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 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     =jis-x0208
 118     =jis-x0208@1978
 119     =jis-x0208@1983
 120     japanese-jisx0212
 121     chinese-gb2312
 122     =jis-x0208@1990
 123     chinese-cns11643-1
 124     chinese-cns11643-2
 125     chinese-cns11643-3
 126     chinese-cns11643-4
 127     chinese-cns11643-5
 128     chinese-cns11643-6
 129     chinese-cns11643-7
 130     =jis-x0213-1
 131     =jis-x0213-1@2000
 132     =jis-x0213-1@2004
 133     =jis-x0213-2
 134     korean-ksc5601
 135     chinese-isoir165
 136     katakana-jisx0201
 137     hebrew-iso8859-8
 138     chinese-gb12345
 139     latin-viscii
 140     ethiopic-ucs
 141     =big5-cdp
 142     =gt
 143     =adobe-japan1-0
 144     =adobe-japan1-1
 145     =adobe-japan1-2
 146     =adobe-japan1-3
 147     =adobe-japan1-4
 148     =adobe-japan1-5
 149     =adobe-japan1-6
 150     =hanyo-denshi/ja
 151     =hanyo-denshi/jb
 152     =hanyo-denshi/jc
 153     =hanyo-denshi/jd
 154     =hanyo-denshi/ft
 155     =hanyo-denshi/ia
 156     =hanyo-denshi/ib
 157     =hanyo-denshi/hg
 158     =hanyo-denshi/jt
 159     =hanyo-denshi/ks
 160     =daikanwa
 161     =daikanwa@rev2
 162     =daikanwa@rev1
 163     =cbeta
 164     =gt-k
 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     =>>>jis-x0208
 178     =>>>jis-x0213-1
 179     =>>>jis-x0213-2
 180     =>>>gt
 181     =>>>adobe-japan1
 182     =>>jis-x0208
 183     =>>jis-x0213-1
 184     =>>jis-x0213-1@2000
 185     =>>jis-x0213-1@2004
 186     =>>jis-x0213-2
 187     =>>jis-x0208@1978
 188     =>>hanyo-denshi/ft
 189     =>>hanyo-denshi/ks
 190     =>>gt
 191     =>>daikanwa
 192     =>>adobe-japan1
 193     =+>jis-x0208
 194     =+>jis-x0213-1
 195     =+>jis-x0213-2
 196     =+>jis-x0208@1978
 197     =+>adobe-japan1
 198     =>jis-x0208
 199     =>jis-x0208@1997
 200     =>jis-x0213-1
 201     =>jis-x0213-1@2000
 202     =>jis-x0213-1@2004
 203     =>jis-x0213-2
 204     ==>ucs@bucs
 205     =>ucs@iso
 206     =>ucs@unicode
 207     =>ucs@jis
 208     =>ucs@cns
 209     =>ucs@ks
 210     =+>ucs@unicode
 211     =+>ucs@jis
 212     =+>ucs@jis/1990
 213     =+>ucs@cns
 214     =+>ucs@ks
 215     =>>ucs@iso
 216     =>>ucs@unicode
 217     =>>ucs@jis
 218     =>>ucs@cns
 219     =>>>ucs@iso
 220     =>>>ucs@unicode
 221     =ucs@iso
 222     =ucs@unicode
 223     =>>big5-cdp
 224     =>>gt-k
 225     =+>gt
 226     =>gt
 227     =>big5-cdp
 228     =>daikanwa
 229     =>cns11643-7
 230     =big5
 231     =big5-eten
 232     =>gt-k
 233     =zinbun-oracle
 234     =>zinbun-oracle
 235     =ruimoku-v6
 236     =>>ruimoku-v6
 237     =jef-china3
 238     =shinjigen))
 239
 240
 241 ;;; @ char-db formatters
 242 ;;;
 243
 244 (defun char-db-make-char-spec (char)
 245   (let (ret char-spec)
 246     (cond ((characterp char)
 247            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 248                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 249                   (setq char-spec (list (cons '=ucs ret)))
 250                   (cond ((setq ret (get-char-attribute char 'name))
 251                          (setq char-spec (cons (cons 'name ret) char-spec))
 252                          )
 253                         ((setq ret (get-char-attribute char 'name*))
 254                          (setq char-spec (cons (cons 'name* ret) char-spec))
 255                          ))
 256                   )
 257                  ((setq ret
 258                         (catch 'tag
 259                           (let ((rest char-db-coded-charset-priority-list)
 260                                 ccs)
 261                             (while rest
 262                               (setq ccs (charset-name
 263                                          (find-charset (car rest))))
 264                               (if (setq ret
 265                                         (encode-char char ccs
 266                                                      'defined-only))
 267                                   (throw 'tag (cons ccs ret)))
 268                               (setq rest (cdr rest))))))
 269                   (setq char-spec (list ret))
 270                   (dolist (ccs (delq (car ret) (charset-list)))
 271                     (if (and (or (charset-iso-final-char ccs)
 272                                  (memq ccs
 273                                        '(=daikanwa
 274                                          =daikanwa@rev2
 275                                          ;; =gt-k
 276                                          =jis-x0208@1997
 277                                          ))
 278                                  (string-match "=ucs@" (symbol-name ccs)))
 279                              (setq ccs (charset-name ccs))
 280                              (null (assq ccs char-spec))
 281                              (setq ret (encode-char char ccs 'defined-only)))
 282                         (setq char-spec (cons (cons ccs ret) char-spec))))
 283                   (if (null char-spec)
 284                       (setq char-spec (split-char char)))
 285                   (cond ((setq ret (get-char-attribute char 'name))
 286                          (setq char-spec (cons (cons 'name ret) char-spec))
 287                          )
 288                         ((setq ret (get-char-attribute char 'name*))
 289                          (setq char-spec (cons (cons 'name* ret) char-spec))
 290                          ))
 291                   )
 292                  ((setq ret (get-char-attribute
 293                              char 'ideographic-combination))
 294                   (setq char-spec
 295                         (cons (cons 'ideographic-combination ret)
 296                               char-spec))
 297                   ))
 298            char-spec)
 299           ((consp char)
 300            char))))
 301
 302 (defun char-db-insert-char-spec (char &optional readable column
 303                                       required-features)
 304   (unless column
 305     (setq column (current-column)))
 306   (let (char-spec temp-char)
 307     (setq char-spec (char-db-make-char-spec char))
 308     (unless (or (characterp char) ; char
 309                 (condition-case nil
 310                     (setq char (find-char char-spec))
 311                   (error nil)))
 312       ;; define temporary character
 313       ;;   Current implementation is dirty.
