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