git.chise.org Git - chise/xemacs-chise.git.1/blob - lisp/utf-2000/char-db-util.el

   1 ;;; char-db-util.el --- Character Database utility -*- coding: utf-8-er; -*-
   2
   3 ;; Copyright (C) 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006,
   4 ;;   2007, 2008, 2009, 2010, 2011, 2012 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     =adobe-japan1-0
 118     =adobe-japan1-1
 119     =adobe-japan1-2
 120     =adobe-japan1-3
 121     =adobe-japan1-4
 122     =adobe-japan1-5
 123     =adobe-japan1-6
 124     =jis-x0208
 125     =jis-x0208@1978
 126     =jis-x0208@1983
 127     japanese-jisx0212
 128     chinese-gb2312
 129     =jis-x0208@1990
 130     chinese-cns11643-1
 131     chinese-cns11643-2
 132     chinese-cns11643-3
 133     chinese-cns11643-4
 134     chinese-cns11643-5
 135     chinese-cns11643-6
 136     chinese-cns11643-7
 137     =jis-x0213-1
 138     =jis-x0213-1@2000
 139     =jis-x0213-1@2004
 140     =jis-x0213-2
 141     korean-ksc5601
 142     chinese-isoir165
 143     katakana-jisx0201
 144     hebrew-iso8859-8
 145     chinese-gb12345
 146     latin-viscii
 147     ethiopic-ucs
 148     =big5-cdp
 149     =hanyo-denshi/ja
 150     =hanyo-denshi/jb
 151     =hanyo-denshi/jc
 152     =hanyo-denshi/jd
 153     =hanyo-denshi/ft
 154     =hanyo-denshi/ia
 155     =hanyo-denshi/ib
 156     =hanyo-denshi/hg
 157     =hanyo-denshi/jt
 158     =hanyo-denshi/ks
 159     =gt
 160     =gt-k
 161     =daikanwa
 162     =daikanwa@rev2
 163     =daikanwa@rev1
 164     =cbeta
 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     =>>>adobe-japan1-0
 178     =>>>adobe-japan1-1
 179     =>>>adobe-japan1-2
 180     =>>>adobe-japan1-3
 181     =>>>adobe-japan1-4
 182     =>>>adobe-japan1-5
 183     =>>>adobe-japan1-6
 184     =>>>jis-x0208
 185     =>>>jis-x0213-1
 186     =>>>jis-x0213-2
 187     =>>>hanyo-denshi/ja
 188     =>>>hanyo-denshi/jb
 189     =>>>hanyo-denshi/jc
 190     =>>>hanyo-denshi/ft
 191     =>>>hanyo-denshi/ib
 192     =>>>hanyo-denshi/hg
 193     =>>>hanyo-denshi/jt
 194     =>>>hanyo-denshi/ks
 195     =>>>gt
 196     =>>adobe-japan1-0
 197     =>>adobe-japan1-1
 198     =>>adobe-japan1-2
 199     =>>adobe-japan1-3
 200     =>>adobe-japan1-4
 201     =>>adobe-japan1-5
 202     =>>adobe-japan1-6
 203     =>>jis-x0208
 204     =>>jis-x0213-1
 205     =>>jis-x0213-1@2000
 206     =>>jis-x0213-1@2004
 207     =>>jis-x0213-2
 208     =>>jis-x0208@1978
 209     =>>hanyo-denshi/ft
 210     =>>hanyo-denshi/jt
 211     =>>hanyo-denshi/ks
 212     =>>gt
 213     =>>daikanwa
 214     =+>jis-x0208
 215     =+>jis-x0213-1
 216     =+>jis-x0213-2
 217     =+>adobe-japan1-0
 218     =+>adobe-japan1-1
 219     =+>adobe-japan1-2
 220     =+>adobe-japan1-3
 221     =+>adobe-japan1-4
 222     =+>adobe-japan1-5
 223     =+>adobe-japan1-6
 224     =+>jis-x0208@1978
 225     =>jis-x0208
 226     =>jis-x0208@1997
 227     =>jis-x0213-1
 228     =>jis-x0213-1@2000
 229     =>jis-x0213-1@2004
 230     =>jis-x0213-2
 231     ==>ucs@bucs
 232     =>ucs@iso
 233     =>ucs@unicode
 234     =>ucs@jis
 235     =>ucs@cns
 236     =>ucs@ks
 237     =+>ucs@iso
 238     =+>ucs@unicode
 239     =+>ucs@jis
 240     =+>ucs@jis/1990
 241     =+>ucs@cns
 242     =+>ucs@ks
 243     =>>ucs@iso
 244     =>>ucs@unicode
 245     =>>ucs@jis
 246     =>>ucs@cns
 247     =>>>ucs@iso
 248     =>>>ucs@unicode
 249     =ucs@iso
 250     =ucs@unicode
 251     =>>big5-cdp
 252     =>>gt-k
 253     =+>gt
 254     =>gt
 255     =>big5-cdp
 256     =>daikanwa
 257     =>daikanwa/ho
 258     =>cns11643-7
 259     =big5
 260     =big5-eten
 261     =>gt-k
 262     =zinbun-oracle
 263     =>zinbun-oracle
 264     =ruimoku-v6
 265     =>>ruimoku-v6
 266     =jef-china3
 267     =shinjigen))
 268
 269
 270 ;;; @ char-db formatters
 271 ;;;
 272
 273 (defun char-db-make-char-spec (char)
 274   (let (ret char-spec)
 275     (cond ((characterp char)
 276            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 277                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 278                   (setq char-spec (list (cons '=ucs ret)))
 279                   (cond ((setq ret (get-char-attribute char 'name))
 280                          (setq char-spec (cons (cons 'name ret) char-spec))
 281                          )
 282                         ((setq ret (get-char-attribute char 'name*))
 283                          (setq char-spec (cons (cons 'name* ret) char-spec))
 284                          ))
 285                   )
 286                  ((encode-char char '=adobe-japan1 'defined-only)
 287                   (setq char-spec nil)
 288                   (dolist (ccs (charset-list))
 289                     (if (and (or (memq ccs
 290                                        '(=adobe-japan1-0
 291                                          =adobe-japan1-1
 292                                          =adobe-japan1-2
 293                                          =adobe-japan1-3
 294                                          =adobe-japan1-4
 295                                          =adobe-japan1-5
 296                                          =adobe-japan1-6
 297                                          ))
 298                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 299                                  (string-match "=ucs@" (symbol-name ccs))
 300                                  )
 301                              (setq ccs (charset-name ccs))
 302                              (null (assq ccs char-spec))
 303                              (setq ret (encode-char char ccs 'defined-only)))
 304                         (setq char-spec (cons (cons ccs ret) char-spec))))
 305                   )
 306                  ((setq ret
 307                         (catch 'tag
 308                           (let ((rest char-db-coded-charset-priority-list)
 309                                 ccs)
 310                             (while rest
 311                               (setq ccs (charset-name
 312                                          (find-charset (car rest))))
 313                               (if (setq ret
 314                                         (encode-char char ccs
 315                                                      'defined-only))
 316                                   (throw 'tag (cons ccs ret)))
 317                               (setq rest (cdr rest))))))
 318                   (setq char-spec (list ret))
 319                   (dolist (ccs (delq (car ret) (charset-list)))
 320                     (if (and (or (charset-iso-final-char ccs)
 321                                  (memq ccs
 322                                        '(=daikanwa
 323                                          =daikanwa@rev2
 324                                          ;; =gt-k
 325                                          =jis-x0208@1997
 326                                          ))
 327                                  (eq (charset-property ccs 'iso-ir) 177)
 328                                  ;; (string-match "=ucs@" (symbol-name ccs))
 329                                  )
 330                              (setq ccs (charset-name ccs))
 331                              (null (assq ccs char-spec))
 332                              (setq ret (encode-char char ccs 'defined-only)))
