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) 1998,1999,2000,2001,2002,2003,2004,2005,2006,2007,
   4 ;;   2008,2009,2010,2011,2012,2013,2014,2015 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    ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?旨 ?喜 ?壴 ; 160
  83 ;  ?旨 ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?喜 ?壴 ; 160
  84    ?鼓 ?豈 ?豆 ?豊 ?豐 ?䖒 ?虍 ?虎 ?虤 ?皿 ; 170
  85    ?𠙴 ?去 ?血 ?丶 ?丹 ?青 ?井 ?皀 ?鬯 ?食 ; 180
  86    ?亼 ?會 ?倉 ?入 ?缶 ?矢 ?高 ?冂 ?𩫏 ?京 ; 190
  87    ?亯 ?𣆪 ?畗 ?㐭 ?嗇 ?來 ?麥 ?夊 ?舛 ?䑞 ; 200
  88    ?韋 ?弟 ?夂 ?久 ?桀 ?木 ?東 ?林 ?才 ?叒 ; 210
  89    ?之 ?帀 ?出 ?𣎵 ?生 ?乇 ?𠂹 ?𠌶 ?華 ?𥝌 ; 220
  90    ?稽 ?巢 ?桼 ?束 ?㯻 ?囗 ?員 ?貝 ?邑 ?𨛜 ; 230
  91    ?日 ?旦 ?倝 ?㫃 ?冥 ?晶 ?月 ?有 ?明 ?囧 ; 240
  92    ?夕 ?多 ?毌 ?𢎘 ?𣐺 ?卣 ?齊 ?朿 ?片 ?鼎 ; 250
  93    ?克 ?彔 ?禾 ?秝 ?黍 ?香 ?米 ?毇 ?臼 ?凶 ; 260
  94    ?𣎳 ?林 ?麻 ?尗 ?耑 ?韭 ?瓜 ?瓠 ?宀 ?宮 ; 270
  95    ?呂 ?穴 ?㝱 ?𤕫 ?冖 ?𠔼 ?冃 ?㒳 ?网 ?襾 ; 280
  96    ?巾 ?巿 ?帛 ?白 ?㡀 ?黹 ?人 ?𠤎 ?匕 ?从 ; 290
  97    ?比 ?北 ?丘 ?㐺 ?𡈼 ?重 ?臥 ?身 ?㐆 ?衣 ; 300
  98    ?裘 ?老 ?毛 ?毳 ?尸 ?尺 ?尾 ?履 ?舟 ?方 ; 310
  99    ?儿 ?兄 ?兂 ?皃 ?𠑹 ?先 ?秃 ?見 ?覞 ?欠 ; 320
 100    ?㱃 ?㳄 ?旡 ?頁 ?𦣻 ?面 ?丏 ?首 ?𥄉 ?須 ; 330
 101    ?彡 ?彣 ?文 ?髟 ?后 ?司 ?卮 ?卩 ?印 ?色 ; 340
 102    ?𠨍 ?辟 ?勹 ?包 ?茍 ?鬼 ?甶 ?厶 ?嵬 ?山 ; 350
 103    ?屾 ?屵 ?广 ?厂 ?丸 ?危 ?石 ?長 ?勿 ?冄 ; 360
 104    ?而 ?豕 ?㣇 ?彑 ?豚 ?豸 ?𤉡 ?易 ?象 ?馬 ; 370
 105    ?𢊁 ?鹿 ?麤 ?㲋 ?兔 ?萈 ?犬 ?㹜 ?鼠 ?能 ; 380
 106    ?熊 ?火 ?炎 ?黑 ?囪 ?焱 ?炙 ?赤 ?大 ?亦 ; 390
 107    ])
 108
 109 (defun shuowen-radical (number)
 110   (aref shuowen-radicals (1- number)))
 111
 112 (defvar char-db-file-coding-system 'utf-8-mcs-er)
 113
 114 (defvar char-db-ignored-attributes '(ideographic-products))
 115
 116 (defvar char-db-coded-charset-priority-list
 117   '(ascii
 118     control-1
 119     latin-iso8859-1
 120     latin-iso8859-2
 121     latin-iso8859-3
 122     latin-iso8859-4
 123     latin-iso8859-9
 124     latin-jisx0201
 125     cyrillic-iso8859-5
 126     greek-iso8859-7
 127     thai-tis620
 128     ;; =mj
 129     ;; =adobe-japan1-0
 130     ;; =adobe-japan1-1
 131     ;; =adobe-japan1-2
 132     ;; =adobe-japan1-3
 133     ;; =adobe-japan1-4
 134     ;; =adobe-japan1-5
 135     ;; =adobe-japan1-6
 136     =jis-x0208
 137     =jis-x0208@1978
 138     =jis-x0208@1983
 139     japanese-jisx0212
 140     chinese-gb2312
 141     =jis-x0208@1990
 142     chinese-cns11643-1
 143     chinese-cns11643-2
 144     chinese-cns11643-3
 145     chinese-cns11643-4
 146     chinese-cns11643-5
 147     chinese-cns11643-6
 148     chinese-cns11643-7
 149     =jis-x0213-1
 150     =jis-x0213-1@2000
 151     =jis-x0213-1@2004
 152     =jis-x0213-2
 153     korean-ksc5601
 154     chinese-isoir165
 155     katakana-jisx0201
 156     hebrew-iso8859-8
 157     chinese-gb12345
 158     latin-viscii
 159     ethiopic-ucs
 160     =big5-cdp
 161     =hanyo-denshi/ja
 162     =hanyo-denshi/jb
 163     =hanyo-denshi/jc
 164     =hanyo-denshi/jd
 165     =hanyo-denshi/ft
 166     =hanyo-denshi/ia
 167     =hanyo-denshi/ib
 168     =hanyo-denshi/hg
 169     =hanyo-denshi/jt
 170     =hanyo-denshi/ks
 171     ;; ==mj
 172     ;; ==adobe-japan1-0
 173     ;; ==adobe-japan1-1
 174     ;; ==adobe-japan1-2
 175     ;; ==adobe-japan1-3
 176     ;; ==adobe-japan1-4
 177     ;; ==adobe-japan1-5
 178     ;; ==adobe-japan1-6
 179     ==jis-x0208
 180     ==jis-x0213-1
 181     ==jis-x0213-2
 182     ==hanyo-denshi/ja
 183     ==hanyo-denshi/jb
 184     ==hanyo-denshi/jc
 185     ==hanyo-denshi/ft
 186     ==hanyo-denshi/ib
 187     ==hanyo-denshi/hg
 188     ==hanyo-denshi/jt
 189     ==hanyo-denshi/ks
 190     =gt
 191     =gt-k
 192     =daikanwa
 193     =daikanwa@rev2
 194     =daikanwa@rev1
 195     =daikanwa/+p
 196     ==daikanwa
 197     =cbeta
 198     ideograph-hanziku-1
 199     ideograph-hanziku-2
 200     ideograph-hanziku-3
 201     ideograph-hanziku-4
 202     ideograph-hanziku-5
 203     ideograph-hanziku-6
 204     ideograph-hanziku-7
 205     ideograph-hanziku-8
 206     ideograph-hanziku-9
 207     ideograph-hanziku-10
 208     ideograph-hanziku-11
 209     ideograph-hanziku-12
 210     ;; =>>>adobe-japan1-0
 211     ;; =>>>adobe-japan1-1
 212     ;; =>>>adobe-japan1-2
 213     ;; =>>>adobe-japan1-3
 214     ;; =>>>adobe-japan1-4
 215     ;; =>>>adobe-japan1-5
 216     ;; =>>>adobe-japan1-6
 217     ;; =>>>jis-x0208
 218     ;; =>>>jis-x0213-1
 219     ;; =>>>jis-x0213-2
 220     ;; =>>>hanyo-denshi/ja
 221     ;; =>>>hanyo-denshi/jb
 222     ;; =>>>hanyo-denshi/jc
 223     ;; =>>>hanyo-denshi/ft
 224     ;; =>>>hanyo-denshi/ib
 225     ;; =>>>hanyo-denshi/hg
 226     ;; =>>>hanyo-denshi/jt
 227     ;; =>>>hanyo-denshi/ks
 228     ;; =>>>gt
 229     =>>adobe-japan1-0
 230     =>>adobe-japan1-1
 231     =>>adobe-japan1-2
 232     =>>adobe-japan1-3
 233     =>>adobe-japan1-4
 234     =>>adobe-japan1-5
 235     =>>adobe-japan1-6
 236     =>>jis-x0208
 237     =>>jis-x0213-1
 238     =>>jis-x0213-1@2000
 239     =>>jis-x0213-1@2004
 240     =>>jis-x0213-2
 241     =>>jis-x0208@1978
 242     =>>hanyo-denshi/ft
 243     =>>hanyo-denshi/jt
 244     =>>hanyo-denshi/ks
 245     =>>gt
 246     =>>daikanwa
 247     =>>cbeta
 248     =+>jis-x0208
 249     =+>jis-x0213-1
 250     =+>jis-x0213-2
 251     =+>adobe-japan1-0
 252     =+>adobe-japan1-1
 253     =+>adobe-japan1-2
 254     =+>adobe-japan1-3
 255     =+>adobe-japan1-4
 256     =+>adobe-japan1-5
 257     =+>adobe-japan1-6
 258     =+>jis-x0208@1978
 259     =>jis-x0208
 260     =>jis-x0208@1997
 261     =>jis-x0213-1
 262     =>jis-x0213-1@2000
 263     =>jis-x0213-1@2004
 264     =>jis-x0213-2
 265     ==>ucs@bucs
 266     =>iwds-1
 267     =>ucs@hanyo-denshi
 268     =>ucs@iso
 269     =>ucs@unicode
 270     =>ucs@jis
 271     =>ucs@cns
 272     =>ucs@ks
 273     =+>ucs@iso
 274     =+>ucs@unicode
 275     =+>ucs@jis
 276     =+>ucs@jis/1990
 277     =+>ucs@cns
 278     =+>ucs@ks
 279     =>>ucs@iso
 280     =>>ucs@unicode
 281     =>>ucs@jis
 282     =>>ucs@cns
 283     =>>>ucs@iso
 284     =>>>ucs@unicode
 285     ==ucs@iso
 286     ==ucs@unicode
 287     ;; ==ucs@cns
 288     ==gb2312
 289     ==ks-x1001
 290     ==cns11643-1
 291     ==cns11643-2
 292     ==cns11643-3
 293     ==cns11643-4
 294     ==cns11643-5
 295     ==cns11643-6
 296     ==cns11643-7
 297     ==gt
 298     ==jis-x0208@1990
 299     ;; ==jis-x0208@1983
 300     ==jis-x0208@1978
 301     ==gt-k
 302     =ucs@iso
 303     =ucs@unicode
 304     =ucs@cns
 305     ==big5-cdp
 306     ==cbeta
 307     =>>big5-cdp
 308     =>>gt-k
 309     =+>gt
 310     =>gt
 311     =>big5-cdp
 312     =>daikanwa
 313     =>daikanwa/ho
 314     =>cns11643-5
 315     =>cns11643-7
 316     =big5
 317     =big5-eten
 318     =>gt-k
 319     =zinbun-oracle
 320     =>zinbun-oracle
 321     =ruimoku-v6
 322     =>>ruimoku-v6
 323     ==ruimoku-v6
 324     =jef-china3
 325     =>cbeta
 326     =shinjigen
 327     =big5-cdp-var-3
 328     =big5-cdp-var-5
 329     =>ucs@iwds-1))
 330
 331
 332 ;;; @ char-db formatters
 333 ;;;
 334
 335 (defun char-db-make-char-spec (char)
 336   (let (ret char-spec)
 337     (cond ((characterp char)
 338            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 339                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 340                   (setq char-spec (list (cons '=ucs ret)))
 341                   (cond ((setq ret (get-char-attribute char 'name))
 342                          (setq char-spec (cons (cons 'name ret) char-spec))
 343                          )
 344                         ((setq ret (get-char-attribute char 'name*))
 345                          (setq char-spec (cons (cons 'name* ret) char-spec))
 346                          ))
 347                   )
 348                  ((encode-char char '=mj 'defined-only)
 349                   (setq char-spec nil)
 350                   (dolist (ccs (charset-list))
 351                     (if (and (or (eq ccs '=mj)
 352                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 353                                  (string-match "=ucs@" (symbol-name ccs))
 354                                  )
 355                              (setq ccs (charset-name ccs))
 356                              (null (assq ccs char-spec))
 357                              (setq ret (encode-char char ccs 'defined-only)))
 358                         (setq char-spec (cons (cons ccs ret) char-spec))))
