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 ;;   2008,2009,2010,2011,2012,2013,2014,2015,2016,2017,2018 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     =+>big5-cdp
 315     =>gt
 316     =>mj
 317     =>big5-cdp
 318     =>daikanwa
 319     =>daikanwa/ho
 320     =>cns11643-5
 321     =>cns11643-7
 322     =big5
 323     =big5-eten
 324     =>gt-k
 325     =zinbun-oracle
 326     =>zinbun-oracle
 327     =ruimoku-v6
 328     =>>ruimoku-v6
 329     ==ruimoku-v6
 330     =jef-china3
 331     =>cbeta
 332     =shinjigen
 333     =ucs-var-001
 334     =ucs-var-002
 335     =ucs-var-003
 336     =ucs-var-004
 337     =ucs-var-005
 338     =ucs-var-006
 339     =ucs-var-008
 340     =ucs-var-010
 341     =ucs-itaiji-001
 342     =ucs-itaiji-002
 343     =ucs-itaiji-003
 344     =ucs-itaiji-004
 345     =ucs-itaiji-005
 346     =ucs-itaiji-006
 347     =ucs-itaiji-007
 348     =ucs-itaiji-008
 349     =ucs-itaiji-009
 350     =ucs-itaiji-010
 351     =ucs-itaiji-011
 352     =>ucs-itaiji-001
 353     =>ucs-itaiji-002
 354     =>ucs-itaiji-003
 355     =>ucs-itaiji-004
 356     =>ucs-itaiji-005
 357     =>ucs-itaiji-006
 358     =>ucs-itaiji-007
 359     =>ucs-itaiji-009
 360     =big5-cdp-var-001
 361     =big5-cdp-var-002
 362     =big5-cdp-var-003
 363     =big5-cdp-var-004
 364     =big5-cdp-var-005
 365     =big5-cdp-itaiji-001
 366     =big5-cdp-itaiji-002
 367     =big5-cdp-itaiji-003
 368     =>big5-cdp-itaiji-001
 369     =>ucs@iwds-1
 370     =>ucs@cognate
 371     =>ucs@component
 372     =>ucs-itaiji-001@iwds-1
 373     =>big5-cdp@component
 374     =>big5-cdp@cognate
 375     ==ucs@gb
 376     =ucs@gb
 377     ==ucs-var-002
 378     =ucs@JP/hanazono))
 379
 380
 381 ;;; @ char-db formatters
 382 ;;;
 383
 384 (defun char-db-make-char-spec (char)
 385   (let (ret char-spec)
 386     (cond ((characterp char)
 387            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 388                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 389                   (setq char-spec (list (cons '=ucs ret)))
 390                   (cond ((setq ret (get-char-attribute char 'name))
 391                          (setq char-spec (cons (cons 'name ret) char-spec))
 392                          )
 393                         ((setq ret (get-char-attribute char 'name*))
 394                          (setq char-spec (cons (cons 'name* ret) char-spec))
 395                          ))
 396                   )
 397                  ((encode-char char '=mj 'defined-only)
 398                   (setq char-spec nil)
 399                   (dolist (ccs (charset-list))
 400                     (if (and (or (eq ccs '=mj)
 401                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 402                                  (string-match "=ucs@" (symbol-name ccs))
 403                                  )
 404                              (setq ccs (charset-name ccs))
 405                              (null (assq ccs char-spec))
 406                              (setq ret (encode-char char ccs 'defined-only)))
 407                         (setq char-spec (cons (cons ccs ret) char-spec))))
 408                   )
 409                  ((encode-char char '==mj 'defined-only)
 410                   (setq char-spec nil)
 411                   (dolist (ccs (charset-list))
 412                     (if (and (or (eq ccs '==mj)
 413                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 414                                  (string-match "=ucs@" (symbol-name ccs))
 415                                  )
 416                              (setq ccs (charset-name ccs))
 417                              (null (assq ccs char-spec))
 418                              (setq ret (encode-char char ccs 'defined-only)))
 419                         (setq char-spec (cons (cons ccs ret) char-spec))))
 420                   )
 421                  ((encode-char char '=adobe-japan1 'defined-only)
 422                   (setq char-spec nil)
 423                   (dolist (ccs (charset-list))
 424                     (if (and (or (memq ccs
 425                                        '(=adobe-japan1-0
 426                                          =adobe-japan1-1
 427                                          =adobe-japan1-2
 428                                          =adobe-japan1-3
 429                                          =adobe-japan1-4
 430                                          =adobe-japan1-5
 431                                          =adobe-japan1-6
 432                                          ))
 433                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 434                                  (string-match "=ucs@" (symbol-name ccs))
 435                                  )
 436                              (setq ccs (charset-name ccs))
 437                              (null (assq ccs char-spec))
 438                              (setq ret (encode-char char ccs 'defined-only)))
 439                         (setq char-spec (cons (cons ccs ret) char-spec))))
 440                   )
 441                  ((encode-char char '==adobe-japan1 'defined-only)
 442                   (setq char-spec nil)
 443                   (dolist (ccs (charset-list))
 444                     (if (and (or (memq ccs
 445                                        '(==adobe-japan1-0
 446                                          ==adobe-japan1-1
 447                                          ==adobe-japan1-2
 448                                          ==adobe-japan1-3
 449                                          ==adobe-japan1-4
 450                                          ==adobe-japan1-5
 451                                          ==adobe-japan1-6
 452                                          ))
 453                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 454                                  (string-match "=ucs@" (symbol-name ccs))
 455                                  )
 456                              (setq ccs (charset-name ccs))
 457                              (null (assq ccs char-spec))
 458                              (setq ret (encode-char char ccs 'defined-only)))
 459                         (setq char-spec (cons (cons ccs ret) char-spec))))
 460                   )
 461                  ((setq ret
 462                         (catch 'tag
 463                           (let ((rest char-db-coded-charset-priority-list)
 464                                 ccs)
 465                             (while rest
 466                               (setq ccs (charset-name
 467                                          (find-charset (car rest))))
 468                               (if (setq ret
 469                                         (encode-char char ccs
 470                                                      'defined-only))
 471                                   (throw 'tag (cons ccs ret)))
 472                               (setq rest (cdr rest))))))
 473                   (setq char-spec (list ret))
 474                   (dolist (ccs (delq (car ret) (charset-list)))
 475                     (if (and (or (charset-iso-final-char ccs)
 476                                  (memq ccs
 477                                        '(=daikanwa
 478                                          =daikanwa@rev2
 479                                          ;; =gt-k
 480                                          =jis-x0208@1997
 481                                          ))
 482                                  (and (eq (charset-property ccs 'iso-ir) 177)
 483                                       (not (eq (charset-name ccs) '=ucs@big5)))
 484                                  ;; (string-match "=ucs@" (symbol-name ccs))
 485                                  )
 486                              (setq ccs (charset-name ccs))
 487                              (null (assq ccs char-spec))
 488                              (setq ret (encode-char char ccs 'defined-only)))
 489                         (setq char-spec (cons (cons ccs ret) char-spec))))
 490                   (if (null char-spec)
 491                       (setq char-spec (split-char char)))
 492                   (cond ((setq ret (get-char-attribute char 'name))
 493                          (setq char-spec (cons (cons 'name ret) char-spec))
 494                          )
 495                         ((setq ret (get-char-attribute char 'name*))
 496                          (setq char-spec (cons (cons 'name* ret) char-spec))
 497                          ))
 498                   )
 499                  ((setq ret (get-char-attribute
 500                              char 'ideographic-combination))
 501                   (setq char-spec
 502                         (cons (cons 'ideographic-combination ret)
 503                               char-spec))
 504                   ))
 505            char-spec)
 506           ((consp char)
 507            char))))
 508
 509 (defun char-db-insert-char-spec (char &optional readable column
 510                                       required-features)
 511   (unless column
 512     (setq column (current-column)))
 513   (let (char-spec temp-char)
 514     (setq char-spec (char-db-make-char-spec char))
 515     (unless (or (characterp char) ; char
 516                 (condition-case nil
 517                     (setq char (find-char char-spec))
 518                   (error nil)))
 519       ;; define temporary character
 520       ;;   Current implementation is dirty.
