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

   1 ;;; char-db-util.el --- Character Database utility -*- coding: utf-8-er; -*-
   2
   3 ;; Copyright (C) 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006,
   4 ;;   2007, 2008, 2009, 2010, 2011, 2012, 2013 MORIOKA Tomohiko.
   5
   6 ;; Author: MORIOKA Tomohiko <tomo@kanji.zinbun.kyoto-u.ac.jp>
   7 ;; Keywords: CHISE, Character Database, ISO/IEC 10646, UCS, Unicode, MULE.
   8
   9 ;; This file is part of XEmacs CHISE.
  10
  11 ;; XEmacs CHISE is free software; you can redistribute it and/or
  12 ;; modify it under the terms of the GNU General Public License as
  13 ;; published by the Free Software Foundation; either version 2, or (at
  14 ;; your option) any later version.
  15
  16 ;; XEmacs CHISE is distributed in the hope that it will be useful,
  17 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  18 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  19 ;; General Public License for more details.
  20
  21 ;; You should have received a copy of the GNU General Public License
  22 ;; along with XEmacs CHISE; see the file COPYING.  If not, write to
  23 ;; the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  24 ;; Boston, MA 02111-1307, USA.
  25
  26 ;;; Code:
  27
  28 (require 'chise-subr)
  29 (require 'ideograph-subr)
  30
  31 (defconst unidata-normative-category-alist
  32   '(("Lu" letter        uppercase)
  33     ("Ll" letter        lowercase)
  34     ("Lt" letter        titlecase)
  35     ("Mn" mark          non-spacing)
  36     ("Mc" mark          spacing-combining)
  37     ("Me" mark          enclosing)
  38     ("Nd" number        decimal-digit)
  39     ("Nl" number        letter)
  40     ("No" number        other)
  41     ("Zs" separator     space)
  42     ("Zl" separator     line)
  43     ("Zp" separator     paragraph)
  44     ("Cc" other         control)
  45     ("Cf" other         format)
  46     ("Cs" other         surrogate)
  47     ("Co" other         private-use)
  48     ("Cn" other         not-assigned)))
  49
  50 (defconst unidata-informative-category-alist
  51   '(("Lm" letter        modifier)
  52     ("Lo" letter        other)
  53     ("Pc" punctuation   connector)
  54     ("Pd" punctuation   dash)
  55     ("Ps" punctuation   open)
  56     ("Pe" punctuation   close)
  57     ("Pi" punctuation   initial-quote)
  58     ("Pf" punctuation   final-quote)
  59     ("Po" punctuation   other)
  60     ("Sm" symbol        math)
  61     ("Sc" symbol        currency)
  62     ("Sk" symbol        modifier)
  63     ("So" symbol        other)
  64     ))
  65
  66 (defconst shuowen-radicals
  67   [?一 ?上 ?示 ?三 ?王 ?玉 ?玨 ?气 ?士 ?丨 ; 010
  68    ?屮 ?艸 ?蓐 ?茻 ?小 ?八 ?釆 ?半 ?牛 ?犛 ; 020
  69    ?告 ?口 ?凵 ?吅 ?哭 ?走 ?止 ?癶 ?步 ?此 ; 030
  70    ?正 ?是 ?辵 ?彳 ?廴 ?㢟 ?行 ?齒 ?牙 ?足 ; 040
  71    ?疋 ?品 ?龠 ?冊 ?㗊 ?舌 ?干 ?谷 ?只 ?㕯 ; 050
  72    ?句 ?丩 ?古 ?十 ?卅 ?言 ?誩 ?音 ?䇂 ?丵 ; 060
  73    ?菐 ?𠬞 ?𠬜 ?共 ?異 ?舁 ?𦥑 ?䢅 ?爨 ?革 ; 070
  74    ?鬲 ?䰜 ?爪 ?𠃨 ?鬥 ?又 ?𠂇 ?㕜 ?支 ?𦘒 ; 080
  75    ?聿 ?畫 ?隶 ?臤 ?臣 ?殳 ?殺 ?𠘧 ?寸 ?皮 ; 090
  76    ?㼱 ?攴 ?敎 ?卜 ?用 ?爻 ?㸚 ?𥄎 ?目 ?䀠 ; 100
  77    ?眉 ?盾 ?自 ?白 ?鼻 ?皕 ?習 ?羽 ?隹 ?奞 ; 110
  78    ?萑 ?𦫳 ?苜 ?羊 ?羴 ?瞿 ?雔 ?雥 ?鳥 ?烏 ; 120
  79    ?𠦒 ?冓 ?幺 ?𢆶 ?叀 ?玄 ?予 ?放 ?𠬪 ?𣦼 ; 130
  80    ?歺 ?死 ?冎 ?骨 ?肉 ?筋 ?刀 ?刃 ?㓞 ?丰 ; 140
  81    ?耒 ?𧢲 ?竹 ?箕 ?丌 ?左 ?工 ?㠭 ?巫 ?甘 ; 150
  82    ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?旨 ?喜 ?壴 ; 160
  83 ;  ?旨 ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?喜 ?壴 ; 160
  84    ?鼓 ?豈 ?豆 ?豊 ?豐 ?䖒 ?虍 ?虎 ?虤 ?皿 ; 170
  85    ?𠙴 ?去 ?血 ?丶 ?丹 ?青 ?井 ?皀 ?鬯 ?食 ; 180
  86    ?亼 ?會 ?倉 ?入 ?缶 ?矢 ?高 ?冂 ?𩫏 ?京 ; 190
  87    ?亯 ?𣆪 ?畗 ?㐭 ?嗇 ?來 ?麥 ?夊 ?舛 ?䑞 ; 200
  88    ?韋 ?弟 ?夂 ?久 ?桀 ?木 ?東 ?林 ?才 ?叒 ; 210
  89    ?之 ?帀 ?出 ?𣎵 ?生 ?乇 ?𠂹 ?𠌶 ?華 ?𥝌 ; 220
  90    ?稽 ?巢 ?桼 ?束 ?㯻 ?囗 ?員 ?貝 ?邑 ?𨛜 ; 230
  91    ?日 ?旦 ?倝 ?㫃 ?冥 ?晶 ?月 ?有 ?明 ?囧 ; 240
  92    ?夕 ?多 ?毌 ?𢎘 ?𣐺 ?卣 ?齊 ?朿 ?片 ?鼎 ; 250
  93    ?克 ?彔 ?禾 ?秝 ?黍 ?香 ?米 ?毇 ?臼 ?凶 ; 260
  94    ?𣎳 ?林 ?麻 ?尗 ?耑 ?韭 ?瓜 ?瓠 ?宀 ?宮 ; 270
  95    ?呂 ?穴 ?㝱 ?𤕫 ?冖 ?𠔼 ?冃 ?㒳 ?网 ?襾 ; 280
  96    ?巾 ?巿 ?帛 ?白 ?㡀 ?黹 ?人 ?𠤎 ?匕 ?从 ; 290
  97    ?比 ?北 ?丘 ?㐺 ?𡈼 ?重 ?臥 ?身 ?㐆 ?衣 ; 300
  98    ])
  99
 100 (defun shuowen-radical (number)
 101   (aref shuowen-radicals (1- number)))
 102
 103 (defvar char-db-file-coding-system 'utf-8-mcs-er)
 104
 105 (defvar char-db-ignored-attributes '(ideographic-products))
 106
 107 (defvar char-db-coded-charset-priority-list
 108   '(ascii
 109     control-1
 110     latin-iso8859-1
 111     latin-iso8859-2
 112     latin-iso8859-3
 113     latin-iso8859-4
 114     latin-iso8859-9
 115     latin-jisx0201
 116     cyrillic-iso8859-5
 117     greek-iso8859-7
 118     thai-tis620
 119     =adobe-japan1-0
 120     =adobe-japan1-1
 121     =adobe-japan1-2
 122     =adobe-japan1-3
 123     =adobe-japan1-4
 124     =adobe-japan1-5
 125     =adobe-japan1-6
 126     =jis-x0208
 127     =jis-x0208@1978
 128     =jis-x0208@1983
 129     japanese-jisx0212
 130     chinese-gb2312
 131     =jis-x0208@1990
 132     chinese-cns11643-1
 133     chinese-cns11643-2
 134     chinese-cns11643-3
 135     chinese-cns11643-4
 136     chinese-cns11643-5
 137     chinese-cns11643-6
 138     chinese-cns11643-7
 139     =jis-x0213-1
 140     =jis-x0213-1@2000
 141     =jis-x0213-1@2004
 142     =jis-x0213-2
 143     korean-ksc5601
 144     chinese-isoir165
 145     katakana-jisx0201
 146     hebrew-iso8859-8
 147     chinese-gb12345
 148     latin-viscii
 149     ethiopic-ucs
 150     =big5-cdp
 151     =hanyo-denshi/ja
 152     =hanyo-denshi/jb
 153     =hanyo-denshi/jc
 154     =hanyo-denshi/jd
 155     =hanyo-denshi/ft
 156     =hanyo-denshi/ia
 157     =hanyo-denshi/ib
 158     =hanyo-denshi/hg
 159     =hanyo-denshi/jt
 160     =hanyo-denshi/ks
 161     =gt
 162     =gt-k
 163     ==adobe-japan1-0
 164     ==adobe-japan1-1
 165     ==adobe-japan1-2
 166     ==adobe-japan1-3
 167     ==adobe-japan1-4
 168     ==adobe-japan1-5
 169     ==adobe-japan1-6
 170     ==jis-x0208
 171     ==jis-x0213-1
 172     ==jis-x0213-2
 173     ==hanyo-denshi/ja
 174     ==hanyo-denshi/jb
 175     ==hanyo-denshi/jc
 176     ==hanyo-denshi/ft
 177     ==hanyo-denshi/ib
 178     ==hanyo-denshi/hg
 179     ==hanyo-denshi/jt
 180     ==hanyo-denshi/ks
 181     =daikanwa
 182     =daikanwa@rev2
 183     =daikanwa@rev1
 184     =cbeta
 185     ideograph-hanziku-1
 186     ideograph-hanziku-2
 187     ideograph-hanziku-3
 188     ideograph-hanziku-4
 189     ideograph-hanziku-5
 190     ideograph-hanziku-6
 191     ideograph-hanziku-7
 192     ideograph-hanziku-8
 193     ideograph-hanziku-9
 194     ideograph-hanziku-10
 195     ideograph-hanziku-11
 196     ideograph-hanziku-12
 197     ;; =>>>adobe-japan1-0
 198     ;; =>>>adobe-japan1-1
 199     ;; =>>>adobe-japan1-2
 200     ;; =>>>adobe-japan1-3
 201     ;; =>>>adobe-japan1-4
 202     ;; =>>>adobe-japan1-5
 203     ;; =>>>adobe-japan1-6
 204     ;; =>>>jis-x0208
 205     ;; =>>>jis-x0213-1
 206     ;; =>>>jis-x0213-2
 207     ;; =>>>hanyo-denshi/ja
 208     ;; =>>>hanyo-denshi/jb
 209     ;; =>>>hanyo-denshi/jc
 210     ;; =>>>hanyo-denshi/ft
 211     ;; =>>>hanyo-denshi/ib
 212     ;; =>>>hanyo-denshi/hg
 213     ;; =>>>hanyo-denshi/jt
 214     ;; =>>>hanyo-denshi/ks
 215     ;; =>>>gt
 216     =>>adobe-japan1-0
 217     =>>adobe-japan1-1
 218     =>>adobe-japan1-2
 219     =>>adobe-japan1-3
 220     =>>adobe-japan1-4
 221     =>>adobe-japan1-5
 222     =>>adobe-japan1-6
 223     =>>jis-x0208
 224     =>>jis-x0213-1
 225     =>>jis-x0213-1@2000
 226     =>>jis-x0213-1@2004
 227     =>>jis-x0213-2
 228     =>>jis-x0208@1978
 229     =>>hanyo-denshi/ft
 230     =>>hanyo-denshi/jt
 231     =>>hanyo-denshi/ks
 232     =>>gt
 233     =>>daikanwa
 234     =+>jis-x0208
 235     =+>jis-x0213-1
 236     =+>jis-x0213-2
 237     =+>adobe-japan1-0
 238     =+>adobe-japan1-1
 239     =+>adobe-japan1-2
 240     =+>adobe-japan1-3
 241     =+>adobe-japan1-4
 242     =+>adobe-japan1-5
 243     =+>adobe-japan1-6
 244     =+>jis-x0208@1978
 245     =>jis-x0208
 246     =>jis-x0208@1997
 247     =>jis-x0213-1
 248     =>jis-x0213-1@2000
 249     =>jis-x0213-1@2004
 250     =>jis-x0213-2
 251     ==>ucs@bucs
 252     =>ucs@iso
 253     =>ucs@unicode
 254     =>ucs@jis
 255     =>ucs@cns
 256     =>ucs@ks
