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 ;;   2008,2009,2010,2011,2012,2013,2014,2015,2016 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     =hanyo-denshi/tk
 172     ;; ==mj
 173     ;; ==adobe-japan1-0
 174     ;; ==adobe-japan1-1
 175     ;; ==adobe-japan1-2
 176     ;; ==adobe-japan1-3
 177     ;; ==adobe-japan1-4
 178     ;; ==adobe-japan1-5
 179     ;; ==adobe-japan1-6
 180     ==jis-x0208
 181     ==jis-x0213-1
 182     ==jis-x0213-2
 183     ==jis-x0212
 184     ==hanyo-denshi/ja
 185     ==hanyo-denshi/jb
 186     ==hanyo-denshi/jc
 187     ==hanyo-denshi/ft
 188     ==hanyo-denshi/ia
 189     ==hanyo-denshi/ib
 190     ==hanyo-denshi/hg
 191     ==hanyo-denshi/jt
 192     ==hanyo-denshi/ks
 193     =gt
 194     =gt-k
 195     =daikanwa
 196     =daikanwa@rev2
 197     =daikanwa@rev1
 198     =daikanwa/+p
 199     ==daikanwa
 200     =cbeta
 201     ideograph-hanziku-1
 202     ideograph-hanziku-2
 203     ideograph-hanziku-3
 204     ideograph-hanziku-4
 205     ideograph-hanziku-5
 206     ideograph-hanziku-6
 207     ideograph-hanziku-7
 208     ideograph-hanziku-8
 209     ideograph-hanziku-9
 210     ideograph-hanziku-10
 211     ideograph-hanziku-11
 212     ideograph-hanziku-12
 213     ;; =>>>adobe-japan1-0
 214     ;; =>>>adobe-japan1-1
 215     ;; =>>>adobe-japan1-2
 216     ;; =>>>adobe-japan1-3
 217     ;; =>>>adobe-japan1-4
 218     ;; =>>>adobe-japan1-5
 219     ;; =>>>adobe-japan1-6
 220     ;; =>>>jis-x0208
 221     ;; =>>>jis-x0213-1
 222     ;; =>>>jis-x0213-2
 223     ;; =>>>hanyo-denshi/ja
 224     ;; =>>>hanyo-denshi/jb
 225     ;; =>>>hanyo-denshi/jc
 226     ;; =>>>hanyo-denshi/ft
 227     ;; =>>>hanyo-denshi/ib
 228     ;; =>>>hanyo-denshi/hg
 229     ;; =>>>hanyo-denshi/jt
 230     ;; =>>>hanyo-denshi/ks
 231     ;; =>>>gt
 232     =>>adobe-japan1-0
 233     =>>adobe-japan1-1
 234     =>>adobe-japan1-2
 235     =>>adobe-japan1-3
 236     =>>adobe-japan1-4
 237     =>>adobe-japan1-5
 238     =>>adobe-japan1-6
 239     =>>jis-x0208
 240     =>>jis-x0213-1
 241     =>>jis-x0213-1@2000
 242     =>>jis-x0213-1@2004
 243     =>>jis-x0213-2
 244     =>>jis-x0208@1978
 245     =>>hanyo-denshi/ft
 246     =>>hanyo-denshi/jt
 247     =>>hanyo-denshi/ks
 248     =>>gt
 249     =>>daikanwa
 250     =>>cbeta
 251     =+>jis-x0208
 252     =+>jis-x0213-1
 253     =+>jis-x0213-2
 254     =+>adobe-japan1-0
 255     =+>adobe-japan1-1
 256     =+>adobe-japan1-2
 257     =+>adobe-japan1-3
 258     =+>adobe-japan1-4
 259     =+>adobe-japan1-5
 260     =+>adobe-japan1-6
 261     =+>jis-x0208@1978
 262     =>jis-x0208
 263     =>jis-x0208@1997
 264     =>jis-x0213-1
 265     =>jis-x0213-1@2000
 266     =>jis-x0213-1@2004
 267     =>jis-x0213-2
 268     ==>ucs@bucs
 269     =>iwds-1
 270     =>ucs@hanyo-denshi
 271     =>ucs@iso
 272     =>ucs@unicode
 273     =>ucs@jis
 274     =>ucs@cns
 275     =>ucs@ks
 276     =+>ucs@iso
 277     =+>ucs@unicode
 278     =+>ucs@jis
 279     =+>ucs@jis/1990
 280     =+>ucs@cns
 281     =+>ucs@ks
 282     =>>ucs@iso
 283     =>>ucs@unicode
 284     =>>ucs@jis
 285     =>>ucs@cns
 286     =>>>ucs@iso
 287     =>>>ucs@unicode
 288     ==ucs@iso
 289     ==ucs@unicode
 290     ;; ==ucs@cns
 291     ==gb2312
 292     ==ks-x1001
 293     ==cns11643-1
 294     ==cns11643-2
 295     ==cns11643-3
 296     ==cns11643-4
 297     ==cns11643-5
 298     ==cns11643-6
 299     ==cns11643-7
 300     ==gt
 301     ==jis-x0208@1990
 302     ;; ==jis-x0208@1983
 303     ==jis-x0208@1978
 304     ==gt-k
 305     =ucs@iso
 306     =ucs@unicode
 307     =ucs@cns
 308     ==big5-cdp
 309     ==cbeta
 310     =>>big5-cdp
 311     =>>gt-k
 312     =+>gt
 313     =>gt
 314     =>mj
 315     =>big5-cdp
 316     =>daikanwa
 317     =>daikanwa/ho
 318     =>cns11643-5
 319     =>cns11643-7
 320     =big5
 321     =big5-eten
 322     =>gt-k
 323     =zinbun-oracle
 324     =>zinbun-oracle
 325     =ruimoku-v6
 326     =>>ruimoku-v6
 327     ==ruimoku-v6
 328     =jef-china3
 329     =>cbeta
 330     =shinjigen
 331     =ucs-var-001
 332     =ucs-var-002
 333     =ucs-var-003
 334     =ucs-var-004
 335     =ucs-var-006
 336     =ucs-var-008
 337     =ucs-var-010
 338     =ucs-itaiji-001
 339     =ucs-itaiji-002
 340     =ucs-itaiji-003
 341     =ucs-itaiji-004
 342     =ucs-itaiji-005
 343     =ucs-itaiji-006
 344     =ucs-itaiji-008
 345     =ucs-itaiji-009
 346     =>big5-cdp-itaiji-001
 347     =big5-cdp-itaiji-001
 348     =big5-cdp-itaiji-002
 349     =big5-cdp-itaiji-003
 350     =big5-cdp-var-003
 351     =big5-cdp-var-004
 352     =big5-cdp-var-005
 353     =>ucs@iwds-1
 354     ==ucs@gb
 355     =ucs@JP/hanazono))
 356
 357
 358 ;;; @ char-db formatters
 359 ;;;
 360
 361 (defun char-db-make-char-spec (char)
 362   (let (ret char-spec)
 363     (cond ((characterp char)
 364            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 365                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 366                   (setq char-spec (list (cons '=ucs ret)))
 367                   (cond ((setq ret (get-char-attribute char 'name))
 368                          (setq char-spec (cons (cons 'name ret) char-spec))
 369                          )
 370                         ((setq ret (get-char-attribute char 'name*))
 371                          (setq char-spec (cons (cons 'name* ret) char-spec))
 372                          ))
 373                   )
 374                  ((encode-char char '=mj 'defined-only)
 375                   (setq char-spec nil)
 376                   (dolist (ccs (charset-list))
 377                     (if (and (or (eq ccs '=mj)
 378                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 379                                  (string-match "=ucs@" (symbol-name ccs))
 380                                  )
 381                              (setq ccs (charset-name ccs))
 382                              (null (assq ccs char-spec))
 383                              (setq ret (encode-char char ccs 'defined-only)))
 384                         (setq char-spec (cons (cons ccs ret) char-spec))))
 385                   )
 386                  ((encode-char char '==mj 'defined-only)
 387                   (setq char-spec nil)
 388                   (dolist (ccs (charset-list))
 389                     (if (and (or (eq ccs '==mj)
 390                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 391                                  (string-match "=ucs@" (symbol-name ccs))
