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-005
 337     =ucs-var-006
 338     =ucs-var-008
 339     =ucs-var-010
 340     =ucs-itaiji-001
 341     =ucs-itaiji-002
 342     =ucs-itaiji-003
 343     =ucs-itaiji-004
 344     =ucs-itaiji-005
 345     =ucs-itaiji-006
 346     =ucs-itaiji-008
 347     =ucs-itaiji-009
 348     =>ucs-itaiji-001
 349     =>ucs-itaiji-002
 350     =>ucs-itaiji-005
 351     =>ucs-itaiji-006
 352     =big5-cdp-var-002
 353     =big5-cdp-var-003
 354     =big5-cdp-var-004
 355     =big5-cdp-var-005
 356     =big5-cdp-itaiji-001
 357     =big5-cdp-itaiji-002
 358     =big5-cdp-itaiji-003
 359     =>big5-cdp-itaiji-001
 360     =>ucs@iwds-1
 361     =>ucs@component
 362     ==ucs@gb
 363     =ucs@JP/hanazono))
 364
 365
 366 ;;; @ char-db formatters
 367 ;;;
 368
 369 (defun char-db-make-char-spec (char)
 370   (let (ret char-spec)
 371     (cond ((characterp char)
 372            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 373                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 374                   (setq char-spec (list (cons '=ucs ret)))
 375                   (cond ((setq ret (get-char-attribute char 'name))
 376                          (setq char-spec (cons (cons 'name ret) char-spec))
 377                          )
 378                         ((setq ret (get-char-attribute char 'name*))
 379                          (setq char-spec (cons (cons 'name* ret) char-spec))
 380                          ))
 381                   )
 382                  ((encode-char char '=mj 'defined-only)
 383                   (setq char-spec nil)
 384                   (dolist (ccs (charset-list))
 385                     (if (and (or (eq ccs '=mj)
 386                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 387                                  (string-match "=ucs@" (symbol-name ccs))
 388                                  )
 389                              (setq ccs (charset-name ccs))
 390                              (null (assq ccs char-spec))
 391                              (setq ret (encode-char char ccs 'defined-only)))
 392                         (setq char-spec (cons (cons ccs ret) char-spec))))
 393                   )
 394                  ((encode-char char '==mj 'defined-only)
 395                   (setq char-spec nil)
 396                   (dolist (ccs (charset-list))
 397                     (if (and (or (eq ccs '==mj)
 398                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 399                                  (string-match "=ucs@" (symbol-name ccs))
 400                                  )
 401                              (setq ccs (charset-name ccs))
 402                              (null (assq ccs char-spec))
 403                              (setq ret (encode-char char ccs 'defined-only)))
 404                         (setq char-spec (cons (cons ccs ret) char-spec))))
 405                   )
 406                  ((encode-char char '=adobe-japan1 'defined-only)
 407                   (setq char-spec nil)
 408                   (dolist (ccs (charset-list))
 409                     (if (and (or (memq ccs
 410                                        '(=adobe-japan1-0
 411                                          =adobe-japan1-1
 412                                          =adobe-japan1-2
 413                                          =adobe-japan1-3
 414                                          =adobe-japan1-4
 415                                          =adobe-japan1-5
 416                                          =adobe-japan1-6
 417                                          ))
 418                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 419                                  (string-match "=ucs@" (symbol-name ccs))
 420                                  )
 421                              (setq ccs (charset-name ccs))
 422                              (null (assq ccs char-spec))
 423                              (setq ret (encode-char char ccs 'defined-only)))
 424                         (setq char-spec (cons (cons ccs ret) char-spec))))
 425                   )
 426                  ((encode-char char '==adobe-japan1 'defined-only)
 427                   (setq char-spec nil)
 428                   (dolist (ccs (charset-list))
 429                     (if (and (or (memq ccs
 430                                        '(==adobe-japan1-0
 431                                          ==adobe-japan1-1
 432                                          ==adobe-japan1-2
 433                                          ==adobe-japan1-3
 434                                          ==adobe-japan1-4
 435                                          ==adobe-japan1-5
 436                                          ==adobe-japan1-6
 437                                          ))
 438                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 439                                  (string-match "=ucs@" (symbol-name ccs))
 440                                  )
 441                              (setq ccs (charset-name ccs))
 442                              (null (assq ccs char-spec))
 443                              (setq ret (encode-char char ccs 'defined-only)))
 444                         (setq char-spec (cons (cons ccs ret) char-spec))))
 445                   )
 446                  ((setq ret
 447                         (catch 'tag
 448                           (let ((rest char-db-coded-charset-priority-list)
 449                                 ccs)
 450                             (while rest
 451                               (setq ccs (charset-name
 452                                          (find-charset (car rest))))
 453                               (if (setq ret
 454                                         (encode-char char ccs
 455                                                      'defined-only))
 456                                   (throw 'tag (cons ccs ret)))
 457                               (setq rest (cdr rest))))))
 458                   (setq char-spec (list ret))
 459                   (dolist (ccs (delq (car ret) (charset-list)))
 460                     (if (and (or (charset-iso-final-char ccs)
 461                                  (memq ccs
 462                                        '(=daikanwa
 463                                          =daikanwa@rev2
 464                                          ;; =gt-k
 465                                          =jis-x0208@1997
 466                                          ))
 467                                  (eq (charset-property ccs 'iso-ir) 177)
 468                                  ;; (string-match "=ucs@" (symbol-name ccs))
 469                                  )
 470                              (setq ccs (charset-name ccs))
 471                              (null (assq ccs char-spec))
 472                              (setq ret (encode-char char ccs 'defined-only)))
 473                         (setq char-spec (cons (cons ccs ret) char-spec))))
 474                   (if (null char-spec)
 475                       (setq char-spec (split-char char)))
 476                   (cond ((setq ret (get-char-attribute char 'name))
 477                          (setq char-spec (cons (cons 'name ret) char-spec))
 478                          )
 479                         ((setq ret (get-char-attribute char 'name*))
 480                          (setq char-spec (cons (cons 'name* ret) char-spec))
 481                          ))
 482                   )
 483                  ((setq ret (get-char-attribute
 484                              char 'ideographic-combination))
 485                   (setq char-spec
 486                         (cons (cons 'ideographic-combination ret)
 487                               char-spec))
 488                   ))
 489            char-spec)
 490           ((consp char)
 491            char))))
 492
 493 (defun char-db-insert-char-spec (char &optional readable column
 494                                       required-features)
 495   (unless column
 496     (setq column (current-column)))
 497   (let (char-spec temp-char)
 498     (setq char-spec (char-db-make-char-spec char))
 499     (unless (or (characterp char) ; char
 500                 (condition-case nil
 501                     (setq char (find-char char-spec))
 502                   (error nil)))
 503       ;; define temporary character
 504       ;;   Current implementation is dirty.