 521       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 522                                          char-spec)))
 523       (remove-char-attribute temp-char 'ideograph-daikanwa)
 524       (setq char temp-char))
 525     (insert-char-attributes char
 526                             readable
 527                             (union (mapcar #'car char-spec)
 528                                    required-features)
 529                             nil 'for-sub-node)
 530     (when temp-char
 531       ;; undefine temporary character
 532       ;;   Current implementation is dirty.
 533       (setq char-spec (char-attribute-alist temp-char))
 534       (while char-spec
 535         (remove-char-attribute temp-char (car (car char-spec)))
 536         (setq char-spec (cdr char-spec))))))
 537
 538 (defun char-db-insert-alist (alist &optional readable column)
 539   (unless column
 540     (setq column (current-column)))
 541   (let ((line-breaking
 542          (concat "\n" (make-string (1+ column) ?\ )))
 543         name value
 544         ret al ; cal
 545         key
 546         lbs cell rest separator)
 547     (insert "(")
 548     (while alist
 549       (setq name (car (car alist))
 550             value (cdr (car alist)))
 551       (cond ((eq name 'char)
 552              (insert "(char . ")
 553              (if (setq ret (condition-case nil
 554                                (find-char value)
 555                              (error nil)))
 556                  (progn
 557                    (setq al nil
 558                          ;; cal nil
 559                          )
 560                    (while value
 561                      (setq key (car (car value)))
 562                      ;; (if (find-charset key)
 563                      ;;     (setq cal (cons key cal))
 564                      (setq al (cons key al))
 565                      ;; )
 566                      (setq value (cdr value)))
 567                    (insert-char-attributes ret
 568                                            readable
 569                                            (or al 'none) ; cal
 570                                            nil 'for-sub-node))
 571                (insert (prin1-to-string value)))
 572              (insert ")")
 573              (insert line-breaking))
 574             ((consp value)
 575              (insert (format "(%-18s " name))
 576              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 577              (while (consp value)
 578                (setq cell (car value))
 579                (if (and (consp cell)
 580                         (consp (car cell))
 581                         (setq ret (condition-case nil
 582                                       (find-char cell)
 583                                     (error nil)))
 584                         )
 585                    (progn
 586                      (setq rest cell
 587                            al nil
 588                            ;; cal nil
 589                            )
 590                      (while rest
 591                        (setq key (car (car rest)))
 592                        ;; (if (find-charset key)
 593                        ;;     (setq cal (cons key cal))
 594                        (setq al (cons key al))
 595                        ;; )
 596                        (setq rest (cdr rest)))
 597                      (if separator
 598                          (insert lbs))
 599                      (insert-char-attributes ret
 600                                              readable
 601                                              al ; cal
 602                                              nil 'for-sub-node)
 603                      (setq separator lbs))
 604                  (if separator
 605                      (insert separator))
 606                  (insert (prin1-to-string cell))
 607                  (setq separator " "))
 608                (setq value (cdr value)))
 609              (insert ")")
 610              (insert line-breaking))
 611             (t
 612              (insert (format "(%-18s . %S)%s"
 613                              name value
 614                              line-breaking))))
 615       (setq alist (cdr alist))))
 616   (insert ")"))
 617
 618 (defun char-db-insert-char-reference (plist &optional readable column)
 619   (unless column
 620     (setq column (current-column)))
 621   (let ((line-breaking
 622          (concat "\n" (make-string (1+ column) ?\ )))
 623         (separator "")
 624         name value)
 625     (insert "(")
 626     (while plist
 627       (setq name (pop plist))
 628       (setq value (pop plist))
 629       (cond ((eq name :char)
 630              (insert separator)
 631              (insert ":char\t")
 632              (cond ((numberp value)
 633                     (setq value (decode-char '=ucs value)))
 634                    ;; ((consp value)
 635                    ;;  (setq value (or (find-char value)
 636                    ;;                  value)))
 637                    )
 638              (char-db-insert-char-spec value readable)
 639              (insert line-breaking)
 640              (setq separator ""))
 641             ((eq name :radical)
 642              (insert (format "%s%s\t%d ; %c%s"
 643                              separator
 644                              name value
 645                              (ideographic-radical value)
 646                              line-breaking))
 647              (setq separator ""))
 648             (t
 649              (insert (format "%s%s\t%S" separator name value))
 650              (setq separator line-breaking)))
 651       ))
 652   (insert ")"))
 653
 654 (defun char-db-decode-isolated-char (ccs code-point)
 655   (let (ret)
 656     (setq ret
 657           (cond ((eq ccs 'arabic-iso8859-6)
 658                  (decode-char ccs code-point))
 659                 ;; ((eq ccs '=gt)
 660                 ;;  (decode-builtin-char '==gt code-point))
 661                 ((and (memq ccs '(=gt-pj-1
 662                                   =gt-pj-2
 663                                   =gt-pj-3
 664                                   =gt-pj-4
 665                                   =gt-pj-5
 666                                   =gt-pj-6
 667                                   =gt-pj-7
 668                                   =gt-pj-8
 669                                   =gt-pj-9
 670                                   =gt-pj-10
 671                                   =gt-pj-11))
 672                       (setq ret (decode-char ccs code-point))
 673                       (setq ret (encode-char ret '=gt 'defined-only)))
 674                  (decode-builtin-char '=gt ret))
 675                 (t
 676                  (decode-builtin-char ccs code-point))))
 677     (cond ((null ret)
 678            (or (decode-char ccs code-point)
 679                (define-char (list (cons ccs code-point)))))
 680           ((and (<= 0 (char-int ret))
 681                 (<= (char-int ret) #x1F))
 682            (decode-char '=ucs (+ #x2400 (char-int ret))))
 683           ((= (char-int ret) #x7F)
 684            ?\u2421)
 685           (t ret))))
 686
 687 (defvar char-db-convert-obsolete-format t)
 688
 689 (defun char-db-insert-ccs-feature (name value line-breaking)
 690   (cond
 691    ((integerp value)
 692     (insert
 693      (format
 694       (cond