 392                                  )
 393                              (setq ccs (charset-name ccs))
 394                              (null (assq ccs char-spec))
 395                              (setq ret (encode-char char ccs 'defined-only)))
 396                         (setq char-spec (cons (cons ccs ret) char-spec))))
 397                   )
 398                  ((encode-char char '=adobe-japan1 'defined-only)
 399                   (setq char-spec nil)
 400                   (dolist (ccs (charset-list))
 401                     (if (and (or (memq ccs
 402                                        '(=adobe-japan1-0
 403                                          =adobe-japan1-1
 404                                          =adobe-japan1-2
 405                                          =adobe-japan1-3
 406                                          =adobe-japan1-4
 407                                          =adobe-japan1-5
 408                                          =adobe-japan1-6
 409                                          ))
 410                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 411                                  (string-match "=ucs@" (symbol-name ccs))
 412                                  )
 413                              (setq ccs (charset-name ccs))
 414                              (null (assq ccs char-spec))
 415                              (setq ret (encode-char char ccs 'defined-only)))
 416                         (setq char-spec (cons (cons ccs ret) char-spec))))
 417                   )
 418                  ((encode-char char '==adobe-japan1 'defined-only)
 419                   (setq char-spec nil)
 420                   (dolist (ccs (charset-list))
 421                     (if (and (or (memq ccs
 422                                        '(==adobe-japan1-0
 423                                          ==adobe-japan1-1
 424                                          ==adobe-japan1-2
 425                                          ==adobe-japan1-3
 426                                          ==adobe-japan1-4
 427                                          ==adobe-japan1-5
 428                                          ==adobe-japan1-6
 429                                          ))
 430                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 431                                  (string-match "=ucs@" (symbol-name ccs))
 432                                  )
 433                              (setq ccs (charset-name ccs))
 434                              (null (assq ccs char-spec))
 435                              (setq ret (encode-char char ccs 'defined-only)))
 436                         (setq char-spec (cons (cons ccs ret) char-spec))))
 437                   )
 438                  ((setq ret
 439                         (catch 'tag
 440                           (let ((rest char-db-coded-charset-priority-list)
 441                                 ccs)
 442                             (while rest
 443                               (setq ccs (charset-name
 444                                          (find-charset (car rest))))
 445                               (if (setq ret
 446                                         (encode-char char ccs
 447                                                      'defined-only))
 448                                   (throw 'tag (cons ccs ret)))
 449                               (setq rest (cdr rest))))))
 450                   (setq char-spec (list ret))
 451                   (dolist (ccs (delq (car ret) (charset-list)))
 452                     (if (and (or (charset-iso-final-char ccs)
 453                                  (memq ccs
 454                                        '(=daikanwa
 455                                          =daikanwa@rev2
 456                                          ;; =gt-k
 457                                          =jis-x0208@1997
 458                                          ))
 459                                  (eq (charset-property ccs 'iso-ir) 177)
 460                                  ;; (string-match "=ucs@" (symbol-name ccs))
 461                                  )
 462                              (setq ccs (charset-name ccs))
 463                              (null (assq ccs char-spec))
 464                              (setq ret (encode-char char ccs 'defined-only)))
 465                         (setq char-spec (cons (cons ccs ret) char-spec))))
 466                   (if (null char-spec)
 467                       (setq char-spec (split-char char)))
 468                   (cond ((setq ret (get-char-attribute char 'name))
 469                          (setq char-spec (cons (cons 'name ret) char-spec))
 470                          )
 471                         ((setq ret (get-char-attribute char 'name*))
 472                          (setq char-spec (cons (cons 'name* ret) char-spec))
 473                          ))
 474                   )
 475                  ((setq ret (get-char-attribute
 476                              char 'ideographic-combination))
 477                   (setq char-spec
 478                         (cons (cons 'ideographic-combination ret)
 479                               char-spec))
 480                   ))
 481            char-spec)
 482           ((consp char)
 483            char))))
 484
 485 (defun char-db-insert-char-spec (char &optional readable column
 486                                       required-features)
 487   (unless column
 488     (setq column (current-column)))
 489   (let (char-spec temp-char)
 490     (setq char-spec (char-db-make-char-spec char))
 491     (unless (or (characterp char) ; char
 492                 (condition-case nil
 493                     (setq char (find-char char-spec))
 494                   (error nil)))
 495       ;; define temporary character
 496       ;;   Current implementation is dirty.
 497       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 498                                          char-spec)))
 499       (remove-char-attribute temp-char 'ideograph-daikanwa)
 500       (setq char temp-char))
 501     (insert-char-attributes char
 502                             readable
 503                             (union (mapcar #'car char-spec)
 504                                    required-features)
 505                             nil 'for-sub-node)
 506     (when temp-char
 507       ;; undefine temporary character
 508       ;;   Current implementation is dirty.