 505       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 506                                          char-spec)))
 507       (remove-char-attribute temp-char 'ideograph-daikanwa)
 508       (setq char temp-char))
 509     (insert-char-attributes char
 510                             readable
 511                             (union (mapcar #'car char-spec)
 512                                    required-features)
 513                             nil 'for-sub-node)
 514     (when temp-char
 515       ;; undefine temporary character
 516       ;;   Current implementation is dirty.
 517       (setq char-spec (char-attribute-alist temp-char))
 518       (while char-spec
 519         (remove-char-attribute temp-char (car (car char-spec)))
 520         (setq char-spec (cdr char-spec))))))
 521
 522 (defun char-db-insert-alist (alist &optional readable column)
 523   (unless column
 524     (setq column (current-column)))
 525   (let ((line-breaking
 526          (concat "\n" (make-string (1+ column) ?\ )))
 527         name value
 528         ret al ; cal
 529         key
 530         lbs cell rest separator)
 531     (insert "(")
 532     (while alist
 533       (setq name (car (car alist))
 534             value (cdr (car alist)))
 535       (cond ((eq name 'char)
 536              (insert "(char . ")
 537              (if (setq ret (condition-case nil
 538                                (find-char value)
 539                              (error nil)))
 540                  (progn
 541                    (setq al nil
 542                          ;; cal nil
 543                          )
 544                    (while value
 545                      (setq key (car (car value)))
 546                      ;; (if (find-charset key)
 547                      ;;     (setq cal (cons key cal))
 548                      (setq al (cons key al))
 549                      ;; )
 550                      (setq value (cdr value)))
 551                    (insert-char-attributes ret
 552                                            readable
 553                                            (or al 'none) ; cal
 554                                            nil 'for-sub-node))
 555                (insert (prin1-to-string value)))
 556              (insert ")")
 557              (insert line-breaking))
 558             ((consp value)
 559              (insert (format "(%-18s " name))
 560              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 561              (while (consp value)
 562                (setq cell (car value))
 563                (if (and (consp cell)
 564                         (consp (car cell))
 565                         (setq ret (condition-case nil
 566                                       (find-char cell)
 567                                     (error nil)))
 568                         )
 569                    (progn
 570                      (setq rest cell
 571                            al nil
 572                            ;; cal nil
 573                            )
 574                      (while rest
 575                        (setq key (car (car rest)))
 576                        ;; (if (find-charset key)
 577                        ;;     (setq cal (cons key cal))
 578                        (setq al (cons key al))
 579                        ;; )
 580                        (setq rest (cdr rest)))
 581                      (if separator
 582                          (insert lbs))
 583                      (insert-char-attributes ret
 584                                              readable
 585                                              al ; cal
 586                                              nil 'for-sub-node)
 587                      (setq separator lbs))
 588                  (if separator
 589                      (insert separator))
 590                  (insert (prin1-to-string cell))
 591                  (setq separator " "))
 592                (setq value (cdr value)))
 593              (insert ")")
 594              (insert line-breaking))
 595             (t
 596              (insert (format "(%-18s . %S)%s"
 597                              name value
 598                              line-breaking))))
 599       (setq alist (cdr alist))))
 600   (insert ")"))
 601
 602 (defun char-db-insert-char-reference (plist &optional readable column)
 603   (unless column
 604     (setq column (current-column)))
 605   (let ((line-breaking
 606          (concat "\n" (make-string (1+ column) ?\ )))
 607         (separator "")
 608         name value)
 609     (insert "(")
 610     (while plist
 611       (setq name (pop plist))
 612       (setq value (pop plist))
 613       (cond ((eq name :char)
 614              (insert separator)
 615              (insert ":char\t")
 616              (cond ((numberp value)
 617                     (setq value (decode-char '=ucs value)))
 618                    ;; ((consp value)
 619                    ;;  (setq value (or (find-char value)
 620                    ;;                  value)))
 621                    )
 622              (char-db-insert-char-spec value readable)
 623              (insert line-breaking)
 624              (setq separator ""))
 625             ((eq name :radical)
 626              (insert (format "%s%s\t%d ; %c%s"
 627                              separator
 628                              name value
 629                              (ideographic-radical value)
 630                              line-breaking))
 631              (setq separator ""))
 632             (t
 633              (insert (format "%s%s\t%S" separator name value))
 634              (setq separator line-breaking)))
 635       ))
 636   (insert ")"))
 637
 638 (defun char-db-decode-isolated-char (ccs code-point)
 639   (let (ret)
 640     (setq ret
 641           (cond ((eq ccs 'arabic-iso8859-6)
 642                  (decode-char ccs code-point))
 643                 ;; ((eq ccs '=gt)
 644                 ;;  (decode-builtin-char '==gt code-point))
 645                 ((and (memq ccs '(=gt-pj-1
 646                                   =gt-pj-2
 647                                   =gt-pj-3
 648                                   =gt-pj-4
 649                                   =gt-pj-5
 650                                   =gt-pj-6
 651                                   =gt-pj-7
 652                                   =gt-pj-8
 653                                   =gt-pj-9
 654                                   =gt-pj-10
 655                                   =gt-pj-11))
 656                       (setq ret (decode-char ccs code-point))
 657                       (setq ret (encode-char ret '=gt 'defined-only)))
 658                  (decode-builtin-char '=gt ret))
 659                 (t
 660                  (decode-builtin-char ccs code-point))))
 661     (cond ((null ret)
 662            (or (decode-char ccs code-point)
 663                (define-char (list (cons ccs code-point)))))
 664           ((and (<= 0 (char-int ret))
 665                 (<= (char-int ret) #x1F))
 666            (decode-char '=ucs (+ #x2400 (char-int ret))))
 667           ((= (char-int ret) #x7F)
 668            ?\u2421)
 669           (t ret))))
 670
 671 (defvar char-db-convert-obsolete-format t)
 672
 673 (defun char-db-insert-ccs-feature (name value line-breaking)
 674   (cond
 675    ((integerp value)
 676     (insert
 677      (format
 678       (cond
