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@iwds-1/normalized
 380     =>ucs-itaiji-001@iwds-1
 381     =>big5-cdp@iwds-1
 382     =>big5-cdp@component
 383     =>big5-cdp@cognate
 384     ==ucs@gb
 385     =ucs@gb
 386     ==ucs-var-002
 387     =ucs@JP/hanazono))
 388
 389
 390 ;;; @ char-db formatters
 391 ;;;
 392
 393 (defun char-db-make-char-spec (char)
 394   (let (ret char-spec)
 395     (cond ((characterp char)
 396            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 397                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 398                   (setq char-spec (list (cons '=ucs ret)))
 399                   (cond ((setq ret (get-char-attribute char 'name))
 400                          (setq char-spec (cons (cons 'name ret) char-spec))
 401                          )
 402                         ((setq ret (get-char-attribute char 'name*))
 403                          (setq char-spec (cons (cons 'name* ret) char-spec))
 404                          ))
 405                   )
 406                  ((encode-char char '=mj 'defined-only)
 407                   (setq char-spec nil)
 408                   (dolist (ccs (charset-list))
 409                     (if (and (or (eq ccs '=mj)
 410                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 411                                  (string-match "=ucs@" (symbol-name ccs))
 412                                  )
 413                              (setq ccs (charset-name ccs))
 414                              (null (assq ccs char-spec))
 415                              (setq ret (encode-char char ccs 'defined-only)))
 416                         (setq char-spec (cons (cons ccs ret) char-spec))))
 417                   )
 418                  ((encode-char char '==mj 'defined-only)
 419                   (setq char-spec nil)
 420                   (dolist (ccs (charset-list))
 421                     (if (and (or (eq ccs '==mj)
 422                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 423                                  (string-match "=ucs@" (symbol-name ccs))
 424                                  )
 425                              (setq ccs (charset-name ccs))
 426                              (null (assq ccs char-spec))
 427                              (setq ret (encode-char char ccs 'defined-only)))
 428                         (setq char-spec (cons (cons ccs ret) char-spec))))
 429                   )
 430                  ((encode-char char '=adobe-japan1 'defined-only)
 431                   (setq char-spec nil)
 432                   (dolist (ccs (charset-list))
 433                     (if (and (or (memq ccs
 434                                        '(=adobe-japan1-0
 435                                          =adobe-japan1-1
 436                                          =adobe-japan1-2
 437                                          =adobe-japan1-3
 438                                          =adobe-japan1-4
 439                                          =adobe-japan1-5
 440                                          =adobe-japan1-6
 441                                          ))
 442                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 443                                  (string-match "=ucs@" (symbol-name ccs))
 444                                  )
 445                              (setq ccs (charset-name ccs))
 446                              (null (assq ccs char-spec))
 447                              (setq ret (encode-char char ccs 'defined-only)))
 448                         (setq char-spec (cons (cons ccs ret) char-spec))))
 449                   )
 450                  ((encode-char char '==adobe-japan1 'defined-only)
 451                   (setq char-spec nil)
 452                   (dolist (ccs (charset-list))
 453                     (if (and (or (memq ccs
 454                                        '(==adobe-japan1-0
 455                                          ==adobe-japan1-1
 456                                          ==adobe-japan1-2
 457                                          ==adobe-japan1-3
 458                                          ==adobe-japan1-4
 459                                          ==adobe-japan1-5
 460                                          ==adobe-japan1-6
 461                                          ))
 462                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 463                                  (string-match "=ucs@" (symbol-name ccs))
 464                                  )
 465                              (setq ccs (charset-name ccs))
 466                              (null (assq ccs char-spec))
 467                              (setq ret (encode-char char ccs 'defined-only)))
 468                         (setq char-spec (cons (cons ccs ret) char-spec))))
 469                   )
 470                  ((setq ret
 471                         (catch 'tag
 472                           (let ((rest char-db-coded-charset-priority-list)
 473                                 ccs)
 474                             (while rest
 475                               (setq ccs (charset-name
 476                                          (find-charset (car rest))))
 477                               (if (setq ret
 478                                         (encode-char char ccs
 479                                                      'defined-only))
 480                                   (throw 'tag (cons ccs ret)))
 481                               (setq rest (cdr rest))))))
 482                   (setq char-spec (list ret))
 483                   (dolist (ccs (delq (car ret) (charset-list)))
 484                     (if (and (or (charset-iso-final-char ccs)
 485                                  (memq ccs
 486                                        '(=daikanwa
 487                                          =daikanwa@rev2
 488                                          ;; =gt-k
 489                                          =jis-x0208@1997
 490                                          ))
 491                                  (and (eq (charset-property ccs 'iso-ir) 177)
 492                                       (not (eq (charset-name ccs) '=ucs@big5)))
 493                                  ;; (string-match "=ucs@" (symbol-name ccs))
 494                                  )
 495                              (setq ccs (charset-name ccs))
 496                              (null (assq ccs char-spec))
 497                              (setq ret (encode-char char ccs 'defined-only)))
 498                         (setq char-spec (cons (cons ccs ret) char-spec))))
 499                   (if (null char-spec)
 500                       (setq char-spec (split-char char)))
 501                   (cond ((setq ret (get-char-attribute char 'name))
 502                          (setq char-spec (cons (cons 'name ret) char-spec))
 503                          )
 504                         ((setq ret (get-char-attribute char 'name*))
 505                          (setq char-spec (cons (cons 'name* ret) char-spec))
 506                          ))
 507                   )
 508                  ((setq ret (get-char-attribute
 509                              char 'ideographic-combination))
 510                   (setq char-spec
 511                         (cons (cons 'ideographic-combination ret)
 512                               char-spec))
 513                   ))
 514            char-spec)
 515           ((consp char)
 516            char))))
 517
 518 (defun char-db-insert-char-spec (char &optional readable column
 519                                       required-features)
 520   (unless column
 521     (setq column (current-column)))
 522   (let (char-spec temp-char)
 523     (setq char-spec (char-db-make-char-spec char))
 524     (unless (or (characterp char) ; char
 525                 (condition-case nil
 526                     (setq char (find-char char-spec))
 527                   (error nil)))
 528       ;; define temporary character
 529       ;;   Current implementation is dirty.