 509       (setq char-spec (char-attribute-alist temp-char))
 510       (while char-spec
 511         (remove-char-attribute temp-char (car (car char-spec)))
 512         (setq char-spec (cdr char-spec))))))
 513
 514 (defun char-db-insert-alist (alist &optional readable column)
 515   (unless column
 516     (setq column (current-column)))
 517   (let ((line-breaking
 518          (concat "\n" (make-string (1+ column) ?\ )))
 519         name value
 520         ret al ; cal
 521         key
 522         lbs cell rest separator)
 523     (insert "(")
 524     (while alist
 525       (setq name (car (car alist))
 526             value (cdr (car alist)))
 527       (cond ((eq name 'char)
 528              (insert "(char . ")
 529              (if (setq ret (condition-case nil
 530                                (find-char value)
 531                              (error nil)))
 532                  (progn
 533                    (setq al nil
 534                          ;; cal nil
 535                          )
 536                    (while value
 537                      (setq key (car (car value)))
 538                      ;; (if (find-charset key)
 539                      ;;     (setq cal (cons key cal))
 540                      (setq al (cons key al))
 541                      ;; )
 542                      (setq value (cdr value)))
 543                    (insert-char-attributes ret
 544                                            readable
 545                                            (or al 'none) ; cal
 546                                            nil 'for-sub-node))
 547                (insert (prin1-to-string value)))
 548              (insert ")")
 549              (insert line-breaking))
 550             ((consp value)
 551              (insert (format "(%-18s " name))
 552              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 553              (while (consp value)
 554                (setq cell (car value))
 555                (if (and (consp cell)
 556                         (consp (car cell))
 557                         (setq ret (condition-case nil
 558                                       (find-char cell)
 559                                     (error nil)))
 560                         )
 561                    (progn
 562                      (setq rest cell
 563                            al nil
 564                            ;; cal nil
 565                            )
 566                      (while rest
 567                        (setq key (car (car rest)))
 568                        ;; (if (find-charset key)
 569                        ;;     (setq cal (cons key cal))
 570                        (setq al (cons key al))
 571                        ;; )
 572                        (setq rest (cdr rest)))
 573                      (if separator
 574                          (insert lbs))
 575                      (insert-char-attributes ret
 576                                              readable
 577                                              al ; cal
 578                                              nil 'for-sub-node)
 579                      (setq separator lbs))
 580                  (if separator
 581                      (insert separator))
 582                  (insert (prin1-to-string cell))
 583                  (setq separator " "))
 584                (setq value (cdr value)))
 585              (insert ")")
 586              (insert line-breaking))
 587             (t
 588              (insert (format "(%-18s . %S)%s"
 589                              name value
 590                              line-breaking))))
 591       (setq alist (cdr alist))))
 592   (insert ")"))
 593
 594 (defun char-db-insert-char-reference (plist &optional readable column)
 595   (unless column
 596     (setq column (current-column)))
 597   (let ((line-breaking
 598          (concat "\n" (make-string (1+ column) ?\ )))
 599         (separator "")
 600         name value)
 601     (insert "(")
 602     (while plist
 603       (setq name (pop plist))
 604       (setq value (pop plist))
 605       (cond ((eq name :char)
 606              (insert separator)
 607              (insert ":char\t")
 608              (cond ((numberp value)
 609                     (setq value (decode-char '=ucs value)))
 610                    ;; ((consp value)
 611                    ;;  (setq value (or (find-char value)
 612                    ;;                  value)))
 613                    )
 614              (char-db-insert-char-spec value readable)
 615              (insert line-breaking)
 616              (setq separator ""))
 617             ((eq name :radical)
 618              (insert (format "%s%s\t%d ; %c%s"
 619                              separator
 620                              name value
 621                              (ideographic-radical value)
 622                              line-breaking))
 623              (setq separator ""))
 624             (t
 625              (insert (format "%s%s\t%S" separator name value))
 626              (setq separator line-breaking)))
 627       ))
 628   (insert ")"))
 629
 630 (defun char-db-decode-isolated-char (ccs code-point)
 631   (let (ret)
 632     (setq ret
 633           (cond ((eq ccs 'arabic-iso8859-6)
 634                  (decode-char ccs code-point))
 635                 ;; ((eq ccs '=gt)
 636                 ;;  (decode-builtin-char '==gt code-point))
 637                 ((and (memq ccs '(=gt-pj-1
 638                                   =gt-pj-2
 639                                   =gt-pj-3
 640                                   =gt-pj-4
 641                                   =gt-pj-5
 642                                   =gt-pj-6
 643                                   =gt-pj-7
 644                                   =gt-pj-8
 645                                   =gt-pj-9
 646                                   =gt-pj-10
 647                                   =gt-pj-11))
 648                       (setq ret (decode-char ccs code-point))
 649                       (setq ret (encode-char ret '=gt 'defined-only)))
 650                  (decode-builtin-char '=gt ret))
 651                 (t
 652                  (decode-builtin-char ccs code-point))))
 653     (cond ((null ret)
 654            (or (decode-char ccs code-point)
 655                (define-char (list (cons ccs code-point)))))
 656           ((and (<= 0 (char-int ret))
 657                 (<= (char-int ret) #x1F))
 658            (decode-char '=ucs (+ #x2400 (char-int ret))))
 659           ((= (char-int ret) #x7F)
 660            ?\u2421)
 661           (t ret))))
 662
 663 (defvar char-db-convert-obsolete-format t)
 664
 665 (defun char-db-insert-ccs-feature (name value line-breaking)
 666   (cond
 667    ((integerp value)
 668     (insert
 669      (format
 670       (cond
 671        ((memq name '(=>iwds-1
 672                      ==shinjigen
 673                      =shinjigen
 674                      =shinjigen@1ed ==shinjigen@1ed
 675                      =shinjigen@rev ==shinjigen@rev
 676                      =shinjigen/+p@rev ==shinjigen/+p@rev
 677                      ===daikanwa/ho ==daikanwa/ho
 678                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 679         "(%-18s .  %04d)\t; %c")
 680        ((eq name '=shinjigen@1ed/24pr)
 681         "(%-18s . %04d)\t; %c")
 682        ((or
 683          (memq name
 684                '(===daikanwa
 685                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 686                  =daikanwa@rev1 =daikanwa@rev2
 687                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 688                  =>>daikanwa/+p
 689                  =daikanwa/+2p =>>daikanwa/+2p
 690                  =gt ==gt ===gt
 691                  =>>gt =+>gt =>gt
 692                  =gt-k ==gt-k ===gt-k
 693                  =>>gt-k =>gt-k
 694                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 695                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 696                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 697                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 698                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 699                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 700                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 701                  =>>adobe-japan1-0 =+>adobe-japan1-0
 702                  =>>adobe-japan1-1 =+>adobe-japan1-1
 703                  =>>adobe-japan1-2 =+>adobe-japan1-2