 359                   )
 360                  ((encode-char char '==mj 'defined-only)
 361                   (setq char-spec nil)
 362                   (dolist (ccs (charset-list))
 363                     (if (and (or (eq ccs '==mj)
 364                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 365                                  (string-match "=ucs@" (symbol-name ccs))
 366                                  )
 367                              (setq ccs (charset-name ccs))
 368                              (null (assq ccs char-spec))
 369                              (setq ret (encode-char char ccs 'defined-only)))
 370                         (setq char-spec (cons (cons ccs ret) char-spec))))
 371                   )
 372                  ((encode-char char '=adobe-japan1 'defined-only)
 373                   (setq char-spec nil)
 374                   (dolist (ccs (charset-list))
 375                     (if (and (or (memq ccs
 376                                        '(=adobe-japan1-0
 377                                          =adobe-japan1-1
 378                                          =adobe-japan1-2
 379                                          =adobe-japan1-3
 380                                          =adobe-japan1-4
 381                                          =adobe-japan1-5
 382                                          =adobe-japan1-6
 383                                          ))
 384                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 385                                  (string-match "=ucs@" (symbol-name ccs))
 386                                  )
 387                              (setq ccs (charset-name ccs))
 388                              (null (assq ccs char-spec))
 389                              (setq ret (encode-char char ccs 'defined-only)))
 390                         (setq char-spec (cons (cons ccs ret) char-spec))))
 391                   )
 392                  ((encode-char char '==adobe-japan1 'defined-only)
 393                   (setq char-spec nil)
 394                   (dolist (ccs (charset-list))
 395                     (if (and (or (memq ccs
 396                                        '(==adobe-japan1-0
 397                                          ==adobe-japan1-1
 398                                          ==adobe-japan1-2
 399                                          ==adobe-japan1-3
 400                                          ==adobe-japan1-4
 401                                          ==adobe-japan1-5
 402                                          ==adobe-japan1-6
 403                                          ))
 404                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 405                                  (string-match "=ucs@" (symbol-name ccs))
 406                                  )
 407                              (setq ccs (charset-name ccs))
 408                              (null (assq ccs char-spec))
 409                              (setq ret (encode-char char ccs 'defined-only)))
 410                         (setq char-spec (cons (cons ccs ret) char-spec))))
 411                   )
 412                  ((setq ret
 413                         (catch 'tag
 414                           (let ((rest char-db-coded-charset-priority-list)
 415                                 ccs)
 416                             (while rest
 417                               (setq ccs (charset-name
 418                                          (find-charset (car rest))))
 419                               (if (setq ret
 420                                         (encode-char char ccs
 421                                                      'defined-only))
 422                                   (throw 'tag (cons ccs ret)))
 423                               (setq rest (cdr rest))))))
 424                   (setq char-spec (list ret))
 425                   (dolist (ccs (delq (car ret) (charset-list)))
 426                     (if (and (or (charset-iso-final-char ccs)
 427                                  (memq ccs
 428                                        '(=daikanwa
 429                                          =daikanwa@rev2
 430                                          ;; =gt-k
 431                                          =jis-x0208@1997
 432                                          ))
 433                                  (eq (charset-property ccs 'iso-ir) 177)
 434                                  ;; (string-match "=ucs@" (symbol-name ccs))
 435                                  )
 436                              (setq ccs (charset-name ccs))
 437                              (null (assq ccs char-spec))
 438                              (setq ret (encode-char char ccs 'defined-only)))
 439                         (setq char-spec (cons (cons ccs ret) char-spec))))
 440                   (if (null char-spec)
 441                       (setq char-spec (split-char char)))
 442                   (cond ((setq ret (get-char-attribute char 'name))
 443                          (setq char-spec (cons (cons 'name ret) char-spec))
 444                          )
 445                         ((setq ret (get-char-attribute char 'name*))
 446                          (setq char-spec (cons (cons 'name* ret) char-spec))
 447                          ))
 448                   )
 449                  ((setq ret (get-char-attribute
 450                              char 'ideographic-combination))
 451                   (setq char-spec
 452                         (cons (cons 'ideographic-combination ret)
 453                               char-spec))
 454                   ))
 455            char-spec)
 456           ((consp char)
 457            char))))
 458
 459 (defun char-db-insert-char-spec (char &optional readable column
 460                                       required-features)
 461   (unless column
 462     (setq column (current-column)))
 463   (let (char-spec temp-char)
 464     (setq char-spec (char-db-make-char-spec char))
 465     (unless (or (characterp char) ; char
 466                 (condition-case nil
 467                     (setq char (find-char char-spec))
 468                   (error nil)))
 469       ;; define temporary character
 470       ;;   Current implementation is dirty.
 471       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 472                                          char-spec)))
 473       (remove-char-attribute temp-char 'ideograph-daikanwa)
 474       (setq char temp-char))
 475     (insert-char-attributes char
 476                             readable
 477                             (union (mapcar #'car char-spec)
 478                                    required-features)
 479                             nil 'for-sub-node)
 480     (when temp-char
 481       ;; undefine temporary character
 482       ;;   Current implementation is dirty.
 483       (setq char-spec (char-attribute-alist temp-char))
 484       (while char-spec
 485         (remove-char-attribute temp-char (car (car char-spec)))
 486         (setq char-spec (cdr char-spec))))))
 487
 488 (defun char-db-insert-alist (alist &optional readable column)
 489   (unless column
 490     (setq column (current-column)))
 491   (let ((line-breaking
 492          (concat "\n" (make-string (1+ column) ?\ )))
 493         name value
 494         ret al ; cal
 495         key
 496         lbs cell rest separator)
 497     (insert "(")
 498     (while alist
 499       (setq name (car (car alist))
 500             value (cdr (car alist)))
 501       (cond ((eq name 'char)
 502              (insert "(char . ")
 503              (if (setq ret (condition-case nil
 504                                (find-char value)
 505                              (error nil)))
 506                  (progn
 507                    (setq al nil
 508                          ;; cal nil
 509                          )
 510                    (while value
 511                      (setq key (car (car value)))
 512                      ;; (if (find-charset key)
 513                      ;;     (setq cal (cons key cal))
 514                      (setq al (cons key al))
 515                      ;; )
 516                      (setq value (cdr value)))
 517                    (insert-char-attributes ret