 333                         (setq char-spec (cons (cons ccs ret) char-spec))))
 334                   (if (null char-spec)
 335                       (setq char-spec (split-char char)))
 336                   (cond ((setq ret (get-char-attribute char 'name))
 337                          (setq char-spec (cons (cons 'name ret) char-spec))
 338                          )
 339                         ((setq ret (get-char-attribute char 'name*))
 340                          (setq char-spec (cons (cons 'name* ret) char-spec))
 341                          ))
 342                   )
 343                  ((setq ret (get-char-attribute
 344                              char 'ideographic-combination))
 345                   (setq char-spec
 346                         (cons (cons 'ideographic-combination ret)
 347                               char-spec))
 348                   ))
 349            char-spec)
 350           ((consp char)
 351            char))))
 352
 353 (defun char-db-insert-char-spec (char &optional readable column
 354                                       required-features)
 355   (unless column
 356     (setq column (current-column)))
 357   (let (char-spec temp-char)
 358     (setq char-spec (char-db-make-char-spec char))
 359     (unless (or (characterp char) ; char
 360                 (condition-case nil
 361                     (setq char (find-char char-spec))
 362                   (error nil)))
 363       ;; define temporary character
 364       ;;   Current implementation is dirty.
 365       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 366                                          char-spec)))
 367       (remove-char-attribute temp-char 'ideograph-daikanwa)
 368       (setq char temp-char))
 369     (insert-char-attributes char
 370                             readable
 371                             (union (mapcar #'car char-spec)
 372                                    required-features)
 373                             nil 'for-sub-node)
 374     (when temp-char
 375       ;; undefine temporary character
 376       ;;   Current implementation is dirty.
 377       (setq char-spec (char-attribute-alist temp-char))
 378       (while char-spec
 379         (remove-char-attribute temp-char (car (car char-spec)))
 380         (setq char-spec (cdr char-spec))))))
 381
 382 (defun char-db-insert-alist (alist &optional readable column)
 383   (unless column
 384     (setq column (current-column)))
 385   (let ((line-breaking
 386          (concat "\n" (make-string (1+ column) ?\ )))
 387         name value
 388         ret al ; cal
 389         key
 390         lbs cell rest separator)
 391     (insert "(")
 392     (while alist
 393       (setq name (car (car alist))
 394             value (cdr (car alist)))
 395       (cond ((eq name 'char)
 396              (insert "(char . ")
 397              (if (setq ret (condition-case nil
 398                                (find-char value)
 399                              (error nil)))
 400                  (progn
 401                    (setq al nil
 402                          ;; cal nil
 403                          )
 404                    (while value
 405                      (setq key (car (car value)))
 406                      ;; (if (find-charset key)
 407                      ;;     (setq cal (cons key cal))
 408                      (setq al (cons key al))
 409                      ;; )
 410                      (setq value (cdr value)))
 411                    (insert-char-attributes ret
 412                                            readable
 413                                            (or al 'none) ; cal
 414                                            nil 'for-sub-node))
 415                (insert (prin1-to-string value)))
 416              (insert ")")
 417              (insert line-breaking))
 418             ((consp value)
 419              (insert (format "(%-18s " name))
 420              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 421              (while (consp value)
 422                (setq cell (car value))
 423                (if (and (consp cell)
 424                         (consp (car cell))
 425                         (setq ret (condition-case nil
 426                                       (find-char cell)
 427                                     (error nil)))
 428                         )
 429                    (progn
 430                      (setq rest cell
 431                            al nil
 432                            ;; cal nil
 433                            )
 434                      (while rest
 435                        (setq key (car (car rest)))
 436                        ;; (if (find-charset key)
 437                        ;;     (setq cal (cons key cal))
 438                        (setq al (cons key al))
 439                        ;; )
 440                        (setq rest (cdr rest)))
 441                      (if separator
 442                          (insert lbs))
 443                      (insert-char-attributes ret
 444                                              readable
 445                                              al ; cal
 446                                              nil 'for-sub-node)
 447                      (setq separator lbs))
 448                  (if separator
 449                      (insert separator))
 450                  (insert (prin1-to-string cell))
 451                  (setq separator " "))
 452                (setq value (cdr value)))
 453              (insert ")")
 454              (insert line-breaking))
 455             (t
 456              (insert (format "(%-18s . %S)%s"
 457                              name value
 458                              line-breaking))))
 459       (setq alist (cdr alist))))
 460   (insert ")"))
 461
 462 (defun char-db-insert-char-reference (plist &optional readable column)
 463   (unless column
 464     (setq column (current-column)))
 465   (let ((line-breaking
 466          (concat "\n" (make-string (1+ column) ?\ )))
 467         (separator "")
 468         name value)
 469     (insert "(")
 470     (while plist
 471       (setq name (pop plist))
 472       (setq value (pop plist))
 473       (cond ((eq name :char)
 474              (insert separator)
 475              (insert ":char\t")
 476              (cond ((numberp value)
 477                     (setq value (decode-char '=ucs value)))
 478                    ;; ((consp value)
 479                    ;;  (setq value (or (find-char value)
 480                    ;;                  value)))
 481                    )
 482              (char-db-insert-char-spec value readable)
 483              (insert line-breaking)
 484              (setq separator ""))
 485             ((eq name :radical)
 486              (insert (format "%s%s\t%d ; %c%s"
 487                              separator
 488                              name value
 489                              (ideographic-radical value)
 490                              line-breaking))
 491              (setq separator ""))
 492             (t
 493              (insert (format "%s%s\t%S" separator name value))
 494              (setq separator line-breaking)))
 495       ))
 496   (insert ")"))
 497
 498 (defun char-db-decode-isolated-char (ccs code-point)
