git.chise.org Git - chise/xemacs-chise.git-/blob - lisp/utf-2000/char-db-util.el

   1 ;;; char-db-util.el --- Character Database utility -*- coding: utf-8-er; -*-
   2
   3 ;; Copyright (C) 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006,
   4 ;;   2007, 2008, 2009, 2010, 2011 MORIOKA Tomohiko.
   5
   6 ;; Author: MORIOKA Tomohiko <tomo@kanji.zinbun.kyoto-u.ac.jp>
   7 ;; Keywords: CHISE, Character Database, ISO/IEC 10646, UCS, Unicode, MULE.
   8
   9 ;; This file is part of XEmacs CHISE.
  10
  11 ;; XEmacs CHISE is free software; you can redistribute it and/or
  12 ;; modify it under the terms of the GNU General Public License as
  13 ;; published by the Free Software Foundation; either version 2, or (at
  14 ;; your option) any later version.
  15
  16 ;; XEmacs CHISE is distributed in the hope that it will be useful,
  17 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  18 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  19 ;; General Public License for more details.
  20
  21 ;; You should have received a copy of the GNU General Public License
  22 ;; along with XEmacs CHISE; see the file COPYING.  If not, write to
  23 ;; the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  24 ;; Boston, MA 02111-1307, USA.
  25
  26 ;;; Code:
  27
  28 (require 'chise-subr)
  29 (require 'ideograph-subr)
  30
  31 (defconst unidata-normative-category-alist
  32   '(("Lu" letter        uppercase)
  33     ("Ll" letter        lowercase)
  34     ("Lt" letter        titlecase)
  35     ("Mn" mark          non-spacing)
  36     ("Mc" mark          spacing-combining)
  37     ("Me" mark          enclosing)
  38     ("Nd" number        decimal-digit)
  39     ("Nl" number        letter)
  40     ("No" number        other)
  41     ("Zs" separator     space)
  42     ("Zl" separator     line)
  43     ("Zp" separator     paragraph)
  44     ("Cc" other         control)
  45     ("Cf" other         format)
  46     ("Cs" other         surrogate)
  47     ("Co" other         private-use)
  48     ("Cn" other         not-assigned)))
  49
  50 (defconst unidata-informative-category-alist
  51   '(("Lm" letter        modifier)
  52     ("Lo" letter        other)
  53     ("Pc" punctuation   connector)
  54     ("Pd" punctuation   dash)
  55     ("Ps" punctuation   open)
  56     ("Pe" punctuation   close)
  57     ("Pi" punctuation   initial-quote)
  58     ("Pf" punctuation   final-quote)
  59     ("Po" punctuation   other)
  60     ("Sm" symbol        math)
  61     ("Sc" symbol        currency)
  62     ("Sk" symbol        modifier)
  63     ("So" symbol        other)
  64     ))
  65
  66 (defconst shuowen-radicals
  67   [?一 ?上 ?示 ?三 ?王 ?玉 ?玨 ?气 ?士 ?丨 ; 010
  68    ?屮 ?艸 ?蓐 ?茻 ?小 ?八 ?釆 ?半 ?牛 ?犛 ; 020
  69    ?告 ?口 ?凵 ?吅 ?哭 ?走 ?止 ?癶 ?步 ?此 ; 030
  70    ?正 ?是 ?辵 ?彳 ?廴 ?㢟 ?行 ?齒 ?牙 ?足 ; 040
  71    ?疋 ?品 ?龠 ?冊 ?㗊 ?舌 ?干 ?谷 ?只 ?㕯 ; 050
  72    ?句 ?丩 ?古 ?十 ?卅 ?言 ?誩 ?音 ?䇂 ?丵 ; 060
  73    ?菐 ?𠬞 ?𠬜 ?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    ?旨 ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?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/jt
 159     =hanyo-denshi/ks
 160     =daikanwa
 161     =daikanwa@rev2
 162     =daikanwa@rev1
 163     =cbeta
 164     =gt-k
 165     ideograph-hanziku-1
 166     ideograph-hanziku-2
 167     ideograph-hanziku-3
 168     ideograph-hanziku-4
 169     ideograph-hanziku-5
 170     ideograph-hanziku-6
 171     ideograph-hanziku-7
 172     ideograph-hanziku-8
 173     ideograph-hanziku-9
 174     ideograph-hanziku-10
 175     ideograph-hanziku-11
 176     ideograph-hanziku-12
 177     =>>>jis-x0208
 178     =>>>jis-x0213-1
 179     =>>>jis-x0213-2
 180     =>>>gt
 181     =>>>adobe-japan1
 182     =>>jis-x0208
 183     =>>jis-x0213-1
 184     =>>jis-x0213-1@2000
 185     =>>jis-x0213-1@2004
 186     =>>jis-x0213-2
 187     =>>jis-x0208@1978
 188     =>>hanyo-denshi/ft
 189     =>>hanyo-denshi/jt
 190     =>>hanyo-denshi/ks
 191     =>>gt
 192     =>>daikanwa
 193     =>>adobe-japan1
 194     =+>jis-x0208
 195     =+>jis-x0213-1
 196     =+>jis-x0213-2
 197     =+>jis-x0208@1978
 198     =+>adobe-japan1
 199     =>jis-x0208
 200     =>jis-x0208@1997
 201     =>jis-x0213-1
 202     =>jis-x0213-1@2000
 203     =>jis-x0213-1@2004
 204     =>jis-x0213-2
 205     ==>ucs@bucs
 206     =>ucs@iso
 207     =>ucs@unicode
 208     =>ucs@jis
 209     =>ucs@cns
 210     =>ucs@ks
 211     =+>ucs@iso
 212     =+>ucs@unicode
 213     =+>ucs@jis
 214     =+>ucs@jis/1990
 215     =+>ucs@cns
 216     =+>ucs@ks
 217     =>>ucs@iso
 218     =>>ucs@unicode
 219     =>>ucs@jis
 220     =>>ucs@cns
 221     =>>>ucs@iso
 222     =>>>ucs@unicode
 223     =ucs@iso
 224     =ucs@unicode
 225     =>>big5-cdp
 226     =>>gt-k
 227     =+>gt
 228     =>gt
 229     =>big5-cdp
 230     =>daikanwa
 231     =>cns11643-7
 232     =big5
 233     =big5-eten
 234     =>gt-k
 235     =zinbun-oracle
 236     =>zinbun-oracle
 237     =ruimoku-v6
 238     =>>ruimoku-v6
 239     =jef-china3
 240     =shinjigen))
 241
 242
 243 ;;; @ char-db formatters
 244 ;;;
 245
 246 (defun char-db-make-char-spec (char)
 247   (let (ret char-spec)
 248     (cond ((characterp char)
 249            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 250                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 251                   (setq char-spec (list (cons '=ucs ret)))
 252                   (cond ((setq ret (get-char-attribute char 'name))
 253                          (setq char-spec (cons (cons 'name ret) char-spec))
 254                          )
 255                         ((setq ret (get-char-attribute char 'name*))
 256                          (setq char-spec (cons (cons 'name* ret) char-spec))
 257                          ))
 258                   )
 259                  ((setq ret
 260                         (catch 'tag
 261                           (let ((rest char-db-coded-charset-priority-list)
 262                                 ccs)
 263                             (while rest
 264                               (setq ccs (charset-name
 265                                          (find-charset (car rest))))
 266                               (if (setq ret
 267                                         (encode-char char ccs
 268                                                      'defined-only))
 269                                   (throw 'tag (cons ccs ret)))
 270                               (setq rest (cdr rest))))))
 271                   (setq char-spec (list ret))
 272                   (dolist (ccs (delq (car ret) (charset-list)))
 273                     (if (and (or (charset-iso-final-char ccs)
 274                                  (memq ccs
 275                                        '(=daikanwa
 276                                          =daikanwa@rev2
 277                                          ;; =gt-k
 278                                          =jis-x0208@1997
 279                                          ))
 280                                  (string-match "=ucs@" (symbol-name ccs)))
 281                              (setq ccs (charset-name ccs))
 282                              (null (assq ccs char-spec))
