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,
   5 ;;   2020 MORIOKA Tomohiko.
   6
   7 ;; Author: MORIOKA Tomohiko <tomo@kanji.zinbun.kyoto-u.ac.jp>
   8 ;; Keywords: CHISE, Character Database, ISO/IEC 10646, UCS, Unicode, MULE.
   9
  10 ;; This file is part of XEmacs CHISE.
  11
  12 ;; XEmacs CHISE is free software; you can redistribute it and/or
  13 ;; modify it under the terms of the GNU General Public License as
  14 ;; published by the Free Software Foundation; either version 2, or (at
  15 ;; your option) any later version.
  16
  17 ;; XEmacs CHISE is distributed in the hope that it will be useful,
  18 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  19 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  20 ;; General Public License for more details.
  21
  22 ;; You should have received a copy of the GNU General Public License
  23 ;; along with XEmacs CHISE; see the file COPYING.  If not, write to
  24 ;; the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  25 ;; Boston, MA 02111-1307, USA.
  26
  27 ;;; Code:
  28
  29 (require 'chise-subr)
  30 (require 'ideograph-subr)
  31
  32 (defconst unidata-normative-category-alist
  33   '(("Lu" letter        uppercase)
  34     ("Ll" letter        lowercase)
  35     ("Lt" letter        titlecase)
  36     ("Mn" mark          non-spacing)
  37     ("Mc" mark          spacing-combining)
  38     ("Me" mark          enclosing)
  39     ("Nd" number        decimal-digit)
  40     ("Nl" number        letter)
  41     ("No" number        other)
  42     ("Zs" separator     space)
  43     ("Zl" separator     line)
  44     ("Zp" separator     paragraph)
  45     ("Cc" other         control)
  46     ("Cf" other         format)
  47     ("Cs" other         surrogate)
  48     ("Co" other         private-use)
  49     ("Cn" other         not-assigned)))
  50
  51 (defconst unidata-informative-category-alist
  52   '(("Lm" letter        modifier)
  53     ("Lo" letter        other)
  54     ("Pc" punctuation   connector)
  55     ("Pd" punctuation   dash)
  56     ("Ps" punctuation   open)
  57     ("Pe" punctuation   close)
  58     ("Pi" punctuation   initial-quote)
  59     ("Pf" punctuation   final-quote)
  60     ("Po" punctuation   other)
  61     ("Sm" symbol        math)
  62     ("Sc" symbol        currency)
  63     ("Sk" symbol        modifier)
  64     ("So" symbol        other)
  65     ))
  66
  67 (defconst shuowen-radicals
  68   [?一 ?上 ?示 ?三 ?王 ?玉 ?玨 ?气 ?士 ?丨 ; 010
  69    ?屮 ?艸 ?蓐 ?茻 ?小 ?八 ?釆 ?半 ?牛 ?犛 ; 020
  70    ?告 ?口 ?凵 ?吅 ?哭 ?走 ?止 ?癶 ?步 ?此 ; 030
  71    ?正 ?是 ?辵 ?彳 ?廴 ?㢟 ?行 ?齒 ?牙 ?足 ; 040
  72    ?疋 ?品 ?龠 ?冊 ?㗊 ?舌 ?干 ?𧮫 ?只 ?㕯 ; 050
  73    ?句 ?丩 ?古 ?十 ?卅 ?言 ?誩 ?音 ?䇂 ?丵 ; 060
  74    ?菐 ?廾 ?𠬜 ?共 ?異 ?舁 ?𦥑 ?䢅 ?爨 ?革 ; 070
  75    ?鬲 ?䰜 ?爪 ?𠃨 ?鬥 ?又 ?𠂇 ?史 ?支 ?𦘒 ; 080
  76    ?聿 ?畫 ?隶 ?臤 ?臣 ?殳 ?殺 ?𠘧 ?寸 ?皮 ; 090
  77    ?㼱 ?攴 ?敎 ?卜 ?用 ?爻 ?㸚 ?𥄎 ?目 ?䀠 ; 100
  78    ?眉 ?盾 ?自 ?白 ?鼻 ?皕 ?習 ?羽 ?隹 ?奞 ; 110
  79    ?雈 ?𦫳 ?𥄕 ?羊 ?羴 ?瞿 ?雔 ?雥 ?鳥 ?烏 ; 120
  80    ?𠦒 ?冓 ?幺 ?𢆶 ?叀 ?玄 ?予 ?放 ?𠬪 ?𣦼 ; 130
  81    ?歺 ?死 ?冎 ?骨 ?肉 ?筋 ?刀 ?刃 ?㓞 ?丯 ; 140
  82    ?耒 ?角 ?竹 ?箕 ?丌 ?左 ?工 ?㠭 ?巫 ?甘 ; 150
  83    ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?旨 ?喜 ?壴 ; 160
  84 ;  ?旨 ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?喜 ?壴 ; 160
  85    ?鼓 ?豈 ?豆 ?豊 ?豐 ?䖒 ?虍 ?虎 ?虤 ?皿 ; 170
  86    ?𠙴 ?去 ?血 ?丶 ?丹 ?青 ?井 ?皀 ?鬯 ?食 ; 180
  87    ?亼 ?會 ?倉 ?入 ?缶 ?矢 ?高 ?冂 ?𩫏 ?京 ; 190
  88    ?亯 ?𣆪 ?畗 ?㐭 ?嗇 ?來 ?麥 ?夊 ?舛 ?䑞 ; 200
  89    ?韋 ?弟 ?夂 ?久 ?桀 ?木 ?東 ?林 ?才 ?叒 ; 210
  90    ?之 ?帀 ?出 ?𣎵 ?生 ?乇 ?𠂹 ?𠌶 ?華 ?𥝌 ; 220
  91    ?稽 ?巢 ?桼 ?束 ?㯻 ?囗 ?員 ?貝 ?邑 ?𨛜 ; 230
  92    ?日 ?旦 ?倝 ?㫃 ?冥 ?晶 ?月 ?有 ?明 ?囧 ; 240
  93    ?夕 ?多 ?毌 ?𢎘 ?𣐺 ?𠧪 ?齊 ?朿 ?片 ?鼎 ; 250
  94    ?克 ?彔 ?禾 ?秝 ?黍 ?香 ?米 ?毇 ?臼 ?凶 ; 260
  95    ?𣎳 ?𣏟 ?麻 ?尗 ?耑 ?韭 ?瓜 ?瓠 ?宀 ?宮 ; 270
  96    ?呂 ?穴 ?㝱 ?疒 ?冖 ?𠔼 ?冃 ?㒳 ?网 ?襾 ; 280
  97    ?巾 ?巿 ?帛 ?白 ?㡀 ?黹 ?人 ?𠤎 ?匕 ?从 ; 290
  98    ?比 ?北 ?丘 ?㐺 ?𡈼 ?重 ?臥 ?身 ?㐆 ?衣 ; 300
  99    ?裘 ?老 ?毛 ?毳 ?尸 ?尺 ?尾 ?履 ?舟 ?方 ; 310
 100    ?儿 ?兄 ?兂 ?皃 ?𠑹 ?先 ?秃 ?見 ?覞 ?欠 ; 320
 101    ?㱃 ?㳄 ?旡 ?頁 ?𦣻 ?面 ?丏 ?首 ?𥄉 ?須 ; 330
 102    ?彡 ?彣 ?文 ?髟 ?后 ?司 ?卮 ?卩 ?印 ?色 ; 340
 103    ?𠨍 ?辟 ?勹 ?包 ?茍 ?鬼 ?甶 ?厶 ?嵬 ?山 ; 350
 104    ?屾 ?屵 ?广 ?厂 ?丸 ?危 ?石 ?長 ?勿 ?冄 ; 360
 105    ?而 ?豕 ?㣇 ?彑 ?豚 ?豸 ?𤉡 ?易 ?象 ?馬 ; 370
 106    ?𢊁 ?鹿 ?麤 ?㲋 ?兔 ?萈 ?犬 ?㹜 ?鼠 ?能 ; 380
 107    ?熊 ?火 ?炎 ?黑 ?囪 ?焱 ?炙 ?赤 ?大 ?亦 ; 390
 108    ?夨 ?夭 ?交 ?尣 ?壺 ?壹 ?㚔 ?奢 ?亢 ?夲 ; 400
 109    ?夰 ?亣 ?夫 ?立 ?竝 ?囟 ?思 ?心 ?惢 ?水 ; 410
 110    ?沝 ?瀕 ?𡿨 ?巜 ?川 ?泉 ?灥 ?永 ?𠂢 ?谷 ; 420
 111    ?仌 ?雨 ?雲 ?魚 ?𩺰 ?燕 ?龍 ?飛 ?非 ?卂 ; 430
 112    ?𠃉 ?不 ?至 ?西 ?鹵 ?鹽 ?戶 ?門 ?耳 ?𦣝 ; 440
 113    ?手 ?𠦬 ?女 ?毋 ?民 ?丿 ?𠂆 ?乁 ?氏 ?氐 ; 450
 114    ?戈 ?戉 ?我 ?亅 ?琴 ?𠃊 ?亡 ?匸 ?匚 ?曲 ; 460
 115    ?甾 ?瓦 ?弓 ?弜 ?弦 ?系 ?糸 ?素 ?絲 ?率 ; 470
 116    ?虫 ?䖵 ?蟲 ?風 ?它 ?龜 ?黽 ?卵 ?二 ?土 ; 480
 117    ?垚 ?堇 ?里 ?田 ?畕 ?黃 ?男 ?力 ?劦 ?金 ; 490
 118    ?幵 ?勺 ?几 ?且 ?斤 ?斗 ?矛 ?車 ?𠂤 ?𨸏 ; 500
 119    ?𨺅 ?厽 ?四 ?宁 ?叕 ?亞 ?五 ?六 ?七 ?九 ; 510
 120    ?禸 ?嘼 ?甲 ?乙 ?丙 ?丁 ?戊 ?己 ?巴 ?庚 ; 520
 121    ?辛 ?辡 ?壬 ?癸 ?子 ?了 ?孨 ?𠫓 ?丑 ?寅 ; 530
 122    ?卯 ?辰 ?巳 ?午 ?未 ?申 ?酉 ?酋 ?戌 ?亥 ; 540
 123    ])
 124
 125 (defun shuowen-radical (number)
 126   (aref shuowen-radicals (1- number)))
 127
 128 (defvar char-db-file-coding-system 'utf-8-mcs-er)
 129
 130 (defvar char-db-ignored-attributes '(ideographic-products))
 131
 132 (defvar char-db-coded-charset-priority-list
 133   '(ascii
 134     control-1
 135     latin-iso8859-1
 136     latin-iso8859-2
 137     latin-iso8859-3
 138     latin-iso8859-4
 139     latin-iso8859-9
 140     latin-jisx0201
 141     cyrillic-iso8859-5
 142     greek-iso8859-7
 143     thai-tis620
 144     ;; =mj
 145     ;; =adobe-japan1-0
 146     ;; =adobe-japan1-1
 147     ;; =adobe-japan1-2
 148     ;; =adobe-japan1-3
 149     ;; =adobe-japan1-4
 150     ;; =adobe-japan1-5
 151     ;; =adobe-japan1-6
 152     =jis-x0208
 153     =jis-x0208@1978
 154     =jis-x0208@1983
 155     japanese-jisx0212
 156     chinese-gb2312
 157     =jis-x0208@1990
 158     chinese-cns11643-1
 159     chinese-cns11643-2
 160     chinese-cns11643-3
 161     chinese-cns11643-4
 162     chinese-cns11643-5
 163     chinese-cns11643-6
 164     chinese-cns11643-7
 165     =jis-x0213-1
 166     =jis-x0213-1@2000
 167     =jis-x0213-1@2004
 168     =jis-x0213-2
 169     korean-ksc5601
 170     chinese-isoir165
 171     katakana-jisx0201
 172     hebrew-iso8859-8
 173     chinese-gb12345
 174     latin-viscii
 175     ethiopic-ucs
 176     =big5-cdp
 177     =hanyo-denshi/ja
 178     =hanyo-denshi/jb
 179     =hanyo-denshi/jc
 180     =hanyo-denshi/jd
 181     =hanyo-denshi/ft
 182     =hanyo-denshi/ia
 183     =hanyo-denshi/ib
 184     =hanyo-denshi/hg
 185     =hanyo-denshi/jt
 186     =hanyo-denshi/ks
 187     =hanyo-denshi/tk
 188     ;; ==mj
 189     ;; ==adobe-japan1-0
 190     ;; ==adobe-japan1-1
