (1+ code)))))
1356
1357 (defun write-char-range-data-to-file (min max file
1358                                           &optional script excluded-script)
1359   (let ((coding-system-for-write char-db-file-coding-system))
1360     (with-temp-buffer
1361       (insert (format ";; -*- coding: %s -*-\n"
1362                       char-db-file-coding-system))
1363       (insert-char-range-data min max script excluded-script)
1364       (write-region (point-min)(point-max) file))))
1365
1366 (defvar what-character-original-window-configuration)
1367
1368 ;;;###autoload
1369 (defun what-char-definition (char)
1370   (interactive (list (char-after)))
1371   (let ((buf (get-buffer-create "*Character Description*"))
1372         (the-buf (current-buffer))
1373         (win-conf (current-window-configuration)))
1374     (pop-to-buffer buf)
1375     (make-local-variable 'what-character-original-window-configuration)
1376     (setq what-character-original-window-configuration win-conf)
1377     (setq buffer-read-only nil)
1378     (erase-buffer)
1379     (condition-case err
1380         (progn
1381           (insert-char-data-with-variant char 'printable)
1382           (unless (char-attribute-alist char)
1383             (insert (format ";; = %c\n"
1384                             (let* ((rest (split-char char))
1385                                    (ccs (pop rest))
1386                                    (code (pop rest)))
1387                               (while rest
1388                                 (setq code (logior (lsh code 8)
1389                                                    (pop rest))))
1390                               (decode-char ccs code)))))
1391           ;; (char-db-update-comment)
1392           (set-buffer-modified-p nil)
1393           (view-mode the-buf (lambda (buf)
1394                                (set-window-configuration
1395                                 what-character-original-window-configuration)
1396                                ))
1397           (goto-char (point-min)))
1398       (error (progn
1399                (set-window-configuration
1400                 what-character-original-window-configuration)
1401                (signal (car err) (cdr err)))))))
1402
1403
1404 ;;; @ end
1405 ;;;
1406
1407 (provide 'char-db-util)
1408
1409 ;;; char-db-util.el ends here