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

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