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     =hanyo-denshi/ks
 159     =daikanwa
 160     =daikanwa@rev2
 161     =daikanwa@rev1
 162     =cbeta
 163     =gt-k
 164     ideograph-hanziku-1
 165     ideograph-hanziku-2
 166     ideograph-hanziku-3
 167     ideograph-hanziku-4
 168     ideograph-hanziku-5
 169     ideograph-hanziku-6
 170     ideograph-hanziku-7
 171     ideograph-hanziku-8
 172     ideograph-hanziku-9
 173     ideograph-hanziku-10
 174     ideograph-hanziku-11
 175     ideograph-hanziku-12
 176     =>>>jis-x0208
 177     =>>>jis-x0213-1
 178     =>>jis-x0208
 179     =>>jis-x0213-1
 180     =>>jis-x0213-1@2000
 181     =>>jis-x0213-1@2004
 182     =>>jis-x0213-2
 183     =>>jis-x0208@1978
 184     =>>hanyo-denshi/ft
 185     =>>hanyo-denshi/ks
 186     =>>gt
 187     =>>daikanwa
 188     =>jis-x0208@usual
 189     =>jis-x0208
 190     =>jis-x0208@1997
 191     =>jis-x0213-1
 192     =>jis-x0213-1@2000
 193     =>jis-x0213-1@2004
 194     =>jis-x0213-2@usual
 195     =>jis-x0213-2
 196     ==>ucs@bucs
 197     =>ucs@iso
 198     =>ucs@unicode
 199     =>ucs@jis
 200     =>ucs@JP
 201     =>ucs@cns
 202     =>ucs@ks
 203     =>>ucs@iso
 204     =>>ucs@unicode
 205     =>>ucs@jis
 206     =>>ucs@cns
 207     =>>>ucs@iso
 208     =>>>ucs@unicode
 209     =ucs@iso
 210     =ucs@unicode
 211     =>>big5-cdp
 212     =>>gt-k
 213     =>gt
 214     =>big5-cdp
 215     =>daikanwa
 216     =big5
 217     =big5-eten
 218     =>gt-k
 219     =zinbun-oracle
 220     =>zinbun-oracle
 221     =ruimoku-v6
 222     =>>ruimoku-v6
 223     =jef-china3
 224     =shinjigen))
 225
 226
 227 ;;; @ char-db formatters
 228 ;;;
 229
 230 (defun char-db-make-char-spec (char)
 231   (let (ret char-spec)
 232     (cond ((characterp char)
 233            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 234                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 235                   (setq char-spec (list (cons '=ucs ret)))
 236                   (cond ((setq ret (get-char-attribute char 'name))
 237                          (setq char-spec (cons (cons 'name ret) char-spec))
 238                          )
 239                         ((setq ret (get-char-attribute char 'name*))
 240                          (setq char-spec (cons (cons 'name* ret) char-spec))
 241                          ))
 242                   )
 243                  ((setq ret
 244                         (catch 'tag
 245                           (let ((rest char-db-coded-charset-priority-list)
 246                                 ccs)
 247                             (while rest
 248                               (setq ccs (charset-name
 249                                          (find-charset (car rest))))
 250                               (if (setq ret
 251                                         (encode-char char ccs
 252                                                      'defined-only))
 253                                   (throw 'tag (cons ccs ret)))
 254                               (setq rest (cdr rest))))))
 255                   (setq char-spec (list ret))
 256                   (dolist (ccs (delq (car ret) (charset-list)))
 257                     (if (and (or (charset-iso-final-char ccs)
 258                                  (memq ccs
 259                                        '(=daikanwa
 260                                          =daikanwa@rev2
 261                                          ;; =gt-k
 262                                          =jis-x0208@1997
 263                                          ))
 264                                  (string-match "=ucs@" (symbol-name ccs)))
 265                              (setq ccs (charset-name ccs))
 266                              (null (assq ccs char-spec))
 267                              (setq ret (encode-char char ccs 'defined-only)))
 268                         (setq char-spec (cons (cons ccs ret) char-spec))))
 269                   (if (null char-spec)
 270                       (setq char-spec (split-char char)))
 271                   (cond ((setq ret (get-char-attribute char 'name))
 272                          (setq char-spec (cons (cons 'name ret) char-spec))
 273                          )
 274                         ((setq ret (get-char-attribute char 'name*))
 275                          (setq char-spec (cons (cons 'name* ret) char-spec))
 276                          ))
 277                   )
 278                  ((setq ret (get-char-attribute
 279                              char 'ideographic-combination))
 280                   (setq char-spec
 281                         (cons (cons 'ideographic-combination ret)
 282                               char-spec))
 283                   ))
 284            char-spec)
 285           ((consp char)
 286            char))))
 287
 288 (defun char-db-insert-char-spec (char &optional readable column
 289                                       required-features)
 290   (unless column
 291     (setq column (current-column)))
 292   (let (char-spec temp-char)
 293     (setq char-spec (char-db-make-char-spec char))
 294     (unless (or (characterp char) ; char
 295                 (condition-case nil
 296                     (setq char (find-char char-spec))
 297                   (error nil)))
 298       ;; define temporary character
 299       ;;   Current implementation is dirty.
 300       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 301                                          char-spec)))
 302       (remove-char-attribute temp-char 'ideograph-daikanwa)
 303       (setq char temp-char))
 304     (insert-char-attributes char
 305                             readable
 306                             (union (mapcar #'car char-spec)
 307                                    required-features)
 308                             nil 'for-sub-node)
 309     (when temp-char
 310       ;; undefine temporary character
 311       ;;   Current implementation is dirty.