 695        ((memq name '(=>iwds-1
 696                      ==shinjigen
 697                      =shinjigen
 698                      =shinjigen@1ed ==shinjigen@1ed
 699                      =shinjigen@rev ==shinjigen@rev
 700                      =shinjigen/+p@rev ==shinjigen/+p@rev
 701                      ===daikanwa/ho ==daikanwa/ho
 702                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 703         "(%-18s .  %04d)\t; %c")
 704        ((eq name '=shinjigen@1ed/24pr)
 705         "(%-18s . %04d)\t; %c")
 706        ((or
 707          (memq name
 708                '(===daikanwa
 709                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 710                  =daikanwa@rev1 =daikanwa@rev2
 711                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 712                  =>>daikanwa/+p
 713                  =daikanwa/+2p =>>daikanwa/+2p
 714                  =gt ==gt ===gt
 715                  =>>gt =+>gt =>gt
 716                  =gt-k ==gt-k ===gt-k
 717                  =>>gt-k =>gt-k
 718                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 719                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 720                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 721                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 722                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 723                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 724                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 725                  =>>adobe-japan1-0 =+>adobe-japan1-0
 726                  =>>adobe-japan1-1 =+>adobe-japan1-1
 727                  =>>adobe-japan1-2 =+>adobe-japan1-2
 728                  =>>adobe-japan1-3 =+>adobe-japan1-3
 729                  =>>adobe-japan1-4 =+>adobe-japan1-4
 730                  =>>adobe-japan1-5 =+>adobe-japan1-5
 731                  =>>adobe-japan1-6 =+>adobe-japan1-6
 732                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 733                  =zinbun-oracle =>zinbun-oracle
 734                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 735                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 736                  ===hng-khi ===hng-khm ===hng-hok ===hng-kyd ===hng-sok
 737                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 738                  ===hng-sys ===hng-tsu ===hng-tzj
 739                  ===hng-hos ===hng-nak ===hng-jhk
 740                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 741                  ===hng-nkm ===hng-k24 ===hng-nkk
 742                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 743                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 744                  ===hng-okd ===hng-wad ===hng-kmi
 745                  ===hng-zkd ===hng-doh ===hng-jyu
 746                  ===hng-tzs ===hng-kss ===hng-kyo
 747                  ===hng-smk))
 748          ;; (string-match "^=adobe-" (symbol-name name))
 749          )
 750         "(%-18s . %05d)\t; %c")
 751        ((memq name '(=hanyo-denshi/ks
 752                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 753                      =>>hanyo-denshi/ks
 754                      =koseki ==koseki
 755                      =mj ==mj ===mj =>>mj =>mj
 756                      =zihai mojikyo))
 757         "(%-18s . %06d)\t; %c")
 758        ((memq name '(=hanyo-denshi/tk ==hanyo-denshi/tk))
 759         "(%-18s . %08d)\t; %c")
 760        ((>= (charset-dimension name) 2)
 761         "(%-18s . #x%04X)\t; %c")
 762        (t
 763         "(%-18s . #x%02X)\t; %c"))
 764       name
 765       (if (= (charset-iso-graphic-plane name) 1)
 766           (logior value
 767                   (cond ((= (charset-dimension name) 1)
 768                          #x80)
 769                         ((= (charset-dimension name) 2)
 770                          #x8080)
 771                         ((= (charset-dimension name) 3)
 772                          #x808080)
 773                         (t 0)))
 774         value)
 775       (char-db-decode-isolated-char name value)))
 776     (if (and (= (charset-chars name) 94)
 777              (= (charset-dimension name) 2))
 778         (insert (format " [%02d-%02d]"
 779                         (- (lsh value -8) 32)
 780                         (- (logand value 255) 32))))
 781     )
 782    (t
 783     (insert (format "(%-18s . %s)" name value))
 784     ))
 785   (insert line-breaking))
 786
 787 (defun char-db-insert-relation-feature (char name value line-breaking
 788                                              ccss readable)
 789   (insert (format "(%-18s%s " name line-breaking))
 790   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 791         separator cell sources required-features
 792         ret)
 793     (while (consp value)
 794       (setq cell (car value))
 795       (if (integerp cell)
 796           (setq cell (decode-char '=ucs cell)))
 797       (cond
 798        ((eq name '->subsumptive)
 799         (when (or (not (some (lambda (atr)
 800                                (get-char-attribute cell atr))
 801                              char-db-ignored-attributes))
 802                   (some (lambda (ccs)
 803                           (encode-char cell ccs 'defined-only))
 804                         ccss))
 805           (if separator
 806               (insert lbs))
 807           (let ((char-db-ignored-attributes
 808                  (cons '<-subsumptive
 809                        char-db-ignored-attributes)))
 810             (insert-char-attributes cell readable nil nil 'for-sub-node))
 811           (setq separator lbs))
 812         )
 813        ((characterp cell)
 814         (setq sources
 815               (get-char-attribute
 816                char (intern (format "%s*sources" name))))
 817         (setq required-features nil)
 818         (dolist (source sources)
 819           (cond
 820            ((memq source '(JP
 821                            JP/Jouyou
 822                            shinjigen shinjigen@1ed shinjigen@rev))
 823             (setq required-features
 824                   (union required-features
 825                          '(=jis-x0208
 826                            =jis-x0208@1990
 827                            =jis-x0213-1@2000
 828                            =jis-x0213-1@2004
 829                            =jis-x0213-2
 830                            =jis-x0212
 831                            =jis-x0208@1983
 832                            =jis-x0208@1978
 833                            =shinjigen
 834                            =shinjigen@1ed
 835                            =shinjigen@rev
 836                            =shinjigen/+p@rev))))
 837            ((eq source 'CN)
 838             (setq required-features
 839                   (union required-features
 840                          '(=gb2312
 841                            =gb12345
 842                            =iso-ir165)))))
 843           (cond
 844            ((find-charset (setq ret (intern (format "=%s" source))))
 845             (setq required-features
 846                   (cons ret required-features)))
 847            (t (setq required-features
 848                     (cons source required-features)))))
 849         (cond ((string-match "@JP" (symbol-name name))
 850                (setq required-features