 499   (let (ret)
 500     (setq ret
 501           (cond ((eq ccs 'arabic-iso8859-6)
 502                  (decode-char ccs code-point))
 503                 ((and (memq ccs '(=gt-pj-1
 504                                   =gt-pj-2
 505                                   =gt-pj-3
 506                                   =gt-pj-4
 507                                   =gt-pj-5
 508                                   =gt-pj-6
 509                                   =gt-pj-7
 510                                   =gt-pj-8
 511                                   =gt-pj-9
 512                                   =gt-pj-10
 513                                   =gt-pj-11))
 514                       (setq ret (decode-char ccs code-point))
 515                       (setq ret (encode-char ret '=gt 'defined-only)))
 516                  (decode-builtin-char '=gt ret))
 517                 (t
 518                  (decode-builtin-char ccs code-point))))
 519     (cond ((and (<= 0 (char-int ret))
 520                 (<= (char-int ret) #x1F))
 521            (decode-char '=ucs (+ #x2400 (char-int ret))))
 522           ((= (char-int ret) #x7F)
 523            ?\u2421)
 524           (t ret))))
 525
 526 (defvar char-db-convert-obsolete-format t)
 527
 528 (defun char-db-insert-ccs-feature (name value line-breaking)
 529   (cond
 530    ((integerp value)
 531     (insert
 532      (format
 533       (cond
 534        ((memq name '(=shinjigen
 535                      =shinjigen@1ed
 536                      =shinjigen@rev =shinjigen/+p@rev
 537                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 538         "(%-18s .  %04d)\t; %c")
 539        ((eq name '=shinjigen@1ed/24pr)
 540         "(%-18s . %04d)\t; %c")
 541        ((or (memq name '(=daikanwa =>>daikanwa =>daikanwa
 542                          =daikanwa@rev1 =daikanwa@rev2
 543                          =daikanwa/+p =>>daikanwa/+p
 544                          =daikanwa/+2p =>>daikanwa/+2p
 545                          =gt =>>>gt =>>gt =+>gt =>gt
 546                          =gt-k =>>gt-k =>gt-k
 547                          =adobe-japan1-0 =>>>adobe-japan1-0
 548                          =adobe-japan1-1 =>>>adobe-japan1-1
 549                          =adobe-japan1-2 =>>>adobe-japan1-2
 550                          =adobe-japan1-3 =>>>adobe-japan1-3
 551                          =adobe-japan1-4 =>>>adobe-japan1-4
 552                          =adobe-japan1-5 =>>>adobe-japan1-5
 553                          =adobe-japan1-6 =>>>adobe-japan1-6
 554                          =>>adobe-japan1-0 =+>adobe-japan1-0
 555                          =>>adobe-japan1-1 =+>adobe-japan1-1
 556                          =>>adobe-japan1-2 =+>adobe-japan1-2
 557                          =>>adobe-japan1-3 =+>adobe-japan1-3
 558                          =>>adobe-japan1-4 =+>adobe-japan1-4
 559                          =>>adobe-japan1-5 =+>adobe-japan1-5
 560                          =>>adobe-japan1-6 =+>adobe-japan1-6
 561                          =cbeta =>>cbeta
 562                          =zinbun-oracle =>zinbun-oracle))
 563             ;; (string-match "^=adobe-" (symbol-name name))
 564             )
 565         "(%-18s . %05d)\t; %c")
 566        ((memq name '(=hanyo-denshi/ks =>>>hanyo-denshi/ks =>>hanyo-denshi/ks mojikyo))
 567         "(%-18s . %06d)\t; %c")
 568        ((>= (charset-dimension name) 2)
 569         "(%-18s . #x%04X)\t; %c")
 570        (t
 571         "(%-18s . #x%02X)\t; %c"))
 572       name
 573       (if (= (charset-iso-graphic-plane name) 1)
 574           (logior value
 575                   (cond ((= (charset-dimension name) 1)
 576                          #x80)
 577                         ((= (charset-dimension name) 2)
 578                          #x8080)
 579                         ((= (charset-dimension name) 3)
 580                          #x808080)
 581                         (t 0)))
 582         value)
 583       (char-db-decode-isolated-char name value)))
 584     (if (and (= (charset-chars name) 94)
 585              (= (charset-dimension name) 2))
 586         (insert (format " [%02d-%02d]"
 587                         (- (lsh value -8) 32)
 588                         (- (logand value 255) 32))))
 589     )
 590    (t
 591     (insert (format "(%-18s . %s)" name value))
 592     ))
 593   (insert line-breaking))
 594
 595 (defun char-db-insert-relation-feature (char name value line-breaking
 596                                              ccss readable)
 597   (insert (format "(%-18s%s " name line-breaking))
 598   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 599         separator cell sources required-features
 600         ret)
 601     (while (consp value)
 602       (setq cell (car value))
 603       (if (integerp cell)
 604           (setq cell (decode-char '=ucs cell)))
 605       (cond
 606        ((eq name '->subsumptive)
 607         (when (or (not (some (lambda (atr)
 608                                (get-char-attribute cell atr))
 609                              char-db-ignored-attributes))
 610                   (some (lambda (ccs)
 611                           (encode-char cell ccs 'defined-only))
 612                         ccss))
 613           (if separator
 614               (insert lbs))
 615           (let ((char-db-ignored-attributes
 616                  (cons '<-subsumptive
 617                        char-db-ignored-attributes)))
 618             (insert-char-attributes cell readable nil nil 'for-sub-node))
 619           (setq separator lbs))
 620         )
 621        ((characterp cell)
 622         (setq sources
 623               (get-char-attribute
 624                char (intern (format "%s*sources" name))))
 625         (setq required-features nil)
 626         (dolist (source sources)
 627           (cond
 628            ((memq source '(JP
 629                            JP/Jouyou
 630                            shinjigen shinjigen@1ed shinjigen@rev))
 631             (setq required-features
 632                   (union required-features
 633                          '(=jis-x0208
 634                            =jis-x0208@1990
 635                            =jis-x0213-1@2000
 636                            =jis-x0213-1@2004
 637                            =jis-x0213-2
 638                            =jis-x0212
 639                            =jis-x0208@1983
 640                            =jis-x0208@1978
 641                            =shinjigen
 642                            =shinjigen@1ed
 643                            =shinjigen@rev
 644                            =shinjigen/+p@rev))))
 645            ((eq source 'CN)
 646             (setq required-features
 647                   (union required-features
 648                          '(=gb2312
 649                            =gb12345
 650                            =iso-ir165)))))
 651           (cond
 652            ((find-charset (setq ret (intern (format "=%s" source))))
 653             (setq required-features
 654                   (cons ret required-features)))
 655            (t (setq required-features
 656                     (cons source required-features)))))
 657         (cond ((string-match "@JP" (symbol-name name))
 658                (setq required-features
