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     =>big5-cdp@component
 412     =>big5-cdp@cognate
 413     ==ucs@gb
 414     =ucs@gb
 415     ==ucs-var-002
 416     =ucs@JP/hanazono
 417     =koseki
 418     =shuowen-jiguge))
 419
 420
 421 ;;; @ char-db formatters
 422 ;;;
 423
 424 (defun char-db-make-char-spec (char)
 425   (let (ret char-spec)
 426     (cond ((characterp char)
 427            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 428                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 429                   (setq char-spec (list (cons '=ucs ret)))
 430                   (cond ((setq ret (get-char-attribute char 'name))
 431                          (setq char-spec (cons (cons 'name ret) char-spec))
 432                          )
 433                         ((setq ret (get-char-attribute char 'name*))
 434                          (setq char-spec (cons (cons 'name* ret) char-spec))
 435                          ))
 436                   )
 437                  ((encode-char char '=mj 'defined-only)
 438                   (setq char-spec nil)
 439                   (dolist (ccs (charset-list))
 440                     (if (and (or (eq ccs '=mj)
 441                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 442                                  (string-match "=ucs@" (symbol-name ccs))
 443                                  )
 444                              (setq ccs (charset-name ccs))
 445                              (null (assq ccs char-spec))
 446                              (setq ret (encode-char char ccs 'defined-only)))
 447                         (setq char-spec (cons (cons ccs ret) char-spec))))
 448                   )
 449                  ((encode-char char '==mj 'defined-only)
 450                   (setq char-spec nil)
 451                   (dolist (ccs (charset-list))
 452                     (if (and (or (eq ccs '==mj)
 453                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 454                                  (string-match "=ucs@" (symbol-name ccs))
 455                                  )
 456                              (setq ccs (charset-name ccs))
 457                              (null (assq ccs char-spec))
 458                              (setq ret (encode-char char ccs 'defined-only)))
 459                         (setq char-spec (cons (cons ccs ret) char-spec))))
 460                   )
 461                  ((encode-char char '=adobe-japan1 'defined-only)
 462                   (setq char-spec nil)
 463                   (dolist (ccs (charset-list))
 464                     (if (and (or (memq ccs
 465                                        '(=adobe-japan1-0
 466                                          =adobe-japan1-1
 467                                          =adobe-japan1-2
 468                                          =adobe-japan1-3
 469                                          =adobe-japan1-4
 470                                          =adobe-japan1-5
 471                                          =adobe-japan1-6
 472                                          ))
 473                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 474                                  (string-match "=ucs@" (symbol-name ccs))
 475                                  )
 476                              (setq ccs (charset-name ccs))
 477                              (null (assq ccs char-spec))
 478                              (setq ret (encode-char char ccs 'defined-only)))
 479                         (setq char-spec (cons (cons ccs ret) char-spec))))
 480                   )
 481                  ((encode-char char '==adobe-japan1 'defined-only)
 482                   (setq char-spec nil)
 483                   (dolist (ccs (charset-list))
 484                     (if (and (or (memq ccs
 485                                        '(==adobe-japan1-0
 486                                          ==adobe-japan1-1
 487                                          ==adobe-japan1-2
 488                                          ==adobe-japan1-3
 489                                          ==adobe-japan1-4
 490                                          ==adobe-japan1-5
 491                                          ==adobe-japan1-6
 492                                          ))
 493                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 494                                  (string-match "=ucs@" (symbol-name ccs))
 495                                  )
 496                              (setq ccs (charset-name ccs))
 497                              (null (assq ccs char-spec))
 498                              (setq ret (encode-char char ccs 'defined-only)))
 499                         (setq char-spec (cons (cons ccs ret) char-spec))))
 500                   )
 501                  ((setq ret
 502                         (catch 'tag
 503                           (let ((rest char-db-coded-charset-priority-list)
 504                                 ccs)
 505                             (while rest
 506                               (setq ccs (charset-name
 507                                          (find-charset (car rest))))
 508                               (if (setq ret
 509                                         (encode-char char ccs
 510                                                      'defined-only))
 511                                   (throw 'tag (cons ccs ret)))
 512                               (setq rest (cdr rest))))))
 513                   (setq char-spec (list ret))
 514                   (dolist (ccs (delq (car ret) (charset-list)))
 515                     (if (and (or (charset-iso-final-char ccs)
 516                                  (memq ccs
 517                                        '(=daikanwa
 518                                          =daikanwa@rev2
 519                                          ;; =gt-k
 520                                          =jis-x0208@1997
 521                                          ))
 522                                  (and (eq (charset-property ccs 'iso-ir) 177)
 523                                       (not (eq (charset-name ccs) '=ucs@big5)))
 524                                  ;; (string-match "=ucs@" (symbol-name ccs))
 525                                  )
 526                              (setq ccs (charset-name ccs))
 527                              (null (assq ccs char-spec))
 528                              (setq ret (encode-char char ccs 'defined-only)))
 529                         (setq char-spec (cons (cons ccs ret) char-spec))))
 530                   (if (null char-spec)
 531                       (setq char-spec (split-char char)))
 532                   (cond ((setq ret (get-char-attribute char 'name))
 533                          (setq char-spec (cons (cons 'name ret) char-spec))
 534                          )
 535                         ((setq ret (get-char-attribute char 'name*))
 536                          (setq char-spec (cons (cons 'name* ret) char-spec))
 537                          ))
 538                   )
 539                  ((setq ret (get-char-attribute
 540                              char 'ideographic-combination))
 541                   (setq char-spec
 542                         (cons (cons 'ideographic-combination ret)
 543                               char-spec))
 544                   ))
 545            char-spec)
 546           ((consp char)
 547            char))))
 548
 549 (defun char-db-insert-char-spec (char &optional readable column
 550                                       required-features)
 551   (unless column
 552     (setq column (current-column)))
 553   (let (char-spec temp-char)
 554     (setq char-spec (char-db-make-char-spec char))
 555     (unless (or (characterp char) ; char
 556                 (condition-case nil
 557                     (setq char (find-char char-spec))
 558                   (error nil)))
 559       ;; define temporary character
 560       ;;   Current implementation is dirty.