 312       (setq char-spec (char-attribute-alist temp-char))
 313       (while char-spec
 314         (remove-char-attribute temp-char (car (car char-spec)))
 315         (setq char-spec (cdr char-spec))))))
 316
 317 (defun char-db-insert-alist (alist &optional readable column)
 318   (unless column
 319     (setq column (current-column)))
 320   (let ((line-breaking
 321          (concat "\n" (make-string (1+ column) ?\ )))
 322         name value
 323         ret al ; cal
 324         key
 325         lbs cell rest separator)
 326     (insert "(")
 327     (while alist
 328       (setq name (car (car alist))
 329             value (cdr (car alist)))
 330       (cond ((eq name 'char)
 331              (insert "(char . ")
 332              (if (setq ret (condition-case nil
 333                                (find-char value)
 334                              (error nil)))
 335                  (progn
 336                    (setq al nil
 337                          ;; cal nil
 338                          )
 339                    (while value
 340                      (setq key (car (car value)))
 341                      ;; (if (find-charset key)
 342                      ;;     (setq cal (cons key cal))
 343                      (setq al (cons key al))
 344                      ;; )
 345                      (setq value (cdr value)))
 346                    (insert-char-attributes ret
 347                                            readable
 348                                            (or al 'none) ; cal
 349                                            nil 'for-sub-node))
 350                (insert (prin1-to-string value)))
 351              (insert ")")
 352              (insert line-breaking))
 353             ((consp value)
 354              (insert (format "(%-18s " name))
 355              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 356              (while (consp value)
 357                (setq cell (car value))
 358                (if (and (consp cell)
 359                         (consp (car cell))
 360                         (setq ret (condition-case nil
 361                                       (find-char cell)
 362                                     (error nil)))
 363                         )
 364                    (progn
 365                      (setq rest cell
 366                            al nil
 367                            ;; cal nil
 368                            )
 369                      (while rest
 370                        (setq key (car (car rest)))
 371                        ;; (if (find-charset key)
 372                        ;;     (setq cal (cons key cal))
 373                        (setq al (cons key al))
 374                        ;; )
 375                        (setq rest (cdr rest)))
 376                      (if separator
 377                          (insert lbs))
 378                      (insert-char-attributes ret
 379                                              readable
 380                                              al ; cal
 381                                              nil 'for-sub-node)
 382                      (setq separator lbs))
 383                  (if separator
 384                      (insert separator))
 385                  (insert (prin1-to-string cell))
 386                  (setq separator " "))
 387                (setq value (cdr value)))
 388              (insert ")")
 389              (insert line-breaking))
 390             (t
 391              (insert (format "(%-18s . %S)%s"
 392                              name value
 393                              line-breaking))))
 394       (setq alist (cdr alist))))
 395   (insert ")"))
 396
 397 (defun char-db-insert-char-reference (plist &optional readable column)
 398   (unless column
 399     (setq column (current-column)))
 400   (let ((line-breaking
 401          (concat "\n" (make-string (1+ column) ?\ )))
 402         (separator "")
 403         name value)
 404     (insert "(")
 405     (while plist
 406       (setq name (pop plist))
 407       (setq value (pop plist))
 408       (cond ((eq name :char)
 409              (insert separator)
 410              (insert ":char\t")
 411              (cond ((numberp value)
 412                     (setq value (decode-char '=ucs value)))
 413                    ;; ((consp value)
 414                    ;;  (setq value (or (find-char value)
 415                    ;;                  value)))
 416                    )
 417              (char-db-insert-char-spec value readable)
 418              (insert line-breaking)
 419              (setq separator ""))
 420             ((eq name :radical)
 421              (insert (format "%s%s\t%d ; %c%s"
 422                              separator
 423                              name value
 424                              (ideographic-radical value)
 425                              line-breaking))
 426              (setq separator ""))
 427             (t
 428              (insert (format "%s%s\t%S" separator name value))
 429              (setq separator line-breaking)))
 430       ))
 431   (insert ")"))
 432
 433 (defun char-db-decode-isolated-char (ccs code-point)
 434   (let (ret)
 435     (setq ret
 436           (cond ((eq ccs 'arabic-iso8859-6)
 437                  (decode-char ccs code-point))
 438                 ((and (memq ccs '(=gt-pj-1
 439                                   =gt-pj-2
 440                                   =gt-pj-3
 441                                   =gt-pj-4
 442                                   =gt-pj-5
 443                                   =gt-pj-6
 444                                   =gt-pj-7
 445                                   =gt-pj-8
 446                                   =gt-pj-9
 447                                   =gt-pj-10
 448                                   =gt-pj-11))
 449                       (setq ret (decode-char ccs code-point))
 450                       (setq ret (encode-char ret '=gt 'defined-only)))
 451                  (decode-builtin-char '=gt ret))
 452                 (t
 453                  (decode-builtin-char ccs code-point))))
 454     (cond ((and (<= 0 (char-int ret))
 455                 (<= (char-int ret) #x1F))
 456            (decode-char '=ucs (+ #x2400 (char-int ret))))
 457           ((= (char-int ret) #x7F)
 458            ?\u2421)
 459           (t ret))))
 460
 461 (defvar char-db-convert-obsolete-format t)
 462
 463 (defun char-db-insert-ccs-feature (name value line-breaking)
 464   (insert
 465    (format
 466     (cond ((memq name '(=shinjigen
 467                         =shinjigen@1ed
 468                         =shinjigen@rev =shinjigen/+p@rev
 469                         =daikanwa/ho))
 470            "(%-18s .  %04d)\t; %c")
 471           ((eq name '=shinjigen@1ed/24pr)
 472            "(%-18s . %04d)\t; %c")