 314       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 315                                          char-spec)))
 316       (remove-char-attribute temp-char 'ideograph-daikanwa)
 317       (setq char temp-char))
 318     (insert-char-attributes char
 319                             readable
 320                             (union (mapcar #'car char-spec)
 321                                    required-features)
 322                             nil 'for-sub-node)
 323     (when temp-char
 324       ;; undefine temporary character
 325       ;;   Current implementation is dirty.
 326       (setq char-spec (char-attribute-alist temp-char))
 327       (while char-spec
 328         (remove-char-attribute temp-char (car (car char-spec)))
 329         (setq char-spec (cdr char-spec))))))
 330
 331 (defun char-db-insert-alist (alist &optional readable column)
 332   (unless column
 333     (setq column (current-column)))
 334   (let ((line-breaking
 335          (concat "\n" (make-string (1+ column) ?\ )))
 336         name value
 337         ret al ; cal
 338         key
 339         lbs cell rest separator)
 340     (insert "(")
 341     (while alist
 342       (setq name (car (car alist))
 343             value (cdr (car alist)))
 344       (cond ((eq name 'char)
 345              (insert "(char . ")
 346              (if (setq ret (condition-case nil
 347                                (find-char value)
 348                              (error nil)))
 349                  (progn
 350                    (setq al nil
 351                          ;; cal nil
 352                          )
 353                    (while value
 354                      (setq key (car (car value)))
 355                      ;; (if (find-charset key)
 356                      ;;     (setq cal (cons key cal))
 357                      (setq al (cons key al))
 358                      ;; )
 359                      (setq value (cdr value)))
 360                    (insert-char-attributes ret
 361                                            readable
 362                                            (or al 'none) ; cal
 363                                            nil 'for-sub-node))
 364                (insert (prin1-to-string value)))
 365              (insert ")")
 366              (insert line-breaking))
 367             ((consp value)
 368              (insert (format "(%-18s " name))
 369              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 370              (while (consp value)
 371                (setq cell (car value))
 372                (if (and (consp cell)
 373                         (consp (car cell))
 374                         (setq ret (condition-case nil
 375                                       (find-char cell)
 376                                     (error nil)))
 377                         )
 378                    (progn
 379                      (setq rest cell
 380                            al nil
 381                            ;; cal nil
 382                            )
 383                      (while rest
 384                        (setq key (car (car rest)))
 385                        ;; (if (find-charset key)
 386                        ;;     (setq cal (cons key cal))
 387                        (setq al (cons key al))
 388                        ;; )
 389                        (setq rest (cdr rest)))
 390                      (if separator
 391                          (insert lbs))
 392                      (insert-char-attributes ret
 393                                              readable
 394                                              al ; cal
 395                                              nil 'for-sub-node)
 396                      (setq separator lbs))
 397                  (if separator
 398                      (insert separator))
 399                  (insert (prin1-to-string cell))
 400                  (setq separator " "))
 401                (setq value (cdr value)))
 402              (insert ")")
 403              (insert line-breaking))
 404             (t
 405              (insert (format "(%-18s . %S)%s"
 406                              name value
 407                              line-breaking))))
 408       (setq alist (cdr alist))))
 409   (insert ")"))
 410
 411 (defun char-db-insert-char-reference (plist &optional readable column)
 412   (unless column
 413     (setq column (current-column)))
 414   (let ((line-breaking
 415          (concat "\n" (make-string (1+ column) ?\ )))
 416         (separator "")
 417         name value)
 418     (insert "(")
 419     (while plist
 420       (setq name (pop plist))
 421       (setq value (pop plist))
 422       (cond ((eq name :char)
 423              (insert separator)
 424              (insert ":char\t")
 425              (cond ((numberp value)
 426                     (setq value (decode-char '=ucs value)))
 427                    ;; ((consp value)
 428                    ;;  (setq value (or (find-char value)
 429                    ;;                  value)))
 430                    )
 431              (char-db-insert-char-spec value readable)
 432              (insert line-breaking)
 433              (setq separator ""))
 434             ((eq name :radical)
 435              (insert (format "%s%s\t%d ; %c%s"
 436                              separator
 437                              name value
 438                              (ideographic-radical value)
 439                              line-breaking))
 440              (setq separator ""))
 441             (t
 442              (insert (format "%s%s\t%S" separator name value))
 443              (setq separator line-breaking)))
 444       ))
 445   (insert ")"))
 446
 447 (defun char-db-decode-isolated-char (ccs code-point)
 448   (let (ret)
 449     (setq ret
 450           (cond ((eq ccs 'arabic-iso8859-6)
 451                  (decode-char ccs code-point))
 452                 ((and (memq ccs '(=gt-pj-1
 453                                   =gt-pj-2
 454                                   =gt-pj-3
 455                                   =gt-pj-4
 456                                   =gt-pj-5
 457                                   =gt-pj-6
 458                                   =gt-pj-7
 459                                   =gt-pj-8
 460                                   =gt-pj-9
 461                                   =gt-pj-10
 462                                   =gt-pj-11))
 463                       (setq ret (decode-char ccs code-point))
 464                       (setq ret (encode-char ret '=gt 'defined-only)))
 465                  (decode-builtin-char '=gt ret))
 466                 (t
 467                  (decode-builtin-char ccs code-point))))
 468     (cond ((and (<= 0 (char-int ret))
 469                 (<= (char-int ret) #x1F))
 470            (decode-char '=ucs (+ #x2400 (char-int ret))))
 471           ((= (char-int ret) #x7F)
 472            ?\u2421)
 473           (t ret))))
 474
