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, 2021 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 ?𠬞 ?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 ?卣 ?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     =>ruimoku-v6
 346     =jef-china3
 347     =shinjigen
 348     =>cns11643-5
 349     =>cns11643-7
 350     =>gt
 351     =>gt-k
 352     =>daikanwa
 353     =>daikanwa/ho
 354     =>cbeta
 355     =ucs-var-001
 356     =ucs-var-002
 357     =ucs-var-003
 358     =ucs-var-004
 359     =ucs-var-005
 360     =ucs-var-006
 361     =ucs-var-007
 362     =ucs-var-008
 363     =ucs-var-010
 364     =ucs-itaiji-001
 365     =ucs-itaiji-002
 366     =ucs-itaiji-003
 367     =ucs-itaiji-004
 368     =ucs-itaiji-005
 369     =ucs-itaiji-006
 370     =ucs-itaiji-007
 371     =ucs-itaiji-008
 372     =ucs-itaiji-009
 373     =ucs-itaiji-010
 374     =ucs-itaiji-011
 375     =ucs-itaiji-012
 376     =+>ucs-var-003
 377     =>ucs-itaiji-001
 378     =>ucs-itaiji-002
 379     =>ucs-itaiji-003
 380     =>ucs-itaiji-004
 381     =>ucs-itaiji-005
 382     =>ucs-itaiji-006
 383     =>ucs-itaiji-007
 384     =>ucs-itaiji-008
 385     =>ucs-itaiji-009
 386     =>ucs-itaiji-010
 387     =>ucs-itaiji-011
 388     =big5-cdp-var-001
 389     =big5-cdp-var-002
 390     =big5-cdp-var-003
 391     =big5-cdp-var-004
 392     =big5-cdp-var-005
 393     =big5-cdp-var-007
 394     =big5-cdp-var-010
 395     =big5-cdp-itaiji-001
 396     =big5-cdp-itaiji-002
 397     =big5-cdp-itaiji-003
 398     =big5-cdp-itaiji-004
 399     =big5-cdp-itaiji-005
 400     =>big5-cdp-itaiji-001
 401     =>big5-cdp-itaiji-002
 402     =>big5-cdp-itaiji-003
 403     =>big5-cdp-itaiji-004
 404     =>big5-cdp-var-001
 405     =>ucs@iwds-1
 406     =>ucs@cognate
 407     =>ucs@component
 408     =>ucs@iwds-1/normalized
 409     =>ucs-itaiji-001@iwds-1
 410     =>ucs-itaiji-002@iwds-1
 411     =>ucs-itaiji-003@iwds-1
 412     =>ucs-itaiji-006@iwds-1
 413     =>big5-cdp@component
 414     =>big5-cdp@cognate
 415     ==ucs@gb
 416     =ucs@gb
 417     ==ucs-var-002
 418     =ucs@JP/hanazono
 419     =koseki
 420     =shuowen-jiguge
 421     ===shuowen-jiguge5
 422     ===shuowen-jiguge4))
 423
 424
 425 ;;; @ char-db formatters
 426 ;;;
 427
 428 (defun char-db-make-char-spec (char)
 429   (let (ret char-spec)
 430     (cond ((characterp char)
 431            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 432                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 433                   (setq char-spec (list (cons '=ucs ret)))
 434                   (cond ((setq ret (get-char-attribute char 'name))
 435                          (setq char-spec (cons (cons 'name ret) char-spec))
 436                          )
 437                         ((setq ret (get-char-attribute char 'name*))
 438                          (setq char-spec (cons (cons 'name* ret) char-spec))
 439                          ))
 440                   )
 441                  ((encode-char char '=mj 'defined-only)
 442                   (setq char-spec nil)
 443                   (dolist (ccs (charset-list))
 444                     (if (and (or (eq ccs '=mj)
 445                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 446                                  (string-match "=ucs@" (symbol-name ccs))
 447                                  )
 448                              (setq ccs (charset-name ccs))
 449                              (null (assq ccs char-spec))
 450                              (setq ret (encode-char char ccs 'defined-only)))
 451                         (setq char-spec (cons (cons ccs ret) char-spec))))
 452                   )
 453                  ((encode-char char '==mj 'defined-only)
 454                   (setq char-spec nil)
 455                   (dolist (ccs (charset-list))
 456                     (if (and (or (eq ccs '==mj)
 457                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 458                                  (string-match "=ucs@" (symbol-name ccs))
 459                                  )
 460                              (setq ccs (charset-name ccs))
 461                              (null (assq ccs char-spec))
 462                              (setq ret (encode-char char ccs 'defined-only)))
 463                         (setq char-spec (cons (cons ccs ret) char-spec))))
 464                   )
 465                  ((encode-char char '=adobe-japan1 'defined-only)
 466                   (setq char-spec nil)
 467                   (dolist (ccs (charset-list))
 468                     (if (and (or (memq ccs
 469                                        '(=adobe-japan1-0
 470                                          =adobe-japan1-1
 471                                          =adobe-japan1-2
 472                                          =adobe-japan1-3
 473                                          =adobe-japan1-4
 474                                          =adobe-japan1-5
 475                                          =adobe-japan1-6
 476                                          ))
 477                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 478                                  (string-match "=ucs@" (symbol-name ccs))
 479                                  )
 480                              (setq ccs (charset-name ccs))
 481                              (null (assq ccs char-spec))
 482                              (setq ret (encode-char char ccs 'defined-only)))
 483                         (setq char-spec (cons (cons ccs ret) char-spec))))
 484                   )
 485                  ((encode-char char '==adobe-japan1 'defined-only)
 486                   (setq char-spec nil)
 487                   (dolist (ccs (charset-list))
 488                     (if (and (or (memq ccs
 489                                        '(==adobe-japan1-0
 490                                          ==adobe-japan1-1
 491                                          ==adobe-japan1-2
 492                                          ==adobe-japan1-3
 493                                          ==adobe-japan1-4
 494                                          ==adobe-japan1-5
 495                                          ==adobe-japan1-6
 496                                          ))
 497                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 498                                  (string-match "=ucs@" (symbol-name ccs))
 499                                  )
 500                              (setq ccs (charset-name ccs))
 501                              (null (assq ccs char-spec))
 502                              (setq ret (encode-char char ccs 'defined-only)))
 503                         (setq char-spec (cons (cons ccs ret) char-spec))))
 504                   )
 505                  ((setq ret
 506                         (catch 'tag
 507                           (let ((rest char-db-coded-charset-priority-list)
 508                                 ccs)
 509                             (while rest
 510                               (setq ccs (charset-name
 511                                          (find-charset (car rest))))
 512                               (if (setq ret
 513                                         (encode-char char ccs
 514                                                      'defined-only))
 515                                   (throw 'tag (cons ccs ret)))
 516                               (setq rest (cdr rest))))))
 517                   (setq char-spec (list ret))
 518                   (dolist (ccs (delq (car ret) (charset-list)))
 519                     (if (and (or (charset-iso-final-char ccs)
