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 ?廾 ?共 ?異 ?舁 ?𦥑 ?䢅 ?爨 ?革 ; 070
  74    ?鬲 ?䰜 ?爪 ?𠃨 ?鬥 ?又 ?𠂇 ?㕜 ?支 ?𦘒 ; 080
  75    ?聿 ?畫 ?隶 ?臤 ?臣 ?殳 ?殺 ?𠘧 ?寸 ?皮 ; 090
  76    ?㼱 ?攴 ?敎 ?卜 ?用 ?爻 ?㸚 ?𥄎 ?目 ?䀠 ; 100
  77    ?眉 ?盾 ?自 ?白 ?鼻 ?皕 ?習 ?羽 ?隹 ?奞 ; 110
  78    ?萑 ?𦫳 ?苜 ?羊 ?羴 ?瞿 ?雔 ?雥 ?鳥 ?烏 ; 120
  79    ?𠦒 ?冓 ?幺 ?𢆶 ?叀 ?玄 ?予 ?放 ?𠬪 ?𣦼 ; 130
  80    ?歺 ?死 ?冎 ?骨 ?肉 ?筋 ?刀 ?刃 ?㓞 ?丰 ; 140
  81    ?耒 ?𧢲 ?竹 ?箕 ?丌 ?左 ?工 ?㠭 ?巫 ?甘 ; 150
  82    ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?旨 ?喜 ?壴 ; 160
  83    ?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     ideograph-daikanwa-2
 159     ideograph-daikanwa
 160     =cbeta
 161     =gt-k
 162     ideograph-hanziku-1
 163     ideograph-hanziku-2
 164     ideograph-hanziku-3
 165     ideograph-hanziku-4
 166     ideograph-hanziku-5
 167     ideograph-hanziku-6
 168     ideograph-hanziku-7
 169     ideograph-hanziku-8
 170     ideograph-hanziku-9
 171     ideograph-hanziku-10
 172     ideograph-hanziku-11
 173     ideograph-hanziku-12
 174     =>>>jis-x0208
 175     =>>>jis-x0213-1
 176     =>>jis-x0208
 177     =>>jis-x0213-1
 178     =>>jis-x0213-1@2000
 179     =>>jis-x0213-1@2004
 180     =>>jis-x0213-2
 181     =>>jis-x0208@1978
 182     =>>hanyo-denshi/ft
 183     =>>hanyo-denshi/ks
 184     =>>gt
 185     =>>daikanwa
 186     =>jis-x0208@usual
 187     =>jis-x0208
 188     =>jis-x0208@1997
 189     =>jis-x0213-1
 190     =>jis-x0213-1@2000
 191     =>jis-x0213-1@2004
 192     =>jis-x0213-2@usual
 193     =>jis-x0213-2
 194     ==>ucs@bucs
 195     =>ucs@iso
 196     =>ucs@unicode
 197     =>ucs@jis
 198     =>ucs@JP
 199     =>ucs@cns
 200     =>ucs@ks
 201     =>>ucs@iso
 202     =>>ucs@unicode
 203     =>>ucs@jis
 204     =>>ucs@cns
 205     =>>>ucs@iso
 206     =>>>ucs@unicode
 207     =ucs@iso
 208     =ucs@unicode
 209     =>>big5-cdp
 210     =>>gt-k
 211     =>gt
 212     =>big5-cdp
 213     =>daikanwa
 214     =big5
 215     =big5-eten
 216     =>gt-k
 217     =zinbun-oracle
 218     =>zinbun-oracle
 219     =ruimoku-v6
 220     =>>ruimoku-v6
 221     =jef-china3
 222     =shinjigen))
 223
 224
 225 ;;; @ char-db formatters
 226 ;;;
 227
 228 (defun char-db-make-char-spec (char)
 229   (let (ret char-spec)
 230     (cond ((characterp char)
 231            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 232                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 233                   (setq char-spec (list (cons '=ucs ret)))
 234                   (cond ((setq ret (get-char-attribute char 'name))
 235                          (setq char-spec (cons (cons 'name ret) char-spec))
 236                          )
 237                         ((setq ret (get-char-attribute char 'name*))
 238                          (setq char-spec (cons (cons 'name* ret) char-spec))
 239                          ))
 240                   )
 241                  ((setq ret
 242                         (catch 'tag
 243                           (let ((rest char-db-coded-charset-priority-list)
 244                                 ccs)
 245                             (while rest
 246                               (setq ccs (charset-name
 247                                          (find-charset (car rest))))
 248                               (if (setq ret
 249                                         (encode-char char ccs
 250                                                      'defined-only))
 251                                   (throw 'tag (cons ccs ret)))
 252                               (setq rest (cdr rest))))))
 253                   (setq char-spec (list ret))
 254                   (dolist (ccs (delq (car ret) (charset-list)))
 255                     (if (and (or (charset-iso-final-char ccs)
 256                                  (memq ccs
 257                                        '(=daikanwa
 258                                          =daikanwa@rev2
 259                                          ;; =gt-k
 260                                          =jis-x0208@1997
 261                                          ))
 262                                  (string-match "=ucs@" (symbol-name ccs)))
 263                              (setq ccs (charset-name ccs))
 264                              (null (assq ccs char-spec))
 265                              (setq ret (encode-char char ccs 'defined-only)))
 266                         (setq char-spec (cons (cons ccs ret) char-spec))))
 267                   (if (null char-spec)
 268                       (setq char-spec (split-char char)))
 269                   (cond ((setq ret (get-char-attribute char 'name))
 270                          (setq char-spec (cons (cons 'name ret) char-spec))
 271                          )
 272                         ((setq ret (get-char-attribute char 'name*))
 273                          (setq char-spec (cons (cons 'name* ret) char-spec))
 274                          ))
 275                   )
 276                  ((setq ret (get-char-attribute
 277                              char 'ideographic-combination))
 278                   (setq char-spec
 279                         (cons (cons 'ideographic-combination ret)
 280                               char-spec))
 281                   ))
 282            char-spec)
 283           ((consp char)
 284            char))))
 285
 286 (defun char-db-insert-char-spec (char &optional readable column
 287                                       required-features)
 288   (unless column
 289     (setq column (current-column)))
 290   (let (char-spec temp-char)
 291     (setq char-spec (char-db-make-char-spec char))
 292     (unless (or (characterp char) ; char
 293                 (condition-case nil
 294                     (setq char (find-char char-spec))
 295                   (error nil)))
 296       ;; define temporary character
 297       ;;   Current implementation is dirty.
