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, 2022, 2023 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/ip
 186     =hanyo-denshi/jt
 187     =hanyo-denshi/ks
 188     =hanyo-denshi/tk
 189     ;; ==mj
 190     ;; ==adobe-japan1-0
 191     ;; ==adobe-japan1-1
 192     ;; ==adobe-japan1-2
 193     ;; ==adobe-japan1-3
 194     ;; ==adobe-japan1-4
 195     ;; ==adobe-japan1-5
 196     ;; ==adobe-japan1-6
 197     ==jis-x0208
 198     ==jis-x0213-1
 199     ==jis-x0213-2
 200     ==jis-x0212
 201     ==hanyo-denshi/ja
 202     ==hanyo-denshi/jb
 203     ==hanyo-denshi/jc
 204     ==hanyo-denshi/ft
 205     ==hanyo-denshi/ia
 206     ==hanyo-denshi/ib
 207     ==hanyo-denshi/hg
 208     ==hanyo-denshi/jt
 209     ==hanyo-denshi/ks
 210     ==hanyo-denshi/tk
 211     =gt
 212     =gt-k
 213     =daikanwa
 214     =daijiten
 215     =daikanwa@rev2
 216     =daikanwa@rev1
 217     =daikanwa/+p
 218     =daikanwa/ho
 219     ==daikanwa
 220     ==daijiten
 221     =cbeta
 222     ideograph-hanziku-1
 223     ideograph-hanziku-2
 224     ideograph-hanziku-3
 225     ideograph-hanziku-4
 226     ideograph-hanziku-5
 227     ideograph-hanziku-6
 228     ideograph-hanziku-7
 229     ideograph-hanziku-8
 230     ideograph-hanziku-9
 231     ideograph-hanziku-10
 232     ideograph-hanziku-11
 233     ideograph-hanziku-12
 234     ;; =>>>adobe-japan1-0
 235     ;; =>>>adobe-japan1-1
 236     ;; =>>>adobe-japan1-2
 237     ;; =>>>adobe-japan1-3
 238     ;; =>>>adobe-japan1-4
 239     ;; =>>>adobe-japan1-5
 240     ;; =>>>adobe-japan1-6
 241     ;; =>>>jis-x0208
 242     ;; =>>>jis-x0213-1
 243     ;; =>>>jis-x0213-2
 244     ;; =>>>hanyo-denshi/ja
 245     ;; =>>>hanyo-denshi/jb
 246     ;; =>>>hanyo-denshi/jc
 247     ;; =>>>hanyo-denshi/ft
 248     ;; =>>>hanyo-denshi/ib
 249     ;; =>>>hanyo-denshi/hg
 250     ;; =>>>hanyo-denshi/jt
 251     ;; =>>>hanyo-denshi/ks
 252     ;; =>>>gt
 253     =>>adobe-japan1-0
 254     =>>adobe-japan1-1
 255     =>>adobe-japan1-2
 256     =>>adobe-japan1-3
 257     =>>adobe-japan1-4
 258     =>>adobe-japan1-5
 259     =>>adobe-japan1-6
 260     =>>jis-x0208
 261     =>>jis-x0213-1
 262     =>>jis-x0213-1@2000
 263     =>>jis-x0213-1@2004
 264     =>>jis-x0213-2
 265     =>>jis-x0208@1978
 266     =>>hanyo-denshi/ft
 267     =>>hanyo-denshi/jt
 268     =>>hanyo-denshi/ks
 269     =>>gt
 270     =>>daikanwa
 271     =>>cbeta
 272     =+>jis-x0208
 273     =+>jis-x0213-1
 274     =+>jis-x0213-2
 275     =+>adobe-japan1-0
 276     =+>adobe-japan1-1
 277     =+>adobe-japan1-2
 278     =+>adobe-japan1-3
 279     =+>adobe-japan1-4
 280     =+>adobe-japan1-5
 281     =+>adobe-japan1-6
 282     =+>jis-x0208@1978
 283     =>jis-x0208
 284     =>jis-x0208@1997
 285     =>jis-x0213-1
 286     =>jis-x0213-1@2000
 287     =>jis-x0213-1@2004
 288     =>jis-x0213-2
 289     ==>ucs@bucs
 290     ==>daijiten
 291     ;; =>ucs@hanyo-denshi
 292     =>ucs@iso
 293     =>ucs@unicode
 294     =>ucs@jis
 295     =>ucs@cns
 296     =>ucs@ks
 297     =+>ucs@iso
 298     =+>ucs@unicode
 299     =+>ucs@jis
 300     =+>ucs@jis/1990
 301     =+>ucs@cns
 302     =+>ucs@ks
 303     =>>ucs@iso
 304     =>>ucs@unicode
 305     =>>ucs@jis
 306     =>>ucs@cns
 307     =>>>ucs@iso
 308     =>>>ucs@unicode
 309     ==ucs@iso
 310     ==ucs@unicode
 311     ;; ==ucs@cns
 312     ==gb2312
 313     ==ks-x1001
 314     ==cns11643-1
 315     ==cns11643-2
 316     ==cns11643-3
 317     ==cns11643-4
 318     ==cns11643-5
 319     ==cns11643-6
 320     ==cns11643-7
 321     ==gt
 322     ==jis-x0208@1990
 323     ;; ==jis-x0208@1983
 324     ==jis-x0208@1978
 325     ==gt-k
 326     =ucs@iso
 327     =ucs@unicode
 328     =ucs@cns
 329     ==big5-cdp
 330     ==cbeta
 331     =>>big5-cdp
 332     =>>gt-k
 333     =+>gt
 334     =+>big5-cdp
 335     =>big5-cdp
 336     =>big5-cdp@iwds-1
 337     =>mj
 338     =>hanyo-denshi/tk
 339     =>iwds-1
 340     =big5
 341     =big5-eten
 342     =zinbun-oracle
 343     =>zinbun-oracle
 344     =ruimoku-v6
 345     =>>ruimoku-v6
 346     ==ruimoku-v6
 347     =>ruimoku-v6
 348     =jef-china3
 349     =shinjigen
 350     =>cns11643-5
 351     =>cns11643-7
 352     =>gt
 353     =>gt-k
 354     =>daikanwa
 355     =>daikanwa/ho
 356     =>cbeta
 357     =ucs-var-001
 358     =ucs-var-002
 359     =ucs-var-003
 360     =ucs-var-004
 361     =ucs-var-005
 362     =ucs-var-006
 363     =ucs-var-007
 364     =ucs-var-008
 365     =ucs-var-010
 366     =ucs-itaiji-001
 367     =ucs-itaiji-002
 368     =ucs-itaiji-003
 369     =ucs-itaiji-004
 370     =ucs-itaiji-005
 371     =ucs-itaiji-006
 372     =ucs-itaiji-007
 373     =ucs-itaiji-008
 374     =ucs-itaiji-009
 375     =ucs-itaiji-010
 376     =ucs-itaiji-011
 377     =ucs-itaiji-012
 378     =ucs-itaiji-014
 379     =ucs-itaiji-016
 380     =+>ucs-var-003
 381     =>ucs-itaiji-001
 382     =>ucs-itaiji-002
 383     =>ucs-itaiji-003
 384     =>ucs-itaiji-004
 385     =>ucs-itaiji-005
 386     =>ucs-itaiji-006
 387     =>ucs-itaiji-007
 388     =>ucs-itaiji-008
 389     =>ucs-itaiji-009
 390     =>ucs-itaiji-010
 391     =>ucs-itaiji-011
 392     =>ucs-itaiji-012
 393     =>ucs-itaiji-014
 394     =>ucs-itaiji-016
 395     =big5-cdp-var-001
 396     =big5-cdp-var-002
 397     =big5-cdp-var-003
 398     =big5-cdp-var-004
 399     =big5-cdp-var-005
 400     =big5-cdp-var-007
 401     =big5-cdp-var-008
 402     =big5-cdp-var-010
 403     =big5-cdp-itaiji-001
 404     =big5-cdp-itaiji-002
 405     =big5-cdp-itaiji-003
 406     =big5-cdp-itaiji-004
 407     =big5-cdp-itaiji-005
 408     =>big5-cdp-itaiji-001
 409     =>big5-cdp-itaiji-002
 410     =>big5-cdp-itaiji-003
 411     =>big5-cdp-itaiji-004
 412     =>big5-cdp-itaiji-005
 413     =>big5-cdp-itaiji-006
 414     =>big5-cdp-var-001
 415     =>ucs@iwds-1
 416     =>ucs@cognate
 417     =>ucs@component
 418     =>ucs@iwds-1/normalized
 419     =>ucs-itaiji-001@iwds-1
 420     =>ucs-itaiji-002@iwds-1
 421     =>ucs-itaiji-003@iwds-1
 422     =>ucs-itaiji-006@iwds-1
 423     =>big5-cdp@component
 424     =>big5-cdp@cognate
 425     ==ucs@gb
 426     =ucs@gb
 427     ==ucs-var-002
 428     ==ucs-itaiji-002
 429     =ucs@JP/hanazono
 430     =koseki
 431     =shuowen-jiguge
 432     =>shuowen-jiguge
 433     ===shuowen-jiguge5