 518                                            readable
 519                                            (or al 'none) ; cal
 520                                            nil 'for-sub-node))
 521                (insert (prin1-to-string value)))
 522              (insert ")")
 523              (insert line-breaking))
 524             ((consp value)
 525              (insert (format "(%-18s " name))
 526              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 527              (while (consp value)
 528                (setq cell (car value))
 529                (if (and (consp cell)
 530                         (consp (car cell))
 531                         (setq ret (condition-case nil
 532                                       (find-char cell)
 533                                     (error nil)))
 534                         )
 535                    (progn
 536                      (setq rest cell
 537                            al nil
 538                            ;; cal nil
 539                            )
 540                      (while rest
 541                        (setq key (car (car rest)))
 542                        ;; (if (find-charset key)
 543                        ;;     (setq cal (cons key cal))
 544                        (setq al (cons key al))
 545                        ;; )
 546                        (setq rest (cdr rest)))
 547                      (if separator
 548                          (insert lbs))
 549                      (insert-char-attributes ret
 550                                              readable
 551                                              al ; cal
 552                                              nil 'for-sub-node)
 553                      (setq separator lbs))
 554                  (if separator
 555                      (insert separator))
 556                  (insert (prin1-to-string cell))
 557                  (setq separator " "))
 558                (setq value (cdr value)))
 559              (insert ")")
 560              (insert line-breaking))
 561             (t
 562              (insert (format "(%-18s . %S)%s"
 563                              name value
 564                              line-breaking))))
 565       (setq alist (cdr alist))))
 566   (insert ")"))
 567
 568 (defun char-db-insert-char-reference (plist &optional readable column)
 569   (unless column
 570     (setq column (current-column)))
 571   (let ((line-breaking
 572          (concat "\n" (make-string (1+ column) ?\ )))
 573         (separator "")
 574         name value)
 575     (insert "(")
 576     (while plist
 577       (setq name (pop plist))
 578       (setq value (pop plist))
 579       (cond ((eq name :char)
 580              (insert separator)
 581              (insert ":char\t")
 582              (cond ((numberp value)
 583                     (setq value (decode-char '=ucs value)))
 584                    ;; ((consp value)
 585                    ;;  (setq value (or (find-char value)
 586                    ;;                  value)))
 587                    )
 588              (char-db-insert-char-spec value readable)
 589              (insert line-breaking)
 590              (setq separator ""))
 591             ((eq name :radical)
 592              (insert (format "%s%s\t%d ; %c%s"
 593                              separator
 594                              name value
 595                              (ideographic-radical value)
 596                              line-breaking))
 597              (setq separator ""))
 598             (t
 599              (insert (format "%s%s\t%S" separator name value))
 600              (setq separator line-breaking)))
 601       ))
 602   (insert ")"))
 603
 604 (defun char-db-decode-isolated-char (ccs code-point)
 605   (let (ret)
 606     (setq ret
 607           (cond ((eq ccs 'arabic-iso8859-6)
 608                  (decode-char ccs code-point))
 609                 ;; ((eq ccs '=gt)
 610                 ;;  (decode-builtin-char '==gt code-point))
 611                 ((and (memq ccs '(=gt-pj-1
 612                                   =gt-pj-2
 613                                   =gt-pj-3
 614                                   =gt-pj-4
 615                                   =gt-pj-5
 616                                   =gt-pj-6
 617                                   =gt-pj-7
 618                                   =gt-pj-8
 619                                   =gt-pj-9
 620                                   =gt-pj-10
 621                                   =gt-pj-11))
 622                       (setq ret (decode-char ccs code-point))
 623                       (setq ret (encode-char ret '=gt 'defined-only)))
 624                  (decode-builtin-char '=gt ret))
 625                 (t
 626                  (decode-builtin-char ccs code-point))))
 627     (cond ((null ret)
 628            (or (decode-char ccs code-point)
 629                (define-char (list (cons ccs code-point)))))
 630           ((and (<= 0 (char-int ret))
 631                 (<= (char-int ret) #x1F))
 632            (decode-char '=ucs (+ #x2400 (char-int ret))))
 633           ((= (char-int ret) #x7F)
 634            ?\u2421)
 635           (t ret))))
 636
 637 (defvar char-db-convert-obsolete-format t)
 638
 639 (defun char-db-insert-ccs-feature (name value line-breaking)
 640   (cond
 641    ((integerp value)
 642     (insert
 643      (format
 644       (cond
 645        ((memq name '(=>iwds-1
 646                      ==shinjigen
 647                      =shinjigen
 648                      =shinjigen@1ed ==shinjigen@1ed
 649                      =shinjigen@rev ==shinjigen@rev
 650                      =shinjigen/+p@rev ==shinjigen/+p@rev
 651                      ===daikanwa/ho ==daikanwa/ho
 652                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 653         "(%-18s .  %04d)\t; %c")
 654        ((eq name '=shinjigen@1ed/24pr)
 655         "(%-18s . %04d)\t; %c")
 656        ((or (memq name '(===daikanwa
 657                          ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 658                          =daikanwa@rev1 =daikanwa@rev2
 659                          =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 660                          =>>daikanwa/+p
 661                          =daikanwa/+2p =>>daikanwa/+2p
 662                          =gt ==gt ===gt
 663                          =>>gt =+>gt =>gt
 664                          =gt-k ==gt-k ===gt-k
 665                          =>>gt-k =>gt-k
 666                          =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 667                          =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 668                          =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 669                          =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 670                          =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 671                          =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 672                          =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 673                          =>>adobe-japan1-0 =+>adobe-japan1-0
 674                          =>>adobe-japan1-1 =+>adobe-japan1-1
 675                          =>>adobe-japan1-2 =+>adobe-japan1-2
 676                          =>>adobe-japan1-3 =+>adobe-japan1-3
 677                          =>>adobe-japan1-4 =+>adobe-japan1-4
 678                          =>>adobe-japan1-5 =+>adobe-japan1-5
 679                          =>>adobe-japan1-6 =+>adobe-japan1-6
 680                          =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 681                          =zinbun-oracle =>zinbun-oracle
 682                          ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 683                          ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 684                          ===hng-khi ===hng-khm ===hng-sok
 685                          ===hng-yhk ===hng-kak
 686                          ===hng-sys ===hng-tsu
 687                          ===hng-hod ===hng-gok ===hng-nto
 688                          ===hng-nkm ===hng-k24
 689                          ===hng-kcj ===hng-kbk ===hng-sik
 690                          ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 691                          ===hng-okd ===hng-kmi