 851                      (union required-features
 852                             '(=jis-x0208
 853                               =jis-x0208@1990
 854                               =jis-x0213-1-2000
 855                               =jis-x0213-2-2000
 856                               =jis-x0212
 857                               =jis-x0208@1983
 858                               =jis-x0208@1978))))
 859               ((string-match "@CN" (symbol-name name))
 860                (setq required-features
 861                      (union required-features
 862                             '(=gb2312
 863                               =gb12345
 864                               =iso-ir165)))))
 865         (if separator
 866             (insert lbs))
 867         (if readable
 868             (insert (format "%S" cell))
 869           (char-db-insert-char-spec cell readable
 870                                     nil
 871                                     required-features))
 872         (setq separator lbs))
 873        ((consp cell)
 874         (if separator
 875             (insert lbs))
 876         (if (consp (car cell))
 877             (char-db-insert-char-spec cell readable)
 878           (char-db-insert-char-reference cell readable))
 879         (setq separator lbs))
 880        (t
 881         (if separator
 882             (insert separator))
 883         (insert (prin1-to-string cell))
 884         (setq separator " ")))
 885       (setq value (cdr value)))
 886     (insert ")")
 887     (insert line-breaking)))
 888
 889 (defun insert-char-attributes (char &optional readable attributes column
 890                                     for-sub-node)
 891   (unless column
 892     (setq column (current-column)))
 893   (let (name value ; has-long-ccs-name
 894         rest
 895         radical strokes
 896         (line-breaking
 897          (concat "\n" (make-string (1+ column) ?\ )))
 898         lbs cell separator ret
 899         key al cal
 900         dest-ccss ; sources required-features
 901         ccss)
 902     (let (atr-d)
 903       (setq attributes
 904             (sort (if attributes
 905                       (if (consp attributes)
 906                           (progn
 907                             (dolist (name attributes)
 908                               (unless (memq name char-db-ignored-attributes)
 909                                 (if (find-charset name)
 910                                     (push name ccss))
 911                                 (push name atr-d)))
 912                             atr-d))
 913                     (dolist (name (char-attribute-list))
 914                       (unless (memq name char-db-ignored-attributes)
 915                         (if (find-charset name)
 916                             (push name ccss))
 917                         (push name atr-d)))
 918                     atr-d)
 919                   #'char-attribute-name<)))
 920     (insert "(")
 921     (when (memq '<-subsumptive attributes)
 922       (when (or readable (not for-sub-node))
 923         (when (setq value (get-char-attribute char '<-subsumptive))
 924           (char-db-insert-relation-feature char '<-subsumptive value
 925                                            line-breaking
 926                                            ccss readable)))
 927       (setq attributes (delq '<-subsumptive attributes)))
 928     (when (and (memq '<-denotational attributes)
 929                (setq value (get-char-attribute char '<-denotational)))
 930       (char-db-insert-relation-feature char '<-denotational value
 931                                        line-breaking
 932                                        ccss readable)
 933       (setq attributes (delq '<-denotational attributes)))
 934     (when (and (memq '<-denotational@component attributes)
 935                (setq value (get-char-attribute char '<-denotational@component)))
 936       (char-db-insert-relation-feature char '<-denotational@component value
 937                                        line-breaking
 938                                        ccss readable)
 939       (setq attributes (delq '<-denotational@component attributes)))
 940     (when (and (memq 'name attributes)
 941                (setq value (get-char-attribute char 'name)))
 942       (insert (format
 943                (if (> (+ (current-column) (length value)) 48)
 944                    "(name . %S)%s"
 945                  "(name               . %S)%s")
 946                value line-breaking))
 947       (setq attributes (delq 'name attributes))
 948       )
 949     (when (and (memq 'name* attributes)
 950                (setq value (get-char-attribute char 'name*)))
 951       (insert (format
 952                (if (> (+ (current-column) (length value)) 48)
 953                    "(name* . %S)%s"
 954                  "(name*              . %S)%s")
 955                value line-breaking))
 956       (setq attributes (delq 'name* attributes))
 957       )
 958     (when (and (memq 'script attributes)
 959                (setq value (get-char-attribute char 'script)))
 960       (insert (format "(script\t\t%s)%s"
 961                       (mapconcat (function prin1-to-string)
 962                                  value " ")
 963                       line-breaking))
 964       (setq attributes (delq 'script attributes))
 965       )
 966     (dolist (name '(=>ucs =>ucs*))
 967       (when (and (memq name attributes)
 968                  (setq value (get-char-attribute char name)))
 969         (insert (format "(%-18s . #x%04X)\t; %c%s"
 970                         name value (decode-char '=ucs value)
 971                         line-breaking))
 972         (setq attributes (delq name attributes))))
 973     (dolist (name '(=>ucs@gb =>ucs@big5))
 974       (when (and (memq name attributes)
 975                  (setq value (get-char-attribute char name)))
 976         (insert (format "(%-18s . #x%04X)\t; %c%s"
 977                         name value
 978                         (decode-char (intern
 979                                       (concat "="
 980                                               (substring
 981                                                (symbol-name name) 2)))
 982                                      value)
 983                         line-breaking))
 984         (setq attributes (delq name attributes))
 985         ))
 986     ;; (dolist (name '(=>daikanwa))
 987     ;;   (when (and (memq name attributes)
 988     ;;              (setq value (get-char-attribute char name)))
 989     ;;     (insert
 990     ;;      (if (integerp value)
 991     ;;          (format "(%-18s . %05d)\t; %c%s"
 992     ;;                  name value (decode-char '=daikanwa value)
 993     ;;                  line-breaking)
 994     ;;        (format "(%-18s %s)\t; %c%s"
 995     ;;                name
 996     ;;                (mapconcat (function prin1-to-string)
 997     ;;                           value " ")
 998     ;;                (char-representative-of-daikanwa char)
 999     ;;                line-breaking)))
1000     ;;     (setq attributes (delq name attributes))))
1001     (when (and (memq 'general-category attributes)
1002                (setq value (get-char-attribute char 'general-category)))
1003       (insert (format
1004                "(general-category\t%s) ; %s%s"