 704                  =>>adobe-japan1-3 =+>adobe-japan1-3
 705                  =>>adobe-japan1-4 =+>adobe-japan1-4
 706                  =>>adobe-japan1-5 =+>adobe-japan1-5
 707                  =>>adobe-japan1-6 =+>adobe-japan1-6
 708                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 709                  =zinbun-oracle =>zinbun-oracle
 710                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 711                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 712                  ===hng-khi ===hng-khm ===hng-hok ===hng-kyd ===hng-sok
 713                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 714                  ===hng-sys ===hng-tsu ===hng-tzj
 715                  ===hng-hos ===hng-nak ===hng-jhk
 716                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 717                  ===hng-nkm ===hng-k24 ===hng-nkk
 718                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 719                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 720                  ===hng-okd ===hng-wad ===hng-kmi
 721                  ===hng-zkd ===hng-doh ===hng-jyu
 722                  ===hng-tzs ===hng-kss ===hng-kyo
 723                  ===hng-smk))
 724          ;; (string-match "^=adobe-" (symbol-name name))
 725          )
 726         "(%-18s . %05d)\t; %c")
 727        ((memq name '(=hanyo-denshi/ks
 728                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 729                      =>>hanyo-denshi/ks
 730                      =koseki ==koseki
 731                      =mj ==mj ===mj =>>mj =>mj
 732                      =zihai mojikyo))
 733         "(%-18s . %06d)\t; %c")
 734        ((memq name '(=hanyo-denshi/tk ==hanyo-denshi/tk))
 735         "(%-18s . %08d)\t; %c")
 736        ((>= (charset-dimension name) 2)
 737         "(%-18s . #x%04X)\t; %c")
 738        (t
 739         "(%-18s . #x%02X)\t; %c"))
 740       name
 741       (if (= (charset-iso-graphic-plane name) 1)
 742           (logior value
 743                   (cond ((= (charset-dimension name) 1)
 744                          #x80)
 745                         ((= (charset-dimension name) 2)
 746                          #x8080)
 747                         ((= (charset-dimension name) 3)
 748                          #x808080)
 749                         (t 0)))
 750         value)
 751       (char-db-decode-isolated-char name value)))
 752     (if (and (= (charset-chars name) 94)
 753              (= (charset-dimension name) 2))
 754         (insert (format " [%02d-%02d]"
 755                         (- (lsh value -8) 32)
 756                         (- (logand value 255) 32))))
 757     )
 758    (t
 759     (insert (format "(%-18s . %s)" name value))
 760     ))
 761   (insert line-breaking))
 762
 763 (defun char-db-insert-relation-feature (char name value line-breaking
 764                                              ccss readable)
 765   (insert (format "(%-18s%s " name line-breaking))
 766   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 767         separator cell sources required-features
 768         ret)
 769     (while (consp value)
 770       (setq cell (car value))
 771       (if (integerp cell)
 772           (setq cell (decode-char '=ucs cell)))
 773       (cond
 774        ((eq name '->subsumptive)
 775         (when (or (not (some (lambda (atr)
 776                                (get-char-attribute cell atr))
 777                              char-db-ignored-attributes))
 778                   (some (lambda (ccs)
 779                           (encode-char cell ccs 'defined-only))
 780                         ccss))
 781           (if separator
 782               (insert lbs))
 783           (let ((char-db-ignored-attributes
 784                  (cons '<-subsumptive
 785                        char-db-ignored-attributes)))
 786             (insert-char-attributes cell readable nil nil 'for-sub-node))
 787           (setq separator lbs))
 788         )
 789        ((characterp cell)
 790         (setq sources
 791               (get-char-attribute
 792                char (intern (format "%s*sources" name))))
 793         (setq required-features nil)
 794         (dolist (source sources)
 795           (cond
 796            ((memq source '(JP
 797                            JP/Jouyou
 798                            shinjigen shinjigen@1ed shinjigen@rev))
 799             (setq required-features
 800                   (union required-features
 801                          '(=jis-x0208
 802                            =jis-x0208@1990
 803                            =jis-x0213-1@2000
 804                            =jis-x0213-1@2004
 805                            =jis-x0213-2
 806                            =jis-x0212
 807                            =jis-x0208@1983
 808                            =jis-x0208@1978
 809                            =shinjigen
 810                            =shinjigen@1ed
 811                            =shinjigen@rev
 812                            =shinjigen/+p@rev))))
 813            ((eq source 'CN)
 814             (setq required-features
 815                   (union required-features
 816                          '(=gb2312
 817                            =gb12345
 818                            =iso-ir165)))))
 819           (cond
 820            ((find-charset (setq ret (intern (format "=%s" source))))
 821             (setq required-features
 822                   (cons ret required-features)))
 823            (t (setq required-features
 824                     (cons source required-features)))))
 825         (cond ((string-match "@JP" (symbol-name name))
 826                (setq required-features
 827                      (union required-features
 828                             '(=jis-x0208
 829                               =jis-x0208@1990
 830                               =jis-x0213-1-2000
 831                               =jis-x0213-2-2000
 832                               =jis-x0212
 833                               =jis-x0208@1983
 834                               =jis-x0208@1978))))
 835               ((string-match "@CN" (symbol-name name))
 836                (setq required-features
 837                      (union required-features
 838                             '(=gb2312
 839                               =gb12345
 840                               =iso-ir165)))))
 841         (if separator
 842             (insert lbs))
 843         (if readable
 844             (insert (format "%S" cell))
 845           (char-db-insert-char-spec cell readable
 846                                     nil
 847                                     required-features))
 848         (setq separator lbs))
 849        ((consp cell)
 850         (if separator
 851             (insert lbs))
 852         (if (consp (car cell))
 853             (char-db-insert-char-spec cell readable)
 854           (char-db-insert-char-reference cell readable))
 855         (setq separator lbs))
 856        (t
 857         (if separator
 858             (insert separator))
 859         (insert (prin1-to-string cell))
 860         (setq separator " ")))
 861       (setq value (cdr value)))
 862     (insert ")")
 863     (insert line-breaking)))
 864
 865 (defun insert-char-attributes (char &optional readable attributes column
 866                                     for-sub-node)
 867   (unless column
 868     (setq column (current-column)))
 869   (let (name value ; has-long-ccs-name
 870         rest
 871         radical strokes
 872         (line-breaking
 873          (concat "\n" (make-string (1+ column) ?\ )))
 874         lbs cell separator ret
 875         key al cal
 876         dest-ccss ; sources required-features
 877         ccss)
 878     (let (atr-d)
 879       (setq attributes
 880             (sort (if attributes
 881                       (if (consp attributes)
 882                           (progn
 883                             (dolist (name attributes)
 884                               (unless (memq name char-db-ignored-attributes)
 885                                 (if (find-charset name)
 886                                     (push name ccss))
 887                                 (push name atr-d)))
 888                             atr-d))
 889                     (dolist (name (char-attribute-list))
 890                       (unless (memq name char-db-ignored-attributes)
 891                         (if (find-charset name)
 892                             (push name ccss))
 893                         (push name atr-d)))