 191     ;; ==adobe-japan1-2
 192     ;; ==adobe-japan1-3
 193     ;; ==adobe-japan1-4
 194     ;; ==adobe-japan1-5
 195     ;; ==adobe-japan1-6
 196     ==jis-x0208
 197     ==jis-x0213-1
 198     ==jis-x0213-2
 199     ==jis-x0212
 200     ==hanyo-denshi/ja
 201     ==hanyo-denshi/jb
 202     ==hanyo-denshi/jc
 203     ==hanyo-denshi/ft
 204     ==hanyo-denshi/ia
 205     ==hanyo-denshi/ib
 206     ==hanyo-denshi/hg
 207     ==hanyo-denshi/jt
 208     ==hanyo-denshi/ks
 209     ==hanyo-denshi/tk
 210     =gt
 211     =gt-k
 212     =daikanwa
 213     =daijiten
 214     =daikanwa@rev2
 215     =daikanwa@rev1
 216     =daikanwa/+p
 217     =daikanwa/ho
 218     ==daikanwa
 219     ==daijiten
 220     =cbeta
 221     ideograph-hanziku-1
 222     ideograph-hanziku-2
 223     ideograph-hanziku-3
 224     ideograph-hanziku-4
 225     ideograph-hanziku-5
 226     ideograph-hanziku-6
 227     ideograph-hanziku-7
 228     ideograph-hanziku-8
 229     ideograph-hanziku-9
 230     ideograph-hanziku-10
 231     ideograph-hanziku-11
 232     ideograph-hanziku-12
 233     ;; =>>>adobe-japan1-0
 234     ;; =>>>adobe-japan1-1
 235     ;; =>>>adobe-japan1-2
 236     ;; =>>>adobe-japan1-3
 237     ;; =>>>adobe-japan1-4
 238     ;; =>>>adobe-japan1-5
 239     ;; =>>>adobe-japan1-6
 240     ;; =>>>jis-x0208
 241     ;; =>>>jis-x0213-1
 242     ;; =>>>jis-x0213-2
 243     ;; =>>>hanyo-denshi/ja
 244     ;; =>>>hanyo-denshi/jb
 245     ;; =>>>hanyo-denshi/jc
 246     ;; =>>>hanyo-denshi/ft
 247     ;; =>>>hanyo-denshi/ib
 248     ;; =>>>hanyo-denshi/hg
 249     ;; =>>>hanyo-denshi/jt
 250     ;; =>>>hanyo-denshi/ks
 251     ;; =>>>gt
 252     =>>adobe-japan1-0
 253     =>>adobe-japan1-1
 254     =>>adobe-japan1-2
 255     =>>adobe-japan1-3
 256     =>>adobe-japan1-4
 257     =>>adobe-japan1-5
 258     =>>adobe-japan1-6
 259     =>>jis-x0208
 260     =>>jis-x0213-1
 261     =>>jis-x0213-1@2000
 262     =>>jis-x0213-1@2004
 263     =>>jis-x0213-2
 264     =>>jis-x0208@1978
 265     =>>hanyo-denshi/ft
 266     =>>hanyo-denshi/jt
 267     =>>hanyo-denshi/ks
 268     =>>gt
 269     =>>daikanwa
 270     =>>cbeta
 271     =+>jis-x0208
 272     =+>jis-x0213-1
 273     =+>jis-x0213-2
 274     =+>adobe-japan1-0
 275     =+>adobe-japan1-1
 276     =+>adobe-japan1-2
 277     =+>adobe-japan1-3
 278     =+>adobe-japan1-4
 279     =+>adobe-japan1-5
 280     =+>adobe-japan1-6
 281     =+>jis-x0208@1978
 282     =>jis-x0208
 283     =>jis-x0208@1997
 284     =>jis-x0213-1
 285     =>jis-x0213-1@2000
 286     =>jis-x0213-1@2004
 287     =>jis-x0213-2
 288     ==>ucs@bucs
 289     ==>daijiten
 290     ;; =>ucs@hanyo-denshi
 291     =>ucs@iso
 292     =>ucs@unicode
 293     =>ucs@jis
 294     =>ucs@cns
 295     =>ucs@ks
 296     =+>ucs@iso
 297     =+>ucs@unicode
 298     =+>ucs@jis
 299     =+>ucs@jis/1990
 300     =+>ucs@cns
 301     =+>ucs@ks
 302     =>>ucs@iso
 303     =>>ucs@unicode
 304     =>>ucs@jis
 305     =>>ucs@cns
 306     =>>>ucs@iso
 307     =>>>ucs@unicode
 308     ==ucs@iso
 309     ==ucs@unicode
 310     ;; ==ucs@cns
 311     ==gb2312
 312     ==ks-x1001
 313     ==cns11643-1
 314     ==cns11643-2
 315     ==cns11643-3
 316     ==cns11643-4
 317     ==cns11643-5
 318     ==cns11643-6
 319     ==cns11643-7
 320     ==gt
 321     ==jis-x0208@1990
 322     ;; ==jis-x0208@1983
 323     ==jis-x0208@1978
 324     ==gt-k
 325     =ucs@iso
 326     =ucs@unicode
 327     =ucs@cns
 328     ==big5-cdp
 329     ==cbeta
 330     =>>big5-cdp
 331     =>>gt-k
 332     =+>gt
 333     =+>big5-cdp
 334     =>big5-cdp
 335     =>big5-cdp@iwds-1
 336     =>mj
 337     =>iwds-1
 338     =big5
 339     =big5-eten
 340     =zinbun-oracle
 341     =>zinbun-oracle
 342     =ruimoku-v6
 343     =>>ruimoku-v6
 344     ==ruimoku-v6
 345     =jef-china3
 346     =shinjigen
 347     =>cns11643-5
 348     =>cns11643-7
 349     =>gt
 350     =>gt-k
 351     =>daikanwa
 352     =>daikanwa/ho
 353     =>cbeta
 354     =ucs-var-001
 355     =ucs-var-002
 356     =ucs-var-003
 357     =ucs-var-004
 358     =ucs-var-005
 359     =ucs-var-006
 360     =ucs-var-007
 361     =ucs-var-008
 362     =ucs-var-010
 363     =ucs-itaiji-001
 364     =ucs-itaiji-002
 365     =ucs-itaiji-003
 366     =ucs-itaiji-004
 367     =ucs-itaiji-005
 368     =ucs-itaiji-006
 369     =ucs-itaiji-007
 370     =ucs-itaiji-008
 371     =ucs-itaiji-009
 372     =ucs-itaiji-010
 373     =ucs-itaiji-011
 374     =ucs-itaiji-012
 375     =+>ucs-var-003
 376     =>ucs-itaiji-001
 377     =>ucs-itaiji-002
 378     =>ucs-itaiji-003
 379     =>ucs-itaiji-004
 380     =>ucs-itaiji-005
 381     =>ucs-itaiji-006
 382     =>ucs-itaiji-007
 383     =>ucs-itaiji-008
 384     =>ucs-itaiji-009
 385     =>ucs-itaiji-010
 386     =>ucs-itaiji-011
 387     =big5-cdp-var-001
 388     =big5-cdp-var-002
 389     =big5-cdp-var-003
 390     =big5-cdp-var-004
 391     =big5-cdp-var-005
 392     =big5-cdp-var-007
 393     =big5-cdp-var-010
 394     =big5-cdp-itaiji-001
 395     =big5-cdp-itaiji-002
 396     =big5-cdp-itaiji-003
 397     =big5-cdp-itaiji-004
 398     =big5-cdp-itaiji-005
 399     =>big5-cdp-itaiji-001
 400     =>big5-cdp-itaiji-002
 401     =>big5-cdp-itaiji-003
 402     =>big5-cdp-itaiji-004
 403     =>big5-cdp-var-001
 404     =>ucs@iwds-1
 405     =>ucs@cognate
 406     =>ucs@component
 407     =>ucs@iwds-1/normalized
 408     =>ucs-itaiji-001@iwds-1
 409     =>ucs-itaiji-002@iwds-1
 410     =>big5-cdp@component
 411     =>big5-cdp@cognate
 412     ==ucs@gb
 413     =ucs@gb
 414     ==ucs-var-002
 415     =ucs@JP/hanazono))
 416
 417
 418 ;;; @ char-db formatters
 419 ;;;
 420
 421 (defun char-db-make-char-spec (char)
 422   (let (ret char-spec)
 423     (cond ((characterp char)
 424            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 425                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 426                   (setq char-spec (list (cons '=ucs ret)))
 427                   (cond ((setq ret (get-char-attribute char 'name))
 428                          (setq char-spec (cons (cons 'name ret) char-spec))
 429                          )
 430                         ((setq ret (get-char-attribute char 'name*))
 431                          (setq char-spec (cons (cons 'name* ret) char-spec))
 432                          ))
 433                   )
 434                  ((encode-char char '=mj 'defined-only)
 435                   (setq char-spec nil)
 436                   (dolist (ccs (charset-list))
 437                     (if (and (or (eq ccs '=mj)
 438                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 439                                  (string-match "=ucs@" (symbol-name ccs))
 440                                  )