 257     =+>ucs@iso
 258     =+>ucs@unicode
 259     =+>ucs@jis
 260     =+>ucs@jis/1990
 261     =+>ucs@cns
 262     =+>ucs@ks
 263     =>>ucs@iso
 264     =>>ucs@unicode
 265     =>>ucs@jis
 266     =>>ucs@cns
 267     =>>>ucs@iso
 268     =>>>ucs@unicode
 269     ==ucs@iso
 270     ==ucs@unicode
 271     ==gb2312
 272     ==ks-x1001
 273     ==cns11643-1
 274     ==cns11643-2
 275     ==cns11643-3
 276     ==gt
 277     ==jis-x0208@1990
 278     ==gt-k
 279     =ucs@iso
 280     =ucs@unicode
 281     =ucs@cns
 282     =>>big5-cdp
 283     =>>gt-k
 284     =+>gt
 285     =>gt
 286     =>big5-cdp
 287     =>daikanwa
 288     =>daikanwa/ho
 289     =>cns11643-7
 290     =big5
 291     =big5-eten
 292     =>gt-k
 293     =zinbun-oracle
 294     =>zinbun-oracle
 295     =ruimoku-v6
 296     =>>ruimoku-v6
 297     =jef-china3
 298     =shinjigen
 299     =big5-cdp-var-3
 300     =big5-cdp-var-5))
 301
 302
 303 ;;; @ char-db formatters
 304 ;;;
 305
 306 (defun char-db-make-char-spec (char)
 307   (let (ret char-spec)
 308     (cond ((characterp char)
 309            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 310                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 311                   (setq char-spec (list (cons '=ucs ret)))
 312                   (cond ((setq ret (get-char-attribute char 'name))
 313                          (setq char-spec (cons (cons 'name ret) char-spec))
 314                          )
 315                         ((setq ret (get-char-attribute char 'name*))
 316                          (setq char-spec (cons (cons 'name* ret) char-spec))
 317                          ))
 318                   )
 319                  ((encode-char char '=adobe-japan1 'defined-only)
 320                   (setq char-spec nil)
 321                   (dolist (ccs (charset-list))
 322                     (if (and (or (memq ccs
 323                                        '(=adobe-japan1-0
 324                                          =adobe-japan1-1
 325                                          =adobe-japan1-2
 326                                          =adobe-japan1-3
 327                                          =adobe-japan1-4
 328                                          =adobe-japan1-5
 329                                          =adobe-japan1-6
 330                                          ))
 331                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 332                                  (string-match "=ucs@" (symbol-name ccs))
 333                                  )
 334                              (setq ccs (charset-name ccs))
 335                              (null (assq ccs char-spec))
 336                              (setq ret (encode-char char ccs 'defined-only)))
 337                         (setq char-spec (cons (cons ccs ret) char-spec))))
 338                   )
 339                  ((encode-char char '==adobe-japan1 'defined-only)
 340                   (setq char-spec nil)
 341                   (dolist (ccs (charset-list))
 342                     (if (and (or (memq ccs
 343                                        '(==adobe-japan1-0
 344                                          ==adobe-japan1-1
 345                                          ==adobe-japan1-2
 346                                          ==adobe-japan1-3
 347                                          ==adobe-japan1-4
 348                                          ==adobe-japan1-5
 349                                          ==adobe-japan1-6
 350                                          ))
 351                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 352                                  (string-match "=ucs@" (symbol-name ccs))
 353                                  )
 354                              (setq ccs (charset-name ccs))
 355                              (null (assq ccs char-spec))
 356                              (setq ret (encode-char char ccs 'defined-only)))
 357                         (setq char-spec (cons (cons ccs ret) char-spec))))
 358                   )
 359                  ((setq ret
 360                         (catch 'tag
 361                           (let ((rest char-db-coded-charset-priority-list)
 362                                 ccs)
 363                             (while rest
 364                               (setq ccs (charset-name
 365                                          (find-charset (car rest))))
 366                               (if (setq ret
 367                                         (encode-char char ccs
 368                                                      'defined-only))
 369                                   (throw 'tag (cons ccs ret)))
 370                               (setq rest (cdr rest))))))
 371                   (setq char-spec (list ret))
 372                   (dolist (ccs (delq (car ret) (charset-list)))
 373                     (if (and (or (charset-iso-final-char ccs)
 374                                  (memq ccs
 375                                        '(=daikanwa
 376                                          =daikanwa@rev2
 377                                          ;; =gt-k
 378                                          =jis-x0208@1997
 379                                          ))
 380                                  (eq (charset-property ccs 'iso-ir) 177)
 381                                  ;; (string-match "=ucs@" (symbol-name ccs))
 382                                  )
 383                              (setq ccs (charset-name ccs))
 384                              (null (assq ccs char-spec))
 385                              (setq ret (encode-char char ccs 'defined-only)))
 386                         (setq char-spec (cons (cons ccs ret) char-spec))))
 387                   (if (null char-spec)
 388                       (setq char-spec (split-char char)))
 389                   (cond ((setq ret (get-char-attribute char 'name))
 390                          (setq char-spec (cons (cons 'name ret) char-spec))
 391                          )
 392                         ((setq ret (get-char-attribute char 'name*))
 393                          (setq char-spec (cons (cons 'name* ret) char-spec))
 394                          ))
 395                   )
 396                  ((setq ret (get-char-attribute
 397                              char 'ideographic-combination))
 398                   (setq char-spec
 399                         (cons (cons 'ideographic-combination ret)
 400                               char-spec))
 401                   ))
 402            char-spec)
 403           ((consp char)
 404            char))))
 405
 406 (defun char-db-insert-char-spec (char &optional readable column
 407                                       required-features)
 408   (unless column
 409     (setq column (current-column)))
 410   (let (char-spec temp-char)
 411     (setq char-spec (char-db-make-char-spec char))
 412     (unless (or (characterp char) ; char
 413                 (condition-case nil
 414                     (setq char (find-char char-spec))
 415                   (error nil)))
 416       ;; define temporary character
 417       ;;   Current implementation is dirty.
 418       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 419                                          char-spec)))
 420       (remove-char-attribute temp-char 'ideograph-daikanwa)
 421       (setq char temp-char))
 422     (insert-char-attributes char
 423                             readable
 424                             (union (mapcar #'car char-spec)
 425                                    required-features)
 426                             nil 'for-sub-node)
 427     (when temp-char
 428       ;; undefine temporary character
 429       ;;   Current implementation is dirty.
 430       (setq char-spec (char-attribute-alist temp-char))
 431       (while char-spec
 432         (remove-char-attribute temp-char (car (car char-spec)))
 433         (setq char-spec (cdr char-spec))))))
 434
 435 (defun char-db-insert-alist (alist &optional readable column)
 436   (unless column
 437     (setq column (current-column)))
 438   (let ((line-breaking