 520                                  (memq ccs
 521                                        '(=daikanwa
 522                                          =daikanwa@rev2
 523                                          ;; =gt-k
 524                                          =jis-x0208@1997
 525                                          ))
 526                                  (and (eq (charset-property ccs 'iso-ir) 177)
 527                                       (not (eq (charset-name ccs) '=ucs@big5)))
 528                                  ;; (string-match "=ucs@" (symbol-name ccs))
 529                                  )
 530                              (setq ccs (charset-name ccs))
 531                              (null (assq ccs char-spec))
 532                              (setq ret (encode-char char ccs 'defined-only)))
 533                         (setq char-spec (cons (cons ccs ret) char-spec))))
 534                   (if (null char-spec)
 535                       (setq char-spec (split-char char)))
 536                   (cond ((setq ret (get-char-attribute char 'name))
 537                          (setq char-spec (cons (cons 'name ret) char-spec))
 538                          )
 539                         ((setq ret (get-char-attribute char 'name*))
 540                          (setq char-spec (cons (cons 'name* ret) char-spec))
 541                          ))
 542                   )
 543                  ((setq ret (get-char-attribute
 544                              char 'ideographic-combination))
 545                   (setq char-spec
 546                         (cons (cons 'ideographic-combination ret)
 547                               char-spec))
 548                   ))
 549            char-spec)
 550           ((consp char)
 551            char))))
 552
 553 (defun char-db-insert-char-spec (char &optional readable column
 554                                       required-features)
 555   (unless column
 556     (setq column (current-column)))
 557   (let (char-spec temp-char)
 558     (setq char-spec (char-db-make-char-spec char))
 559     (unless (or (characterp char) ; char
 560                 (condition-case nil
 561                     (setq char (find-char char-spec))
 562                   (error nil)))
 563       ;; define temporary character
 564       ;;   Current implementation is dirty.
 565       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 566                                          char-spec)))
 567       (remove-char-attribute temp-char 'ideograph-daikanwa)
 568       (setq char temp-char))
 569     (insert-char-attributes char
 570                             readable
 571                             (union (mapcar #'car char-spec)
 572                                    required-features)
 573                             nil 'for-sub-node)
 574     (when temp-char
 575       ;; undefine temporary character
 576       ;;   Current implementation is dirty.
 577       (setq char-spec (char-attribute-alist temp-char))
 578       (while char-spec
 579         (remove-char-attribute temp-char (car (car char-spec)))
 580         (setq char-spec (cdr char-spec))))))
 581
 582 (defun char-db-insert-alist (alist &optional readable column)
 583   (unless column
 584     (setq column (current-column)))
 585   (let ((line-breaking
 586          (concat "\n" (make-string (1+ column) ?\ )))
 587         name value
 588         ret al ; cal
 589         key
 590         lbs cell rest separator)
 591     (insert "(")
 592     (while alist
 593       (setq name (car (car alist))
 594             value (cdr (car alist)))
 595       (cond ((eq name 'char)
 596              (insert "(char . ")
 597              (if (setq ret (condition-case nil
 598                                (find-char value)
 599                              (error nil)))
 600                  (progn
 601                    (setq al nil
 602                          ;; cal nil
 603                          )
 604                    (while value
 605                      (setq key (car (car value)))
 606                      ;; (if (find-charset key)
 607                      ;;     (setq cal (cons key cal))
 608                      (setq al (cons key al))
 609                      ;; )
 610                      (setq value (cdr value)))
 611                    (insert-char-attributes ret
 612                                            readable
 613                                            (or al 'none) ; cal
 614                                            nil 'for-sub-node))
 615                (insert (prin1-to-string value)))
 616              (insert ")")
 617              (insert line-breaking))
 618             ((consp value)
 619              (insert (format "(%-18s " name))
 620              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 621              (while (consp value)
 622                (setq cell (car value))
 623                (if (and (consp cell)
 624                         (consp (car cell))
 625                         (setq ret (condition-case nil
 626                                       (find-char cell)
 627                                     (error nil)))
 628                         )
 629                    (progn
 630                      (setq rest cell
 631                            al nil
 632                            ;; cal nil
 633                            )
 634                      (while rest
 635                        (setq key (car (car rest)))
 636                        ;; (if (find-charset key)
 637                        ;;     (setq cal (cons key cal))
 638                        (setq al (cons key al))
 639                        ;; )
 640                        (setq rest (cdr rest)))
 641                      (if separator
 642                          (insert lbs))
 643                      (insert-char-attributes ret
 644                                              readable
 645                                              al ; cal
 646                                              nil 'for-sub-node)
 647                      (setq separator lbs))
 648                  (if separator
 649                      (insert separator))
 650                  (insert (prin1-to-string cell))
 651                  (setq separator " "))
 652                (setq value (cdr value)))
 653              (insert ")")
 654              (insert line-breaking))
 655             (t
 656              (insert (format "(%-18s . %S)%s"
 657                              name value
 658                              line-breaking))))
 659       (setq alist (cdr alist))))
 660   (insert ")"))
 661
 662 (defun char-db-insert-char-reference (plist &optional readable column)
 663   (unless column
 664     (setq column (current-column)))
 665   (let ((line-breaking
 666          (concat "\n" (make-string (1+ column) ?\ )))
 667         (separator "")
 668         name value)
 669     (insert "(")
 670     (while plist
 671       (setq name (pop plist))
 672       (setq value (pop plist))
 673       (cond ((eq name :char)
 674              (insert separator)
 675              (insert ":char\t")
 676              (cond ((numberp value)
 677                     (setq value (decode-char '=ucs value)))
 678                    ;; ((consp value)
 679                    ;;  (setq value (or (find-char value)
 680                    ;;                  value)))
 681                    )
 682              (char-db-insert-char-spec value readable)
 683              (insert line-breaking)
 684              (setq separator ""))
 685             ((eq name :radical)
 686              (insert (format "%s%s\t%d ; %c%s"