 441                              (setq ccs (charset-name ccs))
 442                              (null (assq ccs char-spec))
 443                              (setq ret (encode-char char ccs 'defined-only)))
 444                         (setq char-spec (cons (cons ccs ret) char-spec))))
 445                   )
 446                  ((encode-char char '==mj 'defined-only)
 447                   (setq char-spec nil)
 448                   (dolist (ccs (charset-list))
 449                     (if (and (or (eq ccs '==mj)
 450                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 451                                  (string-match "=ucs@" (symbol-name ccs))
 452                                  )
 453                              (setq ccs (charset-name ccs))
 454                              (null (assq ccs char-spec))
 455                              (setq ret (encode-char char ccs 'defined-only)))
 456                         (setq char-spec (cons (cons ccs ret) char-spec))))
 457                   )
 458                  ((encode-char char '=adobe-japan1 'defined-only)
 459                   (setq char-spec nil)
 460                   (dolist (ccs (charset-list))
 461                     (if (and (or (memq ccs
 462                                        '(=adobe-japan1-0
 463                                          =adobe-japan1-1
 464                                          =adobe-japan1-2
 465                                          =adobe-japan1-3
 466                                          =adobe-japan1-4
 467                                          =adobe-japan1-5
 468                                          =adobe-japan1-6
 469                                          ))
 470                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 471                                  (string-match "=ucs@" (symbol-name ccs))
 472                                  )
 473                              (setq ccs (charset-name ccs))
 474                              (null (assq ccs char-spec))
 475                              (setq ret (encode-char char ccs 'defined-only)))
 476                         (setq char-spec (cons (cons ccs ret) char-spec))))
 477                   )
 478                  ((encode-char char '==adobe-japan1 'defined-only)
 479                   (setq char-spec nil)
 480                   (dolist (ccs (charset-list))
 481                     (if (and (or (memq ccs
 482                                        '(==adobe-japan1-0
 483                                          ==adobe-japan1-1
 484                                          ==adobe-japan1-2
 485                                          ==adobe-japan1-3
 486                                          ==adobe-japan1-4
 487                                          ==adobe-japan1-5
 488                                          ==adobe-japan1-6
 489                                          ))
 490                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 491                                  (string-match "=ucs@" (symbol-name ccs))
 492                                  )
 493                              (setq ccs (charset-name ccs))
 494                              (null (assq ccs char-spec))
 495                              (setq ret (encode-char char ccs 'defined-only)))
 496                         (setq char-spec (cons (cons ccs ret) char-spec))))
 497                   )
 498                  ((setq ret
 499                         (catch 'tag
 500                           (let ((rest char-db-coded-charset-priority-list)
 501                                 ccs)
 502                             (while rest
 503                               (setq ccs (charset-name
 504                                          (find-charset (car rest))))
 505                               (if (setq ret
 506                                         (encode-char char ccs
 507                                                      'defined-only))
 508                                   (throw 'tag (cons ccs ret)))
 509                               (setq rest (cdr rest))))))
 510                   (setq char-spec (list ret))
 511                   (dolist (ccs (delq (car ret) (charset-list)))
 512                     (if (and (or (charset-iso-final-char ccs)
 513                                  (memq ccs
 514                                        '(=daikanwa
 515                                          =daikanwa@rev2
 516                                          ;; =gt-k
 517                                          =jis-x0208@1997
 518                                          ))
 519                                  (and (eq (charset-property ccs 'iso-ir) 177)
 520                                       (not (eq (charset-name ccs) '=ucs@big5)))
 521                                  ;; (string-match "=ucs@" (symbol-name ccs))
 522                                  )
 523                              (setq ccs (charset-name ccs))
 524                              (null (assq ccs char-spec))
 525                              (setq ret (encode-char char ccs 'defined-only)))
 526                         (setq char-spec (cons (cons ccs ret) char-spec))))
 527                   (if (null char-spec)
 528                       (setq char-spec (split-char char)))
 529                   (cond ((setq ret (get-char-attribute char 'name))
 530                          (setq char-spec (cons (cons 'name ret) char-spec))
 531                          )
 532                         ((setq ret (get-char-attribute char 'name*))
 533                          (setq char-spec (cons (cons 'name* ret) char-spec))
 534                          ))
 535                   )
 536                  ((setq ret (get-char-attribute
 537                              char 'ideographic-combination))
 538                   (setq char-spec
 539                         (cons (cons 'ideographic-combination ret)
 540                               char-spec))
 541                   ))
 542            char-spec)
 543           ((consp char)
 544            char))))
 545
 546 (defun char-db-insert-char-spec (char &optional readable column
 547                                       required-features)
 548   (unless column
 549     (setq column (current-column)))
 550   (let (char-spec temp-char)
 551     (setq char-spec (char-db-make-char-spec char))
 552     (unless (or (characterp char) ; char
 553                 (condition-case nil
 554                     (setq char (find-char char-spec))
 555                   (error nil)))
 556       ;; define temporary character
 557       ;;   Current implementation is dirty.