 439          (concat "\n" (make-string (1+ column) ?\ )))
 440         name value
 441         ret al ; cal
 442         key
 443         lbs cell rest separator)
 444     (insert "(")
 445     (while alist
 446       (setq name (car (car alist))
 447             value (cdr (car alist)))
 448       (cond ((eq name 'char)
 449              (insert "(char . ")
 450              (if (setq ret (condition-case nil
 451                                (find-char value)
 452                              (error nil)))
 453                  (progn
 454                    (setq al nil
 455                          ;; cal nil
 456                          )
 457                    (while value
 458                      (setq key (car (car value)))
 459                      ;; (if (find-charset key)
 460                      ;;     (setq cal (cons key cal))
 461                      (setq al (cons key al))
 462                      ;; )
 463                      (setq value (cdr value)))
 464                    (insert-char-attributes ret
 465                                            readable
 466                                            (or al 'none) ; cal
 467                                            nil 'for-sub-node))
 468                (insert (prin1-to-string value)))
 469              (insert ")")
 470              (insert line-breaking))
 471             ((consp value)
 472              (insert (format "(%-18s " name))
 473              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 474              (while (consp value)
 475                (setq cell (car value))
 476                (if (and (consp cell)
 477                         (consp (car cell))
 478                         (setq ret (condition-case nil
 479                                       (find-char cell)
 480                                     (error nil)))
 481                         )
 482                    (progn
 483                      (setq rest cell
 484                            al nil
 485                            ;; cal nil
 486                            )
 487                      (while rest
 488                        (setq key (car (car rest)))
 489                        ;; (if (find-charset key)
 490                        ;;     (setq cal (cons key cal))
 491                        (setq al (cons key al))
 492                        ;; )
 493                        (setq rest (cdr rest)))
 494                      (if separator
 495                          (insert lbs))
 496                      (insert-char-attributes ret
 497                                              readable
 498                                              al ; cal
 499                                              nil 'for-sub-node)
 500                      (setq separator lbs))
 501                  (if separator
 502                      (insert separator))
 503                  (insert (prin1-to-string cell))
 504                  (setq separator " "))
 505                (setq value (cdr value)))
 506              (insert ")")
 507              (insert line-breaking))
 508             (t
 509              (insert (format "(%-18s . %S)%s"
 510                              name value
 511                              line-breaking))))
 512       (setq alist (cdr alist))))
 513   (insert ")"))
 514
 515 (defun char-db-insert-char-reference (plist &optional readable column)
 516   (unless column
 517     (setq column (current-column)))
 518   (let ((line-breaking
 519          (concat "\n" (make-string (1+ column) ?\ )))
 520         (separator "")
 521         name value)
 522     (insert "(")
 523     (while plist
 524       (setq name (pop plist))
 525       (setq value (pop plist))
 526       (cond ((eq name :char)
 527              (insert separator)
 528              (insert ":char\t")
 529              (cond ((numberp value)
 530                     (setq value (decode-char '=ucs value)))
 531                    ;; ((consp value)
 532                    ;;  (setq value (or (find-char value)
 533                    ;;                  value)))
 534                    )
 535              (char-db-insert-char-spec value readable)
 536              (insert line-breaking)
 537              (setq separator ""))
 538             ((eq name :radical)
 539              (insert (format "%s%s\t%d ; %c%s"
 540                              separator
 541                              name value
 542                              (ideographic-radical value)
 543                              line-breaking))
 544              (setq separator ""))
 545             (t
 546              (insert (format "%s%s\t%S" separator name value))
 547              (setq separator line-breaking)))
 548       ))
 549   (insert ")"))
 550
 551 (defun char-db-decode-isolated-char (ccs code-point)
 552   (let (ret)
 553     (setq ret
 554           (cond ((eq ccs 'arabic-iso8859-6)
 555                  (decode-char ccs code-point))
 556                 ;; ((eq ccs '=gt)
 557                 ;;  (decode-builtin-char '==gt code-point))
 558                 ((and (memq ccs '(=gt-pj-1
 559                                   =gt-pj-2
 560                                   =gt-pj-3
 561                                   =gt-pj-4
 562                                   =gt-pj-5
 563                                   =gt-pj-6
 564                                   =gt-pj-7
 565                                   =gt-pj-8
 566                                   =gt-pj-9
 567                                   =gt-pj-10
 568                                   =gt-pj-11))
 569                       (setq ret (decode-char ccs code-point))
 570                       (setq ret (encode-char ret '=gt 'defined-only)))
 571                  (decode-builtin-char '=gt ret))
 572                 (t
 573                  (decode-builtin-char ccs code-point))))
 574     (cond ((and (<= 0 (char-int ret))
 575                 (<= (char-int ret) #x1F))
 576            (decode-char '=ucs (+ #x2400 (char-int ret))))
 577           ((= (char-int ret) #x7F)
 578            ?\u2421)
 579           (t ret))))
 580
 581 (defvar char-db-convert-obsolete-format t)
 582
 583 (defun char-db-insert-ccs-feature (name value line-breaking)
 584   (cond
 585    ((integerp value)
 586     (insert
 587      (format
 588       (cond
 589        ((memq name '(==shinjigen
 590                      =shinjigen
 591                      =shinjigen@1ed
 592                      =shinjigen@rev =shinjigen/+p@rev
 593                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 594         "(%-18s .  %04d)\t; %c")
 595        ((eq name '=shinjigen@1ed/24pr)
 596         "(%-18s . %04d)\t; %c")
 597        ((or (memq name '(==daikanwa
 598                          =daikanwa =>>daikanwa =>daikanwa
 599                          =daikanwa@rev1 =daikanwa@rev2
 600                          =daikanwa/+p =>>daikanwa/+p
 601                          =daikanwa/+2p =>>daikanwa/+2p
 602                          =gt ==gt ; =>>>gt
 603                          =>>gt =+>gt =>gt
 604                          =gt-k ==gt-k =>>gt-k =>gt-k
 605                          =adobe-japan1-0 ==adobe-japan1-0 ; =>>>adobe-japan1-0
 606                          =adobe-japan1-1 ==adobe-japan1-1 ; =>>>adobe-japan1-1
 607                          =adobe-japan1-2 ==adobe-japan1-2 ; =>>>adobe-japan1-2
 608                          =adobe-japan1-3 ==adobe-japan1-3 ; =>>>adobe-japan1-3
 609                          =adobe-japan1-4 ==adobe-japan1-4 ; =>>>adobe-japan1-4
 610                          =adobe-japan1-5 ==adobe-japan1-5 ; =>>>adobe-japan1-5
 611                          =adobe-japan1-6 ==adobe-japan1-6 ; =>>>adobe-japan1-6
 612                          =>>adobe-japan1-0 =+>adobe-japan1-0
 613                          =>>adobe-japan1-1 =+>adobe-japan1-1
 614                          =>>adobe-japan1-2 =+>adobe-japan1-2
 615                          =>>adobe-japan1-3 =+>adobe-japan1-3
 616                          =>>adobe-japan1-4 =+>adobe-japan1-4
 617                          =>>adobe-japan1-5 =+>adobe-japan1-5
 618                          =>>adobe-japan1-6 =+>adobe-japan1-6