 298       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 299                                          char-spec)))
 300       (remove-char-attribute temp-char 'ideograph-daikanwa)
 301       (setq char temp-char))
 302     (insert-char-attributes char
 303                             readable
 304                             (union (mapcar #'car char-spec)
 305                                    required-features))
 306     (when temp-char
 307       ;; undefine temporary character
 308       ;;   Current implementation is dirty.
 309       (setq char-spec (char-attribute-alist temp-char))
 310       (while char-spec
 311         (remove-char-attribute temp-char (car (car char-spec)))
 312         (setq char-spec (cdr char-spec))))))
 313
 314 (defun char-db-insert-alist (alist &optional readable column)
 315   (unless column
 316     (setq column (current-column)))
 317   (let ((line-breaking
 318          (concat "\n" (make-string (1+ column) ?\ )))
 319         name value
 320         ret al ; cal
 321         key
 322         lbs cell rest separator)
 323     (insert "(")
 324     (while alist
 325       (setq name (car (car alist))
 326             value (cdr (car alist)))
 327       (cond ((eq name 'char)
 328              (insert "(char . ")
 329              (if (setq ret (condition-case nil
 330                                (find-char value)
 331                              (error nil)))
 332                  (progn
 333                    (setq al nil
 334                          ;; cal nil
 335                          )
 336                    (while value
 337                      (setq key (car (car value)))
 338                      ;; (if (find-charset key)
 339                      ;;     (setq cal (cons key cal))
 340                      (setq al (cons key al))
 341                      ;; )
 342                      (setq value (cdr value)))
 343                    (insert-char-attributes ret
 344                                            readable
 345                                            (or al 'none) ; cal
 346                                            ))
 347                (insert (prin1-to-string value)))
 348              (insert ")")
 349              (insert line-breaking))
 350             ((consp value)
 351              (insert (format "(%-18s " name))
 352              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 353              (while (consp value)
 354                (setq cell (car value))
 355                (if (and (consp cell)
 356                         (consp (car cell))
 357                         (setq ret (condition-case nil
 358                                       (find-char cell)
 359                                     (error nil)))
 360                         )
 361                    (progn
 362                      (setq rest cell
 363                            al nil
 364                            ;; cal nil
 365                            )
 366                      (while rest
 367                        (setq key (car (car rest)))
 368                        ;; (if (find-charset key)
 369                        ;;     (setq cal (cons key cal))
 370                        (setq al (cons key al))
 371                        ;; )
 372                        (setq rest (cdr rest)))
 373                      (if separator
 374                          (insert lbs))
 375                      (insert-char-attributes ret
 376                                              readable
 377                                              al ; cal
 378                                              )
 379                      (setq separator lbs))
 380                  (if separator
 381                      (insert separator))
 382                  (insert (prin1-to-string cell))
 383                  (setq separator " "))
 384                (setq value (cdr value)))
 385              (insert ")")
 386              (insert line-breaking))
 387             (t
 388              (insert (format "(%-18s . %S)%s"
 389                              name value
 390                              line-breaking))))
 391       (setq alist (cdr alist))))
 392   (insert ")"))
 393
 394 (defun char-db-insert-char-reference (plist &optional readable column)
 395   (unless column
 396     (setq column (current-column)))
 397   (let ((line-breaking
 398          (concat "\n" (make-string (1+ column) ?\ )))
 399         (separator "")
 400         name value)
 401     (insert "(")
 402     (while plist
 403       (setq name (pop plist))
 404       (setq value (pop plist))
 405       (cond ((eq name :char)
 406              (insert separator)
 407              (insert ":char\t")
 408              (cond ((numberp value)
 409                     (setq value (decode-char '=ucs value)))
 410                    ;; ((consp value)
 411                    ;;  (setq value (or (find-char value)
 412                    ;;                  value)))
 413                    )
 414              (char-db-insert-char-spec value readable)
 415              (insert line-breaking)
 416              (setq separator ""))
 417             ((eq name :radical)
 418              (insert (format "%s%s\t%d ; %c%s"
 419                              separator
 420                              name value
 421                              (ideographic-radical value)
 422                              line-breaking))
 423              (setq separator ""))
 424             (t
 425              (insert (format "%s%s\t%S" separator name value))
 426              (setq separator line-breaking)))
 427       ))
 428   (insert ")"))
 429
 430 (defun char-db-decode-isolated-char (ccs code-point)
 431   (let (ret)
 432     (setq ret
 433           (cond ((eq ccs 'arabic-iso8859-6)
 434                  (decode-char ccs code-point))
 435                 ((and (memq ccs '(=gt-pj-1
 436                                   =gt-pj-2
 437                                   =gt-pj-3
 438                                   =gt-pj-4
 439                                   =gt-pj-5
 440                                   =gt-pj-6
 441                                   =gt-pj-7
 442                                   =gt-pj-8
 443                                   =gt-pj-9
 444                                   =gt-pj-10
 445                                   =gt-pj-11))
 446                       (setq ret (decode-char ccs code-point))
 447                       (setq ret (encode-char ret '=gt 'defined-only)))
 448                  (decode-builtin-char '=gt ret))
 449                 (t
 450                  (decode-builtin-char ccs code-point))))
 451     (cond ((and (<= 0 (char-int ret))
 452                 (<= (char-int ret) #x1F))
 453            (decode-char '=ucs (+ #x2400 (char-int ret))))
 454           ((= (char-int ret) #x7F)
 455            ?\u2421)
 456           (t ret))))