 530       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 531                                          char-spec)))
 532       (remove-char-attribute temp-char 'ideograph-daikanwa)
 533       (setq char temp-char))
 534     (insert-char-attributes char
 535                             readable
 536                             (union (mapcar #'car char-spec)
 537                                    required-features)
 538                             nil 'for-sub-node)
 539     (when temp-char
 540       ;; undefine temporary character
 541       ;;   Current implementation is dirty.
 542       (setq char-spec (char-attribute-alist temp-char))
 543       (while char-spec
 544         (remove-char-attribute temp-char (car (car char-spec)))
 545         (setq char-spec (cdr char-spec))))))
 546
 547 (defun char-db-insert-alist (alist &optional readable column)
 548   (unless column
 549     (setq column (current-column)))
 550   (let ((line-breaking
 551          (concat "\n" (make-string (1+ column) ?\ )))
 552         name value
 553         ret al ; cal
 554         key
 555         lbs cell rest separator)
 556     (insert "(")
 557     (while alist
 558       (setq name (car (car alist))
 559             value (cdr (car alist)))
 560       (cond ((eq name 'char)
 561              (insert "(char . ")
 562              (if (setq ret (condition-case nil
 563                                (find-char value)
 564                              (error nil)))
 565                  (progn
 566                    (setq al nil
 567                          ;; cal nil
 568                          )
 569                    (while value
 570                      (setq key (car (car value)))
 571                      ;; (if (find-charset key)
 572                      ;;     (setq cal (cons key cal))
 573                      (setq al (cons key al))
 574                      ;; )
 575                      (setq value (cdr value)))
 576                    (insert-char-attributes ret
 577                                            readable
 578                                            (or al 'none) ; cal
 579                                            nil 'for-sub-node))
 580                (insert (prin1-to-string value)))
 581              (insert ")")
 582              (insert line-breaking))
 583             ((consp value)
 584              (insert (format "(%-18s " name))
 585              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 586              (while (consp value)
 587                (setq cell (car value))
 588                (if (and (consp cell)
 589                         (consp (car cell))
 590                         (setq ret (condition-case nil
 591                                       (find-char cell)
 592                                     (error nil)))
 593                         )
 594                    (progn
 595                      (setq rest cell
 596                            al nil
 597                            ;; cal nil
 598                            )
 599                      (while rest
 600                        (setq key (car (car rest)))
 601                        ;; (if (find-charset key)
 602                        ;;     (setq cal (cons key cal))
 603                        (setq al (cons key al))
 604                        ;; )
 605                        (setq rest (cdr rest)))
 606                      (if separator
 607                          (insert lbs))
 608                      (insert-char-attributes ret
 609                                              readable
 610                                              al ; cal
 611                                              nil 'for-sub-node)
 612                      (setq separator lbs))
 613                  (if separator
 614                      (insert separator))
 615                  (insert (prin1-to-string cell))
 616                  (setq separator " "))
 617                (setq value (cdr value)))
 618              (insert ")")
 619              (insert line-breaking))
 620             (t
 621              (insert (format "(%-18s . %S)%s"
 622                              name value
 623                              line-breaking))))
 624       (setq alist (cdr alist))))
 625   (insert ")"))
 626
 627 (defun char-db-insert-char-reference (plist &optional readable column)
 628   (unless column
 629     (setq column (current-column)))
 630   (let ((line-breaking
 631          (concat "\n" (make-string (1+ column) ?\ )))
 632         (separator "")
 633         name value)
 634     (insert "(")
 635     (while plist
 636       (setq name (pop plist))
 637       (setq value (pop plist))
 638       (cond ((eq name :char)
 639              (insert separator)
 640              (insert ":char\t")
 641              (cond ((numberp value)
 642                     (setq value (decode-char '=ucs value)))
 643                    ;; ((consp value)
 644                    ;;  (setq value (or (find-char value)
 645                    ;;                  value)))
 646                    )
 647              (char-db-insert-char-spec value readable)
 648              (insert line-breaking)
 649              (setq separator ""))
 650             ((eq name :radical)
 651              (insert (format "%s%s\t%d ; %c%s"
 652                              separator
 653                              name value
 654                              (ideographic-radical value)
 655                              line-breaking))
 656              (setq separator ""))
 657             (t
 658              (insert (format "%s%s\t%S" separator name value))
 659              (setq separator line-breaking)))
 660       ))
 661   (insert ")"))
 662
 663 (defun char-db-decode-isolated-char (ccs code-point)
 664   (let (ret)
 665     (setq ret
 666           (cond ((eq ccs 'arabic-iso8859-6)
 667                  (decode-char ccs code-point))
 668                 ;; ((eq ccs '=gt)
 669                 ;;  (decode-builtin-char '==gt code-point))
 670                 ((and (memq ccs '(=gt-pj-1
 671                                   =gt-pj-2
 672                                   =gt-pj-3
 673                                   =gt-pj-4
 674                                   =gt-pj-5
 675                                   =gt-pj-6
 676                                   =gt-pj-7
 677                                   =gt-pj-8
 678                                   =gt-pj-9
 679                                   =gt-pj-10
 680                                   =gt-pj-11))
 681                       (setq ret (decode-char ccs code-point))
 682                       (setq ret (encode-char ret '=gt 'defined-only)))
 683                  (decode-builtin-char '=gt ret))
 684                 (t
 685                  (decode-builtin-char ccs code-point))))
 686     (cond ((null ret)
 687            (or (decode-char ccs code-point)
 688                (define-char (list (cons ccs code-point)))))
 689           ((and (<= 0 (char-int ret))
 690                 (<= (char-int ret) #x1F))
 691            (decode-char '=ucs (+ #x2400 (char-int ret))))
 692           ((= (char-int ret) #x7F)
 693            ?\u2421)
 694           (t ret))))
 695
 696 (defvar char-db-convert-obsolete-format t)
 697
 698 (defun char-db-insert-ccs-feature (name value line-breaking)