 283                              (setq ret (encode-char char ccs 'defined-only)))
 284                         (setq char-spec (cons (cons ccs ret) char-spec))))
 285                   (if (null char-spec)
 286                       (setq char-spec (split-char char)))
 287                   (cond ((setq ret (get-char-attribute char 'name))
 288                          (setq char-spec (cons (cons 'name ret) char-spec))
 289                          )
 290                         ((setq ret (get-char-attribute char 'name*))
 291                          (setq char-spec (cons (cons 'name* ret) char-spec))
 292                          ))
 293                   )
 294                  ((setq ret (get-char-attribute
 295                              char 'ideographic-combination))
 296                   (setq char-spec
 297                         (cons (cons 'ideographic-combination ret)
 298                               char-spec))
 299                   ))
 300            char-spec)
 301           ((consp char)
 302            char))))
 303
 304 (defun char-db-insert-char-spec (char &optional readable column
 305                                       required-features)
 306   (unless column
 307     (setq column (current-column)))
 308   (let (char-spec temp-char)
 309     (setq char-spec (char-db-make-char-spec char))
 310     (unless (or (characterp char) ; char
 311                 (condition-case nil
 312                     (setq char (find-char char-spec))
 313                   (error nil)))
 314       ;; define temporary character
 315       ;;   Current implementation is dirty.
 316       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 317                                          char-spec)))
 318       (remove-char-attribute temp-char 'ideograph-daikanwa)
 319       (setq char temp-char))
 320     (insert-char-attributes char
 321                             readable
 322                             (union (mapcar #'car char-spec)
 323                                    required-features)
 324                             nil 'for-sub-node)
 325     (when temp-char
 326       ;; undefine temporary character
 327       ;;   Current implementation is dirty.
 328       (setq char-spec (char-attribute-alist temp-char))
 329       (while char-spec
 330         (remove-char-attribute temp-char (car (car char-spec)))
 331         (setq char-spec (cdr char-spec))))))
 332
 333 (defun char-db-insert-alist (alist &optional readable column)
 334   (unless column
 335     (setq column (current-column)))
 336   (let ((line-breaking
 337          (concat "\n" (make-string (1+ column) ?\ )))
 338         name value
 339         ret al ; cal
 340         key
 341         lbs cell rest separator)
 342     (insert "(")
 343     (while alist
 344       (setq name (car (car alist))
 345             value (cdr (car alist)))
 346       (cond ((eq name 'char)
 347              (insert "(char . ")
 348              (if (setq ret (condition-case nil
 349                                (find-char value)
 350                              (error nil)))
 351                  (progn
 352                    (setq al nil
 353                          ;; cal nil
 354                          )
 355                    (while value
 356                      (setq key (car (car value)))
 357                      ;; (if (find-charset key)
 358                      ;;     (setq cal (cons key cal))
 359                      (setq al (cons key al))
 360                      ;; )
 361                      (setq value (cdr value)))
 362                    (insert-char-attributes ret
 363                                            readable
 364                                            (or al 'none) ; cal
 365                                            nil 'for-sub-node))
 366                (insert (prin1-to-string value)))
 367              (insert ")")
 368              (insert line-breaking))
 369             ((consp value)
 370              (insert (format "(%-18s " name))
 371              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 372              (while (consp value)
 373                (setq cell (car value))
 374                (if (and (consp cell)
 375                         (consp (car cell))
 376                         (setq ret (condition-case nil
 377                                       (find-char cell)
 378                                     (error nil)))
 379                         )
 380                    (progn
 381                      (setq rest cell
 382                            al nil
 383                            ;; cal nil
 384                            )
 385                      (while rest
 386                        (setq key (car (car rest)))
 387                        ;; (if (find-charset key)
 388                        ;;     (setq cal (cons key cal))
 389                        (setq al (cons key al))
 390                        ;; )
 391                        (setq rest (cdr rest)))
 392                      (if separator
 393                          (insert lbs))
 394                      (insert-char-attributes ret
 395                                              readable
 396                                              al ; cal
 397                                              nil 'for-sub-node)
 398                      (setq separator lbs))
 399                  (if separator
 400                      (insert separator))
 401                  (insert (prin1-to-string cell))
 402                  (setq separator " "))
 403                (setq value (cdr value)))
 404              (insert ")")
 405              (insert line-breaking))
 406             (t
 407              (insert (format "(%-18s . %S)%s"
 408                              name value
 409                              line-breaking))))
 410       (setq alist (cdr alist))))
 411   (insert ")"))
 412
 413 (defun char-db-insert-char-reference (plist &optional readable column)
 414   (unless column
 415     (setq column (current-column)))
 416   (let ((line-breaking
 417          (concat "\n" (make-string (1+ column) ?\ )))
 418         (separator "")
 419         name value)
 420     (insert "(")
 421     (while plist
 422       (setq name (pop plist))
 423       (setq value (pop plist))
 424       (cond ((eq name :char)
 425              (insert separator)
 426              (insert ":char\t")
 427              (cond ((numberp value)
 428                     (setq value (decode-char '=ucs value)))
 429                    ;; ((consp value)
 430                    ;;  (setq value (or (find-char value)
 431                    ;;                  value)))
 432                    )
 433              (char-db-insert-char-spec value readable)
 434              (insert line-breaking)
 435              (setq separator ""))
 436             ((eq name :radical)
 437              (insert (format "%s%s\t%d ; %c%s"
 438                              separator
 439                              name value
 440                              (ideographic-radical value)
 441                              line-breaking))
 442              (setq separator ""))
 443             (t
 444              (insert (format "%s%s\t%S" separator name value))