 679        ((memq name '(=>iwds-1
 680                      ==shinjigen
 681                      =shinjigen
 682                      =shinjigen@1ed ==shinjigen@1ed
 683                      =shinjigen@rev ==shinjigen@rev
 684                      =shinjigen/+p@rev ==shinjigen/+p@rev
 685                      ===daikanwa/ho ==daikanwa/ho
 686                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 687         "(%-18s .  %04d)\t; %c")
 688        ((eq name '=shinjigen@1ed/24pr)
 689         "(%-18s . %04d)\t; %c")
 690        ((or
 691          (memq name
 692                '(===daikanwa
 693                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 694                  =daikanwa@rev1 =daikanwa@rev2
 695                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 696                  =>>daikanwa/+p
 697                  =daikanwa/+2p =>>daikanwa/+2p
 698                  =gt ==gt ===gt
 699                  =>>gt =+>gt =>gt
 700                  =gt-k ==gt-k ===gt-k
 701                  =>>gt-k =>gt-k
 702                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 703                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 704                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 705                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 706                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 707                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 708                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 709                  =>>adobe-japan1-0 =+>adobe-japan1-0
 710                  =>>adobe-japan1-1 =+>adobe-japan1-1
 711                  =>>adobe-japan1-2 =+>adobe-japan1-2
 712                  =>>adobe-japan1-3 =+>adobe-japan1-3
 713                  =>>adobe-japan1-4 =+>adobe-japan1-4
 714                  =>>adobe-japan1-5 =+>adobe-japan1-5
 715                  =>>adobe-japan1-6 =+>adobe-japan1-6
 716                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 717                  =zinbun-oracle =>zinbun-oracle
 718                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 719                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 720                  ===hng-khi ===hng-khm ===hng-hok ===hng-kyd ===hng-sok
 721                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 722                  ===hng-sys ===hng-tsu ===hng-tzj
 723                  ===hng-hos ===hng-nak ===hng-jhk
 724                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 725                  ===hng-nkm ===hng-k24 ===hng-nkk
 726                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 727                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 728                  ===hng-okd ===hng-wad ===hng-kmi
 729                  ===hng-zkd ===hng-doh ===hng-jyu
 730                  ===hng-tzs ===hng-kss ===hng-kyo
 731                  ===hng-smk))
 732          ;; (string-match "^=adobe-" (symbol-name name))
 733          )
 734         "(%-18s . %05d)\t; %c")
 735        ((memq name '(=hanyo-denshi/ks
 736                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 737                      =>>hanyo-denshi/ks
 738                      =koseki ==koseki
 739                      =mj ==mj ===mj =>>mj =>mj
 740                      =zihai mojikyo))
 741         "(%-18s . %06d)\t; %c")
 742        ((memq name '(=hanyo-denshi/tk ==hanyo-denshi/tk))
 743         "(%-18s . %08d)\t; %c")
 744        ((>= (charset-dimension name) 2)
 745         "(%-18s . #x%04X)\t; %c")
 746        (t
 747         "(%-18s . #x%02X)\t; %c"))
 748       name
 749       (if (= (charset-iso-graphic-plane name) 1)
 750           (logior value
 751                   (cond ((= (charset-dimension name) 1)
 752                          #x80)
 753                         ((= (charset-dimension name) 2)
 754                          #x8080)
 755                         ((= (charset-dimension name) 3)
 756                          #x808080)
 757                         (t 0)))
 758         value)
 759       (char-db-decode-isolated-char name value)))
 760     (if (and (= (charset-chars name) 94)
 761              (= (charset-dimension name) 2))
 762         (insert (format " [%02d-%02d]"
 763                         (- (lsh value -8) 32)
 764                         (- (logand value 255) 32))))
 765     )
 766    (t
 767     (insert (format "(%-18s . %s)" name value))
 768     ))
 769   (insert line-breaking))
 770
 771 (defun char-db-insert-relation-feature (char name value line-breaking
 772                                              ccss readable)
 773   (insert (format "(%-18s%s " name line-breaking))
 774   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 775         separator cell sources required-features
 776         ret)
 777     (if (characterp value)
 778         (setq value (list value)))
 779     (while (consp value)
 780       (setq cell (car value))
 781       (if (integerp cell)
 782           (setq cell (decode-char '=ucs cell)))
 783       (cond
 784        ((eq name '->subsumptive)
 785         (when (or (not (some (lambda (atr)
 786                                (get-char-attribute cell atr))
 787                              char-db-ignored-attributes))
 788                   (some (lambda (ccs)
 789                           (encode-char cell ccs 'defined-only))
 790                         ccss))
 791           (if separator
 792               (insert lbs))
 793           (let ((char-db-ignored-attributes
 794                  (cons '<-subsumptive
 795                        char-db-ignored-attributes)))
 796             (insert-char-attributes cell readable nil nil 'for-sub-node))
 797           (setq separator lbs))
 798         )
 799        ((characterp cell)
 800         (setq sources
 801               (get-char-attribute
 802                char (intern (format "%s*sources" name))))
 803         (setq required-features nil)
 804         (dolist (source sources)
 805           (cond
 806            ((memq source '(JP
 807                            JP/Jouyou
 808                            shinjigen shinjigen@1ed shinjigen@rev))
 809             (setq required-features
 810                   (union required-features
 811                          '(=jis-x0208
 812                            =jis-x0208@1990
 813                            =jis-x0213-1@2000
 814                            =jis-x0213-1@2004
 815                            =jis-x0213-2
 816                            =jis-x0212
 817                            =jis-x0208@1983
 818                            =jis-x0208@1978
 819                            =shinjigen
 820                            =shinjigen@1ed
 821                            =shinjigen@rev
 822                            =shinjigen/+p@rev))))
 823            ((eq source 'CN)
 824             (setq required-features
 825                   (union required-features
 826                          '(=gb2312
 827                            =gb12345
 828                            =iso-ir165)))))
 829           (cond
 830            ((find-charset (setq ret (intern (format "=%s" source))))
 831             (setq required-features
 832                   (cons ret required-features)))
 833            (t (setq required-features
 834                     (cons source required-features)))))
 835         (cond ((string-match "@JP" (symbol-name name))