 561       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 562                                          char-spec)))
 563       (remove-char-attribute temp-char 'ideograph-daikanwa)
 564       (setq char temp-char))
 565     (insert-char-attributes char
 566                             readable
 567                             (union (mapcar #'car char-spec)
 568                                    required-features)
 569                             nil 'for-sub-node)
 570     (when temp-char
 571       ;; undefine temporary character
 572       ;;   Current implementation is dirty.
 573       (setq char-spec (char-attribute-alist temp-char))
 574       (while char-spec
 575         (remove-char-attribute temp-char (car (car char-spec)))
 576         (setq char-spec (cdr char-spec))))))
 577
 578 (defun char-db-insert-alist (alist &optional readable column)
 579   (unless column
 580     (setq column (current-column)))
 581   (let ((line-breaking
 582          (concat "\n" (make-string (1+ column) ?\ )))
 583         name value
 584         ret al ; cal
 585         key
 586         lbs cell rest separator)
 587     (insert "(")
 588     (while alist
 589       (setq name (car (car alist))
 590             value (cdr (car alist)))
 591       (cond ((eq name 'char)
 592              (insert "(char . ")
 593              (if (setq ret (condition-case nil
 594                                (find-char value)
 595                              (error nil)))
 596                  (progn
 597                    (setq al nil
 598                          ;; cal nil
 599                          )
 600                    (while value
 601                      (setq key (car (car value)))
 602                      ;; (if (find-charset key)
 603                      ;;     (setq cal (cons key cal))
 604                      (setq al (cons key al))
 605                      ;; )
 606                      (setq value (cdr value)))
 607                    (insert-char-attributes ret
 608                                            readable
 609                                            (or al 'none) ; cal
 610                                            nil 'for-sub-node))
 611                (insert (prin1-to-string value)))
 612              (insert ")")
 613              (insert line-breaking))
 614             ((consp value)
 615              (insert (format "(%-18s " name))
 616              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 617              (while (consp value)
 618                (setq cell (car value))
 619                (if (and (consp cell)
 620                         (consp (car cell))
 621                         (setq ret (condition-case nil
 622                                       (find-char cell)
 623                                     (error nil)))
 624                         )
 625                    (progn
 626                      (setq rest cell
 627                            al nil
 628                            ;; cal nil
 629                            )
 630                      (while rest
 631                        (setq key (car (car rest)))
 632                        ;; (if (find-charset key)
 633                        ;;     (setq cal (cons key cal))
 634                        (setq al (cons key al))
 635                        ;; )
 636                        (setq rest (cdr rest)))
 637                      (if separator
 638                          (insert lbs))
 639                      (insert-char-attributes ret
 640                                              readable
 641                                              al ; cal
 642                                              nil 'for-sub-node)
 643                      (setq separator lbs))
 644                  (if separator
 645                      (insert separator))
 646                  (insert (prin1-to-string cell))
 647                  (setq separator " "))
 648                (setq value (cdr value)))
 649              (insert ")")
 650              (insert line-breaking))
 651             (t
 652              (insert (format "(%-18s . %S)%s"
 653                              name value
 654                              line-breaking))))
 655       (setq alist (cdr alist))))
 656   (insert ")"))
 657
 658 (defun char-db-insert-char-reference (plist &optional readable column)
 659   (unless column
 660     (setq column (current-column)))
 661   (let ((line-breaking
 662          (concat "\n" (make-string (1+ column) ?\ )))
 663         (separator "")
 664         name value)
 665     (insert "(")
 666     (while plist
 667       (setq name (pop plist))
 668       (setq value (pop plist))
 669       (cond ((eq name :char)
 670              (insert separator)
 671              (insert ":char\t")
 672              (cond ((numberp value)
 673                     (setq value (decode-char '=ucs value)))
 674                    ;; ((consp value)
 675                    ;;  (setq value (or (find-char value)
 676                    ;;                  value)))
 677                    )
 678              (char-db-insert-char-spec value readable)
 679              (insert line-breaking)
 680              (setq separator ""))
 681             ((eq name :radical)
 682              (insert (format "%s%s\t%d ; %c%s"
 683                              separator
 684                              name value
 685                              (ideographic-radical value)
 686                              line-breaking))
 687              (setq separator ""))
 688             (t
 689              (insert (format "%s%s\t%S" separator name value))
 690              (setq separator line-breaking)))
 691       ))
 692   (insert ")"))
 693
 694 (defun char-db-decode-isolated-char (ccs code-point)
 695   (let (ret)
 696     (setq ret
 697           (cond ((eq ccs 'arabic-iso8859-6)
 698                  (decode-char ccs code-point))
 699                 ;; ((eq ccs '=gt)
 700                 ;;  (decode-builtin-char '==gt code-point))
 701                 ((and (memq ccs '(=gt-pj-1
 702                                   =gt-pj-2
 703                                   =gt-pj-3
 704                                   =gt-pj-4
 705                                   =gt-pj-5
 706                                   =gt-pj-6
 707                                   =gt-pj-7
 708                                   =gt-pj-8
 709                                   =gt-pj-9
 710                                   =gt-pj-10
 711                                   =gt-pj-11))
 712                       (setq ret (decode-char ccs code-point))
 713                       (setq ret (encode-char ret '=gt 'defined-only)))
 714                  (decode-builtin-char '=gt ret))
 715                 (t
 716                  (decode-builtin-char ccs code-point))))
 717     (cond ((null ret)
 718            (or (decode-char ccs code-point)
 719                (define-char (list (cons ccs code-point)))))
 720           ((and (<= 0 (char-int ret))
 721                 (<= (char-int ret) #x1F))
 722            (decode-char '=ucs (+ #x2400 (char-int ret))))
 723           ((= (char-int ret) #x7F)
 724            ?\u2421)
 725           (t ret))))
 726
 727 (defvar char-db-convert-obsolete-format t)
 728