 434     ===shuowen-jiguge4
 435     ===chise-hdic-ktb-seal))
 436
 437
 438 ;;; @ char-db formatters
 439 ;;;
 440
 441 (defun char-db-make-char-spec (char)
 442   (let (ret char-spec)
 443     (cond ((characterp char)
 444            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 445                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 446                   (setq char-spec (list (cons '=ucs ret)))
 447                   (cond ((setq ret (get-char-attribute char 'name))
 448                          (setq char-spec (cons (cons 'name ret) char-spec))
 449                          )
 450                         ((setq ret (get-char-attribute char 'name*))
 451                          (setq char-spec (cons (cons 'name* ret) char-spec))
 452                          ))
 453                   )
 454                  ((encode-char char '=mj 'defined-only)
 455                   (setq char-spec nil)
 456                   (dolist (ccs (charset-list))
 457                     (if (and (or (eq ccs '=mj)
 458                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 459                                  (string-match "=ucs@" (symbol-name ccs))
 460                                  )
 461                              (setq ccs (charset-name ccs))
 462                              (null (assq ccs char-spec))
 463                              (setq ret (encode-char char ccs 'defined-only)))
 464                         (setq char-spec (cons (cons ccs ret) char-spec))))
 465                   )
 466                  ((encode-char char '==mj 'defined-only)
 467                   (setq char-spec nil)
 468                   (dolist (ccs (charset-list))
 469                     (if (and (or (eq ccs '==mj)
 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                  ((encode-char char '==adobe-japan1 'defined-only)
 499                   (setq char-spec nil)
 500                   (dolist (ccs (charset-list))
 501                     (if (and (or (memq ccs
 502                                        '(==adobe-japan1-0
 503                                          ==adobe-japan1-1
 504                                          ==adobe-japan1-2
 505                                          ==adobe-japan1-3
 506                                          ==adobe-japan1-4
 507                                          ==adobe-japan1-5
 508                                          ==adobe-japan1-6
 509                                          ))
 510                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 511                                  (string-match "=ucs@" (symbol-name ccs))
 512                                  )
 513                              (setq ccs (charset-name ccs))
 514                              (null (assq ccs char-spec))
 515                              (setq ret (encode-char char ccs 'defined-only)))
 516                         (setq char-spec (cons (cons ccs ret) char-spec))))
 517                   )
 518                  ((setq ret
 519                         (catch 'tag
 520                           (let ((rest char-db-coded-charset-priority-list)
 521                                 ccs)
 522                             (while rest
 523                               (setq ccs (charset-name
 524                                          (find-charset (car rest))))
 525                               (if (setq ret
 526                                         (encode-char char ccs
 527                                                      'defined-only))
 528                                   (throw 'tag (cons ccs ret)))
 529                               (setq rest (cdr rest))))))
 530                   (setq char-spec (list ret))
 531                   (dolist (ccs (delq (car ret) (charset-list)))
 532                     (if (and (or (charset-iso-final-char ccs)
 533                                  (memq ccs
 534                                        '(=daikanwa
 535                                          =daikanwa@rev2
 536                                          ;; =gt-k
 537                                          =jis-x0208@1997
 538                                          ))
 539                                  (and (eq (charset-property ccs 'iso-ir) 177)
 540                                       (not (eq (charset-name ccs) '=ucs@big5)))
 541                                  ;; (string-match "=ucs@" (symbol-name ccs))
 542                                  )
 543                              (setq ccs (charset-name ccs))
 544                              (null (assq ccs char-spec))
 545                              (setq ret (encode-char char ccs 'defined-only)))
 546                         (setq char-spec (cons (cons ccs ret) char-spec))))
 547                   (if (null char-spec)
 548                       (setq char-spec (split-char char)))
 549                   (cond ((setq ret (get-char-attribute char 'name))
 550                          (setq char-spec (cons (cons 'name ret) char-spec))
 551                          )
 552                         ((setq ret (get-char-attribute char 'name*))
 553                          (setq char-spec (cons (cons 'name* ret) char-spec))
 554                          ))
 555                   )
 556                  ((setq ret (get-char-attribute
 557                              char 'ideographic-combination))
 558                   (setq char-spec
 559                         (cons (cons 'ideographic-combination ret)
 560                               char-spec))
 561                   ))
 562            char-spec)
 563           ((consp char)
 564            char))))
 565
 566 (defun char-db-insert-char-spec (char &optional readable column
 567                                       required-features)
 568   (unless column
 569     (setq column (current-column)))
 570   (let (char-spec temp-char)
 571     (setq char-spec (char-db-make-char-spec char))
 572     (unless (or (characterp char) ; char
 573                 (condition-case nil
 574                     (setq char (find-char char-spec))
 575                   (error nil)))
 576       ;; define temporary character
 577       ;;   Current implementation is dirty.