 687                              separator
 688                              name value
 689                              (ideographic-radical value)
 690                              line-breaking))
 691              (setq separator ""))
 692             (t
 693              (insert (format "%s%s\t%S" separator name value))
 694              (setq separator line-breaking)))
 695       ))
 696   (insert ")"))
 697
 698 (defun char-db-decode-isolated-char (ccs code-point)
 699   (let (ret)
 700     (setq ret
 701           (cond ((eq ccs 'arabic-iso8859-6)
 702                  (decode-char ccs code-point))
 703                 ;; ((eq ccs '=gt)
 704                 ;;  (decode-builtin-char '==gt code-point))
 705                 ((and (memq ccs '(=gt-pj-1
 706                                   =gt-pj-2
 707                                   =gt-pj-3
 708                                   =gt-pj-4
 709                                   =gt-pj-5
 710                                   =gt-pj-6
 711                                   =gt-pj-7
 712                                   =gt-pj-8
 713                                   =gt-pj-9
 714                                   =gt-pj-10
 715                                   =gt-pj-11))
 716                       (setq ret (decode-char ccs code-point))
 717                       (setq ret (encode-char ret '=gt 'defined-only)))
 718                  (decode-builtin-char '=gt ret))
 719                 (t
 720                  (decode-builtin-char ccs code-point))))
 721     (cond ((null ret)
 722            (or (decode-char ccs code-point)
 723                (define-char (list (cons ccs code-point)))))
 724           ((and (<= 0 (char-int ret))
 725                 (<= (char-int ret) #x1F))
 726            (decode-char '=ucs (+ #x2400 (char-int ret))))
 727           ((= (char-int ret) #x7F)
 728            ?\u2421)
 729           (t ret))))
 730
 731 (defvar char-db-convert-obsolete-format t)
 732
 733 (defun char-db-insert-ccs-feature (name value line-breaking)
 734   (cond
 735    ((integerp value)
 736     (insert
 737      (format
 738       (cond
 739        ((memq name '(=>iwds-1
 740                      ==shinjigen
 741                      =shinjigen
 742                      =shinjigen@1ed ==shinjigen@1ed
 743                      =shinjigen@rev ==shinjigen@rev
 744                      =shinjigen/+p@rev ==shinjigen/+p@rev
 745                      ===daikanwa/ho ==daikanwa/ho
 746                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 747         "(%-18s .  %04d)\t; %c")
 748        ((eq name '=shinjigen@1ed/24pr)
 749         "(%-18s . %04d)\t; %c")
 750        ((or
 751          (memq name
 752                '(===daikanwa
 753                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 754                  =daikanwa@rev1 =daikanwa@rev2
 755                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 756                  =>>daikanwa/+p
 757                  =daikanwa/+2p =>>daikanwa/+2p
 758                  =gt ==gt ===gt
 759                  =>>gt =+>gt =>gt
 760                  =gt-k ==gt-k ===gt-k
 761                  =>>gt-k =>gt-k
 762                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 763                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 764                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 765                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 766                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 767                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 768                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 769                  =>>adobe-japan1-0 =+>adobe-japan1-0
 770                  =>>adobe-japan1-1 =+>adobe-japan1-1
 771                  =>>adobe-japan1-2 =+>adobe-japan1-2
 772                  =>>adobe-japan1-3 =+>adobe-japan1-3
 773                  =>>adobe-japan1-4 =+>adobe-japan1-4
 774                  =>>adobe-japan1-5 =+>adobe-japan1-5
 775                  =>>adobe-japan1-6 =+>adobe-japan1-6
 776                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 777                  =zinbun-oracle =>zinbun-oracle
 778                  =daijiten ==daijiten ===daijiten ==>daijiten
 779                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 780                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 781                  ===hng-khi ===hng-khm ===hng-fhs ===hng-hok
 782                  ===hng-kyd ===hng-sok
 783                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 784                  ===hng-sys ===hng-tsu ===hng-tzj
 785                  ===hng-hos ===hng-kkh ===hng-nak ===hng-jhk
 786                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 787                  ===hng-nkm ===hng-k24 ===hng-ini ===hng-nkk
 788                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 789                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 790                  ===hng-okd ===hng-wad ===hng-kmi
 791                  ===hng-zkd ===hng-doh ===hng-jyu ===hng-tzs
 792                  ===hng-sai ===hng-kad ===hng-kss
 793                  ===hng-kyo ===hng-ykk ===hng-saa
 794                  ===hng-sab ===hng-wks ===hng-wke ===hng-smk
 795                  ===hng-sgs ===hng-sts ===hng-kkd
 796                  =shuowen-jiguge ===shuowen-jiguge4 ===shuowen-jiguge5))
 797          ;; (string-match "^=adobe-" (symbol-name name))
 798          )
 799         "(%-18s . %05d)\t; %c")
 800        ((memq name '(=hanyo-denshi/ks
 801                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 802                      =>>hanyo-denshi/ks
 803                      =koseki ==koseki
 804                      =mj ==mj ===mj =>>mj =>mj
 805                      =zihai mojikyo))
 806         "(%-18s . %06d)\t; %c")
 807        ((memq name '(=hanyo-denshi/tk ==hanyo-denshi/tk))
 808         "(%-18s . %08d)\t; %c")
 809        ((>= (charset-dimension name) 2)
 810         "(%-18s . #x%04X)\t; %c")
 811        (t
 812         "(%-18s . #x%02X)\t; %c"))
 813       name
 814       (if (= (charset-iso-graphic-plane name) 1)
 815           (logior value
 816                   (cond ((= (charset-dimension name) 1)
 817                          #x80)
 818                         ((= (charset-dimension name) 2)
 819                          #x8080)
 820                         ((= (charset-dimension name) 3)
 821                          #x808080)
 822                         (t 0)))
 823         value)
 824       (char-db-decode-isolated-char name value)))
 825     (if (and (= (charset-chars name) 94)
 826              (= (charset-dimension name) 2))
 827         (insert (format " [%02d-%02d]"
 828                         (- (lsh value -8) 32)
 829                         (- (logand value 255) 32))))
 830     )
 831    (t
 832     (insert (format "(%-18s . %s)" name value))
 833     ))
 834   (insert line-breaking))
 835
 836 (defun char-db-insert-relation-feature (char name value line-breaking
 837                                              ccss readable)
 838   (insert (format "(%-18s%s " name line-breaking))
 839   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 840         separator cell sources required-features
 841         ret)
 842     (while (consp value)
 843       (setq cell (car value))
 844       (if (integerp cell)
 845           (setq cell (decode-char '=ucs cell)))
 846       (cond
 847        ((eq name '->subsumptive)
 848         (when (or (not (some (lambda (atr)
 849                                (get-char-attribute cell atr))
