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 ;;   2008,2009,2010,2011,2012,2013,2014,2015,2016,2017,2018,2019 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 ?𧮫 ?只 ?㕯 ; 050
  72    ?句 ?丩 ?古 ?十 ?卅 ?言 ?誩 ?音 ?䇂 ?丵 ; 060
  73    ?菐 ?廾 ?𠬜 ?共 ?異 ?舁 ?𦥑 ?䢅 ?爨 ?革 ; 070
  74    ?鬲 ?䰜 ?爪 ?𠃨 ?鬥 ?又 ?𠂇 ?史 ?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    ?雈 ?𦫳 ?𥄕 ?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 ?丯 ; 140
  81    ?耒 ?角 ?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 ?𠧪 ?齊 ?朿 ?片 ?鼎 ; 250
  93    ?克 ?彔 ?禾 ?秝 ?黍 ?香 ?米 ?毇 ?臼 ?凶 ; 260
  94    ?𣎳 ?𣏟 ?麻 ?尗 ?耑 ?韭 ?瓜 ?瓠 ?宀 ?宮 ; 270
  95    ?呂 ?穴 ?㝱 ?疒 ?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    ?夨 ?夭 ?交 ?尣 ?壺 ?壹 ?㚔 ?奢 ?亢 ?夲 ; 400
 108    ])
 109
 110 (defun shuowen-radical (number)
 111   (aref shuowen-radicals (1- number)))
 112
 113 (defvar char-db-file-coding-system 'utf-8-mcs-er)
 114
 115 (defvar char-db-ignored-attributes '(ideographic-products))
 116
 117 (defvar char-db-coded-charset-priority-list
 118   '(ascii
 119     control-1
 120     latin-iso8859-1
 121     latin-iso8859-2
 122     latin-iso8859-3
 123     latin-iso8859-4
 124     latin-iso8859-9
 125     latin-jisx0201
 126     cyrillic-iso8859-5
 127     greek-iso8859-7
 128     thai-tis620
 129     ;; =mj
 130     ;; =adobe-japan1-0
 131     ;; =adobe-japan1-1
 132     ;; =adobe-japan1-2
 133     ;; =adobe-japan1-3
 134     ;; =adobe-japan1-4
 135     ;; =adobe-japan1-5
 136     ;; =adobe-japan1-6
 137     =jis-x0208
 138     =jis-x0208@1978
 139     =jis-x0208@1983
 140     japanese-jisx0212
 141     chinese-gb2312
 142     =jis-x0208@1990
 143     chinese-cns11643-1
 144     chinese-cns11643-2
 145     chinese-cns11643-3
 146     chinese-cns11643-4
 147     chinese-cns11643-5
 148     chinese-cns11643-6
 149     chinese-cns11643-7
 150     =jis-x0213-1
 151     =jis-x0213-1@2000
 152     =jis-x0213-1@2004
 153     =jis-x0213-2
 154     korean-ksc5601
 155     chinese-isoir165
 156     katakana-jisx0201
 157     hebrew-iso8859-8
 158     chinese-gb12345
 159     latin-viscii
 160     ethiopic-ucs
 161     =big5-cdp
 162     =hanyo-denshi/ja
 163     =hanyo-denshi/jb
 164     =hanyo-denshi/jc
 165     =hanyo-denshi/jd
 166     =hanyo-denshi/ft
 167     =hanyo-denshi/ia
 168     =hanyo-denshi/ib
 169     =hanyo-denshi/hg
 170     =hanyo-denshi/jt
 171     =hanyo-denshi/ks
 172     =hanyo-denshi/tk
 173     ;; ==mj
 174     ;; ==adobe-japan1-0
 175     ;; ==adobe-japan1-1
 176     ;; ==adobe-japan1-2
 177     ;; ==adobe-japan1-3
 178     ;; ==adobe-japan1-4
 179     ;; ==adobe-japan1-5
 180     ;; ==adobe-japan1-6
 181     ==jis-x0208
 182     ==jis-x0213-1
 183     ==jis-x0213-2
 184     ==jis-x0212
 185     ==hanyo-denshi/ja
 186     ==hanyo-denshi/jb
 187     ==hanyo-denshi/jc
 188     ==hanyo-denshi/ft
 189     ==hanyo-denshi/ia
 190     ==hanyo-denshi/ib
 191     ==hanyo-denshi/hg
 192     ==hanyo-denshi/jt
 193     ==hanyo-denshi/ks
 194     ==hanyo-denshi/tk
 195     =gt
 196     =gt-k
 197     =daikanwa
 198     =daijiten
 199     =daikanwa@rev2
 200     =daikanwa@rev1
 201     =daikanwa/+p
 202     ==daikanwa
 203     ==daijiten
 204     =cbeta
 205     ideograph-hanziku-1
 206     ideograph-hanziku-2
 207     ideograph-hanziku-3
 208     ideograph-hanziku-4
 209     ideograph-hanziku-5
 210     ideograph-hanziku-6
 211     ideograph-hanziku-7
 212     ideograph-hanziku-8
 213     ideograph-hanziku-9
 214     ideograph-hanziku-10
 215     ideograph-hanziku-11
 216     ideograph-hanziku-12
 217     ;; =>>>adobe-japan1-0
 218     ;; =>>>adobe-japan1-1
 219     ;; =>>>adobe-japan1-2
 220     ;; =>>>adobe-japan1-3
 221     ;; =>>>adobe-japan1-4
 222     ;; =>>>adobe-japan1-5
 223     ;; =>>>adobe-japan1-6
 224     ;; =>>>jis-x0208
 225     ;; =>>>jis-x0213-1
 226     ;; =>>>jis-x0213-2
 227     ;; =>>>hanyo-denshi/ja
 228     ;; =>>>hanyo-denshi/jb
 229     ;; =>>>hanyo-denshi/jc
 230     ;; =>>>hanyo-denshi/ft
 231     ;; =>>>hanyo-denshi/ib
 232     ;; =>>>hanyo-denshi/hg
 233     ;; =>>>hanyo-denshi/jt
 234     ;; =>>>hanyo-denshi/ks
 235     ;; =>>>gt
 236     =>>adobe-japan1-0
 237     =>>adobe-japan1-1
 238     =>>adobe-japan1-2
 239     =>>adobe-japan1-3
 240     =>>adobe-japan1-4
 241     =>>adobe-japan1-5
 242     =>>adobe-japan1-6
 243     =>>jis-x0208
 244     =>>jis-x0213-1
 245     =>>jis-x0213-1@2000
 246     =>>jis-x0213-1@2004
 247     =>>jis-x0213-2
 248     =>>jis-x0208@1978
 249     =>>hanyo-denshi/ft
 250     =>>hanyo-denshi/jt
 251     =>>hanyo-denshi/ks
 252     =>>gt
 253     =>>daikanwa
 254     =>>cbeta
 255     =+>jis-x0208
 256     =+>jis-x0213-1
 257     =+>jis-x0213-2
 258     =+>adobe-japan1-0
 259     =+>adobe-japan1-1
 260     =+>adobe-japan1-2
 261     =+>adobe-japan1-3
 262     =+>adobe-japan1-4
 263     =+>adobe-japan1-5
 264     =+>adobe-japan1-6
 265     =+>jis-x0208@1978
 266     =>jis-x0208
 267     =>jis-x0208@1997
 268     =>jis-x0213-1
 269     =>jis-x0213-1@2000
 270     =>jis-x0213-1@2004
 271     =>jis-x0213-2
 272     ==>ucs@bucs
 273     ==>daijiten
 274     =>iwds-1
 275     ;; =>ucs@hanyo-denshi
 276     =>ucs@iso
 277     =>ucs@unicode
 278     =>ucs@jis
 279     =>ucs@cns
 280     =>ucs@ks
 281     =+>ucs@iso
 282     =+>ucs@unicode
 283     =+>ucs@jis
 284     =+>ucs@jis/1990
 285     =+>ucs@cns
 286     =+>ucs@ks
 287     =>>ucs@iso
 288     =>>ucs@unicode
 289     =>>ucs@jis
 290     =>>ucs@cns
 291     =>>>ucs@iso
 292     =>>>ucs@unicode
 293     ==ucs@iso
 294     ==ucs@unicode
 295     ;; ==ucs@cns
 296     ==gb2312
 297     ==ks-x1001
 298     ==cns11643-1
 299     ==cns11643-2
 300     ==cns11643-3
 301     ==cns11643-4
 302     ==cns11643-5
 303     ==cns11643-6
 304     ==cns11643-7
 305     ==gt
 306     ==jis-x0208@1990
 307     ;; ==jis-x0208@1983
 308     ==jis-x0208@1978
 309     ==gt-k
 310     =ucs@iso
 311     =ucs@unicode
 312     =ucs@cns
 313     ==big5-cdp
 314     ==cbeta
 315     =>>big5-cdp
 316     =>>gt-k
 317     =+>gt
 318     =+>big5-cdp
 319     =>gt
 320     =>mj
 321     =>big5-cdp
 322     =>daikanwa
 323     =>daikanwa/ho
 324     =>cns11643-5
 325     =>cns11643-7
 326     =big5
 327     =big5-eten
 328     =>gt-k
 329     =zinbun-oracle
 330     =>zinbun-oracle
 331     =ruimoku-v6
 332     =>>ruimoku-v6
 333     ==ruimoku-v6
 334     =jef-china3
 335     =>cbeta
 336     =shinjigen
 337     =ucs-var-001
 338     =ucs-var-002
 339     =ucs-var-003
 340     =ucs-var-004
 341     =ucs-var-005
 342     =ucs-var-006
 343     =ucs-var-008
 344     =ucs-var-010
 345     =ucs-itaiji-001
 346     =ucs-itaiji-002
 347     =ucs-itaiji-003
 348     =ucs-itaiji-004
 349     =ucs-itaiji-005
 350     =ucs-itaiji-006
 351     =ucs-itaiji-007
 352     =ucs-itaiji-008
 353     =ucs-itaiji-009
 354     =ucs-itaiji-010
 355     =ucs-itaiji-011
 356     =ucs-itaiji-012
 357     =>ucs-itaiji-001
 358     =>ucs-itaiji-002
 359     =>ucs-itaiji-003
 360     =>ucs-itaiji-004
 361     =>ucs-itaiji-005
 362     =>ucs-itaiji-006
 363     =>ucs-itaiji-007
 364     =>ucs-itaiji-009
 365     =big5-cdp-var-001
 366     =big5-cdp-var-002
 367     =big5-cdp-var-003
 368     =big5-cdp-var-004
 369     =big5-cdp-var-005
 370     =big5-cdp-var-010
 371     =big5-cdp-itaiji-001
 372     =big5-cdp-itaiji-002
 373     =big5-cdp-itaiji-003
 374     =>big5-cdp-itaiji-001
 375     =>big5-cdp-var-001
 376     =>ucs@iwds-1
 377     =>ucs@cognate
 378     =>ucs@component
 379     =>ucs-itaiji-001@iwds-1
 380     =>big5-cdp@iwds-1
 381     =>big5-cdp@component
 382     =>big5-cdp@cognate
 383     ==ucs@gb
 384     =ucs@gb
 385     ==ucs-var-002
 386     =ucs@JP/hanazono))
 387
 388
 389 ;;; @ char-db formatters
 390 ;;;
 391
 392 (defun char-db-make-char-spec (char)
 393   (let (ret char-spec)
 394     (cond ((characterp char)
 395            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 396                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 397                   (setq char-spec (list (cons '=ucs ret)))
 398                   (cond ((setq ret (get-char-attribute char 'name))
 399                          (setq char-spec (cons (cons 'name ret) char-spec))
 400                          )
 401                         ((setq ret (get-char-attribute char 'name*))
 402                          (setq char-spec (cons (cons 'name* ret) char-spec))
 403                          ))