 475 (defvar char-db-convert-obsolete-format t)
 476
 477 (defun char-db-insert-ccs-feature (name value line-breaking)
 478   (cond
 479    ((integerp value)
 480     (insert
 481      (format
 482       (cond
 483        ((memq name '(=shinjigen
 484                      =shinjigen@1ed
 485                      =shinjigen@rev =shinjigen/+p@rev
 486                      =daikanwa/ho))
 487         "(%-18s .  %04d)\t; %c")
 488        ((eq name '=shinjigen@1ed/24pr)
 489         "(%-18s . %04d)\t; %c")
 490        ((or (memq name '(=daikanwa =>>daikanwa =>daikanwa
 491                          =daikanwa@rev1 =daikanwa@rev2
 492                          =daikanwa/+p =daikanwa/+2p
 493                          =gt =>>>gt =>>gt =+>gt =>gt
 494                          =gt-k =>>gt-k =>gt-k
 495                          =adobe-japan1-0 =>>>adobe-japan1-0 =>>adobe-japan1-0
 496                          =adobe-japan1-1 =>>>adobe-japan1-1 =>>adobe-japan1-1
 497                          =adobe-japan1-2 =>>>adobe-japan1-2 =>>adobe-japan1-2
 498                          =adobe-japan1-3 =>>>adobe-japan1-3 =>>adobe-japan1-3
 499                          =adobe-japan1-4 =>>>adobe-japan1-4 =>>adobe-japan1-4
 500                          =adobe-japan1-5 =>>>adobe-japan1-5 =>>adobe-japan1-5
 501                          =adobe-japan1-6 =>>>adobe-japan1-6 =>>adobe-japan1-6
 502                          =+>adobe-japan1
 503                          =cbeta =>>cbeta
 504                          =zinbun-oracle =>zinbun-oracle))
 505             ;; (string-match "^=adobe-" (symbol-name name))
 506             )
 507         "(%-18s . %05d)\t; %c")
 508        ((memq name '(=hanyo-denshi/ks =>>hanyo-denshi/ks mojikyo))
 509         "(%-18s . %06d)\t; %c")
 510        ((>= (charset-dimension name) 2)
 511         "(%-18s . #x%04X)\t; %c")
 512        (t
 513         "(%-18s . #x%02X)\t; %c"))
 514       name
 515       (if (= (charset-iso-graphic-plane name) 1)
 516           (logior value
 517                   (cond ((= (charset-dimension name) 1)
 518                          #x80)
 519                         ((= (charset-dimension name) 2)
 520                          #x8080)
 521                         ((= (charset-dimension name) 3)
 522                          #x808080)
 523                         (t 0)))
 524         value)
 525       (char-db-decode-isolated-char name value)))
 526     (if (and (= (charset-chars name) 94)
 527              (= (charset-dimension name) 2))
 528         (insert (format " [%02d-%02d]"
 529                         (- (lsh value -8) 32)
 530                         (- (logand value 255) 32))))
 531     )
 532    (t
 533     (insert (format "(%-18s . %s)" name value))
 534     ))
 535   (insert line-breaking))
 536
 537 (defun char-db-insert-relation-feature (char name value line-breaking
 538                                              ccss readable)
 539   (insert (format "(%-18s%s " name line-breaking))
 540   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 541         separator cell sources required-features
 542         ret)
 543     (while (consp value)
 544       (setq cell (car value))
 545       (if (integerp cell)
 546           (setq cell (decode-char '=ucs cell)))
 547       (cond
 548        ((eq name '->subsumptive)
 549         (when (or (not (some (lambda (atr)
 550                                (get-char-attribute cell atr))
 551                              char-db-ignored-attributes))
 552                   (some (lambda (ccs)
 553                           (encode-char cell ccs 'defined-only))
 554                         ccss))
 555           (if separator
 556               (insert lbs))
 557           (let ((char-db-ignored-attributes
 558                  (cons '<-subsumptive
 559                        char-db-ignored-attributes)))
 560             (insert-char-attributes cell readable nil nil 'for-sub-node))
 561           (setq separator lbs))
 562         )
 563        ((characterp cell)
 564         (setq sources
 565               (get-char-attribute
 566                char (intern (format "%s*sources" name))))
 567         (setq required-features nil)
 568         (dolist (source sources)
 569           (cond
 570            ((memq source '(JP
 571                            JP/Jouyou
 572                            shinjigen shinjigen@1ed shinjigen@rev))
 573             (setq required-features
 574                   (union required-features
 575                          '(=jis-x0208
 576                            =jis-x0208@1990
 577                            =jis-x0213-1@2000
 578                            =jis-x0213-1@2004
 579                            =jis-x0213-2
 580                            =jis-x0212
 581                            =jis-x0208@1983
 582                            =jis-x0208@1978
 583                            =shinjigen))))
 584            ((eq source 'CN)
 585             (setq required-features
 586                   (union required-features
 587                          '(=gb2312
 588                            =gb12345
 589                            =iso-ir165)))))
 590           (cond
 591            ((find-charset (setq ret (intern (format "=%s" source))))
 592             (setq required-features
 593                   (cons ret required-features)))
 594            (t (setq required-features
 595                     (cons source required-features)))))
 596         (cond ((string-match "@JP" (symbol-name name))
 597                (setq required-features
 598                      (union required-features
 599                             '(=jis-x0208
 600                               =jis-x0208@1990
 601                               =jis-x0213-1-2000
 602                               =jis-x0213-2-2000
 603                               =jis-x0212
 604                               =jis-x0208@1983
 605                               =jis-x0208@1978))))
 606               ((string-match "@CN" (symbol-name name))
 607                (setq required-features
 608                      (union required-features
 609                             '(=gb2312
 610                               =gb12345
 611                               =iso-ir165)))))
 612         (if separator
 613             (insert lbs))
 614         (if readable
 615             (insert (format "%S" cell))
 616           (char-db-insert-char-spec cell readable
 617                                     nil
 618                                     required-features))
 619         (setq separator lbs))
 620        ((consp cell)
 621         (if separator
 622             (insert lbs))
 623         (if (consp (car cell))
 624             (char-db-insert-char-spec cell readable)
 625           (char-db-insert-char-reference cell readable))
 626         (setq separator lbs))
 627        (t
 628         (if separator
 629             (insert separator))
 630         (insert (prin1-to-string cell))
 631         (setq separator " ")))
 632       (setq value (cdr value)))