 850                              char-db-ignored-attributes))
 851                   (some (lambda (ccs)
 852                           (encode-char cell ccs 'defined-only))
 853                         ccss))
 854           (if separator
 855               (insert lbs))
 856           (let ((char-db-ignored-attributes
 857                  (cons '<-subsumptive
 858                        char-db-ignored-attributes)))
 859             (insert-char-attributes cell readable nil nil 'for-sub-node))
 860           (setq separator lbs))
 861         )
 862        ((characterp cell)
 863         (setq sources
 864               (get-char-attribute
 865                char (intern (format "%s*sources" name))))
 866         (setq required-features nil)
 867         (dolist (source sources)
 868           (cond
 869            ((memq source '(JP
 870                            JP/Jouyou
 871                            shinjigen shinjigen@1ed shinjigen@rev))
 872             (setq required-features
 873                   (union required-features
 874                          '(=jis-x0208
 875                            =jis-x0208@1990
 876                            =jis-x0213-1@2000
 877                            =jis-x0213-1@2004
 878                            =jis-x0213-2
 879                            =jis-x0212
 880                            =jis-x0208@1983
 881                            =jis-x0208@1978
 882                            =shinjigen
 883                            =shinjigen@1ed
 884                            =shinjigen@rev
 885                            =shinjigen/+p@rev))))
 886            ((eq source 'CN)
 887             (setq required-features
 888                   (union required-features
 889                          '(=gb2312
 890                            =gb12345
 891                            =iso-ir165)))))
 892           (cond
 893            ((find-charset (setq ret (intern (format "=%s" source))))
 894             (setq required-features
 895                   (cons ret required-features)))
 896            (t (setq required-features
 897                     (cons source required-features)))))
 898         (cond ((string-match "@JP" (symbol-name name))
 899                (setq required-features
 900                      (union required-features
 901                             '(=jis-x0208
 902                               =jis-x0208@1990
 903                               =jis-x0213-1-2000
 904                               =jis-x0213-2-2000
 905                               =jis-x0212
 906                               =jis-x0208@1983
 907                               =jis-x0208@1978))))
 908               ((string-match "@CN" (symbol-name name))
 909                (setq required-features
 910                      (union required-features
 911                             '(=gb2312
 912                               =gb12345
 913                               =iso-ir165)))))
 914         (if separator
 915             (insert lbs))
 916         (if readable
 917             (insert (format "%S" cell))
 918           (char-db-insert-char-spec cell readable
 919                                     nil
 920                                     required-features))
 921         (setq separator lbs))
 922        ((consp cell)
 923         (if separator
 924             (insert lbs))
 925         (if (consp (car cell))
 926             (char-db-insert-char-spec cell readable)
 927           (char-db-insert-char-reference cell readable))
 928         (setq separator lbs))
 929        (t
 930         (if separator
 931             (insert separator))
 932         (insert (prin1-to-string cell))
 933         (setq separator " ")))
 934       (setq value (cdr value)))
 935     (insert ")")
 936     (insert line-breaking)))
 937
 938 (defun insert-char-attributes (char &optional readable attributes column
 939                                     for-sub-node)
 940   (unless column
 941     (setq column (current-column)))
 942   (let (name value ; has-long-ccs-name
 943         rest
 944         radical strokes
 945         (line-breaking
 946          (concat "\n" (make-string (1+ column) ?\ )))
 947         lbs cell separator ret
 948         key al cal
 949         dest-ccss ; sources required-features
 950         ccss)
 951     (let (atr-d)
 952       (setq attributes
 953             (sort (if attributes
 954                       (if (consp attributes)
 955                           (progn
 956                             (dolist (name attributes)
 957                               (unless (memq name char-db-ignored-attributes)
 958                                 (if (find-charset name)
 959                                     (push name ccss))
 960                                 (push name atr-d)))
 961                             atr-d))
 962                     (dolist (name (char-attribute-list))
 963                       (unless (memq name char-db-ignored-attributes)
 964                         (if (find-charset name)
 965                             (push name ccss))
 966                         (push name atr-d)))
 967                     atr-d)
 968                   #'char-attribute-name<)))
 969     (insert "(")
 970     (when (memq '<-subsumptive attributes)
 971       (when (or readable (not for-sub-node))
 972         (when (setq value (get-char-attribute char '<-subsumptive))
 973           (char-db-insert-relation-feature char '<-subsumptive value
 974                                            line-breaking
 975                                            ccss readable)))
 976       (setq attributes (delq '<-subsumptive attributes)))
 977     (when (and (memq '<-denotational attributes)
 978                (setq value (get-char-attribute char '<-denotational)))
 979       (char-db-insert-relation-feature char '<-denotational value
 980                                        line-breaking
 981                                        ccss readable)
 982       (setq attributes (delq '<-denotational attributes)))
 983     (when (and (memq '<-denotational@component attributes)
 984                (setq value (get-char-attribute char '<-denotational@component)))
 985       (char-db-insert-relation-feature char '<-denotational@component value
 986                                        line-breaking
 987                                        ccss readable)
 988       (setq attributes (delq '<-denotational@component attributes)))
 989     (when (and (memq '<-denotational@usage attributes)
 990                (setq value (get-char-attribute char '<-denotational@usage)))
 991       (char-db-insert-relation-feature char '<-denotational@usage value
 992                                        line-breaking
 993                                        ccss readable)
 994       (setq attributes (delq '<-denotational@usage attributes)))
 995     (when (and (memq 'name attributes)
 996                (setq value (get-char-attribute char 'name)))
 997       (insert (format
 998                (if (> (+ (current-column) (length value)) 48)
 999                    "(name . %S)%s"
1000                  "(name               . %S)%s")
1001                value line-breaking))
1002       (setq attributes (delq 'name attributes))
1003       )
1004     (when (and (memq 'name* attributes)
1005                (setq value (get-char-attribute char 'name*)))
1006       (insert (format
1007                (if (> (+ (current-column) (length value)) 48)
1008                    "(name* . %S)%s"
1009                  "(name*              . %S)%s")
1010                value line-breaking))
1011       (setq attributes (delq 'name* attributes))