 473           ((or (memq name '(=daikanwa =>>daikanwa =>daikanwa
 474                             =daikanwa@rev1 =daikanwa@rev2
 475                             =daikanwa/+p =daikanwa/+2p
 476                             =gt =>>>gt =>>gt =>gt
 477                             =gt-k =>>gt-k =>gt-k
 478                             =>>adobe-japan1
 479                             =cbeta =>>cbeta
 480                             =zinbun-oracle =>zinbun-oracle))
 481                (string-match "^=adobe-" (symbol-name name)))
 482            "(%-18s . %05d)\t; %c")
 483           ((memq name '(=hanyo-denshi/ks =>>hanyo-denshi/ks mojikyo))
 484            "(%-18s . %06d)\t; %c")
 485           ((>= (charset-dimension name) 2)
 486            "(%-18s . #x%04X)\t; %c")
 487           (t
 488            "(%-18s . #x%02X)\t; %c"))
 489     name
 490     (if (= (charset-iso-graphic-plane name) 1)
 491         (logior value
 492                 (cond ((= (charset-dimension name) 1)
 493                        #x80)
 494                       ((= (charset-dimension name) 2)
 495                        #x8080)
 496                       ((= (charset-dimension name) 3)
 497                        #x808080)
 498                       (t 0)))
 499       value)
 500     (char-db-decode-isolated-char name value)))
 501   (if (and (= (charset-chars name) 94)
 502            (= (charset-dimension name) 2))
 503       (insert (format " [%02d-%02d]"
 504                       (- (lsh value -8) 32)
 505                       (- (logand value 255) 32))))
 506   (insert line-breaking))
 507
 508 (defun char-db-insert-relation-feature (char name value line-breaking
 509                                              ccss readable)
 510   (insert (format "(%-18s%s " name line-breaking))
 511   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 512         separator cell sources required-features
 513         ret)
 514     (while (consp value)
 515       (setq cell (car value))
 516       (if (integerp cell)
 517           (setq cell (decode-char '=ucs cell)))
 518       (cond
 519        ((eq name '->subsumptive)
 520         (when (or (not (some (lambda (atr)
 521                                (get-char-attribute cell atr))
 522                              char-db-ignored-attributes))
 523                   (some (lambda (ccs)
 524                           (encode-char cell ccs 'defined-only))
 525                         ccss))
 526           (if separator
 527               (insert lbs))
 528           (let ((char-db-ignored-attributes
 529                  (cons '<-subsumptive
 530                        char-db-ignored-attributes)))
 531             (insert-char-attributes cell readable nil nil 'for-sub-node))
 532           (setq separator lbs))
 533         )
 534        ((characterp cell)
 535         (setq sources
 536               (get-char-attribute
 537                char (intern (format "%s*sources" name))))
 538         (setq required-features nil)
 539         (dolist (source sources)
 540           (cond
 541            ((memq source '(JP
 542                            JP/Jouyou
 543                            shinjigen shinjigen@1ed shinjigen@rev))
 544             (setq required-features
 545                   (union required-features
 546                          '(=jis-x0208
 547                            =jis-x0208@1990
 548                            =jis-x0213-1@2000
 549                            =jis-x0213-1@2004
 550                            =jis-x0213-2
 551                            =jis-x0212
 552                            =jis-x0208@1983
 553                            =jis-x0208@1978
 554                            =shinjigen))))
 555            ((eq source 'CN)
 556             (setq required-features
 557                   (union required-features
 558                          '(=gb2312
 559                            =gb12345
 560                            =iso-ir165)))))
 561           (cond
 562            ((find-charset (setq ret (intern (format "=%s" source))))
 563             (setq required-features
 564                   (cons ret required-features)))
 565            (t (setq required-features
 566                     (cons source required-features)))))
 567         (cond ((string-match "@JP" (symbol-name name))
 568                (setq required-features
 569                      (union required-features
 570                             '(=jis-x0208
 571                               =jis-x0208@1990
 572                               =jis-x0213-1-2000
 573                               =jis-x0213-2-2000
 574                               =jis-x0212
 575                               =jis-x0208@1983
 576                               =jis-x0208@1978))))
 577               ((string-match "@CN" (symbol-name name))
 578                (setq required-features
 579                      (union required-features
 580                             '(=gb2312
 581                               =gb12345
 582                               =iso-ir165)))))
 583         (if separator
 584             (insert lbs))
 585         (if readable
 586             (insert (format "%S" cell))
 587           (char-db-insert-char-spec cell readable
 588                                     nil
 589                                     required-features))
 590         (setq separator lbs))
 591        ((consp cell)
 592         (if separator
 593             (insert lbs))
 594         (if (consp (car cell))
 595             (char-db-insert-char-spec cell readable)
 596           (char-db-insert-char-reference cell readable))
 597         (setq separator lbs))
 598        (t
 599         (if separator
 600             (insert separator))
 601         (insert (prin1-to-string cell))
 602         (setq separator " ")))
 603       (setq value (cdr value)))
 604     (insert ")")
 605     (insert line-breaking)))
 606
 607 (defun insert-char-attributes (char &optional readable attributes column
 608                                     for-sub-node)
 609   (unless column
 610     (setq column (current-column)))
 611   (let (name value ; has-long-ccs-name
 612         rest
 613         radical strokes
 614         (line-breaking
 615          (concat "\n" (make-string (1+ column) ?\ )))
 616         lbs cell separator ret
 617         key al cal
 618         dest-ccss ; sources required-features
 619         ccss)
 620     (let (atr-d)
 621       (setq attributes
 622             (sort (if attributes
 623                       (if (consp attributes)
 624                           (progn
 625                             (dolist (name attributes)
 626                               (unless (memq name char-db-ignored-attributes)
 627                                 (if (find-charset name)
 628                                     (push name ccss))
 629                                 (push name atr-d)))