 558       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 559                                          char-spec)))
 560       (remove-char-attribute temp-char 'ideograph-daikanwa)
 561       (setq char temp-char))
 562     (insert-char-attributes char
 563                             readable
 564                             (union (mapcar #'car char-spec)
 565                                    required-features)
 566                             nil 'for-sub-node)
 567     (when temp-char
 568       ;; undefine temporary character
 569       ;;   Current implementation is dirty.
 570       (setq char-spec (char-attribute-alist temp-char))
 571       (while char-spec
 572         (remove-char-attribute temp-char (car (car char-spec)))
 573         (setq char-spec (cdr char-spec))))))
 574
 575 (defun char-db-insert-alist (alist &optional readable column)
 576   (unless column
 577     (setq column (current-column)))
 578   (let ((line-breaking
 579          (concat "\n" (make-string (1+ column) ?\ )))
 580         name value
 581         ret al ; cal
 582         key
 583         lbs cell rest separator)
 584     (insert "(")
 585     (while alist
 586       (setq name (car (car alist))
 587             value (cdr (car alist)))
 588       (cond ((eq name 'char)
 589              (insert "(char . ")
 590              (if (setq ret (condition-case nil
 591                                (find-char value)
 592                              (error nil)))
 593                  (progn
 594                    (setq al nil
 595                          ;; cal nil
 596                          )
 597                    (while value
 598                      (setq key (car (car value)))
 599                      ;; (if (find-charset key)
 600                      ;;     (setq cal (cons key cal))
 601                      (setq al (cons key al))
 602                      ;; )
 603                      (setq value (cdr value)))
 604                    (insert-char-attributes ret
 605                                            readable
 606                                            (or al 'none) ; cal
 607                                            nil 'for-sub-node))
 608                (insert (prin1-to-string value)))
 609              (insert ")")
 610              (insert line-breaking))
 611             ((consp value)
 612              (insert (format "(%-18s " name))
 613              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 614              (while (consp value)
 615                (setq cell (car value))
 616                (if (and (consp cell)
 617                         (consp (car cell))
 618                         (setq ret (condition-case nil
 619                                       (find-char cell)
 620                                     (error nil)))
 621                         )
 622                    (progn
 623                      (setq rest cell
 624                            al nil
 625                            ;; cal nil
 626                            )
 627                      (while rest
 628                        (setq key (car (car rest)))
 629                        ;; (if (find-charset key)
 630                        ;;     (setq cal (cons key cal))
 631                        (setq al (cons key al))
 632                        ;; )
 633                        (setq rest (cdr rest)))
 634                      (if separator
 635                          (insert lbs))
 636                      (insert-char-attributes ret
 637                                              readable
 638                                              al ; cal
 639                                              nil 'for-sub-node)
 640                      (setq separator lbs))
 641                  (if separator
 642                      (insert separator))
 643                  (insert (prin1-to-string cell))
 644                  (setq separator " "))
 645                (setq value (cdr value)))
 646              (insert ")")
 647              (insert line-breaking))
 648             (t
 649              (insert (format "(%-18s . %S)%s"
 650                              name value
 651                              line-breaking))))
 652       (setq alist (cdr alist))))
 653   (insert ")"))
 654
 655 (defun char-db-insert-char-reference (plist &optional readable column)
 656   (unless column
 657     (setq column (current-column)))
 658   (let ((line-breaking
 659          (concat "\n" (make-string (1+ column) ?\ )))
 660         (separator "")
 661         name value)
 662     (insert "(")
 663     (while plist
 664       (setq name (pop plist))
 665       (setq value (pop plist))
 666       (cond ((eq name :char)
 667              (insert separator)
 668              (insert ":char\t")
 669              (cond ((numberp value)
 670                     (setq value (decode-char '=ucs value)))
 671                    ;; ((consp value)
 672                    ;;  (setq value (or (find-char value)
 673                    ;;                  value)))
 674                    )
 675              (char-db-insert-char-spec value readable)
 676              (insert line-breaking)
 677              (setq separator ""))
 678             ((eq name :radical)
 679              (insert (format "%s%s\t%d ; %c%s"
 680                              separator
 681                              name value
 682                              (ideographic-radical value)
 683                              line-breaking))
 684              (setq separator ""))
 685             (t
 686              (insert (format "%s%s\t%S" separator name value))
 687              (setq separator line-breaking)))
 688       ))
 689   (insert ")"))
 690
 691 (defun char-db-decode-isolated-char (ccs code-point)
 692   (let (ret)
 693     (setq ret
 694           (cond ((eq ccs 'arabic-iso8859-6)
 695                  (decode-char ccs code-point))
 696                 ;; ((eq ccs '=gt)
 697                 ;;  (decode-builtin-char '==gt code-point))
 698                 ((and (memq ccs '(=gt-pj-1
 699                                   =gt-pj-2
 700                                   =gt-pj-3
 701                                   =gt-pj-4
 702                                   =gt-pj-5
 703                                   =gt-pj-6
 704                                   =gt-pj-7
 705                                   =gt-pj-8
 706                                   =gt-pj-9
 707                                   =gt-pj-10
 708                                   =gt-pj-11))
 709                       (setq ret (decode-char ccs code-point))
 710                       (setq ret (encode-char ret '=gt 'defined-only)))
 711                  (decode-builtin-char '=gt ret))
 712                 (t
 713                  (decode-builtin-char ccs code-point))))
 714     (cond ((null ret)
 715            (or (decode-char ccs code-point)
 716                (define-char (list (cons ccs code-point)))))
 717           ((and (<= 0 (char-int ret))
 718                 (<= (char-int ret) #x1F))
 719            (decode-char '=ucs (+ #x2400 (char-int ret))))
 720           ((= (char-int ret) #x7F)
 721            ?\u2421)
 722           (t ret))))
 723
 724 (defvar char-db-convert-obsolete-format t)
 725
 726 (defun char-db-insert-ccs-feature (name value line-breaking)
 727   (cond
 728    ((integerp value)
 729     (insert
 730      (format
 731       (cond
 732        ((memq name '(=>iwds-1
 733                      ==shinjigen
 734                      =shinjigen
 735                      =shinjigen@1ed ==shinjigen@1ed
 736                      =shinjigen@rev ==shinjigen@rev
 737                      =shinjigen/+p@rev ==shinjigen/+p@rev
 738                      ===daikanwa/ho ==daikanwa/ho
 739                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 740         "(%-18s .  %04d)\t; %c")
 741        ((eq name '=shinjigen@1ed/24pr)
 742         "(%-18s . %04d)\t; %c")
 743        ((or
 744          (memq name
 745                '(===daikanwa
 746                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 747                  =daikanwa@rev1 =daikanwa@rev2
 748                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 749                  =>>daikanwa/+p
 750                  =daikanwa/+2p =>>daikanwa/+2p
 751                  =gt ==gt ===gt
 752                  =>>gt =+>gt =>gt
 753                  =gt-k ==gt-k ===gt-k
 754                  =>>gt-k =>gt-k
 755                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 756                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 757                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 758                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 759                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 760                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 761                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 762                  =>>adobe-japan1-0 =+>adobe-japan1-0
 763                  =>>adobe-japan1-1 =+>adobe-japan1-1
 764                  =>>adobe-japan1-2 =+>adobe-japan1-2
 765                  =>>adobe-japan1-3 =+>adobe-japan1-3
 766                  =>>adobe-japan1-4 =+>adobe-japan1-4
 767                  =>>adobe-japan1-5 =+>adobe-japan1-5
 768                  =>>adobe-japan1-6 =+>adobe-japan1-6
 769                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 770                  =zinbun-oracle =>zinbun-oracle
 771                  =daijiten ==daijiten ===daijiten ==>daijiten
 772                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 773                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 774                  ===hng-khi ===hng-khm ===hng-fhs ===hng-hok
 775                  ===hng-kyd ===hng-sok
 776                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 777                  ===hng-sys ===hng-tsu ===hng-tzj
 778                  ===hng-hos ===hng-kkh ===hng-nak ===hng-jhk
 779                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 780                  ===hng-nkm ===hng-k24 ===hng-ini ===hng-nkk