 619                          =cbeta =>>cbeta
 620                          =zinbun-oracle =>zinbun-oracle))
 621             ;; (string-match "^=adobe-" (symbol-name name))
 622             )
 623         "(%-18s . %05d)\t; %c")
 624        ((memq name '(=hanyo-denshi/ks
 625                      ==hanyo-denshi/ks ; =>>>hanyo-denshi/ks
 626                      =>>hanyo-denshi/ks
 627                      =zihai mojikyo))
 628         "(%-18s . %06d)\t; %c")
 629        ((>= (charset-dimension name) 2)
 630         "(%-18s . #x%04X)\t; %c")
 631        (t
 632         "(%-18s . #x%02X)\t; %c"))
 633       name
 634       (if (= (charset-iso-graphic-plane name) 1)
 635           (logior value
 636                   (cond ((= (charset-dimension name) 1)
 637                          #x80)
 638                         ((= (charset-dimension name) 2)
 639                          #x8080)
 640                         ((= (charset-dimension name) 3)
 641                          #x808080)
 642                         (t 0)))
 643         value)
 644       (char-db-decode-isolated-char name value)))
 645     (if (and (= (charset-chars name) 94)
 646              (= (charset-dimension name) 2))
 647         (insert (format " [%02d-%02d]"
 648                         (- (lsh value -8) 32)
 649                         (- (logand value 255) 32))))
 650     )
 651    (t
 652     (insert (format "(%-18s . %s)" name value))
 653     ))
 654   (insert line-breaking))
 655
 656 (defun char-db-insert-relation-feature (char name value line-breaking
 657                                              ccss readable)
 658   (insert (format "(%-18s%s " name line-breaking))
 659   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 660         separator cell sources required-features
 661         ret)
 662     (while (consp value)
 663       (setq cell (car value))
 664       (if (integerp cell)
 665           (setq cell (decode-char '=ucs cell)))
 666       (cond
 667        ((eq name '->subsumptive)
 668         (when (or (not (some (lambda (atr)
 669                                (get-char-attribute cell atr))
 670                              char-db-ignored-attributes))
 671                   (some (lambda (ccs)
 672                           (encode-char cell ccs 'defined-only))
 673                         ccss))
 674           (if separator
 675               (insert lbs))
 676           (let ((char-db-ignored-attributes
 677                  (cons '<-subsumptive
 678                        char-db-ignored-attributes)))
 679             (insert-char-attributes cell readable nil nil 'for-sub-node))
 680           (setq separator lbs))
 681         )
 682        ((characterp cell)
 683         (setq sources
 684               (get-char-attribute
 685                char (intern (format "%s*sources" name))))
 686         (setq required-features nil)
 687         (dolist (source sources)
 688           (cond
 689            ((memq source '(JP
 690                            JP/Jouyou
 691                            shinjigen shinjigen@1ed shinjigen@rev))
 692             (setq required-features
 693                   (union required-features
 694                          '(=jis-x0208
 695                            =jis-x0208@1990
 696                            =jis-x0213-1@2000
 697                            =jis-x0213-1@2004
 698                            =jis-x0213-2
 699                            =jis-x0212
 700                            =jis-x0208@1983
 701                            =jis-x0208@1978
 702                            =shinjigen
 703                            =shinjigen@1ed
 704                            =shinjigen@rev
 705                            =shinjigen/+p@rev))))
 706            ((eq source 'CN)
 707             (setq required-features
 708                   (union required-features
 709                          '(=gb2312
 710                            =gb12345
 711                            =iso-ir165)))))
 712           (cond
 713            ((find-charset (setq ret (intern (format "=%s" source))))
 714             (setq required-features
 715                   (cons ret required-features)))
 716            (t (setq required-features
 717                     (cons source required-features)))))
 718         (cond ((string-match "@JP" (symbol-name name))
 719                (setq required-features
 720                      (union required-features
 721                             '(=jis-x0208
 722                               =jis-x0208@1990
 723                               =jis-x0213-1-2000
 724                               =jis-x0213-2-2000
 725                               =jis-x0212
 726                               =jis-x0208@1983
 727                               =jis-x0208@1978))))
 728               ((string-match "@CN" (symbol-name name))
 729                (setq required-features
 730                      (union required-features
 731                             '(=gb2312
 732                               =gb12345
 733                               =iso-ir165)))))
 734         (if separator
 735             (insert lbs))
 736         (if readable
 737             (insert (format "%S" cell))
 738           (char-db-insert-char-spec cell readable
 739                                     nil
 740                                     required-features))
 741         (setq separator lbs))
 742        ((consp cell)
 743         (if separator
 744             (insert lbs))
 745         (if (consp (car cell))
 746             (char-db-insert-char-spec cell readable)
 747           (char-db-insert-char-reference cell readable))
 748         (setq separator lbs))
 749        (t
 750         (if separator
 751             (insert separator))
 752         (insert (prin1-to-string cell))
 753         (setq separator " ")))
 754       (setq value (cdr value)))
 755     (insert ")")
 756     (insert line-breaking)))
 757
 758 (defun insert-char-attributes (char &optional readable attributes column
 759                                     for-sub-node)
 760   (unless column
 761     (setq column (current-column)))
 762   (let (name value ; has-long-ccs-name
 763         rest
 764         radical strokes
 765         (line-breaking
 766          (concat "\n" (make-string (1+ column) ?\ )))
 767         lbs cell separator ret
 768         key al cal
 769         dest-ccss ; sources required-features
 770         ccss)
 771     (let (atr-d)
 772       (setq attributes
 773             (sort (if attributes
 774                       (if (consp attributes)
 775                           (progn
 776                             (dolist (name attributes)
 777                               (unless (memq name char-db-ignored-attributes)
 778                                 (if (find-charset name)
 779                                     (push name ccss))
 780                                 (push name atr-d)))
 781                             atr-d))
 782                     (dolist (name (char-attribute-list))
 783                       (unless (memq name char-db-ignored-attributes)
 784                         (if (find-charset name)
 785                             (push name ccss))
 786                         (push name atr-d)))
 787                     atr-d)
 788                   #'char-attribute-name<)))
 789     (insert "(")
 790     (when (memq '<-subsumptive attributes)
 791       (when (or readable (not for-sub-node))
 792         (when (setq value (get-char-attribute char '<-subsumptive))
 793           (char-db-insert-relation-feature char '<-subsumptive value
 794                                            line-breaking
 795                                            ccss readable)))
 796       (setq attributes (delq '<-subsumptive attributes)))
 797     (when (and (memq '<-denotational attributes)
 798                (setq value (get-char-attribute char '<-denotational)))
 799       (char-db-insert-relation-feature char '<-denotational value
 800                                        line-breaking
 801                                        ccss readable)
 802       (setq attributes (delq '<-denotational attributes)))