 699   (cond
 700    ((integerp value)
 701     (insert
 702      (format
 703       (cond
 704        ((memq name '(=>iwds-1
 705                      ==shinjigen
 706                      =shinjigen
 707                      =shinjigen@1ed ==shinjigen@1ed
 708                      =shinjigen@rev ==shinjigen@rev
 709                      =shinjigen/+p@rev ==shinjigen/+p@rev
 710                      ===daikanwa/ho ==daikanwa/ho
 711                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 712         "(%-18s .  %04d)\t; %c")
 713        ((eq name '=shinjigen@1ed/24pr)
 714         "(%-18s . %04d)\t; %c")
 715        ((or
 716          (memq name
 717                '(===daikanwa
 718                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 719                  =daikanwa@rev1 =daikanwa@rev2
 720                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 721                  =>>daikanwa/+p
 722                  =daikanwa/+2p =>>daikanwa/+2p
 723                  =gt ==gt ===gt
 724                  =>>gt =+>gt =>gt
 725                  =gt-k ==gt-k ===gt-k
 726                  =>>gt-k =>gt-k
 727                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 728                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 729                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 730                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 731                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 732                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 733                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 734                  =>>adobe-japan1-0 =+>adobe-japan1-0
 735                  =>>adobe-japan1-1 =+>adobe-japan1-1
 736                  =>>adobe-japan1-2 =+>adobe-japan1-2
 737                  =>>adobe-japan1-3 =+>adobe-japan1-3
 738                  =>>adobe-japan1-4 =+>adobe-japan1-4
 739                  =>>adobe-japan1-5 =+>adobe-japan1-5
 740                  =>>adobe-japan1-6 =+>adobe-japan1-6
 741                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 742                  =zinbun-oracle =>zinbun-oracle
 743                  =daijiten ==daijiten ===daijiten ==>daijiten
 744                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 745                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 746                  ===hng-khi ===hng-khm ===hng-fhs ===hng-hok
 747                  ===hng-kyd ===hng-sok
 748                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 749                  ===hng-sys ===hng-tsu ===hng-tzj
 750                  ===hng-hos ===hng-kkh ===hng-nak ===hng-jhk
 751                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 752                  ===hng-nkm ===hng-k24 ===hng-ini ===hng-nkk
 753                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 754                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 755                  ===hng-okd ===hng-wad ===hng-kmi
 756                  ===hng-zkd ===hng-doh ===hng-jyu ===hng-tzs
 757                  ===hng-sai ===hng-kad ===hng-kss
 758                  ===hng-kyo ===hng-ykk ===hng-saa
 759                  ===hng-sab ===hng-wks ===hng-wke ===hng-smk
 760                  ===hng-sgs ===hng-sts
 761                  =shuowen-jiguge ===shuowen-jiguge4 ===shuowen-jiguge5))
 762          ;; (string-match "^=adobe-" (symbol-name name))
 763          )
 764         "(%-18s . %05d)\t; %c")
 765        ((memq name '(=hanyo-denshi/ks
 766                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 767                      =>>hanyo-denshi/ks
 768                      =koseki ==koseki
 769                      =mj ==mj ===mj =>>mj =>mj
 770                      =zihai mojikyo))
 771         "(%-18s . %06d)\t; %c")
 772        ((memq name '(=hanyo-denshi/tk ==hanyo-denshi/tk))
 773         "(%-18s . %08d)\t; %c")
 774        ((>= (charset-dimension name) 2)
 775         "(%-18s . #x%04X)\t; %c")
 776        (t
 777         "(%-18s . #x%02X)\t; %c"))
 778       name
 779       (if (= (charset-iso-graphic-plane name) 1)
 780           (logior value
 781                   (cond ((= (charset-dimension name) 1)
 782                          #x80)
 783                         ((= (charset-dimension name) 2)
 784                          #x8080)
 785                         ((= (charset-dimension name) 3)
 786                          #x808080)
 787                         (t 0)))
 788         value)
 789       (char-db-decode-isolated-char name value)))
 790     (if (and (= (charset-chars name) 94)
 791              (= (charset-dimension name) 2))
 792         (insert (format " [%02d-%02d]"
 793                         (- (lsh value -8) 32)
 794                         (- (logand value 255) 32))))
 795     )
 796    (t
 797     (insert (format "(%-18s . %s)" name value))
 798     ))
 799   (insert line-breaking))
 800
 801 (defun char-db-insert-relation-feature (char name value line-breaking
 802                                              ccss readable)
 803   (insert (format "(%-18s%s " name line-breaking))
 804   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 805         separator cell sources required-features
 806         ret)
 807     (while (consp value)
 808       (setq cell (car value))
 809       (if (integerp cell)
 810           (setq cell (decode-char '=ucs cell)))
 811       (cond
 812        ((eq name '->subsumptive)
 813         (when (or (not (some (lambda (atr)
 814                                (get-char-attribute cell atr))
 815                              char-db-ignored-attributes))
 816                   (some (lambda (ccs)
 817                           (encode-char cell ccs 'defined-only))
 818                         ccss))
 819           (if separator
 820               (insert lbs))
 821           (let ((char-db-ignored-attributes
 822                  (cons '<-subsumptive
 823                        char-db-ignored-attributes)))
 824             (insert-char-attributes cell readable nil nil 'for-sub-node))
 825           (setq separator lbs))
 826         )
 827        ((characterp cell)
 828         (setq sources
 829               (get-char-attribute
 830                char (intern (format "%s*sources" name))))
 831         (setq required-features nil)
 832         (dolist (source sources)
 833           (cond
 834            ((memq source '(JP
 835                            JP/Jouyou
 836                            shinjigen shinjigen@1ed shinjigen@rev))
 837             (setq required-features
 838                   (union required-features
 839                          '(=jis-x0208
 840                            =jis-x0208@1990
 841                            =jis-x0213-1@2000
 842                            =jis-x0213-1@2004
 843                            =jis-x0213-2
 844                            =jis-x0212
 845                            =jis-x0208@1983
 846                            =jis-x0208@1978
 847                            =shinjigen
 848                            =shinjigen@1ed
 849                            =shinjigen@rev
 850                            =shinjigen/+p@rev))))
 851            ((eq source 'CN)
 852             (setq required-features