 457
 458 (defvar char-db-convert-obsolete-format t)
 459
 460 (defun char-db-insert-ccs-feature (name value line-breaking)
 461   (insert
 462    (format
 463     (cond ((memq name '(=shinjigen
 464                         =shinjigen@1ed
 465                         =shinjigen@rev =shinjigen/+p@rev
 466                         =daikanwa/ho))
 467            "(%-18s .  %04d)\t; %c")
 468           ((eq name '=shinjigen@1ed/24pr)
 469            "(%-18s . %04d)\t; %c")
 470           ((or (memq name '(=daikanwa =>>daikanwa =>daikanwa
 471                             =daikanwa@rev1 =daikanwa@rev2
 472                             =daikanwa/+p =daikanwa/+2p
 473                             =gt =>>>gt =>>gt =>gt
 474                             =gt-k =>>gt-k =>gt-k
 475                             =cbeta =zinbun-oracle =>zinbun-oracle))
 476                (string-match "^=adobe-" (symbol-name name)))
 477            "(%-18s . %05d)\t; %c")
 478           ((memq name '(=hanyo-denshi/ks =>>hanyo-denshi/ks mojikyo))
 479            "(%-18s . %06d)\t; %c")
 480           ((>= (charset-dimension name) 2)
 481            "(%-18s . #x%04X)\t; %c")
 482           (t
 483            "(%-18s . #x%02X)\t; %c"))
 484     name
 485     (if (= (charset-iso-graphic-plane name) 1)
 486         (logior value
 487                 (cond ((= (charset-dimension name) 1)
 488                        #x80)
 489                       ((= (charset-dimension name) 2)
 490                        #x8080)
 491                       ((= (charset-dimension name) 3)
 492                        #x808080)
 493                       (t 0)))
 494       value)
 495     (char-db-decode-isolated-char name value)))
 496   (if (and (= (charset-chars name) 94)
 497            (= (charset-dimension name) 2))
 498       (insert (format " [%02d-%02d]"
 499                       (- (lsh value -8) 32)
 500                       (- (logand value 255) 32))))
 501   (insert line-breaking))
 502
 503 (defun char-db-insert-relation-feature (char name value line-breaking
 504                                              ccss readable)
 505   (insert (format "(%-18s%s " name line-breaking))
 506   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 507         separator cell sources required-features
 508         ret)
 509     (while (consp value)
 510       (setq cell (car value))
 511       (if (integerp cell)
 512           (setq cell (decode-char '=ucs cell)))
 513       (cond
 514        ((eq name '->subsumptive)
 515         (when (or (not (some (lambda (atr)
 516                                (get-char-attribute cell atr))
 517                              char-db-ignored-attributes))
 518                   (some (lambda (ccs)
 519                           (encode-char cell ccs 'defined-only))
 520                         ccss))
 521           (if separator
 522               (insert lbs))
 523           (let ((char-db-ignored-attributes
 524                  (cons '<-subsumptive
 525                        char-db-ignored-attributes)))
 526             (insert-char-attributes cell readable))
 527           (setq separator lbs))
 528         )
 529        ((characterp cell)
 530         (setq sources
 531               (get-char-attribute
 532                char (intern (format "%s*sources" name))))
 533         (setq required-features nil)
 534         (dolist (source sources)
 535           (cond
 536            ((memq source '(JP
 537                            JP/Jouyou
 538                            shinjigen shinjigen@1ed shinjigen@rev))
 539             (setq required-features
 540                   (union required-features
 541                          '(=jis-x0208
 542                            =jis-x0208@1990
 543                            =jis-x0213-1@2000
 544                            =jis-x0213-1@2004
 545                            =jis-x0213-2
 546                            =jis-x0212
 547                            =jis-x0208@1983
 548                            =jis-x0208@1978
 549                            =shinjigen))))
 550            ((eq source 'CN)
 551             (setq required-features
 552                   (union required-features
 553                          '(=gb2312
 554                            =gb12345
 555                            =iso-ir165)))))
 556           (cond
 557            ((find-charset (setq ret (intern (format "=%s" source))))
 558             (setq required-features
 559                   (cons ret required-features)))
 560            (t (setq required-features
 561                     (cons source required-features)))))
 562         (cond ((string-match "@JP" (symbol-name name))
 563                (setq required-features
 564                      (union required-features
 565                             '(=jis-x0208
 566                               =jis-x0208@1990
 567                               =jis-x0213-1-2000
 568                               =jis-x0213-2-2000
 569                               =jis-x0212
 570                               =jis-x0208@1983
 571                               =jis-x0208@1978))))
 572               ((string-match "@CN" (symbol-name name))
 573                (setq required-features
 574                      (union required-features
 575                             '(=gb2312
 576                               =gb12345
 577                               =iso-ir165)))))
 578         (if separator
 579             (insert lbs))
 580         (if readable
 581             (insert (format "%S" cell))
 582           (char-db-insert-char-spec cell readable
 583                                     nil
 584                                     required-features))
 585         (setq separator lbs))
 586        ((consp cell)
 587         (if separator
 588             (insert lbs))
 589         (if (consp (car cell))
 590             (char-db-insert-char-spec cell readable)
 591           (char-db-insert-char-reference cell readable))
 592         (setq separator lbs))
 593        (t
 594         (if separator
 595             (insert separator))
 596         (insert (prin1-to-string cell))
 597         (setq separator " ")))
 598       (setq value (cdr value)))
 599     (insert ")")
 600     (insert line-breaking)))
 601
 602 (defun insert-char-attributes (char &optional readable attributes column)
 603   (unless column
 604     (setq column (current-column)))
 605   (let (name value ; has-long-ccs-name
 606         rest
 607         radical strokes
 608         (line-breaking
 609          (concat "\n" (make-string (1+ column) ?\ )))
 610         lbs cell separator ret
 611         key al cal
 612         dest-ccss ; sources required-features
 613         ccss)
 614     (let (atr-d)
 615       (setq attributes
 616             (sort (if attributes
 617                       (if (consp attributes)
 618                           (progn