 894                     atr-d)
 895                   #'char-attribute-name<)))
 896     (insert "(")
 897     (when (memq '<-subsumptive attributes)
 898       (when (or readable (not for-sub-node))
 899         (when (setq value (get-char-attribute char '<-subsumptive))
 900           (char-db-insert-relation-feature char '<-subsumptive value
 901                                            line-breaking
 902                                            ccss readable)))
 903       (setq attributes (delq '<-subsumptive attributes)))
 904     (when (and (memq '<-denotational attributes)
 905                (setq value (get-char-attribute char '<-denotational)))
 906       (char-db-insert-relation-feature char '<-denotational value
 907                                        line-breaking
 908                                        ccss readable)
 909       (setq attributes (delq '<-denotational attributes)))
 910     (when (and (memq '<-denotational@component attributes)
 911                (setq value (get-char-attribute char '<-denotational@component)))
 912       (char-db-insert-relation-feature char '<-denotational@component value
 913                                        line-breaking
 914                                        ccss readable)
 915       (setq attributes (delq '<-denotational@component attributes)))
 916     (when (and (memq 'name attributes)
 917                (setq value (get-char-attribute char 'name)))
 918       (insert (format
 919                (if (> (+ (current-column) (length value)) 48)
 920                    "(name . %S)%s"
 921                  "(name               . %S)%s")
 922                value line-breaking))
 923       (setq attributes (delq 'name attributes))
 924       )
 925     (when (and (memq 'name* attributes)
 926                (setq value (get-char-attribute char 'name*)))
 927       (insert (format
 928                (if (> (+ (current-column) (length value)) 48)
 929                    "(name* . %S)%s"
 930                  "(name*              . %S)%s")
 931                value line-breaking))
 932       (setq attributes (delq 'name* attributes))
 933       )
 934     (when (and (memq 'script attributes)
 935                (setq value (get-char-attribute char 'script)))
 936       (insert (format "(script\t\t%s)%s"
 937                       (mapconcat (function prin1-to-string)
 938                                  value " ")
 939                       line-breaking))
 940       (setq attributes (delq 'script attributes))
 941       )
 942     (dolist (name '(=>ucs =>ucs*))
 943       (when (and (memq name attributes)
 944                  (setq value (get-char-attribute char name)))
 945         (insert (format "(%-18s . #x%04X)\t; %c%s"
 946                         name value (decode-char '=ucs value)
 947                         line-breaking))
 948         (setq attributes (delq name attributes))))
 949     (dolist (name '(=>ucs@gb =>ucs@big5))
 950       (when (and (memq name attributes)
 951                  (setq value (get-char-attribute char name)))
 952         (insert (format "(%-18s . #x%04X)\t; %c%s"
 953                         name value
 954                         (decode-char (intern
 955                                       (concat "="
 956                                               (substring
 957                                                (symbol-name name) 2)))
 958                                      value)
 959                         line-breaking))
 960         (setq attributes (delq name attributes))
 961         ))
 962     ;; (dolist (name '(=>daikanwa))
 963     ;;   (when (and (memq name attributes)
 964     ;;              (setq value (get-char-attribute char name)))
 965     ;;     (insert
 966     ;;      (if (integerp value)
 967     ;;          (format "(%-18s . %05d)\t; %c%s"
 968     ;;                  name value (decode-char '=daikanwa value)
 969     ;;                  line-breaking)
 970     ;;        (format "(%-18s %s)\t; %c%s"
 971     ;;                name
 972     ;;                (mapconcat (function prin1-to-string)
 973     ;;                           value " ")
 974     ;;                (char-representative-of-daikanwa char)
 975     ;;                line-breaking)))
 976     ;;     (setq attributes (delq name attributes))))
 977     (when (and (memq 'general-category attributes)
 978                (setq value (get-char-attribute char 'general-category)))
 979       (insert (format
 980                "(general-category\t%s) ; %s%s"
 981                (mapconcat (lambda (cell)
 982                             (format "%S" cell))
 983                           value " ")
 984                (cond ((rassoc value unidata-normative-category-alist)
 985                       "Normative Category")
 986                      ((rassoc value unidata-informative-category-alist)
 987                       "Informative Category")
 988                      (t
 989                       "Unknown Category"))
 990                line-breaking))
 991       (setq attributes (delq 'general-category attributes))
 992       )
 993     (when (and (memq 'bidi-category attributes)
 994                (setq value (get-char-attribute char 'bidi-category)))
 995       (insert (format "(bidi-category\t. %S)%s"
 996                       value
 997                       line-breaking))
 998       (setq attributes (delq 'bidi-category attributes))
 999       )
1000     (unless (or (not (memq 'mirrored attributes))
1001                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
1002                     'empty))
1003       (insert (format "(mirrored\t\t. %S)%s"
1004                       value
1005                       line-breaking))
1006       (setq attributes (delq 'mirrored attributes))
1007       )
1008     (cond
1009      ((and (memq 'decimal-digit-value attributes)
1010            (setq value (get-char-attribute char 'decimal-digit-value)))
1011       (insert (format "(decimal-digit-value . %S)%s"
1012                       value
1013                       line-breaking))
1014       (setq attributes (delq 'decimal-digit-value attributes))
1015       (when (and (memq 'digit-value attributes)
1016                  (setq value (get-char-attribute char 'digit-value)))
1017         (insert (format "(digit-value\t . %S)%s"
1018                         value
1019                         line-breaking))
1020         (setq attributes (delq 'digit-value attributes))
1021         )
1022       (when (and (memq 'numeric-value attributes)
1023                  (setq value (get-char-attribute char 'numeric-value)))
1024         (insert (format "(numeric-value\t . %S)%s"
1025                         value
1026                         line-breaking))
1027         (setq attributes (delq 'numeric-value attributes))
1028         )
1029       )
1030      (t
1031       (when (and (memq 'digit-value attributes)
1032                  (setq value (get-char-attribute char 'digit-value)))
1033         (insert (format "(digit-value\t. %S)%s"
1034                         value
1035                         line-breaking))
1036         (setq attributes (delq 'digit-value attributes))
1037         )
1038       (when (and (memq 'numeric-value attributes)
1039                  (setq value (get-char-attribute char 'numeric-value)))
1040         (insert (format "(numeric-value\t. %S)%s"
1041                         value
1042                         line-breaking))
1043         (setq attributes (delq 'numeric-value attributes))
1044         )))
1045     (when (and (memq 'iso-10646-comment attributes)
1046                (setq value (get-char-attribute char 'iso-10646-comment)))
1047       (insert (format "(iso-10646-comment\t. %S)%s"
1048                       value
1049                       line-breaking))
1050       (setq attributes (delq 'iso-10646-comment attributes))
1051       )
1052     (when (and (memq 'morohashi-daikanwa attributes)
1053                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1054       (insert (format "(morohashi-daikanwa\t%s)%s"
1055                       (mapconcat (function prin1-to-string) value " ")
1056                       line-breaking))
1057       (setq attributes (delq 'morohashi-daikanwa attributes))
1058       )
1059     (setq radical nil
1060           strokes nil)
1061     (when (and (memq 'ideographic-radical attributes)
1062                (setq value (get-char-attribute char 'ideographic-radical)))
1063       (setq radical value)
1064       (insert (format "(ideographic-radical . %S)\t; %c%s"
