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     =daikanwa
 159     =daikanwa@rev2
 160     =daikanwa@rev1
 161     =cbeta
 162     =gt-k
 163     ideograph-hanziku-1
 164     ideograph-hanziku-2
 165     ideograph-hanziku-3
 166     ideograph-hanziku-4
 167     ideograph-hanziku-5
 168     ideograph-hanziku-6
 169     ideograph-hanziku-7
 170     ideograph-hanziku-8
 171     ideograph-hanziku-9
 172     ideograph-hanziku-10
 173     ideograph-hanziku-11
 174     ideograph-hanziku-12
 175     =>>>jis-x0208
 176     =>>>jis-x0213-1
 177     =>>jis-x0208
 178     =>>jis-x0213-1
 179     =>>jis-x0213-1@2000
 180     =>>jis-x0213-1@2004
 181     =>>jis-x0213-2
 182     =>>jis-x0208@1978
 183     =>>hanyo-denshi/ft
 184     =>>hanyo-denshi/ks
 185     =>>gt
 186     =>>daikanwa
 187     =>jis-x0208@usual
 188     =>jis-x0208
 189     =>jis-x0208@1997
 190     =>jis-x0213-1
 191     =>jis-x0213-1@2000
 192     =>jis-x0213-1@2004
 193     =>jis-x0213-2@usual
 194     =>jis-x0213-2
 195     ==>ucs@bucs
 196     =>ucs@iso
 197     =>ucs@unicode
 198     =>ucs@jis
 199     =>ucs@JP
 200     =>ucs@cns
 201     =>ucs@ks
 202     =>>ucs@iso
 203     =>>ucs@unicode
 204     =>>ucs@jis
 205     =>>ucs@cns
 206     =>>>ucs@iso
 207     =>>>ucs@unicode
 208     =ucs@iso
 209     =ucs@unicode
 210     =>>big5-cdp
 211     =>>gt-k
 212     =>gt
 213     =>big5-cdp
 214     =>daikanwa
 215     =big5
 216     =big5-eten
 217     =>gt-k
 218     =zinbun-oracle
 219     =>zinbun-oracle
 220     =ruimoku-v6
 221     =>>ruimoku-v6
 222     =jef-china3
 223     =shinjigen))
 224
 225
 226 ;;; @ char-db formatters
 227 ;;;
 228
 229 (defun char-db-make-char-spec (char)
 230   (let (ret char-spec)
 231     (cond ((characterp char)
 232            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 233                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 234                   (setq char-spec (list (cons '=ucs ret)))
 235                   (cond ((setq ret (get-char-attribute char 'name))
 236                          (setq char-spec (cons (cons 'name ret) char-spec))
 237                          )
 238                         ((setq ret (get-char-attribute char 'name*))
 239                          (setq char-spec (cons (cons 'name* ret) char-spec))
 240                          ))
 241                   )
 242                  ((setq ret
 243                         (catch 'tag
 244                           (let ((rest char-db-coded-charset-priority-list)
 245                                 ccs)
 246                             (while rest
 247                               (setq ccs (charset-name
 248                                          (find-charset (car rest))))
 249                               (if (setq ret
 250                                         (encode-char char ccs
 251                                                      'defined-only))
 252                                   (throw 'tag (cons ccs ret)))
 253                               (setq rest (cdr rest))))))
 254                   (setq char-spec (list ret))
 255                   (dolist (ccs (delq (car ret) (charset-list)))
 256                     (if (and (or (charset-iso-final-char ccs)
 257                                  (memq ccs
 258                                        '(=daikanwa
 259                                          =daikanwa@rev2
 260                                          ;; =gt-k
 261                                          =jis-x0208@1997
 262                                          ))
 263                                  (string-match "=ucs@" (symbol-name ccs)))
 264                              (setq ccs (charset-name ccs))
 265                              (null (assq ccs char-spec))
 266                              (setq ret (encode-char char ccs 'defined-only)))
 267                         (setq char-spec (cons (cons ccs ret) char-spec))))
 268                   (if (null char-spec)
 269                       (setq char-spec (split-char char)))
 270                   (cond ((setq ret (get-char-attribute char 'name))
 271                          (setq char-spec (cons (cons 'name ret) char-spec))
 272                          )
 273                         ((setq ret (get-char-attribute char 'name*))
 274                          (setq char-spec (cons (cons 'name* ret) char-spec))
 275                          ))
 276                   )
 277                  ((setq ret (get-char-attribute
 278                              char 'ideographic-combination))
 279                   (setq char-spec
 280                         (cons (cons 'ideographic-combination ret)
 281                               char-spec))
 282                   ))
 283            char-spec)
 284           ((consp char)
 285            char))))
 286
 287 (defun char-db-insert-char-spec (char &optional readable column
 288                                       required-features)
 289   (unless column
 290     (setq column (current-column)))
 291   (let (char-spec temp-char)
 292     (setq char-spec (char-db-make-char-spec char))
 293     (unless (or (characterp char) ; char
 294                 (condition-case nil
 295                     (setq char (find-char char-spec))
 296                   (error nil)))
 297       ;; define temporary character
 298       ;;   Current implementation is dirty.
 299       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 300                                          char-spec)))
 301       (remove-char-attribute temp-char 'ideograph-daikanwa)
 302       (setq char temp-char))
 303     (insert-char-attributes char
 304                             readable
 305                             (union (mapcar #'car char-spec)
 306                                    required-features)
 307                             nil 'for-sub-node)
 308     (when temp-char
 309       ;; undefine temporary character
 310       ;;   Current implementation is dirty.