 445              (setq separator line-breaking)))
 446       ))
 447   (insert ")"))
 448
 449 (defun char-db-decode-isolated-char (ccs code-point)
 450   (let (ret)
 451     (setq ret
 452           (cond ((eq ccs 'arabic-iso8859-6)
 453                  (decode-char ccs code-point))
 454                 ((and (memq ccs '(=gt-pj-1
 455                                   =gt-pj-2
 456                                   =gt-pj-3
 457                                   =gt-pj-4
 458                                   =gt-pj-5
 459                                   =gt-pj-6
 460                                   =gt-pj-7
 461                                   =gt-pj-8
 462                                   =gt-pj-9
 463                                   =gt-pj-10
 464                                   =gt-pj-11))
 465                       (setq ret (decode-char ccs code-point))
 466                       (setq ret (encode-char ret '=gt 'defined-only)))
 467                  (decode-builtin-char '=gt ret))
 468                 (t
 469                  (decode-builtin-char ccs code-point))))
 470     (cond ((and (<= 0 (char-int ret))
 471                 (<= (char-int ret) #x1F))
 472            (decode-char '=ucs (+ #x2400 (char-int ret))))
 473           ((= (char-int ret) #x7F)
 474            ?\u2421)
 475           (t ret))))
 476
 477 (defvar char-db-convert-obsolete-format t)
 478
 479 (defun char-db-insert-ccs-feature (name value line-breaking)
 480   (cond
 481    ((integerp value)
 482     (insert
 483      (format
 484       (cond
 485        ((memq name '(=shinjigen
 486                      =shinjigen@1ed
 487                      =shinjigen@rev =shinjigen/+p@rev
 488                      =daikanwa/ho))
 489         "(%-18s .  %04d)\t; %c")
 490        ((eq name '=shinjigen@1ed/24pr)
 491         "(%-18s . %04d)\t; %c")
 492        ((or (memq name '(=daikanwa =>>daikanwa =>daikanwa
 493                          =daikanwa@rev1 =daikanwa@rev2
 494                          =daikanwa/+p =daikanwa/+2p
 495                          =gt =>>>gt =>>gt =+>gt =>gt
 496                          =gt-k =>>gt-k =>gt-k
 497                          =adobe-japan1-0 =>>>adobe-japan1-0 =>>adobe-japan1-0
 498                          =adobe-japan1-1 =>>>adobe-japan1-1 =>>adobe-japan1-1
 499                          =adobe-japan1-2 =>>>adobe-japan1-2 =>>adobe-japan1-2
 500                          =adobe-japan1-3 =>>>adobe-japan1-3 =>>adobe-japan1-3
 501                          =adobe-japan1-4 =>>>adobe-japan1-4 =>>adobe-japan1-4
 502                          =adobe-japan1-5 =>>>adobe-japan1-5 =>>adobe-japan1-5
 503                          =adobe-japan1-6 =>>>adobe-japan1-6 =>>adobe-japan1-6
 504                          =+>adobe-japan1
 505                          =cbeta =>>cbeta
 506                          =zinbun-oracle =>zinbun-oracle))
 507             ;; (string-match "^=adobe-" (symbol-name name))
 508             )
 509         "(%-18s . %05d)\t; %c")
 510        ((memq name '(=hanyo-denshi/ks =>>hanyo-denshi/ks mojikyo))
 511         "(%-18s . %06d)\t; %c")
 512        ((>= (charset-dimension name) 2)
 513         "(%-18s . #x%04X)\t; %c")
 514        (t
 515         "(%-18s . #x%02X)\t; %c"))
 516       name
 517       (if (= (charset-iso-graphic-plane name) 1)
 518           (logior value
 519                   (cond ((= (charset-dimension name) 1)
 520                          #x80)
 521                         ((= (charset-dimension name) 2)
 522                          #x8080)
 523                         ((= (charset-dimension name) 3)
 524                          #x808080)
 525                         (t 0)))
 526         value)
 527       (char-db-decode-isolated-char name value)))
 528     (if (and (= (charset-chars name) 94)
 529              (= (charset-dimension name) 2))
 530         (insert (format " [%02d-%02d]"
 531                         (- (lsh value -8) 32)
 532                         (- (logand value 255) 32))))
 533     )
 534    (t
 535     (insert (format "(%-18s . %s)" name value))
 536     ))
 537   (insert line-breaking))
 538
 539 (defun char-db-insert-relation-feature (char name value line-breaking
 540                                              ccss readable)
 541   (insert (format "(%-18s%s " name line-breaking))
 542   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 543         separator cell sources required-features
 544         ret)
 545     (while (consp value)
 546       (setq cell (car value))
 547       (if (integerp cell)
 548           (setq cell (decode-char '=ucs cell)))
 549       (cond
 550        ((eq name '->subsumptive)
 551         (when (or (not (some (lambda (atr)
 552                                (get-char-attribute cell atr))
 553                              char-db-ignored-attributes))
 554                   (some (lambda (ccs)
 555                           (encode-char cell ccs 'defined-only))
 556                         ccss))
 557           (if separator
 558               (insert lbs))
 559           (let ((char-db-ignored-attributes
 560                  (cons '<-subsumptive
 561                        char-db-ignored-attributes)))
 562             (insert-char-attributes cell readable nil nil 'for-sub-node))
 563           (setq separator lbs))
 564         )
 565        ((characterp cell)
 566         (setq sources
 567               (get-char-attribute
 568                char (intern (format "%s*sources" name))))
 569         (setq required-features nil)
 570         (dolist (source sources)
 571           (cond
 572            ((memq source '(JP
 573                            JP/Jouyou
 574                            shinjigen shinjigen@1ed shinjigen@rev))
 575             (setq required-features
 576                   (union required-features
 577                          '(=jis-x0208
 578                            =jis-x0208@1990
 579                            =jis-x0213-1@2000
 580                            =jis-x0213-1@2004
 581                            =jis-x0213-2
 582                            =jis-x0212
 583                            =jis-x0208@1983
 584                            =jis-x0208@1978
 585                            =shinjigen))))
 586            ((eq source 'CN)
 587             (setq required-features
 588                   (union required-features
 589                          '(=gb2312
 590                            =gb12345
 591                            =iso-ir165)))))
 592           (cond
 593            ((find-charset (setq ret (intern (format "=%s" source))))
 594             (setq required-features
 595                   (cons ret required-features)))
 596            (t (setq required-features
 597                     (cons source required-features)))))
 598         (cond ((string-match "@JP" (symbol-name name))
 599                (setq required-features
 600                      (union required-features
 601                             '(=jis-x0208
 602                               =jis-x0208@1990
 603                               =jis-x0213-1-2000
 604                               =jis-x0213-2-2000
 605                               =jis-x0212
 606                               =jis-x0208@1983