 692                          ===hng-zkd ===hng-jyu
 693                          ===hng-kss ===hng-kyo
 694                          ===hng-smk))
 695             ;; (string-match "^=adobe-" (symbol-name name))
 696             )
 697         "(%-18s . %05d)\t; %c")
 698        ((memq name '(=hanyo-denshi/ks
 699                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 700                      =>>hanyo-denshi/ks
 701                      =koseki
 702                      =mj ==mj ===mj =>>mj
 703                      =zihai mojikyo))
 704         "(%-18s . %06d)\t; %c")
 705        ((>= (charset-dimension name) 2)
 706         "(%-18s . #x%04X)\t; %c")
 707        (t
 708         "(%-18s . #x%02X)\t; %c"))
 709       name
 710       (if (= (charset-iso-graphic-plane name) 1)
 711           (logior value
 712                   (cond ((= (charset-dimension name) 1)
 713                          #x80)
 714                         ((= (charset-dimension name) 2)
 715                          #x8080)
 716                         ((= (charset-dimension name) 3)
 717                          #x808080)
 718                         (t 0)))
 719         value)
 720       (char-db-decode-isolated-char name value)))
 721     (if (and (= (charset-chars name) 94)
 722              (= (charset-dimension name) 2))
 723         (insert (format " [%02d-%02d]"
 724                         (- (lsh value -8) 32)
 725                         (- (logand value 255) 32))))
 726     )
 727    (t
 728     (insert (format "(%-18s . %s)" name value))
 729     ))
 730   (insert line-breaking))
 731
 732 (defun char-db-insert-relation-feature (char name value line-breaking
 733                                              ccss readable)
 734   (insert (format "(%-18s%s " name line-breaking))
 735   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 736         separator cell sources required-features
 737         ret)
 738     (while (consp value)
 739       (setq cell (car value))
 740       (if (integerp cell)
 741           (setq cell (decode-char '=ucs cell)))
 742       (cond
 743        ((eq name '->subsumptive)
 744         (when (or (not (some (lambda (atr)
 745                                (get-char-attribute cell atr))
 746                              char-db-ignored-attributes))
 747                   (some (lambda (ccs)
 748                           (encode-char cell ccs 'defined-only))
 749                         ccss))
 750           (if separator
 751               (insert lbs))
 752           (let ((char-db-ignored-attributes
 753                  (cons '<-subsumptive
 754                        char-db-ignored-attributes)))
 755             (insert-char-attributes cell readable nil nil 'for-sub-node))
 756           (setq separator lbs))
 757         )
 758        ((characterp cell)
 759         (setq sources
 760               (get-char-attribute
 761                char (intern (format "%s*sources" name))))
 762         (setq required-features nil)
 763         (dolist (source sources)
 764           (cond
 765            ((memq source '(JP
 766                            JP/Jouyou
 767                            shinjigen shinjigen@1ed shinjigen@rev))
 768             (setq required-features
 769                   (union required-features
 770                          '(=jis-x0208
 771                            =jis-x0208@1990
 772                            =jis-x0213-1@2000
 773                            =jis-x0213-1@2004
 774                            =jis-x0213-2
 775                            =jis-x0212
 776                            =jis-x0208@1983
 777                            =jis-x0208@1978
 778                            =shinjigen
 779                            =shinjigen@1ed
 780                            =shinjigen@rev
 781                            =shinjigen/+p@rev))))
 782            ((eq source 'CN)
 783             (setq required-features
 784                   (union required-features
 785                          '(=gb2312
 786                            =gb12345
 787                            =iso-ir165)))))
 788           (cond
 789            ((find-charset (setq ret (intern (format "=%s" source))))
 790             (setq required-features
 791                   (cons ret required-features)))
 792            (t (setq required-features
 793                     (cons source required-features)))))
 794         (cond ((string-match "@JP" (symbol-name name))
 795                (setq required-features
 796                      (union required-features
 797                             '(=jis-x0208
 798                               =jis-x0208@1990
 799                               =jis-x0213-1-2000
 800                               =jis-x0213-2-2000
 801                               =jis-x0212
 802                               =jis-x0208@1983
 803                               =jis-x0208@1978))))
 804               ((string-match "@CN" (symbol-name name))
 805                (setq required-features
 806                      (union required-features
 807                             '(=gb2312
 808                               =gb12345
 809                               =iso-ir165)))))
 810         (if separator
 811             (insert lbs))
 812         (if readable
 813             (insert (format "%S" cell))
 814           (char-db-insert-char-spec cell readable
 815                                     nil
 816                                     required-features))
 817         (setq separator lbs))
 818        ((consp cell)
 819         (if separator
 820             (insert lbs))
 821         (if (consp (car cell))
 822             (char-db-insert-char-spec cell readable)
 823           (char-db-insert-char-reference cell readable))
 824         (setq separator lbs))
 825        (t
 826         (if separator
 827             (insert separator))
 828         (insert (prin1-to-string cell))
 829         (setq separator " ")))
 830       (setq value (cdr value)))
 831     (insert ")")
 832     (insert line-breaking)))
 833
 834 (defun insert-char-attributes (char &optional readable attributes column
 835                                     for-sub-node)
 836   (unless column
 837     (setq column (current-column)))
 838   (let (name value ; has-long-ccs-name
 839         rest
 840         radical strokes
 841         (line-breaking
 842          (concat "\n" (make-string (1+ column) ?\ )))
 843         lbs cell separator ret
 844         key al cal
 845         dest-ccss ; sources required-features
 846         ccss)
 847     (let (atr-d)
 848       (setq attributes
 849             (sort (if attributes
 850                       (if (consp attributes)
 851                           (progn
 852                             (dolist (name attributes)
 853                               (unless (memq name char-db-ignored-attributes)
 854                                 (if (find-charset name)
 855                                     (push name ccss))
 856                                 (push name atr-d)))
 857                             atr-d))
 858                     (dolist (name (char-attribute-list))
 859                       (unless (memq name char-db-ignored-attributes)
 860                         (if (find-charset name)
 861                             (push name ccss))
 862                         (push name atr-d)))
 863                     atr-d)
 864                   #'char-attribute-name<)))
 865     (insert "(")
 866     (when (memq '<-subsumptive attributes)
 867       (when (or readable (not for-sub-node))
 868         (when (setq value (get-char-attribute char '<-subsumptive))
 869           (char-db-insert-relation-feature char '<-subsumptive value
 870                                            line-breaking
 871                                            ccss readable)))
 872       (setq attributes (delq '<-subsumptive attributes)))
 873     (when (and (memq '<-denotational attributes)
 874                (setq value (get-char-attribute char '<-denotational)))
 875       (char-db-insert-relation-feature char '<-denotational value
 876                                        line-breaking
 877                                        ccss readable)