1065                       radical
1066                       (ideographic-radical radical)
1067                       line-breaking))
1068       (setq attributes (delq 'ideographic-radical attributes))
1069       )
1070     (when (and (memq 'shuowen-radical attributes)
1071                (setq value (get-char-attribute char 'shuowen-radical)))
1072       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1073                       value
1074                       (shuowen-radical value)
1075                       line-breaking))
1076       (setq attributes (delq 'shuowen-radical attributes))
1077       )
1078     (let (key)
1079       (dolist (domain
1080                (append
1081                 char-db-feature-domains
1082                 (let (dest domain)
1083                   (dolist (feature (char-attribute-list))
1084                     (setq feature (symbol-name feature))
1085                     (when (string-match
1086                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1087                            feature)
1088                       (setq domain (intern (match-string 2 feature)))
1089                      (unless (memq domain dest)
1090                        (setq dest (cons domain dest)))))
1091                   (sort dest #'string<))))
1092         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1093         (when (and (memq key attributes)
1094                    (setq value (get-char-attribute char key)))
1095           (setq radical value)
1096           (insert (format "(%s . %S)\t; %c%s"
1097                           key
1098                           radical
1099                           (ideographic-radical radical)
1100                           line-breaking))
1101           (setq attributes (delq key attributes))
1102           )
1103         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1104         (when (and (memq key attributes)
1105                    (setq value (get-char-attribute char key)))
1106           (setq strokes value)
1107           (insert (format "(%s . %S)%s"
1108                           key
1109                           strokes
1110                           line-breaking))
1111           (setq attributes (delq key attributes))
1112           )
1113         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1114         (when (and (memq key attributes)
1115                    (setq value (get-char-attribute char key)))
1116           (insert (format "(%s       . %S)%s"
1117                           key
1118                           value
1119                           line-breaking))
1120           (setq attributes (delq key attributes))
1121           )
1122         (dolist (feature '(ideographic-radical
1123                            ideographic-strokes
1124                            total-strokes))
1125           (setq key (intern (format "%s@%s*sources" feature domain)))
1126           (when (and (memq key attributes)
1127                      (setq value (get-char-attribute char key)))
1128             (insert (format "(%s%s" key line-breaking))
1129             (dolist (cell value)
1130               (insert (format " %s" cell)))
1131             (insert ")")
1132             (insert line-breaking)
1133             (setq attributes (delq key attributes))
1134             ))
1135         ))
1136     (when (and (memq 'ideographic-strokes attributes)
1137                (setq value (get-char-attribute char 'ideographic-strokes)))
1138       (setq strokes value)
1139       (insert (format "(ideographic-strokes . %S)%s"
1140                       strokes
1141                       line-breaking))
1142       (setq attributes (delq 'ideographic-strokes attributes))
1143       )
1144     (when (and (memq 'kangxi-radical attributes)
1145                (setq value (get-char-attribute char 'kangxi-radical)))
1146       (unless (eq value radical)
1147         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1148                         value
1149                         (ideographic-radical value)
1150                         line-breaking))
1151         (or radical
1152             (setq radical value)))
1153       (setq attributes (delq 'kangxi-radical attributes))
1154       )
1155     (when (and (memq 'kangxi-strokes attributes)
1156                (setq value (get-char-attribute char 'kangxi-strokes)))
1157       (unless (eq value strokes)
1158         (insert (format "(kangxi-strokes\t . %S)%s"
1159                         value
1160                         line-breaking))
1161         (or strokes
1162             (setq strokes value)))
1163       (setq attributes (delq 'kangxi-strokes attributes))
1164       )
1165     ;; (when (and (memq 'japanese-radical attributes)
1166     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1167     ;;   (unless (eq value radical)
1168     ;;     (insert (format "(japanese-radical\t . %S)\t; %c%s"
1169     ;;                     value
1170     ;;                     (ideographic-radical value)
1171     ;;                     line-breaking))
1172     ;;     (or radical
1173     ;;         (setq radical value)))
1174     ;;   (setq attributes (delq 'japanese-radical attributes))
1175     ;;   )
1176     (when (and (memq 'japanese-strokes attributes)
1177                (setq value (get-char-attribute char 'japanese-strokes)))
1178       (unless (eq value strokes)
1179         (insert (format "(japanese-strokes\t . %S)%s"
1180                         value
1181                         line-breaking))
1182         (or strokes
1183             (setq strokes value)))
1184       (setq attributes (delq 'japanese-strokes attributes))
1185       )
1186     (when (and (memq 'cns-radical attributes)
1187                (setq value (get-char-attribute char 'cns-radical)))
1188       (insert (format "(cns-radical\t . %S)\t; %c%s"
1189                       value
1190                       (ideographic-radical value)
1191                       line-breaking))
1192       (setq attributes (delq 'cns-radical attributes))
1193       )
1194     (when (and (memq 'cns-strokes attributes)
1195                (setq value (get-char-attribute char 'cns-strokes)))
1196       (unless (eq value strokes)
1197         (insert (format "(cns-strokes\t . %S)%s"
1198                         value
1199                         line-breaking))
1200         (or strokes
1201             (setq strokes value)))
1202       (setq attributes (delq 'cns-strokes attributes))
1203       )
1204     ;; (when (and (memq 'shinjigen-1-radical attributes)
1205     ;;            (setq value (get-char-attribute char 'shinjigen-1-radical)))
1206     ;;   (unless (eq value radical)
1207     ;;     (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
1208     ;;                     value
1209     ;;                     (ideographic-radical value)
1210     ;;                     line-breaking))
1211     ;;     (or radical
1212     ;;         (setq radical value)))
1213     ;;   (setq attributes (delq 'shinjigen-1-radical attributes))
1214     ;;   )
1215     ;; (when (and (memq 'ideographic- attributes)
1216     ;;            (setq value (get-char-attribute char 'ideographic-)))
1217     ;;   (insert "(ideographic-       ")
1218     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1219     ;;         separator nil)
1220     ;;   (while (consp value)
1221     ;;     (setq cell (car value))
1222     ;;     (if (integerp cell)
1223     ;;         (setq cell (decode-char '=ucs cell)))
1224     ;;     (cond ((characterp cell)
1225     ;;            (if separator
1226     ;;                (insert lbs))
1227     ;;            (if readable
1228     ;;                (insert (format "%S" cell))
1229     ;;              (char-db-insert-char-spec cell readable))
1230     ;;            (setq separator lbs))
1231     ;;           ((consp cell)
1232     ;;            (if separator
1233     ;;                (insert lbs))
1234     ;;            (if (consp (car cell))
1235     ;;                (char-db-insert-char-spec cell readable)
1236     ;;              (char-db-insert-char-reference cell readable))
1237     ;;            (setq separator lbs))
1238     ;;           (t
1239     ;;            (if separator
1240     ;;                (insert separator))
1241     ;;            (insert (prin1-to-string cell))
1242     ;;            (setq separator " ")))
1243     ;;     (setq value (cdr value)))
1244     ;;   (insert ")")
1245     ;;   (insert line-breaking)
1246     ;;   (setq attributes (delq 'ideographic- attributes)))
1247     (when (and (memq 'total-strokes attributes)