 607                               =jis-x0208@1978))))
 608               ((string-match "@CN" (symbol-name name))
 609                (setq required-features
 610                      (union required-features
 611                             '(=gb2312
 612                               =gb12345
 613                               =iso-ir165)))))
 614         (if separator
 615             (insert lbs))
 616         (if readable
 617             (insert (format "%S" cell))
 618           (char-db-insert-char-spec cell readable
 619                                     nil
 620                                     required-features))
 621         (setq separator lbs))
 622        ((consp cell)
 623         (if separator
 624             (insert lbs))
 625         (if (consp (car cell))
 626             (char-db-insert-char-spec cell readable)
 627           (char-db-insert-char-reference cell readable))
 628         (setq separator lbs))
 629        (t
 630         (if separator
 631             (insert separator))
 632         (insert (prin1-to-string cell))
 633         (setq separator " ")))
 634       (setq value (cdr value)))
 635     (insert ")")
 636     (insert line-breaking)))
 637
 638 (defun insert-char-attributes (char &optional readable attributes column
 639                                     for-sub-node)
 640   (unless column
 641     (setq column (current-column)))
 642   (let (name value ; has-long-ccs-name
 643         rest
 644         radical strokes
 645         (line-breaking
 646          (concat "\n" (make-string (1+ column) ?\ )))
 647         lbs cell separator ret
 648         key al cal
 649         dest-ccss ; sources required-features
 650         ccss)
 651     (let (atr-d)
 652       (setq attributes
 653             (sort (if attributes
 654                       (if (consp attributes)
 655                           (progn
 656                             (dolist (name attributes)
 657                               (unless (memq name char-db-ignored-attributes)
 658                                 (if (find-charset name)
 659                                     (push name ccss))
 660                                 (push name atr-d)))
 661                             atr-d))
 662                     (dolist (name (char-attribute-list))
 663                       (unless (memq name char-db-ignored-attributes)
 664                         (if (find-charset name)
 665                             (push name ccss))
 666                         (push name atr-d)))
 667                     atr-d)
 668                   #'char-attribute-name<)))
 669     (insert "(")
 670     (when (memq '<-subsumptive attributes)
 671       (when (or readable (not for-sub-node))
 672         (when (setq value (get-char-attribute char '<-subsumptive))
 673           (char-db-insert-relation-feature char '<-subsumptive value
 674                                            line-breaking
 675                                            ccss readable)))
 676       (setq attributes (delq '<-subsumptive attributes)))
 677     (when (and (memq '<-denotational attributes)
 678                (setq value (get-char-attribute char '<-denotational)))
 679       (char-db-insert-relation-feature char '<-denotational value
 680                                        line-breaking
 681                                        ccss readable)
 682       (setq attributes (delq '<-denotational attributes)))
 683     (when (and (memq 'name attributes)
 684                (setq value (get-char-attribute char 'name)))
 685       (insert (format
 686                (if (> (+ (current-column) (length value)) 48)
 687                    "(name . %S)%s"
 688                  "(name               . %S)%s")
 689                value line-breaking))
 690       (setq attributes (delq 'name attributes))
 691       )
 692     (when (and (memq 'name* attributes)
 693                (setq value (get-char-attribute char 'name*)))
 694       (insert (format
 695                (if (> (+ (current-column) (length value)) 48)
 696                    "(name* . %S)%s"
 697                  "(name*              . %S)%s")
 698                value line-breaking))
 699       (setq attributes (delq 'name* attributes))
 700       )
 701     (when (and (memq 'script attributes)
 702                (setq value (get-char-attribute char 'script)))
 703       (insert (format "(script\t\t%s)%s"
 704                       (mapconcat (function prin1-to-string)
 705                                  value " ")
 706                       line-breaking))
 707       (setq attributes (delq 'script attributes))
 708       )
 709     (dolist (name '(=>ucs =>ucs*))
 710       (when (and (memq name attributes)
 711                  (setq value (get-char-attribute char name)))
 712         (insert (format "(%-18s . #x%04X)\t; %c%s"
 713                         name value (decode-char '=ucs value)
 714                         line-breaking))
 715         (setq attributes (delq name attributes))))
 716     (dolist (name '(=>ucs@gb =>ucs@big5))
 717       (when (and (memq name attributes)
 718                  (setq value (get-char-attribute char name)))
 719         (insert (format "(%-18s . #x%04X)\t; %c%s"
 720                         name value
 721                         (decode-char (intern
 722                                       (concat "="
 723                                               (substring
 724                                                (symbol-name name) 2)))
 725                                      value)
 726                         line-breaking))
 727         (setq attributes (delq name attributes))
 728         ))
 729     ;; (dolist (name '(=>daikanwa))
 730     ;;   (when (and (memq name attributes)
 731     ;;              (setq value (get-char-attribute char name)))
 732     ;;     (insert
 733     ;;      (if (integerp value)
 734     ;;          (format "(%-18s . %05d)\t; %c%s"
 735     ;;                  name value (decode-char '=daikanwa value)
 736     ;;                  line-breaking)
 737     ;;        (format "(%-18s %s)\t; %c%s"
 738     ;;                name
 739     ;;                (mapconcat (function prin1-to-string)
 740     ;;                           value " ")
 741     ;;                (char-representative-of-daikanwa char)
 742     ;;                line-breaking)))
 743     ;;     (setq attributes (delq name attributes))))
 744     (when (and (memq 'general-category attributes)
 745                (setq value (get-char-attribute char 'general-category)))
 746       (insert (format
 747                "(general-category\t%s) ; %s%s"
 748                (mapconcat (lambda (cell)
 749                             (format "%S" cell))
 750                           value " ")
 751                (cond ((rassoc value unidata-normative-category-alist)
 752                       "Normative Category")
 753                      ((rassoc value unidata-informative-category-alist)
 754                       "Informative Category")
 755                      (t
 756                       "Unknown Category"))
 757                line-breaking))
 758       (setq attributes (delq 'general-category attributes))
 759       )