 404                   )
 405                  ((encode-char char '=mj 'defined-only)
 406                   (setq char-spec nil)
 407                   (dolist (ccs (charset-list))
 408                     (if (and (or (eq ccs '=mj)
 409                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 410                                  (string-match "=ucs@" (symbol-name ccs))
 411                                  )
 412                              (setq ccs (charset-name ccs))
 413                              (null (assq ccs char-spec))
 414                              (setq ret (encode-char char ccs 'defined-only)))
 415                         (setq char-spec (cons (cons ccs ret) char-spec))))
 416                   )
 417                  ((encode-char char '==mj 'defined-only)
 418                   (setq char-spec nil)
 419                   (dolist (ccs (charset-list))
 420                     (if (and (or (eq ccs '==mj)
 421                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 422                                  (string-match "=ucs@" (symbol-name ccs))
 423                                  )
 424                              (setq ccs (charset-name ccs))
 425                              (null (assq ccs char-spec))
 426                              (setq ret (encode-char char ccs 'defined-only)))
 427                         (setq char-spec (cons (cons ccs ret) char-spec))))
 428                   )
 429                  ((encode-char char '=adobe-japan1 'defined-only)
 430                   (setq char-spec nil)
 431                   (dolist (ccs (charset-list))
 432                     (if (and (or (memq ccs
 433                                        '(=adobe-japan1-0
 434                                          =adobe-japan1-1
 435                                          =adobe-japan1-2
 436                                          =adobe-japan1-3
 437                                          =adobe-japan1-4
 438                                          =adobe-japan1-5
 439                                          =adobe-japan1-6
 440                                          ))
 441                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 442                                  (string-match "=ucs@" (symbol-name ccs))
 443                                  )
 444                              (setq ccs (charset-name ccs))
 445                              (null (assq ccs char-spec))
 446                              (setq ret (encode-char char ccs 'defined-only)))
 447                         (setq char-spec (cons (cons ccs ret) char-spec))))
 448                   )
 449                  ((encode-char char '==adobe-japan1 'defined-only)
 450                   (setq char-spec nil)
 451                   (dolist (ccs (charset-list))
 452                     (if (and (or (memq ccs
 453                                        '(==adobe-japan1-0
 454                                          ==adobe-japan1-1
 455                                          ==adobe-japan1-2
 456                                          ==adobe-japan1-3
 457                                          ==adobe-japan1-4
 458                                          ==adobe-japan1-5
 459                                          ==adobe-japan1-6
 460                                          ))
 461                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 462                                  (string-match "=ucs@" (symbol-name ccs))
 463                                  )
 464                              (setq ccs (charset-name ccs))
 465                              (null (assq ccs char-spec))
 466                              (setq ret (encode-char char ccs 'defined-only)))
 467                         (setq char-spec (cons (cons ccs ret) char-spec))))
 468                   )
 469                  ((setq ret
 470                         (catch 'tag
 471                           (let ((rest char-db-coded-charset-priority-list)
 472                                 ccs)
 473                             (while rest
 474                               (setq ccs (charset-name
 475                                          (find-charset (car rest))))
 476                               (if (setq ret
 477                                         (encode-char char ccs
 478                                                      'defined-only))
 479                                   (throw 'tag (cons ccs ret)))
 480                               (setq rest (cdr rest))))))
 481                   (setq char-spec (list ret))
 482                   (dolist (ccs (delq (car ret) (charset-list)))
 483                     (if (and (or (charset-iso-final-char ccs)
 484                                  (memq ccs
 485                                        '(=daikanwa
 486                                          =daikanwa@rev2
 487                                          ;; =gt-k
 488                                          =jis-x0208@1997
 489                                          ))
 490                                  (and (eq (charset-property ccs 'iso-ir) 177)
 491                                       (not (eq (charset-name ccs) '=ucs@big5)))
 492                                  ;; (string-match "=ucs@" (symbol-name ccs))
 493                                  )
 494                              (setq ccs (charset-name ccs))
 495                              (null (assq ccs char-spec))
 496                              (setq ret (encode-char char ccs 'defined-only)))
 497                         (setq char-spec (cons (cons ccs ret) char-spec))))
 498                   (if (null char-spec)
 499                       (setq char-spec (split-char char)))
 500                   (cond ((setq ret (get-char-attribute char 'name))
 501                          (setq char-spec (cons (cons 'name ret) char-spec))
 502                          )
 503                         ((setq ret (get-char-attribute char 'name*))
 504                          (setq char-spec (cons (cons 'name* ret) char-spec))
 505                          ))
 506                   )
 507                  ((setq ret (get-char-attribute
 508                              char 'ideographic-combination))
 509                   (setq char-spec
 510                         (cons (cons 'ideographic-combination ret)
 511                               char-spec))
 512                   ))
 513            char-spec)
 514           ((consp char)
 515            char))))
 516
 517 (defun char-db-insert-char-spec (char &optional readable column
 518                                       required-features)
 519   (unless column
 520     (setq column (current-column)))
 521   (let (char-spec temp-char)
 522     (setq char-spec (char-db-make-char-spec char))
 523     (unless (or (characterp char) ; char
 524                 (condition-case nil
 525                     (setq char (find-char char-spec))
 526                   (error nil)))
 527       ;; define temporary character
 528       ;;   Current implementation is dirty.
 529       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 530                                          char-spec)))
 531       (remove-char-attribute temp-char 'ideograph-daikanwa)
 532       (setq char temp-char))
 533     (insert-char-attributes char
 534                             readable
 535                             (union (mapcar #'car char-spec)
 536                                    required-features)
 537                             nil 'for-sub-node)
 538     (when temp-char
 539       ;; undefine temporary character
 540       ;;   Current implementation is dirty.
 541       (setq char-spec (char-attribute-alist temp-char))
 542       (while char-spec
 543         (remove-char-attribute temp-char (car (car char-spec)))
 544         (setq char-spec (cdr char-spec))))))
 545
 546 (defun char-db-insert-alist (alist &optional readable column)
 547   (unless column
 548     (setq column (current-column)))
 549   (let ((line-breaking
 550          (concat "\n" (make-string (1+ column) ?\ )))
 551         name value
 552         ret al ; cal
 553         key
 554         lbs cell rest separator)
 555     (insert "(")
 556     (while alist
 557       (setq name (car (car alist))
 558             value (cdr (car alist)))
