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