 729 (defun char-db-insert-ccs-feature (name value line-breaking)
 730   (cond
 731    ((integerp value)
 732     (insert
 733      (format
 734       (cond
 735        ((memq name '(=>iwds-1
 736                      ==shinjigen
 737                      =shinjigen
 738                      =shinjigen@1ed ==shinjigen@1ed
 739                      =shinjigen@rev ==shinjigen@rev
 740                      =shinjigen/+p@rev ==shinjigen/+p@rev
 741                      ===daikanwa/ho ==daikanwa/ho
 742                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 743         "(%-18s .  %04d)\t; %c")
 744        ((eq name '=shinjigen@1ed/24pr)
 745         "(%-18s . %04d)\t; %c")
 746        ((or
 747          (memq name
 748                '(===daikanwa
 749                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 750                  =daikanwa@rev1 =daikanwa@rev2
 751                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 752                  =>>daikanwa/+p
 753                  =daikanwa/+2p =>>daikanwa/+2p
 754                  =gt ==gt ===gt
 755                  =>>gt =+>gt =>gt
 756                  =gt-k ==gt-k ===gt-k
 757                  =>>gt-k =>gt-k
 758                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 759                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 760                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 761                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 762                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 763                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 764                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 765                  =>>adobe-japan1-0 =+>adobe-japan1-0
 766                  =>>adobe-japan1-1 =+>adobe-japan1-1
 767                  =>>adobe-japan1-2 =+>adobe-japan1-2
 768                  =>>adobe-japan1-3 =+>adobe-japan1-3
 769                  =>>adobe-japan1-4 =+>adobe-japan1-4
 770                  =>>adobe-japan1-5 =+>adobe-japan1-5
 771                  =>>adobe-japan1-6 =+>adobe-japan1-6
 772                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 773                  =zinbun-oracle =>zinbun-oracle
 774                  =daijiten ==daijiten ===daijiten ==>daijiten
 775                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 776                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 777                  ===hng-khi ===hng-khm ===hng-fhs ===hng-hok
 778                  ===hng-kyd ===hng-sok
 779                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 780                  ===hng-sys ===hng-tsu ===hng-tzj
 781                  ===hng-hos ===hng-kkh ===hng-nak ===hng-jhk
 782                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 783                  ===hng-nkm ===hng-k24 ===hng-ini ===hng-nkk
 784                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 785                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 786                  ===hng-okd ===hng-wad ===hng-kmi
 787                  ===hng-zkd ===hng-doh ===hng-jyu ===hng-tzs
 788                  ===hng-sai ===hng-kad ===hng-kss
 789                  ===hng-kyo ===hng-ykk ===hng-saa
 790                  ===hng-sab ===hng-wks ===hng-wke ===hng-smk
 791                  ===hng-sgs ===hng-sts ===hng-kkd
 792                  =shuowen-jiguge ===shuowen-jiguge4 ===shuowen-jiguge5))
 793          ;; (string-match "^=adobe-" (symbol-name name))
 794          )
 795         "(%-18s . %05d)\t; %c")
 796        ((memq name '(=hanyo-denshi/ks
 797                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 798                      =>>hanyo-denshi/ks
 799                      =koseki ==koseki
 800                      =mj ==mj ===mj =>>mj =>mj
 801                      =zihai mojikyo))
 802         "(%-18s . %06d)\t; %c")
 803        ((memq name '(=hanyo-denshi/tk ==hanyo-denshi/tk))
 804         "(%-18s . %08d)\t; %c")
 805        ((>= (charset-dimension name) 2)
 806         "(%-18s . #x%04X)\t; %c")
 807        (t
 808         "(%-18s . #x%02X)\t; %c"))
 809       name
 810       (if (= (charset-iso-graphic-plane name) 1)
 811           (logior value
 812                   (cond ((= (charset-dimension name) 1)
 813                          #x80)
 814                         ((= (charset-dimension name) 2)
 815                          #x8080)
 816                         ((= (charset-dimension name) 3)
 817                          #x808080)
 818                         (t 0)))
 819         value)
 820       (char-db-decode-isolated-char name value)))
 821     (if (and (= (charset-chars name) 94)
 822              (= (charset-dimension name) 2))
 823         (insert (format " [%02d-%02d]"
 824                         (- (lsh value -8) 32)
 825                         (- (logand value 255) 32))))
 826     )
 827    (t
 828     (insert (format "(%-18s . %s)" name value))
 829     ))
 830   (insert line-breaking))
 831
 832 (defun char-db-insert-relation-feature (char name value line-breaking
 833                                              ccss readable)
 834   (insert (format "(%-18s%s " name line-breaking))
 835   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 836         separator cell sources required-features
 837         ret)
 838     (while (consp value)
 839       (setq cell (car value))
 840       (if (integerp cell)
 841           (setq cell (decode-char '=ucs cell)))
 842       (cond
 843        ((eq name '->subsumptive)
 844         (when (or (not (some (lambda (atr)
 845                                (get-char-attribute cell atr))
 846                              char-db-ignored-attributes))
 847                   (some (lambda (ccs)
 848                           (encode-char cell ccs 'defined-only))
 849                         ccss))
 850           (if separator
 851               (insert lbs))
 852           (let ((char-db-ignored-attributes
 853                  (cons '<-subsumptive
 854                        char-db-ignored-attributes)))
 855             (insert-char-attributes cell readable nil nil 'for-sub-node))
 856           (setq separator lbs))
 857         )
 858        ((characterp cell)
 859         (setq sources
 860               (get-char-attribute
 861                char (intern (format "%s*sources" name))))
 862         (setq required-features nil)
 863         (dolist (source sources)
 864           (cond
 865            ((memq source '(JP
 866                            JP/Jouyou
 867                            shinjigen shinjigen@1ed shinjigen@rev))
 868             (setq required-features
 869                   (union required-features
 870                          '(=jis-x0208
 871                            =jis-x0208@1990
 872                            =jis-x0213-1@2000
 873                            =jis-x0213-1@2004
 874                            =jis-x0213-2
 875                            =jis-x0212
 876                            =jis-x0208@1983
 877                            =jis-x0208@1978
 878                            =shinjigen
 879                            =shinjigen@1ed
 880                            =shinjigen@rev
 881                            =shinjigen/+p@rev))))
 882            ((eq source 'CN)
