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

   1 ;;; char-db-util.el --- Character Database utility -*- coding: utf-8-er; -*-
   2
   3 ;; Copyright (C) 1998,1999,2000,2001,2002,2003,2004,2005,2006,2007,
   4 ;;   2008,2009,2010,2011,2012,2013,2014,2015,2016,2017 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     ==hanyo-denshi/tk
 194     =gt
 195     =gt-k
 196     =daikanwa
 197     =daikanwa@rev2
 198     =daikanwa@rev1
 199     =daikanwa/+p
 200     ==daikanwa
 201     =cbeta
 202     ideograph-hanziku-1
 203     ideograph-hanziku-2
 204     ideograph-hanziku-3
 205     ideograph-hanziku-4
 206     ideograph-hanziku-5
 207     ideograph-hanziku-6
 208     ideograph-hanziku-7
 209     ideograph-hanziku-8
 210     ideograph-hanziku-9
 211     ideograph-hanziku-10
 212     ideograph-hanziku-11
 213     ideograph-hanziku-12
 214     ;; =>>>adobe-japan1-0
 215     ;; =>>>adobe-japan1-1
 216     ;; =>>>adobe-japan1-2
 217     ;; =>>>adobe-japan1-3
 218     ;; =>>>adobe-japan1-4
 219     ;; =>>>adobe-japan1-5
 220     ;; =>>>adobe-japan1-6
 221     ;; =>>>jis-x0208
 222     ;; =>>>jis-x0213-1
 223     ;; =>>>jis-x0213-2
 224     ;; =>>>hanyo-denshi/ja
 225     ;; =>>>hanyo-denshi/jb
 226     ;; =>>>hanyo-denshi/jc
 227     ;; =>>>hanyo-denshi/ft
 228     ;; =>>>hanyo-denshi/ib
 229     ;; =>>>hanyo-denshi/hg
 230     ;; =>>>hanyo-denshi/jt
 231     ;; =>>>hanyo-denshi/ks
 232     ;; =>>>gt
 233     =>>adobe-japan1-0
 234     =>>adobe-japan1-1
 235     =>>adobe-japan1-2
 236     =>>adobe-japan1-3
 237     =>>adobe-japan1-4
 238     =>>adobe-japan1-5
 239     =>>adobe-japan1-6
 240     =>>jis-x0208
 241     =>>jis-x0213-1
 242     =>>jis-x0213-1@2000
 243     =>>jis-x0213-1@2004
 244     =>>jis-x0213-2
 245     =>>jis-x0208@1978
 246     =>>hanyo-denshi/ft
 247     =>>hanyo-denshi/jt
 248     =>>hanyo-denshi/ks
 249     =>>gt
 250     =>>daikanwa
 251     =>>cbeta
 252     =+>jis-x0208
 253     =+>jis-x0213-1
 254     =+>jis-x0213-2
 255     =+>adobe-japan1-0
 256     =+>adobe-japan1-1
 257     =+>adobe-japan1-2
 258     =+>adobe-japan1-3
 259     =+>adobe-japan1-4
 260     =+>adobe-japan1-5
 261     =+>adobe-japan1-6
 262     =+>jis-x0208@1978
 263     =>jis-x0208
 264     =>jis-x0208@1997
 265     =>jis-x0213-1
 266     =>jis-x0213-1@2000
 267     =>jis-x0213-1@2004
 268     =>jis-x0213-2
 269     ==>ucs@bucs
 270     =>iwds-1
 271     =>ucs@hanyo-denshi
 272     =>ucs@iso
 273     =>ucs@unicode
 274     =>ucs@jis
 275     =>ucs@cns
 276     =>ucs@ks
 277     =+>ucs@iso
 278     =+>ucs@unicode
 279     =+>ucs@jis
 280     =+>ucs@jis/1990
 281     =+>ucs@cns
 282     =+>ucs@ks
 283     =>>ucs@iso
 284     =>>ucs@unicode
 285     =>>ucs@jis
 286     =>>ucs@cns
 287     =>>>ucs@iso
 288     =>>>ucs@unicode
 289     ==ucs@iso
 290     ==ucs@unicode
 291     ;; ==ucs@cns
 292     ==gb2312
 293     ==ks-x1001
 294     ==cns11643-1
 295     ==cns11643-2
 296     ==cns11643-3
 297     ==cns11643-4
 298     ==cns11643-5
 299     ==cns11643-6
 300     ==cns11643-7
 301     ==gt
 302     ==jis-x0208@1990
 303     ;; ==jis-x0208@1983
 304     ==jis-x0208@1978
 305     ==gt-k
 306     =ucs@iso
 307     =ucs@unicode
 308     =ucs@cns
 309     ==big5-cdp
 310     ==cbeta
 311     =>>big5-cdp
 312     =>>gt-k
 313     =+>gt
 314     =>gt
 315     =>mj
 316     =>big5-cdp
 317     =>daikanwa
 318     =>daikanwa/ho
 319     =>cns11643-5
 320     =>cns11643-7
 321     =big5
 322     =big5-eten
 323     =>gt-k
 324     =zinbun-oracle
 325     =>zinbun-oracle
 326     =ruimoku-v6
 327     =>>ruimoku-v6
 328     ==ruimoku-v6
 329     =jef-china3
 330     =>cbeta
 331     =shinjigen
 332     =ucs-var-001
 333     =ucs-var-002
 334     =ucs-var-003
 335     =ucs-var-004
 336     =ucs-var-006
 337     =ucs-var-008
 338     =ucs-var-010
 339     =ucs-itaiji-001
 340     =ucs-itaiji-002
 341     =ucs-itaiji-003
 342     =ucs-itaiji-004
 343     =ucs-itaiji-005
 344     =ucs-itaiji-006
 345     =ucs-itaiji-008
 346     =ucs-itaiji-009
 347     =>ucs-itaiji-001
 348     =>ucs-itaiji-002
 349     =>ucs-itaiji-005
 350     =>big5-cdp-itaiji-001
 351     =big5-cdp-itaiji-001
 352     =big5-cdp-itaiji-002
 353     =big5-cdp-itaiji-003
 354     =big5-cdp-var-003
 355     =big5-cdp-var-004
 356     =big5-cdp-var-005
 357     =>ucs@iwds-1
 358     =>ucs@component
 359     ==ucs@gb
 360     =ucs@JP/hanazono))
 361
 362
 363 ;;; @ char-db formatters
 364 ;;;
 365
 366 (defun char-db-make-char-spec (char)
 367   (let (ret char-spec)
 368     (cond ((characterp char)
 369            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 370                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 371                   (setq char-spec (list (cons '=ucs ret)))
 372                   (cond ((setq ret (get-char-attribute char 'name))
 373                          (setq char-spec (cons (cons 'name ret) char-spec))
 374                          )
 375                         ((setq ret (get-char-attribute char 'name*))
 376                          (setq char-spec (cons (cons 'name* ret) char-spec))
 377                          ))
 378                   )
 379                  ((encode-char char '=mj 'defined-only)
 380                   (setq char-spec nil)
 381                   (dolist (ccs (charset-list))
 382                     (if (and (or (eq ccs '=mj)
 383                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 384                                  (string-match "=ucs@" (symbol-name ccs))
 385                                  )
 386                              (setq ccs (charset-name ccs))
 387                              (null (assq ccs char-spec))
 388                              (setq ret (encode-char char ccs 'defined-only)))
 389                         (setq char-spec (cons (cons ccs ret) char-spec))))
 390                   )
 391                  ((encode-char char '==mj 'defined-only)
 392                   (setq char-spec nil)
 393                   (dolist (ccs (charset-list))
 394                     (if (and (or (eq ccs '==mj)
 395                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 396                                  (string-match "=ucs@" (symbol-name ccs))
 397                                  )
 398                              (setq ccs (charset-name ccs))
 399                              (null (assq ccs char-spec))