 311       (setq char-spec (char-attribute-alist temp-char))
 312       (while char-spec
 313         (remove-char-attribute temp-char (car (car char-spec)))
 314         (setq char-spec (cdr char-spec))))))
 315
 316 (defun char-db-insert-alist (alist &optional readable column)
 317   (unless column
 318     (setq column (current-column)))
 319   (let ((line-breaking
 320          (concat "\n" (make-string (1+ column) ?\ )))
 321         name value
 322         ret al ; cal
 323         key
 324         lbs cell rest separator)
 325     (insert "(")
 326     (while alist
 327       (setq name (car (car alist))
 328             value (cdr (car alist)))
 329       (cond ((eq name 'char)
 330              (insert "(char . ")
 331              (if (setq ret (condition-case nil
 332                                (find-char value)
 333                              (error nil)))
 334                  (progn
 335                    (setq al nil
 336                          ;; cal nil
 337                          )
 338                    (while value
 339                      (setq key (car (car value)))
 340                      ;; (if (find-charset key)
 341                      ;;     (setq cal (cons key cal))
 342                      (setq al (cons key al))
 343                      ;; )
 344                      (setq value (cdr value)))
 345                    (insert-char-attributes ret
 346                                            readable
 347                                            (or al 'none) ; cal
 348                                            nil 'for-sub-node))
 349                (insert (prin1-to-string value)))
 350              (insert ")")
 351              (insert line-breaking))
 352             ((consp value)
 353              (insert (format "(%-18s " name))
 354              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 355              (while (consp value)
 356                (setq cell (car value))
 357                (if (and (consp cell)
 358                         (consp (car cell))
 359                         (setq ret (condition-case nil
 360                                       (find-char cell)
 361                                     (error nil)))
 362                         )
 363                    (progn
 364                      (setq rest cell
 365                            al nil
 366                            ;; cal nil
 367                            )
 368                      (while rest
 369                        (setq key (car (car rest)))
 370                        ;; (if (find-charset key)
 371                        ;;     (setq cal (cons key cal))
 372                        (setq al (cons key al))
 373                        ;; )
 374                        (setq rest (cdr rest)))
 375                      (if separator
 376                          (insert lbs))
 377                      (insert-char-attributes ret
 378                                              readable
 379                                              al ; cal
 380                                              nil 'for-sub-node)
 381                      (setq separator lbs))
 382                  (if separator
 383                      (insert separator))
 384                  (insert (prin1-to-string cell))
 385                  (setq separator " "))
 386                (setq value (cdr value)))
 387              (insert ")")
 388              (insert line-breaking))
 389             (t
 390              (insert (format "(%-18s . %S)%s"
 391                              name value
 392                              line-breaking))))
 393       (setq alist (cdr alist))))
 394   (insert ")"))
 395
 396 (defun char-db-insert-char-reference (plist &optional readable column)
 397   (unless column
 398     (setq column (current-column)))
 399   (let ((line-breaking
 400          (concat "\n" (make-string (1+ column) ?\ )))
 401         (separator "")
 402         name value)
 403     (insert "(")
 404     (while plist
 405       (setq name (pop plist))
 406       (setq value (pop plist))
 407       (cond ((eq name :char)
 408              (insert separator)
 409              (insert ":char\t")
 410              (cond ((numberp value)
 411                     (setq value (decode-char '=ucs value)))
 412                    ;; ((consp value)
 413                    ;;  (setq value (or (find-char value)
 414                    ;;                  value)))
 415                    )
 416              (char-db-insert-char-spec value readable)
 417              (insert line-breaking)
 418              (setq separator ""))
 419             ((eq name :radical)
 420              (insert (format "%s%s\t%d ; %c%s"
 421                              separator
 422                              name value
 423                              (ideographic-radical value)
 424                              line-breaking))
 425              (setq separator ""))
 426             (t
 427              (insert (format "%s%s\t%S" separator name value))
 428              (setq separator line-breaking)))
 429       ))
 430   (insert ")"))
 431
 432 (defun char-db-decode-isolated-char (ccs code-point)
 433   (let (ret)
 434     (setq ret
 435           (cond ((eq ccs 'arabic-iso8859-6)
 436                  (decode-char ccs code-point))
 437                 ((and (memq ccs '(=gt-pj-1
 438                                   =gt-pj-2
 439                                   =gt-pj-3
 440                                   =gt-pj-4
 441                                   =gt-pj-5
 442                                   =gt-pj-6
 443                                   =gt-pj-7
 444                                   =gt-pj-8
 445                                   =gt-pj-9
 446                                   =gt-pj-10
 447                                   =gt-pj-11))
 448                       (setq ret (decode-char ccs code-point))
 449                       (setq ret (encode-char ret '=gt 'defined-only)))
 450                  (decode-builtin-char '=gt ret))
 451                 (t
 452                  (decode-builtin-char ccs code-point))))
 453     (cond ((and (<= 0 (char-int ret))
 454                 (<= (char-int ret) #x1F))
 455            (decode-char '=ucs (+ #x2400 (char-int ret))))
 456           ((= (char-int ret) #x7F)
 457            ?\u2421)
 458           (t ret))))
 459
 460 (defvar char-db-convert-obsolete-format t)
 461
 462 (defun char-db-insert-ccs-feature (name value line-breaking)
 463   (insert
 464    (format
 465     (cond ((memq name '(=shinjigen
 466                         =shinjigen@1ed
 467                         =shinjigen@rev =shinjigen/+p@rev
 468                         =daikanwa/ho))
 469            "(%-18s .  %04d)\t; %c")
 470           ((eq name '=shinjigen@1ed/24pr)
 471            "(%-18s . %04d)\t; %c")