 836                (setq required-features
 837                      (union required-features
 838                             '(=jis-x0208
 839                               =jis-x0208@1990
 840                               =jis-x0213-1-2000
 841                               =jis-x0213-2-2000
 842                               =jis-x0212
 843                               =jis-x0208@1983
 844                               =jis-x0208@1978))))
 845               ((string-match "@CN" (symbol-name name))
 846                (setq required-features
 847                      (union required-features
 848                             '(=gb2312
 849                               =gb12345
 850                               =iso-ir165)))))
 851         (if separator
 852             (insert lbs))
 853         (if readable
 854             (insert (format "%S" cell))
 855           (char-db-insert-char-spec cell readable
 856                                     nil
 857                                     required-features))
 858         (setq separator lbs))
 859        ((consp cell)
 860         (if separator
 861             (insert lbs))
 862         (if (consp (car cell))
 863             (char-db-insert-char-spec cell readable)
 864           (char-db-insert-char-reference cell readable))
 865         (setq separator lbs))
 866        (t
 867         (if separator
 868             (insert separator))
 869         (insert (prin1-to-string cell))
 870         (setq separator " ")))
 871       (setq value (cdr value)))
 872     (insert ")")
 873     (insert line-breaking)))
 874
 875 (defun insert-char-attributes (char &optional readable attributes column
 876                                     for-sub-node)
 877   (unless column
 878     (setq column (current-column)))
 879   (let (name value ; has-long-ccs-name
 880         rest
 881         radical strokes
 882         (line-breaking
 883          (concat "\n" (make-string (1+ column) ?\ )))
 884         lbs cell separator ret
 885         key al cal
 886         dest-ccss ; sources required-features
 887         ccss)
 888     (let (atr-d)
 889       (setq attributes
 890             (sort (if attributes
 891                       (if (consp attributes)
 892                           (progn
 893                             (dolist (name attributes)
 894                               (unless (memq name char-db-ignored-attributes)
 895                                 (if (find-charset name)
 896                                     (push name ccss))
 897                                 (push name atr-d)))
 898                             atr-d))
 899                     (dolist (name (char-attribute-list))
 900                       (unless (memq name char-db-ignored-attributes)
 901                         (if (find-charset name)
 902                             (push name ccss))
 903                         (push name atr-d)))
 904                     atr-d)
 905                   #'char-attribute-name<)))
 906     (insert "(")
 907     (when (memq '<-subsumptive attributes)
 908       (when (or readable (not for-sub-node))
 909         (when (setq value (get-char-attribute char '<-subsumptive))
 910           (char-db-insert-relation-feature char '<-subsumptive value
 911                                            line-breaking
 912                                            ccss readable)))
 913       (setq attributes (delq '<-subsumptive attributes)))
 914     (when (and (memq '<-denotational attributes)
 915                (setq value (get-char-attribute char '<-denotational)))
 916       (char-db-insert-relation-feature char '<-denotational value
 917                                        line-breaking
 918                                        ccss readable)
 919       (setq attributes (delq '<-denotational attributes)))
 920     (when (and (memq '<-denotational@component attributes)
 921                (setq value (get-char-attribute char '<-denotational@component)))
 922       (char-db-insert-relation-feature char '<-denotational@component value
 923                                        line-breaking
 924                                        ccss readable)
 925       (setq attributes (delq '<-denotational@component attributes)))
 926     (when (and (memq 'name attributes)
 927                (setq value (get-char-attribute char 'name)))
 928       (insert (format
 929                (if (> (+ (current-column) (length value)) 48)
 930                    "(name . %S)%s"
 931                  "(name               . %S)%s")
 932                value line-breaking))
 933       (setq attributes (delq 'name attributes))
 934       )
 935     (when (and (memq 'name* attributes)
 936                (setq value (get-char-attribute char 'name*)))
 937       (insert (format
 938                (if (> (+ (current-column) (length value)) 48)
 939                    "(name* . %S)%s"
 940                  "(name*              . %S)%s")
 941                value line-breaking))
 942       (setq attributes (delq 'name* attributes))
 943       )
 944     (when (and (memq 'script attributes)
 945                (setq value (get-char-attribute char 'script)))
 946       (insert (format "(script\t\t%s)%s"
 947                       (mapconcat (function prin1-to-string)
 948                                  value " ")
 949                       line-breaking))
 950       (setq attributes (delq 'script attributes))
 951       )
 952     (dolist (name '(=>ucs =>ucs*))
 953       (when (and (memq name attributes)
 954                  (setq value (get-char-attribute char name)))
 955         (insert (format "(%-18s . #x%04X)\t; %c%s"
 956                         name value (decode-char '=ucs value)
 957                         line-breaking))
 958         (setq attributes (delq name attributes))))
 959     (dolist (name '(=>ucs@gb =>ucs@big5))
 960       (when (and (memq name attributes)
 961                  (setq value (get-char-attribute char name)))
 962         (insert (format "(%-18s . #x%04X)\t; %c%s"
 963                         name value
 964                         (decode-char (intern
 965                                       (concat "="
 966                                               (substring
 967                                                (symbol-name name) 2)))
 968                                      value)
 969                         line-breaking))
 970         (setq attributes (delq name attributes))
 971         ))
 972     ;; (dolist (name '(=>daikanwa))
 973     ;;   (when (and (memq name attributes)
 974     ;;              (setq value (get-char-attribute char name)))
 975     ;;     (insert
 976     ;;      (if (integerp value)
 977     ;;          (format "(%-18s . %05d)\t; %c%s"
 978     ;;                  name value (decode-char '=daikanwa value)
 979     ;;                  line-breaking)
 980     ;;        (format "(%-18s %s)\t; %c%s"
 981     ;;                name
 982     ;;                (mapconcat (function prin1-to-string)
 983     ;;                           value " ")
 984     ;;                (char-representative-of-daikanwa char)
 985     ;;                line-breaking)))
 986     ;;     (setq attributes (delq name attributes))))
 987     (when (and (memq 'general-category attributes)
 988                (setq value (get-char-attribute char 'general-category)))
 989       (insert (format
 990                "(general-category\t%s) ; %s%s"