 400                              (setq ret (encode-char char ccs 'defined-only)))
 401                         (setq char-spec (cons (cons ccs ret) char-spec))))
 402                   )
 403                  ((encode-char char '=adobe-japan1 'defined-only)
 404                   (setq char-spec nil)
 405                   (dolist (ccs (charset-list))
 406                     (if (and (or (memq ccs
 407                                        '(=adobe-japan1-0
 408                                          =adobe-japan1-1
 409                                          =adobe-japan1-2
 410                                          =adobe-japan1-3
 411                                          =adobe-japan1-4
 412                                          =adobe-japan1-5
 413                                          =adobe-japan1-6
 414                                          ))
 415                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 416                                  (string-match "=ucs@" (symbol-name ccs))
 417                                  )
 418                              (setq ccs (charset-name ccs))
 419                              (null (assq ccs char-spec))
 420                              (setq ret (encode-char char ccs 'defined-only)))
 421                         (setq char-spec (cons (cons ccs ret) char-spec))))
 422                   )
 423                  ((encode-char char '==adobe-japan1 'defined-only)
 424                   (setq char-spec nil)
 425                   (dolist (ccs (charset-list))
 426                     (if (and (or (memq ccs
 427                                        '(==adobe-japan1-0
 428                                          ==adobe-japan1-1
 429                                          ==adobe-japan1-2
 430                                          ==adobe-japan1-3
 431                                          ==adobe-japan1-4
 432                                          ==adobe-japan1-5
 433                                          ==adobe-japan1-6
 434                                          ))
 435                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 436                                  (string-match "=ucs@" (symbol-name ccs))
 437                                  )
 438                              (setq ccs (charset-name ccs))
 439                              (null (assq ccs char-spec))
 440                              (setq ret (encode-char char ccs 'defined-only)))
 441                         (setq char-spec (cons (cons ccs ret) char-spec))))
 442                   )
 443                  ((setq ret
 444                         (catch 'tag
 445                           (let ((rest char-db-coded-charset-priority-list)
 446                                 ccs)
 447                             (while rest
 448                               (setq ccs (charset-name
 449                                          (find-charset (car rest))))
 450                               (if (setq ret
 451                                         (encode-char char ccs
 452                                                      'defined-only))
 453                                   (throw 'tag (cons ccs ret)))
 454                               (setq rest (cdr rest))))))
 455                   (setq char-spec (list ret))
 456                   (dolist (ccs (delq (car ret) (charset-list)))
 457                     (if (and (or (charset-iso-final-char ccs)
 458                                  (memq ccs
 459                                        '(=daikanwa
 460                                          =daikanwa@rev2
 461                                          ;; =gt-k
 462                                          =jis-x0208@1997
 463                                          ))
 464                                  (eq (charset-property ccs 'iso-ir) 177)
 465                                  ;; (string-match "=ucs@" (symbol-name ccs))
 466                                  )
 467                              (setq ccs (charset-name ccs))
 468                              (null (assq ccs char-spec))
 469                              (setq ret (encode-char char ccs 'defined-only)))
 470                         (setq char-spec (cons (cons ccs ret) char-spec))))
 471                   (if (null char-spec)
 472                       (setq char-spec (split-char char)))
 473                   (cond ((setq ret (get-char-attribute char 'name))
 474                          (setq char-spec (cons (cons 'name ret) char-spec))
 475                          )
 476                         ((setq ret (get-char-attribute char 'name*))
 477                          (setq char-spec (cons (cons 'name* ret) char-spec))
 478                          ))
 479                   )
 480                  ((setq ret (get-char-attribute
 481                              char 'ideographic-combination))
 482                   (setq char-spec
 483                         (cons (cons 'ideographic-combination ret)
 484                               char-spec))
 485                   ))
 486            char-spec)
 487           ((consp char)
 488            char))))
 489
 490 (defun char-db-insert-char-spec (char &optional readable column
 491                                       required-features)
 492   (unless column
 493     (setq column (current-column)))
 494   (let (char-spec temp-char)
 495     (setq char-spec (char-db-make-char-spec char))
 496     (unless (or (characterp char) ; char
 497                 (condition-case nil
 498                     (setq char (find-char char-spec))
 499                   (error nil)))
 500       ;; define temporary character
 501       ;;   Current implementation is dirty.
 502       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 503                                          char-spec)))
 504       (remove-char-attribute temp-char 'ideograph-daikanwa)
 505       (setq char temp-char))
 506     (insert-char-attributes char
 507                             readable
 508                             (union (mapcar #'car char-spec)
 509                                    required-features)
 510                             nil 'for-sub-node)
 511     (when temp-char
 512       ;; undefine temporary character
 513       ;;   Current implementation is dirty.