 853                   (union required-features
 854                          '(=gb2312
 855                            =gb12345
 856                            =iso-ir165)))))
 857           (cond
 858            ((find-charset (setq ret (intern (format "=%s" source))))
 859             (setq required-features
 860                   (cons ret required-features)))
 861            (t (setq required-features
 862                     (cons source required-features)))))
 863         (cond ((string-match "@JP" (symbol-name name))
 864                (setq required-features
 865                      (union required-features
 866                             '(=jis-x0208
 867                               =jis-x0208@1990
 868                               =jis-x0213-1-2000
 869                               =jis-x0213-2-2000
 870                               =jis-x0212
 871                               =jis-x0208@1983
 872                               =jis-x0208@1978))))
 873               ((string-match "@CN" (symbol-name name))
 874                (setq required-features
 875                      (union required-features
 876                             '(=gb2312
 877                               =gb12345
 878                               =iso-ir165)))))
 879         (if separator
 880             (insert lbs))
 881         (if readable
 882             (insert (format "%S" cell))
 883           (char-db-insert-char-spec cell readable
 884                                     nil
 885                                     required-features))
 886         (setq separator lbs))
 887        ((consp cell)
 888         (if separator
 889             (insert lbs))
 890         (if (consp (car cell))
 891             (char-db-insert-char-spec cell readable)
 892           (char-db-insert-char-reference cell readable))
 893         (setq separator lbs))
 894        (t
 895         (if separator
 896             (insert separator))
 897         (insert (prin1-to-string cell))
 898         (setq separator " ")))
 899       (setq value (cdr value)))
 900     (insert ")")
 901     (insert line-breaking)))
 902
 903 (defun insert-char-attributes (char &optional readable attributes column
 904                                     for-sub-node)
 905   (unless column
 906     (setq column (current-column)))
 907   (let (name value ; has-long-ccs-name
 908         rest
 909         radical strokes
 910         (line-breaking
 911          (concat "\n" (make-string (1+ column) ?\ )))
 912         lbs cell separator ret
 913         key al cal
 914         dest-ccss ; sources required-features
 915         ccss)
 916     (let (atr-d)
 917       (setq attributes
 918             (sort (if attributes
 919                       (if (consp attributes)
 920                           (progn
 921                             (dolist (name attributes)
 922                               (unless (memq name char-db-ignored-attributes)
 923                                 (if (find-charset name)
 924                                     (push name ccss))
 925                                 (push name atr-d)))
 926                             atr-d))
 927                     (dolist (name (char-attribute-list))
 928                       (unless (memq name char-db-ignored-attributes)
 929                         (if (find-charset name)
 930                             (push name ccss))
 931                         (push name atr-d)))
 932                     atr-d)
 933                   #'char-attribute-name<)))
 934     (insert "(")
 935     (when (memq '<-subsumptive attributes)
 936       (when (or readable (not for-sub-node))
 937         (when (setq value (get-char-attribute char '<-subsumptive))
 938           (char-db-insert-relation-feature char '<-subsumptive value
 939                                            line-breaking
 940                                            ccss readable)))
 941       (setq attributes (delq '<-subsumptive attributes)))
 942     (when (and (memq '<-denotational attributes)
 943                (setq value (get-char-attribute char '<-denotational)))
 944       (char-db-insert-relation-feature char '<-denotational value
 945                                        line-breaking
 946                                        ccss readable)
 947       (setq attributes (delq '<-denotational attributes)))
 948     (when (and (memq '<-denotational@component attributes)
 949                (setq value (get-char-attribute char '<-denotational@component)))
 950       (char-db-insert-relation-feature char '<-denotational@component value
 951                                        line-breaking
 952                                        ccss readable)
 953       (setq attributes (delq '<-denotational@component attributes)))
 954     (when (and (memq '<-denotational@usage attributes)
 955                (setq value (get-char-attribute char '<-denotational@usage)))
 956       (char-db-insert-relation-feature char '<-denotational@usage value
 957                                        line-breaking
 958                                        ccss readable)
 959       (setq attributes (delq '<-denotational@usage attributes)))
 960     (when (and (memq 'name attributes)
 961                (setq value (get-char-attribute char 'name)))
 962       (insert (format
 963                (if (> (+ (current-column) (length value)) 48)
 964                    "(name . %S)%s"
 965                  "(name               . %S)%s")
 966                value line-breaking))
 967       (setq attributes (delq 'name attributes))
 968       )
 969     (when (and (memq 'name* attributes)
 970                (setq value (get-char-attribute char 'name*)))
 971       (insert (format
 972                (if (> (+ (current-column) (length value)) 48)
 973                    "(name* . %S)%s"
 974                  "(name*              . %S)%s")
 975                value line-breaking))
 976       (setq attributes (delq 'name* attributes))
 977       )
 978     (when (and (memq 'script attributes)
 979                (setq value (get-char-attribute char 'script)))
 980       (insert (format "(script\t\t%s)%s"
 981                       (mapconcat (function prin1-to-string)
 982                                  value " ")
 983                       line-breaking))
 984       (setq attributes (delq 'script attributes))
 985       )
 986     (dolist (name '(=>ucs =>ucs*))
 987       (when (and (memq name attributes)
 988                  (setq value (get-char-attribute char name)))
 989         (insert (format "(%-18s . #x%04X)\t; %c%s"
 990                         name value (decode-char '=ucs value)
 991                         line-breaking))
 992         (setq attributes (delq name attributes))))
 993     (dolist (name '(=>ucs@gb =>ucs@big5))
 994       (when (and (memq name attributes)
 995                  (setq value (get-char-attribute char name)))
 996         (insert (format "(%-18s . #x%04X)\t; %c%s"
 997                         name value
 998                         (decode-char (intern
 999                                       (concat "="
1000                                               (substring
1001                                                (symbol-name name) 2)))
1002                                      value)
1003                         line-breaking))