 803     (when (and (memq 'name attributes)
 804                (setq value (get-char-attribute char 'name)))
 805       (insert (format
 806                (if (> (+ (current-column) (length value)) 48)
 807                    "(name . %S)%s"
 808                  "(name               . %S)%s")
 809                value line-breaking))
 810       (setq attributes (delq 'name attributes))
 811       )
 812     (when (and (memq 'name* attributes)
 813                (setq value (get-char-attribute char 'name*)))
 814       (insert (format
 815                (if (> (+ (current-column) (length value)) 48)
 816                    "(name* . %S)%s"
 817                  "(name*              . %S)%s")
 818                value line-breaking))
 819       (setq attributes (delq 'name* attributes))
 820       )
 821     (when (and (memq 'script attributes)
 822                (setq value (get-char-attribute char 'script)))
 823       (insert (format "(script\t\t%s)%s"
 824                       (mapconcat (function prin1-to-string)
 825                                  value " ")
 826                       line-breaking))
 827       (setq attributes (delq 'script attributes))
 828       )
 829     (dolist (name '(=>ucs =>ucs*))
 830       (when (and (memq name attributes)
 831                  (setq value (get-char-attribute char name)))
 832         (insert (format "(%-18s . #x%04X)\t; %c%s"
 833                         name value (decode-char '=ucs value)
 834                         line-breaking))
 835         (setq attributes (delq name attributes))))
 836     (dolist (name '(=>ucs@gb =>ucs@big5))
 837       (when (and (memq name attributes)
 838                  (setq value (get-char-attribute char name)))
 839         (insert (format "(%-18s . #x%04X)\t; %c%s"
 840                         name value
 841                         (decode-char (intern
 842                                       (concat "="
 843                                               (substring
 844                                                (symbol-name name) 2)))
 845                                      value)
 846                         line-breaking))
 847         (setq attributes (delq name attributes))
 848         ))
 849     ;; (dolist (name '(=>daikanwa))
 850     ;;   (when (and (memq name attributes)
 851     ;;              (setq value (get-char-attribute char name)))
 852     ;;     (insert
 853     ;;      (if (integerp value)
 854     ;;          (format "(%-18s . %05d)\t; %c%s"
 855     ;;                  name value (decode-char '=daikanwa value)
 856     ;;                  line-breaking)
 857     ;;        (format "(%-18s %s)\t; %c%s"
 858     ;;                name
 859     ;;                (mapconcat (function prin1-to-string)
 860     ;;                           value " ")
 861     ;;                (char-representative-of-daikanwa char)
 862     ;;                line-breaking)))
 863     ;;     (setq attributes (delq name attributes))))
 864     (when (and (memq 'general-category attributes)
 865                (setq value (get-char-attribute char 'general-category)))
 866       (insert (format
 867                "(general-category\t%s) ; %s%s"
 868                (mapconcat (lambda (cell)
 869                             (format "%S" cell))
 870                           value " ")
 871                (cond ((rassoc value unidata-normative-category-alist)
 872                       "Normative Category")
 873                      ((rassoc value unidata-informative-category-alist)
 874                       "Informative Category")
 875                      (t
 876                       "Unknown Category"))
 877                line-breaking))
 878       (setq attributes (delq 'general-category attributes))
 879       )
 880     (when (and (memq 'bidi-category attributes)
 881                (setq value (get-char-attribute char 'bidi-category)))
 882       (insert (format "(bidi-category\t. %S)%s"
 883                       value
 884                       line-breaking))
 885       (setq attributes (delq 'bidi-category attributes))
 886       )
 887     (unless (or (not (memq 'mirrored attributes))
 888                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
 889                     'empty))
 890       (insert (format "(mirrored\t\t. %S)%s"
 891                       value
 892                       line-breaking))
 893       (setq attributes (delq 'mirrored attributes))
 894       )
 895     (cond
 896      ((and (memq 'decimal-digit-value attributes)
 897            (setq value (get-char-attribute char 'decimal-digit-value)))
 898       (insert (format "(decimal-digit-value . %S)%s"
 899                       value
 900                       line-breaking))
 901       (setq attributes (delq 'decimal-digit-value attributes))
 902       (when (and (memq 'digit-value attributes)
 903                  (setq value (get-char-attribute char 'digit-value)))
 904         (insert (format "(digit-value\t . %S)%s"
 905                         value
 906                         line-breaking))
 907         (setq attributes (delq 'digit-value attributes))
 908         )
 909       (when (and (memq 'numeric-value attributes)
 910                  (setq value (get-char-attribute char 'numeric-value)))
 911         (insert (format "(numeric-value\t . %S)%s"
 912                         value
 913                         line-breaking))
 914         (setq attributes (delq 'numeric-value attributes))
 915         )
 916       )
 917      (t
 918       (when (and (memq 'digit-value attributes)
 919                  (setq value (get-char-attribute char 'digit-value)))
 920         (insert (format "(digit-value\t. %S)%s"
 921                         value
 922                         line-breaking))
 923         (setq attributes (delq 'digit-value attributes))
 924         )
 925       (when (and (memq 'numeric-value attributes)
 926                  (setq value (get-char-attribute char 'numeric-value)))
 927         (insert (format "(numeric-value\t. %S)%s"
 928                         value
 929                         line-breaking))
 930         (setq attributes (delq 'numeric-value attributes))
 931         )))
 932     (when (and (memq 'iso-10646-comment attributes)
 933                (setq value (get-char-attribute char 'iso-10646-comment)))
 934       (insert (format "(iso-10646-comment\t. %S)%s"
 935                       value
 936                       line-breaking))
 937       (setq attributes (delq 'iso-10646-comment attributes))
 938       )
 939     (when (and (memq 'morohashi-daikanwa attributes)
 940                (setq value (get-char-attribute char 'morohashi-daikanwa)))
 941       (insert (format "(morohashi-daikanwa\t%s)%s"
 942                       (mapconcat (function prin1-to-string) value " ")
 943                       line-breaking))
 944       (setq attributes (delq 'morohashi-daikanwa attributes))
 945       )
 946     (setq radical nil
 947           strokes nil)
 948     (when (and (memq 'ideographic-radical attributes)
 949                (setq value (get-char-attribute char 'ideographic-radical)))
 950       (setq radical value)
 951       (insert (format "(ideographic-radical . %S)\t; %c%s"
 952                       radical
 953                       (ideographic-radical radical)
 954                       line-breaking))
 955       (setq attributes (delq 'ideographic-radical attributes))
 956       )
 957     (when (and (memq 'shuowen-radical attributes)
 958                (setq value (get-char-attribute char 'shuowen-radical)))
 959       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
 960                       value
 961                       (shuowen-radical value)
 962                       line-breaking))
 963       (setq attributes (delq 'shuowen-radical attributes))
 964       )
 965     (let (key)
 966       (dolist (domain
 967                (append
 968                 char-db-feature-domains
 969                 (let (dest domain)
 970                   (dolist (feature (char-attribute-list))
 971                     (setq feature (symbol-name feature))
 972                     (when (string-match