 878       (setq attributes (delq '<-denotational attributes)))
 879     (when (and (memq '<-denotational@component attributes)
 880                (setq value (get-char-attribute char '<-denotational@component)))
 881       (char-db-insert-relation-feature char '<-denotational@component value
 882                                        line-breaking
 883                                        ccss readable)
 884       (setq attributes (delq '<-denotational@component attributes)))
 885     (when (and (memq 'name attributes)
 886                (setq value (get-char-attribute char 'name)))
 887       (insert (format
 888                (if (> (+ (current-column) (length value)) 48)
 889                    "(name . %S)%s"
 890                  "(name               . %S)%s")
 891                value line-breaking))
 892       (setq attributes (delq 'name attributes))
 893       )
 894     (when (and (memq 'name* attributes)
 895                (setq value (get-char-attribute char 'name*)))
 896       (insert (format
 897                (if (> (+ (current-column) (length value)) 48)
 898                    "(name* . %S)%s"
 899                  "(name*              . %S)%s")
 900                value line-breaking))
 901       (setq attributes (delq 'name* attributes))
 902       )
 903     (when (and (memq 'script attributes)
 904                (setq value (get-char-attribute char 'script)))
 905       (insert (format "(script\t\t%s)%s"
 906                       (mapconcat (function prin1-to-string)
 907                                  value " ")
 908                       line-breaking))
 909       (setq attributes (delq 'script attributes))
 910       )
 911     (dolist (name '(=>ucs =>ucs*))
 912       (when (and (memq name attributes)
 913                  (setq value (get-char-attribute char name)))
 914         (insert (format "(%-18s . #x%04X)\t; %c%s"
 915                         name value (decode-char '=ucs value)
 916                         line-breaking))
 917         (setq attributes (delq name attributes))))
 918     (dolist (name '(=>ucs@gb =>ucs@big5))
 919       (when (and (memq name attributes)
 920                  (setq value (get-char-attribute char name)))
 921         (insert (format "(%-18s . #x%04X)\t; %c%s"
 922                         name value
 923                         (decode-char (intern
 924                                       (concat "="
 925                                               (substring
 926                                                (symbol-name name) 2)))
 927                                      value)
 928                         line-breaking))
 929         (setq attributes (delq name attributes))
 930         ))
 931     ;; (dolist (name '(=>daikanwa))
 932     ;;   (when (and (memq name attributes)
 933     ;;              (setq value (get-char-attribute char name)))
 934     ;;     (insert
 935     ;;      (if (integerp value)
 936     ;;          (format "(%-18s . %05d)\t; %c%s"
 937     ;;                  name value (decode-char '=daikanwa value)
 938     ;;                  line-breaking)
 939     ;;        (format "(%-18s %s)\t; %c%s"
 940     ;;                name
 941     ;;                (mapconcat (function prin1-to-string)
 942     ;;                           value " ")
 943     ;;                (char-representative-of-daikanwa char)
 944     ;;                line-breaking)))
 945     ;;     (setq attributes (delq name attributes))))
 946     (when (and (memq 'general-category attributes)
 947                (setq value (get-char-attribute char 'general-category)))
 948       (insert (format
 949                "(general-category\t%s) ; %s%s"
 950                (mapconcat (lambda (cell)
 951                             (format "%S" cell))
 952                           value " ")
 953                (cond ((rassoc value unidata-normative-category-alist)
 954                       "Normative Category")
 955                      ((rassoc value unidata-informative-category-alist)
 956                       "Informative Category")
 957                      (t
 958                       "Unknown Category"))
 959                line-breaking))
 960       (setq attributes (delq 'general-category attributes))
 961       )
 962     (when (and (memq 'bidi-category attributes)
 963                (setq value (get-char-attribute char 'bidi-category)))
 964       (insert (format "(bidi-category\t. %S)%s"
 965                       value
 966                       line-breaking))
 967       (setq attributes (delq 'bidi-category attributes))
 968       )
 969     (unless (or (not (memq 'mirrored attributes))
 970                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
 971                     'empty))
 972       (insert (format "(mirrored\t\t. %S)%s"
 973                       value
 974                       line-breaking))
 975       (setq attributes (delq 'mirrored attributes))
 976       )
 977     (cond
 978      ((and (memq 'decimal-digit-value attributes)
 979            (setq value (get-char-attribute char 'decimal-digit-value)))
 980       (insert (format "(decimal-digit-value . %S)%s"
 981                       value
 982                       line-breaking))
 983       (setq attributes (delq 'decimal-digit-value attributes))
 984       (when (and (memq 'digit-value attributes)
 985                  (setq value (get-char-attribute char 'digit-value)))
 986         (insert (format "(digit-value\t . %S)%s"
 987                         value
 988                         line-breaking))
 989         (setq attributes (delq 'digit-value attributes))
 990         )
 991       (when (and (memq 'numeric-value attributes)
 992                  (setq value (get-char-attribute char 'numeric-value)))
 993         (insert (format "(numeric-value\t . %S)%s"
 994                         value
 995                         line-breaking))
 996         (setq attributes (delq 'numeric-value attributes))
 997         )
 998       )
 999      (t
1000       (when (and (memq 'digit-value attributes)
1001                  (setq value (get-char-attribute char 'digit-value)))
1002         (insert (format "(digit-value\t. %S)%s"
1003                         value
1004                         line-breaking))
1005         (setq attributes (delq 'digit-value attributes))
1006         )
1007       (when (and (memq 'numeric-value attributes)
1008                  (setq value (get-char-attribute char 'numeric-value)))
1009         (insert (format "(numeric-value\t. %S)%s"
1010                         value
1011                         line-breaking))
1012         (setq attributes (delq 'numeric-value attributes))
1013         )))
1014     (when (and (memq 'iso-10646-comment attributes)
1015                (setq value (get-char-attribute char 'iso-10646-comment)))
1016       (insert (format "(iso-10646-comment\t. %S)%s"
1017                       value
1018                       line-breaking))
1019       (setq attributes (delq 'iso-10646-comment attributes))
1020       )
1021     (when (and (memq 'morohashi-daikanwa attributes)
1022                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1023       (insert (format "(morohashi-daikanwa\t%s)%s"
1024                       (mapconcat (function prin1-to-string) value " ")
1025                       line-breaking))
1026       (setq attributes (delq 'morohashi-daikanwa attributes))
1027       )
1028     (setq radical nil
1029           strokes nil)
1030     (when (and (memq 'ideographic-radical attributes)
1031                (setq value (get-char-attribute char 'ideographic-radical)))
1032       (setq radical value)
1033       (insert (format "(ideographic-radical . %S)\t; %c%s"
1034                       radical
1035                       (ideographic-radical radical)
1036                       line-breaking))
1037       (setq attributes (delq 'ideographic-radical attributes))
1038       )
1039     (when (and (memq 'shuowen-radical attributes)
1040                (setq value (get-char-attribute char 'shuowen-radical)))
1041       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1042                       value
1043                       (shuowen-radical value)
1044                       line-breaking))
1045       (setq attributes (delq 'shuowen-radical attributes))