 633     (insert ")")
 634     (insert line-breaking)))
 635
 636 (defun insert-char-attributes (char &optional readable attributes column
 637                                     for-sub-node)
 638   (unless column
 639     (setq column (current-column)))
 640   (let (name value ; has-long-ccs-name
 641         rest
 642         radical strokes
 643         (line-breaking
 644          (concat "\n" (make-string (1+ column) ?\ )))
 645         lbs cell separator ret
 646         key al cal
 647         dest-ccss ; sources required-features
 648         ccss)
 649     (let (atr-d)
 650       (setq attributes
 651             (sort (if attributes
 652                       (if (consp attributes)
 653                           (progn
 654                             (dolist (name attributes)
 655                               (unless (memq name char-db-ignored-attributes)
 656                                 (if (find-charset name)
 657                                     (push name ccss))
 658                                 (push name atr-d)))
 659                             atr-d))
 660                     (dolist (name (char-attribute-list))
 661                       (unless (memq name char-db-ignored-attributes)
 662                         (if (find-charset name)
 663                             (push name ccss))
 664                         (push name atr-d)))
 665                     atr-d)
 666                   #'char-attribute-name<)))
 667     (insert "(")
 668     (when (memq '<-subsumptive attributes)
 669       (when (or readable (not for-sub-node))
 670         (when (setq value (get-char-attribute char '<-subsumptive))
 671           (char-db-insert-relation-feature char '<-subsumptive value
 672                                            line-breaking
 673                                            ccss readable)))
 674       (setq attributes (delq '<-subsumptive attributes)))
 675     (when (and (memq '<-denotational attributes)
 676                (setq value (get-char-attribute char '<-denotational)))
 677       (char-db-insert-relation-feature char '<-denotational value
 678                                        line-breaking
 679                                        ccss readable)
 680       (setq attributes (delq '<-denotational attributes)))
 681     (when (and (memq 'name attributes)
 682                (setq value (get-char-attribute char 'name)))
 683       (insert (format
 684                (if (> (+ (current-column) (length value)) 48)
 685                    "(name . %S)%s"
 686                  "(name               . %S)%s")
 687                value line-breaking))
 688       (setq attributes (delq 'name attributes))
 689       )
 690     (when (and (memq 'name* attributes)
 691                (setq value (get-char-attribute char 'name*)))
 692       (insert (format
 693                (if (> (+ (current-column) (length value)) 48)
 694                    "(name* . %S)%s"
 695                  "(name*              . %S)%s")
 696                value line-breaking))
 697       (setq attributes (delq 'name* attributes))
 698       )
 699     (when (and (memq 'script attributes)
 700                (setq value (get-char-attribute char 'script)))
 701       (insert (format "(script\t\t%s)%s"
 702                       (mapconcat (function prin1-to-string)
 703                                  value " ")
 704                       line-breaking))
 705       (setq attributes (delq 'script attributes))
 706       )
 707     (dolist (name '(=>ucs =>ucs*))
 708       (when (and (memq name attributes)
 709                  (setq value (get-char-attribute char name)))
 710         (insert (format "(%-18s . #x%04X)\t; %c%s"
 711                         name value (decode-char '=ucs value)
 712                         line-breaking))
 713         (setq attributes (delq name attributes))))
 714     (dolist (name '(=>ucs@gb =>ucs@big5))
 715       (when (and (memq name attributes)
 716                  (setq value (get-char-attribute char name)))
 717         (insert (format "(%-18s . #x%04X)\t; %c%s"
 718                         name value
 719                         (decode-char (intern
 720                                       (concat "="
 721                                               (substring
 722                                                (symbol-name name) 2)))
 723                                      value)
 724                         line-breaking))
 725         (setq attributes (delq name attributes))
 726         ))
 727     ;; (dolist (name '(=>daikanwa))
 728     ;;   (when (and (memq name attributes)
 729     ;;              (setq value (get-char-attribute char name)))
 730     ;;     (insert
 731     ;;      (if (integerp value)
 732     ;;          (format "(%-18s . %05d)\t; %c%s"
 733     ;;                  name value (decode-char '=daikanwa value)
 734     ;;                  line-breaking)
 735     ;;        (format "(%-18s %s)\t; %c%s"
 736     ;;                name
 737     ;;                (mapconcat (function prin1-to-string)
 738     ;;                           value " ")
 739     ;;                (char-representative-of-daikanwa char)
 740     ;;                line-breaking)))
 741     ;;     (setq attributes (delq name attributes))))
 742     (when (and (memq 'general-category attributes)
 743                (setq value (get-char-attribute char 'general-category)))
 744       (insert (format
 745                "(general-category\t%s) ; %s%s"
 746                (mapconcat (lambda (cell)
 747                             (format "%S" cell))
 748                           value " ")
 749                (cond ((rassoc value unidata-normative-category-alist)
 750                       "Normative Category")
 751                      ((rassoc value unidata-informative-category-alist)
 752                       "Informative Category")
 753                      (t
 754                       "Unknown Category"))
 755                line-breaking))
 756       (setq attributes (delq 'general-category attributes))
 757       )
 758     (when (and (memq 'bidi-category attributes)
 759                (setq value (get-char-attribute char 'bidi-category)))
 760       (insert (format "(bidi-category\t. %S)%s"
 761                       value
 762                       line-breaking))
 763       (setq attributes (delq 'bidi-category attributes))
 764       )
 765     (unless (or (not (memq 'mirrored attributes))
 766                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
 767                     'empty))
 768       (insert (format "(mirrored\t\t. %S)%s"
 769                       value
 770                       line-breaking))
 771       (setq attributes (delq 'mirrored attributes))
 772       )
 773     (cond
 774      ((and (memq 'decimal-digit-value attributes)
 775            (setq value (get-char-attribute char 'decimal-digit-value)))
 776       (insert (format "(decimal-digit-value . %S)%s"
 777                       value