 973                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
 974                            feature)
 975                       (setq domain (intern (match-string 2 feature)))
 976                      (unless (memq domain dest)
 977                        (setq dest (cons domain dest)))))
 978                   (sort dest #'string<))))
 979         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
 980         (when (and (memq key attributes)
 981                    (setq value (get-char-attribute char key)))
 982           (setq radical value)
 983           (insert (format "(%s . %S)\t; %c%s"
 984                           key
 985                           radical
 986                           (ideographic-radical radical)
 987                           line-breaking))
 988           (setq attributes (delq key attributes))
 989           )
 990         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
 991         (when (and (memq key attributes)
 992                    (setq value (get-char-attribute char key)))
 993           (setq strokes value)
 994           (insert (format "(%s . %S)%s"
 995                           key
 996                           strokes
 997                           line-breaking))
 998           (setq attributes (delq key attributes))
 999           )
1000         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1001         (when (and (memq key attributes)
1002                    (setq value (get-char-attribute char key)))
1003           (insert (format "(%s       . %S)%s"
1004                           key
1005                           value
1006                           line-breaking))
1007           (setq attributes (delq key attributes))
1008           )
1009         (dolist (feature '(ideographic-radical
1010                            ideographic-strokes
1011                            total-strokes))
1012           (setq key (intern (format "%s@%s*sources" feature domain)))
1013           (when (and (memq key attributes)
1014                      (setq value (get-char-attribute char key)))
1015             (insert (format "(%s%s" key line-breaking))
1016             (dolist (cell value)
1017               (insert (format " %s" cell)))
1018             (insert ")")
1019             (insert line-breaking)
1020             (setq attributes (delq key attributes))
1021             ))
1022         ))
1023     (when (and (memq 'ideographic-strokes attributes)
1024                (setq value (get-char-attribute char 'ideographic-strokes)))
1025       (setq strokes value)
1026       (insert (format "(ideographic-strokes . %S)%s"
1027                       strokes
1028                       line-breaking))
1029       (setq attributes (delq 'ideographic-strokes attributes))
1030       )
1031     (when (and (memq 'kangxi-radical attributes)
1032                (setq value (get-char-attribute char 'kangxi-radical)))
1033       (unless (eq value radical)
1034         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1035                         value
1036                         (ideographic-radical value)
1037                         line-breaking))
1038         (or radical
1039             (setq radical value)))
1040       (setq attributes (delq 'kangxi-radical attributes))
1041       )
1042     (when (and (memq 'kangxi-strokes attributes)
1043                (setq value (get-char-attribute char 'kangxi-strokes)))
1044       (unless (eq value strokes)
1045         (insert (format "(kangxi-strokes\t . %S)%s"
1046                         value
1047                         line-breaking))
1048         (or strokes
1049             (setq strokes value)))
1050       (setq attributes (delq 'kangxi-strokes attributes))
1051       )
1052     (when (and (memq 'japanese-radical attributes)
1053                (setq value (get-char-attribute char 'japanese-radical)))
1054       (unless (eq value radical)
1055         (insert (format "(japanese-radical\t . %S)\t; %c%s"
1056                         value
1057                         (ideographic-radical value)
1058                         line-breaking))
1059         (or radical
1060             (setq radical value)))
1061       (setq attributes (delq 'japanese-radical attributes))
1062       )
1063     (when (and (memq 'japanese-strokes attributes)
1064                (setq value (get-char-attribute char 'japanese-strokes)))
1065       (unless (eq value strokes)
1066         (insert (format "(japanese-strokes\t . %S)%s"
1067                         value
1068                         line-breaking))
1069         (or strokes
1070             (setq strokes value)))
1071       (setq attributes (delq 'japanese-strokes attributes))
1072       )
1073     (when (and (memq 'cns-radical attributes)
1074                (setq value (get-char-attribute char 'cns-radical)))
1075       (insert (format "(cns-radical\t . %S)\t; %c%s"
1076                       value
1077                       (ideographic-radical value)
1078                       line-breaking))
1079       (setq attributes (delq 'cns-radical attributes))
1080       )
1081     (when (and (memq 'cns-strokes attributes)
1082                (setq value (get-char-attribute char 'cns-strokes)))
1083       (unless (eq value strokes)
1084         (insert (format "(cns-strokes\t . %S)%s"
1085                         value
1086                         line-breaking))
1087         (or strokes
1088             (setq strokes value)))
1089       (setq attributes (delq 'cns-strokes attributes))
1090       )
1091     ;; (when (and (memq 'shinjigen-1-radical attributes)
1092     ;;            (setq value (get-char-attribute char 'shinjigen-1-radical)))
1093     ;;   (unless (eq value radical)
1094     ;;     (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
1095     ;;                     value
1096     ;;                     (ideographic-radical value)
1097     ;;                     line-breaking))
1098     ;;     (or radical
1099     ;;         (setq radical value)))
1100     ;;   (setq attributes (delq 'shinjigen-1-radical attributes))
1101     ;;   )
1102     (when (and (memq 'ideographic- attributes)
1103                (setq value (get-char-attribute char 'ideographic-)))
1104       (insert "(ideographic-       ")
1105       (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1106             separator nil)
1107       (while (consp value)
1108         (setq cell (car value))
1109         (if (integerp cell)
1110             (setq cell (decode-char '=ucs cell)))
1111         (cond ((characterp cell)
1112                (if separator
1113                    (insert lbs))
1114                (if readable
1115                    (insert (format "%S" cell))
1116                  (char-db-insert-char-spec cell readable))
1117                (setq separator lbs))
1118               ((consp cell)
1119                (if separator
1120                    (insert lbs))
1121                (if (consp (car cell))
1122                    (char-db-insert-char-spec cell readable)
1123                  (char-db-insert-char-reference cell readable))
1124                (setq separator lbs))
1125               (t
1126                (if separator
1127                    (insert separator))
1128                (insert (prin1-to-string cell))
1129                (setq separator " ")))
1130         (setq value (cdr value)))
1131       (insert ")")
1132       (insert line-breaking)
1133       (setq attributes (delq 'ideographic- attributes)))
1134     (when (and (memq 'total-strokes attributes)
1135                (setq value (get-char-attribute char 'total-strokes)))
1136       (insert (format "(total-strokes       . %S)%s"
1137                       value
1138                       line-breaking))
1139       (setq attributes (delq 'total-strokes attributes))
1140       )
1141     (when (and (memq '->ideograph attributes)
1142                (setq value (get-char-attribute char '->ideograph)))
1143       (insert (format "(->ideograph\t%s)%s"
1144                       (mapconcat (lambda (code)
1145                                    (cond ((symbolp code)
1146                                           (symbol-name code))