1004         (setq attributes (delq name attributes))
1005         ))
1006     ;; (dolist (name '(=>daikanwa))
1007     ;;   (when (and (memq name attributes)
1008     ;;              (setq value (get-char-attribute char name)))
1009     ;;     (insert
1010     ;;      (if (integerp value)
1011     ;;          (format "(%-18s . %05d)\t; %c%s"
1012     ;;                  name value (decode-char '=daikanwa value)
1013     ;;                  line-breaking)
1014     ;;        (format "(%-18s %s)\t; %c%s"
1015     ;;                name
1016     ;;                (mapconcat (function prin1-to-string)
1017     ;;                           value " ")
1018     ;;                (char-representative-of-daikanwa char)
1019     ;;                line-breaking)))
1020     ;;     (setq attributes (delq name attributes))))
1021     (when (and (memq 'general-category attributes)
1022                (setq value (get-char-attribute char 'general-category)))
1023       (insert (format
1024                "(general-category\t%s) ; %s%s"
1025                (mapconcat (lambda (cell)
1026                             (format "%S" cell))
1027                           value " ")
1028                (cond ((rassoc value unidata-normative-category-alist)
1029                       "Normative Category")
1030                      ((rassoc value unidata-informative-category-alist)
1031                       "Informative Category")
1032                      (t
1033                       "Unknown Category"))
1034                line-breaking))
1035       (setq attributes (delq 'general-category attributes))
1036       )
1037     (when (and (memq 'bidi-category attributes)
1038                (setq value (get-char-attribute char 'bidi-category)))
1039       (insert (format "(bidi-category\t. %S)%s"
1040                       value
1041                       line-breaking))
1042       (setq attributes (delq 'bidi-category attributes))
1043       )
1044     (unless (or (not (memq 'mirrored attributes))
1045                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
1046                     'empty))
1047       (insert (format "(mirrored\t\t. %S)%s"
1048                       value
1049                       line-breaking))
1050       (setq attributes (delq 'mirrored attributes))
1051       )
1052     (cond
1053      ((and (memq 'decimal-digit-value attributes)
1054            (setq value (get-char-attribute char 'decimal-digit-value)))
1055       (insert (format "(decimal-digit-value . %S)%s"
1056                       value
1057                       line-breaking))
1058       (setq attributes (delq 'decimal-digit-value attributes))
1059       (when (and (memq 'digit-value attributes)
1060                  (setq value (get-char-attribute char 'digit-value)))
1061         (insert (format "(digit-value\t . %S)%s"
1062                         value
1063                         line-breaking))
1064         (setq attributes (delq 'digit-value attributes))
1065         )
1066       (when (and (memq 'numeric-value attributes)
1067                  (setq value (get-char-attribute char 'numeric-value)))
1068         (insert (format "(numeric-value\t . %S)%s"
1069                         value
1070                         line-breaking))
1071         (setq attributes (delq 'numeric-value attributes))
1072         )
1073       )
1074      (t
1075       (when (and (memq 'digit-value attributes)
1076                  (setq value (get-char-attribute char 'digit-value)))
1077         (insert (format "(digit-value\t. %S)%s"
1078                         value
1079                         line-breaking))
1080         (setq attributes (delq 'digit-value attributes))
1081         )
1082       (when (and (memq 'numeric-value attributes)
1083                  (setq value (get-char-attribute char 'numeric-value)))
1084         (insert (format "(numeric-value\t. %S)%s"
1085                         value
1086                         line-breaking))
1087         (setq attributes (delq 'numeric-value attributes))
1088         )))
1089     (when (and (memq 'iso-10646-comment attributes)
1090                (setq value (get-char-attribute char 'iso-10646-comment)))
1091       (insert (format "(iso-10646-comment\t. %S)%s"
1092                       value
1093                       line-breaking))
1094       (setq attributes (delq 'iso-10646-comment attributes))
1095       )
1096     (when (and (memq 'morohashi-daikanwa attributes)
1097                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1098       (insert (format "(morohashi-daikanwa\t%s)%s"
1099                       (mapconcat (function prin1-to-string) value " ")
1100                       line-breaking))
1101       (setq attributes (delq 'morohashi-daikanwa attributes))
1102       )
1103     (setq radical nil
1104           strokes nil)
1105     (when (and (memq 'ideographic-radical attributes)
1106                (setq value (get-char-attribute char 'ideographic-radical)))
1107       (setq radical value)
1108       (insert (format "(ideographic-radical . %S)\t; %c%s"
1109                       radical
1110                       (ideographic-radical radical)
1111                       line-breaking))
1112       (setq attributes (delq 'ideographic-radical attributes))
1113       )
1114     (when (and (memq 'shuowen-radical attributes)
1115                (setq value (get-char-attribute char 'shuowen-radical)))
1116       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1117                       value
1118                       (shuowen-radical value)
1119                       line-breaking))
1120       (setq attributes (delq 'shuowen-radical attributes))
1121       )
1122     (let (key)
1123       (dolist (domain
1124                (append
1125                 char-db-feature-domains
1126                 (let (dest domain)
1127                   (dolist (feature (char-attribute-list))
1128                     (setq feature (symbol-name feature))
1129                     (when (string-match
1130                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1131                            feature)
1132                       (setq domain (intern (match-string 2 feature)))
1133                      (unless (memq domain dest)
1134                        (setq dest (cons domain dest)))))
1135                   (sort dest #'string<))))
1136         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1137         (when (and (memq key attributes)
1138                    (setq value (get-char-attribute char key)))
1139           (setq radical value)
1140           (insert (format "(%s . %S)\t; %c%s"
1141                           key
1142                           radical
1143                           (ideographic-radical radical)
1144                           line-breaking))
1145           (setq attributes (delq key attributes))
1146           )
1147         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1148         (when (and (memq key attributes)
1149                    (setq value (get-char-attribute char key)))
1150           (setq strokes value)
1151           (insert (format "(%s . %S)%s"
1152                           key
1153                           strokes
1154                           line-breaking))
1155           (setq attributes (delq key attributes))