 760     (when (and (memq 'bidi-category attributes)
 761                (setq value (get-char-attribute char 'bidi-category)))
 762       (insert (format "(bidi-category\t. %S)%s"
 763                       value
 764                       line-breaking))
 765       (setq attributes (delq 'bidi-category attributes))
 766       )
 767     (unless (or (not (memq 'mirrored attributes))
 768                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
 769                     'empty))
 770       (insert (format "(mirrored\t\t. %S)%s"
 771                       value
 772                       line-breaking))
 773       (setq attributes (delq 'mirrored attributes))
 774       )
 775     (cond
 776      ((and (memq 'decimal-digit-value attributes)
 777            (setq value (get-char-attribute char 'decimal-digit-value)))
 778       (insert (format "(decimal-digit-value . %S)%s"
 779                       value
 780                       line-breaking))
 781       (setq attributes (delq 'decimal-digit-value attributes))
 782       (when (and (memq 'digit-value attributes)
 783                  (setq value (get-char-attribute char 'digit-value)))
 784         (insert (format "(digit-value\t . %S)%s"
 785                         value
 786                         line-breaking))
 787         (setq attributes (delq 'digit-value attributes))
 788         )
 789       (when (and (memq 'numeric-value attributes)
 790                  (setq value (get-char-attribute char 'numeric-value)))
 791         (insert (format "(numeric-value\t . %S)%s"
 792                         value
 793                         line-breaking))
 794         (setq attributes (delq 'numeric-value attributes))
 795         )
 796       )
 797      (t
 798       (when (and (memq 'digit-value attributes)
 799                  (setq value (get-char-attribute char 'digit-value)))
 800         (insert (format "(digit-value\t. %S)%s"
 801                         value
 802                         line-breaking))
 803         (setq attributes (delq 'digit-value attributes))
 804         )
 805       (when (and (memq 'numeric-value attributes)
 806                  (setq value (get-char-attribute char 'numeric-value)))
 807         (insert (format "(numeric-value\t. %S)%s"
 808                         value
 809                         line-breaking))
 810         (setq attributes (delq 'numeric-value attributes))
 811         )))
 812     (when (and (memq 'iso-10646-comment attributes)
 813                (setq value (get-char-attribute char 'iso-10646-comment)))
 814       (insert (format "(iso-10646-comment\t. %S)%s"
 815                       value
 816                       line-breaking))
 817       (setq attributes (delq 'iso-10646-comment attributes))
 818       )
 819     (when (and (memq 'morohashi-daikanwa attributes)
 820                (setq value (get-char-attribute char 'morohashi-daikanwa)))
 821       (insert (format "(morohashi-daikanwa\t%s)%s"
 822                       (mapconcat (function prin1-to-string) value " ")
 823                       line-breaking))
 824       (setq attributes (delq 'morohashi-daikanwa attributes))
 825       )
 826     (setq radical nil
 827           strokes nil)
 828     (when (and (memq 'ideographic-radical attributes)
 829                (setq value (get-char-attribute char 'ideographic-radical)))
 830       (setq radical value)
 831       (insert (format "(ideographic-radical . %S)\t; %c%s"
 832                       radical
 833                       (ideographic-radical radical)
 834                       line-breaking))
 835       (setq attributes (delq 'ideographic-radical attributes))
 836       )
 837     (when (and (memq 'shuowen-radical attributes)
 838                (setq value (get-char-attribute char 'shuowen-radical)))
 839       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
 840                       value
 841                       (shuowen-radical value)
 842                       line-breaking))
 843       (setq attributes (delq 'shuowen-radical attributes))
 844       )
 845     (let (key)
 846       (dolist (domain
 847                (append
 848                 char-db-feature-domains
 849                 (let (dest domain)
 850                   (dolist (feature (char-attribute-list))
 851                     (setq feature (symbol-name feature))
 852                     (when (string-match
 853                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
 854                            feature)
 855                       (setq domain (intern (match-string 2 feature)))
 856                      (unless (memq domain dest)
 857                        (setq dest (cons domain dest)))))
 858                   (sort dest #'string<))))
 859         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
 860         (when (and (memq key attributes)
 861                    (setq value (get-char-attribute char key)))
 862           (setq radical value)
 863           (insert (format "(%s . %S)\t; %c%s"
 864                           key
 865                           radical
 866                           (ideographic-radical radical)
 867                           line-breaking))
 868           (setq attributes (delq key attributes))
 869           )
 870         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
 871         (when (and (memq key attributes)
 872                    (setq value (get-char-attribute char key)))
 873           (setq strokes value)
 874           (insert (format "(%s . %S)%s"
 875                           key
 876                           strokes
 877                           line-breaking))
 878           (setq attributes (delq key attributes))
 879           )
 880         (setq key (intern (format "%s@%s" 'total-strokes domain)))
 881         (when (and (memq key attributes)
 882                    (setq value (get-char-attribute char key)))
 883           (insert (format "(%s       . %S)%s"
 884                           key
 885                           value
 886                           line-breaking))
 887           (setq attributes (delq key attributes))
 888           )
 889         (dolist (feature '(ideographic-radical
 890                            ideographic-strokes
 891                            total-strokes))
 892           (setq key (intern (format "%s@%s*sources" feature domain)))
 893           (when (and (memq key attributes)
 894                      (setq value (get-char-attribute char key)))
 895             (insert (format "(%s%s" key line-breaking))
 896             (dolist (cell value)
 897               (insert (format " %s" cell)))
 898             (insert ")")
 899             (insert line-breaking)
 900             (setq attributes (delq key attributes))
 901             ))
 902         ))
 903     (when (and (memq 'ideographic-strokes attributes)
 904                (setq value (get-char-attribute char 'ideographic-strokes)))
 905       (setq strokes value)
 906       (insert (format "(ideographic-strokes . %S)%s"
 907                       strokes
 908                       line-breaking))
 909       (setq attributes (delq 'ideographic-strokes attributes))
 910       )
 911     (when (and (memq 'kangxi-radical attributes)