1005                (mapconcat (lambda (cell)
1006                             (format "%S" cell))
1007                           value " ")
1008                (cond ((rassoc value unidata-normative-category-alist)
1009                       "Normative Category")
1010                      ((rassoc value unidata-informative-category-alist)
1011                       "Informative Category")
1012                      (t
1013                       "Unknown Category"))
1014                line-breaking))
1015       (setq attributes (delq 'general-category attributes))
1016       )
1017     (when (and (memq 'bidi-category attributes)
1018                (setq value (get-char-attribute char 'bidi-category)))
1019       (insert (format "(bidi-category\t. %S)%s"
1020                       value
1021                       line-breaking))
1022       (setq attributes (delq 'bidi-category attributes))
1023       )
1024     (unless (or (not (memq 'mirrored attributes))
1025                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
1026                     'empty))
1027       (insert (format "(mirrored\t\t. %S)%s"
1028                       value
1029                       line-breaking))
1030       (setq attributes (delq 'mirrored attributes))
1031       )
1032     (cond
1033      ((and (memq 'decimal-digit-value attributes)
1034            (setq value (get-char-attribute char 'decimal-digit-value)))
1035       (insert (format "(decimal-digit-value . %S)%s"
1036                       value
1037                       line-breaking))
1038       (setq attributes (delq 'decimal-digit-value attributes))
1039       (when (and (memq 'digit-value attributes)
1040                  (setq value (get-char-attribute char 'digit-value)))
1041         (insert (format "(digit-value\t . %S)%s"
1042                         value
1043                         line-breaking))
1044         (setq attributes (delq 'digit-value attributes))
1045         )
1046       (when (and (memq 'numeric-value attributes)
1047                  (setq value (get-char-attribute char 'numeric-value)))
1048         (insert (format "(numeric-value\t . %S)%s"
1049                         value
1050                         line-breaking))
1051         (setq attributes (delq 'numeric-value attributes))
1052         )
1053       )
1054      (t
1055       (when (and (memq 'digit-value attributes)
1056                  (setq value (get-char-attribute char 'digit-value)))
1057         (insert (format "(digit-value\t. %S)%s"
1058                         value
1059                         line-breaking))
1060         (setq attributes (delq 'digit-value attributes))
1061         )
1062       (when (and (memq 'numeric-value attributes)
1063                  (setq value (get-char-attribute char 'numeric-value)))
1064         (insert (format "(numeric-value\t. %S)%s"
1065                         value
1066                         line-breaking))
1067         (setq attributes (delq 'numeric-value attributes))
1068         )))
1069     (when (and (memq 'iso-10646-comment attributes)
1070                (setq value (get-char-attribute char 'iso-10646-comment)))
1071       (insert (format "(iso-10646-comment\t. %S)%s"
1072                       value
1073                       line-breaking))
1074       (setq attributes (delq 'iso-10646-comment attributes))
1075       )
1076     (when (and (memq 'morohashi-daikanwa attributes)
1077                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1078       (insert (format "(morohashi-daikanwa\t%s)%s"
1079                       (mapconcat (function prin1-to-string) value " ")
1080                       line-breaking))
1081       (setq attributes (delq 'morohashi-daikanwa attributes))
1082       )
1083     (setq radical nil
1084           strokes nil)
1085     (when (and (memq 'ideographic-radical attributes)
1086                (setq value (get-char-attribute char 'ideographic-radical)))
1087       (setq radical value)
1088       (insert (format "(ideographic-radical . %S)\t; %c%s"
1089                       radical
1090                       (ideographic-radical radical)
1091                       line-breaking))
1092       (setq attributes (delq 'ideographic-radical attributes))
1093       )
1094     (when (and (memq 'shuowen-radical attributes)
1095                (setq value (get-char-attribute char 'shuowen-radical)))
1096       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1097                       value
1098                       (shuowen-radical value)
1099                       line-breaking))
1100       (setq attributes (delq 'shuowen-radical attributes))
1101       )
1102     (let (key)
1103       (dolist (domain
1104                (append
1105                 char-db-feature-domains
1106                 (let (dest domain)
1107                   (dolist (feature (char-attribute-list))
1108                     (setq feature (symbol-name feature))
1109                     (when (string-match
1110                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1111                            feature)
1112                       (setq domain (intern (match-string 2 feature)))
1113                      (unless (memq domain dest)
1114                        (setq dest (cons domain dest)))))
1115                   (sort dest #'string<))))
1116         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1117         (when (and (memq key attributes)
1118                    (setq value (get-char-attribute char key)))
1119           (setq radical value)
1120           (insert (format "(%s . %S)\t; %c%s"
1121                           key
1122                           radical
1123                           (ideographic-radical radical)
1124                           line-breaking))
1125           (setq attributes (delq key attributes))
1126           )
1127         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1128         (when (and (memq key attributes)
1129                    (setq value (get-char-attribute char key)))
1130           (setq strokes value)
1131           (insert (format "(%s . %S)%s"
1132                           key
1133                           strokes
1134                           line-breaking))
1135           (setq attributes (delq key attributes))
1136           )
1137         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1138         (when (and (memq key attributes)
1139                    (setq value (get-char-attribute char key)))
1140           (insert (format "(%s       . %S)%s"
1141                           key
1142                           value
1143                           line-breaking))
1144           (setq attributes (delq key attributes))
1145           )
1146         (dolist (feature '(ideographic-radical
1147                            ideographic-strokes
1148                            total-strokes))
1149           (setq key (intern (format "%s@%s*sources" feature domain)))
1150           (when (and (memq key attributes)
1151                      (setq value (get-char-attribute char key)))
1152             (insert (format "(%s%s" key line-breaking))
1153             (dolist (cell value)
1154               (insert (format " %s" cell)))
1155             (insert ")")
1156             (insert line-breaking)
1157             (setq attributes (delq key attributes))
1158             ))
1159         ))
1160     (when (and (memq 'ideographic-strokes attributes)