 514       (setq char-spec (char-attribute-alist temp-char))
 515       (while char-spec
 516         (remove-char-attribute temp-char (car (car char-spec)))
 517         (setq char-spec (cdr char-spec))))))
 518
 519 (defun char-db-insert-alist (alist &optional readable column)
 520   (unless column
 521     (setq column (current-column)))
 522   (let ((line-breaking
 523          (concat "\n" (make-string (1+ column) ?\ )))
 524         name value
 525         ret al ; cal
 526         key
 527         lbs cell rest separator)
 528     (insert "(")
 529     (while alist
 530       (setq name (car (car alist))
 531             value (cdr (car alist)))
 532       (cond ((eq name 'char)
 533              (insert "(char . ")
 534              (if (setq ret (condition-case nil
 535                                (find-char value)
 536                              (error nil)))
 537                  (progn
 538                    (setq al nil
 539                          ;; cal nil
 540                          )
 541                    (while value
 542                      (setq key (car (car value)))
 543                      ;; (if (find-charset key)
 544                      ;;     (setq cal (cons key cal))
 545                      (setq al (cons key al))
 546                      ;; )
 547                      (setq value (cdr value)))
 548                    (insert-char-attributes ret
 549                                            readable
 550                                            (or al 'none) ; cal
 551                                            nil 'for-sub-node))
 552                (insert (prin1-to-string value)))
 553              (insert ")")
 554              (insert line-breaking))
 555             ((consp value)
 556              (insert (format "(%-18s " name))
 557              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 558              (while (consp value)
 559                (setq cell (car value))
 560                (if (and (consp cell)
 561                         (consp (car cell))
 562                         (setq ret (condition-case nil
 563                                       (find-char cell)
 564                                     (error nil)))
 565                         )
 566                    (progn
 567                      (setq rest cell
 568                            al nil
 569                            ;; cal nil
 570                            )
 571                      (while rest
 572                        (setq key (car (car rest)))
 573                        ;; (if (find-charset key)
 574                        ;;     (setq cal (cons key cal))
 575                        (setq al (cons key al))
 576                        ;; )
 577                        (setq rest (cdr rest)))
 578                      (if separator
 579                          (insert lbs))
 580                      (insert-char-attributes ret
 581                                              readable
 582                                              al ; cal
 583                                              nil 'for-sub-node)
 584                      (setq separator lbs))
 585                  (if separator
 586                      (insert separator))
 587                  (insert (prin1-to-string cell))
 588                  (setq separator " "))
 589                (setq value (cdr value)))
 590              (insert ")")
 591              (insert line-breaking))
 592             (t
 593              (insert (format "(%-18s . %S)%s"
 594                              name value
 595                              line-breaking))))
 596       (setq alist (cdr alist))))
 597   (insert ")"))
 598
 599 (defun char-db-insert-char-reference (plist &optional readable column)
 600   (unless column
 601     (setq column (current-column)))
 602   (let ((line-breaking
 603          (concat "\n" (make-string (1+ column) ?\ )))
 604         (separator "")
 605         name value)
 606     (insert "(")
 607     (while plist
 608       (setq name (pop plist))
 609       (setq value (pop plist))
 610       (cond ((eq name :char)
 611              (insert separator)
 612              (insert ":char\t")
 613              (cond ((numberp value)
 614                     (setq value (decode-char '=ucs value)))
 615                    ;; ((consp value)
 616                    ;;  (setq value (or (find-char value)
 617                    ;;                  value)))
 618                    )
 619              (char-db-insert-char-spec value readable)
 620              (insert line-breaking)
 621              (setq separator ""))
 622             ((eq name :radical)
 623              (insert (format "%s%s\t%d ; %c%s"
 624                              separator
 625                              name value
 626                              (ideographic-radical value)
 627                              line-breaking))
 628              (setq separator ""))
 629             (t
 630              (insert (format "%s%s\t%S" separator name value))
 631              (setq separator line-breaking)))
 632       ))
 633   (insert ")"))
 634
 635 (defun char-db-decode-isolated-char (ccs code-point)
 636   (let (ret)
 637     (setq ret
 638           (cond ((eq ccs 'arabic-iso8859-6)
 639                  (decode-char ccs code-point))
 640                 ;; ((eq ccs '=gt)
 641                 ;;  (decode-builtin-char '==gt code-point))
 642                 ((and (memq ccs '(=gt-pj-1
 643                                   =gt-pj-2
 644                                   =gt-pj-3
 645                                   =gt-pj-4
 646                                   =gt-pj-5
 647                                   =gt-pj-6
 648                                   =gt-pj-7
 649                                   =gt-pj-8
 650                                   =gt-pj-9
 651                                   =gt-pj-10
 652                                   =gt-pj-11))
 653                       (setq ret (decode-char ccs code-point))
 654                       (setq ret (encode-char ret '=gt 'defined-only)))
 655                  (decode-builtin-char '=gt ret))
 656                 (t
 657                  (decode-builtin-char ccs code-point))))
 658     (cond ((null ret)
 659            (or (decode-char ccs code-point)
 660                (define-char (list (cons ccs code-point)))))
 661           ((and (<= 0 (char-int ret))
 662                 (<= (char-int ret) #x1F))
 663            (decode-char '=ucs (+ #x2400 (char-int ret))))
 664           ((= (char-int ret) #x7F)
 665            ?\u2421)
 666           (t ret))))
 667
 668 (defvar char-db-convert-obsolete-format t)
 669
 670 (defun char-db-insert-ccs-feature (name value line-breaking)
 671   (cond
 672    ((integerp value)
 673     (insert
 674      (format
 675       (cond
 676        ((memq name '(=>iwds-1
 677                      ==shinjigen
 678                      =shinjigen
 679                      =shinjigen@1ed ==shinjigen@1ed
 680                      =shinjigen@rev ==shinjigen@rev
 681                      =shinjigen/+p@rev ==shinjigen/+p@rev
 682                      ===daikanwa/ho ==daikanwa/ho
 683                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 684         "(%-18s .  %04d)\t; %c")
 685        ((eq name '=shinjigen@1ed/24pr)
 686         "(%-18s . %04d)\t; %c")
 687        ((or
 688          (memq name
 689                '(===daikanwa
 690                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 691                  =daikanwa@rev1 =daikanwa@rev2
 692                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 693                  =>>daikanwa/+p
 694                  =daikanwa/+2p =>>daikanwa/+2p
 695                  =gt ==gt ===gt
 696                  =>>gt =+>gt =>gt
 697                  =gt-k ==gt-k ===gt-k
 698                  =>>gt-k =>gt-k
 699                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 700                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 701                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 702                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 703                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 704                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 705                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 706                  =>>adobe-japan1-0 =+>adobe-japan1-0
 707                  =>>adobe-japan1-1 =+>adobe-japan1-1
 708                  =>>adobe-japan1-2 =+>adobe-japan1-2