1046       )
1047     (let (key)
1048       (dolist (domain
1049                (append
1050                 char-db-feature-domains
1051                 (let (dest domain)
1052                   (dolist (feature (char-attribute-list))
1053                     (setq feature (symbol-name feature))
1054                     (when (string-match
1055                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1056                            feature)
1057                       (setq domain (intern (match-string 2 feature)))
1058                      (unless (memq domain dest)
1059                        (setq dest (cons domain dest)))))
1060                   (sort dest #'string<))))
1061         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1062         (when (and (memq key attributes)
1063                    (setq value (get-char-attribute char key)))
1064           (setq radical value)
1065           (insert (format "(%s . %S)\t; %c%s"
1066                           key
1067                           radical
1068                           (ideographic-radical radical)
1069                           line-breaking))
1070           (setq attributes (delq key attributes))
1071           )
1072         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1073         (when (and (memq key attributes)
1074                    (setq value (get-char-attribute char key)))
1075           (setq strokes value)
1076           (insert (format "(%s . %S)%s"
1077                           key
1078                           strokes
1079                           line-breaking))
1080           (setq attributes (delq key attributes))
1081           )
1082         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1083         (when (and (memq key attributes)
1084                    (setq value (get-char-attribute char key)))
1085           (insert (format "(%s       . %S)%s"
1086                           key
1087                           value
1088                           line-breaking))
1089           (setq attributes (delq key attributes))
1090           )
1091         (dolist (feature '(ideographic-radical
1092                            ideographic-strokes
1093                            total-strokes))
1094           (setq key (intern (format "%s@%s*sources" feature domain)))
1095           (when (and (memq key attributes)
1096                      (setq value (get-char-attribute char key)))
1097             (insert (format "(%s%s" key line-breaking))
1098             (dolist (cell value)
1099               (insert (format " %s" cell)))
1100             (insert ")")
1101             (insert line-breaking)
1102             (setq attributes (delq key attributes))
1103             ))
1104         ))
1105     (when (and (memq 'ideographic-strokes attributes)
1106                (setq value (get-char-attribute char 'ideographic-strokes)))
1107       (setq strokes value)
1108       (insert (format "(ideographic-strokes . %S)%s"
1109                       strokes
1110                       line-breaking))
1111       (setq attributes (delq 'ideographic-strokes attributes))
1112       )
1113     (when (and (memq 'kangxi-radical attributes)
1114                (setq value (get-char-attribute char 'kangxi-radical)))
1115       (unless (eq value radical)
1116         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1117                         value
1118                         (ideographic-radical value)
1119                         line-breaking))
1120         (or radical
1121             (setq radical value)))
1122       (setq attributes (delq 'kangxi-radical attributes))
1123       )
1124     (when (and (memq 'kangxi-strokes attributes)
1125                (setq value (get-char-attribute char 'kangxi-strokes)))
1126       (unless (eq value strokes)
1127         (insert (format "(kangxi-strokes\t . %S)%s"
1128                         value
1129                         line-breaking))
1130         (or strokes
1131             (setq strokes value)))
1132       (setq attributes (delq 'kangxi-strokes attributes))
1133       )
1134     (when (and (memq 'japanese-radical attributes)
1135                (setq value (get-char-attribute char 'japanese-radical)))
1136       (unless (eq value radical)
1137         (insert (format "(japanese-radical\t . %S)\t; %c%s"
1138                         value
1139                         (ideographic-radical value)
1140                         line-breaking))
1141         (or radical
1142             (setq radical value)))
1143       (setq attributes (delq 'japanese-radical attributes))
1144       )
1145     (when (and (memq 'japanese-strokes attributes)
1146                (setq value (get-char-attribute char 'japanese-strokes)))
1147       (unless (eq value strokes)
1148         (insert (format "(japanese-strokes\t . %S)%s"
1149                         value
1150                         line-breaking))
1151         (or strokes
1152             (setq strokes value)))
1153       (setq attributes (delq 'japanese-strokes attributes))
1154       )
1155     (when (and (memq 'cns-radical attributes)
1156                (setq value (get-char-attribute char 'cns-radical)))
1157       (insert (format "(cns-radical\t . %S)\t; %c%s"
1158                       value
1159                       (ideographic-radical value)
1160                       line-breaking))
1161       (setq attributes (delq 'cns-radical attributes))
1162       )
1163     (when (and (memq 'cns-strokes attributes)
1164                (setq value (get-char-attribute char 'cns-strokes)))
1165       (unless (eq value strokes)
1166         (insert (format "(cns-strokes\t . %S)%s"
1167                         value
1168                         line-breaking))
1169         (or strokes
1170             (setq strokes value)))
1171       (setq attributes (delq 'cns-strokes attributes))
1172       )
1173     ;; (when (and (memq 'shinjigen-1-radical attributes)
1174     ;;            (setq value (get-char-attribute char 'shinjigen-1-radical)))
1175     ;;   (unless (eq value radical)
1176     ;;     (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
1177     ;;                     value
1178     ;;                     (ideographic-radical value)
1179     ;;                     line-breaking))
1180     ;;     (or radical
1181     ;;         (setq radical value)))
1182     ;;   (setq attributes (delq 'shinjigen-1-radical attributes))
1183     ;;   )
1184     (when (and (memq 'ideographic- attributes)
1185                (setq value (get-char-attribute char 'ideographic-)))
1186       (insert "(ideographic-       ")
1187       (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1188             separator nil)
1189       (while (consp value)
1190         (setq cell (car value))
1191         (if (integerp cell)
1192             (setq cell (decode-char '=ucs cell)))
1193         (cond ((characterp cell)
1194                (if separator
1195                    (insert lbs))
1196                (if readable
1197                    (insert (format "%S" cell))
1198                  (char-db-insert-char-spec cell readable))
1199                (setq separator lbs))
1200               ((consp cell)
1201                (if separator
1202                    (insert lbs))
1203                (if (consp (car cell))
1204                    (char-db-insert-char-spec cell readable)
1205                  (char-db-insert-char-reference cell readable))
1206                (setq separator lbs))
1207               (t
1208                (if separator
1209                    (insert separator))
1210                (insert (prin1-to-string cell))
1211                (setq separator " ")))
1212         (setq value (cdr value)))
1213       (insert ")")
1214       (insert line-breaking)
1215       (setq attributes (delq 'ideographic- attributes)))
1216     (when (and (memq 'total-strokes attributes)
1217                (setq value (get-char-attribute char 'total-strokes)))
1218       (insert (format "(total-strokes       . %S)%s"
1219                       value
1220                       line-breaking))
1221       (setq attributes (delq 'total-strokes attributes))
1222       )
1223     (when (and (memq '->ideograph attributes)
1224                (setq value (get-char-attribute char '->ideograph)))