1161                (setq value (get-char-attribute char 'ideographic-strokes)))
1162       (setq strokes value)
1163       (insert (format "(ideographic-strokes . %S)%s"
1164                       strokes
1165                       line-breaking))
1166       (setq attributes (delq 'ideographic-strokes attributes))
1167       )
1168     (when (and (memq 'kangxi-radical attributes)
1169                (setq value (get-char-attribute char 'kangxi-radical)))
1170       (unless (eq value radical)
1171         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1172                         value
1173                         (ideographic-radical value)
1174                         line-breaking))
1175         (or radical
1176             (setq radical value)))
1177       (setq attributes (delq 'kangxi-radical attributes))
1178       )
1179     (when (and (memq 'kangxi-strokes attributes)
1180                (setq value (get-char-attribute char 'kangxi-strokes)))
1181       (unless (eq value strokes)
1182         (insert (format "(kangxi-strokes\t . %S)%s"
1183                         value
1184                         line-breaking))
1185         (or strokes
1186             (setq strokes value)))
1187       (setq attributes (delq 'kangxi-strokes attributes))
1188       )
1189     ;; (when (and (memq 'japanese-radical attributes)
1190     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1191     ;;   (unless (eq value radical)
1192     ;;     (insert (format "(japanese-radical\t . %S)\t; %c%s"
1193     ;;                     value
1194     ;;                     (ideographic-radical value)
1195     ;;                     line-breaking))
1196     ;;     (or radical
1197     ;;         (setq radical value)))
1198     ;;   (setq attributes (delq 'japanese-radical attributes))
1199     ;;   )
1200     (when (and (memq 'japanese-strokes attributes)
1201                (setq value (get-char-attribute char 'japanese-strokes)))
1202       (unless (eq value strokes)
1203         (insert (format "(japanese-strokes\t . %S)%s"
1204                         value
1205                         line-breaking))
1206         (or strokes
1207             (setq strokes value)))
1208       (setq attributes (delq 'japanese-strokes attributes))
1209       )
1210     (when (and (memq 'cns-radical attributes)
1211                (setq value (get-char-attribute char 'cns-radical)))
1212       (insert (format "(cns-radical\t . %S)\t; %c%s"
1213                       value
1214                       (ideographic-radical value)
1215                       line-breaking))
1216       (setq attributes (delq 'cns-radical attributes))
1217       )
1218     (when (and (memq 'cns-strokes attributes)
1219                (setq value (get-char-attribute char 'cns-strokes)))
1220       (unless (eq value strokes)
1221         (insert (format "(cns-strokes\t . %S)%s"
1222                         value
1223                         line-breaking))
1224         (or strokes
1225             (setq strokes value)))
1226       (setq attributes (delq 'cns-strokes attributes))
1227       )
1228     ;; (when (and (memq 'shinjigen-1-radical attributes)
1229     ;;            (setq value (get-char-attribute char 'shinjigen-1-radical)))
1230     ;;   (unless (eq value radical)
1231     ;;     (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
1232     ;;                     value
1233     ;;                     (ideographic-radical value)
1234     ;;                     line-breaking))
1235     ;;     (or radical
1236     ;;         (setq radical value)))
1237     ;;   (setq attributes (delq 'shinjigen-1-radical attributes))
1238     ;;   )
1239     ;; (when (and (memq 'ideographic- attributes)
1240     ;;            (setq value (get-char-attribute char 'ideographic-)))
1241     ;;   (insert "(ideographic-       ")
1242     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1243     ;;         separator nil)
1244     ;;   (while (consp value)
1245     ;;     (setq cell (car value))
1246     ;;     (if (integerp cell)
1247     ;;         (setq cell (decode-char '=ucs cell)))
1248     ;;     (cond ((characterp cell)
1249     ;;            (if separator
1250     ;;                (insert lbs))
1251     ;;            (if readable
1252     ;;                (insert (format "%S" cell))
1253     ;;              (char-db-insert-char-spec cell readable))
1254     ;;            (setq separator lbs))
1255     ;;           ((consp cell)
1256     ;;            (if separator
1257     ;;                (insert lbs))
1258     ;;            (if (consp (car cell))
1259     ;;                (char-db-insert-char-spec cell readable)
1260     ;;              (char-db-insert-char-reference cell readable))
1261     ;;            (setq separator lbs))
1262     ;;           (t
1263     ;;            (if separator
1264     ;;                (insert separator))
1265     ;;            (insert (prin1-to-string cell))
1266     ;;            (setq separator " ")))
1267     ;;     (setq value (cdr value)))
1268     ;;   (insert ")")
1269     ;;   (insert line-breaking)
1270     ;;   (setq attributes (delq 'ideographic- attributes)))
1271     (when (and (memq 'total-strokes attributes)
1272                (setq value (get-char-attribute char 'total-strokes)))
1273       (insert (format "(total-strokes       . %S)%s"
1274                       value
1275                       line-breaking))
1276       (setq attributes (delq 'total-strokes attributes))
1277       )
1278     ;; (when (and (memq '->ideograph attributes)
1279     ;;            (setq value (get-char-attribute char '->ideograph)))
1280     ;;   (insert (format "(->ideograph\t%s)%s"
1281     ;;                   (mapconcat (lambda (code)
1282     ;;                                (cond ((symbolp code)
1283     ;;                                       (symbol-name code))
1284     ;;                                      ((integerp code)
1285     ;;                                       (format "#x%04X" code))
1286     ;;                                      (t
1287     ;;                                       (format "%s %S"
1288     ;;                                               line-breaking code))))
1289     ;;                              value " ")
1290     ;;                   line-breaking))
1291     ;;   (setq attributes (delq '->ideograph attributes))
1292     ;;   )
1293     ;; (when (and (memq '->decomposition attributes)
1294     ;;            (setq value (get-char-attribute char '->decomposition)))
1295     ;;   (insert (format "(->decomposition\t%s)%s"
1296     ;;                   (mapconcat (lambda (code)
1297     ;;                                (cond ((symbolp code)
1298     ;;                                       (symbol-name code))
1299     ;;                                      ((characterp code)
1300     ;;                                       (if readable
1301     ;;                                           (format "%S" code)
1302     ;;                                         (format "#x%04X"
1303     ;;                                                 (char-int code))
1304     ;;                                         ))
1305     ;;                                      ((integerp code)
1306     ;;                                       (format "#x%04X" code))
1307     ;;                                      (t
1308     ;;                                       (format "%s%S" line-breaking code))))