 781                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 782                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 783                  ===hng-okd ===hng-wad ===hng-kmi
 784                  ===hng-zkd ===hng-doh ===hng-jyu ===hng-tzs
 785                  ===hng-sai ===hng-kad ===hng-kss
 786                  ===hng-kyo ===hng-ykk ===hng-saa
 787                  ===hng-sab ===hng-wks ===hng-wke ===hng-smk
 788                  ===hng-sgs ===hng-sts
 789                  =shuowen-jiguge ===shuowen-jiguge4 ===shuowen-jiguge5))
 790          ;; (string-match "^=adobe-" (symbol-name name))
 791          )
 792         "(%-18s . %05d)\t; %c")
 793        ((memq name '(=hanyo-denshi/ks
 794                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 795                      =>>hanyo-denshi/ks
 796                      =koseki ==koseki
 797                      =mj ==mj ===mj =>>mj =>mj
 798                      =zihai mojikyo))
 799         "(%-18s . %06d)\t; %c")
 800        ((memq name '(=hanyo-denshi/tk ==hanyo-denshi/tk))
 801         "(%-18s . %08d)\t; %c")
 802        ((>= (charset-dimension name) 2)
 803         "(%-18s . #x%04X)\t; %c")
 804        (t
 805         "(%-18s . #x%02X)\t; %c"))
 806       name
 807       (if (= (charset-iso-graphic-plane name) 1)
 808           (logior value
 809                   (cond ((= (charset-dimension name) 1)
 810                          #x80)
 811                         ((= (charset-dimension name) 2)
 812                          #x8080)
 813                         ((= (charset-dimension name) 3)
 814                          #x808080)
 815                         (t 0)))
 816         value)
 817       (char-db-decode-isolated-char name value)))
 818     (if (and (= (charset-chars name) 94)
 819              (= (charset-dimension name) 2))
 820         (insert (format " [%02d-%02d]"
 821                         (- (lsh value -8) 32)
 822                         (- (logand value 255) 32))))
 823     )
 824    (t
 825     (insert (format "(%-18s . %s)" name value))
 826     ))
 827   (insert line-breaking))
 828
 829 (defun char-db-insert-relation-feature (char name value line-breaking
 830                                              ccss readable)
 831   (insert (format "(%-18s%s " name line-breaking))
 832   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 833         separator cell sources required-features
 834         ret)
 835     (while (consp value)
 836       (setq cell (car value))
 837       (if (integerp cell)
 838           (setq cell (decode-char '=ucs cell)))
 839       (cond
 840        ((eq name '->subsumptive)
 841         (when (or (not (some (lambda (atr)
 842                                (get-char-attribute cell atr))
 843                              char-db-ignored-attributes))
 844                   (some (lambda (ccs)
 845                           (encode-char cell ccs 'defined-only))
 846                         ccss))
 847           (if separator
 848               (insert lbs))
 849           (let ((char-db-ignored-attributes
 850                  (cons '<-subsumptive
 851                        char-db-ignored-attributes)))
 852             (insert-char-attributes cell readable nil nil 'for-sub-node))
 853           (setq separator lbs))
 854         )
 855        ((characterp cell)
 856         (setq sources
 857               (get-char-attribute
 858                char (intern (format "%s*sources" name))))
 859         (setq required-features nil)
 860         (dolist (source sources)
 861           (cond
 862            ((memq source '(JP
 863                            JP/Jouyou
 864                            shinjigen shinjigen@1ed shinjigen@rev))
 865             (setq required-features
 866                   (union required-features
 867                          '(=jis-x0208
 868                            =jis-x0208@1990
 869                            =jis-x0213-1@2000
 870                            =jis-x0213-1@2004
 871                            =jis-x0213-2
 872                            =jis-x0212
 873                            =jis-x0208@1983
 874                            =jis-x0208@1978
 875                            =shinjigen
 876                            =shinjigen@1ed
 877                            =shinjigen@rev
 878                            =shinjigen/+p@rev))))
 879            ((eq source 'CN)
 880             (setq required-features
 881                   (union required-features
 882                          '(=gb2312
 883                            =gb12345
 884                            =iso-ir165)))))
 885           (cond
 886            ((find-charset (setq ret (intern (format "=%s" source))))
 887             (setq required-features
 888                   (cons ret required-features)))
 889            (t (setq required-features
 890                     (cons source required-features)))))
 891         (cond ((string-match "@JP" (symbol-name name))
 892                (setq required-features
 893                      (union required-features
 894                             '(=jis-x0208
 895                               =jis-x0208@1990
 896                               =jis-x0213-1-2000
 897                               =jis-x0213-2-2000
 898                               =jis-x0212
 899                               =jis-x0208@1983
 900                               =jis-x0208@1978))))
 901               ((string-match "@CN" (symbol-name name))
 902                (setq required-features
 903                      (union required-features
 904                             '(=gb2312
 905                               =gb12345
 906                               =iso-ir165)))))
 907         (if separator
 908             (insert lbs))
 909         (if readable
 910             (insert (format "%S" cell))
 911           (char-db-insert-char-spec cell readable
 912                                     nil
 913                                     required-features))
 914         (setq separator lbs))
 915        ((consp cell)
 916         (if separator
 917             (insert lbs))
 918         (if (consp (car cell))
 919             (char-db-insert-char-spec cell readable)
 920           (char-db-insert-char-reference cell readable))
 921         (setq separator lbs))
 922        (t
 923         (if separator
 924             (insert separator))
 925         (insert (prin1-to-string cell))
 926         (setq separator " ")))
 927       (setq value (cdr value)))
 928     (insert ")")
 929     (insert line-breaking)))
 930
 931 (defun insert-char-attributes (char &optional readable attributes column
 932                                     for-sub-node)
 933   (unless column
 934     (setq column (current-column)))
 935   (let (name value ; has-long-ccs-name
 936         rest
 937         radical strokes
 938         (line-breaking
 939          (concat "\n" (make-string (1+ column) ?\ )))
 940         lbs cell separator ret
 941         key al cal
 942         dest-ccss ; sources required-features
 943         ccss)
 944     (let (atr-d)
 945       (setq attributes
 946             (sort (if attributes
 947                       (if (consp attributes)
 948                           (progn
 949                             (dolist (name attributes)
 950                               (unless (memq name char-db-ignored-attributes)
 951                                 (if (find-charset name)
 952                                     (push name ccss))
 953                                 (push name atr-d)))
 954                             atr-d))
 955                     (dolist (name (char-attribute-list))
 956                       (unless (memq name char-db-ignored-attributes)
 957                         (if (find-charset name)
 958                             (push name ccss))
 959                         (push name atr-d)))
 960                     atr-d)
 961                   #'char-attribute-name<)))
 962     (insert "(")
 963     (when (memq '<-subsumptive attributes)
 964       (when (or readable (not for-sub-node))
 965         (when (setq value (get-char-attribute char '<-subsumptive))
 966           (char-db-insert-relation-feature char '<-subsumptive value
 967                                            line-breaking
 968                                            ccss readable)))
 969       (setq attributes (delq '<-subsumptive attributes)))
 970     (when (and (memq '<-denotational attributes)
 971                (setq value (get-char-attribute char '<-denotational)))
 972       (char-db-insert-relation-feature char '<-denotational value
 973                                        line-breaking
 974                                        ccss readable)
 975       (setq attributes (delq '<-denotational attributes)))
 976     (when (and (memq '<-denotational@component attributes)
 977                (setq value (get-char-attribute char '<-denotational@component)))
 978       (char-db-insert-relation-feature char '<-denotational@component value
 979                                        line-breaking
 980                                        ccss readable)
 981       (setq attributes (delq '<-denotational@component attributes)))
 982     (when (and (memq '<-denotational@usage attributes)
 983                (setq value (get-char-attribute char '<-denotational@usage)))
 984       (char-db-insert-relation-feature char '<-denotational@usage value
 985                                        line-breaking
 986                                        ccss readable)
 987       (setq attributes (delq '<-denotational@usage attributes)))
 988     (when (and (memq 'name attributes)
 989                (setq value (get-char-attribute char 'name)))
 990       (insert (format
 991                (if (> (+ (current-column) (length value)) 48)
 992                    "(name . %S)%s"
 993                  "(name               . %S)%s")
 994                value line-breaking))
 995       (setq attributes (delq 'name attributes))
 996       )
 997     (when (and (memq 'name* attributes)
 998                (setq value (get-char-attribute char 'name*)))
 999       (insert (format
1000                (if (> (+ (current-column) (length value)) 48)
1001                    "(name* . %S)%s"
1002                  "(name*              . %S)%s")