1012       )
1013     (when (and (memq 'script attributes)
1014                (setq value (get-char-attribute char 'script)))
1015       (insert (format "(script\t\t%s)%s"
1016                       (mapconcat (function prin1-to-string)
1017                                  value " ")
1018                       line-breaking))
1019       (setq attributes (delq 'script attributes))
1020       )
1021     (dolist (name '(=>ucs =>ucs*))
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 (decode-char '=ucs value)
1026                         line-breaking))
1027         (setq attributes (delq name attributes))))
1028     (dolist (name '(=>ucs@gb =>ucs@big5))
1029       (when (and (memq name attributes)
1030                  (setq value (get-char-attribute char name)))
1031         (insert (format "(%-18s . #x%04X)\t; %c%s"
1032                         name value
1033                         (decode-char (intern
1034                                       (concat "="
1035                                               (substring
1036                                                (symbol-name name) 2)))
1037                                      value)
1038                         line-breaking))
1039         (setq attributes (delq name attributes))
1040         ))
1041     ;; (dolist (name '(=>daikanwa))
1042     ;;   (when (and (memq name attributes)
1043     ;;              (setq value (get-char-attribute char name)))
1044     ;;     (insert
1045     ;;      (if (integerp value)
1046     ;;          (format "(%-18s . %05d)\t; %c%s"
1047     ;;                  name value (decode-char '=daikanwa value)
1048     ;;                  line-breaking)
1049     ;;        (format "(%-18s %s)\t; %c%s"
1050     ;;                name
1051     ;;                (mapconcat (function prin1-to-string)
1052     ;;                           value " ")
1053     ;;                (char-representative-of-daikanwa char)
1054     ;;                line-breaking)))
1055     ;;     (setq attributes (delq name attributes))))
1056     (when (and (memq 'general-category attributes)
1057                (setq value (get-char-attribute char 'general-category)))
1058       (insert (format
1059                "(general-category\t%s) ; %s%s"
1060                (mapconcat (lambda (cell)
1061                             (format "%S" cell))
1062                           value " ")
1063                (cond ((rassoc value unidata-normative-category-alist)
1064                       "Normative Category")
1065                      ((rassoc value unidata-informative-category-alist)
1066                       "Informative Category")
1067                      (t
1068                       "Unknown Category"))
1069                line-breaking))
1070       (setq attributes (delq 'general-category attributes))
1071       )
1072     (when (and (memq 'bidi-category attributes)
1073                (setq value (get-char-attribute char 'bidi-category)))
1074       (insert (format "(bidi-category\t. %S)%s"
1075                       value
1076                       line-breaking))
1077       (setq attributes (delq 'bidi-category attributes))
1078       )
1079     (unless (or (not (memq 'mirrored attributes))
1080                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
1081                     'empty))
1082       (insert (format "(mirrored\t\t. %S)%s"
1083                       value
1084                       line-breaking))
1085       (setq attributes (delq 'mirrored attributes))
1086       )
1087     (cond
1088      ((and (memq 'decimal-digit-value attributes)
1089            (setq value (get-char-attribute char 'decimal-digit-value)))
1090       (insert (format "(decimal-digit-value . %S)%s"
1091                       value
1092                       line-breaking))
1093       (setq attributes (delq 'decimal-digit-value attributes))
1094       (when (and (memq 'digit-value attributes)
1095                  (setq value (get-char-attribute char 'digit-value)))
1096         (insert (format "(digit-value\t . %S)%s"
1097                         value
1098                         line-breaking))
1099         (setq attributes (delq 'digit-value attributes))
1100         )
1101       (when (and (memq 'numeric-value attributes)
1102                  (setq value (get-char-attribute char 'numeric-value)))
1103         (insert (format "(numeric-value\t . %S)%s"
1104                         value
1105                         line-breaking))
1106         (setq attributes (delq 'numeric-value attributes))
1107         )
1108       )
1109      (t
1110       (when (and (memq 'digit-value attributes)
1111                  (setq value (get-char-attribute char 'digit-value)))
1112         (insert (format "(digit-value\t. %S)%s"
1113                         value
1114                         line-breaking))
1115         (setq attributes (delq 'digit-value attributes))
1116         )
1117       (when (and (memq 'numeric-value attributes)
1118                  (setq value (get-char-attribute char 'numeric-value)))
1119         (insert (format "(numeric-value\t. %S)%s"
1120                         value
1121                         line-breaking))
1122         (setq attributes (delq 'numeric-value attributes))
1123         )))
1124     (when (and (memq 'iso-10646-comment attributes)
1125                (setq value (get-char-attribute char 'iso-10646-comment)))
1126       (insert (format "(iso-10646-comment\t. %S)%s"
1127                       value
1128                       line-breaking))
1129       (setq attributes (delq 'iso-10646-comment attributes))
1130       )
1131     (when (and (memq 'morohashi-daikanwa attributes)
1132                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1133       (insert (format "(morohashi-daikanwa\t%s)%s"
1134                       (mapconcat (function prin1-to-string) value " ")
1135                       line-breaking))
1136       (setq attributes (delq 'morohashi-daikanwa attributes))
1137       )
1138     (setq radical nil
1139           strokes nil)
1140     (when (and (memq 'ideographic-radical attributes)
1141                (setq value (get-char-attribute char 'ideographic-radical)))
1142       (setq radical value)
1143       (insert (format "(ideographic-radical . %S)\t; %c%s"
1144                       radical
1145                       (ideographic-radical radical)
1146                       line-breaking))
1147       (setq attributes (delq 'ideographic-radical attributes))
1148       )
1149     (when (and (memq 'shuowen-radical attributes)
1150                (setq value (get-char-attribute char 'shuowen-radical)))
1151       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1152                       value
1153                       (shuowen-radical value)
1154                       line-breaking))
1155       (setq attributes (delq 'shuowen-radical attributes))
1156       )
1157     (let (key)
1158       (dolist (domain
1159                (append
1160                 char-db-feature-domains
1161                 (let (dest domain)
1162                   (dolist (feature (char-attribute-list))
1163                     (setq feature (symbol-name feature))
1164                     (when (string-match
1165                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1166                            feature)
1167                       (setq domain (intern (match-string 2 feature)))
1168                      (unless (memq domain dest)
1169                        (setq dest (cons domain dest)))))
1170                   (sort dest #'string<))))
1171         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1172         (when (and (memq key attributes)