 709                  =>>adobe-japan1-3 =+>adobe-japan1-3
 710                  =>>adobe-japan1-4 =+>adobe-japan1-4
 711                  =>>adobe-japan1-5 =+>adobe-japan1-5
 712                  =>>adobe-japan1-6 =+>adobe-japan1-6
 713                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 714                  =zinbun-oracle =>zinbun-oracle
 715                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 716                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 717                  ===hng-khi ===hng-khm ===hng-hok ===hng-kyd ===hng-sok
 718                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 719                  ===hng-sys ===hng-tsu ===hng-tzj
 720                  ===hng-hos ===hng-nak ===hng-jhk
 721                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 722                  ===hng-nkm ===hng-k24 ===hng-nkk
 723                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 724                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 725                  ===hng-okd ===hng-wad ===hng-kmi
 726                  ===hng-zkd ===hng-doh ===hng-jyu
 727                  ===hng-tzs ===hng-kss ===hng-kyo
 728                  ===hng-smk))
 729          ;; (string-match "^=adobe-" (symbol-name name))
 730          )
 731         "(%-18s . %05d)\t; %c")
 732        ((memq name '(=hanyo-denshi/ks
 733                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 734                      =>>hanyo-denshi/ks
 735                      =koseki ==koseki
 736                      =mj ==mj ===mj =>>mj =>mj
 737                      =zihai mojikyo))
 738         "(%-18s . %06d)\t; %c")
 739        ((memq name '(=hanyo-denshi/tk ==hanyo-denshi/tk))
 740         "(%-18s . %08d)\t; %c")
 741        ((>= (charset-dimension name) 2)
 742         "(%-18s . #x%04X)\t; %c")
 743        (t
 744         "(%-18s . #x%02X)\t; %c"))
 745       name
 746       (if (= (charset-iso-graphic-plane name) 1)
 747           (logior value
 748                   (cond ((= (charset-dimension name) 1)
 749                          #x80)
 750                         ((= (charset-dimension name) 2)
 751                          #x8080)
 752                         ((= (charset-dimension name) 3)
 753                          #x808080)
 754                         (t 0)))
 755         value)
 756       (char-db-decode-isolated-char name value)))
 757     (if (and (= (charset-chars name) 94)
 758              (= (charset-dimension name) 2))
 759         (insert (format " [%02d-%02d]"
 760                         (- (lsh value -8) 32)
 761                         (- (logand value 255) 32))))
 762     )
 763    (t
 764     (insert (format "(%-18s . %s)" name value))
 765     ))
 766   (insert line-breaking))
 767
 768 (defun char-db-insert-relation-feature (char name value line-breaking
 769                                              ccss readable)
 770   (insert (format "(%-18s%s " name line-breaking))
 771   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 772         separator cell sources required-features
 773         ret)
 774     (if (characterp value)
 775         (setq value (list value)))
 776     (while (consp value)
 777       (setq cell (car value))
 778       (if (integerp cell)
 779           (setq cell (decode-char '=ucs cell)))
 780       (cond
 781        ((eq name '->subsumptive)
 782         (when (or (not (some (lambda (atr)
 783                                (get-char-attribute cell atr))
 784                              char-db-ignored-attributes))
 785                   (some (lambda (ccs)
 786                           (encode-char cell ccs 'defined-only))
 787                         ccss))
 788           (if separator
 789               (insert lbs))
 790           (let ((char-db-ignored-attributes
 791                  (cons '<-subsumptive
 792                        char-db-ignored-attributes)))
 793             (insert-char-attributes cell readable nil nil 'for-sub-node))
 794           (setq separator lbs))
 795         )
 796        ((characterp cell)
 797         (setq sources
 798               (get-char-attribute
 799                char (intern (format "%s*sources" name))))
 800         (setq required-features nil)
 801         (dolist (source sources)
 802           (cond
 803            ((memq source '(JP
 804                            JP/Jouyou
 805                            shinjigen shinjigen@1ed shinjigen@rev))
 806             (setq required-features
 807                   (union required-features
 808                          '(=jis-x0208
 809                            =jis-x0208@1990
 810                            =jis-x0213-1@2000
 811                            =jis-x0213-1@2004
 812                            =jis-x0213-2
 813                            =jis-x0212
 814                            =jis-x0208@1983
 815                            =jis-x0208@1978
 816                            =shinjigen
 817                            =shinjigen@1ed
 818                            =shinjigen@rev
 819                            =shinjigen/+p@rev))))
 820            ((eq source 'CN)
 821             (setq required-features
 822                   (union required-features
 823                          '(=gb2312
 824                            =gb12345
 825                            =iso-ir165)))))
 826           (cond
 827            ((find-charset (setq ret (intern (format "=%s" source))))
 828             (setq required-features
 829                   (cons ret required-features)))
 830            (t (setq required-features
 831                     (cons source required-features)))))
 832         (cond ((string-match "@JP" (symbol-name name))
 833                (setq required-features
 834                      (union required-features
 835                             '(=jis-x0208
 836                               =jis-x0208@1990
 837                               =jis-x0213-1-2000
 838                               =jis-x0213-2-2000
 839                               =jis-x0212
 840                               =jis-x0208@1983
 841                               =jis-x0208@1978))))
 842               ((string-match "@CN" (symbol-name name))
 843                (setq required-features
 844                      (union required-features
 845                             '(=gb2312
 846                               =gb12345
 847                               =iso-ir165)))))
 848         (if separator
 849             (insert lbs))
 850         (if readable
 851             (insert (format "%S" cell))
 852           (char-db-insert-char-spec cell readable
 853                                     nil
 854                                     required-features))
 855         (setq separator lbs))
 856        ((consp cell)
 857         (if separator
 858             (insert lbs))
 859         (if (consp (car cell))
 860             (char-db-insert-char-spec cell readable)
 861           (char-db-insert-char-reference cell readable))
 862         (setq separator lbs))
 863        (t
 864         (if separator
 865             (insert separator))
 866         (insert (prin1-to-string cell))
 867         (setq separator " ")))
 868       (setq value (cdr value)))
 869     (insert ")")
 870     (insert line-breaking)))
 871
 872 (defun insert-char-attributes (char &optional readable attributes column
 873                                     for-sub-node)
 874   (unless column
 875     (setq column (current-column)))
 876   (let (name value ; has-long-ccs-name
 877         rest
 878         radical strokes
 879         (line-breaking
 880          (concat "\n" (make-string (1+ column) ?\ )))
 881         lbs cell separator ret
 882         key al cal
 883         dest-ccss ; sources required-features
 884         ccss)
 885     (let (atr-d)
 886       (setq attributes
 887             (sort (if attributes
 888                       (if (consp attributes)
 889                           (progn
 890                             (dolist (name attributes)
 891                               (unless (memq name char-db-ignored-attributes)
 892                                 (if (find-charset name)
 893                                     (push name ccss))
 894                                 (push name atr-d)))
 895                             atr-d))
 896                     (dolist (name (char-attribute-list))
 897                       (unless (memq name char-db-ignored-attributes)
 898                         (if (find-charset name)