 659                      (union required-features
 660                             '(=jis-x0208
 661                               =jis-x0208@1990
 662                               =jis-x0213-1-2000
 663                               =jis-x0213-2-2000
 664                               =jis-x0212
 665                               =jis-x0208@1983
 666                               =jis-x0208@1978))))
 667               ((string-match "@CN" (symbol-name name))
 668                (setq required-features
 669                      (union required-features
 670                             '(=gb2312
 671                               =gb12345
 672                               =iso-ir165)))))
 673         (if separator
 674             (insert lbs))
 675         (if readable
 676             (insert (format "%S" cell))
 677           (char-db-insert-char-spec cell readable
 678                                     nil
 679                                     required-features))
 680         (setq separator lbs))
 681        ((consp cell)
 682         (if separator
 683             (insert lbs))
 684         (if (consp (car cell))
 685             (char-db-insert-char-spec cell readable)
 686           (char-db-insert-char-reference cell readable))
 687         (setq separator lbs))
 688        (t
 689         (if separator
 690             (insert separator))
 691         (insert (prin1-to-string cell))
 692         (setq separator " ")))
 693       (setq value (cdr value)))
 694     (insert ")")
 695     (insert line-breaking)))
 696
 697 (defun insert-char-attributes (char &optional readable attributes column
 698                                     for-sub-node)
 699   (unless column
 700     (setq column (current-column)))
 701   (let (name value ; has-long-ccs-name
 702         rest
 703         radical strokes
 704         (line-breaking
 705          (concat "\n" (make-string (1+ column) ?\ )))
 706         lbs cell separator ret
 707         key al cal
 708         dest-ccss ; sources required-features
 709         ccss)
 710     (let (atr-d)
 711       (setq attributes
 712             (sort (if attributes
 713                       (if (consp attributes)
 714                           (progn
 715                             (dolist (name attributes)
 716                               (unless (memq name char-db-ignored-attributes)
 717                                 (if (find-charset name)
 718                                     (push name ccss))
 719                                 (push name atr-d)))
 720                             atr-d))
 721                     (dolist (name (char-attribute-list))
 722                       (unless (memq name char-db-ignored-attributes)
 723                         (if (find-charset name)
 724                             (push name ccss))
 725                         (push name atr-d)))
 726                     atr-d)
 727                   #'char-attribute-name<)))
 728     (insert "(")
 729     (when (memq '<-subsumptive attributes)
 730       (when (or readable (not for-sub-node))
 731         (when (setq value (get-char-attribute char '<-subsumptive))
 732           (char-db-insert-relation-feature char '<-subsumptive value
 733                                            line-breaking
 734                                            ccss readable)))
 735       (setq attributes (delq '<-subsumptive attributes)))
 736     (when (and (memq '<-denotational attributes)
 737                (setq value (get-char-attribute char '<-denotational)))
 738       (char-db-insert-relation-feature char '<-denotational value
 739                                        line-breaking
 740                                        ccss readable)
 741       (setq attributes (delq '<-denotational attributes)))
 742     (when (and (memq 'name attributes)
 743                (setq value (get-char-attribute char 'name)))
 744       (insert (format
 745                (if (> (+ (current-column) (length value)) 48)
 746                    "(name . %S)%s"
 747                  "(name               . %S)%s")
 748                value line-breaking))
 749       (setq attributes (delq 'name attributes))
 750       )
 751     (when (and (memq 'name* attributes)
 752                (setq value (get-char-attribute char 'name*)))
 753       (insert (format
 754                (if (> (+ (current-column) (length value)) 48)
 755                    "(name* . %S)%s"
 756                  "(name*              . %S)%s")
 757                value line-breaking))
 758       (setq attributes (delq 'name* attributes))
 759       )
 760     (when (and (memq 'script attributes)
 761                (setq value (get-char-attribute char 'script)))
 762       (insert (format "(script\t\t%s)%s"
 763                       (mapconcat (function prin1-to-string)
 764                                  value " ")
 765                       line-breaking))
 766       (setq attributes (delq 'script attributes))
 767       )
 768     (dolist (name '(=>ucs =>ucs*))
 769       (when (and (memq name attributes)
 770                  (setq value (get-char-attribute char name)))
 771         (insert (format "(%-18s . #x%04X)\t; %c%s"
 772                         name value (decode-char '=ucs value)
 773                         line-breaking))
 774         (setq attributes (delq name attributes))))
 775     (dolist (name '(=>ucs@gb =>ucs@big5))
 776       (when (and (memq name attributes)
 777                  (setq value (get-char-attribute char name)))
 778         (insert (format "(%-18s . #x%04X)\t; %c%s"
 779                         name value
 780                         (decode-char (intern
 781                                       (concat "="
 782                                               (substring
 783                                                (symbol-name name) 2)))
 784                                      value)
 785                         line-breaking))
 786         (setq attributes (delq name attributes))
 787         ))
 788     ;; (dolist (name '(=>daikanwa))
 789     ;;   (when (and (memq name attributes)
 790     ;;              (setq value (get-char-attribute char name)))
 791     ;;     (insert
 792     ;;      (if (integerp value)
 793     ;;          (format "(%-18s . %05d)\t; %c%s"
 794     ;;                  name value (decode-char '=daikanwa value)
 795     ;;                  line-breaking)
 796     ;;        (format "(%-18s %s)\t; %c%s"
 797     ;;                name
 798     ;;                (mapconcat (function prin1-to-string)
 799     ;;                           value " ")
 800     ;;                (char-representative-of-daikanwa char)
 801     ;;                line-breaking)))
 802     ;;     (setq attributes (delq name attributes))))
 803     (when (and (memq 'general-category attributes)
 804                (setq value (get-char-attribute char 'general-category)))
 805       (insert (format
 806                "(general-category\t%s) ; %s%s"
 807                (mapconcat (lambda (cell)
 808                             (format "%S" cell))
 809                           value " ")
 810                (cond ((rassoc value unidata-normative-category-alist)
 811                       "Normative Category")
 812                      ((rassoc value unidata-informative-category-alist)
