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