 883             (setq required-features
 884                   (union required-features
 885                          '(=gb2312
 886                            =gb12345
 887                            =iso-ir165)))))
 888           (cond
 889            ((find-charset (setq ret (intern (format "=%s" source))))
 890             (setq required-features
 891                   (cons ret required-features)))
 892            (t (setq required-features
 893                     (cons source required-features)))))
 894         (cond ((string-match "@JP" (symbol-name name))
 895                (setq required-features
 896                      (union required-features
 897                             '(=jis-x0208
 898                               =jis-x0208@1990
 899                               =jis-x0213-1-2000
 900                               =jis-x0213-2-2000
 901                               =jis-x0212
 902                               =jis-x0208@1983
 903                               =jis-x0208@1978))))
 904               ((string-match "@CN" (symbol-name name))
 905                (setq required-features
 906                      (union required-features
 907                             '(=gb2312
 908                               =gb12345
 909                               =iso-ir165)))))
 910         (if separator
 911             (insert lbs))
 912         (if readable
 913             (insert (format "%S" cell))
 914           (char-db-insert-char-spec cell readable
 915                                     nil
 916                                     required-features))
 917         (setq separator lbs))
 918        ((consp cell)
 919         (if separator
 920             (insert lbs))
 921         (if (consp (car cell))
 922             (char-db-insert-char-spec cell readable)
 923           (char-db-insert-char-reference cell readable))
 924         (setq separator lbs))
 925        (t
 926         (if separator
 927             (insert separator))
 928         (insert (prin1-to-string cell))
 929         (setq separator " ")))
 930       (setq value (cdr value)))
 931     (insert ")")
 932     (insert line-breaking)))
 933
 934 (defun insert-char-attributes (char &optional readable attributes column
 935                                     for-sub-node)
 936   (unless column
 937     (setq column (current-column)))
 938   (let (name value ; has-long-ccs-name
 939         rest
 940         radical strokes
 941         (line-breaking
 942          (concat "\n" (make-string (1+ column) ?\ )))
 943         lbs cell separator ret
 944         key al cal
 945         dest-ccss ; sources required-features
 946         ccss)
 947     (let (atr-d)
 948       (setq attributes
 949             (sort (if attributes
 950                       (if (consp attributes)
 951                           (progn
 952                             (dolist (name attributes)
 953                               (unless (memq name char-db-ignored-attributes)
 954                                 (if (find-charset name)
 955                                     (push name ccss))
 956                                 (push name atr-d)))
 957                             atr-d))
 958                     (dolist (name (char-attribute-list))
 959                       (unless (memq name char-db-ignored-attributes)
 960                         (if (find-charset name)
 961                             (push name ccss))
 962                         (push name atr-d)))
 963                     atr-d)
 964                   #'char-attribute-name<)))
 965     (insert "(")
 966     (when (memq '<-subsumptive attributes)
 967       (when (or readable (not for-sub-node))
 968         (when (setq value (get-char-attribute char '<-subsumptive))
 969           (char-db-insert-relation-feature char '<-subsumptive value
 970                                            line-breaking
 971                                            ccss readable)))
 972       (setq attributes (delq '<-subsumptive attributes)))
 973     (when (and (memq '<-denotational attributes)
 974                (setq value (get-char-attribute char '<-denotational)))
 975       (char-db-insert-relation-feature char '<-denotational value
 976                                        line-breaking
 977                                        ccss readable)
 978       (setq attributes (delq '<-denotational attributes)))
 979     (when (and (memq '<-denotational@component attributes)
 980                (setq value (get-char-attribute char '<-denotational@component)))
 981       (char-db-insert-relation-feature char '<-denotational@component value
 982                                        line-breaking
 983                                        ccss readable)
 984       (setq attributes (delq '<-denotational@component attributes)))
 985     (when (and (memq '<-denotational@usage attributes)
 986                (setq value (get-char-attribute char '<-denotational@usage)))
 987       (char-db-insert-relation-feature char '<-denotational@usage value
 988                                        line-breaking
 989                                        ccss readable)
 990       (setq attributes (delq '<-denotational@usage attributes)))
 991     (when (and (memq 'name attributes)
 992                (setq value (get-char-attribute char 'name)))
 993       (insert (format
 994                (if (> (+ (current-column) (length value)) 48)
 995                    "(name . %S)%s"
 996                  "(name               . %S)%s")
 997                value line-breaking))
 998       (setq attributes (delq 'name attributes))
 999       )
1000     (when (and (memq 'name* attributes)
1001                (setq value (get-char-attribute char 'name*)))
1002       (insert (format
1003                (if (> (+ (current-column) (length value)) 48)
1004                    "(name* . %S)%s"
1005                  "(name*              . %S)%s")
1006                value line-breaking))
1007       (setq attributes (delq 'name* attributes))
1008       )
1009     (when (and (memq 'script attributes)
1010                (setq value (get-char-attribute char 'script)))
1011       (insert (format "(script\t\t%s)%s"
1012                       (mapconcat (function prin1-to-string)
1013                                  value " ")
1014                       line-breaking))
1015       (setq attributes (delq 'script attributes))
1016       )
1017     (dolist (name '(=>ucs =>ucs*))
1018       (when (and (memq name attributes)
1019                  (setq value (get-char-attribute char name)))
1020         (insert (format "(%-18s . #x%04X)\t; %c%s"
1021                         name value (decode-char '=ucs value)
1022                         line-breaking))
1023         (setq attributes (delq name attributes))))
1024     (dolist (name '(=>ucs@gb =>ucs@big5))
1025       (when (and (memq name attributes)
1026                  (setq value (get-char-attribute char name)))
1027         (insert (format "(%-18s . #x%04X)\t; %c%s"
1028                         name value
1029                         (decode-char (intern
1030                                       (concat "="
1031                                               (substring
1032                                                (symbol-name name) 2)))