 472           ((or (memq name '(=daikanwa =>>daikanwa =>daikanwa
 473                             =daikanwa@rev1 =daikanwa@rev2
 474                             =daikanwa/+p =daikanwa/+2p
 475                             =gt =>>>gt =>>gt =>gt
 476                             =gt-k =>>gt-k =>gt-k
 477                             =>>adobe-japan1
 478                             =cbeta =zinbun-oracle =>zinbun-oracle))
 479                (string-match "^=adobe-" (symbol-name name)))
 480            "(%-18s . %05d)\t; %c")
 481           ((memq name '(=hanyo-denshi/ks =>>hanyo-denshi/ks mojikyo))
 482            "(%-18s . %06d)\t; %c")
 483           ((>= (charset-dimension name) 2)
 484            "(%-18s . #x%04X)\t; %c")
 485           (t
 486            "(%-18s . #x%02X)\t; %c"))
 487     name
 488     (if (= (charset-iso-graphic-plane name) 1)
 489         (logior value
 490                 (cond ((= (charset-dimension name) 1)
 491                        #x80)
 492                       ((= (charset-dimension name) 2)
 493                        #x8080)
 494                       ((= (charset-dimension name) 3)
 495                        #x808080)
 496                       (t 0)))
 497       value)
 498     (char-db-decode-isolated-char name value)))
 499   (if (and (= (charset-chars name) 94)
 500            (= (charset-dimension name) 2))
 501       (insert (format " [%02d-%02d]"
 502                       (- (lsh value -8) 32)
 503                       (- (logand value 255) 32))))
 504   (insert line-breaking))
 505
 506 (defun char-db-insert-relation-feature (char name value line-breaking
 507                                              ccss readable)
 508   (insert (format "(%-18s%s " name line-breaking))
 509   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 510         separator cell sources required-features
 511         ret)
 512     (while (consp value)
 513       (setq cell (car value))
 514       (if (integerp cell)
 515           (setq cell (decode-char '=ucs cell)))
 516       (cond
 517        ((eq name '->subsumptive)
 518         (when (or (not (some (lambda (atr)
 519                                (get-char-attribute cell atr))
 520                              char-db-ignored-attributes))
 521                   (some (lambda (ccs)
 522                           (encode-char cell ccs 'defined-only))
 523                         ccss))
 524           (if separator
 525               (insert lbs))
 526           (let ((char-db-ignored-attributes
 527                  (cons '<-subsumptive
 528                        char-db-ignored-attributes)))
 529             (insert-char-attributes cell readable nil nil 'for-sub-node))
 530           (setq separator lbs))
 531         )
 532        ((characterp cell)
 533         (setq sources
 534               (get-char-attribute
 535                char (intern (format "%s*sources" name))))
 536         (setq required-features nil)
 537         (dolist (source sources)
 538           (cond
 539            ((memq source '(JP
 540                            JP/Jouyou
 541                            shinjigen shinjigen@1ed shinjigen@rev))
 542             (setq required-features
 543                   (union required-features
 544                          '(=jis-x0208
 545                            =jis-x0208@1990
 546                            =jis-x0213-1@2000
 547                            =jis-x0213-1@2004
 548                            =jis-x0213-2
 549                            =jis-x0212
 550                            =jis-x0208@1983
 551                            =jis-x0208@1978
 552                            =shinjigen))))
 553            ((eq source 'CN)
 554             (setq required-features
 555                   (union required-features
 556                          '(=gb2312
 557                            =gb12345
 558                            =iso-ir165)))))
 559           (cond
 560            ((find-charset (setq ret (intern (format "=%s" source))))
 561             (setq required-features
 562                   (cons ret required-features)))
 563            (t (setq required-features
 564                     (cons source required-features)))))
 565         (cond ((string-match "@JP" (symbol-name name))
 566                (setq required-features
 567                      (union required-features
 568                             '(=jis-x0208
 569                               =jis-x0208@1990
 570                               =jis-x0213-1-2000
 571                               =jis-x0213-2-2000
 572                               =jis-x0212
 573                               =jis-x0208@1983
 574                               =jis-x0208@1978))))
 575               ((string-match "@CN" (symbol-name name))
 576                (setq required-features
 577                      (union required-features
 578                             '(=gb2312
 579                               =gb12345
 580                               =iso-ir165)))))
 581         (if separator
 582             (insert lbs))
 583         (if readable
 584             (insert (format "%S" cell))
 585           (char-db-insert-char-spec cell readable
 586                                     nil
 587                                     required-features))
 588         (setq separator lbs))
 589        ((consp cell)
 590         (if separator
 591             (insert lbs))
 592         (if (consp (car cell))
 593             (char-db-insert-char-spec cell readable)
 594           (char-db-insert-char-reference cell readable))
 595         (setq separator lbs))
 596        (t
 597         (if separator
 598             (insert separator))
 599         (insert (prin1-to-string cell))
 600         (setq separator " ")))
 601       (setq value (cdr value)))
 602     (insert ")")
 603     (insert line-breaking)))
 604
 605 (defun insert-char-attributes (char &optional readable attributes column
 606                                     for-sub-node)
 607   (unless column
 608     (setq column (current-column)))
 609   (let (name value ; has-long-ccs-name
 610         rest
 611         radical strokes
 612         (line-breaking
 613          (concat "\n" (make-string (1+ column) ?\ )))
 614         lbs cell separator ret
 615         key al cal
 616         dest-ccss ; sources required-features
 617         ccss)
 618     (let (atr-d)
 619       (setq attributes
 620             (sort (if attributes
 621                       (if (consp attributes)
 622                           (progn
 623                             (dolist (name attributes)
 624                               (unless (memq name char-db-ignored-attributes)
 625                                 (if (find-charset name)
 626                                     (push name ccss))
 627                                 (push name atr-d)))