 619                             (dolist (name attributes)
 620                               (unless (memq name char-db-ignored-attributes)
 621                                 (if (find-charset name)
 622                                     (push name ccss))
 623                                 (push name atr-d)))
 624                             atr-d))
 625                     (dolist (name (char-attribute-list))
 626                       (unless (memq name char-db-ignored-attributes)
 627                         (if (find-charset name)
 628                             (push name ccss))
 629                         (push name atr-d)))
 630                     atr-d)
 631                   #'char-attribute-name<)))
 632     (insert "(")
 633     (when (memq '<-subsumptive attributes)
 634       (when readable
 635         (when (setq value (get-char-attribute char '<-subsumptive))
 636           (char-db-insert-relation-feature char '<-subsumptive value
 637                                            line-breaking
 638                                            ccss readable)))
 639       (setq attributes (delq '<-subsumptive attributes)))
 640     (when (and (memq '<-denotational attributes)
 641                (setq value (get-char-attribute char '<-denotational)))
 642       (char-db-insert-relation-feature char '<-denotational value
 643                                        line-breaking
 644                                        ccss readable)
 645       (setq attributes (delq '<-denotational attributes)))
 646     (when (and (memq 'name attributes)
 647                (setq value (get-char-attribute char 'name)))
 648       (insert (format
 649                (if (> (+ (current-column) (length value)) 48)
 650                    "(name . %S)%s"
 651                  "(name               . %S)%s")
 652                value line-breaking))
 653       (setq attributes (delq 'name attributes))
 654       )
 655     (when (and (memq 'name* attributes)
 656                (setq value (get-char-attribute char 'name*)))
 657       (insert (format
 658                (if (> (+ (current-column) (length value)) 48)
 659                    "(name* . %S)%s"
 660                  "(name*              . %S)%s")
 661                value line-breaking))
 662       (setq attributes (delq 'name* attributes))
 663       )
 664     (when (and (memq 'script attributes)
 665                (setq value (get-char-attribute char 'script)))
 666       (insert (format "(script\t\t%s)%s"
 667                       (mapconcat (function prin1-to-string)
 668                                  value " ")
 669                       line-breaking))
 670       (setq attributes (delq 'script attributes))
 671       )
 672     (dolist (name '(=>ucs =>ucs*))
 673       (when (and (memq name attributes)
 674                  (setq value (get-char-attribute char name)))
 675         (insert (format "(%-18s . #x%04X)\t; %c%s"
 676                         name value (decode-char '=ucs value)
 677                         line-breaking))
 678         (setq attributes (delq name attributes))))
 679     (dolist (name '(=>ucs@gb =>ucs@big5))
 680       (when (and (memq name attributes)
 681                  (setq value (get-char-attribute char name)))
 682         (insert (format "(%-18s . #x%04X)\t; %c%s"
 683                         name value
 684                         (decode-char (intern
 685                                       (concat "="
 686                                               (substring
 687                                                (symbol-name name) 2)))
 688                                      value)
 689                         line-breaking))
 690         (setq attributes (delq name attributes))
 691         ))
 692     ;; (dolist (name '(=>daikanwa))
 693     ;;   (when (and (memq name attributes)
 694     ;;              (setq value (get-char-attribute char name)))
 695     ;;     (insert
 696     ;;      (if (integerp value)
 697     ;;          (format "(%-18s . %05d)\t; %c%s"
 698     ;;                  name value (decode-char '=daikanwa value)
 699     ;;                  line-breaking)
 700     ;;        (format "(%-18s %s)\t; %c%s"
 701     ;;                name
 702     ;;                (mapconcat (function prin1-to-string)
 703     ;;                           value " ")
 704     ;;                (char-representative-of-daikanwa char)
 705     ;;                line-breaking)))
 706     ;;     (setq attributes (delq name attributes))))
 707     (when (and (memq 'general-category attributes)
 708                (setq value (get-char-attribute char 'general-category)))
 709       (insert (format
 710                "(general-category\t%s) ; %s%s"
 711                (mapconcat (lambda (cell)
 712                             (format "%S" cell))
 713                           value " ")
 714                (cond ((rassoc value unidata-normative-category-alist)
 715                       "Normative Category")
 716                      ((rassoc value unidata-informative-category-alist)
 717                       "Informative Category")
 718                      (t
 719                       "Unknown Category"))
 720                line-breaking))
 721       (setq attributes (delq 'general-category attributes))
 722       )
 723     (when (and (memq 'bidi-category attributes)
 724                (setq value (get-char-attribute char 'bidi-category)))
 725       (insert (format "(bidi-category\t. %S)%s"
 726                       value
 727                       line-breaking))
 728       (setq attributes (delq 'bidi-category attributes))
 729       )
 730     (unless (or (not (memq 'mirrored attributes))
 731                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
 732                     'empty))
 733       (insert (format "(mirrored\t\t. %S)%s"
 734                       value
 735                       line-breaking))
 736       (setq attributes (delq 'mirrored attributes))
 737       )
 738     (cond
 739      ((and (memq 'decimal-digit-value attributes)
 740            (setq value (get-char-attribute char 'decimal-digit-value)))
 741       (insert (format "(decimal-digit-value . %S)%s"
 742                       value
 743                       line-breaking))
 744       (setq attributes (delq 'decimal-digit-value attributes))
 745       (when (and (memq 'digit-value attributes)
 746                  (setq value (get-char-attribute char 'digit-value)))
 747         (insert (format "(digit-value\t . %S)%s"
 748                         value
 749                         line-breaking))
 750         (setq attributes (delq 'digit-value attributes))
 751         )
 752       (when (and (memq 'numeric-value attributes)
 753                  (setq value (get-char-attribute char 'numeric-value)))
 754         (insert (format "(numeric-value\t . %S)%s"