 559       (cond ((eq name 'char)
 560              (insert "(char . ")
 561              (if (setq ret (condition-case nil
 562                                (find-char value)
 563                              (error nil)))
 564                  (progn
 565                    (setq al nil
 566                          ;; cal nil
 567                          )
 568                    (while value
 569                      (setq key (car (car value)))
 570                      ;; (if (find-charset key)
 571                      ;;     (setq cal (cons key cal))
 572                      (setq al (cons key al))
 573                      ;; )
 574                      (setq value (cdr value)))
 575                    (insert-char-attributes ret
 576                                            readable
 577                                            (or al 'none) ; cal
 578                                            nil 'for-sub-node))
 579                (insert (prin1-to-string value)))
 580              (insert ")")
 581              (insert line-breaking))
 582             ((consp value)
 583              (insert (format "(%-18s " name))
 584              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 585              (while (consp value)
 586                (setq cell (car value))
 587                (if (and (consp cell)
 588                         (consp (car cell))
 589                         (setq ret (condition-case nil
 590                                       (find-char cell)
 591                                     (error nil)))
 592                         )
 593                    (progn
 594                      (setq rest cell
 595                            al nil
 596                            ;; cal nil
 597                            )
 598                      (while rest
 599                        (setq key (car (car rest)))
 600                        ;; (if (find-charset key)
 601                        ;;     (setq cal (cons key cal))
 602                        (setq al (cons key al))
 603                        ;; )
 604                        (setq rest (cdr rest)))
 605                      (if separator
 606                          (insert lbs))
 607                      (insert-char-attributes ret
 608                                              readable
 609                                              al ; cal
 610                                              nil 'for-sub-node)
 611                      (setq separator lbs))
 612                  (if separator
 613                      (insert separator))
 614                  (insert (prin1-to-string cell))
 615                  (setq separator " "))
 616                (setq value (cdr value)))
 617              (insert ")")
 618              (insert line-breaking))
 619             (t
 620              (insert (format "(%-18s . %S)%s"
 621                              name value
 622                              line-breaking))))
 623       (setq alist (cdr alist))))
 624   (insert ")"))
 625
 626 (defun char-db-insert-char-reference (plist &optional readable column)
 627   (unless column
 628     (setq column (current-column)))
 629   (let ((line-breaking
 630          (concat "\n" (make-string (1+ column) ?\ )))
 631         (separator "")
 632         name value)
 633     (insert "(")
 634     (while plist
 635       (setq name (pop plist))
 636       (setq value (pop plist))
 637       (cond ((eq name :char)
 638              (insert separator)
 639              (insert ":char\t")
 640              (cond ((numberp value)
 641                     (setq value (decode-char '=ucs value)))
 642                    ;; ((consp value)
 643                    ;;  (setq value (or (find-char value)
 644                    ;;                  value)))
 645                    )
 646              (char-db-insert-char-spec value readable)
 647              (insert line-breaking)
 648              (setq separator ""))
 649             ((eq name :radical)
 650              (insert (format "%s%s\t%d ; %c%s"
 651                              separator
 652                              name value
 653                              (ideographic-radical value)
 654                              line-breaking))
 655              (setq separator ""))
 656             (t
 657              (insert (format "%s%s\t%S" separator name value))
 658              (setq separator line-breaking)))
 659       ))
 660   (insert ")"))
 661
 662 (defun char-db-decode-isolated-char (ccs code-point)
 663   (let (ret)
 664     (setq ret
 665           (cond ((eq ccs 'arabic-iso8859-6)
 666                  (decode-char ccs code-point))
 667                 ;; ((eq ccs '=gt)
 668                 ;;  (decode-builtin-char '==gt code-point))
 669                 ((and (memq ccs '(=gt-pj-1
 670                                   =gt-pj-2
 671                                   =gt-pj-3
 672                                   =gt-pj-4
 673                                   =gt-pj-5
 674                                   =gt-pj-6
 675                                   =gt-pj-7
 676                                   =gt-pj-8
 677                                   =gt-pj-9
 678                                   =gt-pj-10
 679                                   =gt-pj-11))
 680                       (setq ret (decode-char ccs code-point))
 681                       (setq ret (encode-char ret '=gt 'defined-only)))
 682                  (decode-builtin-char '=gt ret))
 683                 (t
 684                  (decode-builtin-char ccs code-point))))
 685     (cond ((null ret)
 686            (or (decode-char ccs code-point)
 687                (define-char (list (cons ccs code-point)))))
 688           ((and (<= 0 (char-int ret))
 689                 (<= (char-int ret) #x1F))
 690            (decode-char '=ucs (+ #x2400 (char-int ret))))
 691           ((= (char-int ret) #x7F)
 692            ?\u2421)
 693           (t ret))))
 694
 695 (defvar char-db-convert-obsolete-format t)
 696
 697 (defun char-db-insert-ccs-feature (name value line-breaking)
 698   (cond
 699    ((integerp value)
 700     (insert
 701      (format
 702       (cond
 703        ((memq name '(=>iwds-1
 704                      ==shinjigen
 705                      =shinjigen
 706                      =shinjigen@1ed ==shinjigen@1ed
 707                      =shinjigen@rev ==shinjigen@rev
 708                      =shinjigen/+p@rev ==shinjigen/+p@rev
 709                      ===daikanwa/ho ==daikanwa/ho
 710                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 711         "(%-18s .  %04d)\t; %c")
 712        ((eq name '=shinjigen@1ed/24pr)
 713         "(%-18s . %04d)\t; %c")
 714        ((or
 715          (memq name
 716                '(===daikanwa
 717                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 718                  =daikanwa@rev1 =daikanwa@rev2
 719                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 720                  =>>daikanwa/+p
 721                  =daikanwa/+2p =>>daikanwa/+2p
 722                  =gt ==gt ===gt
 723                  =>>gt =+>gt =>gt
 724                  =gt-k ==gt-k ===gt-k
 725                  =>>gt-k =>gt-k
 726                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 727                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 728                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 729                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 730                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 731                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 732                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 733                  =>>adobe-japan1-0 =+>adobe-japan1-0
 734                  =>>adobe-japan1-1 =+>adobe-japan1-1
 735                  =>>adobe-japan1-2 =+>adobe-japan1-2
 736                  =>>adobe-japan1-3 =+>adobe-japan1-3
 737                  =>>adobe-japan1-4 =+>adobe-japan1-4
 738                  =>>adobe-japan1-5 =+>adobe-japan1-5
 739                  =>>adobe-japan1-6 =+>adobe-japan1-6
 740                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 741                  =zinbun-oracle =>zinbun-oracle
 742                  =daijiten ==daijiten ===daijiten ==>daijiten