 630                             atr-d))
 631                     (dolist (name (char-attribute-list))
 632                       (unless (memq name char-db-ignored-attributes)
 633                         (if (find-charset name)
 634                             (push name ccss))
 635                         (push name atr-d)))
 636                     atr-d)
 637                   #'char-attribute-name<)))
 638     (insert "(")
 639     (when (memq '<-subsumptive attributes)
 640       (when (or readable (not for-sub-node))
 641         (when (setq value (get-char-attribute char '<-subsumptive))
 642           (char-db-insert-relation-feature char '<-subsumptive value
 643                                            line-breaking
 644                                            ccss readable)))
 645       (setq attributes (delq '<-subsumptive attributes)))
 646     (when (and (memq '<-denotational attributes)
 647                (setq value (get-char-attribute char '<-denotational)))
 648       (char-db-insert-relation-feature char '<-denotational value
 649                                        line-breaking
 650                                        ccss readable)
 651       (setq attributes (delq '<-denotational attributes)))
 652     (when (and (memq 'name attributes)
 653                (setq value (get-char-attribute char 'name)))
 654       (insert (format
 655                (if (> (+ (current-column) (length value)) 48)
 656                    "(name . %S)%s"
 657                  "(name               . %S)%s")
 658                value line-breaking))
 659       (setq attributes (delq 'name attributes))
 660       )
 661     (when (and (memq 'name* attributes)
 662                (setq value (get-char-attribute char 'name*)))
 663       (insert (format
 664                (if (> (+ (current-column) (length value)) 48)
 665                    "(name* . %S)%s"
 666                  "(name*              . %S)%s")
 667                value line-breaking))
 668       (setq attributes (delq 'name* attributes))
 669       )
 670     (when (and (memq 'script attributes)
 671                (setq value (get-char-attribute char 'script)))
 672       (insert (format "(script\t\t%s)%s"
 673                       (mapconcat (function prin1-to-string)
 674                                  value " ")
 675                       line-breaking))
 676       (setq attributes (delq 'script attributes))
 677       )
 678     (dolist (name '(=>ucs =>ucs*))
 679       (when (and (memq name attributes)
 680                  (setq value (get-char-attribute char name)))
 681         (insert (format "(%-18s . #x%04X)\t; %c%s"
 682                         name value (decode-char '=ucs value)
 683                         line-breaking))
 684         (setq attributes (delq name attributes))))
 685     (dolist (name '(=>ucs@gb =>ucs@big5))
 686       (when (and (memq name attributes)
 687                  (setq value (get-char-attribute char name)))
 688         (insert (format "(%-18s . #x%04X)\t; %c%s"
 689                         name value
 690                         (decode-char (intern
 691                                       (concat "="
 692                                               (substring
 693                                                (symbol-name name) 2)))
 694                                      value)
 695                         line-breaking))
 696         (setq attributes (delq name attributes))
 697         ))
 698     ;; (dolist (name '(=>daikanwa))
 699     ;;   (when (and (memq name attributes)
 700     ;;              (setq value (get-char-attribute char name)))
 701     ;;     (insert
 702     ;;      (if (integerp value)
 703     ;;          (format "(%-18s . %05d)\t; %c%s"
 704     ;;                  name value (decode-char '=daikanwa value)
 705     ;;                  line-breaking)
 706     ;;        (format "(%-18s %s)\t; %c%s"
 707     ;;                name
 708     ;;                (mapconcat (function prin1-to-string)
 709     ;;                           value " ")
 710     ;;                (char-representative-of-daikanwa char)
 711     ;;                line-breaking)))
 712     ;;     (setq attributes (delq name attributes))))
 713     (when (and (memq 'general-category attributes)
 714                (setq value (get-char-attribute char 'general-category)))
 715       (insert (format
 716                "(general-category\t%s) ; %s%s"
 717                (mapconcat (lambda (cell)
 718                             (format "%S" cell))
 719                           value " ")
 720                (cond ((rassoc value unidata-normative-category-alist)
 721                       "Normative Category")
 722                      ((rassoc value unidata-informative-category-alist)
 723                       "Informative Category")
 724                      (t
 725                       "Unknown Category"))
 726                line-breaking))
 727       (setq attributes (delq 'general-category attributes))
 728       )
 729     (when (and (memq 'bidi-category attributes)
 730                (setq value (get-char-attribute char 'bidi-category)))
 731       (insert (format "(bidi-category\t. %S)%s"
 732                       value
 733                       line-breaking))
 734       (setq attributes (delq 'bidi-category attributes))
 735       )
 736     (unless (or (not (memq 'mirrored attributes))
 737                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
 738                     'empty))
 739       (insert (format "(mirrored\t\t. %S)%s"
 740                       value
 741                       line-breaking))
 742       (setq attributes (delq 'mirrored attributes))
 743       )
 744     (cond
 745      ((and (memq 'decimal-digit-value attributes)
 746            (setq value (get-char-attribute char 'decimal-digit-value)))
 747       (insert (format "(decimal-digit-value . %S)%s"
 748                       value
 749                       line-breaking))
 750       (setq attributes (delq 'decimal-digit-value attributes))
 751       (when (and (memq 'digit-value attributes)
 752                  (setq value (get-char-attribute char 'digit-value)))
 753         (insert (format "(digit-value\t . %S)%s"
 754                         value
 755                         line-breaking))
 756         (setq attributes (delq 'digit-value attributes))
 757         )
 758       (when (and (memq 'numeric-value attributes)
 759                  (setq value (get-char-attribute char 'numeric-value)))
 760         (insert (format "(numeric-value\t . %S)%s"
 761                         value
 762                         line-breaking))
 763         (setq attributes (delq 'numeric-value attributes))
 764         )
 765       )
 766      (t
 767       (when (and (memq 'digit-value attributes)