 991                (mapconcat (lambda (cell)
 992                             (format "%S" cell))
 993                           value " ")
 994                (cond ((rassoc value unidata-normative-category-alist)
 995                       "Normative Category")
 996                      ((rassoc value unidata-informative-category-alist)
 997                       "Informative Category")
 998                      (t
 999                       "Unknown Category"))
1000                line-breaking))
1001       (setq attributes (delq 'general-category attributes))
1002       )
1003     (when (and (memq 'bidi-category attributes)
1004                (setq value (get-char-attribute char 'bidi-category)))
1005       (insert (format "(bidi-category\t. %S)%s"
1006                       value
1007                       line-breaking))
1008       (setq attributes (delq 'bidi-category attributes))
1009       )
1010     (unless (or (not (memq 'mirrored attributes))
1011                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
1012                     'empty))
1013       (insert (format "(mirrored\t\t. %S)%s"
1014                       value
1015                       line-breaking))
1016       (setq attributes (delq 'mirrored attributes))
1017       )
1018     (cond
1019      ((and (memq 'decimal-digit-value attributes)
1020            (setq value (get-char-attribute char 'decimal-digit-value)))
1021       (insert (format "(decimal-digit-value . %S)%s"
1022                       value
1023                       line-breaking))
1024       (setq attributes (delq 'decimal-digit-value attributes))
1025       (when (and (memq 'digit-value attributes)
1026                  (setq value (get-char-attribute char 'digit-value)))
1027         (insert (format "(digit-value\t . %S)%s"
1028                         value
1029                         line-breaking))
1030         (setq attributes (delq 'digit-value attributes))
1031         )
1032       (when (and (memq 'numeric-value attributes)
1033                  (setq value (get-char-attribute char 'numeric-value)))
1034         (insert (format "(numeric-value\t . %S)%s"
1035                         value
1036                         line-breaking))
1037         (setq attributes (delq 'numeric-value attributes))
1038         )
1039       )
1040      (t
1041       (when (and (memq 'digit-value attributes)
1042                  (setq value (get-char-attribute char 'digit-value)))
1043         (insert (format "(digit-value\t. %S)%s"
1044                         value
1045                         line-breaking))
1046         (setq attributes (delq 'digit-value attributes))
1047         )
1048       (when (and (memq 'numeric-value attributes)
1049                  (setq value (get-char-attribute char 'numeric-value)))
1050         (insert (format "(numeric-value\t. %S)%s"
1051                         value
1052                         line-breaking))
1053         (setq attributes (delq 'numeric-value attributes))
1054         )))
1055     (when (and (memq 'iso-10646-comment attributes)
1056                (setq value (get-char-attribute char 'iso-10646-comment)))
1057       (insert (format "(iso-10646-comment\t. %S)%s"
1058                       value
1059                       line-breaking))
1060       (setq attributes (delq 'iso-10646-comment attributes))
1061       )
1062     (when (and (memq 'morohashi-daikanwa attributes)
1063                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1064       (insert (format "(morohashi-daikanwa\t%s)%s"
1065                       (mapconcat (function prin1-to-string) value " ")
1066                       line-breaking))
1067       (setq attributes (delq 'morohashi-daikanwa attributes))
1068       )
1069     (setq radical nil
1070           strokes nil)
1071     (when (and (memq 'ideographic-radical attributes)
1072                (setq value (get-char-attribute char 'ideographic-radical)))
1073       (setq radical value)
1074       (insert (format "(ideographic-radical . %S)\t; %c%s"
1075                       radical
1076                       (ideographic-radical radical)
1077                       line-breaking))
1078       (setq attributes (delq 'ideographic-radical attributes))
1079       )
1080     (when (and (memq 'shuowen-radical attributes)
1081                (setq value (get-char-attribute char 'shuowen-radical)))
1082       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1083                       value
1084                       (shuowen-radical value)
1085                       line-breaking))
1086       (setq attributes (delq 'shuowen-radical attributes))
1087       )
1088     (let (key)
1089       (dolist (domain
1090                (append
1091                 char-db-feature-domains
1092                 (let (dest domain)
1093                   (dolist (feature (char-attribute-list))
1094                     (setq feature (symbol-name feature))
1095                     (when (string-match
1096                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1097                            feature)
1098                       (setq domain (intern (match-string 2 feature)))
1099                      (unless (memq domain dest)
1100                        (setq dest (cons domain dest)))))
1101                   (sort dest #'string<))))
1102         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1103         (when (and (memq key attributes)
1104                    (setq value (get-char-attribute char key)))
1105           (setq radical value)
1106           (insert (format "(%s . %S)\t; %c%s"
1107                           key
1108                           radical
1109                           (ideographic-radical radical)
1110                           line-breaking))
1111           (setq attributes (delq key attributes))
1112           )
1113         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1114         (when (and (memq key attributes)
1115                    (setq value (get-char-attribute char key)))
1116           (setq strokes value)
1117           (insert (format "(%s . %S)%s"
1118                           key
1119                           strokes
1120                           line-breaking))
1121           (setq attributes (delq key attributes))
1122           )
1123         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1124         (when (and (memq key attributes)
1125                    (setq value (get-char-attribute char key)))
1126           (insert (format "(%s       . %S)%s"
1127                           key
1128                           value
1129                           line-breaking))
1130           (setq attributes (delq key attributes))
1131           )
1132         (dolist (feature '(ideographic-radical
1133                            ideographic-strokes
1134                            total-strokes))
1135           (setq key (intern (format "%s@%s*sources" feature domain)))
1136           (when (and (memq key attributes)
1137                      (setq value (get-char-attribute char key)))
1138             (insert (format "(%s%s" key line-breaking))
1139             (dolist (cell value)
1140               (insert (format " %s" cell)))
1141             (insert ")")
1142             (insert line-breaking)
1143             (setq attributes (delq key attributes))
1144             ))
1145         ))
1146     (when (and (memq 'ideographic-strokes attributes)