 778                       line-breaking))
 779       (setq attributes (delq 'decimal-digit-value attributes))
 780       (when (and (memq 'digit-value attributes)
 781                  (setq value (get-char-attribute char 'digit-value)))
 782         (insert (format "(digit-value\t . %S)%s"
 783                         value
 784                         line-breaking))
 785         (setq attributes (delq 'digit-value attributes))
 786         )
 787       (when (and (memq 'numeric-value attributes)
 788                  (setq value (get-char-attribute char 'numeric-value)))
 789         (insert (format "(numeric-value\t . %S)%s"
 790                         value
 791                         line-breaking))
 792         (setq attributes (delq 'numeric-value attributes))
 793         )
 794       )
 795      (t
 796       (when (and (memq 'digit-value attributes)
 797                  (setq value (get-char-attribute char 'digit-value)))
 798         (insert (format "(digit-value\t. %S)%s"
 799                         value
 800                         line-breaking))
 801         (setq attributes (delq 'digit-value attributes))
 802         )
 803       (when (and (memq 'numeric-value attributes)
 804                  (setq value (get-char-attribute char 'numeric-value)))
 805         (insert (format "(numeric-value\t. %S)%s"
 806                         value
 807                         line-breaking))
 808         (setq attributes (delq 'numeric-value attributes))
 809         )))
 810     (when (and (memq 'iso-10646-comment attributes)
 811                (setq value (get-char-attribute char 'iso-10646-comment)))
 812       (insert (format "(iso-10646-comment\t. %S)%s"
 813                       value
 814                       line-breaking))
 815       (setq attributes (delq 'iso-10646-comment attributes))
 816       )
 817     (when (and (memq 'morohashi-daikanwa attributes)
 818                (setq value (get-char-attribute char 'morohashi-daikanwa)))
 819       (insert (format "(morohashi-daikanwa\t%s)%s"
 820                       (mapconcat (function prin1-to-string) value " ")
 821                       line-breaking))
 822       (setq attributes (delq 'morohashi-daikanwa attributes))
 823       )
 824     (setq radical nil
 825           strokes nil)
 826     (when (and (memq 'ideographic-radical attributes)
 827                (setq value (get-char-attribute char 'ideographic-radical)))
 828       (setq radical value)
 829       (insert (format "(ideographic-radical . %S)\t; %c%s"
 830                       radical
 831                       (ideographic-radical radical)
 832                       line-breaking))
 833       (setq attributes (delq 'ideographic-radical attributes))
 834       )
 835     (when (and (memq 'shuowen-radical attributes)
 836                (setq value (get-char-attribute char 'shuowen-radical)))
 837       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
 838                       value
 839                       (shuowen-radical value)
 840                       line-breaking))
 841       (setq attributes (delq 'shuowen-radical attributes))
 842       )
 843     (let (key)
 844       (dolist (domain
 845                (append
 846                 char-db-feature-domains
 847                 (let (dest domain)
 848                   (dolist (feature (char-attribute-list))
 849                     (setq feature (symbol-name feature))
 850                     (when (string-match
 851                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
 852                            feature)
 853                       (setq domain (intern (match-string 2 feature)))
 854                      (unless (memq domain dest)
 855                        (setq dest (cons domain dest)))))
 856                   (sort dest #'string<))))
 857         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
 858         (when (and (memq key attributes)
 859                    (setq value (get-char-attribute char key)))
 860           (setq radical value)
 861           (insert (format "(%s . %S)\t; %c%s"
 862                           key
 863                           radical
 864                           (ideographic-radical radical)
 865                           line-breaking))
 866           (setq attributes (delq key attributes))
 867           )
 868         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
 869         (when (and (memq key attributes)
 870                    (setq value (get-char-attribute char key)))
 871           (setq strokes value)
 872           (insert (format "(%s . %S)%s"
 873                           key
 874                           strokes
 875                           line-breaking))
 876           (setq attributes (delq key attributes))
 877           )
 878         (setq key (intern (format "%s@%s" 'total-strokes domain)))
 879         (when (and (memq key attributes)
 880                    (setq value (get-char-attribute char key)))
 881           (insert (format "(%s       . %S)%s"
 882                           key
 883                           value
 884                           line-breaking))
 885           (setq attributes (delq key attributes))
 886           )
 887         (dolist (feature '(ideographic-radical
 888                            ideographic-strokes
 889                            total-strokes))
 890           (setq key (intern (format "%s@%s*sources" feature domain)))
 891           (when (and (memq key attributes)
 892                      (setq value (get-char-attribute char key)))
 893             (insert (format "(%s%s" key line-breaking))
 894             (dolist (cell value)
 895               (insert (format " %s" cell)))
 896             (insert ")")
 897             (insert line-breaking)
 898             (setq attributes (delq key attributes))
 899             ))
 900         ))
 901     (when (and (memq 'ideographic-strokes attributes)
 902                (setq value (get-char-attribute char 'ideographic-strokes)))
 903       (setq strokes value)
 904       (insert (format "(ideographic-strokes . %S)%s"
 905                       strokes
 906                       line-breaking))
 907       (setq attributes (delq 'ideographic-strokes attributes))
 908       )
 909     (when (and (memq 'kangxi-radical attributes)
 910                (setq value (get-char-attribute char 'kangxi-radical)))
 911       (unless (eq value radical)
 912         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
 913                         value
 914                         (ideographic-radical value)
 915                         line-breaking))
 916         (or radical
 917             (setq radical value)))
 918       (setq attributes (delq 'kangxi-radical attributes))
 919       )
 920     (when (and (memq 'kangxi-strokes attributes)
 921                (setq value (get-char-attribute char 'kangxi-strokes)))
 922       (unless (eq value strokes)
 923         (insert (format "(kangxi-strokes\t . %S)%s"