 768                  (setq value (get-char-attribute char 'digit-value)))
 769         (insert (format "(digit-value\t. %S)%s"
 770                         value
 771                         line-breaking))
 772         (setq attributes (delq 'digit-value attributes))
 773         )
 774       (when (and (memq 'numeric-value attributes)
 775                  (setq value (get-char-attribute char 'numeric-value)))
 776         (insert (format "(numeric-value\t. %S)%s"
 777                         value
 778                         line-breaking))
 779         (setq attributes (delq 'numeric-value attributes))
 780         )))
 781     (when (and (memq 'iso-10646-comment attributes)
 782                (setq value (get-char-attribute char 'iso-10646-comment)))
 783       (insert (format "(iso-10646-comment\t. %S)%s"
 784                       value
 785                       line-breaking))
 786       (setq attributes (delq 'iso-10646-comment attributes))
 787       )
 788     (when (and (memq 'morohashi-daikanwa attributes)
 789                (setq value (get-char-attribute char 'morohashi-daikanwa)))
 790       (insert (format "(morohashi-daikanwa\t%s)%s"
 791                       (mapconcat (function prin1-to-string) value " ")
 792                       line-breaking))
 793       (setq attributes (delq 'morohashi-daikanwa attributes))
 794       )
 795     (setq radical nil
 796           strokes nil)
 797     (when (and (memq 'ideographic-radical attributes)
 798                (setq value (get-char-attribute char 'ideographic-radical)))
 799       (setq radical value)
 800       (insert (format "(ideographic-radical . %S)\t; %c%s"
 801                       radical
 802                       (ideographic-radical radical)
 803                       line-breaking))
 804       (setq attributes (delq 'ideographic-radical attributes))
 805       )
 806     (when (and (memq 'shuowen-radical attributes)
 807                (setq value (get-char-attribute char 'shuowen-radical)))
 808       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
 809                       value
 810                       (shuowen-radical value)
 811                       line-breaking))
 812       (setq attributes (delq 'shuowen-radical attributes))
 813       )
 814     (let (key)
 815       (dolist (domain
 816                (append
 817                 char-db-feature-domains
 818                 (let (dest domain)
 819                   (dolist (feature (char-attribute-list))
 820                     (setq feature (symbol-name feature))
 821                     (when (string-match
 822                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
 823                            feature)
 824                       (setq domain (intern (match-string 2 feature)))
 825                      (unless (memq domain dest)
 826                        (setq dest (cons domain dest)))))
 827                   (sort dest #'string<))))
 828         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
 829         (when (and (memq key attributes)
 830                    (setq value (get-char-attribute char key)))
 831           (setq radical value)
 832           (insert (format "(%s . %S)\t; %c%s"
 833                           key
 834                           radical
 835                           (ideographic-radical radical)
 836                           line-breaking))
 837           (setq attributes (delq key attributes))
 838           )
 839         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
 840         (when (and (memq key attributes)
 841                    (setq value (get-char-attribute char key)))
 842           (setq strokes value)
 843           (insert (format "(%s . %S)%s"
 844                           key
 845                           strokes
 846                           line-breaking))
 847           (setq attributes (delq key attributes))
 848           )
 849         (setq key (intern (format "%s@%s" 'total-strokes domain)))
 850         (when (and (memq key attributes)
 851                    (setq value (get-char-attribute char key)))
 852           (insert (format "(%s       . %S)%s"
 853                           key
 854                           value
 855                           line-breaking))
 856           (setq attributes (delq key attributes))
 857           )
 858         (dolist (feature '(ideographic-radical
 859                            ideographic-strokes
 860                            total-strokes))
 861           (setq key (intern (format "%s@%s*sources" feature domain)))
 862           (when (and (memq key attributes)
 863                      (setq value (get-char-attribute char key)))
 864             (insert (format "(%s%s" key line-breaking))
 865             (dolist (cell value)
 866               (insert (format " %s" cell)))
 867             (insert ")")
 868             (insert line-breaking)
 869             (setq attributes (delq key attributes))
 870             ))
 871         ))
 872     (when (and (memq 'ideographic-strokes attributes)
 873                (setq value (get-char-attribute char 'ideographic-strokes)))
 874       (setq strokes value)
 875       (insert (format "(ideographic-strokes . %S)%s"
 876                       strokes
 877                       line-breaking))
 878       (setq attributes (delq 'ideographic-strokes attributes))
 879       )
 880     (when (and (memq 'kangxi-radical attributes)
 881                (setq value (get-char-attribute char 'kangxi-radical)))
 882       (unless (eq value radical)
 883         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
 884                         value
 885                         (ideographic-radical value)
 886                         line-breaking))
 887         (or radical
 888             (setq radical value)))
 889       (setq attributes (delq 'kangxi-radical attributes))
 890       )
 891     (when (and (memq 'kangxi-strokes attributes)
 892                (setq value (get-char-attribute char 'kangxi-strokes)))
 893       (unless (eq value strokes)
 894         (insert (format "(kangxi-strokes\t . %S)%s"
 895                         value
 896                         line-breaking))
 897         (or strokes
 898             (setq strokes value)))
 899       (setq attributes (delq 'kangxi-strokes attributes))
 900       )
 901     (when (and (memq 'japanese-radical attributes)
 902                (setq value (get-char-attribute char 'japanese-radical)))
 903       (unless (eq value radical)
 904         (insert (format "(japanese-radical\t . %S)\t; %c%s"
 905                         value
 906                         (ideographic-radical value)
 907                         line-breaking))
 908         (or radical
 909             (setq radical value)))
 910       (setq attributes (delq 'japanese-radical attributes))
 911       )
 912     (when (and (memq 'japanese-strokes attributes)
