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