 755                         value
 756                         line-breaking))
 757         (setq attributes (delq 'numeric-value attributes))
 758         )
 759       )
 760      (t
 761       (when (and (memq 'digit-value attributes)
 762                  (setq value (get-char-attribute char 'digit-value)))
 763         (insert (format "(digit-value\t. %S)%s"
 764                         value
 765                         line-breaking))
 766         (setq attributes (delq 'digit-value attributes))
 767         )
 768       (when (and (memq 'numeric-value attributes)
 769                  (setq value (get-char-attribute char 'numeric-value)))
 770         (insert (format "(numeric-value\t. %S)%s"
 771                         value
 772                         line-breaking))
 773         (setq attributes (delq 'numeric-value attributes))
 774         )))
 775     (when (and (memq 'iso-10646-comment attributes)
 776                (setq value (get-char-attribute char 'iso-10646-comment)))
 777       (insert (format "(iso-10646-comment\t. %S)%s"
 778                       value
 779                       line-breaking))
 780       (setq attributes (delq 'iso-10646-comment attributes))
 781       )
 782     (when (and (memq 'morohashi-daikanwa attributes)
 783                (setq value (get-char-attribute char 'morohashi-daikanwa)))
 784       (insert (format "(morohashi-daikanwa\t%s)%s"
 785                       (mapconcat (function prin1-to-string) value " ")
 786                       line-breaking))
 787       (setq attributes (delq 'morohashi-daikanwa attributes))
 788       )
 789     (setq radical nil
 790           strokes nil)
 791     (when (and (memq 'ideographic-radical attributes)
 792                (setq value (get-char-attribute char 'ideographic-radical)))
 793       (setq radical value)
 794       (insert (format "(ideographic-radical . %S)\t; %c%s"
 795                       radical
 796                       (ideographic-radical radical)
 797                       line-breaking))
 798       (setq attributes (delq 'ideographic-radical attributes))
 799       )
 800     (when (and (memq 'shuowen-radical attributes)
 801                (setq value (get-char-attribute char 'shuowen-radical)))
 802       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
 803                       value
 804                       (shuowen-radical value)
 805                       line-breaking))
 806       (setq attributes (delq 'shuowen-radical attributes))
 807       )
 808     (let (key)
 809       (dolist (domain
 810                (append
 811                 char-db-feature-domains
 812                 (let (dest domain)
 813                   (dolist (feature (char-attribute-list))
 814                     (setq feature (symbol-name feature))
 815                     (when (string-match
 816                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
 817                            feature)
 818                       (setq domain (intern (match-string 2 feature)))
 819                      (unless (memq domain dest)
 820                        (setq dest (cons domain dest)))))
 821                   (sort dest #'string<))))
 822         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
 823         (when (and (memq key attributes)
 824                    (setq value (get-char-attribute char key)))
 825           (setq radical value)
 826           (insert (format "(%s . %S)\t; %c%s"
 827                           key
 828                           radical
 829                           (ideographic-radical radical)
 830                           line-breaking))
 831           (setq attributes (delq key attributes))
 832           )
 833         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
 834         (when (and (memq key attributes)
 835                    (setq value (get-char-attribute char key)))
 836           (setq strokes value)
 837           (insert (format "(%s . %S)%s"
 838                           key
 839                           strokes
 840                           line-breaking))
 841           (setq attributes (delq key attributes))
 842           )
 843         (setq key (intern (format "%s@%s" 'total-strokes domain)))
 844         (when (and (memq key attributes)
 845                    (setq value (get-char-attribute char key)))
 846           (insert (format "(%s       . %S)%s"
 847                           key
 848                           value
 849                           line-breaking))
 850           (setq attributes (delq key attributes))
 851           )
 852         (dolist (feature '(ideographic-radical
 853                            ideographic-strokes
 854                            total-strokes))
 855           (setq key (intern (format "%s@%s*sources" feature domain)))
 856           (when (and (memq key attributes)
 857                      (setq value (get-char-attribute char key)))
 858             (insert (format "(%s%s" key line-breaking))
 859             (dolist (cell value)
 860               (insert (format " %s" cell)))
 861             (insert ")")
 862             (insert line-breaking)
 863             (setq attributes (delq key attributes))
 864             ))
 865         ))
 866     (when (and (memq 'ideographic-strokes attributes)
 867                (setq value (get-char-attribute char 'ideographic-strokes)))
 868       (setq strokes value)
 869       (insert (format "(ideographic-strokes . %S)%s"
 870                       strokes
 871                       line-breaking))
 872       (setq attributes (delq 'ideographic-strokes attributes))
 873       )
 874     (when (and (memq 'kangxi-radical attributes)
 875                (setq value (get-char-attribute char 'kangxi-radical)))
 876       (unless (eq value radical)
 877         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
 878                         value
 879                         (ideographic-radical value)
 880                         line-breaking))
 881         (or radical
 882             (setq radical value)))
 883       (setq attributes (delq 'kangxi-radical attributes))
 884       )
 885     (when (and (memq 'kangxi-strokes attributes)
 886                (setq value (get-char-attribute char 'kangxi-strokes)))
 887       (unless (eq value strokes)
 888         (insert (format "(kangxi-strokes\t . %S)%s"
 889                         value
 890                         line-breaking))
 891         (or strokes
 892             (setq strokes value)))
 893       (setq attributes (delq 'kangxi-strokes attributes))
 894       )
 895     (when (and (memq 'japanese-radical attributes)
 896                (setq value (get-char-attribute char 'japanese-radical)))
 897       (unless (eq value radical)
 898         (insert (format "(japanese-radical\t . %S)\t; %c%s"
 899                         value