1033                                      value)
1034                         line-breaking))
1035         (setq attributes (delq name attributes))
1036         ))
1037     ;; (dolist (name '(=>daikanwa))
1038     ;;   (when (and (memq name attributes)
1039     ;;              (setq value (get-char-attribute char name)))
1040     ;;     (insert
1041     ;;      (if (integerp value)
1042     ;;          (format "(%-18s . %05d)\t; %c%s"
1043     ;;                  name value (decode-char '=daikanwa value)
1044     ;;                  line-breaking)
1045     ;;        (format "(%-18s %s)\t; %c%s"
1046     ;;                name
1047     ;;                (mapconcat (function prin1-to-string)
1048     ;;                           value " ")
1049     ;;                (char-representative-of-daikanwa char)
1050     ;;                line-breaking)))
1051     ;;     (setq attributes (delq name attributes))))
1052     (when (and (memq 'general-category attributes)
1053                (setq value (get-char-attribute char 'general-category)))
1054       (insert (format
1055                "(general-category\t%s) ; %s%s"
1056                (mapconcat (lambda (cell)
1057                             (format "%S" cell))
1058                           value " ")
1059                (cond ((rassoc value unidata-normative-category-alist)
1060                       "Normative Category")
1061                      ((rassoc value unidata-informative-category-alist)
1062                       "Informative Category")
1063                      (t
1064                       "Unknown Category"))
1065                line-breaking))
1066       (setq attributes (delq 'general-category attributes))
1067       )
1068     (when (and (memq 'bidi-category attributes)
1069                (setq value (get-char-attribute char 'bidi-category)))
1070       (insert (format "(bidi-category\t. %S)%s"
1071                       value
1072                       line-breaking))
1073       (setq attributes (delq 'bidi-category attributes))
1074       )
1075     (unless (or (not (memq 'mirrored attributes))
1076                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
1077                     'empty))
1078       (insert (format "(mirrored\t\t. %S)%s"
1079                       value
1080                       line-breaking))
1081       (setq attributes (delq 'mirrored attributes))
1082       )
1083     (cond
1084      ((and (memq 'decimal-digit-value attributes)
1085            (setq value (get-char-attribute char 'decimal-digit-value)))
1086       (insert (format "(decimal-digit-value . %S)%s"
1087                       value
1088                       line-breaking))
1089       (setq attributes (delq 'decimal-digit-value attributes))
1090       (when (and (memq 'digit-value attributes)
1091                  (setq value (get-char-attribute char 'digit-value)))
1092         (insert (format "(digit-value\t . %S)%s"
1093                         value
1094                         line-breaking))
1095         (setq attributes (delq 'digit-value attributes))
1096         )
1097       (when (and (memq 'numeric-value attributes)
1098                  (setq value (get-char-attribute char 'numeric-value)))
1099         (insert (format "(numeric-value\t . %S)%s"
1100                         value
1101                         line-breaking))
1102         (setq attributes (delq 'numeric-value attributes))
1103         )
1104       )
1105      (t
1106       (when (and (memq 'digit-value attributes)
1107                  (setq value (get-char-attribute char 'digit-value)))
1108         (insert (format "(digit-value\t. %S)%s"
1109                         value
1110                         line-breaking))
1111         (setq attributes (delq 'digit-value attributes))
1112         )
1113       (when (and (memq 'numeric-value attributes)
1114                  (setq value (get-char-attribute char 'numeric-value)))
1115         (insert (format "(numeric-value\t. %S)%s"
1116                         value
1117                         line-breaking))
1118         (setq attributes (delq 'numeric-value attributes))
1119         )))
1120     (when (and (memq 'iso-10646-comment attributes)
1121                (setq value (get-char-attribute char 'iso-10646-comment)))
1122       (insert (format "(iso-10646-comment\t. %S)%s"
1123                       value
1124                       line-breaking))
1125       (setq attributes (delq 'iso-10646-comment attributes))
1126       )
1127     (when (and (memq 'morohashi-daikanwa attributes)
1128                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1129       (insert (format "(morohashi-daikanwa\t%s)%s"
1130                       (mapconcat (function prin1-to-string) value " ")
1131                       line-breaking))
1132       (setq attributes (delq 'morohashi-daikanwa attributes))
1133       )
1134     (setq radical nil
1135           strokes nil)
1136     (when (and (memq 'ideographic-radical attributes)
1137                (setq value (get-char-attribute char 'ideographic-radical)))
1138       (setq radical value)
1139       (insert (format "(ideographic-radical . %S)\t; %c%s"
1140                       radical
1141                       (ideographic-radical radical)
1142                       line-breaking))
1143       (setq attributes (delq 'ideographic-radical attributes))
1144       )
1145     (when (and (memq 'shuowen-radical attributes)
1146                (setq value (get-char-attribute char 'shuowen-radical)))
1147       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1148                       value
1149                       (shuowen-radical value)
1150                       line-breaking))
1151       (setq attributes (delq 'shuowen-radical attributes))
1152       )
1153     (let (key)
1154       (dolist (domain
1155                (append
1156                 char-db-feature-domains
1157                 (let (dest domain)
1158                   (dolist (feature (char-attribute-list))
1159                     (setq feature (symbol-name feature))
1160                     (when (string-match
1161                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1162                            feature)
1163                       (setq domain (intern (match-string 2 feature)))
1164                      (unless (memq domain dest)
1165                        (setq dest (cons domain dest)))))
1166                   (sort dest #'string<))))
1167         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1168         (when (and (memq key attributes)
1169                    (setq value (get-char-attribute char key)))
1170           (setq radical value)
1171           (insert (format "(%s . %S)\t; %c%s"
1172                           key
1173                           radical
1174                           (ideographic-radical radical)
1175                           line-breaking))
1176           (setq attributes (delq key attributes))
1177           )
1178         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1179         (when (and (memq key attributes)
1180                    (setq value (get-char-attribute char key)))
1181           (setq strokes value)
1182           (insert (format "(%s . %S)%s"
1183                           key
1184                           strokes