1147                (setq value (get-char-attribute char 'ideographic-strokes)))
1148       (setq strokes value)
1149       (insert (format "(ideographic-strokes . %S)%s"
1150                       strokes
1151                       line-breaking))
1152       (setq attributes (delq 'ideographic-strokes attributes))
1153       )
1154     (when (and (memq 'kangxi-radical attributes)
1155                (setq value (get-char-attribute char 'kangxi-radical)))
1156       (unless (eq value radical)
1157         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1158                         value
1159                         (ideographic-radical value)
1160                         line-breaking))
1161         (or radical
1162             (setq radical value)))
1163       (setq attributes (delq 'kangxi-radical attributes))
1164       )
1165     (when (and (memq 'kangxi-strokes attributes)
1166                (setq value (get-char-attribute char 'kangxi-strokes)))
1167       (unless (eq value strokes)
1168         (insert (format "(kangxi-strokes\t . %S)%s"
1169                         value
1170                         line-breaking))
1171         (or strokes
1172             (setq strokes value)))
1173       (setq attributes (delq 'kangxi-strokes attributes))
1174       )
1175     ;; (when (and (memq 'japanese-radical attributes)
1176     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1177     ;;   (unless (eq value radical)
1178     ;;     (insert (format "(japanese-radical\t . %S)\t; %c%s"
1179     ;;                     value
1180     ;;                     (ideographic-radical value)
1181     ;;                     line-breaking))
1182     ;;     (or radical
1183     ;;         (setq radical value)))
1184     ;;   (setq attributes (delq 'japanese-radical attributes))
1185     ;;   )
1186     (when (and (memq 'japanese-strokes attributes)
1187                (setq value (get-char-attribute char 'japanese-strokes)))
1188       (unless (eq value strokes)
1189         (insert (format "(japanese-strokes\t . %S)%s"
1190                         value
1191                         line-breaking))
1192         (or strokes
1193             (setq strokes value)))
1194       (setq attributes (delq 'japanese-strokes attributes))
1195       )
1196     (when (and (memq 'cns-radical attributes)
1197                (setq value (get-char-attribute char 'cns-radical)))
1198       (insert (format "(cns-radical\t . %S)\t; %c%s"
1199                       value
1200                       (ideographic-radical value)
1201                       line-breaking))
1202       (setq attributes (delq 'cns-radical attributes))
1203       )
1204     (when (and (memq 'cns-strokes attributes)
1205                (setq value (get-char-attribute char 'cns-strokes)))
1206       (unless (eq value strokes)
1207         (insert (format "(cns-strokes\t . %S)%s"
1208                         value
1209                         line-breaking))
1210         (or strokes
1211             (setq strokes value)))
1212       (setq attributes (delq 'cns-strokes attributes))
1213       )
1214     ;; (when (and (memq 'shinjigen-1-radical attributes)
1215     ;;            (setq value (get-char-attribute char 'shinjigen-1-radical)))
1216     ;;   (unless (eq value radical)
1217     ;;     (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
1218     ;;                     value
1219     ;;                     (ideographic-radical value)
1220     ;;                     line-breaking))
1221     ;;     (or radical
1222     ;;         (setq radical value)))
1223     ;;   (setq attributes (delq 'shinjigen-1-radical attributes))
1224     ;;   )
1225     ;; (when (and (memq 'ideographic- attributes)
1226     ;;            (setq value (get-char-attribute char 'ideographic-)))
1227     ;;   (insert "(ideographic-       ")
1228     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1229     ;;         separator nil)
1230     ;;   (while (consp value)
1231     ;;     (setq cell (car value))
1232     ;;     (if (integerp cell)
1233     ;;         (setq cell (decode-char '=ucs cell)))
1234     ;;     (cond ((characterp cell)
1235     ;;            (if separator
1236     ;;                (insert lbs))
1237     ;;            (if readable
1238     ;;                (insert (format "%S" cell))
1239     ;;              (char-db-insert-char-spec cell readable))
1240     ;;            (setq separator lbs))
1241     ;;           ((consp cell)
1242     ;;            (if separator
1243     ;;                (insert lbs))
1244     ;;            (if (consp (car cell))
1245     ;;                (char-db-insert-char-spec cell readable)
1246     ;;              (char-db-insert-char-reference cell readable))
1247     ;;            (setq separator lbs))
1248     ;;           (t
1249     ;;            (if separator
1250     ;;                (insert separator))
1251     ;;            (insert (prin1-to-string cell))
1252     ;;            (setq separator " ")))
1253     ;;     (setq value (cdr value)))
1254     ;;   (insert ")")
1255     ;;   (insert line-breaking)
1256     ;;   (setq attributes (delq 'ideographic- attributes)))
1257     (when (and (memq 'total-strokes attributes)
1258                (setq value (get-char-attribute char 'total-strokes)))
1259       (insert (format "(total-strokes       . %S)%s"
1260                       value
1261                       line-breaking))
1262       (setq attributes (delq 'total-strokes attributes))
1263       )
1264     ;; (when (and (memq '->ideograph attributes)
1265     ;;            (setq value (get-char-attribute char '->ideograph)))
1266     ;;   (insert (format "(->ideograph\t%s)%s"
1267     ;;                   (mapconcat (lambda (code)
1268     ;;                                (cond ((symbolp code)
1269     ;;                                       (symbol-name code))
1270     ;;                                      ((integerp code)
1271     ;;                                       (format "#x%04X" code))
1272     ;;                                      (t
1273     ;;                                       (format "%s %S"
1274     ;;                                               line-breaking code))))
1275     ;;                              value " ")
1276     ;;                   line-breaking))
1277     ;;   (setq attributes (delq '->ideograph attributes))
1278     ;;   )
1279     ;; (when (and (memq '->decomposition attributes)
1280     ;;            (setq value (get-char-attribute char '->decomposition)))
1281     ;;   (insert (format "(->decomposition\t%s)%s"
1282     ;;                   (mapconcat (lambda (code)
1283     ;;                                (cond ((symbolp code)
1284     ;;                                       (symbol-name code))
1285     ;;                                      ((characterp code)
1286     ;;                                       (if readable
1287     ;;                                           (format "%S" code)
1288     ;;                                         (format "#x%04X"
1289     ;;                                                 (char-int code))
1290     ;;                                         ))
1291     ;;                                      ((integerp code)
1292     ;;                                       (format "#x%04X" code))
1293     ;;                                      (t
1294     ;;                                       (format "%s%S" line-breaking code))))