 900                         (ideographic-radical value)
 901                         line-breaking))
 902         (or radical
 903             (setq radical value)))
 904       (setq attributes (delq 'japanese-radical attributes))
 905       )
 906     (when (and (memq 'japanese-strokes attributes)
 907                (setq value (get-char-attribute char 'japanese-strokes)))
 908       (unless (eq value strokes)
 909         (insert (format "(japanese-strokes\t . %S)%s"
 910                         value
 911                         line-breaking))
 912         (or strokes
 913             (setq strokes value)))
 914       (setq attributes (delq 'japanese-strokes attributes))
 915       )
 916     (when (and (memq 'cns-radical attributes)
 917                (setq value (get-char-attribute char 'cns-radical)))
 918       (insert (format "(cns-radical\t . %S)\t; %c%s"
 919                       value
 920                       (ideographic-radical value)
 921                       line-breaking))
 922       (setq attributes (delq 'cns-radical attributes))
 923       )
 924     (when (and (memq 'cns-strokes attributes)
 925                (setq value (get-char-attribute char 'cns-strokes)))
 926       (unless (eq value strokes)
 927         (insert (format "(cns-strokes\t . %S)%s"
 928                         value
 929                         line-breaking))
 930         (or strokes
 931             (setq strokes value)))
 932       (setq attributes (delq 'cns-strokes attributes))
 933       )
 934     (when (and (memq 'shinjigen-1-radical attributes)
 935                (setq value (get-char-attribute char 'shinjigen-1-radical)))
 936       (unless (eq value radical)
 937         (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
 938                         value
 939                         (ideographic-radical value)
 940                         line-breaking))
 941         (or radical
 942             (setq radical value)))
 943       (setq attributes (delq 'shinjigen-1-radical attributes))
 944       )
 945     (when (and (memq 'ideographic- attributes)
 946                (setq value (get-char-attribute char 'ideographic-)))
 947       (insert "(ideographic-       ")
 948       (setq lbs (concat "\n" (make-string (current-column) ?\ ))
 949             separator nil)
 950       (while (consp value)
 951         (setq cell (car value))
 952         (if (integerp cell)
 953             (setq cell (decode-char '=ucs cell)))
 954         (cond ((characterp cell)
 955                (if separator
 956                    (insert lbs))
 957                (if readable
 958                    (insert (format "%S" cell))
 959                  (char-db-insert-char-spec cell readable))
 960                (setq separator lbs))
 961               ((consp cell)
 962                (if separator
 963                    (insert lbs))
 964                (if (consp (car cell))
 965                    (char-db-insert-char-spec cell readable)
 966                  (char-db-insert-char-reference cell readable))
 967                (setq separator lbs))
 968               (t
 969                (if separator
 970                    (insert separator))
 971                (insert (prin1-to-string cell))
 972                (setq separator " ")))
 973         (setq value (cdr value)))
 974       (insert ")")
 975       (insert line-breaking)
 976       (setq attributes (delq 'ideographic- attributes)))
 977     (when (and (memq 'total-strokes attributes)
 978                (setq value (get-char-attribute char 'total-strokes)))
 979       (insert (format "(total-strokes       . %S)%s"
 980                       value
 981                       line-breaking))
 982       (setq attributes (delq 'total-strokes attributes))
 983       )
 984     (when (and (memq '->ideograph attributes)
 985                (setq value (get-char-attribute char '->ideograph)))
 986       (insert (format "(->ideograph\t%s)%s"
 987                       (mapconcat (lambda (code)
 988                                    (cond ((symbolp code)
 989                                           (symbol-name code))
 990                                          ((integerp code)
 991                                           (format "#x%04X" code))
 992                                          (t
 993                                           (format "%s %S"
 994                                                   line-breaking code))))
 995                                  value " ")
 996                       line-breaking))
 997       (setq attributes (delq '->ideograph attributes))
 998       )
 999     ;; (when (and (memq '->decomposition attributes)
1000     ;;            (setq value (get-char-attribute char '->decomposition)))
1001     ;;   (insert (format "(->decomposition\t%s)%s"
1002     ;;                   (mapconcat (lambda (code)
1003     ;;                                (cond ((symbolp code)
1004     ;;                                       (symbol-name code))
1005     ;;                                      ((characterp code)
1006     ;;                                       (if readable
1007     ;;                                           (format "%S" code)
1008     ;;                                         (format "#x%04X"
1009     ;;                                                 (char-int code))
1010     ;;                                         ))
1011     ;;                                      ((integerp code)
1012     ;;                                       (format "#x%04X" code))
1013     ;;                                      (t
1014     ;;                                       (format "%s%S" line-breaking code))))
1015     ;;                              value " ")
1016     ;;                   line-breaking))
1017     ;;   (setq attributes (delq '->decomposition attributes))
1018     ;;   )
1019     (if (equal (get-char-attribute char '->titlecase)
1020                (get-char-attribute char '->uppercase))
1021         (setq attributes (delq '->titlecase attributes)))
1022     (when (and (memq '->mojikyo attributes)
1023                (setq value (get-char-attribute char '->mojikyo)))
1024       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1025                       value (decode-char 'mojikyo value)
1026                       line-breaking))
1027       (setq attributes (delq '->mojikyo attributes))
1028       )
1029     (when (and (memq 'hanyu-dazidian-vol attributes)
1030                (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1031       (insert (format "(hanyu-dazidian-vol  . %d)%s"
1032                       value line-breaking))
1033       (setq attributes (delq 'hanyu-dazidian-vol attributes))
1034       )
1035     (when (and (memq 'hanyu-dazidian-page attributes)
1036                (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1037       (insert (format "(hanyu-dazidian-page . %d)%s"