1156           )
1157         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1158         (when (and (memq key attributes)
1159                    (setq value (get-char-attribute char key)))
1160           (insert (format "(%s       . %S)%s"
1161                           key
1162                           value
1163                           line-breaking))
1164           (setq attributes (delq key attributes))
1165           )
1166         (dolist (feature '(ideographic-radical
1167                            ideographic-strokes
1168                            total-strokes))
1169           (setq key (intern (format "%s@%s*sources" feature domain)))
1170           (when (and (memq key attributes)
1171                      (setq value (get-char-attribute char key)))
1172             (insert (format "(%s%s" key line-breaking))
1173             (dolist (cell value)
1174               (insert (format " %s" cell)))
1175             (insert ")")
1176             (insert line-breaking)
1177             (setq attributes (delq key attributes))
1178             ))
1179         ))
1180     (when (and (memq 'ideographic-strokes attributes)
1181                (setq value (get-char-attribute char 'ideographic-strokes)))
1182       (setq strokes value)
1183       (insert (format "(ideographic-strokes . %S)%s"
1184                       strokes
1185                       line-breaking))
1186       (setq attributes (delq 'ideographic-strokes attributes))
1187       )
1188     (when (and (memq 'kangxi-radical attributes)
1189                (setq value (get-char-attribute char 'kangxi-radical)))
1190       (unless (eq value radical)
1191         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1192                         value
1193                         (ideographic-radical value)
1194                         line-breaking))
1195         (or radical
1196             (setq radical value)))
1197       (setq attributes (delq 'kangxi-radical attributes))
1198       )
1199     (when (and (memq 'kangxi-strokes attributes)
1200                (setq value (get-char-attribute char 'kangxi-strokes)))
1201       (unless (eq value strokes)
1202         (insert (format "(kangxi-strokes\t . %S)%s"
1203                         value
1204                         line-breaking))
1205         (or strokes
1206             (setq strokes value)))
1207       (setq attributes (delq 'kangxi-strokes attributes))
1208       )
1209     ;; (when (and (memq 'japanese-radical attributes)
1210     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1211     ;;   (unless (eq value radical)
1212     ;;     (insert (format "(japanese-radical\t . %S)\t; %c%s"
1213     ;;                     value
1214     ;;                     (ideographic-radical value)
1215     ;;                     line-breaking))
1216     ;;     (or radical
1217     ;;         (setq radical value)))
1218     ;;   (setq attributes (delq 'japanese-radical attributes))
1219     ;;   )
1220     (when (and (memq 'japanese-strokes attributes)
1221                (setq value (get-char-attribute char 'japanese-strokes)))
1222       (unless (eq value strokes)
1223         (insert (format "(japanese-strokes\t . %S)%s"
1224                         value
1225                         line-breaking))
1226         (or strokes
1227             (setq strokes value)))
1228       (setq attributes (delq 'japanese-strokes attributes))
1229       )
1230     (when (and (memq 'cns-radical attributes)
1231                (setq value (get-char-attribute char 'cns-radical)))
1232       (insert (format "(cns-radical\t . %S)\t; %c%s"
1233                       value
1234                       (ideographic-radical value)
1235                       line-breaking))
1236       (setq attributes (delq 'cns-radical attributes))
1237       )
1238     (when (and (memq 'cns-strokes attributes)
1239                (setq value (get-char-attribute char 'cns-strokes)))
1240       (unless (eq value strokes)
1241         (insert (format "(cns-strokes\t . %S)%s"
1242                         value
1243                         line-breaking))
1244         (or strokes
1245             (setq strokes value)))
1246       (setq attributes (delq 'cns-strokes attributes))
1247       )
1248     ;; (when (and (memq 'ideographic- attributes)
1249     ;;            (setq value (get-char-attribute char 'ideographic-)))
1250     ;;   (insert "(ideographic-       ")
1251     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1252     ;;         separator nil)
1253     ;;   (while (consp value)
1254     ;;     (setq cell (car value))
1255     ;;     (if (integerp cell)
1256     ;;         (setq cell (decode-char '=ucs cell)))
1257     ;;     (cond ((characterp cell)
1258     ;;            (if separator
1259     ;;                (insert lbs))
1260     ;;            (if readable
1261     ;;                (insert (format "%S" cell))
1262     ;;              (char-db-insert-char-spec cell readable))
1263     ;;            (setq separator lbs))
1264     ;;           ((consp cell)
1265     ;;            (if separator
1266     ;;                (insert lbs))
1267     ;;            (if (consp (car cell))
1268     ;;                (char-db-insert-char-spec cell readable)
1269     ;;              (char-db-insert-char-reference cell readable))
1270     ;;            (setq separator lbs))
1271     ;;           (t
1272     ;;            (if separator
1273     ;;                (insert separator))
1274     ;;            (insert (prin1-to-string cell))
1275     ;;            (setq separator " ")))
1276     ;;     (setq value (cdr value)))
1277     ;;   (insert ")")
1278     ;;   (insert line-breaking)
1279     ;;   (setq attributes (delq 'ideographic- attributes)))
1280     (when (and (memq 'total-strokes attributes)
1281                (setq value (get-char-attribute char 'total-strokes)))
1282       (insert (format "(total-strokes       . %S)%s"
1283                       value
1284                       line-breaking))
1285       (setq attributes (delq 'total-strokes attributes))
1286       )
1287     ;; (when (and (memq '->ideograph attributes)
1288     ;;            (setq value (get-char-attribute char '->ideograph)))
1289     ;;   (insert (format "(->ideograph\t%s)%s"
1290     ;;                   (mapconcat (lambda (code)
1291     ;;                                (cond ((symbolp code)
1292     ;;                                       (symbol-name code))
1293     ;;                                      ((integerp code)
1294     ;;                                       (format "#x%04X" code))
1295     ;;                                      (t
1296     ;;                                       (format "%s %S"
1297     ;;                                               line-breaking code))))
1298     ;;                              value " ")
1299     ;;                   line-breaking))
1300     ;;   (setq attributes (delq '->ideograph attributes))
1301     ;;   )
1302     ;; (when (and (memq '->decomposition attributes)
1303     ;;            (setq value (get-char-attribute char '->decomposition)))
1304     ;;   (insert (format "(->decomposition\t%s)%s"
1305     ;;                   (mapconcat (lambda (code)
1306     ;;                                (cond ((symbolp code)
1307     ;;                                       (symbol-name code))