 578       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 579                                          char-spec)))
 580       (remove-char-attribute temp-char 'ideograph-daikanwa)
 581       (setq char temp-char))
 582     (insert-char-attributes char
 583                             readable
 584                             (union (mapcar #'car char-spec)
 585                                    required-features)
 586                             nil 'for-sub-node)
 587     (when temp-char
 588       ;; undefine temporary character
 589       ;;   Current implementation is dirty.
 590       (setq char-spec (char-attribute-alist temp-char))
 591       (while char-spec
 592         (remove-char-attribute temp-char (car (car char-spec)))
 593         (setq char-spec (cdr char-spec))))))
 594
 595 (defun char-db-insert-alist (alist &optional readable column)
 596   (unless column
 597     (setq column (current-column)))
 598   (let ((line-breaking
 599          (concat "\n" (make-string (1+ column) ?\ )))
 600         name value
 601         ret al ; cal
 602         key
 603         lbs cell rest separator)
 604     (insert "(")
 605     (while alist
 606       (setq name (car (car alist))
 607             value (cdr (car alist)))
 608       (cond ((eq name 'char)
 609              (insert "(char . ")
 610              (if (setq ret (condition-case nil
 611                                (find-char value)
 612                              (error nil)))
 613                  (progn
 614                    (setq al nil
 615                          ;; cal nil
 616                          )
 617                    (while value
 618                      (setq key (car (car value)))
 619                      ;; (if (find-charset key)
 620                      ;;     (setq cal (cons key cal))
 621                      (setq al (cons key al))
 622                      ;; )
 623                      (setq value (cdr value)))
 624                    (insert-char-attributes ret
 625                                            readable
 626                                            (or al 'none) ; cal
 627                                            nil 'for-sub-node))
 628                (insert (prin1-to-string value)))
 629              (insert ")")
 630              (insert line-breaking))
 631             ((consp value)
 632              (insert (format "(%-18s " name))
 633              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 634              (while (consp value)
 635                (setq cell (car value))
 636                (if (and (consp cell)
 637                         (consp (car cell))
 638                         (setq ret (condition-case nil
 639                                       (find-char cell)
 640                                     (error nil)))
 641                         )
 642                    (progn
 643                      (setq rest cell
 644                            al nil
 645                            ;; cal nil
 646                            )
 647                      (while rest
 648                        (setq key (car (car rest)))
 649                        ;; (if (find-charset key)
 650                        ;;     (setq cal (cons key cal))
 651                        (setq al (cons key al))
 652                        ;; )
 653                        (setq rest (cdr rest)))
 654                      (if separator
 655                          (insert lbs))
 656                      (insert-char-attributes ret
 657                                              readable
 658                                              al ; cal
 659                                              nil 'for-sub-node)
 660                      (setq separator lbs))
 661                  (if separator
 662                      (insert separator))
 663                  (insert (prin1-to-string cell))
 664                  (setq separator " "))
 665                (setq value (cdr value)))
 666              (insert ")")
 667              (insert line-breaking))
 668             (t
 669              (insert (format "(%-18s . %S)%s"
 670                              name value
 671                              line-breaking))))
 672       (setq alist (cdr alist))))
 673   (insert ")"))
 674
 675 (defun char-db-insert-char-reference (plist &optional readable column)
 676   (unless column
 677     (setq column (current-column)))
 678   (let ((line-breaking
 679          (concat "\n" (make-string (1+ column) ?\ )))
 680         (separator "")
 681         name value)
 682     (insert "(")
 683     (while plist
 684       (setq name (pop plist))
 685       (setq value (pop plist))
 686       (cond ((eq name :char)
 687              (insert separator)
 688              (insert ":char\t")
 689              (cond ((numberp value)
 690                     (setq value (decode-char '=ucs value)))
 691                    ;; ((consp value)
 692                    ;;  (setq value (or (find-char value)
 693                    ;;                  value)))
 694                    )
 695              (char-db-insert-char-spec value readable)
 696              (insert line-breaking)
 697              (setq separator ""))
 698             ((eq name :radical)
 699              (insert (format "%s%s\t%d ; %c%s"
 700                              separator
 701                              name value
 702                              (ideographic-radical value)
 703                              line-breaking))
 704              (setq separator ""))
 705             (t
 706              (insert (format "%s%s\t%S" separator name value))
 707              (setq separator line-breaking)))
 708       ))
 709   (insert ")"))
 710
 711 (defun char-db-decode-isolated-char (ccs code-point)
 712   (let (ret)
 713     (setq ret
 714           (cond ((eq ccs 'arabic-iso8859-6)
 715                  (decode-char ccs code-point))
 716                 ;; ((eq ccs '=gt)
 717                 ;;  (decode-builtin-char '==gt code-point))
 718                 ((and (memq ccs '(=gt-pj-1
 719                                   =gt-pj-2
 720                                   =gt-pj-3
 721                                   =gt-pj-4
 722                                   =gt-pj-5
 723                                   =gt-pj-6
 724                                   =gt-pj-7
 725                                   =gt-pj-8
 726                                   =gt-pj-9
 727                                   =gt-pj-10
 728                                   =gt-pj-11))
 729                       (setq ret (decode-char ccs code-point))
 730                       (setq ret (encode-char ret '=gt 'defined-only)))
 731                  (decode-builtin-char '=gt ret))
 732                 (t
 733                  (decode-builtin-char ccs code-point))))
 734     (cond ((null ret)
 735            (or (decode-char ccs code-point)
 736                (define-char (list (cons ccs code-point)))))
 737           ((and (<= 0 (char-int ret))
 738                 (<= (char-int ret) #x1F))
 739            (decode-char '=ucs (+ #x2400 (char-int ret))))
 740           ((= (char-int ret) #x7F)
 741            ?\u2421)
 742           (t ret))))
 743
 744 (defvar char-db-convert-obsolete-format t)
 745
 746 (defun char-db-insert-ccs-feature (name value line-breaking)
 747   (cond
 748    ((integerp value)
 749     (insert
 750      (format
 751       (cond
 752        ((memq name '(=>iwds-1
 753                      ==shinjigen
 754                      =shinjigen
 755                      =shinjigen@1ed ==shinjigen@1ed
 756                      =shinjigen@rev ==shinjigen@rev
 757                      =shinjigen/+p@rev ==shinjigen/+p@rev
 758                      ===daikanwa/ho ==daikanwa/ho
 759                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 760         "(%-18s .  %04d)\t; %c")
 761        ((eq name '=shinjigen@1ed/24pr)
 762         "(%-18s . %04d)\t; %c")
 763        ((or
 764          (memq name
 765                '(===daikanwa
 766                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 767                  =daikanwa@rev1 =daikanwa@rev2
 768                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 769                  =>>daikanwa/+p
 770                  =daikanwa/+2p =>>daikanwa/+2p
 771                  =gt ==gt ===gt
 772                  =>>gt =+>gt =>gt
 773                  =gt-k ==gt-k ===gt-k
 774                  =>>gt-k =>gt-k