1185                           line-breaking))
1186           (setq attributes (delq key attributes))
1187           )
1188         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1189         (when (and (memq key attributes)
1190                    (setq value (get-char-attribute char key)))
1191           (insert (format "(%s       . %S)%s"
1192                           key
1193                           value
1194                           line-breaking))
1195           (setq attributes (delq key attributes))
1196           )
1197         (dolist (feature '(ideographic-radical
1198                            ideographic-strokes
1199                            total-strokes))
1200           (setq key (intern (format "%s@%s*sources" feature domain)))
1201           (when (and (memq key attributes)
1202                      (setq value (get-char-attribute char key)))
1203             (insert (format "(%s%s" key line-breaking))
1204             (dolist (cell value)
1205               (insert (format " %s" cell)))
1206             (insert ")")
1207             (insert line-breaking)
1208             (setq attributes (delq key attributes))
1209             ))
1210         ))
1211     (when (and (memq 'ideographic-strokes attributes)
1212                (setq value (get-char-attribute char 'ideographic-strokes)))
1213       (setq strokes value)
1214       (insert (format "(ideographic-strokes . %S)%s"
1215                       strokes
1216                       line-breaking))
1217       (setq attributes (delq 'ideographic-strokes attributes))
1218       )
1219     (when (and (memq 'kangxi-radical attributes)
1220                (setq value (get-char-attribute char 'kangxi-radical)))
1221       (unless (eq value radical)
1222         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1223                         value
1224                         (ideographic-radical value)
1225                         line-breaking))
1226         (or radical
1227             (setq radical value)))
1228       (setq attributes (delq 'kangxi-radical attributes))
1229       )
1230     (when (and (memq 'kangxi-strokes attributes)
1231                (setq value (get-char-attribute char 'kangxi-strokes)))
1232       (unless (eq value strokes)
1233         (insert (format "(kangxi-strokes\t . %S)%s"
1234                         value
1235                         line-breaking))
1236         (or strokes
1237             (setq strokes value)))
1238       (setq attributes (delq 'kangxi-strokes attributes))
1239       )
1240     ;; (when (and (memq 'japanese-radical attributes)
1241     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1242     ;;   (unless (eq value radical)
1243     ;;     (insert (format "(japanese-radical\t . %S)\t; %c%s"
1244     ;;                     value
1245     ;;                     (ideographic-radical value)
1246     ;;                     line-breaking))
1247     ;;     (or radical
1248     ;;         (setq radical value)))
1249     ;;   (setq attributes (delq 'japanese-radical attributes))
1250     ;;   )
1251     (when (and (memq 'japanese-strokes attributes)
1252                (setq value (get-char-attribute char 'japanese-strokes)))
1253       (unless (eq value strokes)
1254         (insert (format "(japanese-strokes\t . %S)%s"
1255                         value
1256                         line-breaking))
1257         (or strokes
1258             (setq strokes value)))
1259       (setq attributes (delq 'japanese-strokes attributes))
1260       )
1261     (when (and (memq 'cns-radical attributes)
1262                (setq value (get-char-attribute char 'cns-radical)))
1263       (insert (format "(cns-radical\t . %S)\t; %c%s"
1264                       value
1265                       (ideographic-radical value)
1266                       line-breaking))
1267       (setq attributes (delq 'cns-radical attributes))
1268       )
1269     (when (and (memq 'cns-strokes attributes)
1270                (setq value (get-char-attribute char 'cns-strokes)))
1271       (unless (eq value strokes)
1272         (insert (format "(cns-strokes\t . %S)%s"
1273                         value
1274                         line-breaking))
1275         (or strokes
1276             (setq strokes value)))
1277       (setq attributes (delq 'cns-strokes attributes))
1278       )
1279     ;; (when (and (memq 'ideographic- attributes)
1280     ;;            (setq value (get-char-attribute char 'ideographic-)))
1281     ;;   (insert "(ideographic-       ")
1282     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1283     ;;         separator nil)
1284     ;;   (while (consp value)
1285     ;;     (setq cell (car value))
1286     ;;     (if (integerp cell)
1287     ;;         (setq cell (decode-char '=ucs cell)))
1288     ;;     (cond ((characterp cell)
1289     ;;            (if separator
1290     ;;                (insert lbs))
1291     ;;            (if readable
1292     ;;                (insert (format "%S" cell))
1293     ;;              (char-db-insert-char-spec cell readable))
1294     ;;            (setq separator lbs))
1295     ;;           ((consp cell)
1296     ;;            (if separator
1297     ;;                (insert lbs))
1298     ;;            (if (consp (car cell))
1299     ;;                (char-db-insert-char-spec cell readable)
1300     ;;              (char-db-insert-char-reference cell readable))
1301     ;;            (setq separator lbs))
1302     ;;           (t
1303     ;;            (if separator
1304     ;;                (insert separator))
1305     ;;            (insert (prin1-to-string cell))
1306     ;;            (setq separator " ")))
1307     ;;     (setq value (cdr value)))
1308     ;;   (insert ")")
1309     ;;   (insert line-breaking)
1310     ;;   (setq attributes (delq 'ideographic- attributes)))
1311     (when (and (memq 'total-strokes attributes)
1312                (setq value (get-char-attribute char 'total-strokes)))
1313       (insert (format "(total-strokes       . %S)%s"
1314                       value
1315                       line-breaking))
1316       (setq attributes (delq 'total-strokes attributes))
1317       )
1318     ;; (when (and (memq '->ideograph attributes)
1319     ;;            (setq value (get-char-attribute char '->ideograph)))
1320     ;;   (insert (format "(->ideograph\t%s)%s"
1321     ;;                   (mapconcat (lambda (code)
1322     ;;                                (cond ((symbolp code)
1323     ;;                                       (symbol-name code))
1324     ;;                                      ((integerp code)
1325     ;;                                       (format "#x%04X" code))
1326     ;;                                      (t
1327     ;;                                       (format "%s %S"
1328     ;;                                               line-breaking code))))
1329     ;;                              value " ")
1330     ;;                   line-breaking))
1331     ;;   (setq attributes (delq '->ideograph attributes))
1332     ;;   )
1333     ;; (when (and (memq '->decomposition attributes)
1334     ;;            (setq value (get-char-attribute char '->decomposition)))
1335     ;;   (insert (format "(->decomposition\t%s)%s"
1336     ;;                   (mapconcat (lambda (code)