1309     ;;                              value " ")
1310     ;;                   line-breaking))
1311     ;;   (setq attributes (delq '->decomposition attributes))
1312     ;;   )
1313     (if (equal (get-char-attribute char '->titlecase)
1314                (get-char-attribute char '->uppercase))
1315         (setq attributes (delq '->titlecase attributes)))
1316     (when (and (memq '->mojikyo attributes)
1317                (setq value (get-char-attribute char '->mojikyo)))
1318       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1319                       value (decode-char 'mojikyo value)
1320                       line-breaking))
1321       (setq attributes (delq '->mojikyo attributes))
1322       )
1323     ;; (when (and (memq 'hanyu-dazidian-vol attributes)
1324     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1325     ;;   (insert (format "(hanyu-dazidian-vol  . %d)%s"
1326     ;;                   value line-breaking))
1327     ;;   (setq attributes (delq 'hanyu-dazidian-vol attributes))
1328     ;;   )
1329     ;; (when (and (memq 'hanyu-dazidian-page attributes)
1330     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1331     ;;   (insert (format "(hanyu-dazidian-page . %d)%s"
1332     ;;                   value line-breaking))
1333     ;;   (setq attributes (delq 'hanyu-dazidian-page attributes))
1334     ;;   )
1335     ;; (when (and (memq 'hanyu-dazidian-char attributes)
1336     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1337     ;;   (insert (format "(hanyu-dazidian-char . %d)%s"
1338     ;;                   value line-breaking))
1339     ;;   (setq attributes (delq 'hanyu-dazidian-char attributes))
1340     ;;   )
1341     (unless readable
1342       (dolist (ignored '(composition
1343                          ->denotational <-subsumptive ->ucs-unified
1344                          ->ideographic-component-forms))
1345         (setq attributes (delq ignored attributes))))
1346     (while attributes
1347       (setq name (car attributes))
1348       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1349                   'value-is-empty)
1350         (cond ((setq ret (find-charset name))
1351                (setq name (charset-name ret))
1352                (when (not (memq name dest-ccss))
1353                  (setq dest-ccss (cons name dest-ccss))
1354                  (char-db-insert-ccs-feature name value line-breaking))
1355                )
1356               ((string-match "^=>ucs@" (symbol-name name))
1357                (insert (format "(%-18s . #x%04X)\t; %c%s"
1358                                name value (decode-char '=ucs value)
1359                                line-breaking))
1360                )
1361               ((eq name 'jisx0208-1978/4X)
1362                (insert (format "(%-18s . #x%04X)%s"
1363                                name value
1364                                line-breaking))
1365                )
1366               ((and
1367                 (not readable)
1368                 (not (eq name '->subsumptive))
1369                 (not (eq name '->uppercase))
1370                 (not (eq name '->lowercase))
1371                 (not (eq name '->titlecase))
1372                 (not (eq name '->canonical))
1373                 (not (eq name '->Bopomofo))
1374                 (not (eq name '->mistakable))
1375                 (not (eq name '->ideographic-variants))
1376                 (null (get-char-attribute
1377                        char (intern (format "%s*sources" name))))
1378                 (not (string-match "\\*sources$" (symbol-name name)))
1379                 (null (get-char-attribute
1380                        char (intern (format "%s*note" name))))
1381                 (not (string-match "\\*note$" (symbol-name name)))
1382                 (or (eq name '<-identical)
1383                     (eq name '<-uppercase)
1384                     (eq name '<-lowercase)
1385                     (eq name '<-titlecase)
1386                     (eq name '<-canonical)
1387                     (eq name '<-ideographic-variants)
1388                     ;; (eq name '<-synonyms)
1389                     (string-match "^<-synonyms" (symbol-name name))
1390                     (eq name '<-mistakable)
1391                     (when (string-match "^->" (symbol-name name))
1392                       (cond
1393                        ((string-match "^->fullwidth" (symbol-name name))
1394                         (not (and (consp value)
1395                                   (characterp (car value))
1396                                   (encode-char
1397                                    (car value) '=ucs 'defined-only)))
1398                         )
1399                        (t)))
1400                     ))
1401                )
1402               ((or (eq name 'ideographic-structure)
1403                    (eq name 'ideographic-combination)
1404                    ;; (eq name 'ideographic-)
1405                    (eq name '=decomposition)
1406                    (char-feature-base-name= '=decomposition name)
1407                    (char-feature-base-name= '=>decomposition name)
1408                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1409                    ;;               (symbol-name name))
1410                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1411                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1412                                  (symbol-name name))
1413                    )
1414                (char-db-insert-relation-feature char name value
1415                                                 line-breaking
1416                                                 ccss readable))
1417               ;; ((memq name '(ideograph=
1418               ;;               original-ideograph-of
1419               ;;               ancient-ideograph-of
1420               ;;               vulgar-ideograph-of
1421               ;;               wrong-ideograph-of
1422               ;;               ;; simplified-ideograph-of
1423               ;;               ideographic-variants
1424               ;;               ;; ideographic-different-form-of
1425               ;;               ))
1426               ;;  (insert (format "(%-18s%s " name line-breaking))
1427               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1428               ;;        separator nil)
1429               ;;  (while (consp value)
1430               ;;    (setq cell (car value))
1431               ;;    (if (and (consp cell)
1432               ;;             (consp (car cell)))
1433               ;;        (progn
1434               ;;          (if separator
1435               ;;              (insert lbs))
1436               ;;          (char-db-insert-alist cell readable)
1437               ;;          (setq separator lbs))
1438               ;;      (if separator
1439               ;;          (insert separator))
1440               ;;      (insert (prin1-to-string cell))
1441               ;;      (setq separator " "))
1442               ;;    (setq value (cdr value)))
1443               ;;  (insert ")")
1444               ;;  (insert line-breaking))
1445               ((consp value)
1446                (insert (format "(%-18s " name))
1447                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1448                      separator nil)
1449                (while (consp value)
1450                  (setq cell (car value))