1173                    (setq value (get-char-attribute char key)))
1174           (setq radical value)
1175           (insert (format "(%s . %S)\t; %c%s"
1176                           key
1177                           radical
1178                           (ideographic-radical radical)
1179                           line-breaking))
1180           (setq attributes (delq key attributes))
1181           )
1182         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1183         (when (and (memq key attributes)
1184                    (setq value (get-char-attribute char key)))
1185           (setq strokes value)
1186           (insert (format "(%s . %S)%s"
1187                           key
1188                           strokes
1189                           line-breaking))
1190           (setq attributes (delq key attributes))
1191           )
1192         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1193         (when (and (memq key attributes)
1194                    (setq value (get-char-attribute char key)))
1195           (insert (format "(%s       . %S)%s"
1196                           key
1197                           value
1198                           line-breaking))
1199           (setq attributes (delq key attributes))
1200           )
1201         (dolist (feature '(ideographic-radical
1202                            ideographic-strokes
1203                            total-strokes))
1204           (setq key (intern (format "%s@%s*sources" feature domain)))
1205           (when (and (memq key attributes)
1206                      (setq value (get-char-attribute char key)))
1207             (insert (format "(%s%s" key line-breaking))
1208             (dolist (cell value)
1209               (insert (format " %s" cell)))
1210             (insert ")")
1211             (insert line-breaking)
1212             (setq attributes (delq key attributes))
1213             ))
1214         ))
1215     (when (and (memq 'ideographic-strokes attributes)
1216                (setq value (get-char-attribute char 'ideographic-strokes)))
1217       (setq strokes value)
1218       (insert (format "(ideographic-strokes . %S)%s"
1219                       strokes
1220                       line-breaking))
1221       (setq attributes (delq 'ideographic-strokes attributes))
1222       )
1223     (when (and (memq 'kangxi-radical attributes)
1224                (setq value (get-char-attribute char 'kangxi-radical)))
1225       (unless (eq value radical)
1226         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1227                         value
1228                         (ideographic-radical value)
1229                         line-breaking))
1230         (or radical
1231             (setq radical value)))
1232       (setq attributes (delq 'kangxi-radical attributes))
1233       )
1234     (when (and (memq 'kangxi-strokes attributes)
1235                (setq value (get-char-attribute char 'kangxi-strokes)))
1236       (unless (eq value strokes)
1237         (insert (format "(kangxi-strokes\t . %S)%s"
1238                         value
1239                         line-breaking))
1240         (or strokes
1241             (setq strokes value)))
1242       (setq attributes (delq 'kangxi-strokes attributes))
1243       )
1244     ;; (when (and (memq 'japanese-radical attributes)
1245     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1246     ;;   (unless (eq value radical)
1247     ;;     (insert (format "(japanese-radical\t . %S)\t; %c%s"
1248     ;;                     value
1249     ;;                     (ideographic-radical value)
1250     ;;                     line-breaking))
1251     ;;     (or radical
1252     ;;         (setq radical value)))
1253     ;;   (setq attributes (delq 'japanese-radical attributes))
1254     ;;   )
1255     (when (and (memq 'japanese-strokes attributes)
1256                (setq value (get-char-attribute char 'japanese-strokes)))
1257       (unless (eq value strokes)
1258         (insert (format "(japanese-strokes\t . %S)%s"
1259                         value
1260                         line-breaking))
1261         (or strokes
1262             (setq strokes value)))
1263       (setq attributes (delq 'japanese-strokes attributes))
1264       )
1265     (when (and (memq 'cns-radical attributes)
1266                (setq value (get-char-attribute char 'cns-radical)))
1267       (insert (format "(cns-radical\t . %S)\t; %c%s"
1268                       value
1269                       (ideographic-radical value)
1270                       line-breaking))
1271       (setq attributes (delq 'cns-radical attributes))
1272       )
1273     (when (and (memq 'cns-strokes attributes)
1274                (setq value (get-char-attribute char 'cns-strokes)))
1275       (unless (eq value strokes)
1276         (insert (format "(cns-strokes\t . %S)%s"
1277                         value
1278                         line-breaking))
1279         (or strokes
1280             (setq strokes value)))
1281       (setq attributes (delq 'cns-strokes attributes))
1282       )
1283     ;; (when (and (memq 'ideographic- attributes)
1284     ;;            (setq value (get-char-attribute char 'ideographic-)))
1285     ;;   (insert "(ideographic-       ")
1286     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1287     ;;         separator nil)
1288     ;;   (while (consp value)
1289     ;;     (setq cell (car value))
1290     ;;     (if (integerp cell)
1291     ;;         (setq cell (decode-char '=ucs cell)))
1292     ;;     (cond ((characterp cell)
1293     ;;            (if separator
1294     ;;                (insert lbs))
1295     ;;            (if readable
1296     ;;                (insert (format "%S" cell))
1297     ;;              (char-db-insert-char-spec cell readable))
1298     ;;            (setq separator lbs))
1299     ;;           ((consp cell)
1300     ;;            (if separator
1301     ;;                (insert lbs))
1302     ;;            (if (consp (car cell))
1303     ;;                (char-db-insert-char-spec cell readable)
1304     ;;              (char-db-insert-char-reference cell readable))
1305     ;;            (setq separator lbs))
1306     ;;           (t
1307     ;;            (if separator
1308     ;;                (insert separator))
1309     ;;            (insert (prin1-to-string cell))
1310     ;;            (setq separator " ")))
1311     ;;     (setq value (cdr value)))
1312     ;;   (insert ")")
1313     ;;   (insert line-breaking)
1314     ;;   (setq attributes (delq 'ideographic- attributes)))
1315     (when (and (memq 'total-strokes attributes)
1316                (setq value (get-char-attribute char 'total-strokes)))
1317       (insert (format "(total-strokes       . %S)%s"
1318                       value
1319                       line-breaking))
1320       (setq attributes (delq 'total-strokes attributes))
1321       )
1322     ;; (when (and (memq '->ideograph attributes)
1323     ;;            (setq value (get-char-attribute char '->ideograph)))
1324     ;;   (insert (format "(->ideograph\t%s)%s"
1325     ;;                   (mapconcat (lambda (code)
1326     ;;                                (cond ((symbolp code)
1327     ;;                                       (symbol-name code))
1328     ;;                                      ((integerp code)
1329     ;;                                       (format "#x%04X" code))
1330     ;;                                      (t
1331     ;;                                       (format "%s %S"
1332     ;;                                               line-breaking code))))