 743                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 744                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 745                  ===hng-khi ===hng-khm ===hng-fhs ===hng-hok
 746                  ===hng-kyd ===hng-sok
 747                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 748                  ===hng-sys ===hng-tsu ===hng-tzj
 749                  ===hng-hos ===hng-kkh ===hng-nak ===hng-jhk
 750                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 751                  ===hng-nkm ===hng-k24 ===hng-ini ===hng-nkk
 752                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 753                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 754                  ===hng-okd ===hng-wad ===hng-kmi
 755                  ===hng-zkd ===hng-doh ===hng-jyu ===hng-tzs
 756                  ===hng-sai ===hng-kad ===hng-kss
 757                  ===hng-kyo ===hng-ykk ===hng-saa
 758                  ===hng-sab ===hng-wks ===hng-wke ===hng-smk
 759                  ===hng-sgs ===hng-sts
 760                  =shuowen-jiguge ===shuowen-jiguge4 ===shuowen-jiguge5))
 761          ;; (string-match "^=adobe-" (symbol-name name))
 762          )
 763         "(%-18s . %05d)\t; %c")
 764        ((memq name '(=hanyo-denshi/ks
 765                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 766                      =>>hanyo-denshi/ks
 767                      =koseki ==koseki
 768                      =mj ==mj ===mj =>>mj =>mj
 769                      =zihai mojikyo))
 770         "(%-18s . %06d)\t; %c")
 771        ((memq name '(=hanyo-denshi/tk ==hanyo-denshi/tk))
 772         "(%-18s . %08d)\t; %c")
 773        ((>= (charset-dimension name) 2)
 774         "(%-18s . #x%04X)\t; %c")
 775        (t
 776         "(%-18s . #x%02X)\t; %c"))
 777       name
 778       (if (= (charset-iso-graphic-plane name) 1)
 779           (logior value
 780                   (cond ((= (charset-dimension name) 1)
 781                          #x80)
 782                         ((= (charset-dimension name) 2)
 783                          #x8080)
 784                         ((= (charset-dimension name) 3)
 785                          #x808080)
 786                         (t 0)))
 787         value)
 788       (char-db-decode-isolated-char name value)))
 789     (if (and (= (charset-chars name) 94)
 790              (= (charset-dimension name) 2))
 791         (insert (format " [%02d-%02d]"
 792                         (- (lsh value -8) 32)
 793                         (- (logand value 255) 32))))
 794     )
 795    (t
 796     (insert (format "(%-18s . %s)" name value))
 797     ))
 798   (insert line-breaking))
 799
 800 (defun char-db-insert-relation-feature (char name value line-breaking
 801                                              ccss readable)
 802   (insert (format "(%-18s%s " name line-breaking))
 803   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 804         separator cell sources required-features
 805         ret)
 806     (while (consp value)
 807       (setq cell (car value))
 808       (if (integerp cell)
 809           (setq cell (decode-char '=ucs cell)))
 810       (cond
 811        ((eq name '->subsumptive)
 812         (when (or (not (some (lambda (atr)
 813                                (get-char-attribute cell atr))
 814                              char-db-ignored-attributes))
 815                   (some (lambda (ccs)
 816                           (encode-char cell ccs 'defined-only))
 817                         ccss))
 818           (if separator
 819               (insert lbs))
 820           (let ((char-db-ignored-attributes
 821                  (cons '<-subsumptive
 822                        char-db-ignored-attributes)))
 823             (insert-char-attributes cell readable nil nil 'for-sub-node))
 824           (setq separator lbs))
 825         )
 826        ((characterp cell)
 827         (setq sources
 828               (get-char-attribute
 829                char (intern (format "%s*sources" name))))
 830         (setq required-features nil)
 831         (dolist (source sources)
 832           (cond
 833            ((memq source '(JP
 834                            JP/Jouyou
 835                            shinjigen shinjigen@1ed shinjigen@rev))
 836             (setq required-features
 837                   (union required-features
 838                          '(=jis-x0208
 839                            =jis-x0208@1990
 840                            =jis-x0213-1@2000
 841                            =jis-x0213-1@2004
 842                            =jis-x0213-2
 843                            =jis-x0212
 844                            =jis-x0208@1983
 845                            =jis-x0208@1978
 846                            =shinjigen
 847                            =shinjigen@1ed
 848                            =shinjigen@rev
 849                            =shinjigen/+p@rev))))
 850            ((eq source 'CN)
 851             (setq required-features
 852                   (union required-features
 853                          '(=gb2312
 854                            =gb12345
 855                            =iso-ir165)))))
 856           (cond
 857            ((find-charset (setq ret (intern (format "=%s" source))))
 858             (setq required-features
 859                   (cons ret required-features)))
 860            (t (setq required-features
 861                     (cons source required-features)))))
 862         (cond ((string-match "@JP" (symbol-name name))
 863                (setq required-features
 864                      (union required-features
 865                             '(=jis-x0208
 866                               =jis-x0208@1990
 867                               =jis-x0213-1-2000
 868                               =jis-x0213-2-2000
 869                               =jis-x0212
 870                               =jis-x0208@1983
 871                               =jis-x0208@1978))))
 872               ((string-match "@CN" (symbol-name name))
 873                (setq required-features
 874                      (union required-features
 875                             '(=gb2312
 876                               =gb12345
 877                               =iso-ir165)))))
 878         (if separator
 879             (insert lbs))
 880         (if readable
 881             (insert (format "%S" cell))
 882           (char-db-insert-char-spec cell readable
 883                                     nil
 884                                     required-features))
 885         (setq separator lbs))
 886        ((consp cell)
 887         (if separator
 888             (insert lbs))
 889         (if (consp (car cell))
 890             (char-db-insert-char-spec cell readable)
 891           (char-db-insert-char-reference cell readable))
 892         (setq separator lbs))
 893        (t
 894         (if separator
 895             (insert separator))
 896         (insert (prin1-to-string cell))
 897         (setq separator " ")))
 898       (setq value (cdr value)))
 899     (insert ")")
 900     (insert line-breaking)))
 901
 902 (defun insert-char-attributes (char &optional readable attributes column
 903                                     for-sub-node)
 904   (unless column
 905     (setq column (current-column)))
 906   (let (name value ; has-long-ccs-name
 907         rest
 908         radical strokes
 909         (line-breaking
 910          (concat "\n" (make-string (1+ column) ?\ )))
 911         lbs cell separator ret
 912         key al cal
 913         dest-ccss ; sources required-features
 914         ccss)
 915     (let (atr-d)
 916       (setq attributes
 917             (sort (if attributes
 918                       (if (consp attributes)
 919                           (progn
 920                             (dolist (name attributes)
 921                               (unless (memq name char-db-ignored-attributes)
 922                                 (if (find-charset name)
 923                                     (push name ccss))
 924                                 (push name atr-d)))
 925                             atr-d))
 926                     (dolist (name (char-attribute-list))
 927                       (unless (memq name char-db-ignored-attributes)