1248                (setq value (get-char-attribute char 'total-strokes)))
1249       (insert (format "(total-strokes       . %S)%s"
1250                       value
1251                       line-breaking))
1252       (setq attributes (delq 'total-strokes attributes))
1253       )
1254     ;; (when (and (memq '->ideograph attributes)
1255     ;;            (setq value (get-char-attribute char '->ideograph)))
1256     ;;   (insert (format "(->ideograph\t%s)%s"
1257     ;;                   (mapconcat (lambda (code)
1258     ;;                                (cond ((symbolp code)
1259     ;;                                       (symbol-name code))
1260     ;;                                      ((integerp code)
1261     ;;                                       (format "#x%04X" code))
1262     ;;                                      (t
1263     ;;                                       (format "%s %S"
1264     ;;                                               line-breaking code))))
1265     ;;                              value " ")
1266     ;;                   line-breaking))
1267     ;;   (setq attributes (delq '->ideograph attributes))
1268     ;;   )
1269     ;; (when (and (memq '->decomposition attributes)
1270     ;;            (setq value (get-char-attribute char '->decomposition)))
1271     ;;   (insert (format "(->decomposition\t%s)%s"
1272     ;;                   (mapconcat (lambda (code)
1273     ;;                                (cond ((symbolp code)
1274     ;;                                       (symbol-name code))
1275     ;;                                      ((characterp code)
1276     ;;                                       (if readable
1277     ;;                                           (format "%S" code)
1278     ;;                                         (format "#x%04X"
1279     ;;                                                 (char-int code))
1280     ;;                                         ))
1281     ;;                                      ((integerp code)
1282     ;;                                       (format "#x%04X" code))
1283     ;;                                      (t
1284     ;;                                       (format "%s%S" line-breaking code))))
1285     ;;                              value " ")
1286     ;;                   line-breaking))
1287     ;;   (setq attributes (delq '->decomposition attributes))
1288     ;;   )
1289     (if (equal (get-char-attribute char '->titlecase)
1290                (get-char-attribute char '->uppercase))
1291         (setq attributes (delq '->titlecase attributes)))
1292     (when (and (memq '->mojikyo attributes)
1293                (setq value (get-char-attribute char '->mojikyo)))
1294       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1295                       value (decode-char 'mojikyo value)
1296                       line-breaking))
1297       (setq attributes (delq '->mojikyo attributes))
1298       )
1299     ;; (when (and (memq 'hanyu-dazidian-vol attributes)
1300     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1301     ;;   (insert (format "(hanyu-dazidian-vol  . %d)%s"
1302     ;;                   value line-breaking))
1303     ;;   (setq attributes (delq 'hanyu-dazidian-vol attributes))
1304     ;;   )
1305     ;; (when (and (memq 'hanyu-dazidian-page attributes)
1306     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1307     ;;   (insert (format "(hanyu-dazidian-page . %d)%s"
1308     ;;                   value line-breaking))
1309     ;;   (setq attributes (delq 'hanyu-dazidian-page attributes))
1310     ;;   )
1311     ;; (when (and (memq 'hanyu-dazidian-char attributes)
1312     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1313     ;;   (insert (format "(hanyu-dazidian-char . %d)%s"
1314     ;;                   value line-breaking))
1315     ;;   (setq attributes (delq 'hanyu-dazidian-char attributes))
1316     ;;   )
1317     (unless readable
1318       (dolist (ignored '(composition
1319                          ->denotational <-subsumptive ->ucs-unified
1320                          ->ideographic-component-forms))
1321         (setq attributes (delq ignored attributes))))
1322     (while attributes
1323       (setq name (car attributes))
1324       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1325                   'value-is-empty)
1326         (cond ((setq ret (find-charset name))
1327                (setq name (charset-name ret))
1328                (when (not (memq name dest-ccss))
1329                  (setq dest-ccss (cons name dest-ccss))
1330                  (char-db-insert-ccs-feature name value line-breaking))
1331                )
1332               ((string-match "^=>ucs@" (symbol-name name))
1333                (insert (format "(%-18s . #x%04X)\t; %c%s"
1334                                name value (decode-char '=ucs value)
1335                                line-breaking))
1336                )
1337               ((eq name 'jisx0208-1978/4X)
1338                (insert (format "(%-18s . #x%04X)%s"
1339                                name value
1340                                line-breaking))
1341                )
1342               ((and
1343                 (not readable)
1344                 (not (eq name '->subsumptive))
1345                 (not (eq name '->uppercase))
1346                 (not (eq name '->lowercase))
1347                 (not (eq name '->titlecase))
1348                 (not (eq name '->canonical))
1349                 (not (eq name '->Bopomofo))
1350                 (not (eq name '->mistakable))
1351                 (not (eq name '->ideographic-variants))
1352                 (null (get-char-attribute
1353                        char (intern (format "%s*sources" name))))
1354                 (not (string-match "\\*sources$" (symbol-name name)))
1355                 (null (get-char-attribute
1356                        char (intern (format "%s*note" name))))
1357                 (not (string-match "\\*note$" (symbol-name name)))
1358                 (or (eq name '<-identical)
1359                     (eq name '<-uppercase)
1360                     (eq name '<-lowercase)
1361                     (eq name '<-titlecase)
1362                     (eq name '<-canonical)
1363                     (eq name '<-ideographic-variants)
1364                     ;; (eq name '<-synonyms)
1365                     (string-match "^<-synonyms" (symbol-name name))
1366                     (eq name '<-mistakable)
1367                     (when (string-match "^->" (symbol-name name))
1368                       (cond
1369                        ((string-match "^->fullwidth" (symbol-name name))
1370                         (not (and (consp value)
1371                                   (characterp (car value))
1372                                   (encode-char
1373                                    (car value) '=ucs 'defined-only)))
1374                         )
1375                        (t)))
1376                     ))
1377                )
1378               ((or (eq name 'ideographic-structure)
1379                    (eq name 'ideographic-combination)
1380                    ;; (eq name 'ideographic-)
1381                    (eq name '=decomposition)
1382                    (char-feature-base-name= '=decomposition name)
1383                    (char-feature-base-name= '=>decomposition name)
1384                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1385                    ;;               (symbol-name name))
1386                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1387                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1388                                  (symbol-name name))
1389                    )
1390                (char-db-insert-relation-feature char name value
1391                                                 line-breaking
1392                                                 ccss readable))
1393               ;; ((memq name '(ideograph=
1394               ;;               original-ideograph-of
1395               ;;               ancient-ideograph-of
1396               ;;               vulgar-ideograph-of
1397               ;;               wrong-ideograph-of
1398               ;;               ;; simplified-ideograph-of
1399               ;;               ideographic-variants
1400               ;;               ;; ideographic-different-form-of
1401               ;;               ))
1402               ;;  (insert (format "(%-18s%s " name line-breaking))