1147                                          ((integerp code)
1148                                           (format "#x%04X" code))
1149                                          (t
1150                                           (format "%s %S"
1151                                                   line-breaking code))))
1152                                  value " ")
1153                       line-breaking))
1154       (setq attributes (delq '->ideograph attributes))
1155       )
1156     ;; (when (and (memq '->decomposition attributes)
1157     ;;            (setq value (get-char-attribute char '->decomposition)))
1158     ;;   (insert (format "(->decomposition\t%s)%s"
1159     ;;                   (mapconcat (lambda (code)
1160     ;;                                (cond ((symbolp code)
1161     ;;                                       (symbol-name code))
1162     ;;                                      ((characterp code)
1163     ;;                                       (if readable
1164     ;;                                           (format "%S" code)
1165     ;;                                         (format "#x%04X"
1166     ;;                                                 (char-int code))
1167     ;;                                         ))
1168     ;;                                      ((integerp code)
1169     ;;                                       (format "#x%04X" code))
1170     ;;                                      (t
1171     ;;                                       (format "%s%S" line-breaking code))))
1172     ;;                              value " ")
1173     ;;                   line-breaking))
1174     ;;   (setq attributes (delq '->decomposition attributes))
1175     ;;   )
1176     (if (equal (get-char-attribute char '->titlecase)
1177                (get-char-attribute char '->uppercase))
1178         (setq attributes (delq '->titlecase attributes)))
1179     (when (and (memq '->mojikyo attributes)
1180                (setq value (get-char-attribute char '->mojikyo)))
1181       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1182                       value (decode-char 'mojikyo value)
1183                       line-breaking))
1184       (setq attributes (delq '->mojikyo attributes))
1185       )
1186     (when (and (memq 'hanyu-dazidian-vol attributes)
1187                (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1188       (insert (format "(hanyu-dazidian-vol  . %d)%s"
1189                       value line-breaking))
1190       (setq attributes (delq 'hanyu-dazidian-vol attributes))
1191       )
1192     (when (and (memq 'hanyu-dazidian-page attributes)
1193                (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1194       (insert (format "(hanyu-dazidian-page . %d)%s"
1195                       value line-breaking))
1196       (setq attributes (delq 'hanyu-dazidian-page attributes))
1197       )
1198     (when (and (memq 'hanyu-dazidian-char attributes)
1199                (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1200       (insert (format "(hanyu-dazidian-char . %d)%s"
1201                       value line-breaking))
1202       (setq attributes (delq 'hanyu-dazidian-char attributes))
1203       )
1204     (unless readable
1205       (dolist (ignored '(composition
1206                          ->denotational <-subsumptive ->ucs-unified
1207                          ->ideographic-component-forms))
1208         (setq attributes (delq ignored attributes))))
1209     (while attributes
1210       (setq name (car attributes))
1211       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1212                   'value-is-empty)
1213         (cond ((setq ret (find-charset name))
1214                (setq name (charset-name ret))
1215                (when (not (memq name dest-ccss))
1216                  (setq dest-ccss (cons name dest-ccss))
1217                  (char-db-insert-ccs-feature name value line-breaking))
1218                )
1219               ((string-match "^=>ucs@" (symbol-name name))
1220                (insert (format "(%-18s . #x%04X)\t; %c%s"
1221                                name value (decode-char '=ucs value)
1222                                line-breaking))
1223                )
1224               ((eq name 'jisx0208-1978/4X)
1225                (insert (format "(%-18s . #x%04X)%s"
1226                                name value
1227                                line-breaking))
1228                )
1229               ((and
1230                 (not readable)
1231                 (not (eq name '->subsumptive))
1232                 (not (eq name '->uppercase))
1233                 (not (eq name '->lowercase))
1234                 (not (eq name '->titlecase))
1235                 (not (eq name '->canonical))
1236                 (not (eq name '->Bopomofo))
1237                 (not (eq name '->mistakable))
1238                 (not (eq name '->ideographic-variants))
1239                 (null (get-char-attribute
1240                        char (intern (format "%s*sources" name))))
1241                 (not (string-match "\\*sources$" (symbol-name name)))
1242                 (null (get-char-attribute
1243                        char (intern (format "%s*note" name))))
1244                 (not (string-match "\\*note$" (symbol-name name)))
1245                 (or (eq name '<-identical)
1246                     (eq name '<-uppercase)
1247                     (eq name '<-lowercase)
1248                     (eq name '<-titlecase)
1249                     (eq name '<-canonical)
1250                     (eq name '<-ideographic-variants)
1251                     ;; (eq name '<-synonyms)
1252                     (string-match "^<-synonyms" (symbol-name name))
1253                     (eq name '<-mistakable)
1254                     (when (string-match "^->" (symbol-name name))
1255                       (cond
1256                        ((string-match "^->fullwidth" (symbol-name name))
1257                         (not (and (consp value)
1258                                   (characterp (car value))
1259                                   (encode-char
1260                                    (car value) '=ucs 'defined-only)))
1261                         )
1262                        (t)))
1263                     ))
1264                )
1265               ((or (eq name 'ideographic-structure)
1266                    (eq name 'ideographic-combination)
1267                    (eq name 'ideographic-)
1268                    (eq name '=decomposition)
1269                    (char-feature-base-name= '=decomposition name)
1270                    (char-feature-base-name= '=>decomposition name)
1271                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1272                    ;;               (symbol-name name))
1273                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1274                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1275                                  (symbol-name name))
1276                    )
1277                (char-db-insert-relation-feature char name value
1278                                                 line-breaking
1279                                                 ccss readable))
1280               ((memq name '(ideograph=
1281                             original-ideograph-of
1282                             ancient-ideograph-of
1283                             vulgar-ideograph-of
1284                             wrong-ideograph-of
1285                             ;; simplified-ideograph-of
1286                             ideographic-variants
1287                             ;; ideographic-different-form-of
1288                             ))
1289                (insert (format "(%-18s%s " name line-breaking))
1290                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1291                      separator nil)
1292                (while (consp value)
1293                  (setq cell (car value))
1294                  (if (and (consp cell)
1295                           (consp (car cell)))
1296                      (progn
1297                        (if separator