 928                         (if (find-charset name)
 929                             (push name ccss))
 930                         (push name atr-d)))
 931                     atr-d)
 932                   #'char-attribute-name<)))
 933     (insert "(")
 934     (when (memq '<-subsumptive attributes)
 935       (when (or readable (not for-sub-node))
 936         (when (setq value (get-char-attribute char '<-subsumptive))
 937           (char-db-insert-relation-feature char '<-subsumptive value
 938                                            line-breaking
 939                                            ccss readable)))
 940       (setq attributes (delq '<-subsumptive attributes)))
 941     (when (and (memq '<-denotational attributes)
 942                (setq value (get-char-attribute char '<-denotational)))
 943       (char-db-insert-relation-feature char '<-denotational value
 944                                        line-breaking
 945                                        ccss readable)
 946       (setq attributes (delq '<-denotational attributes)))
 947     (when (and (memq '<-denotational@component attributes)
 948                (setq value (get-char-attribute char '<-denotational@component)))
 949       (char-db-insert-relation-feature char '<-denotational@component value
 950                                        line-breaking
 951                                        ccss readable)
 952       (setq attributes (delq '<-denotational@component attributes)))
 953     (when (and (memq '<-denotational@usage attributes)
 954                (setq value (get-char-attribute char '<-denotational@usage)))
 955       (char-db-insert-relation-feature char '<-denotational@usage value
 956                                        line-breaking
 957                                        ccss readable)
 958       (setq attributes (delq '<-denotational@usage attributes)))
 959     (when (and (memq 'name attributes)
 960                (setq value (get-char-attribute char 'name)))
 961       (insert (format
 962                (if (> (+ (current-column) (length value)) 48)
 963                    "(name . %S)%s"
 964                  "(name               . %S)%s")
 965                value line-breaking))
 966       (setq attributes (delq 'name attributes))
 967       )
 968     (when (and (memq 'name* attributes)
 969                (setq value (get-char-attribute char 'name*)))
 970       (insert (format
 971                (if (> (+ (current-column) (length value)) 48)
 972                    "(name* . %S)%s"
 973                  "(name*              . %S)%s")
 974                value line-breaking))
 975       (setq attributes (delq 'name* attributes))
 976       )
 977     (when (and (memq 'script attributes)
 978                (setq value (get-char-attribute char 'script)))
 979       (insert (format "(script\t\t%s)%s"
 980                       (mapconcat (function prin1-to-string)
 981                                  value " ")
 982                       line-breaking))
 983       (setq attributes (delq 'script attributes))
 984       )
 985     (dolist (name '(=>ucs =>ucs*))
 986       (when (and (memq name attributes)
 987                  (setq value (get-char-attribute char name)))
 988         (insert (format "(%-18s . #x%04X)\t; %c%s"
 989                         name value (decode-char '=ucs value)
 990                         line-breaking))
 991         (setq attributes (delq name attributes))))
 992     (dolist (name '(=>ucs@gb =>ucs@big5))
 993       (when (and (memq name attributes)
 994                  (setq value (get-char-attribute char name)))
 995         (insert (format "(%-18s . #x%04X)\t; %c%s"
 996                         name value
 997                         (decode-char (intern
 998                                       (concat "="
 999                                               (substring
1000                                                (symbol-name name) 2)))
1001                                      value)
1002                         line-breaking))
1003         (setq attributes (delq name attributes))
1004         ))
1005     ;; (dolist (name '(=>daikanwa))
1006     ;;   (when (and (memq name attributes)
1007     ;;              (setq value (get-char-attribute char name)))
1008     ;;     (insert
1009     ;;      (if (integerp value)
1010     ;;          (format "(%-18s . %05d)\t; %c%s"
1011     ;;                  name value (decode-char '=daikanwa value)
1012     ;;                  line-breaking)
1013     ;;        (format "(%-18s %s)\t; %c%s"
1014     ;;                name
1015     ;;                (mapconcat (function prin1-to-string)
1016     ;;                           value " ")
1017     ;;                (char-representative-of-daikanwa char)
1018     ;;                line-breaking)))
1019     ;;     (setq attributes (delq name attributes))))
1020     (when (and (memq 'general-category attributes)
1021                (setq value (get-char-attribute char 'general-category)))
1022       (insert (format
1023                "(general-category\t%s) ; %s%s"
1024                (mapconcat (lambda (cell)
1025                             (format "%S" cell))
1026                           value " ")
1027                (cond ((rassoc value unidata-normative-category-alist)
1028                       "Normative Category")
1029                      ((rassoc value unidata-informative-category-alist)
1030                       "Informative Category")
1031                      (t
1032                       "Unknown Category"))
1033                line-breaking))
1034       (setq attributes (delq 'general-category attributes))
1035       )
1036     (when (and (memq 'bidi-category attributes)
1037                (setq value (get-char-attribute char 'bidi-category)))
1038       (insert (format "(bidi-category\t. %S)%s"
1039                       value
1040                       line-breaking))
1041       (setq attributes (delq 'bidi-category attributes))
1042       )
1043     (unless (or (not (memq 'mirrored attributes))
1044                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
1045                     'empty))
1046       (insert (format "(mirrored\t\t. %S)%s"
1047                       value
1048                       line-breaking))
1049       (setq attributes (delq 'mirrored attributes))
1050       )
1051     (cond
1052      ((and (memq 'decimal-digit-value attributes)
1053            (setq value (get-char-attribute char 'decimal-digit-value)))
1054       (insert (format "(decimal-digit-value . %S)%s"
1055                       value
1056                       line-breaking))
1057       (setq attributes (delq 'decimal-digit-value attributes))
1058       (when (and (memq 'digit-value attributes)
1059                  (setq value (get-char-attribute char 'digit-value)))
1060         (insert (format "(digit-value\t . %S)%s"
1061                         value
1062                         line-breaking))
1063         (setq attributes (delq 'digit-value attributes))
1064         )
1065       (when (and (memq 'numeric-value attributes)
1066                  (setq value (get-char-attribute char 'numeric-value)))
1067         (insert (format "(numeric-value\t . %S)%s"
1068                         value
1069                         line-breaking))
1070         (setq attributes (delq 'numeric-value attributes))
1071         )
1072       )
1073      (t
1074       (when (and (memq 'digit-value attributes)
1075                  (setq value (get-char-attribute char 'digit-value)))
1076         (insert (format "(digit-value\t. %S)%s"
1077                         value
1078                         line-breaking))
1079         (setq attributes (delq 'digit-value attributes))
1080         )
1081       (when (and (memq 'numeric-value attributes)
1082                  (setq value (get-char-attribute char 'numeric-value)))
1083         (insert (format "(numeric-value\t. %S)%s"
1084                         value
1085                         line-breaking))
1086         (setq attributes (delq 'numeric-value attributes))
1087         )))
1088     (when (and (memq 'iso-10646-comment attributes)
1089                (setq value (get-char-attribute char 'iso-10646-comment)))
1090       (insert (format "(iso-10646-comment\t. %S)%s"
1091                       value
