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