 813                       "Informative Category")
 814                      (t
 815                       "Unknown Category"))
 816                line-breaking))
 817       (setq attributes (delq 'general-category attributes))
 818       )
 819     (when (and (memq 'bidi-category attributes)
 820                (setq value (get-char-attribute char 'bidi-category)))
 821       (insert (format "(bidi-category\t. %S)%s"
 822                       value
 823                       line-breaking))
 824       (setq attributes (delq 'bidi-category attributes))
 825       )
 826     (unless (or (not (memq 'mirrored attributes))
 827                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
 828                     'empty))
 829       (insert (format "(mirrored\t\t. %S)%s"
 830                       value
 831                       line-breaking))
 832       (setq attributes (delq 'mirrored attributes))
 833       )
 834     (cond
 835      ((and (memq 'decimal-digit-value attributes)
 836            (setq value (get-char-attribute char 'decimal-digit-value)))
 837       (insert (format "(decimal-digit-value . %S)%s"
 838                       value
 839                       line-breaking))
 840       (setq attributes (delq 'decimal-digit-value attributes))
 841       (when (and (memq 'digit-value attributes)
 842                  (setq value (get-char-attribute char 'digit-value)))
 843         (insert (format "(digit-value\t . %S)%s"
 844                         value
 845                         line-breaking))
 846         (setq attributes (delq 'digit-value attributes))
 847         )
 848       (when (and (memq 'numeric-value attributes)
 849                  (setq value (get-char-attribute char 'numeric-value)))
 850         (insert (format "(numeric-value\t . %S)%s"
 851                         value
 852                         line-breaking))
 853         (setq attributes (delq 'numeric-value attributes))
 854         )
 855       )
 856      (t
 857       (when (and (memq 'digit-value attributes)
 858                  (setq value (get-char-attribute char 'digit-value)))
 859         (insert (format "(digit-value\t. %S)%s"
 860                         value
 861                         line-breaking))
 862         (setq attributes (delq 'digit-value attributes))
 863         )
 864       (when (and (memq 'numeric-value attributes)
 865                  (setq value (get-char-attribute char 'numeric-value)))
 866         (insert (format "(numeric-value\t. %S)%s"
 867                         value
 868                         line-breaking))
 869         (setq attributes (delq 'numeric-value attributes))
 870         )))
 871     (when (and (memq 'iso-10646-comment attributes)
 872                (setq value (get-char-attribute char 'iso-10646-comment)))
 873       (insert (format "(iso-10646-comment\t. %S)%s"
 874                       value
 875                       line-breaking))
 876       (setq attributes (delq 'iso-10646-comment attributes))
 877       )
 878     (when (and (memq 'morohashi-daikanwa attributes)
 879                (setq value (get-char-attribute char 'morohashi-daikanwa)))
 880       (insert (format "(morohashi-daikanwa\t%s)%s"
 881                       (mapconcat (function prin1-to-string) value " ")
 882                       line-breaking))
 883       (setq attributes (delq 'morohashi-daikanwa attributes))
 884       )
 885     (setq radical nil
 886           strokes nil)
 887     (when (and (memq 'ideographic-radical attributes)
 888                (setq value (get-char-attribute char 'ideographic-radical)))
 889       (setq radical value)
 890       (insert (format "(ideographic-radical . %S)\t; %c%s"
 891                       radical
 892                       (ideographic-radical radical)
 893                       line-breaking))
 894       (setq attributes (delq 'ideographic-radical attributes))
 895       )
 896     (when (and (memq 'shuowen-radical attributes)
 897                (setq value (get-char-attribute char 'shuowen-radical)))
 898       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
 899                       value
 900                       (shuowen-radical value)
 901                       line-breaking))
 902       (setq attributes (delq 'shuowen-radical attributes))
 903       )
 904     (let (key)
 905       (dolist (domain
 906                (append
 907                 char-db-feature-domains
 908                 (let (dest domain)
 909                   (dolist (feature (char-attribute-list))
 910                     (setq feature (symbol-name feature))
 911                     (when (string-match
 912                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
 913                            feature)
 914                       (setq domain (intern (match-string 2 feature)))
 915                      (unless (memq domain dest)
 916                        (setq dest (cons domain dest)))))
 917                   (sort dest #'string<))))
 918         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
 919         (when (and (memq key attributes)
 920                    (setq value (get-char-attribute char key)))
 921           (setq radical value)
 922           (insert (format "(%s . %S)\t; %c%s"
 923                           key
 924                           radical
 925                           (ideographic-radical radical)
 926                           line-breaking))
 927           (setq attributes (delq key attributes))
 928           )
 929         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
 930         (when (and (memq key attributes)
 931                    (setq value (get-char-attribute char key)))
 932           (setq strokes value)
 933           (insert (format "(%s . %S)%s"
 934                           key
 935                           strokes
 936                           line-breaking))
 937           (setq attributes (delq key attributes))
 938           )
 939         (setq key (intern (format "%s@%s" 'total-strokes domain)))
 940         (when (and (memq key attributes)
 941                    (setq value (get-char-attribute char key)))
 942           (insert (format "(%s       . %S)%s"
 943                           key
 944                           value
 945                           line-breaking))
 946           (setq attributes (delq key attributes))
 947           )
 948         (dolist (feature '(ideographic-radical
 949                            ideographic-strokes
 950                            total-strokes))
 951           (setq key (intern (format "%s@%s*sources" feature domain)))
 952           (when (and (memq key attributes)
 953                      (setq value (get-char-attribute char key)))
 954             (insert (format "(%s%s" key line-breaking))
 955             (dolist (cell value)
 956               (insert (format " %s" cell)))
 957             (insert ")")
 958             (insert line-breaking)
 959             (setq attributes (delq key attributes))
 960             ))
 961         ))
 962     (when (and (memq 'ideographic-strokes attributes)
 963                (setq value (get-char-attribute char 'ideographic-strokes)))
 964       (setq strokes value)
 965       (insert (format "(ideographic-strokes . %S)%s"