 628                             atr-d))
 629                     (dolist (name (char-attribute-list))
 630                       (unless (memq name char-db-ignored-attributes)
 631                         (if (find-charset name)
 632                             (push name ccss))
 633                         (push name atr-d)))
 634                     atr-d)
 635                   #'char-attribute-name<)))
 636     (insert "(")
 637     (when (memq '<-subsumptive attributes)
 638       (when (or readable (not for-sub-node))
 639         (when (setq value (get-char-attribute char '<-subsumptive))
 640           (char-db-insert-relation-feature char '<-subsumptive value
 641                                            line-breaking
 642                                            ccss readable)))
 643       (setq attributes (delq '<-subsumptive attributes)))
 644     (when (and (memq '<-denotational attributes)
 645                (setq value (get-char-attribute char '<-denotational)))
 646       (char-db-insert-relation-feature char '<-denotational value
 647                                        line-breaking
 648                                        ccss readable)
 649       (setq attributes (delq '<-denotational attributes)))
 650     (when (and (memq 'name attributes)
 651                (setq value (get-char-attribute char 'name)))
 652       (insert (format
 653                (if (> (+ (current-column) (length value)) 48)
 654                    "(name . %S)%s"
 655                  "(name               . %S)%s")
 656                value line-breaking))
 657       (setq attributes (delq 'name attributes))
 658       )
 659     (when (and (memq 'name* attributes)
 660                (setq value (get-char-attribute char 'name*)))
 661       (insert (format
 662                (if (> (+ (current-column) (length value)) 48)
 663                    "(name* . %S)%s"
 664                  "(name*              . %S)%s")
 665                value line-breaking))
 666       (setq attributes (delq 'name* attributes))
 667       )
 668     (when (and (memq 'script attributes)
 669                (setq value (get-char-attribute char 'script)))
 670       (insert (format "(script\t\t%s)%s"
 671                       (mapconcat (function prin1-to-string)
 672                                  value " ")
 673                       line-breaking))
 674       (setq attributes (delq 'script attributes))
 675       )
 676     (dolist (name '(=>ucs =>ucs*))
 677       (when (and (memq name attributes)
 678                  (setq value (get-char-attribute char name)))
 679         (insert (format "(%-18s . #x%04X)\t; %c%s"
 680                         name value (decode-char '=ucs value)
 681                         line-breaking))
 682         (setq attributes (delq name attributes))))
 683     (dolist (name '(=>ucs@gb =>ucs@big5))
 684       (when (and (memq name attributes)
 685                  (setq value (get-char-attribute char name)))
 686         (insert (format "(%-18s . #x%04X)\t; %c%s"
 687                         name value
 688                         (decode-char (intern
 689                                       (concat "="
 690                                               (substring
 691                                                (symbol-name name) 2)))
 692                                      value)
 693                         line-breaking))
 694         (setq attributes (delq name attributes))
 695         ))
 696     ;; (dolist (name '(=>daikanwa))
 697     ;;   (when (and (memq name attributes)
 698     ;;              (setq value (get-char-attribute char name)))
 699     ;;     (insert
 700     ;;      (if (integerp value)
 701     ;;          (format "(%-18s . %05d)\t; %c%s"
 702     ;;                  name value (decode-char '=daikanwa value)
 703     ;;                  line-breaking)
 704     ;;        (format "(%-18s %s)\t; %c%s"
 705     ;;                name
 706     ;;                (mapconcat (function prin1-to-string)
 707     ;;                           value " ")
 708     ;;                (char-representative-of-daikanwa char)
 709     ;;                line-breaking)))
 710     ;;     (setq attributes (delq name attributes))))
 711     (when (and (memq 'general-category attributes)
 712                (setq value (get-char-attribute char 'general-category)))
 713       (insert (format
 714                "(general-category\t%s) ; %s%s"
 715                (mapconcat (lambda (cell)
 716                             (format "%S" cell))
 717                           value " ")
 718                (cond ((rassoc value unidata-normative-category-alist)
 719                       "Normative Category")
 720                      ((rassoc value unidata-informative-category-alist)
 721                       "Informative Category")
 722                      (t
 723                       "Unknown Category"))
 724                line-breaking))
 725       (setq attributes (delq 'general-category attributes))
 726       )
 727     (when (and (memq 'bidi-category attributes)
 728                (setq value (get-char-attribute char 'bidi-category)))
 729       (insert (format "(bidi-category\t. %S)%s"
 730                       value
 731                       line-breaking))
 732       (setq attributes (delq 'bidi-category attributes))
 733       )
 734     (unless (or (not (memq 'mirrored attributes))
 735                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
 736                     'empty))
 737       (insert (format "(mirrored\t\t. %S)%s"
 738                       value
 739                       line-breaking))
 740       (setq attributes (delq 'mirrored attributes))
 741       )
 742     (cond
 743      ((and (memq 'decimal-digit-value attributes)
 744            (setq value (get-char-attribute char 'decimal-digit-value)))
 745       (insert (format "(decimal-digit-value . %S)%s"
 746                       value
 747                       line-breaking))
 748       (setq attributes (delq 'decimal-digit-value attributes))
 749       (when (and (memq 'digit-value attributes)
 750                  (setq value (get-char-attribute char 'digit-value)))
 751         (insert (format "(digit-value\t . %S)%s"
 752                         value
 753                         line-breaking))
 754         (setq attributes (delq 'digit-value attributes))
 755         )
 756       (when (and (memq 'numeric-value attributes)
 757                  (setq value (get-char-attribute char 'numeric-value)))
 758         (insert (format "(numeric-value\t . %S)%s"
 759                         value
 760                         line-breaking))
 761         (setq attributes (delq 'numeric-value attributes))
 762         )
 763       )
 764      (t
 765       (when (and (memq 'digit-value attributes)
 766                  (setq value (get-char-attribute char 'digit-value)))