 899                             (push name ccss))
 900                         (push name atr-d)))
 901                     atr-d)
 902                   #'char-attribute-name<)))
 903     (insert "(")
 904     (when (memq '<-subsumptive attributes)
 905       (when (or readable (not for-sub-node))
 906         (when (setq value (get-char-attribute char '<-subsumptive))
 907           (char-db-insert-relation-feature char '<-subsumptive value
 908                                            line-breaking
 909                                            ccss readable)))
 910       (setq attributes (delq '<-subsumptive attributes)))
 911     (when (and (memq '<-denotational attributes)
 912                (setq value (get-char-attribute char '<-denotational)))
 913       (char-db-insert-relation-feature char '<-denotational value
 914                                        line-breaking
 915                                        ccss readable)
 916       (setq attributes (delq '<-denotational attributes)))
 917     (when (and (memq '<-denotational@component attributes)
 918                (setq value (get-char-attribute char '<-denotational@component)))
 919       (char-db-insert-relation-feature char '<-denotational@component value
 920                                        line-breaking
 921                                        ccss readable)
 922       (setq attributes (delq '<-denotational@component attributes)))
 923     (when (and (memq 'name attributes)
 924                (setq value (get-char-attribute char 'name)))
 925       (insert (format
 926                (if (> (+ (current-column) (length value)) 48)
 927                    "(name . %S)%s"
 928                  "(name               . %S)%s")
 929                value line-breaking))
 930       (setq attributes (delq 'name attributes))
 931       )
 932     (when (and (memq 'name* attributes)
 933                (setq value (get-char-attribute char 'name*)))
 934       (insert (format
 935                (if (> (+ (current-column) (length value)) 48)
 936                    "(name* . %S)%s"
 937                  "(name*              . %S)%s")
 938                value line-breaking))
 939       (setq attributes (delq 'name* attributes))
 940       )
 941     (when (and (memq 'script attributes)
 942                (setq value (get-char-attribute char 'script)))
 943       (insert (format "(script\t\t%s)%s"
 944                       (mapconcat (function prin1-to-string)
 945                                  value " ")
 946                       line-breaking))
 947       (setq attributes (delq 'script attributes))
 948       )
 949     (dolist (name '(=>ucs =>ucs*))
 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 (decode-char '=ucs value)
 954                         line-breaking))
 955         (setq attributes (delq name attributes))))
 956     (dolist (name '(=>ucs@gb =>ucs@big5))
 957       (when (and (memq name attributes)
 958                  (setq value (get-char-attribute char name)))
 959         (insert (format "(%-18s . #x%04X)\t; %c%s"
 960                         name value
 961                         (decode-char (intern
 962                                       (concat "="
 963                                               (substring
 964                                                (symbol-name name) 2)))
 965                                      value)
 966                         line-breaking))
 967         (setq attributes (delq name attributes))
 968         ))
 969     ;; (dolist (name '(=>daikanwa))
 970     ;;   (when (and (memq name attributes)
 971     ;;              (setq value (get-char-attribute char name)))
 972     ;;     (insert
 973     ;;      (if (integerp value)
 974     ;;          (format "(%-18s . %05d)\t; %c%s"
 975     ;;                  name value (decode-char '=daikanwa value)
 976     ;;                  line-breaking)
 977     ;;        (format "(%-18s %s)\t; %c%s"
 978     ;;                name
 979     ;;                (mapconcat (function prin1-to-string)
 980     ;;                           value " ")
 981     ;;                (char-representative-of-daikanwa char)
 982     ;;                line-breaking)))
 983     ;;     (setq attributes (delq name attributes))))
 984     (when (and (memq 'general-category attributes)
 985                (setq value (get-char-attribute char 'general-category)))
 986       (insert (format
 987                "(general-category\t%s) ; %s%s"
 988                (mapconcat (lambda (cell)
 989                             (format "%S" cell))
 990                           value " ")
 991                (cond ((rassoc value unidata-normative-category-alist)
 992                       "Normative Category")
 993                      ((rassoc value unidata-informative-category-alist)
 994                       "Informative Category")
 995                      (t
 996                       "Unknown Category"))
 997                line-breaking))
 998       (setq attributes (delq 'general-category attributes))
 999       )
1000     (when (and (memq 'bidi-category attributes)
1001                (setq value (get-char-attribute char 'bidi-category)))
1002       (insert (format "(bidi-category\t. %S)%s"
1003                       value
1004                       line-breaking))
1005       (setq attributes (delq 'bidi-category attributes))
1006       )
1007     (unless (or (not (memq 'mirrored attributes))
1008                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
1009                     'empty))
1010       (insert (format "(mirrored\t\t. %S)%s"
1011                       value
1012                       line-breaking))
1013       (setq attributes (delq 'mirrored attributes))
1014       )
1015     (cond
1016      ((and (memq 'decimal-digit-value attributes)
1017            (setq value (get-char-attribute char 'decimal-digit-value)))
1018       (insert (format "(decimal-digit-value . %S)%s"
1019                       value
1020                       line-breaking))
1021       (setq attributes (delq 'decimal-digit-value attributes))
1022       (when (and (memq 'digit-value attributes)
1023                  (setq value (get-char-attribute char 'digit-value)))
1024         (insert (format "(digit-value\t . %S)%s"
1025                         value
1026                         line-breaking))
1027         (setq attributes (delq 'digit-value attributes))
1028         )
1029       (when (and (memq 'numeric-value attributes)
1030                  (setq value (get-char-attribute char 'numeric-value)))
1031         (insert (format "(numeric-value\t . %S)%s"
1032                         value
1033                         line-breaking))
1034         (setq attributes (delq 'numeric-value attributes))
1035         )
1036       )
1037      (t
1038       (when (and (memq 'digit-value attributes)
1039                  (setq value (get-char-attribute char 'digit-value)))
1040         (insert (format "(digit-value\t. %S)%s"
1041                         value
1042                         line-breaking))
1043         (setq attributes (delq 'digit-value attributes))
1044         )
1045       (when (and (memq 'numeric-value attributes)
1046                  (setq value (get-char-attribute char 'numeric-value)))
1047         (insert (format "(numeric-value\t. %S)%s"
1048                         value
1049                         line-breaking))
1050         (setq attributes (delq 'numeric-value attributes))
1051         )))
1052     (when (and (memq 'iso-10646-comment attributes)
1053                (setq value (get-char-attribute char 'iso-10646-comment)))
1054       (insert (format "(iso-10646-comment\t. %S)%s"
1055                       value
1056                       line-breaking))
1057       (setq attributes (delq 'iso-10646-comment attributes))
1058       )
1059     (when (and (memq 'morohashi-daikanwa attributes)
1060                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1061       (insert (format "(morohashi-daikanwa\t%s)%s"
1062                       (mapconcat (function prin1-to-string) value " ")
1063                       line-breaking))
1064       (setq attributes (delq 'morohashi-daikanwa attributes))
1065       )
1066     (setq radical nil
1067           strokes nil)
1068     (when (and (memq 'ideographic-radical attributes)
1069                (setq value (get-char-attribute char 'ideographic-radical)))
1070       (setq radical value)
1071       (insert (format "(ideographic-radical . %S)\t; %c%s"
1072                       radical