1038                       value line-breaking))
1039       (setq attributes (delq 'hanyu-dazidian-page attributes))
1040       )
1041     (when (and (memq 'hanyu-dazidian-char attributes)
1042                (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1043       (insert (format "(hanyu-dazidian-char . %d)%s"
1044                       value line-breaking))
1045       (setq attributes (delq 'hanyu-dazidian-char attributes))
1046       )
1047     (unless readable
1048       (dolist (ignored '(composition
1049                          ->denotational <-subsumptive ->ucs-unified
1050                          ->ideographic-component-forms))
1051         (setq attributes (delq ignored attributes))))
1052     (while attributes
1053       (setq name (car attributes))
1054       (if (setq value (get-char-attribute char name))
1055           (cond ((setq ret (find-charset name))
1056                  (setq name (charset-name ret))
1057                  (if (and (not (memq name dest-ccss))
1058                           (prog1
1059                               (setq value (get-char-attribute char name))
1060                             (setq dest-ccss (cons name dest-ccss))))
1061                      (char-db-insert-ccs-feature name value line-breaking))
1062                  )
1063                 ((string-match "^=>ucs@" (symbol-name name))
1064                  (insert (format "(%-18s . #x%04X)\t; %c%s"
1065                                  name value (decode-char '=ucs value)
1066                                  line-breaking))
1067                  )
1068                 ((eq name 'jisx0208-1978/4X)
1069                  (insert (format "(%-18s . #x%04X)%s"
1070                                  name value
1071                                  line-breaking))
1072                  )
1073                 ((and
1074                   (not readable)
1075                   (not (eq name '->subsumptive))
1076                   (not (eq name '->uppercase))
1077                   (not (eq name '->lowercase))
1078                   (not (eq name '->titlecase))
1079                   (not (eq name '->canonical))
1080                   (not (eq name '->Bopomofo))
1081                   (not (eq name '->mistakable))
1082                   (not (eq name '->ideographic-variants))
1083                   (null (get-char-attribute
1084                          char (intern (format "%s*sources" name))))
1085                   (not (string-match "\\*sources$" (symbol-name name)))
1086                   (null (get-char-attribute
1087                          char (intern (format "%s*note" name))))
1088                   (not (string-match "\\*note$" (symbol-name name)))
1089                   (or (eq name '<-identical)
1090                       (eq name '<-uppercase)
1091                       (eq name '<-lowercase)
1092                       (eq name '<-titlecase)
1093                       (eq name '<-canonical)
1094                       (eq name '<-ideographic-variants)
1095                       ;; (eq name '<-synonyms)
1096                       (string-match "^<-synonyms" (symbol-name name))
1097                       (eq name '<-mistakable)
1098                       (when (string-match "^->" (symbol-name name))
1099                         (cond
1100                          ((string-match "^->fullwidth" (symbol-name name))
1101                           (not (and (consp value)
1102                                     (characterp (car value))
1103                                     (encode-char
1104                                      (car value) '=ucs 'defined-only)))
1105                           )
1106                          (t)))
1107                       ))
1108                  )
1109                 ((or (eq name 'ideographic-structure)
1110                      (eq name 'ideographic-combination)
1111                      (eq name 'ideographic-)
1112                      (eq name '=decomposition)
1113                      (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1114                                    (symbol-name name))
1115                      (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1116                      (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1117                                    (symbol-name name))
1118                      )
1119                  (char-db-insert-relation-feature char name value
1120                                                   line-breaking
1121                                                   ccss readable))
1122                 ((memq name '(ideograph=
1123                               original-ideograph-of
1124                               ancient-ideograph-of
1125                               vulgar-ideograph-of
1126                               wrong-ideograph-of
1127                               ;; simplified-ideograph-of
1128                               ideographic-variants
1129                               ;; ideographic-different-form-of
1130                               ))
1131                  (insert (format "(%-18s%s " name line-breaking))
1132                  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1133                        separator nil)
1134                  (while (consp value)
1135                    (setq cell (car value))
1136                    (if (and (consp cell)
1137                             (consp (car cell)))
1138                        (progn
1139                          (if separator
1140                              (insert lbs))
1141                          (char-db-insert-alist cell readable)
1142                          (setq separator lbs))
1143                      (if separator
1144                          (insert separator))
1145                      (insert (prin1-to-string cell))
1146                      (setq separator " "))
1147                    (setq value (cdr value)))
1148                  (insert ")")
1149                  (insert line-breaking))
1150                 ((consp value)
1151                  (insert (format "(%-18s " name))
1152                  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1153                        separator nil)
1154                  (while (consp value)
1155                    (setq cell (car value))
1156                    (if (and (consp cell)
1157                             (consp (car cell))
1158                             (setq ret (condition-case nil
1159                                           (find-char cell)
1160                                         (error nil))))
1161                        (progn
1162                          (setq rest cell
1163                                al nil
1164                                cal nil)
1165                          (while rest
1166                            (setq key (car (car rest)))