 767         (insert (format "(digit-value\t. %S)%s"
 768                         value
 769                         line-breaking))
 770         (setq attributes (delq 'digit-value attributes))
 771         )
 772       (when (and (memq 'numeric-value attributes)
 773                  (setq value (get-char-attribute char 'numeric-value)))
 774         (insert (format "(numeric-value\t. %S)%s"
 775                         value
 776                         line-breaking))
 777         (setq attributes (delq 'numeric-value attributes))
 778         )))
 779     (when (and (memq 'iso-10646-comment attributes)
 780                (setq value (get-char-attribute char 'iso-10646-comment)))
 781       (insert (format "(iso-10646-comment\t. %S)%s"
 782                       value
 783                       line-breaking))
 784       (setq attributes (delq 'iso-10646-comment attributes))
 785       )
 786     (when (and (memq 'morohashi-daikanwa attributes)
 787                (setq value (get-char-attribute char 'morohashi-daikanwa)))
 788       (insert (format "(morohashi-daikanwa\t%s)%s"
 789                       (mapconcat (function prin1-to-string) value " ")
 790                       line-breaking))
 791       (setq attributes (delq 'morohashi-daikanwa attributes))
 792       )
 793     (setq radical nil
 794           strokes nil)
 795     (when (and (memq 'ideographic-radical attributes)
 796                (setq value (get-char-attribute char 'ideographic-radical)))
 797       (setq radical value)
 798       (insert (format "(ideographic-radical . %S)\t; %c%s"
 799                       radical
 800                       (ideographic-radical radical)
 801                       line-breaking))
 802       (setq attributes (delq 'ideographic-radical attributes))
 803       )
 804     (when (and (memq 'shuowen-radical attributes)
 805                (setq value (get-char-attribute char 'shuowen-radical)))
 806       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
 807                       value
 808                       (shuowen-radical value)
 809                       line-breaking))
 810       (setq attributes (delq 'shuowen-radical attributes))
 811       )
 812     (let (key)
 813       (dolist (domain
 814                (append
 815                 char-db-feature-domains
 816                 (let (dest domain)
 817                   (dolist (feature (char-attribute-list))
 818                     (setq feature (symbol-name feature))
 819                     (when (string-match
 820                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
 821                            feature)
 822                       (setq domain (intern (match-string 2 feature)))
 823                      (unless (memq domain dest)
 824                        (setq dest (cons domain dest)))))
 825                   (sort dest #'string<))))
 826         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
 827         (when (and (memq key attributes)
 828                    (setq value (get-char-attribute char key)))
 829           (setq radical value)
 830           (insert (format "(%s . %S)\t; %c%s"
 831                           key
 832                           radical
 833                           (ideographic-radical radical)
 834                           line-breaking))
 835           (setq attributes (delq key attributes))
 836           )
 837         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
 838         (when (and (memq key attributes)
 839                    (setq value (get-char-attribute char key)))
 840           (setq strokes value)
 841           (insert (format "(%s . %S)%s"
 842                           key
 843                           strokes
 844                           line-breaking))
 845           (setq attributes (delq key attributes))
 846           )
 847         (setq key (intern (format "%s@%s" 'total-strokes domain)))
 848         (when (and (memq key attributes)
 849                    (setq value (get-char-attribute char key)))
 850           (insert (format "(%s       . %S)%s"
 851                           key
 852                           value
 853                           line-breaking))
 854           (setq attributes (delq key attributes))
 855           )
 856         (dolist (feature '(ideographic-radical
 857                            ideographic-strokes
 858                            total-strokes))
 859           (setq key (intern (format "%s@%s*sources" feature domain)))
 860           (when (and (memq key attributes)
 861                      (setq value (get-char-attribute char key)))
 862             (insert (format "(%s%s" key line-breaking))
 863             (dolist (cell value)
 864               (insert (format " %s" cell)))
 865             (insert ")")
 866             (insert line-breaking)
 867             (setq attributes (delq key attributes))
 868             ))
 869         ))
 870     (when (and (memq 'ideographic-strokes attributes)
 871                (setq value (get-char-attribute char 'ideographic-strokes)))
 872       (setq strokes value)
 873       (insert (format "(ideographic-strokes . %S)%s"
 874                       strokes
 875                       line-breaking))
 876       (setq attributes (delq 'ideographic-strokes attributes))
 877       )
 878     (when (and (memq 'kangxi-radical attributes)
 879                (setq value (get-char-attribute char 'kangxi-radical)))
 880       (unless (eq value radical)
 881         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
 882                         value
 883                         (ideographic-radical value)
 884                         line-breaking))
 885         (or radical
 886             (setq radical value)))
 887       (setq attributes (delq 'kangxi-radical attributes))
 888       )
 889     (when (and (memq 'kangxi-strokes attributes)
 890                (setq value (get-char-attribute char 'kangxi-strokes)))
 891       (unless (eq value strokes)
 892         (insert (format "(kangxi-strokes\t . %S)%s"
 893                         value
 894                         line-breaking))
 895         (or strokes
 896             (setq strokes value)))
 897       (setq attributes (delq 'kangxi-strokes attributes))
 898       )
 899     (when (and (memq 'japanese-radical attributes)
 900                (setq value (get-char-attribute char 'japanese-radical)))
 901       (unless (eq value radical)
 902         (insert (format "(japanese-radical\t . %S)\t; %c%s"
 903                         value
 904                         (ideographic-radical value)
 905                         line-breaking))
 906         (or radical
 907             (setq radical value)))
 908       (setq attributes (delq 'japanese-radical attributes))
 909       )
 910     (when (and (memq 'japanese-strokes attributes)