 775                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 776                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 777                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 778                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 779                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 780                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 781                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 782                  =>>adobe-japan1-0 =+>adobe-japan1-0
 783                  =>>adobe-japan1-1 =+>adobe-japan1-1
 784                  =>>adobe-japan1-2 =+>adobe-japan1-2
 785                  =>>adobe-japan1-3 =+>adobe-japan1-3
 786                  =>>adobe-japan1-4 =+>adobe-japan1-4
 787                  =>>adobe-japan1-5 =+>adobe-japan1-5
 788                  =>>adobe-japan1-6 =+>adobe-japan1-6
 789                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 790                  =zinbun-oracle =>zinbun-oracle
 791                  =daijiten ==daijiten ===daijiten ==>daijiten
 792                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 793                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 794                  ===hng-khi ===hng-khm ===hng-fhs ===hng-hok
 795                  ===hng-kyd ===hng-sok
 796                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 797                  ===hng-sys ===hng-tsu ===hng-tzj
 798                  ===hng-hos ===hng-kkh ===hng-nak ===hng-jhk
 799                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 800                  ===hng-nkm ===hng-k24 ===hng-ini ===hng-nkk
 801                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 802                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 803                  ===hng-okd ===hng-wad ===hng-kmi
 804                  ===hng-zkd ===hng-doh ===hng-jyu ===hng-tzs
 805                  ===hng-sai ===hng-kad ===hng-kss
 806                  ===hng-kyo ===hng-ykk ===hng-saa
 807                  ===hng-sab ===hng-wks ===hng-wke ===hng-smk
 808                  ===hng-sgs ===hng-sts ===hng-kkd
 809                  =>shuowen-jiguge =shuowen-jiguge
 810                  ===shuowen-jiguge4 ===shuowen-jiguge5 ===shuowen-jiguge-A30))
 811          ;; (string-match "^=adobe-" (symbol-name name))
 812          )
 813         "(%-18s . %05d)\t; %c")
 814        ((memq name '(=hanyo-denshi/ks
 815                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 816                      =>>hanyo-denshi/ks
 817                      =koseki ==koseki
 818                      =mj ==mj ===mj =>>mj =>mj
 819                      =zihai mojikyo))
 820         "(%-18s . %06d)\t; %c")
 821        ((memq name '(=hanyo-denshi/tk =>hanyo-denshi/tk ==hanyo-denshi/tk))
 822         "(%-18s . %08d)\t; %c")
 823        ((>= (charset-dimension name) 2)
 824         "(%-18s . #x%04X)\t; %c")
 825        (t
 826         "(%-18s . #x%02X)\t; %c"))
 827       name
 828       (if (= (charset-iso-graphic-plane name) 1)
 829           (logior value
 830                   (cond ((= (charset-dimension name) 1)
 831                          #x80)
 832                         ((= (charset-dimension name) 2)
 833                          #x8080)
 834                         ((= (charset-dimension name) 3)
 835                          #x808080)
 836                         (t 0)))
 837         value)
 838       (char-db-decode-isolated-char name value)))
 839     (if (and (= (charset-chars name) 94)
 840              (= (charset-dimension name) 2))
 841         (insert (format " [%02d-%02d]"
 842                         (- (lsh value -8) 32)
 843                         (- (logand value 255) 32))))
 844     )
 845    (t
 846     (insert (format "(%-18s . %s)" name value))
 847     ))
 848   (insert line-breaking))
 849
 850 (defun char-db-insert-relation-feature (char name value line-breaking
 851                                              ccss readable)
 852   (insert (format "(%-18s%s " name line-breaking))
 853   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 854         separator cell sources required-features
 855         ret)
 856     (while (consp value)
 857       (setq cell (car value))
 858       (if (integerp cell)
 859           (setq cell (decode-char '=ucs cell)))
 860       (cond
 861        ((eq name '->subsumptive)
 862         (when (or (not (some (lambda (atr)
 863                                (get-char-attribute cell atr))
 864                              char-db-ignored-attributes))
 865                   (some (lambda (ccs)
 866                           (encode-char cell ccs 'defined-only))
 867                         ccss))
 868           (if separator
 869               (insert lbs))
 870           (let ((char-db-ignored-attributes
 871                  (cons '<-subsumptive
 872                        char-db-ignored-attributes)))
 873             (insert-char-attributes cell readable nil nil 'for-sub-node))
 874           (setq separator lbs))
 875         )
 876        ((characterp cell)
 877         (setq sources
 878               (get-char-attribute
 879                char (intern (format "%s*sources" name))))
 880         (setq required-features nil)
 881         (dolist (source sources)
 882           (cond
 883            ((memq source '(JP
 884                            JP/Jouyou
 885                            shinjigen shinjigen@1ed shinjigen@rev))
 886             (setq required-features
 887                   (union required-features
 888                          '(=jis-x0208
 889                            =jis-x0208@1990
 890                            =jis-x0213-1@2000
 891                            =jis-x0213-1@2004
 892                            =jis-x0213-2
 893                            =jis-x0212
 894                            =jis-x0208@1983
 895                            =jis-x0208@1978
 896                            =shinjigen
 897                            =shinjigen@1ed
 898                            =shinjigen@rev
 899                            =shinjigen/+p@rev))))
 900            ((eq source 'CN)
 901             (setq required-features
 902                   (union required-features
 903                          '(=gb2312
 904                            =gb12345
 905                            =iso-ir165)))))
 906           (cond
 907            ((find-charset (setq ret (intern (format "=%s" source))))
 908             (setq required-features
 909                   (cons ret required-features)))
 910            (t (setq required-features
 911                     (cons source required-features)))))
 912         (cond ((string-match "@JP" (symbol-name name))
 913                (setq required-features
 914                      (union required-features
 915                             '(=jis-x0208
 916                               =jis-x0208@1990
 917                               =jis-x0213-1-2000
 918                               =jis-x0213-2-2000
 919                               =jis-x0212
 920                               =jis-x0208@1983
 921                               =jis-x0208@1978))))
 922               ((string-match "@CN" (symbol-name name))
 923                (setq required-features
 924                      (union required-features
 925                             '(=gb2312
 926                               =gb12345
 927                               =iso-ir165)))))
 928         (if separator
 929             (insert lbs))
 930         (if readable
 931             (insert (format "%S" cell))
 932           (char-db-insert-char-spec cell readable
 933                                     nil
 934                                     required-features))
 935         (setq separator lbs))
 936        ((consp cell)
 937         (if separator
 938             (insert lbs))
 939         (if (consp (car cell))
 940             (char-db-insert-char-spec cell readable)
 941           (char-db-insert-char-reference cell readable))
 942         (setq separator lbs))
 943        (t
 944         (if separator
 945             (insert separator))
 946         (insert (prin1-to-string cell))
 947         (setq separator " ")))
 948       (setq value (cdr value)))
 949     (insert ")")
 950     (insert line-breaking)))
 951
 952 (defun insert-char-attributes (char &optional readable attributes column
 953                                     for-sub-node)
 954   (unless column
 955     (setq column (current-column)))
 956   (let (name value ; has-long-ccs-name
 957         rest