1337     ;;                                (cond ((symbolp code)
1338     ;;                                       (symbol-name code))
1339     ;;                                      ((characterp code)
1340     ;;                                       (if readable
1341     ;;                                           (format "%S" code)
1342     ;;                                         (format "#x%04X"
1343     ;;                                                 (char-int code))
1344     ;;                                         ))
1345     ;;                                      ((integerp code)
1346     ;;                                       (format "#x%04X" code))
1347     ;;                                      (t
1348     ;;                                       (format "%s%S" line-breaking code))))
1349     ;;                              value " ")
1350     ;;                   line-breaking))
1351     ;;   (setq attributes (delq '->decomposition attributes))
1352     ;;   )
1353     (if (equal (get-char-attribute char '->titlecase)
1354                (get-char-attribute char '->uppercase))
1355         (setq attributes (delq '->titlecase attributes)))
1356     (when (and (memq '->mojikyo attributes)
1357                (setq value (get-char-attribute char '->mojikyo)))
1358       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1359                       value (decode-char 'mojikyo value)
1360                       line-breaking))
1361       (setq attributes (delq '->mojikyo attributes))
1362       )
1363     (unless readable
1364       (dolist (ignored '(composition
1365                          ->denotational <-subsumptive ->ucs-unified
1366                          ;; ->ideographic-component-forms
1367                          ))
1368         (setq attributes (delq ignored attributes))))
1369     (while attributes
1370       (setq name (car attributes))
1371       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1372                   'value-is-empty)
1373         (cond ((setq ret (find-charset name))
1374                (setq name (charset-name ret))
1375                (when (not (memq name dest-ccss))
1376                  (setq dest-ccss (cons name dest-ccss))
1377                  (char-db-insert-ccs-feature name value line-breaking))
1378                )
1379               ((string-match "^=>ucs@" (symbol-name name))
1380                (insert (format "(%-18s . #x%04X)\t; %c%s"
1381                                name value (decode-char '=ucs value)
1382                                line-breaking))
1383                )
1384               ((eq name 'jisx0208-1978/4X)
1385                (insert (format "(%-18s . #x%04X)%s"
1386                                name value
1387                                line-breaking))
1388                )
1389               ((and
1390                 (not readable)
1391                 (not (eq name '->subsumptive))
1392                 (not (eq name '->uppercase))
1393                 (not (eq name '->lowercase))
1394                 (not (eq name '->titlecase))
1395                 (not (eq name '->canonical))
1396                 (not (eq name '->Bopomofo))
1397                 (not (eq name '->mistakable))
1398                 (not (eq name '->ideographic-variants))
1399                 (null (get-char-attribute
1400                        char (intern (format "%s*sources" name))))
1401                 (not (string-match "\\*sources$" (symbol-name name)))
1402                 (null (get-char-attribute
1403                        char (intern (format "%s*note" name))))
1404                 (not (string-match "\\*note$" (symbol-name name)))
1405                 (or (eq name '<-identical)
1406                     (eq name '<-uppercase)
1407                     (eq name '<-lowercase)
1408                     (eq name '<-titlecase)
1409                     (eq name '<-canonical)
1410                     (eq name '<-ideographic-variants)
1411                     ;; (eq name '<-synonyms)
1412                     (string-match "^<-synonyms" (symbol-name name))
1413                     (eq name '<-mistakable)
1414                     (when (string-match "^->" (symbol-name name))
1415                       (cond
1416                        ((string-match "^->fullwidth" (symbol-name name))
1417                         (not (and (consp value)
1418                                   (characterp (car value))
1419                                   (encode-char
1420                                    (car value) '=ucs 'defined-only)))
1421                         )
1422                        (t)))
1423                     ))
1424                )
1425               ((or (eq name 'ideographic-structure)
1426                    (eq name 'ideographic-structure@apparent)
1427                    (eq name 'ideographic-combination)
1428                    ;; (eq name 'ideographic-)
1429                    (eq name '=decomposition)
1430                    (char-feature-base-name= '=decomposition name)
1431                    (char-feature-base-name= '=>decomposition name)
1432                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1433                    ;;               (symbol-name name))
1434                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1435                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1436                                  (symbol-name name))
1437                    )
1438                (char-db-insert-relation-feature char name value
1439                                                 line-breaking
1440                                                 ccss readable))
1441               ;; ((memq name '(ideograph=
1442               ;;               original-ideograph-of
1443               ;;               ancient-ideograph-of
1444               ;;               vulgar-ideograph-of
1445               ;;               wrong-ideograph-of
1446               ;;               ;; simplified-ideograph-of
1447               ;;               ideographic-variants
1448               ;;               ;; ideographic-different-form-of
1449               ;;               ))
1450               ;;  (insert (format "(%-18s%s " name line-breaking))
1451               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1452               ;;        separator nil)
1453               ;;  (while (consp value)
1454               ;;    (setq cell (car value))
1455               ;;    (if (and (consp cell)
1456               ;;             (consp (car cell)))
1457               ;;        (progn
1458               ;;          (if separator
1459               ;;              (insert lbs))
1460               ;;          (char-db-insert-alist cell readable)
1461               ;;          (setq separator lbs))
1462               ;;      (if separator
1463               ;;          (insert separator))
1464               ;;      (insert (prin1-to-string cell))
1465               ;;      (setq separator " "))
1466               ;;    (setq value (cdr value)))
1467               ;;  (insert ")")
1468               ;;  (insert line-breaking))
1469               ((consp value)
1470                (insert (format "(%-18s " name))
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                           (setq ret (condition-case nil