 966                       strokes
 967                       line-breaking))
 968       (setq attributes (delq 'ideographic-strokes attributes))
 969       )
 970     (when (and (memq 'kangxi-radical attributes)
 971                (setq value (get-char-attribute char 'kangxi-radical)))
 972       (unless (eq value radical)
 973         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
 974                         value
 975                         (ideographic-radical value)
 976                         line-breaking))
 977         (or radical
 978             (setq radical value)))
 979       (setq attributes (delq 'kangxi-radical attributes))
 980       )
 981     (when (and (memq 'kangxi-strokes attributes)
 982                (setq value (get-char-attribute char 'kangxi-strokes)))
 983       (unless (eq value strokes)
 984         (insert (format "(kangxi-strokes\t . %S)%s"
 985                         value
 986                         line-breaking))
 987         (or strokes
 988             (setq strokes value)))
 989       (setq attributes (delq 'kangxi-strokes attributes))
 990       )
 991     (when (and (memq 'japanese-radical attributes)
 992                (setq value (get-char-attribute char 'japanese-radical)))
 993       (unless (eq value radical)
 994         (insert (format "(japanese-radical\t . %S)\t; %c%s"
 995                         value
 996                         (ideographic-radical value)
 997                         line-breaking))
 998         (or radical
 999             (setq radical value)))
1000       (setq attributes (delq 'japanese-radical attributes))
1001       )
1002     (when (and (memq 'japanese-strokes attributes)
1003                (setq value (get-char-attribute char 'japanese-strokes)))
1004       (unless (eq value strokes)
1005         (insert (format "(japanese-strokes\t . %S)%s"
1006                         value
1007                         line-breaking))
1008         (or strokes
1009             (setq strokes value)))
1010       (setq attributes (delq 'japanese-strokes attributes))
1011       )
1012     (when (and (memq 'cns-radical attributes)
1013                (setq value (get-char-attribute char 'cns-radical)))
1014       (insert (format "(cns-radical\t . %S)\t; %c%s"
1015                       value
1016                       (ideographic-radical value)
1017                       line-breaking))
1018       (setq attributes (delq 'cns-radical attributes))
1019       )
1020     (when (and (memq 'cns-strokes attributes)
1021                (setq value (get-char-attribute char 'cns-strokes)))
1022       (unless (eq value strokes)
1023         (insert (format "(cns-strokes\t . %S)%s"
1024                         value
1025                         line-breaking))
1026         (or strokes
1027             (setq strokes value)))
1028       (setq attributes (delq 'cns-strokes attributes))
1029       )
1030     ;; (when (and (memq 'shinjigen-1-radical attributes)
1031     ;;            (setq value (get-char-attribute char 'shinjigen-1-radical)))
1032     ;;   (unless (eq value radical)
1033     ;;     (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
1034     ;;                     value
1035     ;;                     (ideographic-radical value)
1036     ;;                     line-breaking))
1037     ;;     (or radical
1038     ;;         (setq radical value)))
1039     ;;   (setq attributes (delq 'shinjigen-1-radical attributes))
1040     ;;   )
1041     (when (and (memq 'ideographic- attributes)
1042                (setq value (get-char-attribute char 'ideographic-)))
1043       (insert "(ideographic-       ")
1044       (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1045             separator nil)
1046       (while (consp value)
1047         (setq cell (car value))
1048         (if (integerp cell)
1049             (setq cell (decode-char '=ucs cell)))
1050         (cond ((characterp cell)
1051                (if separator
1052                    (insert lbs))
1053                (if readable
1054                    (insert (format "%S" cell))
1055                  (char-db-insert-char-spec cell readable))
1056                (setq separator lbs))
1057               ((consp cell)
1058                (if separator
1059                    (insert lbs))
1060                (if (consp (car cell))
1061                    (char-db-insert-char-spec cell readable)
1062                  (char-db-insert-char-reference cell readable))
1063                (setq separator lbs))
1064               (t
1065                (if separator
1066                    (insert separator))
1067                (insert (prin1-to-string cell))
1068                (setq separator " ")))
1069         (setq value (cdr value)))
1070       (insert ")")
1071       (insert line-breaking)
1072       (setq attributes (delq 'ideographic- attributes)))
1073     (when (and (memq 'total-strokes attributes)
1074                (setq value (get-char-attribute char 'total-strokes)))
1075       (insert (format "(total-strokes       . %S)%s"
1076                       value
1077                       line-breaking))
1078       (setq attributes (delq 'total-strokes attributes))
1079       )
1080     (when (and (memq '->ideograph attributes)
1081                (setq value (get-char-attribute char '->ideograph)))
1082       (insert (format "(->ideograph\t%s)%s"
1083                       (mapconcat (lambda (code)
1084                                    (cond ((symbolp code)
1085                                           (symbol-name code))
1086                                          ((integerp code)
1087                                           (format "#x%04X" code))
1088                                          (t
1089                                           (format "%s %S"
1090                                                   line-breaking code))))
1091                                  value " ")
1092                       line-breaking))
1093       (setq attributes (delq '->ideograph attributes))
1094       )
1095     ;; (when (and (memq '->decomposition attributes)
1096     ;;            (setq value (get-char-attribute char '->decomposition)))
1097     ;;   (insert (format "(->decomposition\t%s)%s"
1098     ;;                   (mapconcat (lambda (code)
1099     ;;                                (cond ((symbolp code)
1100     ;;                                       (symbol-name code))
1101     ;;                                      ((characterp code)
1102     ;;                                       (if readable
1103     ;;                                           (format "%S" code)
1104     ;;                                         (format "#x%04X"
1105     ;;                                                 (char-int code))
1106     ;;                                         ))
1107     ;;                                      ((integerp code)
1108     ;;                                       (format "#x%04X" code))
1109     ;;                                      (t
1110     ;;                                       (format "%s%S" line-breaking code))))
1111     ;;                              value " ")
1112     ;;                   line-breaking))
1113     ;;   (setq attributes (delq '->decomposition attributes))
1114     ;;   )