 958         radical strokes
 959         (line-breaking
 960          (concat "\n" (make-string (1+ column) ?\ )))
 961         lbs cell separator ret
 962         key al cal
 963         dest-ccss ; sources required-features
 964         ccss)
 965     (let (atr-d)
 966       (setq attributes
 967             (sort (if attributes
 968                       (if (consp attributes)
 969                           (progn
 970                             (dolist (name attributes)
 971                               (unless (memq name char-db-ignored-attributes)
 972                                 (if (find-charset name)
 973                                     (push name ccss))
 974                                 (push name atr-d)))
 975                             atr-d))
 976                     (dolist (name (char-attribute-list))
 977                       (unless (memq name char-db-ignored-attributes)
 978                         (if (find-charset name)
 979                             (push name ccss))
 980                         (push name atr-d)))
 981                     atr-d)
 982                   #'char-attribute-name<)))
 983     (insert "(")
 984     (when (memq '<-subsumptive attributes)
 985       (when (or readable (not for-sub-node))
 986         (when (setq value (get-char-attribute char '<-subsumptive))
 987           (char-db-insert-relation-feature char '<-subsumptive value
 988                                            line-breaking
 989                                            ccss readable)))
 990       (setq attributes (delq '<-subsumptive attributes)))
 991     (when (and (memq '<-denotational attributes)
 992                (setq value (get-char-attribute char '<-denotational)))
 993       (char-db-insert-relation-feature char '<-denotational value
 994                                        line-breaking
 995                                        ccss readable)
 996       (setq attributes (delq '<-denotational attributes)))
 997     (when (and (memq '<-denotational@component attributes)
 998                (setq value (get-char-attribute char '<-denotational@component)))
 999       (char-db-insert-relation-feature char '<-denotational@component value
1000                                        line-breaking
1001                                        ccss readable)
1002       (setq attributes (delq '<-denotational@component attributes)))
1003     (when (and (memq '<-denotational@usage attributes)
1004                (setq value (get-char-attribute char '<-denotational@usage)))
1005       (char-db-insert-relation-feature char '<-denotational@usage value
1006                                        line-breaking
1007                                        ccss readable)
1008       (setq attributes (delq '<-denotational@usage attributes)))
1009     (when (and (memq 'name attributes)
1010                (setq value (get-char-attribute char 'name)))
1011       (insert (format
1012                (if (> (+ (current-column) (length value)) 48)
1013                    "(name . %S)%s"
1014                  "(name               . %S)%s")
1015                value line-breaking))
1016       (setq attributes (delq 'name attributes))
1017       )
1018     (when (and (memq 'name* attributes)
1019                (setq value (get-char-attribute char 'name*)))
1020       (insert (format
1021                (if (> (+ (current-column) (length value)) 48)
1022                    "(name* . %S)%s"
1023                  "(name*              . %S)%s")
1024                value line-breaking))
1025       (setq attributes (delq 'name* attributes))
1026       )
1027     (when (and (memq 'script attributes)
1028                (setq value (get-char-attribute char 'script)))
1029       (insert (format "(script\t\t%s)%s"
1030                       (mapconcat (function prin1-to-string)
1031                                  value " ")
1032                       line-breaking))
1033       (setq attributes (delq 'script attributes))
1034       )
1035     (dolist (name '(=>ucs =>ucs*))
1036       (when (and (memq name attributes)
1037                  (setq value (get-char-attribute char name)))
1038         (insert (format "(%-18s . #x%04X)\t; %c%s"
1039                         name value (decode-char '=ucs value)
1040                         line-breaking))
1041         (setq attributes (delq name attributes))))
1042     (dolist (name '(=>ucs@gb =>ucs@big5))
1043       (when (and (memq name attributes)
1044                  (setq value (get-char-attribute char name)))
1045         (insert (format "(%-18s . #x%04X)\t; %c%s"
1046                         name value
1047                         (decode-char (intern
1048                                       (concat "="
1049                                               (substring
1050                                                (symbol-name name) 2)))
1051                                      value)
1052                         line-breaking))
1053         (setq attributes (delq name attributes))
1054         ))
1055     ;; (dolist (name '(=>daikanwa))
1056     ;;   (when (and (memq name attributes)
1057     ;;              (setq value (get-char-attribute char name)))
1058     ;;     (insert
1059     ;;      (if (integerp value)
1060     ;;          (format "(%-18s . %05d)\t; %c%s"
1061     ;;                  name value (decode-char '=daikanwa value)
1062     ;;                  line-breaking)
1063     ;;        (format "(%-18s %s)\t; %c%s"
1064     ;;                name
1065     ;;                (mapconcat (function prin1-to-string)
1066     ;;                           value " ")
1067     ;;                (char-representative-of-daikanwa char)
1068     ;;                line-breaking)))
1069     ;;     (setq attributes (delq name attributes))))
1070     (when (and (memq 'general-category attributes)
1071                (setq value (get-char-attribute char 'general-category)))
1072       (insert (format
1073                "(general-category\t%s) ; %s%s"
1074                (mapconcat (lambda (cell)
1075                             (format "%S" cell))
1076                           value " ")
1077                (cond ((rassoc value unidata-normative-category-alist)
1078                       "Normative Category")
1079                      ((rassoc value unidata-informative-category-alist)
1080                       "Informative Category")
1081                      (t
1082                       "Unknown Category"))
1083                line-breaking))
1084       (setq attributes (delq 'general-category attributes))
1085       )
1086     (when (and (memq 'bidi-category attributes)
1087                (setq value (get-char-attribute char 'bidi-category)))
1088       (insert (format "(bidi-category\t. %S)%s"
1089                       value
1090                       line-breaking))
1091       (setq attributes (delq 'bidi-category attributes))
1092       )
1093     (unless (or (not (memq 'mirrored attributes))
1094                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
1095                     'empty))
1096       (insert (format "(mirrored\t\t. %S)%s"
1097                       value
1098                       line-breaking))
1099       (setq attributes (delq 'mirrored attributes))
1100       )
1101     (cond
1102      ((and (memq 'decimal-digit-value attributes)
1103            (setq value (get-char-attribute char 'decimal-digit-value)))
1104       (insert (format "(decimal-digit-value . %S)%s"
1105                       value
1106                       line-breaking))
1107       (setq attributes (delq 'decimal-digit-value attributes))
1108       (when (and (memq 'digit-value attributes)
1109                  (setq value (get-char-attribute char 'digit-value)))
1110         (insert (format "(digit-value\t . %S)%s"
1111                         value
1112                         line-breaking))
1113         (setq attributes (delq 'digit-value attributes))
1114         )
1115       (when (and (memq 'numeric-value attributes)
1116                  (setq value (get-char-attribute char 'numeric-value)))
1117         (insert (format "(numeric-value\t . %S)%s"
1118                         value
1119                         line-breaking))
1120         (setq attributes (delq 'numeric-value attributes))
1121         )
1122       )
1123      (t
1124       (when (and (memq 'digit-value attributes)
1125                  (setq value (get-char-attribute char 'digit-value)))
1126         (insert (format "(digit-value\t. %S)%s"
1127                         value
1128                         line-breaking))