1403               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1404               ;;        separator nil)
1405               ;;  (while (consp value)
1406               ;;    (setq cell (car value))
1407               ;;    (if (and (consp cell)
1408               ;;             (consp (car cell)))
1409               ;;        (progn
1410               ;;          (if separator
1411               ;;              (insert lbs))
1412               ;;          (char-db-insert-alist cell readable)
1413               ;;          (setq separator lbs))
1414               ;;      (if separator
1415               ;;          (insert separator))
1416               ;;      (insert (prin1-to-string cell))
1417               ;;      (setq separator " "))
1418               ;;    (setq value (cdr value)))
1419               ;;  (insert ")")
1420               ;;  (insert line-breaking))
1421               ((consp value)
1422                (insert (format "(%-18s " name))
1423                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1424                      separator nil)
1425                (while (consp value)
1426                  (setq cell (car value))
1427                  (if (and (consp cell)
1428                           (consp (car cell))
1429                           (setq ret (condition-case nil
1430                                         (find-char cell)
1431                                       (error nil))))
1432                      (progn
1433                        (setq rest cell
1434                              al nil
1435                              cal nil)
1436                        (while rest
1437                          (setq key (car (car rest)))
1438                          (if (find-charset key)
1439                              (setq cal (cons key cal))
1440                            (setq al (cons key al)))
1441                          (setq rest (cdr rest)))
1442                        (if separator
1443                            (insert lbs))
1444                        (insert-char-attributes ret
1445                                                readable
1446                                                al ; cal
1447                                                nil 'for-sub-node)
1448                        (setq separator lbs))
1449                    (setq ret (prin1-to-string cell))
1450                    (if separator
1451                        (if (< (+ (current-column)
1452                                  (length ret)
1453                                  (length separator))
1454                               76)
1455                            (insert separator)
1456                          (insert lbs)))
1457                    (insert ret)
1458                    (setq separator " "))
1459                  (setq value (cdr value)))
1460                (insert ")")
1461                (insert line-breaking))
1462               (t
1463                (insert (format "(%-18s" name))
1464                (setq ret (prin1-to-string value))
1465                (unless (< (+ (current-column)
1466                              (length ret)
1467                              3)
1468                           76)
1469                  (insert line-breaking))
1470                (insert " . " ret ")" line-breaking)
1471                ;; (insert (format "(%-18s . %S)%s"
1472                ;;                 name value
1473                ;;                 line-breaking))
1474                )
1475               ))
1476       (setq attributes (cdr attributes)))
1477     (insert ")")))
1478
1479 (defun insert-char-data (char &optional readable
1480                               attributes)
1481   (save-restriction
1482     (narrow-to-region (point)(point))
1483     (insert "(define-char
1484   '")
1485     (insert-char-attributes char readable attributes)
1486     (insert ")\n")
1487     (goto-char (point-min))
1488     (while (re-search-forward "[ \t]+$" nil t)
1489       (replace-match ""))
1490     ;; from tabify.
1491     (goto-char (point-min))
1492     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1493       (let ((column (current-column))
1494             (indent-tabs-mode t))
1495         (delete-region (match-beginning 0) (point))
1496         (indent-to column)))
1497     (goto-char (point-max))
1498     ;; (tabify (point-min)(point-max))
1499     ))
1500
1501 (defun insert-char-data-with-variant (char &optional printable
1502                                            no-ucs-unified
1503                                            script excluded-script)
1504   (insert-char-data char printable)
1505   (let ((variants (char-variants char))
1506         rest
1507         variant vs ret)
1508     (setq variants (sort variants #'<))
1509     (setq rest variants)
1510     (setq variants (cons char variants))
1511     (while rest
1512       (setq variant (car rest))
1513       (unless (get-char-attribute variant '<-subsumptive)
1514         (if (and (or (null script)
1515                      (null (setq vs (get-char-attribute variant 'script)))
1516                      (memq script vs))
1517                  (or (null excluded-script)
1518                      (null (setq vs (get-char-attribute variant 'script)))
1519                      (not (memq excluded-script vs))))
1520             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1521               (insert-char-data variant printable)
1522               (if (setq ret (char-variants variant))
1523                   (while ret
1524                     (or (memq (car ret) variants)
1525                         ;; (get-char-attribute (car ret) '<-subsumptive)
1526                         (setq rest (nconc rest (list (car ret)))))
1527                     (setq ret (cdr ret)))))))
1528       (setq rest (cdr rest)))))
1529
1530 (defun insert-char-range-data (min max &optional script excluded-script)
1531   (let ((code min)
1532         char)
1533     (while (<= code max)
1534       (setq char (decode-char '=ucs code))
1535       (if (encode-char char '=ucs 'defined-only)
1536           (insert-char-data-with-variant char nil 'no-ucs-unified
1537                                          script excluded-script))
1538       (setq code (1+ code)))))
1539
1540 (defun write-char-range-data-to-file (min max file
1541                                           &optional script excluded-script)
1542   (let ((coding-system-for-write char-db-file-coding-system))
1543     (with-temp-buffer
1544       (insert (format ";; -*- coding: %s -*-\n"
1545                       char-db-file-coding-system))
1546       (insert-char-range-data min max script excluded-script)
1547       (write-region (point-min)(point-max) file))))
1548
1549 (defvar what-character-original-window-configuration)
1550
1551 ;;;###autoload
1552 (defun what-char-definition (char)
1553   (interactive (list (char-after)))
1554   (let ((buf (get-buffer-create "*Character Description*"))
1555         (the-buf (current-buffer))
1556         (win-conf (current-window-configuration)))
1557     (pop-to-buffer buf)
1558     (make-local-variable 'what-character-original-window-configuration)
1559     (setq what-character-original-window-configuration win-conf)
1560     (setq buffer-read-only nil)
1561     (erase-buffer)
1562     (condition-case err
1563         (progn
1564           (insert-char-data-with-variant char 'printable)
1565           (unless (char-attribute-alist char)
1566             (insert (format ";; = %c\n"
1567                             (let* ((rest (split-char char))
1568                                    (ccs (pop rest))
1569                                    (code (pop rest)))
1570                               (while rest
1571                                 (setq code (logior (lsh code 8)
1572                                                    (pop rest))))
1573                               (decode-char ccs code)))))
1574           ;; (char-db-update-comment)
1575           (set-buffer-modified-p nil)
1576           (view-mode the-buf (lambda (buf)
1577                                (set-window-configuration
1578                                 what-character-original-window-configuration)
1579                                ))
1580           (goto-char (point-min)))
1581       (error (progn
1582                (set-window-configuration
1583                 what-character-original-window-configuration)
1584                (signal (car err) (cdr err)))))))
1585
1586
1587 ;;; @ end
1588 ;;;
1589
1590 (provide 'char-db-util)
1591
1592 ;;; char-db-util.el ends here