1115     (if (equal (get-char-attribute char '->titlecase)
1116                (get-char-attribute char '->uppercase))
1117         (setq attributes (delq '->titlecase attributes)))
1118     (when (and (memq '->mojikyo attributes)
1119                (setq value (get-char-attribute char '->mojikyo)))
1120       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1121                       value (decode-char 'mojikyo value)
1122                       line-breaking))
1123       (setq attributes (delq '->mojikyo attributes))
1124       )
1125     (when (and (memq 'hanyu-dazidian-vol attributes)
1126                (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1127       (insert (format "(hanyu-dazidian-vol  . %d)%s"
1128                       value line-breaking))
1129       (setq attributes (delq 'hanyu-dazidian-vol attributes))
1130       )
1131     (when (and (memq 'hanyu-dazidian-page attributes)
1132                (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1133       (insert (format "(hanyu-dazidian-page . %d)%s"
1134                       value line-breaking))
1135       (setq attributes (delq 'hanyu-dazidian-page attributes))
1136       )
1137     (when (and (memq 'hanyu-dazidian-char attributes)
1138                (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1139       (insert (format "(hanyu-dazidian-char . %d)%s"
1140                       value line-breaking))
1141       (setq attributes (delq 'hanyu-dazidian-char attributes))
1142       )
1143     (unless readable
1144       (dolist (ignored '(composition
1145                          ->denotational <-subsumptive ->ucs-unified
1146                          ->ideographic-component-forms))
1147         (setq attributes (delq ignored attributes))))
1148     (while attributes
1149       (setq name (car attributes))
1150       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1151                   'value-is-empty)
1152         (cond ((setq ret (find-charset name))
1153                (setq name (charset-name ret))
1154                (when (not (memq name dest-ccss))
1155                  (setq dest-ccss (cons name dest-ccss))
1156                  (char-db-insert-ccs-feature name value line-breaking))
1157                )
1158               ((string-match "^=>ucs@" (symbol-name name))
1159                (insert (format "(%-18s . #x%04X)\t; %c%s"
1160                                name value (decode-char '=ucs value)
1161                                line-breaking))
1162                )
1163               ((eq name 'jisx0208-1978/4X)
1164                (insert (format "(%-18s . #x%04X)%s"
1165                                name value
1166                                line-breaking))
1167                )
1168               ((and
1169                 (not readable)
1170                 (not (eq name '->subsumptive))
1171                 (not (eq name '->uppercase))
1172                 (not (eq name '->lowercase))
1173                 (not (eq name '->titlecase))
1174                 (not (eq name '->canonical))
1175                 (not (eq name '->Bopomofo))
1176                 (not (eq name '->mistakable))
1177                 (not (eq name '->ideographic-variants))
1178                 (null (get-char-attribute
1179                        char (intern (format "%s*sources" name))))
1180                 (not (string-match "\\*sources$" (symbol-name name)))
1181                 (null (get-char-attribute
1182                        char (intern (format "%s*note" name))))
1183                 (not (string-match "\\*note$" (symbol-name name)))
1184                 (or (eq name '<-identical)
1185                     (eq name '<-uppercase)
1186                     (eq name '<-lowercase)
1187                     (eq name '<-titlecase)
1188                     (eq name '<-canonical)
1189                     (eq name '<-ideographic-variants)
1190                     ;; (eq name '<-synonyms)
1191                     (string-match "^<-synonyms" (symbol-name name))
1192                     (eq name '<-mistakable)
1193                     (when (string-match "^->" (symbol-name name))
1194                       (cond
1195                        ((string-match "^->fullwidth" (symbol-name name))
1196                         (not (and (consp value)
1197                                   (characterp (car value))
1198                                   (encode-char
1199                                    (car value) '=ucs 'defined-only)))
1200                         )
1201                        (t)))
1202                     ))
1203                )
1204               ((or (eq name 'ideographic-structure)
1205                    (eq name 'ideographic-combination)
1206                    (eq name 'ideographic-)
1207                    (eq name '=decomposition)
1208                    (char-feature-base-name= '=decomposition name)
1209                    (char-feature-base-name= '=>decomposition name)
1210                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1211                    ;;               (symbol-name name))
1212                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1213                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1214                                  (symbol-name name))
1215                    )
1216                (char-db-insert-relation-feature char name value
1217                                                 line-breaking
1218                                                 ccss readable))
1219               ((memq name '(ideograph=
1220                             original-ideograph-of
1221                             ancient-ideograph-of
1222                             vulgar-ideograph-of
1223                             wrong-ideograph-of
1224                             ;; simplified-ideograph-of
1225                             ideographic-variants
1226                             ;; ideographic-different-form-of
1227                             ))
1228                (insert (format "(%-18s%s " name line-breaking))
1229                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1230                      separator nil)
1231                (while (consp value)
1232                  (setq cell (car value))
1233                  (if (and (consp cell)
1234                           (consp (car cell)))
1235                      (progn
1236                        (if separator
1237                            (insert lbs))
1238                        (char-db-insert-alist cell readable)
1239                        (setq separator lbs))
1240                    (if separator
1241                        (insert separator))
1242                    (insert (prin1-to-string cell))
1243                    (setq separator " "))
1244                  (setq value (cdr value)))
1245                (insert ")")
1246                (insert line-breaking))
1247               ((consp value)
1248                (insert (format "(%-18s " name))
1249                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1250                      separator nil)
1251                (while (consp value)
1252                  (setq cell (car value))
1253                  (if (and (consp cell)
1254                           (consp (car cell))