 911                (setq value (get-char-attribute char 'japanese-strokes)))
 912       (unless (eq value strokes)
 913         (insert (format "(japanese-strokes\t . %S)%s"
 914                         value
 915                         line-breaking))
 916         (or strokes
 917             (setq strokes value)))
 918       (setq attributes (delq 'japanese-strokes attributes))
 919       )
 920     (when (and (memq 'cns-radical attributes)
 921                (setq value (get-char-attribute char 'cns-radical)))
 922       (insert (format "(cns-radical\t . %S)\t; %c%s"
 923                       value
 924                       (ideographic-radical value)
 925                       line-breaking))
 926       (setq attributes (delq 'cns-radical attributes))
 927       )
 928     (when (and (memq 'cns-strokes attributes)
 929                (setq value (get-char-attribute char 'cns-strokes)))
 930       (unless (eq value strokes)
 931         (insert (format "(cns-strokes\t . %S)%s"
 932                         value
 933                         line-breaking))
 934         (or strokes
 935             (setq strokes value)))
 936       (setq attributes (delq 'cns-strokes attributes))
 937       )
 938     (when (and (memq 'shinjigen-1-radical attributes)
 939                (setq value (get-char-attribute char 'shinjigen-1-radical)))
 940       (unless (eq value radical)
 941         (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
 942                         value
 943                         (ideographic-radical value)
 944                         line-breaking))
 945         (or radical
 946             (setq radical value)))
 947       (setq attributes (delq 'shinjigen-1-radical attributes))
 948       )
 949     (when (and (memq 'ideographic- attributes)
 950                (setq value (get-char-attribute char 'ideographic-)))
 951       (insert "(ideographic-       ")
 952       (setq lbs (concat "\n" (make-string (current-column) ?\ ))
 953             separator nil)
 954       (while (consp value)
 955         (setq cell (car value))
 956         (if (integerp cell)
 957             (setq cell (decode-char '=ucs cell)))
 958         (cond ((characterp cell)
 959                (if separator
 960                    (insert lbs))
 961                (if readable
 962                    (insert (format "%S" cell))
 963                  (char-db-insert-char-spec cell readable))
 964                (setq separator lbs))
 965               ((consp cell)
 966                (if separator
 967                    (insert lbs))
 968                (if (consp (car cell))
 969                    (char-db-insert-char-spec cell readable)
 970                  (char-db-insert-char-reference cell readable))
 971                (setq separator lbs))
 972               (t
 973                (if separator
 974                    (insert separator))
 975                (insert (prin1-to-string cell))
 976                (setq separator " ")))
 977         (setq value (cdr value)))
 978       (insert ")")
 979       (insert line-breaking)
 980       (setq attributes (delq 'ideographic- attributes)))
 981     (when (and (memq 'total-strokes attributes)
 982                (setq value (get-char-attribute char 'total-strokes)))
 983       (insert (format "(total-strokes       . %S)%s"
 984                       value
 985                       line-breaking))
 986       (setq attributes (delq 'total-strokes attributes))
 987       )
 988     (when (and (memq '->ideograph attributes)
 989                (setq value (get-char-attribute char '->ideograph)))
 990       (insert (format "(->ideograph\t%s)%s"
 991                       (mapconcat (lambda (code)
 992                                    (cond ((symbolp code)
 993                                           (symbol-name code))
 994                                          ((integerp code)
 995                                           (format "#x%04X" code))
 996                                          (t
 997                                           (format "%s %S"
 998                                                   line-breaking code))))
 999                                  value " ")
1000                       line-breaking))
1001       (setq attributes (delq '->ideograph attributes))
1002       )
1003     ;; (when (and (memq '->decomposition attributes)
1004     ;;            (setq value (get-char-attribute char '->decomposition)))
1005     ;;   (insert (format "(->decomposition\t%s)%s"
1006     ;;                   (mapconcat (lambda (code)
1007     ;;                                (cond ((symbolp code)
1008     ;;                                       (symbol-name code))
1009     ;;                                      ((characterp code)
1010     ;;                                       (if readable
1011     ;;                                           (format "%S" code)
1012     ;;                                         (format "#x%04X"
1013     ;;                                                 (char-int code))
1014     ;;                                         ))
1015     ;;                                      ((integerp code)
1016     ;;                                       (format "#x%04X" code))
1017     ;;                                      (t
1018     ;;                                       (format "%s%S" line-breaking code))))
1019     ;;                              value " ")
1020     ;;                   line-breaking))
1021     ;;   (setq attributes (delq '->decomposition attributes))
1022     ;;   )
1023     (if (equal (get-char-attribute char '->titlecase)
1024                (get-char-attribute char '->uppercase))
1025         (setq attributes (delq '->titlecase attributes)))
1026     (when (and (memq '->mojikyo attributes)
1027                (setq value (get-char-attribute char '->mojikyo)))
1028       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1029                       value (decode-char 'mojikyo value)
1030                       line-breaking))
1031       (setq attributes (delq '->mojikyo attributes))
1032       )
1033     (when (and (memq 'hanyu-dazidian-vol attributes)
1034                (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1035       (insert (format "(hanyu-dazidian-vol  . %d)%s"
1036                       value line-breaking))
1037       (setq attributes (delq 'hanyu-dazidian-vol attributes))
1038       )
1039     (when (and (memq 'hanyu-dazidian-page attributes)
1040                (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1041       (insert (format "(hanyu-dazidian-page . %d)%s"
1042                       value line-breaking))
1043       (setq attributes (delq 'hanyu-dazidian-page attributes))
1044       )
1045     (when (and (memq 'hanyu-dazidian-char attributes)
