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