 913                (setq value (get-char-attribute char 'japanese-strokes)))
 914       (unless (eq value strokes)
 915         (insert (format "(japanese-strokes\t . %S)%s"
 916                         value
 917                         line-breaking))
 918         (or strokes
 919             (setq strokes value)))
 920       (setq attributes (delq 'japanese-strokes attributes))
 921       )
 922     (when (and (memq 'cns-radical attributes)
 923                (setq value (get-char-attribute char 'cns-radical)))
 924       (insert (format "(cns-radical\t . %S)\t; %c%s"
 925                       value
 926                       (ideographic-radical value)
 927                       line-breaking))
 928       (setq attributes (delq 'cns-radical attributes))
 929       )
 930     (when (and (memq 'cns-strokes attributes)
 931                (setq value (get-char-attribute char 'cns-strokes)))
 932       (unless (eq value strokes)
 933         (insert (format "(cns-strokes\t . %S)%s"
 934                         value
 935                         line-breaking))
 936         (or strokes
 937             (setq strokes value)))
 938       (setq attributes (delq 'cns-strokes attributes))
 939       )
 940     (when (and (memq 'shinjigen-1-radical attributes)
 941                (setq value (get-char-attribute char 'shinjigen-1-radical)))
 942       (unless (eq value radical)
 943         (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
 944                         value
 945                         (ideographic-radical value)
 946                         line-breaking))
 947         (or radical
 948             (setq radical value)))
 949       (setq attributes (delq 'shinjigen-1-radical attributes))
 950       )
 951     (when (and (memq 'ideographic- attributes)
 952                (setq value (get-char-attribute char 'ideographic-)))
 953       (insert "(ideographic-       ")
 954       (setq lbs (concat "\n" (make-string (current-column) ?\ ))
 955             separator nil)
 956       (while (consp value)
 957         (setq cell (car value))
 958         (if (integerp cell)
 959             (setq cell (decode-char '=ucs cell)))
 960         (cond ((characterp cell)
 961                (if separator
 962                    (insert lbs))
 963                (if readable
 964                    (insert (format "%S" cell))
 965                  (char-db-insert-char-spec cell readable))
 966                (setq separator lbs))
 967               ((consp cell)
 968                (if separator
 969                    (insert lbs))
 970                (if (consp (car cell))
 971                    (char-db-insert-char-spec cell readable)
 972                  (char-db-insert-char-reference cell readable))
 973                (setq separator lbs))
 974               (t
 975                (if separator
 976                    (insert separator))
 977                (insert (prin1-to-string cell))
 978                (setq separator " ")))
 979         (setq value (cdr value)))
 980       (insert ")")
 981       (insert line-breaking)
 982       (setq attributes (delq 'ideographic- attributes)))
 983     (when (and (memq 'total-strokes attributes)
 984                (setq value (get-char-attribute char 'total-strokes)))
 985       (insert (format "(total-strokes       . %S)%s"
 986                       value
 987                       line-breaking))
 988       (setq attributes (delq 'total-strokes attributes))
 989       )
 990     (when (and (memq '->ideograph attributes)
 991                (setq value (get-char-attribute char '->ideograph)))
 992       (insert (format "(->ideograph\t%s)%s"
 993                       (mapconcat (lambda (code)
 994                                    (cond ((symbolp code)
 995                                           (symbol-name code))
 996                                          ((integerp code)
 997                                           (format "#x%04X" code))
 998                                          (t
 999                                           (format "%s %S"
1000                                                   line-breaking code))))
1001                                  value " ")
1002                       line-breaking))
1003       (setq attributes (delq '->ideograph attributes))
1004       )
1005     ;; (when (and (memq '->decomposition attributes)
1006     ;;            (setq value (get-char-attribute char '->decomposition)))
1007     ;;   (insert (format "(->decomposition\t%s)%s"
1008     ;;                   (mapconcat (lambda (code)
1009     ;;                                (cond ((symbolp code)
1010     ;;                                       (symbol-name code))
1011     ;;                                      ((characterp code)
1012     ;;                                       (if readable
1013     ;;                                           (format "%S" code)
1014     ;;                                         (format "#x%04X"
1015     ;;                                                 (char-int code))
1016     ;;                                         ))
1017     ;;                                      ((integerp code)
1018     ;;                                       (format "#x%04X" code))
1019     ;;                                      (t
1020     ;;                                       (format "%s%S" line-breaking code))))
1021     ;;                              value " ")
1022     ;;                   line-breaking))
1023     ;;   (setq attributes (delq '->decomposition attributes))
1024     ;;   )
1025     (if (equal (get-char-attribute char '->titlecase)
1026                (get-char-attribute char '->uppercase))
1027         (setq attributes (delq '->titlecase attributes)))
1028     (when (and (memq '->mojikyo attributes)
1029                (setq value (get-char-attribute char '->mojikyo)))
1030       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1031                       value (decode-char 'mojikyo value)
1032                       line-breaking))
1033       (setq attributes (delq '->mojikyo attributes))
1034       )
1035     (when (and (memq 'hanyu-dazidian-vol attributes)
1036                (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1037       (insert (format "(hanyu-dazidian-vol  . %d)%s"
1038                       value line-breaking))
1039       (setq attributes (delq 'hanyu-dazidian-vol attributes))
1040       )
1041     (when (and (memq 'hanyu-dazidian-page attributes)
1042                (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1043       (insert (format "(hanyu-dazidian-page . %d)%s"
1044                       value line-breaking))
1045       (setq attributes (delq 'hanyu-dazidian-page attributes))
1046       )
1047     (when (and (memq 'hanyu-dazidian-char attributes)