 912                (setq value (get-char-attribute char 'kangxi-radical)))
 913       (unless (eq value radical)
 914         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
 915                         value
 916                         (ideographic-radical value)
 917                         line-breaking))
 918         (or radical
 919             (setq radical value)))
 920       (setq attributes (delq 'kangxi-radical attributes))
 921       )
 922     (when (and (memq 'kangxi-strokes attributes)
 923                (setq value (get-char-attribute char 'kangxi-strokes)))
 924       (unless (eq value strokes)
 925         (insert (format "(kangxi-strokes\t . %S)%s"
 926                         value
 927                         line-breaking))
 928         (or strokes
 929             (setq strokes value)))
 930       (setq attributes (delq 'kangxi-strokes attributes))
 931       )
 932     (when (and (memq 'japanese-radical attributes)
 933                (setq value (get-char-attribute char 'japanese-radical)))
 934       (unless (eq value radical)
 935         (insert (format "(japanese-radical\t . %S)\t; %c%s"
 936                         value
 937                         (ideographic-radical value)
 938                         line-breaking))
 939         (or radical
 940             (setq radical value)))
 941       (setq attributes (delq 'japanese-radical attributes))
 942       )
 943     (when (and (memq 'japanese-strokes attributes)
 944                (setq value (get-char-attribute char 'japanese-strokes)))
 945       (unless (eq value strokes)
 946         (insert (format "(japanese-strokes\t . %S)%s"
 947                         value
 948                         line-breaking))
 949         (or strokes
 950             (setq strokes value)))
 951       (setq attributes (delq 'japanese-strokes attributes))
 952       )
 953     (when (and (memq 'cns-radical attributes)
 954                (setq value (get-char-attribute char 'cns-radical)))
 955       (insert (format "(cns-radical\t . %S)\t; %c%s"
 956                       value
 957                       (ideographic-radical value)
 958                       line-breaking))
 959       (setq attributes (delq 'cns-radical attributes))
 960       )
 961     (when (and (memq 'cns-strokes attributes)
 962                (setq value (get-char-attribute char 'cns-strokes)))
 963       (unless (eq value strokes)
 964         (insert (format "(cns-strokes\t . %S)%s"
 965                         value
 966                         line-breaking))
 967         (or strokes
 968             (setq strokes value)))
 969       (setq attributes (delq 'cns-strokes attributes))
 970       )
 971     (when (and (memq 'shinjigen-1-radical attributes)
 972                (setq value (get-char-attribute char 'shinjigen-1-radical)))
 973       (unless (eq value radical)
 974         (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
 975                         value
 976                         (ideographic-radical value)
 977                         line-breaking))
 978         (or radical
 979             (setq radical value)))
 980       (setq attributes (delq 'shinjigen-1-radical attributes))
 981       )
 982     (when (and (memq 'ideographic- attributes)
 983                (setq value (get-char-attribute char 'ideographic-)))
 984       (insert "(ideographic-       ")
 985       (setq lbs (concat "\n" (make-string (current-column) ?\ ))
 986             separator nil)
 987       (while (consp value)
 988         (setq cell (car value))
 989         (if (integerp cell)
 990             (setq cell (decode-char '=ucs cell)))
 991         (cond ((characterp cell)
 992                (if separator
 993                    (insert lbs))
 994                (if readable
 995                    (insert (format "%S" cell))
 996                  (char-db-insert-char-spec cell readable))
 997                (setq separator lbs))
 998               ((consp cell)
 999                (if separator
1000                    (insert lbs))
1001                (if (consp (car cell))
1002                    (char-db-insert-char-spec cell readable)
1003                  (char-db-insert-char-reference cell readable))
1004                (setq separator lbs))
1005               (t
1006                (if separator
1007                    (insert separator))
1008                (insert (prin1-to-string cell))
1009                (setq separator " ")))
1010         (setq value (cdr value)))
1011       (insert ")")
1012       (insert line-breaking)
1013       (setq attributes (delq 'ideographic- attributes)))
1014     (when (and (memq 'total-strokes attributes)
1015                (setq value (get-char-attribute char 'total-strokes)))
1016       (insert (format "(total-strokes       . %S)%s"
1017                       value
1018                       line-breaking))
1019       (setq attributes (delq 'total-strokes attributes))
1020       )
1021     (when (and (memq '->ideograph attributes)
1022                (setq value (get-char-attribute char '->ideograph)))
1023       (insert (format "(->ideograph\t%s)%s"
1024                       (mapconcat (lambda (code)
1025                                    (cond ((symbolp code)
1026                                           (symbol-name code))
1027                                          ((integerp code)
1028                                           (format "#x%04X" code))
1029                                          (t
1030                                           (format "%s %S"
1031                                                   line-breaking code))))
1032                                  value " ")
1033                       line-breaking))
1034       (setq attributes (delq '->ideograph attributes))
1035       )
1036     ;; (when (and (memq '->decomposition attributes)
1037     ;;            (setq value (get-char-attribute char '->decomposition)))
1038     ;;   (insert (format "(->decomposition\t%s)%s"
1039     ;;                   (mapconcat (lambda (code)
1040     ;;                                (cond ((symbolp code)
1041     ;;                                       (symbol-name code))
1042     ;;                                      ((characterp code)
1043     ;;                                       (if readable
1044     ;;                                           (format "%S" code)
1045     ;;                                         (format "#x%04X"
1046     ;;                                                 (char-int code))
1047     ;;                                         ))
1048     ;;                                      ((integerp code)
1049     ;;                                       (format "#x%04X" code))
1050     ;;                                      (t
1051     ;;                                       (format "%s%S" line-breaking code))))
1052     ;;                              value " ")
1053     ;;                   line-breaking))
1054     ;;   (setq attributes (delq '->decomposition attributes))
1055     ;;   )
1056     (if (equal (get-char-attribute char '->titlecase)
1057                (get-char-attribute char '->uppercase))
1058         (setq attributes (delq '->titlecase attributes)))