1092                       line-breaking))
1093       (setq attributes (delq 'iso-10646-comment attributes))
1094       )
1095     (when (and (memq 'morohashi-daikanwa attributes)
1096                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1097       (insert (format "(morohashi-daikanwa\t%s)%s"
1098                       (mapconcat (function prin1-to-string) value " ")
1099                       line-breaking))
1100       (setq attributes (delq 'morohashi-daikanwa attributes))
1101       )
1102     (setq radical nil
1103           strokes nil)
1104     (when (and (memq 'ideographic-radical attributes)
1105                (setq value (get-char-attribute char 'ideographic-radical)))
1106       (setq radical value)
1107       (insert (format "(ideographic-radical . %S)\t; %c%s"
1108                       radical
1109                       (ideographic-radical radical)
1110                       line-breaking))
1111       (setq attributes (delq 'ideographic-radical attributes))
1112       )
1113     (when (and (memq 'shuowen-radical attributes)
1114                (setq value (get-char-attribute char 'shuowen-radical)))
1115       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1116                       value
1117                       (shuowen-radical value)
1118                       line-breaking))
1119       (setq attributes (delq 'shuowen-radical attributes))
1120       )
1121     (let (key)
1122       (dolist (domain
1123                (append
1124                 char-db-feature-domains
1125                 (let (dest domain)
1126                   (dolist (feature (char-attribute-list))
1127                     (setq feature (symbol-name feature))
1128                     (when (string-match
1129                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1130                            feature)
1131                       (setq domain (intern (match-string 2 feature)))
1132                      (unless (memq domain dest)
1133                        (setq dest (cons domain dest)))))
1134                   (sort dest #'string<))))
1135         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1136         (when (and (memq key attributes)
1137                    (setq value (get-char-attribute char key)))
1138           (setq radical value)
1139           (insert (format "(%s . %S)\t; %c%s"
1140                           key
1141                           radical
1142                           (ideographic-radical radical)
1143                           line-breaking))
1144           (setq attributes (delq key attributes))
1145           )
1146         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1147         (when (and (memq key attributes)
1148                    (setq value (get-char-attribute char key)))
1149           (setq strokes value)
1150           (insert (format "(%s . %S)%s"
1151                           key
1152                           strokes
1153                           line-breaking))
1154           (setq attributes (delq key attributes))
1155           )
1156         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1157         (when (and (memq key attributes)
1158                    (setq value (get-char-attribute char key)))
1159           (insert (format "(%s       . %S)%s"
1160                           key
1161                           value
1162                           line-breaking))
1163           (setq attributes (delq key attributes))
1164           )
1165         (dolist (feature '(ideographic-radical
1166                            ideographic-strokes
1167                            total-strokes))
1168           (setq key (intern (format "%s@%s*sources" feature domain)))
1169           (when (and (memq key attributes)
1170                      (setq value (get-char-attribute char key)))
1171             (insert (format "(%s%s" key line-breaking))
1172             (dolist (cell value)
1173               (insert (format " %s" cell)))
1174             (insert ")")
1175             (insert line-breaking)
1176             (setq attributes (delq key attributes))
1177             ))
1178         ))
1179     (when (and (memq 'ideographic-strokes attributes)
1180                (setq value (get-char-attribute char 'ideographic-strokes)))
1181       (setq strokes value)
1182       (insert (format "(ideographic-strokes . %S)%s"
1183                       strokes
1184                       line-breaking))
1185       (setq attributes (delq 'ideographic-strokes attributes))
1186       )
1187     (when (and (memq 'kangxi-radical attributes)
1188                (setq value (get-char-attribute char 'kangxi-radical)))
1189       (unless (eq value radical)
1190         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1191                         value
1192                         (ideographic-radical value)
1193                         line-breaking))
1194         (or radical
1195             (setq radical value)))
1196       (setq attributes (delq 'kangxi-radical attributes))
1197       )
1198     (when (and (memq 'kangxi-strokes attributes)
1199                (setq value (get-char-attribute char 'kangxi-strokes)))
1200       (unless (eq value strokes)
1201         (insert (format "(kangxi-strokes\t . %S)%s"
1202                         value
1203                         line-breaking))
1204         (or strokes
1205             (setq strokes value)))
1206       (setq attributes (delq 'kangxi-strokes attributes))
1207       )
1208     ;; (when (and (memq 'japanese-radical attributes)
1209     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1210     ;;   (unless (eq value radical)
1211     ;;     (insert (format "(japanese-radical\t . %S)\t; %c%s"
1212     ;;                     value
1213     ;;                     (ideographic-radical value)
1214     ;;                     line-breaking))
1215     ;;     (or radical
1216     ;;         (setq radical value)))
1217     ;;   (setq attributes (delq 'japanese-radical attributes))
1218     ;;   )
1219     (when (and (memq 'japanese-strokes attributes)
1220                (setq value (get-char-attribute char 'japanese-strokes)))
1221       (unless (eq value strokes)
1222         (insert (format "(japanese-strokes\t . %S)%s"
1223                         value
1224                         line-breaking))
1225         (or strokes
1226             (setq strokes value)))
1227       (setq attributes (delq 'japanese-strokes attributes))
1228       )
1229     (when (and (memq 'cns-radical attributes)
1230                (setq value (get-char-attribute char 'cns-radical)))
1231       (insert (format "(cns-radical\t . %S)\t; %c%s"
1232                       value
1233                       (ideographic-radical value)
1234                       line-breaking))
1235       (setq attributes (delq 'cns-radical attributes))
1236       )
1237     (when (and (memq 'cns-strokes attributes)
1238                (setq value (get-char-attribute char 'cns-strokes)))
1239       (unless (eq value strokes)
1240         (insert (format "(cns-strokes\t . %S)%s"
1241                         value
1242                         line-breaking))
1243         (or strokes
1244             (setq strokes value)))
1245       (setq attributes (delq 'cns-strokes attributes))
1246       )
1247     ;; (when (and (memq 'ideographic- attributes)
1248     ;;            (setq value (get-char-attribute char 'ideographic-)))
1249     ;;   (insert "(ideographic-       ")
1250     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1251     ;;         separator nil)
1252     ;;   (while (consp value)
1253     ;;     (setq cell (car value))
1254     ;;     (if (integerp cell)
1255     ;;         (setq cell (decode-char '=ucs cell)))
1256     ;;     (cond ((characterp cell)
1257     ;;            (if separator
1258     ;;                (insert lbs))
1259     ;;            (if readable
1260     ;;                (insert (format "%S" cell))
1261     ;;              (char-db-insert-char-spec cell readable))
1262     ;;            (setq separator lbs))
1263     ;;           ((consp cell)
1264     ;;            (if separator
1265     ;;                (insert lbs))
1266     ;;            (if (consp (car cell))
1267     ;;                (char-db-insert-char-spec cell readable)
1268     ;;              (char-db-insert-char-reference cell readable))