1129         (setq attributes (delq 'digit-value attributes))
1130         )
1131       (when (and (memq 'numeric-value attributes)
1132                  (setq value (get-char-attribute char 'numeric-value)))
1133         (insert (format "(numeric-value\t. %S)%s"
1134                         value
1135                         line-breaking))
1136         (setq attributes (delq 'numeric-value attributes))
1137         )))
1138     (when (and (memq 'iso-10646-comment attributes)
1139                (setq value (get-char-attribute char 'iso-10646-comment)))
1140       (insert (format "(iso-10646-comment\t. %S)%s"
1141                       value
1142                       line-breaking))
1143       (setq attributes (delq 'iso-10646-comment attributes))
1144       )
1145     (when (and (memq 'morohashi-daikanwa attributes)
1146                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1147       (insert (format "(morohashi-daikanwa\t%s)%s"
1148                       (mapconcat (function prin1-to-string) value " ")
1149                       line-breaking))
1150       (setq attributes (delq 'morohashi-daikanwa attributes))
1151       )
1152     (setq radical nil
1153           strokes nil)
1154     (when (and (memq 'ideographic-radical attributes)
1155                (setq value (get-char-attribute char 'ideographic-radical)))
1156       (setq radical value)
1157       (insert (format "(ideographic-radical . %S)\t; %c%s"
1158                       radical
1159                       (ideographic-radical radical)
1160                       line-breaking))
1161       (setq attributes (delq 'ideographic-radical attributes))
1162       )
1163     (when (and (memq 'shuowen-radical attributes)
1164                (setq value (get-char-attribute char 'shuowen-radical)))
1165       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1166                       value
1167                       (shuowen-radical value)
1168                       line-breaking))
1169       (setq attributes (delq 'shuowen-radical attributes))
1170       )
1171     (let (key)
1172       (dolist (domain
1173                (append
1174                 char-db-feature-domains
1175                 (let (dest domain)
1176                   (dolist (feature (char-attribute-list))
1177                     (setq feature (symbol-name feature))
1178                     (when (string-match
1179                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1180                            feature)
1181                       (setq domain (intern (match-string 2 feature)))
1182                      (unless (memq domain dest)
1183                        (setq dest (cons domain dest)))))
1184                   (sort dest #'string<))))
1185         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1186         (when (and (memq key attributes)
1187                    (setq value (get-char-attribute char key)))
1188           (setq radical value)
1189           (insert (format "(%s . %S)\t; %c%s"
1190                           key
1191                           radical
1192                           (ideographic-radical radical)
1193                           line-breaking))
1194           (setq attributes (delq key attributes))
1195           )
1196         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1197         (when (and (memq key attributes)
1198                    (setq value (get-char-attribute char key)))
1199           (setq strokes value)
1200           (insert (format "(%s . %S)%s"
1201                           key
1202                           strokes
1203                           line-breaking))
1204           (setq attributes (delq key attributes))
1205           )
1206         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1207         (when (and (memq key attributes)
1208                    (setq value (get-char-attribute char key)))
1209           (insert (format "(%s       . %S)%s"
1210                           key
1211                           value
1212                           line-breaking))
1213           (setq attributes (delq key attributes))
1214           )
1215         (dolist (feature '(ideographic-radical
1216                            ideographic-strokes
1217                            total-strokes))
1218           (setq key (intern (format "%s@%s*sources" feature domain)))
1219           (when (and (memq key attributes)
1220                      (setq value (get-char-attribute char key)))
1221             (insert (format "(%s%s" key line-breaking))
1222             (dolist (cell value)
1223               (insert (format " %s" cell)))
1224             (insert ")")
1225             (insert line-breaking)
1226             (setq attributes (delq key attributes))
1227             ))
1228         ))
1229     (when (and (memq 'ideographic-strokes attributes)
1230                (setq value (get-char-attribute char 'ideographic-strokes)))
1231       (setq strokes value)
1232       (insert (format "(ideographic-strokes . %S)%s"
1233                       strokes
1234                       line-breaking))
1235       (setq attributes (delq 'ideographic-strokes attributes))
1236       )
1237     (when (and (memq 'kangxi-radical attributes)
1238                (setq value (get-char-attribute char 'kangxi-radical)))
1239       (unless (eq value radical)
1240         (insert (format "(kangxi-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 'kangxi-radical attributes))
1247       )
1248     (when (and (memq 'kangxi-strokes attributes)
1249                (setq value (get-char-attribute char 'kangxi-strokes)))
1250       (unless (eq value strokes)
1251         (insert (format "(kangxi-strokes\t . %S)%s"
1252                         value
1253                         line-breaking))
1254         (or strokes
1255             (setq strokes value)))
1256       (setq attributes (delq 'kangxi-strokes attributes))
1257       )
1258     ;; (when (and (memq 'japanese-radical attributes)
1259     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1260     ;;   (unless (eq value radical)
1261     ;;     (insert (format "(japanese-radical\t . %S)\t; %c%s"
1262     ;;                     value
1263     ;;                     (ideographic-radical value)
1264     ;;                     line-breaking))
1265     ;;     (or radical
1266     ;;         (setq radical value)))
1267     ;;   (setq attributes (delq 'japanese-radical attributes))
1268     ;;   )
1269     (when (and (memq 'japanese-strokes attributes)
1270                (setq value (get-char-attribute char 'japanese-strokes)))
1271       (unless (eq value strokes)
1272         (insert (format "(japanese-strokes\t . %S)%s"
1273                         value
1274                         line-breaking))
1275         (or strokes
1276             (setq strokes value)))
1277       (setq attributes (delq 'japanese-strokes attributes))
1278       )
1279     (when (and (memq 'cns-radical attributes)
1280                (setq value (get-char-attribute char 'cns-radical)))
1281       (insert (format "(cns-radical\t . %S)\t; %c%s"
1282                       value
1283                       (ideographic-radical value)
1284                       line-breaking))
1285       (setq attributes (delq 'cns-radical attributes))
1286       )
1287     (when (and (memq 'cns-strokes attributes)
1288                (setq value (get-char-attribute char 'cns-strokes)))
1289       (unless (eq value strokes)
1290         (insert (format "(cns-strokes\t . %S)%s"
1291                         value
1292                         line-breaking))
1293         (or strokes
1294             (setq strokes value)))
1295       (setq attributes (delq 'cns-strokes attributes))
1296       )
1297     ;; (when (and (memq 'ideographic- attributes)
1298     ;;            (setq value (get-char-attribute char 'ideographic-)))
1299     ;;   (insert "(ideographic-       ")
1300     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1301     ;;         separator nil)
1302     ;;   (while (consp value)
1303     ;;     (setq cell (car value))
1304     ;;     (if (integerp cell)
1305     ;;         (setq cell (decode-char '=ucs cell)))
1306     ;;     (cond ((characterp cell)
1307     ;;            (if separator
1308     ;;                (insert lbs))
1309     ;;            (if readable
1310     ;;                (insert (format "%S" cell))
1311     ;;              (char-db-insert-char-spec cell readable))
1312     ;;            (setq separator lbs))
1313     ;;           ((consp cell)
1314     ;;            (if separator