 924                         value
 925                         line-breaking))
 926         (or strokes
 927             (setq strokes value)))
 928       (setq attributes (delq 'kangxi-strokes attributes))
 929       )
 930     (when (and (memq 'japanese-radical attributes)
 931                (setq value (get-char-attribute char 'japanese-radical)))
 932       (unless (eq value radical)
 933         (insert (format "(japanese-radical\t . %S)\t; %c%s"
 934                         value
 935                         (ideographic-radical value)
 936                         line-breaking))
 937         (or radical
 938             (setq radical value)))
 939       (setq attributes (delq 'japanese-radical attributes))
 940       )
 941     (when (and (memq 'japanese-strokes attributes)
 942                (setq value (get-char-attribute char 'japanese-strokes)))
 943       (unless (eq value strokes)
 944         (insert (format "(japanese-strokes\t . %S)%s"
 945                         value
 946                         line-breaking))
 947         (or strokes
 948             (setq strokes value)))
 949       (setq attributes (delq 'japanese-strokes attributes))
 950       )
 951     (when (and (memq 'cns-radical attributes)
 952                (setq value (get-char-attribute char 'cns-radical)))
 953       (insert (format "(cns-radical\t . %S)\t; %c%s"
 954                       value
 955                       (ideographic-radical value)
 956                       line-breaking))
 957       (setq attributes (delq 'cns-radical attributes))
 958       )
 959     (when (and (memq 'cns-strokes attributes)
 960                (setq value (get-char-attribute char 'cns-strokes)))
 961       (unless (eq value strokes)
 962         (insert (format "(cns-strokes\t . %S)%s"
 963                         value
 964                         line-breaking))
 965         (or strokes
 966             (setq strokes value)))
 967       (setq attributes (delq 'cns-strokes attributes))
 968       )
 969     (when (and (memq 'shinjigen-1-radical attributes)
 970                (setq value (get-char-attribute char 'shinjigen-1-radical)))
 971       (unless (eq value radical)
 972         (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
 973                         value
 974                         (ideographic-radical value)
 975                         line-breaking))
 976         (or radical
 977             (setq radical value)))
 978       (setq attributes (delq 'shinjigen-1-radical attributes))
 979       )
 980     (when (and (memq 'ideographic- attributes)
 981                (setq value (get-char-attribute char 'ideographic-)))
 982       (insert "(ideographic-       ")
 983       (setq lbs (concat "\n" (make-string (current-column) ?\ ))
 984             separator nil)
 985       (while (consp value)
 986         (setq cell (car value))
 987         (if (integerp cell)
 988             (setq cell (decode-char '=ucs cell)))
 989         (cond ((characterp cell)
 990                (if separator
 991                    (insert lbs))
 992                (if readable
 993                    (insert (format "%S" cell))
 994                  (char-db-insert-char-spec cell readable))
 995                (setq separator lbs))
 996               ((consp cell)
 997                (if separator
 998                    (insert lbs))
 999                (if (consp (car cell))
1000                    (char-db-insert-char-spec cell readable)
1001                  (char-db-insert-char-reference cell readable))
1002                (setq separator lbs))
1003               (t
1004                (if separator
1005                    (insert separator))
1006                (insert (prin1-to-string cell))
1007                (setq separator " ")))
1008         (setq value (cdr value)))
1009       (insert ")")
1010       (insert line-breaking)
1011       (setq attributes (delq 'ideographic- attributes)))
1012     (when (and (memq 'total-strokes attributes)
1013                (setq value (get-char-attribute char 'total-strokes)))
1014       (insert (format "(total-strokes       . %S)%s"
1015                       value
1016                       line-breaking))
1017       (setq attributes (delq 'total-strokes attributes))
1018       )
1019     (when (and (memq '->ideograph attributes)
1020                (setq value (get-char-attribute char '->ideograph)))
1021       (insert (format "(->ideograph\t%s)%s"
1022                       (mapconcat (lambda (code)
1023                                    (cond ((symbolp code)
1024                                           (symbol-name code))
1025                                          ((integerp code)
1026                                           (format "#x%04X" code))
1027                                          (t
1028                                           (format "%s %S"
1029                                                   line-breaking code))))
1030                                  value " ")
1031                       line-breaking))
1032       (setq attributes (delq '->ideograph attributes))
1033       )
1034     ;; (when (and (memq '->decomposition attributes)
1035     ;;            (setq value (get-char-attribute char '->decomposition)))
1036     ;;   (insert (format "(->decomposition\t%s)%s"
1037     ;;                   (mapconcat (lambda (code)
1038     ;;                                (cond ((symbolp code)
1039     ;;                                       (symbol-name code))
1040     ;;                                      ((characterp code)
1041     ;;                                       (if readable
1042     ;;                                           (format "%S" code)
1043     ;;                                         (format "#x%04X"
1044     ;;                                                 (char-int code))
1045     ;;                                         ))
1046     ;;                                      ((integerp code)
1047     ;;                                       (format "#x%04X" code))
1048     ;;                                      (t
1049     ;;                                       (format "%s%S" line-breaking code))))
1050     ;;                              value " ")
1051     ;;                   line-breaking))
1052     ;;   (setq attributes (delq '->decomposition attributes))
1053     ;;   )
1054     (if (equal (get-char-attribute char '->titlecase)
1055                (get-char-attribute char '->uppercase))
1056         (setq attributes (delq '->titlecase attributes)))
1057     (when (and (memq '->mojikyo attributes)
1058                (setq value (get-char-attribute char '->mojikyo)))
1059       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1060                       value (decode-char 'mojikyo value)
1061                       line-breaking))
1062       (setq attributes (delq '->mojikyo attributes))
1063       )
1064     (when (and (memq 'hanyu-dazidian-vol attributes)