1073                       (ideographic-radical radical)
1074                       line-breaking))
1075       (setq attributes (delq 'ideographic-radical attributes))
1076       )
1077     (when (and (memq 'shuowen-radical attributes)
1078                (setq value (get-char-attribute char 'shuowen-radical)))
1079       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1080                       value
1081                       (shuowen-radical value)
1082                       line-breaking))
1083       (setq attributes (delq 'shuowen-radical attributes))
1084       )
1085     (let (key)
1086       (dolist (domain
1087                (append
1088                 char-db-feature-domains
1089                 (let (dest domain)
1090                   (dolist (feature (char-attribute-list))
1091                     (setq feature (symbol-name feature))
1092                     (when (string-match
1093                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1094                            feature)
1095                       (setq domain (intern (match-string 2 feature)))
1096                      (unless (memq domain dest)
1097                        (setq dest (cons domain dest)))))
1098                   (sort dest #'string<))))
1099         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1100         (when (and (memq key attributes)
1101                    (setq value (get-char-attribute char key)))
1102           (setq radical value)
1103           (insert (format "(%s . %S)\t; %c%s"
1104                           key
1105                           radical
1106                           (ideographic-radical radical)
1107                           line-breaking))
1108           (setq attributes (delq key attributes))
1109           )
1110         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1111         (when (and (memq key attributes)
1112                    (setq value (get-char-attribute char key)))
1113           (setq strokes value)
1114           (insert (format "(%s . %S)%s"
1115                           key
1116                           strokes
1117                           line-breaking))
1118           (setq attributes (delq key attributes))
1119           )
1120         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1121         (when (and (memq key attributes)
1122                    (setq value (get-char-attribute char key)))
1123           (insert (format "(%s       . %S)%s"
1124                           key
1125                           value
1126                           line-breaking))
1127           (setq attributes (delq key attributes))
1128           )
1129         (dolist (feature '(ideographic-radical
1130                            ideographic-strokes
1131                            total-strokes))
1132           (setq key (intern (format "%s@%s*sources" feature domain)))
1133           (when (and (memq key attributes)
1134                      (setq value (get-char-attribute char key)))
1135             (insert (format "(%s%s" key line-breaking))
1136             (dolist (cell value)
1137               (insert (format " %s" cell)))
1138             (insert ")")
1139             (insert line-breaking)
1140             (setq attributes (delq key attributes))
1141             ))
1142         ))
1143     (when (and (memq 'ideographic-strokes attributes)
1144                (setq value (get-char-attribute char 'ideographic-strokes)))
1145       (setq strokes value)
1146       (insert (format "(ideographic-strokes . %S)%s"
1147                       strokes
1148                       line-breaking))
1149       (setq attributes (delq 'ideographic-strokes attributes))
1150       )
1151     (when (and (memq 'kangxi-radical attributes)
1152                (setq value (get-char-attribute char 'kangxi-radical)))
1153       (unless (eq value radical)
1154         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1155                         value
1156                         (ideographic-radical value)
1157                         line-breaking))
1158         (or radical
1159             (setq radical value)))
1160       (setq attributes (delq 'kangxi-radical attributes))
1161       )
1162     (when (and (memq 'kangxi-strokes attributes)
1163                (setq value (get-char-attribute char 'kangxi-strokes)))
1164       (unless (eq value strokes)
1165         (insert (format "(kangxi-strokes\t . %S)%s"
1166                         value
1167                         line-breaking))
1168         (or strokes
1169             (setq strokes value)))
1170       (setq attributes (delq 'kangxi-strokes attributes))
1171       )
1172     ;; (when (and (memq 'japanese-radical attributes)
1173     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1174     ;;   (unless (eq value radical)
1175     ;;     (insert (format "(japanese-radical\t . %S)\t; %c%s"
1176     ;;                     value
1177     ;;                     (ideographic-radical value)
1178     ;;                     line-breaking))
1179     ;;     (or radical
1180     ;;         (setq radical value)))
1181     ;;   (setq attributes (delq 'japanese-radical attributes))
1182     ;;   )
1183     (when (and (memq 'japanese-strokes attributes)
1184                (setq value (get-char-attribute char 'japanese-strokes)))
1185       (unless (eq value strokes)
1186         (insert (format "(japanese-strokes\t . %S)%s"
1187                         value
1188                         line-breaking))
1189         (or strokes
1190             (setq strokes value)))
1191       (setq attributes (delq 'japanese-strokes attributes))
1192       )
1193     (when (and (memq 'cns-radical attributes)
1194                (setq value (get-char-attribute char 'cns-radical)))
1195       (insert (format "(cns-radical\t . %S)\t; %c%s"
1196                       value
1197                       (ideographic-radical value)
1198                       line-breaking))
1199       (setq attributes (delq 'cns-radical attributes))
1200       )
1201     (when (and (memq 'cns-strokes attributes)
1202                (setq value (get-char-attribute char 'cns-strokes)))
1203       (unless (eq value strokes)
1204         (insert (format "(cns-strokes\t . %S)%s"
1205                         value
1206                         line-breaking))
1207         (or strokes
1208             (setq strokes value)))
1209       (setq attributes (delq 'cns-strokes attributes))
1210       )
1211     ;; (when (and (memq 'shinjigen-1-radical attributes)
1212     ;;            (setq value (get-char-attribute char 'shinjigen-1-radical)))
1213     ;;   (unless (eq value radical)
1214     ;;     (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
1215     ;;                     value
1216     ;;                     (ideographic-radical value)
1217     ;;                     line-breaking))
1218     ;;     (or radical
1219     ;;         (setq radical value)))
1220     ;;   (setq attributes (delq 'shinjigen-1-radical attributes))
1221     ;;   )
1222     ;; (when (and (memq 'ideographic- attributes)
1223     ;;            (setq value (get-char-attribute char 'ideographic-)))
1224     ;;   (insert "(ideographic-       ")
1225     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1226     ;;         separator nil)
1227     ;;   (while (consp value)
1228     ;;     (setq cell (car value))
1229     ;;     (if (integerp cell)
1230     ;;         (setq cell (decode-char '=ucs cell)))
1231     ;;     (cond ((characterp cell)
1232     ;;            (if separator
1233     ;;                (insert lbs))
1234     ;;            (if readable
1235     ;;                (insert (format "%S" cell))
1236     ;;              (char-db-insert-char-spec cell readable))
1237     ;;            (setq separator lbs))
1238     ;;           ((consp cell)
1239     ;;            (if separator
1240     ;;                (insert lbs))
1241     ;;            (if (consp (car cell))
1242     ;;                (char-db-insert-char-spec cell readable)
1243     ;;              (char-db-insert-char-reference cell readable))
1244     ;;            (setq separator lbs))
1245     ;;           (t
1246     ;;            (if separator
1247     ;;                (insert separator))
1248     ;;            (insert (prin1-to-string cell))
1249     ;;            (setq separator " ")))
1250     ;;     (setq value (cdr value)))
1251     ;;   (insert ")")
1252     ;;   (insert line-breaking)
1253     ;;   (setq attributes (delq 'ideographic- attributes)))
1254     (when (and (memq 'total-strokes attributes)