1315     ;;                (insert lbs))
1316     ;;            (if (consp (car cell))
1317     ;;                (char-db-insert-char-spec cell readable)
1318     ;;              (char-db-insert-char-reference cell readable))
1319     ;;            (setq separator lbs))
1320     ;;           (t
1321     ;;            (if separator
1322     ;;                (insert separator))
1323     ;;            (insert (prin1-to-string cell))
1324     ;;            (setq separator " ")))
1325     ;;     (setq value (cdr value)))
1326     ;;   (insert ")")
1327     ;;   (insert line-breaking)
1328     ;;   (setq attributes (delq 'ideographic- attributes)))
1329     (when (and (memq 'total-strokes attributes)
1330                (setq value (get-char-attribute char 'total-strokes)))
1331       (insert (format "(total-strokes       . %S)%s"
1332                       value
1333                       line-breaking))
1334       (setq attributes (delq 'total-strokes attributes))
1335       )
1336     ;; (when (and (memq '->ideograph attributes)
1337     ;;            (setq value (get-char-attribute char '->ideograph)))
1338     ;;   (insert (format "(->ideograph\t%s)%s"
1339     ;;                   (mapconcat (lambda (code)
1340     ;;                                (cond ((symbolp code)
1341     ;;                                       (symbol-name code))
1342     ;;                                      ((integerp code)
1343     ;;                                       (format "#x%04X" code))
1344     ;;                                      (t
1345     ;;                                       (format "%s %S"
1346     ;;                                               line-breaking code))))
1347     ;;                              value " ")
1348     ;;                   line-breaking))
1349     ;;   (setq attributes (delq '->ideograph attributes))
1350     ;;   )
1351     ;; (when (and (memq '->decomposition attributes)
1352     ;;            (setq value (get-char-attribute char '->decomposition)))
1353     ;;   (insert (format "(->decomposition\t%s)%s"
1354     ;;                   (mapconcat (lambda (code)
1355     ;;                                (cond ((symbolp code)
1356     ;;                                       (symbol-name code))
1357     ;;                                      ((characterp code)
1358     ;;                                       (if readable
1359     ;;                                           (format "%S" code)
1360     ;;                                         (format "#x%04X"
1361     ;;                                                 (char-int code))
1362     ;;                                         ))
1363     ;;                                      ((integerp code)
1364     ;;                                       (format "#x%04X" code))
1365     ;;                                      (t
1366     ;;                                       (format "%s%S" line-breaking code))))
1367     ;;                              value " ")
1368     ;;                   line-breaking))
1369     ;;   (setq attributes (delq '->decomposition attributes))
1370     ;;   )
1371     (if (equal (get-char-attribute char '->titlecase)
1372                (get-char-attribute char '->uppercase))
1373         (setq attributes (delq '->titlecase attributes)))
1374     (when (and (memq '->mojikyo attributes)
1375                (setq value (get-char-attribute char '->mojikyo)))
1376       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1377                       value (decode-char 'mojikyo value)
1378                       line-breaking))
1379       (setq attributes (delq '->mojikyo attributes))
1380       )
1381     (unless readable
1382       (dolist (ignored '(composition
1383                          ->denotational <-subsumptive ->ucs-unified
1384                          ;; ->ideographic-component-forms
1385                          ))
1386         (setq attributes (delq ignored attributes))))
1387     (while attributes
1388       (setq name (car attributes))
1389       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1390                   'value-is-empty)
1391         (cond ((setq ret (find-charset name))
1392                (setq name (charset-name ret))
1393                (when (not (memq name dest-ccss))
1394                  (setq dest-ccss (cons name dest-ccss))
1395                  (char-db-insert-ccs-feature name value line-breaking))
1396                )
1397               ((string-match "^=>ucs@" (symbol-name name))
1398                (insert (format "(%-18s . #x%04X)\t; %c%s"
1399                                name value (decode-char '=ucs value)
1400                                line-breaking))
1401                )
1402               ((eq name 'jisx0208-1978/4X)
1403                (insert (format "(%-18s . #x%04X)%s"
1404                                name value
1405                                line-breaking))
1406                )
1407               ((and
1408                 (not readable)
1409                 (not (eq name '->subsumptive))
1410                 (not (eq name '->uppercase))
1411                 (not (eq name '->lowercase))
1412                 (not (eq name '->titlecase))
1413                 (not (eq name '->canonical))
1414                 (not (eq name '->Bopomofo))
1415                 (not (eq name '->mistakable))
1416                 (not (eq name '->ideographic-variants))
1417                 (null (get-char-attribute
1418                        char (intern (format "%s*sources" name))))
1419                 (not (string-match "\\*sources$" (symbol-name name)))
1420                 (null (get-char-attribute
1421                        char (intern (format "%s*note" name))))
1422                 (not (string-match "\\*note$" (symbol-name name)))
1423                 (or (eq name '<-identical)
1424                     (eq name '<-uppercase)
1425                     (eq name '<-lowercase)
1426                     (eq name '<-titlecase)
1427                     (eq name '<-canonical)
1428                     (eq name '<-ideographic-variants)
1429                     ;; (eq name '<-synonyms)
1430                     (string-match "^<-synonyms" (symbol-name name))
1431                     (eq name '<-mistakable)
1432                     (when (string-match "^->" (symbol-name name))
1433                       (cond
1434                        ((string-match "^->fullwidth" (symbol-name name))
1435                         (not (and (consp value)
1436                                   (characterp (car value))
1437                                   (encode-char
1438                                    (car value) '=ucs 'defined-only)))
1439                         )
1440                        (t)))
1441                     ))
1442                )
1443               ((or (eq name 'ideographic-structure)
1444                    (eq name 'ideographic-structure@apparent)
1445                    (eq name 'ideographic-structure@apparent/leftmost)
1446                    (eq name 'ideographic-structure@apparent/rightmost)
1447                    (eq name 'ideographic-combination)
1448                    ;; (eq name 'ideographic-)
1449                    (eq name '=decomposition)
1450                    (char-feature-base-name= '=decomposition name)
1451                    (char-feature-base-name= '=>decomposition name)
1452                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1453                    ;;               (symbol-name name))
1454                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1455                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1456                                  (symbol-name name))
1457                    )
1458                (char-db-insert-relation-feature char name value
1459                                                 line-breaking
1460                                                 ccss readable))
1461               ;; ((memq name '(ideograph=
1462               ;;               original-ideograph-of
1463               ;;               ancient-ideograph-of
1464               ;;               vulgar-ideograph-of
1465               ;;               wrong-ideograph-of
1466               ;;               ;; simplified-ideograph-of
1467               ;;               ideographic-variants
1468               ;;               ;; ideographic-different-form-of
1469               ;;               ))
1470               ;;  (insert (format "(%-18s%s " name line-breaking))