1308     ;;                                      ((characterp code)
1309     ;;                                       (if readable
1310     ;;                                           (format "%S" code)
1311     ;;                                         (format "#x%04X"
1312     ;;                                                 (char-int code))
1313     ;;                                         ))
1314     ;;                                      ((integerp code)
1315     ;;                                       (format "#x%04X" code))
1316     ;;                                      (t
1317     ;;                                       (format "%s%S" line-breaking code))))
1318     ;;                              value " ")
1319     ;;                   line-breaking))
1320     ;;   (setq attributes (delq '->decomposition attributes))
1321     ;;   )
1322     (if (equal (get-char-attribute char '->titlecase)
1323                (get-char-attribute char '->uppercase))
1324         (setq attributes (delq '->titlecase attributes)))
1325     (when (and (memq '->mojikyo attributes)
1326                (setq value (get-char-attribute char '->mojikyo)))
1327       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1328                       value (decode-char 'mojikyo value)
1329                       line-breaking))
1330       (setq attributes (delq '->mojikyo attributes))
1331       )
1332     (unless readable
1333       (dolist (ignored '(composition
1334                          ->denotational <-subsumptive ->ucs-unified
1335                          ;; ->ideographic-component-forms
1336                          ))
1337         (setq attributes (delq ignored attributes))))
1338     (while attributes
1339       (setq name (car attributes))
1340       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1341                   'value-is-empty)
1342         (cond ((setq ret (find-charset name))
1343                (setq name (charset-name ret))
1344                (when (not (memq name dest-ccss))
1345                  (setq dest-ccss (cons name dest-ccss))
1346                  (char-db-insert-ccs-feature name value line-breaking))
1347                )
1348               ((string-match "^=>ucs@" (symbol-name name))
1349                (insert (format "(%-18s . #x%04X)\t; %c%s"
1350                                name value (decode-char '=ucs value)
1351                                line-breaking))
1352                )
1353               ((eq name 'jisx0208-1978/4X)
1354                (insert (format "(%-18s . #x%04X)%s"
1355                                name value
1356                                line-breaking))
1357                )
1358               ((and
1359                 (not readable)
1360                 (not (eq name '->subsumptive))
1361                 (not (eq name '->uppercase))
1362                 (not (eq name '->lowercase))
1363                 (not (eq name '->titlecase))
1364                 (not (eq name '->canonical))
1365                 (not (eq name '->Bopomofo))
1366                 (not (eq name '->mistakable))
1367                 (not (eq name '->ideographic-variants))
1368                 (null (get-char-attribute
1369                        char (intern (format "%s*sources" name))))
1370                 (not (string-match "\\*sources$" (symbol-name name)))
1371                 (null (get-char-attribute
1372                        char (intern (format "%s*note" name))))
1373                 (not (string-match "\\*note$" (symbol-name name)))
1374                 (or (eq name '<-identical)
1375                     (eq name '<-uppercase)
1376                     (eq name '<-lowercase)
1377                     (eq name '<-titlecase)
1378                     (eq name '<-canonical)
1379                     (eq name '<-ideographic-variants)
1380                     ;; (eq name '<-synonyms)
1381                     (string-match "^<-synonyms" (symbol-name name))
1382                     (eq name '<-mistakable)
1383                     (when (string-match "^->" (symbol-name name))
1384                       (cond
1385                        ((string-match "^->fullwidth" (symbol-name name))
1386                         (not (and (consp value)
1387                                   (characterp (car value))
1388                                   (encode-char
1389                                    (car value) '=ucs 'defined-only)))
1390                         )
1391                        (t)))
1392                     ))
1393                )
1394               ((or (eq name 'ideographic-structure)
1395                    (eq name 'ideographic-combination)
1396                    ;; (eq name 'ideographic-)
1397                    (eq name '=decomposition)
1398                    (char-feature-base-name= '=decomposition name)
1399                    (char-feature-base-name= '=>decomposition name)
1400                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1401                    ;;               (symbol-name name))
1402                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1403                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1404                                  (symbol-name name))
1405                    )
1406                (char-db-insert-relation-feature char name value
1407                                                 line-breaking
1408                                                 ccss readable))
1409               ;; ((memq name '(ideograph=
1410               ;;               original-ideograph-of
1411               ;;               ancient-ideograph-of
1412               ;;               vulgar-ideograph-of
1413               ;;               wrong-ideograph-of
1414               ;;               ;; simplified-ideograph-of
1415               ;;               ideographic-variants
1416               ;;               ;; ideographic-different-form-of
1417               ;;               ))
1418               ;;  (insert (format "(%-18s%s " name line-breaking))
1419               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1420               ;;        separator nil)
1421               ;;  (while (consp value)
1422               ;;    (setq cell (car value))
1423               ;;    (if (and (consp cell)
1424               ;;             (consp (car cell)))
1425               ;;        (progn
1426               ;;          (if separator
1427               ;;              (insert lbs))
1428               ;;          (char-db-insert-alist cell readable)
1429               ;;          (setq separator lbs))
1430               ;;      (if separator
1431               ;;          (insert separator))
1432               ;;      (insert (prin1-to-string cell))
1433               ;;      (setq separator " "))
1434               ;;    (setq value (cdr value)))
1435               ;;  (insert ")")
1436               ;;  (insert line-breaking))
1437               ((consp value)
1438                (insert (format "(%-18s " name))
1439                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1440                      separator nil)
1441                (while (consp value)
1442                  (setq cell (car value))
1443                  (if (and (consp cell)
1444                           (consp (car cell))