1059     (when (and (memq '->mojikyo attributes)
1060                (setq value (get-char-attribute char '->mojikyo)))
1061       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1062                       value (decode-char 'mojikyo value)
1063                       line-breaking))
1064       (setq attributes (delq '->mojikyo attributes))
1065       )
1066     (when (and (memq 'hanyu-dazidian-vol attributes)
1067                (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1068       (insert (format "(hanyu-dazidian-vol  . %d)%s"
1069                       value line-breaking))
1070       (setq attributes (delq 'hanyu-dazidian-vol attributes))
1071       )
1072     (when (and (memq 'hanyu-dazidian-page attributes)
1073                (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1074       (insert (format "(hanyu-dazidian-page . %d)%s"
1075                       value line-breaking))
1076       (setq attributes (delq 'hanyu-dazidian-page attributes))
1077       )
1078     (when (and (memq 'hanyu-dazidian-char attributes)
1079                (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1080       (insert (format "(hanyu-dazidian-char . %d)%s"
1081                       value line-breaking))
1082       (setq attributes (delq 'hanyu-dazidian-char attributes))
1083       )
1084     (unless readable
1085       (dolist (ignored '(composition
1086                          ->denotational <-subsumptive ->ucs-unified
1087                          ->ideographic-component-forms))
1088         (setq attributes (delq ignored attributes))))
1089     (while attributes
1090       (setq name (car attributes))
1091       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1092                   'value-is-empty)
1093         (cond ((setq ret (find-charset name))
1094                (setq name (charset-name ret))
1095                (when (not (memq name dest-ccss))
1096                  (setq dest-ccss (cons name dest-ccss))
1097                  (char-db-insert-ccs-feature name value line-breaking))
1098                )
1099               ((string-match "^=>ucs@" (symbol-name name))
1100                (insert (format "(%-18s . #x%04X)\t; %c%s"
1101                                name value (decode-char '=ucs value)
1102                                line-breaking))
1103                )
1104               ((eq name 'jisx0208-1978/4X)
1105                (insert (format "(%-18s . #x%04X)%s"
1106                                name value
1107                                line-breaking))
1108                )
1109               ((and
1110                 (not readable)
1111                 (not (eq name '->subsumptive))
1112                 (not (eq name '->uppercase))
1113                 (not (eq name '->lowercase))
1114                 (not (eq name '->titlecase))
1115                 (not (eq name '->canonical))
1116                 (not (eq name '->Bopomofo))
1117                 (not (eq name '->mistakable))
1118                 (not (eq name '->ideographic-variants))
1119                 (null (get-char-attribute
1120                        char (intern (format "%s*sources" name))))
1121                 (not (string-match "\\*sources$" (symbol-name name)))
1122                 (null (get-char-attribute
1123                        char (intern (format "%s*note" name))))
1124                 (not (string-match "\\*note$" (symbol-name name)))
1125                 (or (eq name '<-identical)
1126                     (eq name '<-uppercase)
1127                     (eq name '<-lowercase)
1128                     (eq name '<-titlecase)
1129                     (eq name '<-canonical)
1130                     (eq name '<-ideographic-variants)
1131                     ;; (eq name '<-synonyms)
1132                     (string-match "^<-synonyms" (symbol-name name))
1133                     (eq name '<-mistakable)
1134                     (when (string-match "^->" (symbol-name name))
1135                       (cond
1136                        ((string-match "^->fullwidth" (symbol-name name))
1137                         (not (and (consp value)
1138                                   (characterp (car value))
1139                                   (encode-char
1140                                    (car value) '=ucs 'defined-only)))
1141                         )
1142                        (t)))
1143                     ))
1144                )
1145               ((or (eq name 'ideographic-structure)
1146                    (eq name 'ideographic-combination)
1147                    (eq name 'ideographic-)
1148                    (eq name '=decomposition)
1149                    (char-feature-base-name= '=decomposition name)
1150                    (char-feature-base-name= '=>decomposition name)
1151                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1152                    ;;               (symbol-name name))
1153                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1154                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1155                                  (symbol-name name))
1156                    )
1157                (char-db-insert-relation-feature char name value
1158                                                 line-breaking
1159                                                 ccss readable))
1160               ((memq name '(ideograph=
1161                             original-ideograph-of
1162                             ancient-ideograph-of
1163                             vulgar-ideograph-of
1164                             wrong-ideograph-of
1165                             ;; simplified-ideograph-of
1166                             ideographic-variants
1167                             ;; ideographic-different-form-of
1168                             ))
1169                (insert (format "(%-18s%s " name line-breaking))
1170                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1171                      separator nil)
1172                (while (consp value)
1173                  (setq cell (car value))
1174                  (if (and (consp cell)
1175                           (consp (car cell)))
1176                      (progn
1177                        (if separator
1178                            (insert lbs))
1179                        (char-db-insert-alist cell readable)
1180                        (setq separator lbs))
1181                    (if separator
1182                        (insert separator))
1183                    (insert (prin1-to-string cell))
1184                    (setq separator " "))
1185                  (setq value (cdr value)))
1186                (insert ")")
1187                (insert line-breaking))
1188               ((consp value)
1189                (insert (format "(%-18s " name))
1190                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1191                      separator nil)
1192                (while (consp value)
1193                  (setq cell (car value))
1194                  (if (and (consp cell)
1195                           (consp (car cell))
1196                           (setq ret (condition-case nil
1197                                         (find-char cell)