1255                (setq value (get-char-attribute char 'total-strokes)))
1256       (insert (format "(total-strokes       . %S)%s"
1257                       value
1258                       line-breaking))
1259       (setq attributes (delq 'total-strokes attributes))
1260       )
1261     ;; (when (and (memq '->ideograph attributes)
1262     ;;            (setq value (get-char-attribute char '->ideograph)))
1263     ;;   (insert (format "(->ideograph\t%s)%s"
1264     ;;                   (mapconcat (lambda (code)
1265     ;;                                (cond ((symbolp code)
1266     ;;                                       (symbol-name code))
1267     ;;                                      ((integerp code)
1268     ;;                                       (format "#x%04X" code))
1269     ;;                                      (t
1270     ;;                                       (format "%s %S"
1271     ;;                                               line-breaking code))))
1272     ;;                              value " ")
1273     ;;                   line-breaking))
1274     ;;   (setq attributes (delq '->ideograph attributes))
1275     ;;   )
1276     ;; (when (and (memq '->decomposition attributes)
1277     ;;            (setq value (get-char-attribute char '->decomposition)))
1278     ;;   (insert (format "(->decomposition\t%s)%s"
1279     ;;                   (mapconcat (lambda (code)
1280     ;;                                (cond ((symbolp code)
1281     ;;                                       (symbol-name code))
1282     ;;                                      ((characterp code)
1283     ;;                                       (if readable
1284     ;;                                           (format "%S" code)
1285     ;;                                         (format "#x%04X"
1286     ;;                                                 (char-int code))
1287     ;;                                         ))
1288     ;;                                      ((integerp code)
1289     ;;                                       (format "#x%04X" code))
1290     ;;                                      (t
1291     ;;                                       (format "%s%S" line-breaking code))))
1292     ;;                              value " ")
1293     ;;                   line-breaking))
1294     ;;   (setq attributes (delq '->decomposition attributes))
1295     ;;   )
1296     (if (equal (get-char-attribute char '->titlecase)
1297                (get-char-attribute char '->uppercase))
1298         (setq attributes (delq '->titlecase attributes)))
1299     (when (and (memq '->mojikyo attributes)
1300                (setq value (get-char-attribute char '->mojikyo)))
1301       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1302                       value (decode-char 'mojikyo value)
1303                       line-breaking))
1304       (setq attributes (delq '->mojikyo attributes))
1305       )
1306     ;; (when (and (memq 'hanyu-dazidian-vol attributes)
1307     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1308     ;;   (insert (format "(hanyu-dazidian-vol  . %d)%s"
1309     ;;                   value line-breaking))
1310     ;;   (setq attributes (delq 'hanyu-dazidian-vol attributes))
1311     ;;   )
1312     ;; (when (and (memq 'hanyu-dazidian-page attributes)
1313     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1314     ;;   (insert (format "(hanyu-dazidian-page . %d)%s"
1315     ;;                   value line-breaking))
1316     ;;   (setq attributes (delq 'hanyu-dazidian-page attributes))
1317     ;;   )
1318     ;; (when (and (memq 'hanyu-dazidian-char attributes)
1319     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1320     ;;   (insert (format "(hanyu-dazidian-char . %d)%s"
1321     ;;                   value line-breaking))
1322     ;;   (setq attributes (delq 'hanyu-dazidian-char attributes))
1323     ;;   )
1324     (unless readable
1325       (dolist (ignored '(composition
1326                          ->denotational <-subsumptive ->ucs-unified
1327                          ->ideographic-component-forms))
1328         (setq attributes (delq ignored attributes))))
1329     (while attributes
1330       (setq name (car attributes))
1331       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1332                   'value-is-empty)
1333         (cond ((setq ret (find-charset name))
1334                (setq name (charset-name ret))
1335                (when (not (memq name dest-ccss))
1336                  (setq dest-ccss (cons name dest-ccss))
1337                  (char-db-insert-ccs-feature name value line-breaking))
1338                )
1339               ((string-match "^=>ucs@" (symbol-name name))
1340                (insert (format "(%-18s . #x%04X)\t; %c%s"
1341                                name value (decode-char '=ucs value)
1342                                line-breaking))
1343                )
1344               ((eq name 'jisx0208-1978/4X)
1345                (insert (format "(%-18s . #x%04X)%s"
1346                                name value
1347                                line-breaking))
1348                )
1349               ((and
1350                 (not readable)
1351                 (not (eq name '->subsumptive))
1352                 (not (eq name '->uppercase))
1353                 (not (eq name '->lowercase))
1354                 (not (eq name '->titlecase))
1355                 (not (eq name '->canonical))
1356                 (not (eq name '->Bopomofo))
1357                 (not (eq name '->mistakable))
1358                 (not (eq name '->ideographic-variants))
1359                 (null (get-char-attribute
1360                        char (intern (format "%s*sources" name))))
1361                 (not (string-match "\\*sources$" (symbol-name name)))
1362                 (null (get-char-attribute
1363                        char (intern (format "%s*note" name))))
1364                 (not (string-match "\\*note$" (symbol-name name)))
1365                 (or (eq name '<-identical)
1366                     (eq name '<-uppercase)
1367                     (eq name '<-lowercase)
1368                     (eq name '<-titlecase)
1369                     (eq name '<-canonical)
1370                     (eq name '<-ideographic-variants)
1371                     ;; (eq name '<-synonyms)
1372                     (string-match "^<-synonyms" (symbol-name name))
1373                     (eq name '<-mistakable)
1374                     (when (string-match "^->" (symbol-name name))
1375                       (cond
1376                        ((string-match "^->fullwidth" (symbol-name name))
1377                         (not (and (consp value)
1378                                   (characterp (car value))
1379                                   (encode-char
1380                                    (car value) '=ucs 'defined-only)))
1381                         )
1382                        (t)))
1383                     ))
1384                )
1385               ((or (eq name 'ideographic-structure)
1386                    (eq name 'ideographic-combination)
1387                    ;; (eq name 'ideographic-)
1388                    (eq name '=decomposition)
1389                    (char-feature-base-name= '=decomposition name)
1390                    (char-feature-base-name= '=>decomposition name)
1391                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1392                    ;;               (symbol-name name))
1393                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1394                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1395                                  (symbol-name name))
1396                    )
1397                (char-db-insert-relation-feature char name value
1398                                                 line-breaking
1399                                                 ccss readable))
1400               ;; ((memq name '(ideograph=
1401               ;;               original-ideograph-of
1402               ;;               ancient-ideograph-of
1403               ;;               vulgar-ideograph-of
1404               ;;               wrong-ideograph-of
1405               ;;               ;; simplified-ideograph-of
1406               ;;               ideographic-variants
1407               ;;               ;; ideographic-different-form-of
1408               ;;               ))
1409               ;;  (insert (format "(%-18s%s " name line-breaking))