1333     ;;                              value " ")
1334     ;;                   line-breaking))
1335     ;;   (setq attributes (delq '->ideograph attributes))
1336     ;;   )
1337     ;; (when (and (memq '->decomposition attributes)
1338     ;;            (setq value (get-char-attribute char '->decomposition)))
1339     ;;   (insert (format "(->decomposition\t%s)%s"
1340     ;;                   (mapconcat (lambda (code)
1341     ;;                                (cond ((symbolp code)
1342     ;;                                       (symbol-name code))
1343     ;;                                      ((characterp code)
1344     ;;                                       (if readable
1345     ;;                                           (format "%S" code)
1346     ;;                                         (format "#x%04X"
1347     ;;                                                 (char-int code))
1348     ;;                                         ))
1349     ;;                                      ((integerp code)
1350     ;;                                       (format "#x%04X" code))
1351     ;;                                      (t
1352     ;;                                       (format "%s%S" line-breaking code))))
1353     ;;                              value " ")
1354     ;;                   line-breaking))
1355     ;;   (setq attributes (delq '->decomposition attributes))
1356     ;;   )
1357     (if (equal (get-char-attribute char '->titlecase)
1358                (get-char-attribute char '->uppercase))
1359         (setq attributes (delq '->titlecase attributes)))
1360     (when (and (memq '->mojikyo attributes)
1361                (setq value (get-char-attribute char '->mojikyo)))
1362       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1363                       value (decode-char 'mojikyo value)
1364                       line-breaking))
1365       (setq attributes (delq '->mojikyo attributes))
1366       )
1367     (unless readable
1368       (dolist (ignored '(composition
1369                          ->denotational <-subsumptive ->ucs-unified
1370                          ;; ->ideographic-component-forms
1371                          ))
1372         (setq attributes (delq ignored attributes))))
1373     (while attributes
1374       (setq name (car attributes))
1375       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1376                   'value-is-empty)
1377         (cond ((setq ret (find-charset name))
1378                (setq name (charset-name ret))
1379                (when (not (memq name dest-ccss))
1380                  (setq dest-ccss (cons name dest-ccss))
1381                  (char-db-insert-ccs-feature name value line-breaking))
1382                )
1383               ((string-match "^=>ucs@" (symbol-name name))
1384                (insert (format "(%-18s . #x%04X)\t; %c%s"
1385                                name value (decode-char '=ucs value)
1386                                line-breaking))
1387                )
1388               ((eq name 'jisx0208-1978/4X)
1389                (insert (format "(%-18s . #x%04X)%s"
1390                                name value
1391                                line-breaking))
1392                )
1393               ((and
1394                 (not readable)
1395                 (not (eq name '->subsumptive))
1396                 (not (eq name '->uppercase))
1397                 (not (eq name '->lowercase))
1398                 (not (eq name '->titlecase))
1399                 (not (eq name '->canonical))
1400                 (not (eq name '->Bopomofo))
1401                 (not (eq name '->mistakable))
1402                 (not (eq name '->ideographic-variants))
1403                 (null (get-char-attribute
1404                        char (intern (format "%s*sources" name))))
1405                 (not (string-match "\\*sources$" (symbol-name name)))
1406                 (null (get-char-attribute
1407                        char (intern (format "%s*note" name))))
1408                 (not (string-match "\\*note$" (symbol-name name)))
1409                 (or (eq name '<-identical)
1410                     (eq name '<-uppercase)
1411                     (eq name '<-lowercase)
1412                     (eq name '<-titlecase)
1413                     (eq name '<-canonical)
1414                     (eq name '<-ideographic-variants)
1415                     ;; (eq name '<-synonyms)
1416                     (string-match "^<-synonyms" (symbol-name name))
1417                     (eq name '<-mistakable)
1418                     (when (string-match "^->" (symbol-name name))
1419                       (cond
1420                        ((string-match "^->fullwidth" (symbol-name name))
1421                         (not (and (consp value)
1422                                   (characterp (car value))
1423                                   (encode-char
1424                                    (car value) '=ucs 'defined-only)))
1425                         )
1426                        (t)))
1427                     ))
1428                )
1429               ((or (eq name 'ideographic-structure)
1430                    (eq name 'ideographic-structure@apparent)
1431                    (eq name 'ideographic-structure@apparent/leftmost)
1432                    (eq name 'ideographic-combination)
1433                    ;; (eq name 'ideographic-)
1434                    (eq name '=decomposition)
1435                    (char-feature-base-name= '=decomposition name)
1436                    (char-feature-base-name= '=>decomposition name)
1437                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1438                    ;;               (symbol-name name))
1439                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1440                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1441                                  (symbol-name name))
1442                    )
1443                (char-db-insert-relation-feature char name value
1444                                                 line-breaking
1445                                                 ccss readable))
1446               ;; ((memq name '(ideograph=
1447               ;;               original-ideograph-of
1448               ;;               ancient-ideograph-of
1449               ;;               vulgar-ideograph-of
1450               ;;               wrong-ideograph-of
1451               ;;               ;; simplified-ideograph-of
1452               ;;               ideographic-variants
1453               ;;               ;; ideographic-different-form-of
1454               ;;               ))
1455               ;;  (insert (format "(%-18s%s " name line-breaking))
1456               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1457               ;;        separator nil)
1458               ;;  (while (consp value)
1459               ;;    (setq cell (car value))
1460               ;;    (if (and (consp cell)
1461               ;;             (consp (car cell)))
1462               ;;        (progn
1463               ;;          (if separator
1464               ;;              (insert lbs))
1465               ;;          (char-db-insert-alist cell readable)
1466               ;;          (setq separator lbs))
1467               ;;      (if separator
1468               ;;          (insert separator))
1469               ;;      (insert (prin1-to-string cell))
1470               ;;      (setq separator " "))
1471               ;;    (setq value (cdr value)))
1472               ;;  (insert ")")
1473               ;;  (insert line-breaking))
1474               ((consp value)