1048                (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1049       (insert (format "(hanyu-dazidian-char . %d)%s"
1050                       value line-breaking))
1051       (setq attributes (delq 'hanyu-dazidian-char attributes))
1052       )
1053     (unless readable
1054       (dolist (ignored '(composition
1055                          ->denotational <-subsumptive ->ucs-unified
1056                          ->ideographic-component-forms))
1057         (setq attributes (delq ignored attributes))))
1058     (while attributes
1059       (setq name (car attributes))
1060       (if (setq value (get-char-attribute char name))
1061           (cond ((setq ret (find-charset name))
1062                  (setq name (charset-name ret))
1063                  (if (and (not (memq name dest-ccss))
1064                           (prog1
1065                               (setq value (get-char-attribute char name))
1066                             (setq dest-ccss (cons name dest-ccss))))
1067                      (char-db-insert-ccs-feature name value line-breaking))
1068                  )
1069                 ((string-match "^=>ucs@" (symbol-name name))
1070                  (insert (format "(%-18s . #x%04X)\t; %c%s"
1071                                  name value (decode-char '=ucs value)
1072                                  line-breaking))
1073                  )
1074                 ((eq name 'jisx0208-1978/4X)
1075                  (insert (format "(%-18s . #x%04X)%s"
1076                                  name value
1077                                  line-breaking))
1078                  )
1079                 ((and
1080                   (not readable)
1081                   (not (eq name '->subsumptive))
1082                   (not (eq name '->uppercase))
1083                   (not (eq name '->lowercase))
1084                   (not (eq name '->titlecase))
1085                   (not (eq name '->canonical))
1086                   (not (eq name '->Bopomofo))
1087                   (not (eq name '->mistakable))
1088                   (not (eq name '->ideographic-variants))
1089                   (null (get-char-attribute
1090                          char (intern (format "%s*sources" name))))
1091                   (not (string-match "\\*sources$" (symbol-name name)))
1092                   (null (get-char-attribute
1093                          char (intern (format "%s*note" name))))
1094                   (not (string-match "\\*note$" (symbol-name name)))
1095                   (or (eq name '<-identical)
1096                       (eq name '<-uppercase)
1097                       (eq name '<-lowercase)
1098                       (eq name '<-titlecase)
1099                       (eq name '<-canonical)
1100                       (eq name '<-ideographic-variants)
1101                       ;; (eq name '<-synonyms)
1102                       (string-match "^<-synonyms" (symbol-name name))
1103                       (eq name '<-mistakable)
1104                       (when (string-match "^->" (symbol-name name))
1105                         (cond
1106                          ((string-match "^->fullwidth" (symbol-name name))
1107                           (not (and (consp value)
1108                                     (characterp (car value))
1109                                     (encode-char
1110                                      (car value) '=ucs 'defined-only)))
1111                           )
1112                          (t)))
1113                       ))
1114                  )
1115                 ((or (eq name 'ideographic-structure)
1116                      (eq name 'ideographic-combination)
1117                      (eq name 'ideographic-)
1118                      (eq name '=decomposition)
1119                      (char-feature-base-name= '=decomposition name)
1120                      (char-feature-base-name= '=>decomposition name)
1121                      ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1122                      ;;               (symbol-name name))
1123                      (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1124                      (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1125                                    (symbol-name name))
1126                      )
1127                  (char-db-insert-relation-feature char name value
1128                                                   line-breaking
1129                                                   ccss readable))
1130                 ((memq name '(ideograph=
1131                               original-ideograph-of
1132                               ancient-ideograph-of
1133                               vulgar-ideograph-of
1134                               wrong-ideograph-of
1135                               ;; simplified-ideograph-of
1136                               ideographic-variants
1137                               ;; ideographic-different-form-of
1138                               ))
1139                  (insert (format "(%-18s%s " name line-breaking))
1140                  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1141                        separator nil)
1142                  (while (consp value)
1143                    (setq cell (car value))
1144                    (if (and (consp cell)
1145                             (consp (car cell)))
1146                        (progn
1147                          (if separator
1148                              (insert lbs))
1149                          (char-db-insert-alist cell readable)
1150                          (setq separator lbs))
1151                      (if separator
1152                          (insert separator))
1153                      (insert (prin1-to-string cell))
1154                      (setq separator " "))
1155                    (setq value (cdr value)))
1156                  (insert ")")
1157                  (insert line-breaking))
1158                 ((consp value)
1159                  (insert (format "(%-18s " name))
1160                  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1161                        separator nil)
1162                  (while (consp value)
1163                    (setq cell (car value))
1164                    (if (and (consp cell)
1165                             (consp (car cell))
1166                             (setq ret (condition-case nil
1167                                           (find-char cell)
1168                                         (error nil))))
1169                        (progn
1170                          (setq rest cell
1171                                al nil
1172                                cal nil)
1173                          (while rest
1174                            (setq key (car (car rest)))
1175                            (if (find-charset key)
1176                                (setq cal (cons key cal))