1225       (insert (format "(->ideograph\t%s)%s"
1226                       (mapconcat (lambda (code)
1227                                    (cond ((symbolp code)
1228                                           (symbol-name code))
1229                                          ((integerp code)
1230                                           (format "#x%04X" code))
1231                                          (t
1232                                           (format "%s %S"
1233                                                   line-breaking code))))
1234                                  value " ")
1235                       line-breaking))
1236       (setq attributes (delq '->ideograph attributes))
1237       )
1238     ;; (when (and (memq '->decomposition attributes)
1239     ;;            (setq value (get-char-attribute char '->decomposition)))
1240     ;;   (insert (format "(->decomposition\t%s)%s"
1241     ;;                   (mapconcat (lambda (code)
1242     ;;                                (cond ((symbolp code)
1243     ;;                                       (symbol-name code))
1244     ;;                                      ((characterp code)
1245     ;;                                       (if readable
1246     ;;                                           (format "%S" code)
1247     ;;                                         (format "#x%04X"
1248     ;;                                                 (char-int code))
1249     ;;                                         ))
1250     ;;                                      ((integerp code)
1251     ;;                                       (format "#x%04X" code))
1252     ;;                                      (t
1253     ;;                                       (format "%s%S" line-breaking code))))
1254     ;;                              value " ")
1255     ;;                   line-breaking))
1256     ;;   (setq attributes (delq '->decomposition attributes))
1257     ;;   )
1258     (if (equal (get-char-attribute char '->titlecase)
1259                (get-char-attribute char '->uppercase))
1260         (setq attributes (delq '->titlecase attributes)))
1261     (when (and (memq '->mojikyo attributes)
1262                (setq value (get-char-attribute char '->mojikyo)))
1263       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1264                       value (decode-char 'mojikyo value)
1265                       line-breaking))
1266       (setq attributes (delq '->mojikyo attributes))
1267       )
1268     (when (and (memq 'hanyu-dazidian-vol attributes)
1269                (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1270       (insert (format "(hanyu-dazidian-vol  . %d)%s"
1271                       value line-breaking))
1272       (setq attributes (delq 'hanyu-dazidian-vol attributes))
1273       )
1274     (when (and (memq 'hanyu-dazidian-page attributes)
1275                (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1276       (insert (format "(hanyu-dazidian-page . %d)%s"
1277                       value line-breaking))
1278       (setq attributes (delq 'hanyu-dazidian-page attributes))
1279       )
1280     (when (and (memq 'hanyu-dazidian-char attributes)
1281                (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1282       (insert (format "(hanyu-dazidian-char . %d)%s"
1283                       value line-breaking))
1284       (setq attributes (delq 'hanyu-dazidian-char attributes))
1285       )
1286     (unless readable
1287       (dolist (ignored '(composition
1288                          ->denotational <-subsumptive ->ucs-unified
1289                          ->ideographic-component-forms))
1290         (setq attributes (delq ignored attributes))))
1291     (while attributes
1292       (setq name (car attributes))
1293       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1294                   'value-is-empty)
1295         (cond ((setq ret (find-charset name))
1296                (setq name (charset-name ret))
1297                (when (not (memq name dest-ccss))
1298                  (setq dest-ccss (cons name dest-ccss))
1299                  (char-db-insert-ccs-feature name value line-breaking))
1300                )
1301               ((string-match "^=>ucs@" (symbol-name name))
1302                (insert (format "(%-18s . #x%04X)\t; %c%s"
1303                                name value (decode-char '=ucs value)
1304                                line-breaking))
1305                )
1306               ((eq name 'jisx0208-1978/4X)
1307                (insert (format "(%-18s . #x%04X)%s"
1308                                name value
1309                                line-breaking))
1310                )
1311               ((and
1312                 (not readable)
1313                 (not (eq name '->subsumptive))
1314                 (not (eq name '->uppercase))
1315                 (not (eq name '->lowercase))
1316                 (not (eq name '->titlecase))
1317                 (not (eq name '->canonical))
1318                 (not (eq name '->Bopomofo))
1319                 (not (eq name '->mistakable))
1320                 (not (eq name '->ideographic-variants))
1321                 (null (get-char-attribute
1322                        char (intern (format "%s*sources" name))))
1323                 (not (string-match "\\*sources$" (symbol-name name)))
1324                 (null (get-char-attribute
1325                        char (intern (format "%s*note" name))))
1326                 (not (string-match "\\*note$" (symbol-name name)))
1327                 (or (eq name '<-identical)
1328                     (eq name '<-uppercase)
1329                     (eq name '<-lowercase)
1330                     (eq name '<-titlecase)
1331                     (eq name '<-canonical)
1332                     (eq name '<-ideographic-variants)
1333                     ;; (eq name '<-synonyms)
1334                     (string-match "^<-synonyms" (symbol-name name))
1335                     (eq name '<-mistakable)
1336                     (when (string-match "^->" (symbol-name name))
1337                       (cond
1338                        ((string-match "^->fullwidth" (symbol-name name))
1339                         (not (and (consp value)
1340                                   (characterp (car value))
1341                                   (encode-char
1342                                    (car value) '=ucs 'defined-only)))
1343                         )
1344                        (t)))
1345                     ))
1346                )
1347               ((or (eq name 'ideographic-structure)
1348                    (eq name 'ideographic-combination)
1349                    (eq name 'ideographic-)
1350                    (eq name '=decomposition)
1351                    (char-feature-base-name= '=decomposition name)
1352                    (char-feature-base-name= '=>decomposition name)
1353                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1354                    ;;               (symbol-name name))
1355                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1356                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1357                                  (symbol-name name))
1358                    )
1359                (char-db-insert-relation-feature char name value
1360                                                 line-breaking
1361                                                 ccss readable))
1362               ((memq name '(ideograph=
1363                             original-ideograph-of
1364                             ancient-ideograph-of
1365                             vulgar-ideograph-of
1366                             wrong-ideograph-of
1367                             ;; simplified-ideograph-of
1368                             ideographic-variants
1369                             ;; ideographic-different-form-of
1370                             ))
1371                (insert (format "(%-18s%s " name line-breaking))
1372                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1373                      separator nil)
1374                (while (consp value)
1375                  (setq cell (car value))
1376                  (if (and (consp cell)