1003                value line-breaking))
1004       (setq attributes (delq 'name* attributes))
1005       )
1006     (when (and (memq 'script attributes)
1007                (setq value (get-char-attribute char 'script)))
1008       (insert (format "(script\t\t%s)%s"
1009                       (mapconcat (function prin1-to-string)
1010                                  value " ")
1011                       line-breaking))
1012       (setq attributes (delq 'script attributes))
1013       )
1014     (dolist (name '(=>ucs =>ucs*))
1015       (when (and (memq name attributes)
1016                  (setq value (get-char-attribute char name)))
1017         (insert (format "(%-18s . #x%04X)\t; %c%s"
1018                         name value (decode-char '=ucs value)
1019                         line-breaking))
1020         (setq attributes (delq name attributes))))
1021     (dolist (name '(=>ucs@gb =>ucs@big5))
1022       (when (and (memq name attributes)
1023                  (setq value (get-char-attribute char name)))
1024         (insert (format "(%-18s . #x%04X)\t; %c%s"
1025                         name value
1026                         (decode-char (intern
1027                                       (concat "="
1028                                               (substring
1029                                                (symbol-name name) 2)))
1030                                      value)
1031                         line-breaking))
1032         (setq attributes (delq name attributes))
1033         ))
1034     ;; (dolist (name '(=>daikanwa))
1035     ;;   (when (and (memq name attributes)
1036     ;;              (setq value (get-char-attribute char name)))
1037     ;;     (insert
1038     ;;      (if (integerp value)
1039     ;;          (format "(%-18s . %05d)\t; %c%s"
1040     ;;                  name value (decode-char '=daikanwa value)
1041     ;;                  line-breaking)
1042     ;;        (format "(%-18s %s)\t; %c%s"
1043     ;;                name
1044     ;;                (mapconcat (function prin1-to-string)
1045     ;;                           value " ")
1046     ;;                (char-representative-of-daikanwa char)
1047     ;;                line-breaking)))
1048     ;;     (setq attributes (delq name attributes))))
1049     (when (and (memq 'general-category attributes)
1050                (setq value (get-char-attribute char 'general-category)))
1051       (insert (format
1052                "(general-category\t%s) ; %s%s"
1053                (mapconcat (lambda (cell)
1054                             (format "%S" cell))
1055                           value " ")
1056                (cond ((rassoc value unidata-normative-category-alist)
1057                       "Normative Category")
1058                      ((rassoc value unidata-informative-category-alist)
1059                       "Informative Category")
1060                      (t
1061                       "Unknown Category"))
1062                line-breaking))
1063       (setq attributes (delq 'general-category attributes))
1064       )
1065     (when (and (memq 'bidi-category attributes)
1066                (setq value (get-char-attribute char 'bidi-category)))
1067       (insert (format "(bidi-category\t. %S)%s"
1068                       value
1069                       line-breaking))
1070       (setq attributes (delq 'bidi-category attributes))
1071       )
1072     (unless (or (not (memq 'mirrored attributes))
1073                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
1074                     'empty))
1075       (insert (format "(mirrored\t\t. %S)%s"
1076                       value
1077                       line-breaking))
1078       (setq attributes (delq 'mirrored attributes))
1079       )
1080     (cond
1081      ((and (memq 'decimal-digit-value attributes)
1082            (setq value (get-char-attribute char 'decimal-digit-value)))
1083       (insert (format "(decimal-digit-value . %S)%s"
1084                       value
1085                       line-breaking))
1086       (setq attributes (delq 'decimal-digit-value attributes))
1087       (when (and (memq 'digit-value attributes)
1088                  (setq value (get-char-attribute char 'digit-value)))
1089         (insert (format "(digit-value\t . %S)%s"
1090                         value
1091                         line-breaking))
1092         (setq attributes (delq 'digit-value attributes))
1093         )
1094       (when (and (memq 'numeric-value attributes)
1095                  (setq value (get-char-attribute char 'numeric-value)))
1096         (insert (format "(numeric-value\t . %S)%s"
1097                         value
1098                         line-breaking))
1099         (setq attributes (delq 'numeric-value attributes))
1100         )
1101       )
1102      (t
1103       (when (and (memq 'digit-value attributes)
1104                  (setq value (get-char-attribute char 'digit-value)))
1105         (insert (format "(digit-value\t. %S)%s"
1106                         value
1107                         line-breaking))
1108         (setq attributes (delq 'digit-value attributes))
1109         )
1110       (when (and (memq 'numeric-value attributes)
1111                  (setq value (get-char-attribute char 'numeric-value)))
1112         (insert (format "(numeric-value\t. %S)%s"
1113                         value
1114                         line-breaking))
1115         (setq attributes (delq 'numeric-value attributes))
1116         )))
1117     (when (and (memq 'iso-10646-comment attributes)
1118                (setq value (get-char-attribute char 'iso-10646-comment)))
1119       (insert (format "(iso-10646-comment\t. %S)%s"
1120                       value
1121                       line-breaking))
1122       (setq attributes (delq 'iso-10646-comment attributes))
1123       )
1124     (when (and (memq 'morohashi-daikanwa attributes)
1125                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1126       (insert (format "(morohashi-daikanwa\t%s)%s"
1127                       (mapconcat (function prin1-to-string) value " ")
1128                       line-breaking))
1129       (setq attributes (delq 'morohashi-daikanwa attributes))
1130       )
1131     (setq radical nil
1132           strokes nil)
1133     (when (and (memq 'ideographic-radical attributes)
1134                (setq value (get-char-attribute char 'ideographic-radical)))
1135       (setq radical value)
1136       (insert (format "(ideographic-radical . %S)\t; %c%s"
1137                       radical
1138                       (ideographic-radical radical)
1139                       line-breaking))
1140       (setq attributes (delq 'ideographic-radical attributes))
1141       )
1142     (when (and (memq 'shuowen-radical attributes)
1143                (setq value (get-char-attribute char 'shuowen-radical)))
1144       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1145                       value
1146                       (shuowen-radical value)
1147                       line-breaking))
1148       (setq attributes (delq 'shuowen-radical attributes))
1149       )
1150     (let (key)
1151       (dolist (domain
1152                (append
1153                 char-db-feature-domains
1154                 (let (dest domain)
1155                   (dolist (feature (char-attribute-list))
1156                     (setq feature (symbol-name feature))
1157                     (when (string-match
1158                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1159                            feature)
1160                       (setq domain (intern (match-string 2 feature)))
1161                      (unless (memq domain dest)
1162                        (setq dest (cons domain dest)))))
1163                   (sort dest #'string<))))
1164         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1165         (when (and (memq key attributes)
1166                    (setq value (get-char-attribute char key)))
1167           (setq radical value)
1168           (insert (format "(%s . %S)\t; %c%s"
1169                           key
1170                           radical
1171                           (ideographic-radical radical)
1172                           line-breaking))
1173           (setq attributes (delq key attributes))
1174           )
1175         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1176         (when (and (memq key attributes)
1177                    (setq value (get-char-attribute char key)))
1178           (setq strokes value)
1179           (insert (format "(%s . %S)%s"
1180                           key
1181                           strokes
1182                           line-breaking))
1183           (setq attributes (delq key attributes))
1184           )
1185         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1186         (when (and (memq key attributes)
1187                    (setq value (get-char-attribute char key)))
1188           (insert (format "(%s       . %S)%s"
1189                           key
1190                           value
1191                           line-breaking))
1192           (setq attributes (delq key attributes))
1193           )
1194         (dolist (feature '(ideographic-radical
1195                            ideographic-strokes
1196                            total-strokes))
1197           (setq key (intern (format "%s@%s*sources" feature domain)))
1198           (when (and (memq key attributes)
1199                      (setq value (get-char-attribute char key)))
1200             (insert (format "(%s%s" key line-breaking))
1201             (dolist (cell value)
1202               (insert (format " %s" cell)))
1203             (insert ")")
1204             (insert line-breaking)
1205             (setq attributes (delq key attributes))
1206             ))
1207         ))
1208     (when (and (memq 'ideographic-strokes attributes)
1209                (setq value (get-char-attribute char 'ideographic-strokes)))
1210       (setq strokes value)
1211       (insert (format "(ideographic-strokes . %S)%s"
1212                       strokes
1213                       line-breaking))
1214       (setq attributes (delq 'ideographic-strokes attributes))
1215       )
