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