1269     ;;            (setq separator lbs))
1270     ;;           (t
1271     ;;            (if separator
1272     ;;                (insert separator))
1273     ;;            (insert (prin1-to-string cell))
1274     ;;            (setq separator " ")))
1275     ;;     (setq value (cdr value)))
1276     ;;   (insert ")")
1277     ;;   (insert line-breaking)
1278     ;;   (setq attributes (delq 'ideographic- attributes)))
1279     (when (and (memq 'total-strokes attributes)
1280                (setq value (get-char-attribute char 'total-strokes)))
1281       (insert (format "(total-strokes       . %S)%s"
1282                       value
1283                       line-breaking))
1284       (setq attributes (delq 'total-strokes attributes))
1285       )
1286     ;; (when (and (memq '->ideograph attributes)
1287     ;;            (setq value (get-char-attribute char '->ideograph)))
1288     ;;   (insert (format "(->ideograph\t%s)%s"
1289     ;;                   (mapconcat (lambda (code)
1290     ;;                                (cond ((symbolp code)
1291     ;;                                       (symbol-name code))
1292     ;;                                      ((integerp code)
1293     ;;                                       (format "#x%04X" code))
1294     ;;                                      (t
1295     ;;                                       (format "%s %S"
1296     ;;                                               line-breaking code))))
1297     ;;                              value " ")
1298     ;;                   line-breaking))
1299     ;;   (setq attributes (delq '->ideograph attributes))
1300     ;;   )
1301     ;; (when (and (memq '->decomposition attributes)
1302     ;;            (setq value (get-char-attribute char '->decomposition)))
1303     ;;   (insert (format "(->decomposition\t%s)%s"
1304     ;;                   (mapconcat (lambda (code)
1305     ;;                                (cond ((symbolp code)
1306     ;;                                       (symbol-name code))
1307     ;;                                      ((characterp code)
1308     ;;                                       (if readable
1309     ;;                                           (format "%S" code)
1310     ;;                                         (format "#x%04X"
1311     ;;                                                 (char-int code))
1312     ;;                                         ))
1313     ;;                                      ((integerp code)
1314     ;;                                       (format "#x%04X" code))
1315     ;;                                      (t
1316     ;;                                       (format "%s%S" line-breaking code))))
1317     ;;                              value " ")
1318     ;;                   line-breaking))
1319     ;;   (setq attributes (delq '->decomposition attributes))
1320     ;;   )
1321     (if (equal (get-char-attribute char '->titlecase)
1322                (get-char-attribute char '->uppercase))
1323         (setq attributes (delq '->titlecase attributes)))
1324     (when (and (memq '->mojikyo attributes)
1325                (setq value (get-char-attribute char '->mojikyo)))
1326       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1327                       value (decode-char 'mojikyo value)
1328                       line-breaking))
1329       (setq attributes (delq '->mojikyo attributes))
1330       )
1331     (unless readable
1332       (dolist (ignored '(composition
1333                          ->denotational <-subsumptive ->ucs-unified
1334                          ;; ->ideographic-component-forms
1335                          ))
1336         (setq attributes (delq ignored attributes))))
1337     (while attributes
1338       (setq name (car attributes))
1339       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1340                   'value-is-empty)
1341         (cond ((setq ret (find-charset name))
1342                (setq name (charset-name ret))
1343                (when (not (memq name dest-ccss))
1344                  (setq dest-ccss (cons name dest-ccss))
1345                  (char-db-insert-ccs-feature name value line-breaking))
1346                )
1347               ((string-match "^=>ucs@" (symbol-name name))
1348                (insert (format "(%-18s . #x%04X)\t; %c%s"
1349                                name value (decode-char '=ucs value)
1350                                line-breaking))
1351                )
1352               ((eq name 'jisx0208-1978/4X)
1353                (insert (format "(%-18s . #x%04X)%s"
1354                                name value
1355                                line-breaking))
1356                )
1357               ((and
1358                 (not readable)
1359                 (not (eq name '->subsumptive))
1360                 (not (eq name '->uppercase))
1361                 (not (eq name '->lowercase))
1362                 (not (eq name '->titlecase))
1363                 (not (eq name '->canonical))
1364                 (not (eq name '->Bopomofo))
1365                 (not (eq name '->mistakable))
1366                 (not (eq name '->ideographic-variants))
1367                 (null (get-char-attribute
1368                        char (intern (format "%s*sources" name))))
1369                 (not (string-match "\\*sources$" (symbol-name name)))
1370                 (null (get-char-attribute
1371                        char (intern (format "%s*note" name))))
1372                 (not (string-match "\\*note$" (symbol-name name)))
1373                 (or (eq name '<-identical)
1374                     (eq name '<-uppercase)
1375                     (eq name '<-lowercase)
1376                     (eq name '<-titlecase)
1377                     (eq name '<-canonical)
1378                     (eq name '<-ideographic-variants)
1379                     ;; (eq name '<-synonyms)
1380                     (string-match "^<-synonyms" (symbol-name name))
1381                     (eq name '<-mistakable)
1382                     (when (string-match "^->" (symbol-name name))
1383                       (cond
1384                        ((string-match "^->fullwidth" (symbol-name name))
1385                         (not (and (consp value)
1386                                   (characterp (car value))
1387                                   (encode-char
1388                                    (car value) '=ucs 'defined-only)))
1389                         )
1390                        (t)))
1391                     ))
1392                )
1393               ((or (eq name 'ideographic-structure)
1394                    (eq name 'ideographic-combination)
1395                    ;; (eq name 'ideographic-)
1396                    (eq name '=decomposition)
1397                    (char-feature-base-name= '=decomposition name)
1398                    (char-feature-base-name= '=>decomposition name)
1399                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1400                    ;;               (symbol-name name))
1401                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1402                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1403                                  (symbol-name name))
1404                    )
1405                (char-db-insert-relation-feature char name value
1406                                                 line-breaking
1407                                                 ccss readable))
1408               ;; ((memq name '(ideograph=
1409               ;;               original-ideograph-of
1410               ;;               ancient-ideograph-of
1411               ;;               vulgar-ideograph-of
1412               ;;               wrong-ideograph-of
1413               ;;               ;; simplified-ideograph-of
1414               ;;               ideographic-variants
1415               ;;               ;; ideographic-different-form-of
1416               ;;               ))
1417               ;;  (insert (format "(%-18s%s " name line-breaking))
1418               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1419               ;;        separator nil)
1420               ;;  (while (consp value)
1421               ;;    (setq cell (car value))
1422               ;;    (if (and (consp cell)