1216     (when (and (memq 'kangxi-radical attributes)
1217                (setq value (get-char-attribute char 'kangxi-radical)))
1218       (unless (eq value radical)
1219         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1220                         value
1221                         (ideographic-radical value)
1222                         line-breaking))
1223         (or radical
1224             (setq radical value)))
1225       (setq attributes (delq 'kangxi-radical attributes))
1226       )
1227     (when (and (memq 'kangxi-strokes attributes)
1228                (setq value (get-char-attribute char 'kangxi-strokes)))
1229       (unless (eq value strokes)
1230         (insert (format "(kangxi-strokes\t . %S)%s"
1231                         value
1232                         line-breaking))
1233         (or strokes
1234             (setq strokes value)))
1235       (setq attributes (delq 'kangxi-strokes attributes))
1236       )
1237     ;; (when (and (memq 'japanese-radical attributes)
1238     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1239     ;;   (unless (eq value radical)
1240     ;;     (insert (format "(japanese-radical\t . %S)\t; %c%s"
1241     ;;                     value
1242     ;;                     (ideographic-radical value)
1243     ;;                     line-breaking))
1244     ;;     (or radical
1245     ;;         (setq radical value)))
1246     ;;   (setq attributes (delq 'japanese-radical attributes))
1247     ;;   )
1248     (when (and (memq 'japanese-strokes attributes)
1249                (setq value (get-char-attribute char 'japanese-strokes)))
1250       (unless (eq value strokes)
1251         (insert (format "(japanese-strokes\t . %S)%s"
1252                         value
1253                         line-breaking))
1254         (or strokes
1255             (setq strokes value)))
1256       (setq attributes (delq 'japanese-strokes attributes))
1257       )
1258     (when (and (memq 'cns-radical attributes)
1259                (setq value (get-char-attribute char 'cns-radical)))
1260       (insert (format "(cns-radical\t . %S)\t; %c%s"
1261                       value
1262                       (ideographic-radical value)
1263                       line-breaking))
1264       (setq attributes (delq 'cns-radical attributes))
1265       )
1266     (when (and (memq 'cns-strokes attributes)
1267                (setq value (get-char-attribute char 'cns-strokes)))
1268       (unless (eq value strokes)
1269         (insert (format "(cns-strokes\t . %S)%s"
1270                         value
1271                         line-breaking))
1272         (or strokes
1273             (setq strokes value)))
1274       (setq attributes (delq 'cns-strokes attributes))
1275       )
1276     ;; (when (and (memq 'ideographic- attributes)
1277     ;;            (setq value (get-char-attribute char 'ideographic-)))
1278     ;;   (insert "(ideographic-       ")
1279     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1280     ;;         separator nil)
1281     ;;   (while (consp value)
1282     ;;     (setq cell (car value))
1283     ;;     (if (integerp cell)
1284     ;;         (setq cell (decode-char '=ucs cell)))
1285     ;;     (cond ((characterp cell)
1286     ;;            (if separator
1287     ;;                (insert lbs))
1288     ;;            (if readable
1289     ;;                (insert (format "%S" cell))
1290     ;;              (char-db-insert-char-spec cell readable))
1291     ;;            (setq separator lbs))
1292     ;;           ((consp cell)
1293     ;;            (if separator
1294     ;;                (insert lbs))
1295     ;;            (if (consp (car cell))
1296     ;;                (char-db-insert-char-spec cell readable)
1297     ;;              (char-db-insert-char-reference cell readable))
1298     ;;            (setq separator lbs))
1299     ;;           (t
1300     ;;            (if separator
1301     ;;                (insert separator))
1302     ;;            (insert (prin1-to-string cell))
1303     ;;            (setq separator " ")))
1304     ;;     (setq value (cdr value)))
1305     ;;   (insert ")")
1306     ;;   (insert line-breaking)
1307     ;;   (setq attributes (delq 'ideographic- attributes)))
1308     (when (and (memq 'total-strokes attributes)
1309                (setq value (get-char-attribute char 'total-strokes)))
1310       (insert (format "(total-strokes       . %S)%s"
1311                       value
1312                       line-breaking))
1313       (setq attributes (delq 'total-strokes attributes))
1314       )
1315     ;; (when (and (memq '->ideograph attributes)
1316     ;;            (setq value (get-char-attribute char '->ideograph)))
1317     ;;   (insert (format "(->ideograph\t%s)%s"
1318     ;;                   (mapconcat (lambda (code)
1319     ;;                                (cond ((symbolp code)
1320     ;;                                       (symbol-name code))
1321     ;;                                      ((integerp code)
1322     ;;                                       (format "#x%04X" code))
1323     ;;                                      (t
1324     ;;                                       (format "%s %S"
1325     ;;                                               line-breaking code))))
1326     ;;                              value " ")
1327     ;;                   line-breaking))
1328     ;;   (setq attributes (delq '->ideograph attributes))
1329     ;;   )
1330     ;; (when (and (memq '->decomposition attributes)
1331     ;;            (setq value (get-char-attribute char '->decomposition)))
1332     ;;   (insert (format "(->decomposition\t%s)%s"
1333     ;;                   (mapconcat (lambda (code)
1334     ;;                                (cond ((symbolp code)
1335     ;;                                       (symbol-name code))
1336     ;;                                      ((characterp code)
1337     ;;                                       (if readable
1338     ;;                                           (format "%S" code)
1339     ;;                                         (format "#x%04X"
1340     ;;                                                 (char-int code))
1341     ;;                                         ))
1342     ;;                                      ((integerp code)
1343     ;;                                       (format "#x%04X" code))
1344     ;;                                      (t
1345     ;;                                       (format "%s%S" line-breaking code))))
1346     ;;                              value " ")
1347     ;;                   line-breaking))
1348     ;;   (setq attributes (delq '->decomposition attributes))
1349     ;;   )
1350     (if (equal (get-char-attribute char '->titlecase)
1351                (get-char-attribute char '->uppercase))
1352         (setq attributes (delq '->titlecase attributes)))
1353     (when (and (memq '->mojikyo attributes)
1354                (setq value (get-char-attribute char '->mojikyo)))
1355       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1356                       value (decode-char 'mojikyo value)
1357                       line-breaking))
1358       (setq attributes (delq '->mojikyo attributes))
1359       )
1360     (unless readable
1361       (dolist (ignored '(composition
1362                          ->denotational <-subsumptive ->ucs-unified
1363                          ;; ->ideographic-component-forms
1364                          ))
1365         (setq attributes (delq ignored attributes))))
1366     (while attributes
1367       (setq name (car attributes))
1368       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1369                   'value-is-empty)
1370         (cond ((setq ret (find-charset name))
1371                (setq name (charset-name ret))
1372                (when (not (memq name dest-ccss))
1373                  (setq dest-ccss (cons name dest-ccss))
1374                  (char-db-insert-ccs-feature name value line-breaking))
1375                )
1376               ((string-match "^=>ucs@" (symbol-name name))
1377                (insert (format "(%-18s . #x%04X)\t; %c%s"
1378                                name value (decode-char '=ucs value)
1379                                line-breaking))
1380                )
1381               ((eq name 'jisx0208-1978/4X)
1382                (insert (format "(%-18s . #x%04X)%s"
1383                                name value
1384                                line-breaking))
1385                )
1386               ((and
1387                 (not readable)
1388                 (not (eq name '->subsumptive))
1389                 (not (eq name '->uppercase))
1390                 (not (eq name '->lowercase))
1391                 (not (eq name '->titlecase))
1392                 (not (eq name '->canonical))
1393                 (not (eq name '->Bopomofo))
1394                 (not (eq name '->mistakable))
1395                 (not (eq name '->ideographic-variants))
1396                 (null (get-char-attribute
1397                        char (intern (format "%s*sources" name))))
1398                 (not (string-match "\\*sources$" (symbol-name name)))
1399                 (null (get-char-attribute
1400                        char (intern (format "%s*note" name))))
1401                 (not (string-match "\\*note$" (symbol-name name)))
1402                 (or (eq name '<-identical)
1403                     (eq name '<-uppercase)
1404                     (eq name '<-lowercase)
1405                     (eq name '<-titlecase)
1406                     (eq name '<-canonical)
1407                     (eq name '<-ideographic-variants)
1408                     ;; (eq name '<-synonyms)
1409                     (string-match "^<-synonyms" (symbol-name name))
1410                     (eq name '<-mistakable)
1411                     (when (string-match "^->" (symbol-name name))
1412                       (cond
1413                        ((string-match "^->fullwidth" (symbol-name name))
1414                         (not (and (consp value)
1415                                   (characterp (car value))
1416                                   (encode-char
1417                                    (car value) '=ucs 'defined-only)))