1445                           (setq ret (condition-case nil
1446                                         (find-char cell)
1447                                       (error nil))))
1448                      (progn
1449                        (setq rest cell
1450                              al nil
1451                              cal nil)
1452                        (while rest
1453                          (setq key (car (car rest)))
1454                          (if (find-charset key)
1455                              (setq cal (cons key cal))
1456                            (setq al (cons key al)))
1457                          (setq rest (cdr rest)))
1458                        (if separator
1459                            (insert lbs))
1460                        (insert-char-attributes ret
1461                                                readable
1462                                                al ; cal
1463                                                nil 'for-sub-node)
1464                        (setq separator lbs))
1465                    (setq ret (prin1-to-string cell))
1466                    (if separator
1467                        (if (< (+ (current-column)
1468                                  (length ret)
1469                                  (length separator))
1470                               76)
1471                            (insert separator)
1472                          (insert lbs)))
1473                    (insert ret)
1474                    (setq separator " "))
1475                  (setq value (cdr value)))
1476                (insert ")")
1477                (insert line-breaking))
1478               (t
1479                (insert (format "(%-18s" name))
1480                (setq ret (prin1-to-string value))
1481                (unless (< (+ (current-column)
1482                              (length ret)
1483                              3)
1484                           76)
1485                  (insert line-breaking))
1486                (insert " . " ret ")" line-breaking)
1487                ;; (insert (format "(%-18s . %S)%s"
1488                ;;                 name value
1489                ;;                 line-breaking))
1490                )
1491               ))
1492       (setq attributes (cdr attributes)))
1493     (insert ")")))
1494
1495 (defun insert-char-data (char &optional readable
1496                               attributes)
1497   (save-restriction
1498     (narrow-to-region (point)(point))
1499     (insert "(define-char
1500   '")
1501     (insert-char-attributes char readable attributes)
1502     (insert ")\n")
1503     (goto-char (point-min))
1504     (while (re-search-forward "[ \t]+$" nil t)
1505       (replace-match ""))
1506     ;; from tabify.
1507     (goto-char (point-min))
1508     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1509       (let ((column (current-column))
1510             (indent-tabs-mode t))
1511         (delete-region (match-beginning 0) (point))
1512         (indent-to column)))
1513     (goto-char (point-max))
1514     ;; (tabify (point-min)(point-max))
1515     ))
1516
1517 (defun insert-char-data-with-variant (char &optional printable
1518                                            no-ucs-unified
1519                                            script excluded-script)
1520   (insert-char-data char printable)
1521   (let ((variants (char-variants char))
1522         rest
1523         variant vs ret)
1524     (setq variants (sort variants #'<))
1525     (setq rest variants)
1526     (setq variants (cons char variants))
1527     (while rest
1528       (setq variant (car rest))
1529       (unless (get-char-attribute variant '<-subsumptive)
1530         (if (and (or (null script)
1531                      (null (setq vs (get-char-attribute variant 'script)))
1532                      (memq script vs))
1533                  (or (null excluded-script)
1534                      (null (setq vs (get-char-attribute variant 'script)))
1535                      (not (memq excluded-script vs))))
1536             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1537               (insert-char-data variant printable)
1538               (if (setq ret (char-variants variant))
1539                   (while ret
1540                     (or (memq (car ret) variants)
1541                         ;; (get-char-attribute (car ret) '<-subsumptive)
1542                         (setq rest (nconc rest (list (car ret)))))
1543                     (setq ret (cdr ret)))))))
1544       (setq rest (cdr rest)))))
1545
1546 (defun insert-char-range-data (min max &optional script excluded-script)
1547   (let ((code min)
1548         char)
1549     (while (<= code max)
1550       (setq char (decode-char '=ucs code))
1551       (if (encode-char char '=ucs 'defined-only)
1552           (insert-char-data-with-variant char nil 'no-ucs-unified
1553                                          script excluded-script))
1554       (setq code (1+ code)))))
1555
1556 (defun write-char-range-data-to-file (min max file
1557                                           &optional script excluded-script)
1558   (let ((coding-system-for-write char-db-file-coding-system))
1559     (with-temp-buffer
1560       (insert (format ";; -*- coding: %s -*-\n"
1561                       char-db-file-coding-system))
1562       (insert-char-range-data min max script excluded-script)
1563       (write-region (point-min)(point-max) file))))
1564
1565 (defvar what-character-original-window-configuration)
1566
1567 ;;;###autoload
1568 (defun what-char-definition (char)
1569   (interactive (list (char-after)))
1570   (let ((buf (get-buffer-create "*Character Description*"))
1571         (the-buf (current-buffer))
1572         (win-conf (current-window-configuration)))
1573     (pop-to-buffer buf)
1574     (make-local-variable 'what-character-original-window-configuration)
1575     (setq what-character-original-window-configuration win-conf)
1576     (setq buffer-read-only nil)
1577     (erase-buffer)
1578     (condition-case err
1579         (progn
1580           (insert-char-data-with-variant char 'printable)
1581           (unless (char-attribute-alist char)
1582             (insert (format ";; = %c\n"
1583                             (let* ((rest (split-char char))
1584                                    (ccs (pop rest))
1585                                    (code (pop rest)))
1586                               (while rest
1587                                 (setq code (logior (lsh code 8)
1588                                                    (pop rest))))
1589                               (decode-char ccs code)))))
1590           ;; (char-db-update-comment)
1591           (set-buffer-modified-p nil)
1592           (view-mode the-buf (lambda (buf)
1593                                (set-window-configuration
1594                                 what-character-original-window-configuration)
1595                                ))
1596           (goto-char (point-min)))
1597       (error (progn
1598                (set-window-configuration
1599                 what-character-original-window-configuration)
1600                (signal (car err) (cdr err)))))))
1601
1602
1603 ;;; @ end
1604 ;;;
1605
1606 (provide 'char-db-util)
1607
1608 ;;; char-db-util.el ends here