1475                (insert (format "(%-18s " name))
1476                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1477                      separator nil)
1478                (while (consp value)
1479                  (setq cell (car value))
1480                  (if (and (consp cell)
1481                           (consp (car cell))
1482                           (setq ret (condition-case nil
1483                                         (find-char cell)
1484                                       (error nil))))
1485                      (progn
1486                        (setq rest cell
1487                              al nil
1488                              cal nil)
1489                        (while rest
1490                          (setq key (car (car rest)))
1491                          (if (find-charset key)
1492                              (setq cal (cons key cal))
1493                            (setq al (cons key al)))
1494                          (setq rest (cdr rest)))
1495                        (if separator
1496                            (insert lbs))
1497                        (insert-char-attributes ret
1498                                                readable
1499                                                al ; cal
1500                                                nil 'for-sub-node)
1501                        (setq separator lbs))
1502                    (setq ret (prin1-to-string cell))
1503                    (if separator
1504                        (if (< (+ (current-column)
1505                                  (length ret)
1506                                  (length separator))
1507                               76)
1508                            (insert separator)
1509                          (insert lbs)))
1510                    (insert ret)
1511                    (setq separator " "))
1512                  (setq value (cdr value)))
1513                (insert ")")
1514                (insert line-breaking))
1515               (t
1516                (insert (format "(%-18s" name))
1517                (setq ret (prin1-to-string value))
1518                (unless (< (+ (current-column)
1519                              (length ret)
1520                              3)
1521                           76)
1522                  (insert line-breaking))
1523                (insert " . " ret ")" line-breaking)
1524                ;; (insert (format "(%-18s . %S)%s"
1525                ;;                 name value
1526                ;;                 line-breaking))
1527                )
1528               ))
1529       (setq attributes (cdr attributes)))
1530     (insert ")")))
1531
1532 (defun insert-char-data (char &optional readable
1533                               attributes)
1534   (save-restriction
1535     (narrow-to-region (point)(point))
1536     (insert "(define-char
1537   '")
1538     (insert-char-attributes char readable attributes)
1539     (insert ")\n")
1540     (goto-char (point-min))
1541     (while (re-search-forward "[ \t]+$" nil t)
1542       (replace-match ""))
1543     ;; from tabify.
1544     (goto-char (point-min))
1545     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1546       (let ((column (current-column))
1547             (indent-tabs-mode t))
1548         (delete-region (match-beginning 0) (point))
1549         (indent-to column)))
1550     (goto-char (point-max))
1551     ;; (tabify (point-min)(point-max))
1552     ))
1553
1554 (defun insert-char-data-with-variant (char &optional printable
1555                                            no-ucs-unified
1556                                            script excluded-script)
1557   (insert-char-data char printable)
1558   (let ((variants (char-variants char))
1559         rest
1560         variant vs ret)
1561     (setq variants (sort variants #'<))
1562     (setq rest variants)
1563     (setq variants (cons char variants))
1564     (while rest
1565       (setq variant (car rest))
1566       (unless (get-char-attribute variant '<-subsumptive)
1567         (if (and (or (null script)
1568                      (null (setq vs (get-char-attribute variant 'script)))
1569                      (memq script vs))
1570                  (or (null excluded-script)
1571                      (null (setq vs (get-char-attribute variant 'script)))
1572                      (not (memq excluded-script vs))))
1573             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1574               (insert-char-data variant printable)
1575               (if (setq ret (char-variants variant))
1576                   (while ret
1577                     (or (memq (car ret) variants)
1578                         ;; (get-char-attribute (car ret) '<-subsumptive)
1579                         (setq rest (nconc rest (list (car ret)))))
1580                     (setq ret (cdr ret)))))))
1581       (setq rest (cdr rest)))))
1582
1583 (defun insert-char-range-data (min max &optional script excluded-script)
1584   (let ((code min)
1585         char)
1586     (while (<= code max)
1587       (setq char (decode-char '=ucs code))
1588       (if (encode-char char '=ucs 'defined-only)
1589           (insert-char-data-with-variant char nil 'no-ucs-unified
1590                                          script excluded-script))
1591       (setq code (1+ code)))))
1592
1593 (defun write-char-range-data-to-file (min max file
1594                                           &optional script excluded-script)
1595   (let ((coding-system-for-write char-db-file-coding-system))
1596     (with-temp-buffer
1597       (insert (format ";; -*- coding: %s -*-\n"
1598                       char-db-file-coding-system))
1599       (insert-char-range-data min max script excluded-script)
1600       (write-region (point-min)(point-max) file))))
1601
1602 (defvar what-character-original-window-configuration)
1603
1604 ;;;###autoload
1605 (defun what-char-definition (char)
1606   (interactive (list (char-after)))
1607   (let ((buf (get-buffer-create "*Character Description*"))
1608         (the-buf (current-buffer))
1609         (win-conf (current-window-configuration)))
1610     (pop-to-buffer buf)
1611     (make-local-variable 'what-character-original-window-configuration)
1612     (setq what-character-original-window-configuration win-conf)
1613     (setq buffer-read-only nil)
1614     (erase-buffer)
1615     (condition-case err
1616         (progn
1617           (insert-char-data-with-variant char 'printable)
1618           (unless (char-attribute-alist char)
1619             (insert (format ";; = %c\n"
1620                             (let* ((rest (split-char char))
1621                                    (ccs (pop rest))
1622                                    (code (pop rest)))
1623                               (while rest
1624                                 (setq code (logior (lsh code 8)
1625                                                    (pop rest))))
1626                               (decode-char ccs code)))))
1627           ;; (char-db-update-comment)
1628           (set-buffer-modified-p nil)
1629           (view-mode the-buf (lambda (buf)
1630                                (set-window-configuration
1631                                 what-character-original-window-configuration)
1632                                ))
1633           (goto-char (point-min)))
1634       (error (progn
1635                (set-window-configuration
1636                 what-character-original-window-configuration)
1637                (signal (car err) (cdr err)))))))
1638
1639
1640 ;;; @ end
1641 ;;;
1642
1643 (provide 'char-db-util)
1644
1645 ;;; char-db-util.el ends here