1295     ;;                              value " ")
1296     ;;                   line-breaking))
1297     ;;   (setq attributes (delq '->decomposition attributes))
1298     ;;   )
1299     (if (equal (get-char-attribute char '->titlecase)
1300                (get-char-attribute char '->uppercase))
1301         (setq attributes (delq '->titlecase attributes)))
1302     (when (and (memq '->mojikyo attributes)
1303                (setq value (get-char-attribute char '->mojikyo)))
1304       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1305                       value (decode-char 'mojikyo value)
1306                       line-breaking))
1307       (setq attributes (delq '->mojikyo attributes))
1308       )
1309     ;; (when (and (memq 'hanyu-dazidian-vol attributes)
1310     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1311     ;;   (insert (format "(hanyu-dazidian-vol  . %d)%s"
1312     ;;                   value line-breaking))
1313     ;;   (setq attributes (delq 'hanyu-dazidian-vol attributes))
1314     ;;   )
1315     ;; (when (and (memq 'hanyu-dazidian-page attributes)
1316     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1317     ;;   (insert (format "(hanyu-dazidian-page . %d)%s"
1318     ;;                   value line-breaking))
1319     ;;   (setq attributes (delq 'hanyu-dazidian-page attributes))
1320     ;;   )
1321     ;; (when (and (memq 'hanyu-dazidian-char attributes)
1322     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1323     ;;   (insert (format "(hanyu-dazidian-char . %d)%s"
1324     ;;                   value line-breaking))
1325     ;;   (setq attributes (delq 'hanyu-dazidian-char attributes))
1326     ;;   )
1327     (unless readable
1328       (dolist (ignored '(composition
1329                          ->denotational <-subsumptive ->ucs-unified
1330                          ->ideographic-component-forms))
1331         (setq attributes (delq ignored attributes))))
1332     (while attributes
1333       (setq name (car attributes))
1334       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1335                   'value-is-empty)
1336         (cond ((setq ret (find-charset name))
1337                (setq name (charset-name ret))
1338                (when (not (memq name dest-ccss))
1339                  (setq dest-ccss (cons name dest-ccss))
1340                  (char-db-insert-ccs-feature name value line-breaking))
1341                )
1342               ((string-match "^=>ucs@" (symbol-name name))
1343                (insert (format "(%-18s . #x%04X)\t; %c%s"
1344                                name value (decode-char '=ucs value)
1345                                line-breaking))
1346                )
1347               ((eq name 'jisx0208-1978/4X)
1348                (insert (format "(%-18s . #x%04X)%s"
1349                                name value
1350                                line-breaking))
1351                )
1352               ((and
1353                 (not readable)
1354                 (not (eq name '->subsumptive))
1355                 (not (eq name '->uppercase))
1356                 (not (eq name '->lowercase))
1357                 (not (eq name '->titlecase))
1358                 (not (eq name '->canonical))
1359                 (not (eq name '->Bopomofo))
1360                 (not (eq name '->mistakable))
1361                 (not (eq name '->ideographic-variants))
1362                 (null (get-char-attribute
1363                        char (intern (format "%s*sources" name))))
1364                 (not (string-match "\\*sources$" (symbol-name name)))
1365                 (null (get-char-attribute
1366                        char (intern (format "%s*note" name))))
1367                 (not (string-match "\\*note$" (symbol-name name)))
1368                 (or (eq name '<-identical)
1369                     (eq name '<-uppercase)
1370                     (eq name '<-lowercase)
1371                     (eq name '<-titlecase)
1372                     (eq name '<-canonical)
1373                     (eq name '<-ideographic-variants)
1374                     ;; (eq name '<-synonyms)
1375                     (string-match "^<-synonyms" (symbol-name name))
1376                     (eq name '<-mistakable)
1377                     (when (string-match "^->" (symbol-name name))
1378                       (cond
1379                        ((string-match "^->fullwidth" (symbol-name name))
1380                         (not (and (consp value)
1381                                   (characterp (car value))
1382                                   (encode-char
1383                                    (car value) '=ucs 'defined-only)))
1384                         )
1385                        (t)))
1386                     ))
1387                )
1388               ((or (eq name 'ideographic-structure)
1389                    (eq name 'ideographic-combination)
1390                    ;; (eq name 'ideographic-)
1391                    (eq name '=decomposition)
1392                    (char-feature-base-name= '=decomposition name)
1393                    (char-feature-base-name= '=>decomposition name)
1394                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1395                    ;;               (symbol-name name))
1396                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1397                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1398                                  (symbol-name name))
1399                    )
1400                (char-db-insert-relation-feature char name value
1401                                                 line-breaking
1402                                                 ccss readable))
1403               ;; ((memq name '(ideograph=
1404               ;;               original-ideograph-of
1405               ;;               ancient-ideograph-of
1406               ;;               vulgar-ideograph-of
1407               ;;               wrong-ideograph-of
1408               ;;               ;; simplified-ideograph-of
1409               ;;               ideographic-variants
1410               ;;               ;; ideographic-different-form-of
1411               ;;               ))
1412               ;;  (insert (format "(%-18s%s " name line-breaking))
1413               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1414               ;;        separator nil)
1415               ;;  (while (consp value)
1416               ;;    (setq cell (car value))
1417               ;;    (if (and (consp cell)
1418               ;;             (consp (car cell)))
1419               ;;        (progn
1420               ;;          (if separator
1421               ;;              (insert lbs))
1422               ;;          (char-db-insert-alist cell readable)
1423               ;;          (setq separator lbs))
1424               ;;      (if separator
1425               ;;          (insert separator))
1426               ;;      (insert (prin1-to-string cell))
1427               ;;      (setq separator " "))
1428               ;;    (setq value (cdr value)))
1429               ;;  (insert ")")
1430               ;;  (insert line-breaking))
1431               ((consp value)
1432                (insert (format "(%-18s " name))
1433                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1434                      separator nil)
1435                (while (consp value)