1065                (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1066       (insert (format "(hanyu-dazidian-vol  . %d)%s"
1067                       value line-breaking))
1068       (setq attributes (delq 'hanyu-dazidian-vol attributes))
1069       )
1070     (when (and (memq 'hanyu-dazidian-page attributes)
1071                (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1072       (insert (format "(hanyu-dazidian-page . %d)%s"
1073                       value line-breaking))
1074       (setq attributes (delq 'hanyu-dazidian-page attributes))
1075       )
1076     (when (and (memq 'hanyu-dazidian-char attributes)
1077                (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1078       (insert (format "(hanyu-dazidian-char . %d)%s"
1079                       value line-breaking))
1080       (setq attributes (delq 'hanyu-dazidian-char attributes))
1081       )
1082     (unless readable
1083       (dolist (ignored '(composition
1084                          ->denotational <-subsumptive ->ucs-unified
1085                          ->ideographic-component-forms))
1086         (setq attributes (delq ignored attributes))))
1087     (while attributes
1088       (setq name (car attributes))
1089       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1090                   'value-is-empty)
1091         (cond ((setq ret (find-charset name))
1092                (setq name (charset-name ret))
1093                (when (not (memq name dest-ccss))
1094                  (setq dest-ccss (cons name dest-ccss))
1095                  (char-db-insert-ccs-feature name value line-breaking))
1096                )
1097               ((string-match "^=>ucs@" (symbol-name name))
1098                (insert (format "(%-18s . #x%04X)\t; %c%s"
1099                                name value (decode-char '=ucs value)
1100                                line-breaking))
1101                )
1102               ((eq name 'jisx0208-1978/4X)
1103                (insert (format "(%-18s . #x%04X)%s"
1104                                name value
1105                                line-breaking))
1106                )
1107               ((and
1108                 (not readable)
1109                 (not (eq name '->subsumptive))
1110                 (not (eq name '->uppercase))
1111                 (not (eq name '->lowercase))
1112                 (not (eq name '->titlecase))
1113                 (not (eq name '->canonical))
1114                 (not (eq name '->Bopomofo))
1115                 (not (eq name '->mistakable))
1116                 (not (eq name '->ideographic-variants))
1117                 (null (get-char-attribute
1118                        char (intern (format "%s*sources" name))))
1119                 (not (string-match "\\*sources$" (symbol-name name)))
1120                 (null (get-char-attribute
1121                        char (intern (format "%s*note" name))))
1122                 (not (string-match "\\*note$" (symbol-name name)))
1123                 (or (eq name '<-identical)
1124                     (eq name '<-uppercase)
1125                     (eq name '<-lowercase)
1126                     (eq name '<-titlecase)
1127                     (eq name '<-canonical)
1128                     (eq name '<-ideographic-variants)
1129                     ;; (eq name '<-synonyms)
1130                     (string-match "^<-synonyms" (symbol-name name))
1131                     (eq name '<-mistakable)
1132                     (when (string-match "^->" (symbol-name name))
1133                       (cond
1134                        ((string-match "^->fullwidth" (symbol-name name))
1135                         (not (and (consp value)
1136                                   (characterp (car value))
1137                                   (encode-char
1138                                    (car value) '=ucs 'defined-only)))
1139                         )
1140                        (t)))
1141                     ))
1142                )
1143               ((or (eq name 'ideographic-structure)
1144                    (eq name 'ideographic-combination)
1145                    (eq name 'ideographic-)
1146                    (eq name '=decomposition)
1147                    (char-feature-base-name= '=decomposition name)
1148                    (char-feature-base-name= '=>decomposition name)
1149                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1150                    ;;               (symbol-name name))
1151                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1152                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1153                                  (symbol-name name))
1154                    )
1155                (char-db-insert-relation-feature char name value
1156                                                 line-breaking
1157                                                 ccss readable))
1158               ((memq name '(ideograph=
1159                             original-ideograph-of
1160                             ancient-ideograph-of
1161                             vulgar-ideograph-of
1162                             wrong-ideograph-of
1163                             ;; simplified-ideograph-of
1164                             ideographic-variants
1165                             ;; ideographic-different-form-of
1166                             ))
1167                (insert (format "(%-18s%s " name line-breaking))
1168                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1169                      separator nil)
1170                (while (consp value)
1171                  (setq cell (car value))
1172                  (if (and (consp cell)
1173                           (consp (car cell)))
1174                      (progn
1175                        (if separator
1176                            (insert lbs))
1177                        (char-db-insert-alist cell readable)
1178                        (setq separator lbs))
1179                    (if separator
1180                        (insert separator))
1181                    (insert (prin1-to-string cell))
1182                    (setq separator " "))
1183                  (setq value (cdr value)))
1184                (insert ")")
1185                (insert line-breaking))
1186               ((consp value)
1187                (insert (format "(%-18s " name))
1188                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1189                      separator nil)
1190                (while (consp value)
1191                  (setq cell (car value))
1192                  (if (and (consp cell)
1193                           (consp (car cell))
1194                           (setq ret (condition-case nil
1195                                         (find-char cell)
1196                                       (error nil))))
1197                      (progn
1198                        (setq rest cell
1199                              al nil
1200                              cal nil)