1177                              (setq al (cons key al)))
1178                            (setq rest (cdr rest)))
1179                          (if separator
1180                              (insert lbs))
1181                          (insert-char-attributes ret
1182                                                  readable
1183                                                  al ; cal
1184                                                  nil 'for-sub-node)
1185                          (setq separator lbs))
1186                      (setq ret (prin1-to-string cell))
1187                      (if separator
1188                          (if (< (+ (current-column)
1189                                    (length ret)
1190                                    (length separator))
1191                                 76)
1192                              (insert separator)
1193                            (insert lbs)))
1194                      (insert ret)
1195                      (setq separator " "))
1196                    (setq value (cdr value)))
1197                  (insert ")")
1198                  (insert line-breaking))
1199                 (t
1200                  (insert (format "(%-18s" name))
1201                  (setq ret (prin1-to-string value))
1202                  (unless (< (+ (current-column)
1203                                (length ret)
1204                                3)
1205                             76)
1206                    (insert line-breaking))
1207                  (insert " . " ret ")" line-breaking)
1208                  ;; (insert (format "(%-18s . %S)%s"
1209                  ;;                 name value
1210                  ;;                 line-breaking))
1211                  )
1212                 ))
1213       (setq attributes (cdr attributes)))
1214     (insert ")")))
1215
1216 (defun insert-char-data (char &optional readable
1217                               attributes)
1218   (save-restriction
1219     (narrow-to-region (point)(point))
1220     (insert "(define-char
1221   '")
1222     (insert-char-attributes char readable attributes)
1223     (insert ")\n")
1224     (goto-char (point-min))
1225     (while (re-search-forward "[ \t]+$" nil t)
1226       (replace-match ""))
1227     ;; from tabify.
1228     (goto-char (point-min))
1229     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1230       (let ((column (current-column))
1231             (indent-tabs-mode t))
1232         (delete-region (match-beginning 0) (point))
1233         (indent-to column)))
1234     (goto-char (point-max))
1235     ;; (tabify (point-min)(point-max))
1236     ))
1237
1238 (defun insert-char-data-with-variant (char &optional printable
1239                                            no-ucs-unified
1240                                            script excluded-script)
1241   (insert-char-data char printable)
1242   (let ((variants (char-variants char))
1243         rest
1244         variant vs ret)
1245     (setq variants (sort variants #'<))
1246     (setq rest variants)
1247     (setq variants (cons char variants))
1248     (while rest
1249       (setq variant (car rest))
1250       (unless (get-char-attribute variant '<-subsumptive)
1251         (if (and (or (null script)
1252                      (null (setq vs (get-char-attribute variant 'script)))
1253                      (memq script vs))
1254                  (or (null excluded-script)
1255                      (null (setq vs (get-char-attribute variant 'script)))
1256                      (not (memq excluded-script vs))))
1257             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1258               (insert-char-data variant printable)
1259               (if (setq ret (char-variants variant))
1260                   (while ret
1261                     (or (memq (car ret) variants)
1262                         ;; (get-char-attribute (car ret) '<-subsumptive)
1263                         (setq rest (nconc rest (list (car ret)))))
1264                     (setq ret (cdr ret)))))))
1265       (setq rest (cdr rest)))))
1266
1267 (defun insert-char-range-data (min max &optional script excluded-script)
1268   (let ((code min)
1269         char)
1270     (while (<= code max)
1271       (setq char (decode-char '=ucs code))
1272       (if (encode-char char '=ucs 'defined-only)
1273           (insert-char-data-with-variant char nil 'no-ucs-unified
1274                                          script excluded-script))
1275       (setq code (1+ code)))))
1276
1277 (defun write-char-range-data-to-file (min max file
1278                                           &optional script excluded-script)
1279   (let ((coding-system-for-write char-db-file-coding-system))
1280     (with-temp-buffer
1281       (insert (format ";; -*- coding: %s -*-\n"
1282                       char-db-file-coding-system))
1283       (insert-char-range-data min max script excluded-script)
1284       (write-region (point-min)(point-max) file))))
1285
1286 (defvar what-character-original-window-configuration)
1287
1288 ;;;###autoload
1289 (defun what-char-definition (char)
1290   (interactive (list (char-after)))
1291   (let ((buf (get-buffer-create "*Character Description*"))
1292         (the-buf (current-buffer))
1293         (win-conf (current-window-configuration)))
1294     (pop-to-buffer buf)
1295     (make-local-variable 'what-character-original-window-configuration)
1296     (setq what-character-original-window-configuration win-conf)
1297     (setq buffer-read-only nil)
1298     (erase-buffer)
1299     (condition-case err
1300         (progn
1301           (insert-char-data-with-variant char 'printable)
1302           (unless (char-attribute-alist char)
1303             (insert (format ";; = %c\n"
1304                             (let* ((rest (split-char char))
1305                                    (ccs (pop rest))
1306                                    (code (pop rest)))
1307                               (while rest
1308                                 (setq code (logior (lsh code 8)
1309                                                    (pop rest))))
1310                               (decode-char ccs code)))))
1311           ;; (char-db-update-comment)
1312           (set-buffer-modified-p nil)
1313           (view-mode the-buf (lambda (buf)
1314                                (set-window-configuration
1315                                 what-character-original-window-configuration)
1316                                ))
1317           (goto-char (point-min)))
1318       (error (progn
1319                (set-window-configuration
1320                 what-character-original-window-configuration)
1321                (signal (car err) (cdr err)))))))
1322
1323
1324 ;;; @ end
1325 ;;;
1326
1327 (provide 'char-db-util)
1328
1329 ;;; char-db-util.el ends here