1423               ;;             (consp (car cell)))
1424               ;;        (progn
1425               ;;          (if separator
1426               ;;              (insert lbs))
1427               ;;          (char-db-insert-alist cell readable)
1428               ;;          (setq separator lbs))
1429               ;;      (if separator
1430               ;;          (insert separator))
1431               ;;      (insert (prin1-to-string cell))
1432               ;;      (setq separator " "))
1433               ;;    (setq value (cdr value)))
1434               ;;  (insert ")")
1435               ;;  (insert line-breaking))
1436               ((consp value)
1437                (insert (format "(%-18s " name))
1438                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1439                      separator nil)
1440                (while (consp value)
1441                  (setq cell (car value))
1442                  (if (and (consp cell)
1443                           (consp (car cell))
1444                           (setq ret (condition-case nil
1445                                         (find-char cell)
1446                                       (error nil))))
1447                      (progn
1448                        (setq rest cell
1449                              al nil
1450                              cal nil)
1451                        (while rest
1452                          (setq key (car (car rest)))
1453                          (if (find-charset key)
1454                              (setq cal (cons key cal))
1455                            (setq al (cons key al)))
1456                          (setq rest (cdr rest)))
1457                        (if separator
1458                            (insert lbs))
1459                        (insert-char-attributes ret
1460                                                readable
1461                                                al ; cal
1462                                                nil 'for-sub-node)
1463                        (setq separator lbs))
1464                    (setq ret (prin1-to-string cell))
1465                    (if separator
1466                        (if (< (+ (current-column)
1467                                  (length ret)
1468                                  (length separator))
1469                               76)
1470                            (insert separator)
1471                          (insert lbs)))
1472                    (insert ret)
1473                    (setq separator " "))
1474                  (setq value (cdr value)))
1475                (insert ")")
1476                (insert line-breaking))
1477               (t
1478                (insert (format "(%-18s" name))
1479                (setq ret (prin1-to-string value))
1480                (unless (< (+ (current-column)
1481                              (length ret)
1482                              3)
1483                           76)
1484                  (insert line-breaking))
1485                (insert " . " ret ")" line-breaking)
1486                ;; (insert (format "(%-18s . %S)%s"
1487                ;;                 name value
1488                ;;                 line-breaking))
1489                )
1490               ))
1491       (setq attributes (cdr attributes)))
1492     (insert ")")))
1493
1494 (defun insert-char-data (char &optional readable
1495                               attributes)
1496   (save-restriction
1497     (narrow-to-region (point)(point))
1498     (insert "(define-char
1499   '")
1500     (insert-char-attributes char readable attributes)
1501     (insert ")\n")
1502     (goto-char (point-min))
1503     (while (re-search-forward "[ \t]+$" nil t)
1504       (replace-match ""))
1505     ;; from tabify.
1506     (goto-char (point-min))
1507     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1508       (let ((column (current-column))
1509             (indent-tabs-mode t))
1510         (delete-region (match-beginning 0) (point))
1511         (indent-to column)))
1512     (goto-char (point-max))
1513     ;; (tabify (point-min)(point-max))
1514     ))
1515
1516 (defun insert-char-data-with-variant (char &optional printable
1517                                            no-ucs-unified
1518                                            script excluded-script)
1519   (insert-char-data char printable)
1520   (let ((variants (char-variants char))
1521         rest
1522         variant vs ret)
1523     (setq variants (sort variants #'<))
1524     (setq rest variants)
1525     (setq variants (cons char variants))
1526     (while rest
1527       (setq variant (car rest))
1528       (unless (get-char-attribute variant '<-subsumptive)
1529         (if (and (or (null script)
1530                      (null (setq vs (get-char-attribute variant 'script)))
1531                      (memq script vs))
1532                  (or (null excluded-script)
1533                      (null (setq vs (get-char-attribute variant 'script)))
1534                      (not (memq excluded-script vs))))
1535             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1536               (insert-char-data variant printable)
1537               (if (setq ret (char-variants variant))
1538                   (while ret
1539                     (or (memq (car ret) variants)
1540                         ;; (get-char-attribute (car ret) '<-subsumptive)
1541                         (setq rest (nconc rest (list (car ret)))))
1542                     (setq ret (cdr ret)))))))
1543       (setq rest (cdr rest)))))
1544
1545 (defun insert-char-range-data (min max &optional script excluded-script)
1546   (let ((code min)
1547         char)
1548     (while (<= code max)
1549       (setq char (decode-char '=ucs code))
1550       (if (encode-char char '=ucs 'defined-only)
1551           (insert-char-data-with-variant char nil 'no-ucs-unified
1552                                          script excluded-script))
1553       (setq code (1+ code)))))
1554
1555 (defun write-char-range-data-to-file (min max file
1556                                           &optional script excluded-script)
1557   (let ((coding-system-for-write char-db-file-coding-system))
1558     (with-temp-buffer
1559       (insert (format ";; -*- coding: %s -*-\n"
1560                       char-db-file-coding-system))
1561       (insert-char-range-data min max script excluded-script)
1562       (write-region (point-min)(point-max) file))))
1563
1564 (defvar what-character-original-window-configuration)
1565
1566 ;;;###autoload
1567 (defun what-char-definition (char)
1568   (interactive (list (char-after)))
1569   (let ((buf (get-buffer-create "*Character Description*"))
1570         (the-buf (current-buffer))
1571         (win-conf (current-window-configuration)))
1572     (pop-to-buffer buf)
1573     (make-local-variable 'what-character-original-window-configuration)
1574     (setq what-character-original-window-configuration win-conf)
1575     (setq buffer-read-only nil)
1576     (erase-buffer)
1577     (condition-case err
1578         (progn
1579           (insert-char-data-with-variant char 'printable)
1580           (unless (char-attribute-alist char)
1581             (insert (format ";; = %c\n"
1582                             (let* ((rest (split-char char))
1583                                    (ccs (pop rest))
1584                                    (code (pop rest)))
1585                               (while rest
1586                                 (setq code (logior (lsh code 8)
1587                                                    (pop rest))))
1588                               (decode-char ccs code)))))
1589           ;; (char-db-update-comment)
1590           (set-buffer-modified-p nil)
1591           (view-mode the-buf (lambda (buf)
1592                                (set-window-configuration
1593                                 what-character-original-window-configuration)
1594                                ))
1595           (goto-char (point-min)))
1596       (error (progn
1597                (set-window-configuration
1598                 what-character-original-window-configuration)
1599                (signal (car err) (cdr err)))))))
1600
1601
1602 ;;; @ end
1603 ;;;
1604
1605 (provide 'char-db-util)
1606
1607 ;;; char-db-util.el ends here