1255                           (setq ret (condition-case nil
1256                                         (find-char cell)
1257                                       (error nil))))
1258                      (progn
1259                        (setq rest cell
1260                              al nil
1261                              cal nil)
1262                        (while rest
1263                          (setq key (car (car rest)))
1264                          (if (find-charset key)
1265                              (setq cal (cons key cal))
1266                            (setq al (cons key al)))
1267                          (setq rest (cdr rest)))
1268                        (if separator
1269                            (insert lbs))
1270                        (insert-char-attributes ret
1271                                                readable
1272                                                al ; cal
1273                                                nil 'for-sub-node)
1274                        (setq separator lbs))
1275                    (setq ret (prin1-to-string cell))
1276                    (if separator
1277                        (if (< (+ (current-column)
1278                                  (length ret)
1279                                  (length separator))
1280                               76)
1281                            (insert separator)
1282                          (insert lbs)))
1283                    (insert ret)
1284                    (setq separator " "))
1285                  (setq value (cdr value)))
1286                (insert ")")
1287                (insert line-breaking))
1288               (t
1289                (insert (format "(%-18s" name))
1290                (setq ret (prin1-to-string value))
1291                (unless (< (+ (current-column)
1292                              (length ret)
1293                              3)
1294                           76)
1295                  (insert line-breaking))
1296                (insert " . " ret ")" line-breaking)
1297                ;; (insert (format "(%-18s . %S)%s"
1298                ;;                 name value
1299                ;;                 line-breaking))
1300                )
1301               ))
1302       (setq attributes (cdr attributes)))
1303     (insert ")")))
1304
1305 (defun insert-char-data (char &optional readable
1306                               attributes)
1307   (save-restriction
1308     (narrow-to-region (point)(point))
1309     (insert "(define-char
1310   '")
1311     (insert-char-attributes char readable attributes)
1312     (insert ")\n")
1313     (goto-char (point-min))
1314     (while (re-search-forward "[ \t]+$" nil t)
1315       (replace-match ""))
1316     ;; from tabify.
1317     (goto-char (point-min))
1318     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1319       (let ((column (current-column))
1320             (indent-tabs-mode t))
1321         (delete-region (match-beginning 0) (point))
1322         (indent-to column)))
1323     (goto-char (point-max))
1324     ;; (tabify (point-min)(point-max))
1325     ))
1326
1327 (defun insert-char-data-with-variant (char &optional printable
1328                                            no-ucs-unified
1329                                            script excluded-script)
1330   (insert-char-data char printable)
1331   (let ((variants (char-variants char))
1332         rest
1333         variant vs ret)
1334     (setq variants (sort variants #'<))
1335     (setq rest variants)
1336     (setq variants (cons char variants))
1337     (while rest
1338       (setq variant (car rest))
1339       (unless (get-char-attribute variant '<-subsumptive)
1340         (if (and (or (null script)
1341                      (null (setq vs (get-char-attribute variant 'script)))
1342                      (memq script vs))
1343                  (or (null excluded-script)
1344                      (null (setq vs (get-char-attribute variant 'script)))
1345                      (not (memq excluded-script vs))))
1346             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1347               (insert-char-data variant printable)
1348               (if (setq ret (char-variants variant))
1349                   (while ret
1350                     (or (memq (car ret) variants)
1351                         ;; (get-char-attribute (car ret) '<-subsumptive)
1352                         (setq rest (nconc rest (list (car ret)))))
1353                     (setq ret (cdr ret)))))))
1354       (setq rest (cdr rest)))))
1355
1356 (defun insert-char-range-data (min max &optional script excluded-script)
1357   (let ((code min)
1358         char)
1359     (while (<= code max)
1360       (setq char (decode-char '=ucs code))
1361       (if (encode-char char '=ucs 'defined-only)
1362           (insert-char-data-with-variant char nil 'no-ucs-unified
1363                                          script excluded-script))
1364       (setq code (1+ code)))))
1365
1366 (defun write-char-range-data-to-file (min max file
1367                                           &optional script excluded-script)
1368   (let ((coding-system-for-write char-db-file-coding-system))
1369     (with-temp-buffer
1370       (insert (format ";; -*- coding: %s -*-\n"
1371                       char-db-file-coding-system))
1372       (insert-char-range-data min max script excluded-script)
1373       (write-region (point-min)(point-max) file))))
1374
1375 (defvar what-character-original-window-configuration)
1376
1377 ;;;###autoload
1378 (defun what-char-definition (char)
1379   (interactive (list (char-after)))
1380   (let ((buf (get-buffer-create "*Character Description*"))
1381         (the-buf (current-buffer))
1382         (win-conf (current-window-configuration)))
1383     (pop-to-buffer buf)
1384     (make-local-variable 'what-character-original-window-configuration)
1385     (setq what-character-original-window-configuration win-conf)
1386     (setq buffer-read-only nil)
1387     (erase-buffer)
1388     (condition-case err
1389         (progn
1390           (insert-char-data-with-variant char 'printable)
1391           (unless (char-attribute-alist char)
1392             (insert (format ";; = %c\n"
1393                             (let* ((rest (split-char char))
1394                                    (ccs (pop rest))
1395                                    (code (pop rest)))
1396                               (while rest
1397                                 (setq code (logior (lsh code 8)
1398                                                    (pop rest))))
1399                               (decode-char ccs code)))))
1400           ;; (char-db-update-comment)
1401           (set-buffer-modified-p nil)
1402           (view-mode the-buf (lambda (buf)
1403                                (set-window-configuration
1404                                 what-character-original-window-configuration)
1405                                ))
1406           (goto-char (point-min)))
1407       (error (progn
1408                (set-window-configuration
1409                 what-character-original-window-configuration)
1410                (signal (car err) (cdr err)))))))
1411
1412
1413 ;;; @ end
1414 ;;;
1415
1416 (provide 'char-db-util)
1417
1418 ;;; char-db-util.el ends here