1046                (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1047       (insert (format "(hanyu-dazidian-char . %d)%s"
1048                       value line-breaking))
1049       (setq attributes (delq 'hanyu-dazidian-char attributes))
1050       )
1051     (unless readable
1052       (dolist (ignored '(composition
1053                          ->denotational <-subsumptive ->ucs-unified
1054                          ->ideographic-component-forms))
1055         (setq attributes (delq ignored attributes))))
1056     (while attributes
1057       (setq name (car attributes))
1058       (if (setq value (get-char-attribute char name))
1059           (cond ((setq ret (find-charset name))
1060                  (setq name (charset-name ret))
1061                  (if (and (not (memq name dest-ccss))
1062                           (prog1
1063                               (setq value (get-char-attribute char name))
1064                             (setq dest-ccss (cons name dest-ccss))))
1065                      (char-db-insert-ccs-feature name value line-breaking))
1066                  )
1067                 ((string-match "^=>ucs@" (symbol-name name))
1068                  (insert (format "(%-18s . #x%04X)\t; %c%s"
1069                                  name value (decode-char '=ucs value)
1070                                  line-breaking))
1071                  )
1072                 ((eq name 'jisx0208-1978/4X)
1073                  (insert (format "(%-18s . #x%04X)%s"
1074                                  name value
1075                                  line-breaking))
1076                  )
1077                 ((and
1078                   (not readable)
1079                   (not (eq name '->subsumptive))
1080                   (not (eq name '->uppercase))
1081                   (not (eq name '->lowercase))
1082                   (not (eq name '->titlecase))
1083                   (not (eq name '->canonical))
1084                   (not (eq name '->Bopomofo))
1085                   (not (eq name '->mistakable))
1086                   (not (eq name '->ideographic-variants))
1087                   (null (get-char-attribute
1088                          char (intern (format "%s*sources" name))))
1089                   (not (string-match "\\*sources$" (symbol-name name)))
1090                   (null (get-char-attribute
1091                          char (intern (format "%s*note" name))))
1092                   (not (string-match "\\*note$" (symbol-name name)))
1093                   (or (eq name '<-identical)
1094                       (eq name '<-uppercase)
1095                       (eq name '<-lowercase)
1096                       (eq name '<-titlecase)
1097                       (eq name '<-canonical)
1098                       (eq name '<-ideographic-variants)
1099                       ;; (eq name '<-synonyms)
1100                       (string-match "^<-synonyms" (symbol-name name))
1101                       (eq name '<-mistakable)
1102                       (when (string-match "^->" (symbol-name name))
1103                         (cond
1104                          ((string-match "^->fullwidth" (symbol-name name))
1105                           (not (and (consp value)
1106                                     (characterp (car value))
1107                                     (encode-char
1108                                      (car value) '=ucs 'defined-only)))
1109                           )
1110                          (t)))
1111                       ))
1112                  )
1113                 ((or (eq name 'ideographic-structure)
1114                      (eq name 'ideographic-combination)
1115                      (eq name 'ideographic-)
1116                      (eq name '=decomposition)
1117                      (char-feature-base-name= '=decomposition name)
1118                      (char-feature-base-name= '=>decomposition name)
1119                      ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1120                      ;;               (symbol-name name))
1121                      (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1122                      (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1123                                    (symbol-name name))
1124                      )
1125                  (char-db-insert-relation-feature char name value
1126                                                   line-breaking
1127                                                   ccss readable))
1128                 ((memq name '(ideograph=
1129                               original-ideograph-of
1130                               ancient-ideograph-of
1131                               vulgar-ideograph-of
1132                               wrong-ideograph-of
1133                               ;; simplified-ideograph-of
1134                               ideographic-variants
1135                               ;; ideographic-different-form-of
1136                               ))
1137                  (insert (format "(%-18s%s " name line-breaking))
1138                  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1139                        separator nil)
1140                  (while (consp value)
1141                    (setq cell (car value))
1142                    (if (and (consp cell)
1143                             (consp (car cell)))
1144                        (progn
1145                          (if separator
1146                              (insert lbs))
1147                          (char-db-insert-alist cell readable)
1148                          (setq separator lbs))
1149                      (if separator
1150                          (insert separator))
1151                      (insert (prin1-to-string cell))
1152                      (setq separator " "))
1153                    (setq value (cdr value)))
1154                  (insert ")")
1155                  (insert line-breaking))
1156                 ((consp value)
1157                  (insert (format "(%-18s " name))
1158                  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1159                        separator nil)
1160                  (while (consp value)
1161                    (setq cell (car value))
1162                    (if (and (consp cell)
1163                             (consp (car cell))
1164                             (setq ret (condition-case nil
1165                                           (find-char cell)
1166                                         (error nil))))
1167                        (progn
1168                          (setq rest cell
1169                                al nil
1170                                cal nil)
1171                          (while rest
1172                            (setq key (car (car rest)))
1173                            (if (find-charset key)
1174                                (setq cal (cons key cal))