1201                        (while rest
1202                          (setq key (car (car rest)))
1203                          (if (find-charset key)
1204                              (setq cal (cons key cal))
1205                            (setq al (cons key al)))
1206                          (setq rest (cdr rest)))
1207                        (if separator
1208                            (insert lbs))
1209                        (insert-char-attributes ret
1210                                                readable
1211                                                al ; cal
1212                                                nil 'for-sub-node)
1213                        (setq separator lbs))
1214                    (setq ret (prin1-to-string cell))
1215                    (if separator
1216                        (if (< (+ (current-column)
1217                                  (length ret)
1218                                  (length separator))
1219                               76)
1220                            (insert separator)
1221                          (insert lbs)))
1222                    (insert ret)
1223                    (setq separator " "))
1224                  (setq value (cdr value)))
1225                (insert ")")
1226                (insert line-breaking))
1227               (t
1228                (insert (format "(%-18s" name))
1229                (setq ret (prin1-to-string value))
1230                (unless (< (+ (current-column)
1231                              (length ret)
1232                              3)
1233                           76)
1234                  (insert line-breaking))
1235                (insert " . " ret ")" line-breaking)
1236                ;; (insert (format "(%-18s . %S)%s"
1237                ;;                 name value
1238                ;;                 line-breaking))
1239                )
1240               ))
1241       (setq attributes (cdr attributes)))
1242     (insert ")")))
1243
1244 (defun insert-char-data (char &optional readable
1245                               attributes)
1246   (save-restriction
1247     (narrow-to-region (point)(point))
1248     (insert "(define-char
1249   '")
1250     (insert-char-attributes char readable attributes)
1251     (insert ")\n")
1252     (goto-char (point-min))
1253     (while (re-search-forward "[ \t]+$" nil t)
1254       (replace-match ""))
1255     ;; from tabify.
1256     (goto-char (point-min))
1257     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1258       (let ((column (current-column))
1259             (indent-tabs-mode t))
1260         (delete-region (match-beginning 0) (point))
1261         (indent-to column)))
1262     (goto-char (point-max))
1263     ;; (tabify (point-min)(point-max))
1264     ))
1265
1266 (defun insert-char-data-with-variant (char &optional printable
1267                                            no-ucs-unified
1268                                            script excluded-script)
1269   (insert-char-data char printable)
1270   (let ((variants (char-variants char))
1271         rest
1272         variant vs ret)
1273     (setq variants (sort variants #'<))
1274     (setq rest variants)
1275     (setq variants (cons char variants))
1276     (while rest
1277       (setq variant (car rest))
1278       (unless (get-char-attribute variant '<-subsumptive)
1279         (if (and (or (null script)
1280                      (null (setq vs (get-char-attribute variant 'script)))
1281                      (memq script vs))
1282                  (or (null excluded-script)
1283                      (null (setq vs (get-char-attribute variant 'script)))
1284                      (not (memq excluded-script vs))))
1285             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1286               (insert-char-data variant printable)
1287               (if (setq ret (char-variants variant))
1288                   (while ret
1289                     (or (memq (car ret) variants)
1290                         ;; (get-char-attribute (car ret) '<-subsumptive)
1291                         (setq rest (nconc rest (list (car ret)))))
1292                     (setq ret (cdr ret)))))))
1293       (setq rest (cdr rest)))))
1294
1295 (defun insert-char-range-data (min max &optional script excluded-script)
1296   (let ((code min)
1297         char)
1298     (while (<= code max)
1299       (setq char (decode-char '=ucs code))
1300       (if (encode-char char '=ucs 'defined-only)
1301           (insert-char-data-with-variant char nil 'no-ucs-unified
1302                                          script excluded-script))
1303       (setq code (1+ code)))))
1304
1305 (defun write-char-range-data-to-file (min max file
1306                                           &optional script excluded-script)
1307   (let ((coding-system-for-write char-db-file-coding-system))
1308     (with-temp-buffer
1309       (insert (format ";; -*- coding: %s -*-\n"
1310                       char-db-file-coding-system))
1311       (insert-char-range-data min max script excluded-script)
1312       (write-region (point-min)(point-max) file))))
1313
1314 (defvar what-character-original-window-configuration)
1315
1316 ;;;###autoload
1317 (defun what-char-definition (char)
1318   (interactive (list (char-after)))
1319   (let ((buf (get-buffer-create "*Character Description*"))
1320         (the-buf (current-buffer))
1321         (win-conf (current-window-configuration)))
1322     (pop-to-buffer buf)
1323     (make-local-variable 'what-character-original-window-configuration)
1324     (setq what-character-original-window-configuration win-conf)
1325     (setq buffer-read-only nil)
1326     (erase-buffer)
1327     (condition-case err
1328         (progn
1329           (insert-char-data-with-variant char 'printable)
1330           (unless (char-attribute-alist char)
1331             (insert (format ";; = %c\n"
1332                             (let* ((rest (split-char char))
1333                                    (ccs (pop rest))
1334                                    (code (pop rest)))
1335                               (while rest
1336                                 (setq code (logior (lsh code 8)
1337                                                    (pop rest))))
1338                               (decode-char ccs code)))))
1339           ;; (char-db-update-comment)
1340           (set-buffer-modified-p nil)
1341           (view-mode the-buf (lambda (buf)
1342                                (set-window-configuration
1343                                 what-character-original-window-configuration)
1344                                ))
1345           (goto-char (point-min)))
1346       (error (progn
1347                (set-window-configuration
1348                 what-character-original-window-configuration)
1349                (signal (car err) (cdr err)))))))
1350
1351
1352 ;;; @ end
1353 ;;;
1354
1355 (provide 'char-db-util)
1356
1357 ;;; char-db-util.el ends here