1478                                         (find-char cell)
1479                                       (error nil))))
1480                      (progn
1481                        (setq rest cell
1482                              al nil
1483                              cal nil)
1484                        (while rest
1485                          (setq key (car (car rest)))
1486                          (if (find-charset key)
1487                              (setq cal (cons key cal))
1488                            (setq al (cons key al)))
1489                          (setq rest (cdr rest)))
1490                        (if separator
1491                            (insert lbs))
1492                        (insert-char-attributes ret
1493                                                readable
1494                                                al ; cal
1495                                                nil 'for-sub-node)
1496                        (setq separator lbs))
1497                    (setq ret (prin1-to-string cell))
1498                    (if separator
1499                        (if (< (+ (current-column)
1500                                  (length ret)
1501                                  (length separator))
1502                               76)
1503                            (insert separator)
1504                          (insert lbs)))
1505                    (insert ret)
1506                    (setq separator " "))
1507                  (setq value (cdr value)))
1508                (insert ")")
1509                (insert line-breaking))
1510               (t
1511                (insert (format "(%-18s" name))
1512                (setq ret (prin1-to-string value))
1513                (unless (< (+ (current-column)
1514                              (length ret)
1515                              3)
1516                           76)
1517                  (insert line-breaking))
1518                (insert " . " ret ")" line-breaking)
1519                ;; (insert (format "(%-18s . %S)%s"
1520                ;;                 name value
1521                ;;                 line-breaking))
1522                )
1523               ))
1524       (setq attributes (cdr attributes)))
1525     (insert ")")))
1526
1527 (defun insert-char-data (char &optional readable
1528                               attributes)
1529   (save-restriction
1530     (narrow-to-region (point)(point))
1531     (insert "(define-char
1532   '")
1533     (insert-char-attributes char readable attributes)
1534     (insert ")\n")
1535     (goto-char (point-min))
1536     (while (re-search-forward "[ \t]+$" nil t)
1537       (replace-match ""))
1538     ;; from tabify.
1539     (goto-char (point-min))
1540     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1541       (let ((column (current-column))
1542             (indent-tabs-mode t))
1543         (delete-region (match-beginning 0) (point))
1544         (indent-to column)))
1545     (goto-char (point-max))
1546     ;; (tabify (point-min)(point-max))
1547     ))
1548
1549 (defun insert-char-data-with-variant (char &optional printable
1550                                            no-ucs-unified
1551                                            script excluded-script)
1552   (insert-char-data char printable)
1553   (let ((variants (char-variants char))
1554         rest
1555         variant vs ret)
1556     (setq variants (sort variants #'<))
1557     (setq rest variants)
1558     (setq variants (cons char variants))
1559     (while rest
1560       (setq variant (car rest))
1561       (unless (get-char-attribute variant '<-subsumptive)
1562         (if (and (or (null script)
1563                      (null (setq vs (get-char-attribute variant 'script)))
1564                      (memq script vs))
1565                  (or (null excluded-script)
1566                      (null (setq vs (get-char-attribute variant 'script)))
1567                      (not (memq excluded-script vs))))
1568             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1569               (insert-char-data variant printable)
1570               (if (setq ret (char-variants variant))
1571                   (while ret
1572                     (or (memq (car ret) variants)
1573                         ;; (get-char-attribute (car ret) '<-subsumptive)
1574                         (setq rest (nconc rest (list (car ret)))))
1575                     (setq ret (cdr ret)))))))
1576       (setq rest (cdr rest)))))
1577
1578 (defun insert-char-range-data (min max &optional script excluded-script)
1579   (let ((code min)
1580         char)
1581     (while (<= code max)
1582       (setq char (decode-char '=ucs code))
1583       (if (encode-char char '=ucs 'defined-only)
1584           (insert-char-data-with-variant char nil 'no-ucs-unified
1585                                          script excluded-script))
1586       (setq code (1+ code)))))
1587
1588 (defun write-char-range-data-to-file (min max file
1589                                           &optional script excluded-script)
1590   (let ((coding-system-for-write char-db-file-coding-system))
1591     (with-temp-buffer
1592       (insert (format ";; -*- coding: %s -*-\n"
1593                       char-db-file-coding-system))
1594       (insert-char-range-data min max script excluded-script)
1595       (write-region (point-min)(point-max) file))))
1596
1597 (defvar what-character-original-window-configuration)
1598
1599 ;;;###autoload
1600 (defun what-char-definition (char)
1601   (interactive (list (char-after)))
1602   (let ((buf (get-buffer-create "*Character Description*"))
1603         (the-buf (current-buffer))
1604         (win-conf (current-window-configuration)))
1605     (pop-to-buffer buf)
1606     (make-local-variable 'what-character-original-window-configuration)
1607     (setq what-character-original-window-configuration win-conf)
1608     (setq buffer-read-only nil)
1609     (erase-buffer)
1610     (condition-case err
1611         (progn
1612           (insert-char-data-with-variant char 'printable)
1613           (unless (char-attribute-alist char)
1614             (insert (format ";; = %c\n"
1615                             (let* ((rest (split-char char))
1616                                    (ccs (pop rest))
1617                                    (code (pop rest)))
1618                               (while rest
1619                                 (setq code (logior (lsh code 8)
1620                                                    (pop rest))))
1621                               (decode-char ccs code)))))
1622           ;; (char-db-update-comment)
1623           (set-buffer-modified-p nil)
1624           (view-mode the-buf (lambda (buf)
1625                                (set-window-configuration
1626                                 what-character-original-window-configuration)
1627                                ))
1628           (goto-char (point-min)))
1629       (error (progn
1630                (set-window-configuration
1631                 what-character-original-window-configuration)
1632                (signal (car err) (cdr err)))))))
1633
1634
1635 ;;; @ end
1636 ;;;
1637
1638 (provide 'char-db-util)
1639
1640 ;;; char-db-util.el ends here