1298                            (insert lbs))
1299                        (char-db-insert-alist cell readable)
1300                        (setq separator lbs))
1301                    (if separator
1302                        (insert separator))
1303                    (insert (prin1-to-string cell))
1304                    (setq separator " "))
1305                  (setq value (cdr value)))
1306                (insert ")")
1307                (insert line-breaking))
1308               ((consp value)
1309                (insert (format "(%-18s " name))
1310                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1311                      separator nil)
1312                (while (consp value)
1313                  (setq cell (car value))
1314                  (if (and (consp cell)
1315                           (consp (car cell))
1316                           (setq ret (condition-case nil
1317                                         (find-char cell)
1318                                       (error nil))))
1319                      (progn
1320                        (setq rest cell
1321                              al nil
1322                              cal nil)
1323                        (while rest
1324                          (setq key (car (car rest)))
1325                          (if (find-charset key)
1326                              (setq cal (cons key cal))
1327                            (setq al (cons key al)))
1328                          (setq rest (cdr rest)))
1329                        (if separator
1330                            (insert lbs))
1331                        (insert-char-attributes ret
1332                                                readable
1333                                                al ; cal
1334                                                nil 'for-sub-node)
1335                        (setq separator lbs))
1336                    (setq ret (prin1-to-string cell))
1337                    (if separator
1338                        (if (< (+ (current-column)
1339                                  (length ret)
1340                                  (length separator))
1341                               76)
1342                            (insert separator)
1343                          (insert lbs)))
1344                    (insert ret)
1345                    (setq separator " "))
1346                  (setq value (cdr value)))
1347                (insert ")")
1348                (insert line-breaking))
1349               (t
1350                (insert (format "(%-18s" name))
1351                (setq ret (prin1-to-string value))
1352                (unless (< (+ (current-column)
1353                              (length ret)
1354                              3)
1355                           76)
1356                  (insert line-breaking))
1357                (insert " . " ret ")" line-breaking)
1358                ;; (insert (format "(%-18s . %S)%s"
1359                ;;                 name value
1360                ;;                 line-breaking))
1361                )
1362               ))
1363       (setq attributes (cdr attributes)))
1364     (insert ")")))
1365
1366 (defun insert-char-data (char &optional readable
1367                               attributes)
1368   (save-restriction
1369     (narrow-to-region (point)(point))
1370     (insert "(define-char
1371   '")
1372     (insert-char-attributes char readable attributes)
1373     (insert ")\n")
1374     (goto-char (point-min))
1375     (while (re-search-forward "[ \t]+$" nil t)
1376       (replace-match ""))
1377     ;; from tabify.
1378     (goto-char (point-min))
1379     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1380       (let ((column (current-column))
1381             (indent-tabs-mode t))
1382         (delete-region (match-beginning 0) (point))
1383         (indent-to column)))
1384     (goto-char (point-max))
1385     ;; (tabify (point-min)(point-max))
1386     ))
1387
1388 (defun insert-char-data-with-variant (char &optional printable
1389                                            no-ucs-unified
1390                                            script excluded-script)
1391   (insert-char-data char printable)
1392   (let ((variants (char-variants char))
1393         rest
1394         variant vs ret)
1395     (setq variants (sort variants #'<))
1396     (setq rest variants)
1397     (setq variants (cons char variants))
1398     (while rest
1399       (setq variant (car rest))
1400       (unless (get-char-attribute variant '<-subsumptive)
1401         (if (and (or (null script)
1402                      (null (setq vs (get-char-attribute variant 'script)))
1403                      (memq script vs))
1404                  (or (null excluded-script)
1405                      (null (setq vs (get-char-attribute variant 'script)))
1406                      (not (memq excluded-script vs))))
1407             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1408               (insert-char-data variant printable)
1409               (if (setq ret (char-variants variant))
1410                   (while ret
1411                     (or (memq (car ret) variants)
1412                         ;; (get-char-attribute (car ret) '<-subsumptive)
1413                         (setq rest (nconc rest (list (car ret)))))
1414                     (setq ret (cdr ret)))))))
1415       (setq rest (cdr rest)))))
1416
1417 (defun insert-char-range-data (min max &optional script excluded-script)
1418   (let ((code min)
1419         char)
1420     (while (<= code max)
1421       (setq char (decode-char '=ucs code))
1422       (if (encode-char char '=ucs 'defined-only)
1423           (insert-char-data-with-variant char nil 'no-ucs-unified
1424                                          script excluded-script))
1425       (setq code (1+ code)))))
1426
1427 (defun write-char-range-data-to-file (min max file
1428                                           &optional script excluded-script)
1429   (let ((coding-system-for-write char-db-file-coding-system))
1430     (with-temp-buffer
1431       (insert (format ";; -*- coding: %s -*-\n"
1432                       char-db-file-coding-system))
1433       (insert-char-range-data min max script excluded-script)
1434       (write-region (point-min)(point-max) file))))
1435
1436 (defvar what-character-original-window-configuration)
1437
1438 ;;;###autoload
1439 (defun what-char-definition (char)
1440   (interactive (list (char-after)))
1441   (let ((buf (get-buffer-create "*Character Description*"))
1442         (the-buf (current-buffer))
1443         (win-conf (current-window-configuration)))
1444     (pop-to-buffer buf)
1445     (make-local-variable 'what-character-original-window-configuration)
1446     (setq what-character-original-window-configuration win-conf)
1447     (setq buffer-read-only nil)
1448     (erase-buffer)
1449     (condition-case err
1450         (progn
1451           (insert-char-data-with-variant char 'printable)
1452           (unless (char-attribute-alist char)
1453             (insert (format ";; = %c\n"
1454                             (let* ((rest (split-char char))
1455                                    (ccs (pop rest))
1456                                    (code (pop rest)))
1457                               (while rest
1458                                 (setq code (logior (lsh code 8)
1459                                                    (pop rest))))
1460                               (decode-char ccs code)))))
1461           ;; (char-db-update-comment)
1462           (set-buffer-modified-p nil)
1463           (view-mode the-buf (lambda (buf)
1464                                (set-window-configuration
1465                                 what-character-original-window-configuration)
1466                                ))
1467           (goto-char (point-min)))
1468       (error (progn
1469                (set-window-configuration
1470                 what-character-original-window-configuration)
1471                (signal (car err) (cdr err)))))))
1472
1473
1474 ;;; @ end
1475 ;;;
1476
1477 (provide 'char-db-util)
1478
1479 ;;; char-db-util.el ends here