1451                  (if (and (consp cell)
1452                           (consp (car cell))
1453                           (setq ret (condition-case nil
1454                                         (find-char cell)
1455                                       (error nil))))
1456                      (progn
1457                        (setq rest cell
1458                              al nil
1459                              cal nil)
1460                        (while rest
1461                          (setq key (car (car rest)))
1462                          (if (find-charset key)
1463                              (setq cal (cons key cal))
1464                            (setq al (cons key al)))
1465                          (setq rest (cdr rest)))
1466                        (if separator
1467                            (insert lbs))
1468                        (insert-char-attributes ret
1469                                                readable
1470                                                al ; cal
1471                                                nil 'for-sub-node)
1472                        (setq separator lbs))
1473                    (setq ret (prin1-to-string cell))
1474                    (if separator
1475                        (if (< (+ (current-column)
1476                                  (length ret)
1477                                  (length separator))
1478                               76)
1479                            (insert separator)
1480                          (insert lbs)))
1481                    (insert ret)
1482                    (setq separator " "))
1483                  (setq value (cdr value)))
1484                (insert ")")
1485                (insert line-breaking))
1486               (t
1487                (insert (format "(%-18s" name))
1488                (setq ret (prin1-to-string value))
1489                (unless (< (+ (current-column)
1490                              (length ret)
1491                              3)
1492                           76)
1493                  (insert line-breaking))
1494                (insert " . " ret ")" line-breaking)
1495                ;; (insert (format "(%-18s . %S)%s"
1496                ;;                 name value
1497                ;;                 line-breaking))
1498                )
1499               ))
1500       (setq attributes (cdr attributes)))
1501     (insert ")")))
1502
1503 (defun insert-char-data (char &optional readable
1504                               attributes)
1505   (save-restriction
1506     (narrow-to-region (point)(point))
1507     (insert "(define-char
1508   '")
1509     (insert-char-attributes char readable attributes)
1510     (insert ")\n")
1511     (goto-char (point-min))
1512     (while (re-search-forward "[ \t]+$" nil t)
1513       (replace-match ""))
1514     ;; from tabify.
1515     (goto-char (point-min))
1516     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1517       (let ((column (current-column))
1518             (indent-tabs-mode t))
1519         (delete-region (match-beginning 0) (point))
1520         (indent-to column)))
1521     (goto-char (point-max))
1522     ;; (tabify (point-min)(point-max))
1523     ))
1524
1525 (defun insert-char-data-with-variant (char &optional printable
1526                                            no-ucs-unified
1527                                            script excluded-script)
1528   (insert-char-data char printable)
1529   (let ((variants (char-variants char))
1530         rest
1531         variant vs ret)
1532     (setq variants (sort variants #'<))
1533     (setq rest variants)
1534     (setq variants (cons char variants))
1535     (while rest
1536       (setq variant (car rest))
1537       (unless (get-char-attribute variant '<-subsumptive)
1538         (if (and (or (null script)
1539                      (null (setq vs (get-char-attribute variant 'script)))
1540                      (memq script vs))
1541                  (or (null excluded-script)
1542                      (null (setq vs (get-char-attribute variant 'script)))
1543                      (not (memq excluded-script vs))))
1544             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1545               (insert-char-data variant printable)
1546               (if (setq ret (char-variants variant))
1547                   (while ret
1548                     (or (memq (car ret) variants)
1549                         ;; (get-char-attribute (car ret) '<-subsumptive)
1550                         (setq rest (nconc rest (list (car ret)))))
1551                     (setq ret (cdr ret)))))))
1552       (setq rest (cdr rest)))))
1553
1554 (defun insert-char-range-data (min max &optional script excluded-script)
1555   (let ((code min)
1556         char)
1557     (while (<= code max)
1558       (setq char (decode-char '=ucs code))
1559       (if (encode-char char '=ucs 'defined-only)
1560           (insert-char-data-with-variant char nil 'no-ucs-unified
1561                                          script excluded-script))
1562       (setq code (1+ code)))))
1563
1564 (defun write-char-range-data-to-file (min max file
1565                                           &optional script excluded-script)
1566   (let ((coding-system-for-write char-db-file-coding-system))
1567     (with-temp-buffer
1568       (insert (format ";; -*- coding: %s -*-\n"
1569                       char-db-file-coding-system))
1570       (insert-char-range-data min max script excluded-script)
1571       (write-region (point-min)(point-max) file))))
1572
1573 (defvar what-character-original-window-configuration)
1574
1575 ;;;###autoload
1576 (defun what-char-definition (char)
1577   (interactive (list (char-after)))
1578   (let ((buf (get-buffer-create "*Character Description*"))
1579         (the-buf (current-buffer))
1580         (win-conf (current-window-configuration)))
1581     (pop-to-buffer buf)
1582     (make-local-variable 'what-character-original-window-configuration)
1583     (setq what-character-original-window-configuration win-conf)
1584     (setq buffer-read-only nil)
1585     (erase-buffer)
1586     (condition-case err
1587         (progn
1588           (insert-char-data-with-variant char 'printable)
1589           (unless (char-attribute-alist char)
1590             (insert (format ";; = %c\n"
1591                             (let* ((rest (split-char char))
1592                                    (ccs (pop rest))
1593                                    (code (pop rest)))
1594                               (while rest
1595                                 (setq code (logior (lsh code 8)
1596                                                    (pop rest))))
1597                               (decode-char ccs code)))))
1598           ;; (char-db-update-comment)
1599           (set-buffer-modified-p nil)
1600           (view-mode the-buf (lambda (buf)
1601                                (set-window-configuration
1602                                 what-character-original-window-configuration)
1603                                ))
1604           (goto-char (point-min)))
1605       (error (progn
1606                (set-window-configuration
1607                 what-character-original-window-configuration)
1608                (signal (car err) (cdr err)))))))
1609
1610
1611 ;;; @ end
1612 ;;;
1613
1614 (provide 'char-db-util)
1615
1616 ;;; char-db-util.el ends here