1418                         )
1419                        (t)))
1420                     ))
1421                )
1422               ((or (eq name 'ideographic-structure)
1423                    (eq name 'ideographic-structure@apparent)
1424                    (eq name 'ideographic-combination)
1425                    ;; (eq name 'ideographic-)
1426                    (eq name '=decomposition)
1427                    (char-feature-base-name= '=decomposition name)
1428                    (char-feature-base-name= '=>decomposition name)
1429                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1430                    ;;               (symbol-name name))
1431                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1432                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1433                                  (symbol-name name))
1434                    )
1435                (char-db-insert-relation-feature char name value
1436                                                 line-breaking
1437                                                 ccss readable))
1438               ;; ((memq name '(ideograph=
1439               ;;               original-ideograph-of
1440               ;;               ancient-ideograph-of
1441               ;;               vulgar-ideograph-of
1442               ;;               wrong-ideograph-of
1443               ;;               ;; simplified-ideograph-of
1444               ;;               ideographic-variants
1445               ;;               ;; ideographic-different-form-of
1446               ;;               ))
1447               ;;  (insert (format "(%-18s%s " name line-breaking))
1448               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1449               ;;        separator nil)
1450               ;;  (while (consp value)
1451               ;;    (setq cell (car value))
1452               ;;    (if (and (consp cell)
1453               ;;             (consp (car cell)))
1454               ;;        (progn
1455               ;;          (if separator
1456               ;;              (insert lbs))
1457               ;;          (char-db-insert-alist cell readable)
1458               ;;          (setq separator lbs))
1459               ;;      (if separator
1460               ;;          (insert separator))
1461               ;;      (insert (prin1-to-string cell))
1462               ;;      (setq separator " "))
1463               ;;    (setq value (cdr value)))
1464               ;;  (insert ")")
1465               ;;  (insert line-breaking))
1466               ((consp value)
1467                (insert (format "(%-18s " name))
1468                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1469                      separator nil)
1470                (while (consp value)
1471                  (setq cell (car value))
1472                  (if (and (consp cell)
1473                           (consp (car cell))
1474                           (setq ret (condition-case nil
1475                                         (find-char cell)
1476                                       (error nil))))
1477                      (progn
1478                        (setq rest cell
1479                              al nil
1480                              cal nil)
1481                        (while rest
1482                          (setq key (car (car rest)))
1483                          (if (find-charset key)
1484                              (setq cal (cons key cal))
1485                            (setq al (cons key al)))
1486                          (setq rest (cdr rest)))
1487                        (if separator
1488                            (insert lbs))
1489                        (insert-char-attributes ret
1490                                                readable
1491                                                al ; cal
1492                                                nil 'for-sub-node)
1493                        (setq separator lbs))
1494                    (setq ret (prin1-to-string cell))
1495                    (if separator
1496                        (if (< (+ (current-column)
1497                                  (length ret)
1498                                  (length separator))
1499                               76)
1500                            (insert separator)
1501                          (insert lbs)))
1502                    (insert ret)
1503                    (setq separator " "))
1504                  (setq value (cdr value)))
1505                (insert ")")
1506                (insert line-breaking))
1507               (t
1508                (insert (format "(%-18s" name))
1509                (setq ret (prin1-to-string value))
1510                (unless (< (+ (current-column)
1511                              (length ret)
1512                              3)
1513                           76)
1514                  (insert line-breaking))
1515                (insert " . " ret ")" line-breaking)
1516                ;; (insert (format "(%-18s . %S)%s"
1517                ;;                 name value
1518                ;;                 line-breaking))
1519                )
1520               ))
1521       (setq attributes (cdr attributes)))
1522     (insert ")")))
1523
1524 (defun insert-char-data (char &optional readable
1525                               attributes)
1526   (save-restriction
1527     (narrow-to-region (point)(point))
1528     (insert "(define-char
1529   '")
1530     (insert-char-attributes char readable attributes)
1531     (insert ")\n")
1532     (goto-char (point-min))
1533     (while (re-search-forward "[ \t]+$" nil t)
1534       (replace-match ""))
1535     ;; from tabify.
1536     (goto-char (point-min))
1537     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1538       (let ((column (current-column))
1539             (indent-tabs-mode t))
1540         (delete-region (match-beginning 0) (point))
1541         (indent-to column)))
1542     (goto-char (point-max))
1543     ;; (tabify (point-min)(point-max))
1544     ))
1545
1546 (defun insert-char-data-with-variant (char &optional printable
1547                                            no-ucs-unified
1548                                            script excluded-script)
1549   (insert-char-data char printable)
1550   (let ((variants (char-variants char))
1551         rest
1552         variant vs ret)
1553     (setq variants (sort variants #'<))
1554     (setq rest variants)
1555     (setq variants (cons char variants))
1556     (while rest
1557       (setq variant (car rest))
1558       (unless (get-char-attribute variant '<-subsumptive)
1559         (if (and (or (null script)
1560                      (null (setq vs (get-char-attribute variant 'script)))
1561                      (memq script vs))
1562                  (or (null excluded-script)
1563                      (null (setq vs (get-char-attribute variant 'script)))
1564                      (not (memq excluded-script vs))))
1565             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1566               (insert-char-data variant printable)
1567               (if (setq ret (char-variants variant))
1568                   (while ret
1569                     (or (memq (car ret) variants)
1570                         ;; (get-char-attribute (car ret) '<-subsumptive)
1571                         (setq rest (nconc rest (list (car ret)))))
1572                     (setq ret (cdr ret)))))))
1573       (setq rest (cdr rest)))))
1574
1575 (defun insert-char-range-data (min max &optional script excluded-script)
1576   (let ((code min)
1577         char)
1578     (while (<= code max)
1579       (setq char (decode-char '=ucs code))
1580       (if (encode-char char '=ucs 'defined-only)
1581           (insert-char-data-with-variant char nil 'no-ucs-unified
1582                                          script excluded-script))
1583       (setq code (1+ code)))))
1584
1585 (defun write-char-range-data-to-file (min max file
1586                                           &optional script excluded-script)
1587   (let ((coding-system-for-write char-db-file-coding-system))
1588     (with-temp-buffer
1589       (insert (format ";; -*- coding: %s -*-\n"
1590                       char-db-file-coding-system))
1591       (insert-char-range-data min max script excluded-script)
1592       (write-region (point-min)(point-max) file))))
1593
1594 (defvar what-character-original-window-configuration)
1595
1596 ;;;###autoload
1597 (defun what-char-definition (char)
1598   (interactive (list (char-after)))
1599   (let ((buf (get-buffer-create "*Character Description*"))
1600         (the-buf (current-buffer))
1601         (win-conf (current-window-configuration)))
1602     (pop-to-buffer buf)
1603     (make-local-variable 'what-character-original-window-configuration)
1604     (setq what-character-original-window-configuration win-conf)
1605     (setq buffer-read-only nil)
1606     (erase-buffer)
1607     (condition-case err
1608         (progn
1609           (insert-char-data-with-variant char 'printable)
1610           (unless (char-attribute-alist char)
1611             (insert (format ";; = %c\n"
1612                             (let* ((rest (split-char char))
1613                                    (ccs (pop rest))
1614                                    (code (pop rest)))
1615                               (while rest
1616                                 (setq code (logior (lsh code 8)
1617                                                    (pop rest))))
1618                               (decode-char ccs code)))))
1619           ;; (char-db-update-comment)
1620           (set-buffer-modified-p nil)
1621           (view-mode the-buf (lambda (buf)
1622                                (set-window-configuration
1623                                 what-character-original-window-configuration)
1624                                ))
1625           (goto-char (point-min)))
1626       (error (progn
1627                (set-window-configuration
1628                 what-character-original-window-configuration)
1629                (signal (car err) (cdr err)))))))
1630
1631
1632 ;;; @ end
1633 ;;;
1634
1635 (provide 'char-db-util)
1636
1637 ;;; char-db-util.el ends here