1175                              (setq al (cons key al)))
1176                            (setq rest (cdr rest)))
1177                          (if separator
1178                              (insert lbs))
1179                          (insert-char-attributes ret
1180                                                  readable
1181                                                  al ; cal
1182                                                  nil 'for-sub-node)
1183                          (setq separator lbs))
1184                      (setq ret (prin1-to-string cell))
1185                      (if separator
1186                          (if (< (+ (current-column)
1187                                    (length ret)
1188                                    (length separator))
1189                                 76)
1190                              (insert separator)
1191                            (insert lbs)))
1192                      (insert ret)
1193                      (setq separator " "))
1194                    (setq value (cdr value)))
1195                  (insert ")")
1196                  (insert line-breaking))
1197                 (t
1198                  (insert (format "(%-18s" name))
1199                  (setq ret (prin1-to-string value))
1200                  (unless (< (+ (current-column)
1201                                (length ret)
1202                                3)
1203                             76)
1204                    (insert line-breaking))
1205                  (insert " . " ret ")" line-breaking)
1206                  ;; (insert (format "(%-18s . %S)%s"
1207                  ;;                 name value
1208                  ;;                 line-breaking))
1209                  )
1210                 ))
1211       (setq attributes (cdr attributes)))
1212     (insert ")")))
1213
1214 (defun insert-char-data (char &optional readable
1215                               attributes)
1216   (save-restriction
1217     (narrow-to-region (point)(point))
1218     (insert "(define-char
1219   '")
1220     (insert-char-attributes char readable attributes)
1221     (insert ")\n")
1222     (goto-char (point-min))
1223     (while (re-search-forward "[ \t]+$" nil t)
1224       (replace-match ""))
1225     ;; from tabify.
1226     (goto-char (point-min))
1227     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1228       (let ((column (current-column))
1229             (indent-tabs-mode t))
1230         (delete-region (match-beginning 0) (point))
1231         (indent-to column)))
1232     (goto-char (point-max))
1233     ;; (tabify (point-min)(point-max))
1234     ))
1235
1236 (defun insert-char-data-with-variant (char &optional printable
1237                                            no-ucs-unified
1238                                            script excluded-script)
1239   (insert-char-data char printable)
1240   (let ((variants (char-variants char))
1241         rest
1242         variant vs ret)
1243     (setq variants (sort variants #'<))
1244     (setq rest variants)
1245     (setq variants (cons char variants))
1246     (while rest
1247       (setq variant (car rest))
1248       (unless (get-char-attribute variant '<-subsumptive)
1249         (if (and (or (null script)
1250                      (null (setq vs (get-char-attribute variant 'script)))
1251                      (memq script vs))
1252                  (or (null excluded-script)
1253                      (null (setq vs (get-char-attribute variant 'script)))
1254                      (not (memq excluded-script vs))))
1255             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1256               (insert-char-data variant printable)
1257               (if (setq ret (char-variants variant))
1258                   (while ret
1259                     (or (memq (car ret) variants)
1260                         ;; (get-char-attribute (car ret) '<-subsumptive)
1261                         (setq rest (nconc rest (list (car ret)))))
1262                     (setq ret (cdr ret)))))))
1263       (setq rest (cdr rest)))))
1264
1265 (defun insert-char-range-data (min max &optional script excluded-script)
1266   (let ((code min)
1267         char)
1268     (while (<= code max)
1269       (setq char (decode-char '=ucs code))
1270       (if (encode-char char '=ucs 'defined-only)
1271           (insert-char-data-with-variant char nil 'no-ucs-unified
1272                                          script excluded-script))
1273       (setq code (1+ code)))))
1274
1275 (defun write-char-range-data-to-file (min max file
1276                                           &optional script excluded-script)
1277   (let ((coding-system-for-write char-db-file-coding-system))
1278     (with-temp-buffer
1279       (insert (format ";; -*- coding: %s -*-\n"
1280                       char-db-file-coding-system))
1281       (insert-char-range-data min max script excluded-script)
1282       (write-region (point-min)(point-max) file))))
1283
1284 (defvar what-character-original-window-configuration)
1285
1286 ;;;###autoload
1287 (defun what-char-definition (char)
1288   (interactive (list (char-after)))
1289   (let ((buf (get-buffer-create "*Character Description*"))
1290         (the-buf (current-buffer))
1291         (win-conf (current-window-configuration)))
1292     (pop-to-buffer buf)
1293     (make-local-variable 'what-character-original-window-configuration)
1294     (setq what-character-original-window-configuration win-conf)
1295     (setq buffer-read-only nil)
1296     (erase-buffer)
1297     (condition-case err
1298         (progn
1299           (insert-char-data-with-variant char 'printable)
1300           (unless (char-attribute-alist char)
1301             (insert (format ";; = %c\n"
1302                             (let* ((rest (split-char char))
1303                                    (ccs (pop rest))
1304                                    (code (pop rest)))
1305                               (while rest
1306                                 (setq code (logior (lsh code 8)
1307                                                    (pop rest))))
1308                               (decode-char ccs code)))))
1309           ;; (char-db-update-comment)
1310           (set-buffer-modified-p nil)
1311           (view-mode the-buf (lambda (buf)
1312                                (set-window-configuration
1313                                 what-character-original-window-configuration)
1314                                ))
1315           (goto-char (point-min)))
1316       (error (progn
1317                (set-window-configuration
1318                 what-character-original-window-configuration)
1319                (signal (car err) (cdr err)))))))
1320
1321
1322 ;;; @ end
1323 ;;;
1324
1325 (provide 'char-db-util)
1326
1327 ;;; char-db-util.el ends here