1410               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1411               ;;        separator nil)
1412               ;;  (while (consp value)
1413               ;;    (setq cell (car value))
1414               ;;    (if (and (consp cell)
1415               ;;             (consp (car cell)))
1416               ;;        (progn
1417               ;;          (if separator
1418               ;;              (insert lbs))
1419               ;;          (char-db-insert-alist cell readable)
1420               ;;          (setq separator lbs))
1421               ;;      (if separator
1422               ;;          (insert separator))
1423               ;;      (insert (prin1-to-string cell))
1424               ;;      (setq separator " "))
1425               ;;    (setq value (cdr value)))
1426               ;;  (insert ")")
1427               ;;  (insert line-breaking))
1428               ((consp value)
1429                (insert (format "(%-18s " name))
1430                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1431                      separator nil)
1432                (while (consp value)
1433                  (setq cell (car value))
1434                  (if (and (consp cell)
1435                           (consp (car cell))
1436                           (setq ret (condition-case nil
1437                                         (find-char cell)
1438                                       (error nil))))
1439                      (progn
1440                        (setq rest cell
1441                              al nil
1442                              cal nil)
1443                        (while rest
1444                          (setq key (car (car rest)))
1445                          (if (find-charset key)
1446                              (setq cal (cons key cal))
1447                            (setq al (cons key al)))
1448                          (setq rest (cdr rest)))
1449                        (if separator
1450                            (insert lbs))
1451                        (insert-char-attributes ret
1452                                                readable
1453                                                al ; cal
1454                                                nil 'for-sub-node)
1455                        (setq separator lbs))
1456                    (setq ret (prin1-to-string cell))
1457                    (if separator
1458                        (if (< (+ (current-column)
1459                                  (length ret)
1460                                  (length separator))
1461                               76)
1462                            (insert separator)
1463                          (insert lbs)))
1464                    (insert ret)
1465                    (setq separator " "))
1466                  (setq value (cdr value)))
1467                (insert ")")
1468                (insert line-breaking))
1469               (t
1470                (insert (format "(%-18s" name))
1471                (setq ret (prin1-to-string value))
1472                (unless (< (+ (current-column)
1473                              (length ret)
1474                              3)
1475                           76)
1476                  (insert line-breaking))
1477                (insert " . " ret ")" line-breaking)
1478                ;; (insert (format "(%-18s . %S)%s"
1479                ;;                 name value
1480                ;;                 line-breaking))
1481                )
1482               ))
1483       (setq attributes (cdr attributes)))
1484     (insert ")")))
1485
1486 (defun insert-char-data (char &optional readable
1487                               attributes)
1488   (save-restriction
1489     (narrow-to-region (point)(point))
1490     (insert "(define-char
1491   '")
1492     (insert-char-attributes char readable attributes)
1493     (insert ")\n")
1494     (goto-char (point-min))
1495     (while (re-search-forward "[ \t]+$" nil t)
1496       (replace-match ""))
1497     ;; from tabify.
1498     (goto-char (point-min))
1499     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1500       (let ((column (current-column))
1501             (indent-tabs-mode t))
1502         (delete-region (match-beginning 0) (point))
1503         (indent-to column)))
1504     (goto-char (point-max))
1505     ;; (tabify (point-min)(point-max))
1506     ))
1507
1508 (defun insert-char-data-with-variant (char &optional printable
1509                                            no-ucs-unified
1510                                            script excluded-script)
1511   (insert-char-data char printable)
1512   (let ((variants (char-variants char))
1513         rest
1514         variant vs ret)
1515     (setq variants (sort variants #'<))
1516     (setq rest variants)
1517     (setq variants (cons char variants))
1518     (while rest
1519       (setq variant (car rest))
1520       (unless (get-char-attribute variant '<-subsumptive)
1521         (if (and (or (null script)
1522                      (null (setq vs (get-char-attribute variant 'script)))
1523                      (memq script vs))
1524                  (or (null excluded-script)
1525                      (null (setq vs (get-char-attribute variant 'script)))
1526                      (not (memq excluded-script vs))))
1527             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1528               (insert-char-data variant printable)
1529               (if (setq ret (char-variants variant))
1530                   (while ret
1531                     (or (memq (car ret) variants)
1532                         ;; (get-char-attribute (car ret) '<-subsumptive)
1533                         (setq rest (nconc rest (list (car ret)))))
1534                     (setq ret (cdr ret)))))))
1535       (setq rest (cdr rest)))))
1536
1537 (defun insert-char-range-data (min max &optional script excluded-script)
1538   (let ((code min)
1539         char)
1540     (while (<= code max)
1541       (setq char (decode-char '=ucs code))
1542       (if (encode-char char '=ucs 'defined-only)
1543           (insert-char-data-with-variant char nil 'no-ucs-unified
1544                                          script excluded-script))
1545       (setq code (1+ code)))))
1546
1547 (defun write-char-range-data-to-file (min max file
1548                                           &optional script excluded-script)
1549   (let ((coding-system-for-write char-db-file-coding-system))
1550     (with-temp-buffer
1551       (insert (format ";; -*- coding: %s -*-\n"
1552                       char-db-file-coding-system))
1553       (insert-char-range-data min max script excluded-script)
1554       (write-region (point-min)(point-max) file))))
1555
1556 (defvar what-character-original-window-configuration)
1557
1558 ;;;###autoload
1559 (defun what-char-definition (char)
1560   (interactive (list (char-after)))
1561   (let ((buf (get-buffer-create "*Character Description*"))
1562         (the-buf (current-buffer))
1563         (win-conf (current-window-configuration)))
1564     (pop-to-buffer buf)
1565     (make-local-variable 'what-character-original-window-configuration)
1566     (setq what-character-original-window-configuration win-conf)
1567     (setq buffer-read-only nil)
1568     (erase-buffer)
1569     (condition-case err
1570         (progn
1571           (insert-char-data-with-variant char 'printable)
1572           (unless (char-attribute-alist char)
1573             (insert (format ";; = %c\n"
1574                             (let* ((rest (split-char char))
1575                                    (ccs (pop rest))
1576                                    (code (pop rest)))
1577                               (while rest
1578                                 (setq code (logior (lsh code 8)
1579                                                    (pop rest))))
1580                               (decode-char ccs code)))))
1581           ;; (char-db-update-comment)
1582           (set-buffer-modified-p nil)
1583           (view-mode the-buf (lambda (buf)
1584                                (set-window-configuration
1585                                 what-character-original-window-configuration)
1586                                ))
1587           (goto-char (point-min)))
1588       (error (progn
1589                (set-window-configuration
1590                 what-character-original-window-configuration)
1591                (signal (car err) (cdr err)))))))
1592
1593
1594 ;;; @ end
1595 ;;;
1596
1597 (provide 'char-db-util)
1598
1599 ;;; char-db-util.el ends here