1167                            (if (find-charset key)
1168                                (setq cal (cons key cal))
1169                              (setq al (cons key al)))
1170                            (setq rest (cdr rest)))
1171                          (if separator
1172                              (insert lbs))
1173                          (insert-char-attributes ret
1174                                                  readable
1175                                                  al cal)
1176                          (setq separator lbs))
1177                      (setq ret (prin1-to-string cell))
1178                      (if separator
1179                          (if (< (+ (current-column)
1180                                    (length ret)
1181                                    (length separator))
1182                                 76)
1183                              (insert separator)
1184                            (insert lbs)))
1185                      (insert ret)
1186                      (setq separator " "))
1187                    (setq value (cdr value)))
1188                  (insert ")")
1189                  (insert line-breaking))
1190                 (t
1191                  (insert (format "(%-18s" name))
1192                  (setq ret (prin1-to-string value))
1193                  (unless (< (+ (current-column)
1194                                (length ret)
1195                                3)
1196                             76)
1197                    (insert line-breaking))
1198                  (insert " . " ret ")" line-breaking)
1199                  ;; (insert (format "(%-18s . %S)%s"
1200                  ;;                 name value
1201                  ;;                 line-breaking))
1202                  )
1203                 ))
1204       (setq attributes (cdr attributes)))
1205     (insert ")")))
1206
1207 (defun insert-char-data (char &optional readable
1208                               attributes)
1209   (save-restriction
1210     (narrow-to-region (point)(point))
1211     (insert "(define-char
1212   '")
1213     (insert-char-attributes char readable attributes)
1214     (insert ")\n")
1215     (goto-char (point-min))
1216     (while (re-search-forward "[ \t]+$" nil t)
1217       (replace-match ""))
1218     ;; from tabify.
1219     (goto-char (point-min))
1220     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1221       (let ((column (current-column))
1222             (indent-tabs-mode t))
1223         (delete-region (match-beginning 0) (point))
1224         (indent-to column)))
1225     (goto-char (point-max))
1226     ;; (tabify (point-min)(point-max))
1227     ))
1228
1229 (defun insert-char-data-with-variant (char &optional printable
1230                                            no-ucs-unified
1231                                            script excluded-script)
1232   (insert-char-data char printable)
1233   (let ((variants (char-variants char))
1234         rest
1235         variant vs ret)
1236     (setq variants (sort variants #'<))
1237     (setq rest variants)
1238     (setq variants (cons char variants))
1239     (while rest
1240       (setq variant (car rest))
1241       (unless (get-char-attribute variant '<-subsumptive)
1242         (if (and (or (null script)
1243                      (null (setq vs (get-char-attribute variant 'script)))
1244                      (memq script vs))
1245                  (or (null excluded-script)
1246                      (null (setq vs (get-char-attribute variant 'script)))
1247                      (not (memq excluded-script vs))))
1248             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1249               (insert-char-data variant printable)
1250               (if (setq ret (char-variants variant))
1251                   (while ret
1252                     (or (memq (car ret) variants)
1253                         ;; (get-char-attribute (car ret) '<-subsumptive)
1254                         (setq rest (nconc rest (list (car ret)))))
1255                     (setq ret (cdr ret)))))))
1256       (setq rest (cdr rest)))))
1257
1258 (defun insert-char-range-data (min max &optional script excluded-script)
1259   (let ((code min)
1260         char)
1261     (while (<= code max)
1262       (setq char (decode-char '=ucs code))
1263       (if (encode-char char '=ucs 'defined-only)
1264           (insert-char-data-with-variant char nil 'no-ucs-unified
1265                                          script excluded-script))
1266       (setq code (1+ code)))))
1267
1268 (defun write-char-range-data-to-file (min max file
1269                                           &optional script excluded-script)
1270   (let ((coding-system-for-write char-db-file-coding-system))
1271     (with-temp-buffer
1272       (insert (format ";; -*- coding: %s -*-\n"
1273                       char-db-file-coding-system))
1274       (insert-char-range-data min max script excluded-script)
1275       (write-region (point-min)(point-max) file))))
1276
1277 (defvar what-character-original-window-configuration)
1278
1279 ;;;###autoload
1280 (defun what-char-definition (char)
1281   (interactive (list (char-after)))
1282   (let ((buf (get-buffer-create "*Character Description*"))
1283         (the-buf (current-buffer))
1284         (win-conf (current-window-configuration)))
1285     (pop-to-buffer buf)
1286     (make-local-variable 'what-character-original-window-configuration)
1287     (setq what-character-original-window-configuration win-conf)
1288     (setq buffer-read-only nil)
1289     (erase-buffer)
1290     (condition-case err
1291         (progn
1292           (insert-char-data-with-variant char 'printable)
1293           (unless (char-attribute-alist char)
1294             (insert (format ";; = %c\n"
1295                             (let* ((rest (split-char char))
1296                                    (ccs (pop rest))
1297                                    (code (pop rest)))
1298                               (while rest
1299                                 (setq code (logior (lsh code 8)
1300                                                    (pop rest))))
1301                               (decode-char ccs code)))))
1302           ;; (char-db-update-comment)
1303           (set-buffer-modified-p nil)
1304           (view-mode the-buf (lambda (buf)
1305                                (set-window-configuration
1306                                 what-character-original-window-configuration)
1307                                ))
1308           (goto-char (point-min)))
1309       (error (progn
1310                (set-window-configuration
1311                 what-character-original-window-configuration)
1312                (signal (car err) (cdr err)))))))
1313
1314
1315 ;;; @ end
1316 ;;;
1317
1318 (provide 'char-db-util)
1319
1320 ;;; char-db-util.el ends here