1471               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1472               ;;        separator nil)
1473               ;;  (while (consp value)
1474               ;;    (setq cell (car value))
1475               ;;    (if (and (consp cell)
1476               ;;             (consp (car cell)))
1477               ;;        (progn
1478               ;;          (if separator
1479               ;;              (insert lbs))
1480               ;;          (char-db-insert-alist cell readable)
1481               ;;          (setq separator lbs))
1482               ;;      (if separator
1483               ;;          (insert separator))
1484               ;;      (insert (prin1-to-string cell))
1485               ;;      (setq separator " "))
1486               ;;    (setq value (cdr value)))
1487               ;;  (insert ")")
1488               ;;  (insert line-breaking))
1489               ((consp value)
1490                (insert (format "(%-18s " name))
1491                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1492                      separator nil)
1493                (while (consp value)
1494                  (setq cell (car value))
1495                  (if (and (consp cell)
1496                           (consp (car cell))
1497                           (setq ret (condition-case nil
1498                                         (find-char cell)
1499                                       (error nil))))
1500                      (progn
1501                        (setq rest cell
1502                              al nil
1503                              cal nil)
1504                        (while rest
1505                          (setq key (car (car rest)))
1506                          (if (find-charset key)
1507                              (setq cal (cons key cal))
1508                            (setq al (cons key al)))
1509                          (setq rest (cdr rest)))
1510                        (if separator
1511                            (insert lbs))
1512                        (insert-char-attributes ret
1513                                                readable
1514                                                al ; cal
1515                                                nil 'for-sub-node)
1516                        (setq separator lbs))
1517                    (setq ret (prin1-to-string cell))
1518                    (if separator
1519                        (if (< (+ (current-column)
1520                                  (length ret)
1521                                  (length separator))
1522                               76)
1523                            (insert separator)
1524                          (insert lbs)))
1525                    (insert ret)
1526                    (setq separator " "))
1527                  (setq value (cdr value)))
1528                (insert ")")
1529                (insert line-breaking))
1530               (t
1531                (insert (format "(%-18s" name))
1532                (setq ret (prin1-to-string value))
1533                (unless (< (+ (current-column)
1534                              (length ret)
1535                              3)
1536                           76)
1537                  (insert line-breaking))
1538                (insert " . " ret ")" line-breaking)
1539                ;; (insert (format "(%-18s . %S)%s"
1540                ;;                 name value
1541                ;;                 line-breaking))
1542                )
1543               ))
1544       (setq attributes (cdr attributes)))
1545     (insert ")")))
1546
1547 (defun insert-char-data (char &optional readable
1548                               attributes)
1549   (save-restriction
1550     (narrow-to-region (point)(point))
1551     (insert "(define-char
1552   '")
1553     (insert-char-attributes char readable attributes)
1554     (insert ")\n")
1555     (goto-char (point-min))
1556     (while (re-search-forward "[ \t]+$" nil t)
1557       (replace-match ""))
1558     ;; from tabify.
1559     (goto-char (point-min))
1560     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1561       (let ((column (current-column))
1562             (indent-tabs-mode t))
1563         (delete-region (match-beginning 0) (point))
1564         (indent-to column)))
1565     (goto-char (point-max))
1566     ;; (tabify (point-min)(point-max))
1567     ))
1568
1569 (defun insert-char-data-with-variant (char &optional printable
1570                                            no-ucs-unified
1571                                            script excluded-script)
1572   (insert-char-data char printable)
1573   (let ((variants (char-variants char))
1574         rest
1575         variant vs ret)
1576     (setq variants (sort variants #'<))
1577     (setq rest variants)
1578     (setq variants (cons char variants))
1579     (while rest
1580       (setq variant (car rest))
1581       (unless (get-char-attribute variant '<-subsumptive)
1582         (if (and (or (null script)
1583                      (null (setq vs (get-char-attribute variant 'script)))
1584                      (memq script vs))
1585                  (or (null excluded-script)
1586                      (null (setq vs (get-char-attribute variant 'script)))
1587                      (not (memq excluded-script vs))))
1588             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1589               (insert-char-data variant printable)
1590               (if (setq ret (char-variants variant))
1591                   (while ret
1592                     (or (memq (car ret) variants)
1593                         ;; (get-char-attribute (car ret) '<-subsumptive)
1594                         (setq rest (nconc rest (list (car ret)))))
1595                     (setq ret (cdr ret)))))))
1596       (setq rest (cdr rest)))))
1597
1598 (defun insert-char-range-data (min max &optional script excluded-script)
1599   (let ((code min)
1600         char)
1601     (while (<= code max)
1602       (setq char (decode-char '=ucs code))
1603       (if (encode-char char '=ucs 'defined-only)
1604           (insert-char-data-with-variant char nil 'no-ucs-unified
1605                                          script excluded-script))
1606       (setq code (1+ code)))))
1607
1608 (defun write-char-range-data-to-file (min max file
1609                                           &optional script excluded-script)
1610   (let ((coding-system-for-write char-db-file-coding-system))
1611     (with-temp-buffer
1612       (insert (format ";; -*- coding: %s -*-\n"
1613                       char-db-file-coding-system))
1614       (insert-char-range-data min max script excluded-script)
1615       (write-region (point-min)(point-max) file))))
1616
1617 (defvar what-character-original-window-configuration)
1618
1619 ;;;###autoload
1620 (defun what-char-definition (char)
1621   (interactive (list (char-after)))
1622   (let ((buf (get-buffer-create "*Character Description*"))
1623         (the-buf (current-buffer))
1624         (win-conf (current-window-configuration)))
1625     (pop-to-buffer buf)
1626     (make-local-variable 'what-character-original-window-configuration)
1627     (setq what-character-original-window-configuration win-conf)
1628     (setq buffer-read-only nil)
1629     (erase-buffer)
1630     (condition-case err
1631         (progn
1632           (insert-char-data-with-variant char 'printable)
1633           (unless (char-attribute-alist char)
1634             (insert (format ";; = %c\n"
1635                             (let* ((rest (split-char char))
1636                                    (ccs (pop rest))
1637                                    (code (pop rest)))
1638                               (while rest
1639                                 (setq code (logior (lsh code 8)
1640                                                    (pop rest))))
1641                               (decode-char ccs code)))))
1642           ;; (char-db-update-comment)
1643           (set-buffer-modified-p nil)
1644           (view-mode the-buf (lambda (buf)
1645                                (set-window-configuration
1646                                 what-character-original-window-configuration)
1647                                ))
1648           (goto-char (point-min)))
1649       (error (progn
1650                (set-window-configuration
1651                 what-character-original-window-configuration)
1652                (signal (car err) (cdr err)))))))
1653
1654
1655 ;;; @ end
1656 ;;;
1657
1658 (provide 'char-db-util)
1659
1660 ;;; char-db-util.el ends here