1198                                       (error nil))))
1199                      (progn
1200                        (setq rest cell
1201                              al nil
1202                              cal nil)
1203                        (while rest
1204                          (setq key (car (car rest)))
1205                          (if (find-charset key)
1206                              (setq cal (cons key cal))
1207                            (setq al (cons key al)))
1208                          (setq rest (cdr rest)))
1209                        (if separator
1210                            (insert lbs))
1211                        (insert-char-attributes ret
1212                                                readable
1213                                                al ; cal
1214                                                nil 'for-sub-node)
1215                        (setq separator lbs))
1216                    (setq ret (prin1-to-string cell))
1217                    (if separator
1218                        (if (< (+ (current-column)
1219                                  (length ret)
1220                                  (length separator))
1221                               76)
1222                            (insert separator)
1223                          (insert lbs)))
1224                    (insert ret)
1225                    (setq separator " "))
1226                  (setq value (cdr value)))
1227                (insert ")")
1228                (insert line-breaking))
1229               (t
1230                (insert (format "(%-18s" name))
1231                (setq ret (prin1-to-string value))
1232                (unless (< (+ (current-column)
1233                              (length ret)
1234                              3)
1235                           76)
1236                  (insert line-breaking))
1237                (insert " . " ret ")" line-breaking)
1238                ;; (insert (format "(%-18s . %S)%s"
1239                ;;                 name value
1240                ;;                 line-breaking))
1241                )
1242               ))
1243       (setq attributes (cdr attributes)))
1244     (insert ")")))
1245
1246 (defun insert-char-data (char &optional readable
1247                               attributes)
1248   (save-restriction
1249     (narrow-to-region (point)(point))
1250     (insert "(define-char
1251   '")
1252     (insert-char-attributes char readable attributes)
1253     (insert ")\n")
1254     (goto-char (point-min))
1255     (while (re-search-forward "[ \t]+$" nil t)
1256       (replace-match ""))
1257     ;; from tabify.
1258     (goto-char (point-min))
1259     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1260       (let ((column (current-column))
1261             (indent-tabs-mode t))
1262         (delete-region (match-beginning 0) (point))
1263         (indent-to column)))
1264     (goto-char (point-max))
1265     ;; (tabify (point-min)(point-max))
1266     ))
1267
1268 (defun insert-char-data-with-variant (char &optional printable
1269                                            no-ucs-unified
1270                                            script excluded-script)
1271   (insert-char-data char printable)
1272   (let ((variants (char-variants char))
1273         rest
1274         variant vs ret)
1275     (setq variants (sort variants #'<))
1276     (setq rest variants)
1277     (setq variants (cons char variants))
1278     (while rest
1279       (setq variant (car rest))
1280       (unless (get-char-attribute variant '<-subsumptive)
1281         (if (and (or (null script)
1282                      (null (setq vs (get-char-attribute variant 'script)))
1283                      (memq script vs))
1284                  (or (null excluded-script)
1285                      (null (setq vs (get-char-attribute variant 'script)))
1286                      (not (memq excluded-script vs))))
1287             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1288               (insert-char-data variant printable)
1289               (if (setq ret (char-variants variant))
1290                   (while ret
1291                     (or (memq (car ret) variants)
1292                         ;; (get-char-attribute (car ret) '<-subsumptive)
1293                         (setq rest (nconc rest (list (car ret)))))
1294                     (setq ret (cdr ret)))))))
1295       (setq rest (cdr rest)))))
1296
1297 (defun insert-char-range-data (min max &optional script excluded-script)
1298   (let ((code min)
1299         char)
1300     (while (<= code max)
1301       (setq char (decode-char '=ucs code))
1302       (if (encode-char char '=ucs 'defined-only)
1303           (insert-char-data-with-variant char nil 'no-ucs-unified
1304                                          script excluded-script))
1305       (setq code (1+ code)))))
1306
1307 (defun write-char-range-data-to-file (min max file
1308                                           &optional script excluded-script)
1309   (let ((coding-system-for-write char-db-file-coding-system))
1310     (with-temp-buffer
1311       (insert (format ";; -*- coding: %s -*-\n"
1312                       char-db-file-coding-system))
1313       (insert-char-range-data min max script excluded-script)
1314       (write-region (point-min)(point-max) file))))
1315
1316 (defvar what-character-original-window-configuration)
1317
1318 ;;;###autoload
1319 (defun what-char-definition (char)
1320   (interactive (list (char-after)))
1321   (let ((buf (get-buffer-create "*Character Description*"))
1322         (the-buf (current-buffer))
1323         (win-conf (current-window-configuration)))
1324     (pop-to-buffer buf)
1325     (make-local-variable 'what-character-original-window-configuration)
1326     (setq what-character-original-window-configuration win-conf)
1327     (setq buffer-read-only nil)
1328     (erase-buffer)
1329     (condition-case err
1330         (progn
1331           (insert-char-data-with-variant char 'printable)
1332           (unless (char-attribute-alist char)
1333             (insert (format ";; = %c\n"
1334                             (let* ((rest (split-char char))
1335                                    (ccs (pop rest))
1336                                    (code (pop rest)))
1337                               (while rest
1338                                 (setq code (logior (lsh code 8)
1339                                                    (pop rest))))
1340                               (decode-char ccs code)))))
1341           ;; (char-db-update-comment)
1342           (set-buffer-modified-p nil)
1343           (view-mode the-buf (lambda (buf)
1344                                (set-window-configuration
1345                                 what-character-original-window-configuration)
1346                                ))
1347           (goto-char (point-min)))
1348       (error (progn
1349                (set-window-configuration
1350                 what-character-original-window-configuration)
1351                (signal (car err) (cdr err)))))))
1352
1353
1354 ;;; @ end
1355 ;;;
1356
1357 (provide 'char-db-util)
1358
1359 ;;; char-db-util.el ends here