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