1377                           (consp (car cell)))
1378                      (progn
1379                        (if separator
1380                            (insert lbs))
1381                        (char-db-insert-alist cell readable)
1382                        (setq separator lbs))
1383                    (if separator
1384                        (insert separator))
1385                    (insert (prin1-to-string cell))
1386                    (setq separator " "))
1387                  (setq value (cdr value)))
1388                (insert ")")
1389                (insert line-breaking))
1390               ((consp value)
1391                (insert (format "(%-18s " name))
1392                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1393                      separator nil)
1394                (while (consp value)
1395                  (setq cell (car value))
1396                  (if (and (consp cell)
1397                           (consp (car cell))
1398                           (setq ret (condition-case nil
1399                                         (find-char cell)
1400                                       (error nil))))
1401                      (progn
1402                        (setq rest cell
1403                              al nil
1404                              cal nil)
1405                        (while rest
1406                          (setq key (car (car rest)))
1407                          (if (find-charset key)
1408                              (setq cal (cons key cal))
1409                            (setq al (cons key al)))
1410                          (setq rest (cdr rest)))
1411                        (if separator
1412                            (insert lbs))
1413                        (insert-char-attributes ret
1414                                                readable
1415                                                al ; cal
1416                                                nil 'for-sub-node)
1417                        (setq separator lbs))
1418                    (setq ret (prin1-to-string cell))
1419                    (if separator
1420                        (if (< (+ (current-column)
1421                                  (length ret)
1422                                  (length separator))
1423                               76)
1424                            (insert separator)
1425                          (insert lbs)))
1426                    (insert ret)
1427                    (setq separator " "))
1428                  (setq value (cdr value)))
1429                (insert ")")
1430                (insert line-breaking))
1431               (t
1432                (insert (format "(%-18s" name))
1433                (setq ret (prin1-to-string value))
1434                (unless (< (+ (current-column)
1435                              (length ret)
1436                              3)
1437                           76)
1438                  (insert line-breaking))
1439                (insert " . " ret ")" line-breaking)
1440                ;; (insert (format "(%-18s . %S)%s"
1441                ;;                 name value
1442                ;;                 line-breaking))
1443                )
1444               ))
1445       (setq attributes (cdr attributes)))
1446     (insert ")")))
1447
1448 (defun insert-char-data (char &optional readable
1449                               attributes)
1450   (save-restriction
1451     (narrow-to-region (point)(point))
1452     (insert "(define-char
1453   '")
1454     (insert-char-attributes char readable attributes)
1455     (insert ")\n")
1456     (goto-char (point-min))
1457     (while (re-search-forward "[ \t]+$" nil t)
1458       (replace-match ""))
1459     ;; from tabify.
1460     (goto-char (point-min))
1461     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1462       (let ((column (current-column))
1463             (indent-tabs-mode t))
1464         (delete-region (match-beginning 0) (point))
1465         (indent-to column)))
1466     (goto-char (point-max))
1467     ;; (tabify (point-min)(point-max))
1468     ))
1469
1470 (defun insert-char-data-with-variant (char &optional printable
1471                                            no-ucs-unified
1472                                            script excluded-script)
1473   (insert-char-data char printable)
1474   (let ((variants (char-variants char))
1475         rest
1476         variant vs ret)
1477     (setq variants (sort variants #'<))
1478     (setq rest variants)
1479     (setq variants (cons char variants))
1480     (while rest
1481       (setq variant (car rest))
1482       (unless (get-char-attribute variant '<-subsumptive)
1483         (if (and (or (null script)
1484                      (null (setq vs (get-char-attribute variant 'script)))
1485                      (memq script vs))
1486                  (or (null excluded-script)
1487                      (null (setq vs (get-char-attribute variant 'script)))
1488                      (not (memq excluded-script vs))))
1489             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1490               (insert-char-data variant printable)
1491               (if (setq ret (char-variants variant))
1492                   (while ret
1493                     (or (memq (car ret) variants)
1494                         ;; (get-char-attribute (car ret) '<-subsumptive)
1495                         (setq rest (nconc rest (list (car ret)))))
1496                     (setq ret (cdr ret)))))))
1497       (setq rest (cdr rest)))))
1498
1499 (defun insert-char-range-data (min max &optional script excluded-script)
1500   (let ((code min)
1501         char)
1502     (while (<= code max)
1503       (setq char (decode-char '=ucs code))
1504       (if (encode-char char '=ucs 'defined-only)
1505           (insert-char-data-with-variant char nil 'no-ucs-unified
1506                                          script excluded-script))
1507       (setq code (1+ code)))))
1508
1509 (defun write-char-range-data-to-file (min max file
1510                                           &optional script excluded-script)
1511   (let ((coding-system-for-write char-db-file-coding-system))
1512     (with-temp-buffer
1513       (insert (format ";; -*- coding: %s -*-\n"
1514                       char-db-file-coding-system))
1515       (insert-char-range-data min max script excluded-script)
1516       (write-region (point-min)(point-max) file))))
1517
1518 (defvar what-character-original-window-configuration)
1519
1520 ;;;###autoload
1521 (defun what-char-definition (char)
1522   (interactive (list (char-after)))
1523   (let ((buf (get-buffer-create "*Character Description*"))
1524         (the-buf (current-buffer))
1525         (win-conf (current-window-configuration)))
1526     (pop-to-buffer buf)
1527     (make-local-variable 'what-character-original-window-configuration)
1528     (setq what-character-original-window-configuration win-conf)
1529     (setq buffer-read-only nil)
1530     (erase-buffer)
1531     (condition-case err
1532         (progn
1533           (insert-char-data-with-variant char 'printable)
1534           (unless (char-attribute-alist char)
1535             (insert (format ";; = %c\n"
1536                             (let* ((rest (split-char char))
1537                                    (ccs (pop rest))
1538                                    (code (pop rest)))
1539                               (while rest
1540                                 (setq code (logior (lsh code 8)
1541                                                    (pop rest))))
1542                               (decode-char ccs code)))))
1543           ;; (char-db-update-comment)
1544           (set-buffer-modified-p nil)
1545           (view-mode the-buf (lambda (buf)
1546                                (set-window-configuration
1547                                 what-character-original-window-configuration)
1548                                ))
1549           (goto-char (point-min)))
1550       (error (progn
1551                (set-window-configuration
1552                 what-character-original-window-configuration)
1553                (signal (car err) (cdr err)))))))
1554
1555
1556 ;;; @ end
1557 ;;;
1558
1559 (provide 'char-db-util)
1560
1561 ;;; char-db-util.el ends here