1436                  (setq cell (car value))
1437                  (if (and (consp cell)
1438                           (consp (car cell))
1439                           (setq ret (condition-case nil
1440                                         (find-char cell)
1441                                       (error nil))))
1442                      (progn
1443                        (setq rest cell
1444                              al nil
1445                              cal nil)
1446                        (while rest
1447                          (setq key (car (car rest)))
1448                          (if (find-charset key)
1449                              (setq cal (cons key cal))
1450                            (setq al (cons key al)))
1451                          (setq rest (cdr rest)))
1452                        (if separator
1453                            (insert lbs))
1454                        (insert-char-attributes ret
1455                                                readable
1456                                                al ; cal
1457                                                nil 'for-sub-node)
1458                        (setq separator lbs))
1459                    (setq ret (prin1-to-string cell))
1460                    (if separator
1461                        (if (< (+ (current-column)
1462                                  (length ret)
1463                                  (length separator))
1464                               76)
1465                            (insert separator)
1466                          (insert lbs)))
1467                    (insert ret)
1468                    (setq separator " "))
1469                  (setq value (cdr value)))
1470                (insert ")")
1471                (insert line-breaking))
1472               (t
1473                (insert (format "(%-18s" name))
1474                (setq ret (prin1-to-string value))
1475                (unless (< (+ (current-column)
1476                              (length ret)
1477                              3)
1478                           76)
1479                  (insert line-breaking))
1480                (insert " . " ret ")" line-breaking)
1481                ;; (insert (format "(%-18s . %S)%s"
1482                ;;                 name value
1483                ;;                 line-breaking))
1484                )
1485               ))
1486       (setq attributes (cdr attributes)))
1487     (insert ")")))
1488
1489 (defun insert-char-data (char &optional readable
1490                               attributes)
1491   (save-restriction
1492     (narrow-to-region (point)(point))
1493     (insert "(define-char
1494   '")
1495     (insert-char-attributes char readable attributes)
1496     (insert ")\n")
1497     (goto-char (point-min))
1498     (while (re-search-forward "[ \t]+$" nil t)
1499       (replace-match ""))
1500     ;; from tabify.
1501     (goto-char (point-min))
1502     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1503       (let ((column (current-column))
1504             (indent-tabs-mode t))
1505         (delete-region (match-beginning 0) (point))
1506         (indent-to column)))
1507     (goto-char (point-max))
1508     ;; (tabify (point-min)(point-max))
1509     ))
1510
1511 (defun insert-char-data-with-variant (char &optional printable
1512                                            no-ucs-unified
1513                                            script excluded-script)
1514   (insert-char-data char printable)
1515   (let ((variants (char-variants char))
1516         rest
1517         variant vs ret)
1518     (setq variants (sort variants #'<))
1519     (setq rest variants)
1520     (setq variants (cons char variants))
1521     (while rest
1522       (setq variant (car rest))
1523       (unless (get-char-attribute variant '<-subsumptive)
1524         (if (and (or (null script)
1525                      (null (setq vs (get-char-attribute variant 'script)))
1526                      (memq script vs))
1527                  (or (null excluded-script)
1528                      (null (setq vs (get-char-attribute variant 'script)))
1529                      (not (memq excluded-script vs))))
1530             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1531               (insert-char-data variant printable)
1532               (if (setq ret (char-variants variant))
1533                   (while ret
1534                     (or (memq (car ret) variants)
1535                         ;; (get-char-attribute (car ret) '<-subsumptive)
1536                         (setq rest (nconc rest (list (car ret)))))
1537                     (setq ret (cdr ret)))))))
1538       (setq rest (cdr rest)))))
1539
1540 (defun insert-char-range-data (min max &optional script excluded-script)
1541   (let ((code min)
1542         char)
1543     (while (<= code max)
1544       (setq char (decode-char '=ucs code))
1545       (if (encode-char char '=ucs 'defined-only)
1546           (insert-char-data-with-variant char nil 'no-ucs-unified
1547                                          script excluded-script))
1548       (setq code (1+ code)))))
1549
1550 (defun write-char-range-data-to-file (min max file
1551                                           &optional script excluded-script)
1552   (let ((coding-system-for-write char-db-file-coding-system))
1553     (with-temp-buffer
1554       (insert (format ";; -*- coding: %s -*-\n"
1555                       char-db-file-coding-system))
1556       (insert-char-range-data min max script excluded-script)
1557       (write-region (point-min)(point-max) file))))
1558
1559 (defvar what-character-original-window-configuration)
1560
1561 ;;;###autoload
1562 (defun what-char-definition (char)
1563   (interactive (list (char-after)))
1564   (let ((buf (get-buffer-create "*Character Description*"))
1565         (the-buf (current-buffer))
1566         (win-conf (current-window-configuration)))
1567     (pop-to-buffer buf)
1568     (make-local-variable 'what-character-original-window-configuration)
1569     (setq what-character-original-window-configuration win-conf)
1570     (setq buffer-read-only nil)
1571     (erase-buffer)
1572     (condition-case err
1573         (progn
1574           (insert-char-data-with-variant char 'printable)
1575           (unless (char-attribute-alist char)
1576             (insert (format ";; = %c\n"
1577                             (let* ((rest (split-char char))
1578                                    (ccs (pop rest))
1579                                    (code (pop rest)))
1580                               (while rest
1581                                 (setq code (logior (lsh code 8)
1582                                                    (pop rest))))
1583                               (decode-char ccs code)))))
1584           ;; (char-db-update-comment)
1585           (set-buffer-modified-p nil)
1586           (view-mode the-buf (lambda (buf)
1587                                (set-window-configuration
1588                                 what-character-original-window-configuration)
1589                                ))
1590           (goto-char (point-min)))
1591       (error (progn
1592                (set-window-configuration
1593                 what-character-original-window-configuration)
1594                (signal (car err) (cdr err)))))))
1595
1596
1597 ;;; @ end
1598 ;;;
1599
1600 (provide 'char-db-util)
1601
1602 ;;; char-db-util.el ends here