lisp/utf-2000/char-db-util.el

   1 ;;; char-db-util.el --- Character Database utility -*- coding: utf-8-er; -*-
   2
   3 ;; Copyright (C) 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007,
   4 ;;   2008, 2009, 2010, 2011, 2012, 2013, 2014, 2015, 2016, 2017, 2018, 2019,
   5 ;;   2020, 2021 MORIOKA Tomohiko.
   6
   7 ;; Author: MORIOKA Tomohiko <tomo@kanji.zinbun.kyoto-u.ac.jp>
   8 ;; Keywords: CHISE, Character Database, ISO/IEC 10646, UCS, Unicode, MULE.
   9
  10 ;; This file is part of XEmacs CHISE.
  11
  12 ;; XEmacs CHISE is free software; you can redistribute it and/or
  13 ;; modify it under the terms of the GNU General Public License as
  14 ;; published by the Free Software Foundation; either version 2, or (at
  15 ;; your option) any later version.
  16
  17 ;; XEmacs CHISE is distributed in the hope that it will be useful,
  18 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  19 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  20 ;; General Public License for more details.
  21
  22 ;; You should have received a copy of the GNU General Public License
  23 ;; along with XEmacs CHISE; see the file COPYING.  If not, write to
  24 ;; the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  25 ;; Boston, MA 02111-1307, USA.
  26
  27 ;;; Code:
  28
  29 (require 'chise-subr)
  30 (require 'ideograph-subr)
  31
  32 (defconst unidata-normative-category-alist
  33   '(("Lu" letter        uppercase)
  34     ("Ll" letter        lowercase)
  35     ("Lt" letter        titlecase)
  36     ("Mn" mark          non-spacing)
  37     ("Mc" mark          spacing-combining)
  38     ("Me" mark          enclosing)
  39     ("Nd" number        decimal-digit)
  40     ("Nl" number        letter)
  41     ("No" number        other)
  42     ("Zs" separator     space)
  43     ("Zl" separator     line)
  44     ("Zp" separator     paragraph)
  45     ("Cc" other         control)
  46     ("Cf" other         format)
  47     ("Cs" other         surrogate)
  48     ("Co" other         private-use)
  49     ("Cn" other         not-assigned)))
  50
  51 (defconst unidata-informative-category-alist
  52   '(("Lm" letter        modifier)
  53     ("Lo" letter        other)
  54     ("Pc" punctuation   connector)
  55     ("Pd" punctuation   dash)
  56     ("Ps" punctuation   open)
  57     ("Pe" punctuation   close)
  58     ("Pi" punctuation   initial-quote)
  59     ("Pf" punctuation   final-quote)
  60     ("Po" punctuation   other)
  61     ("Sm" symbol        math)
  62     ("Sc" symbol        currency)
  63     ("Sk" symbol        modifier)
  64     ("So" symbol        other)
  65     ))
  66
  67 (defconst shuowen-radicals
  68   [?一 ?上 ?示 ?三 ?王 ?玉 ?玨 ?气 ?士 ?丨 ; 010
  69    ?屮 ?艸 ?蓐 ?茻 ?小 ?八 ?釆 ?半 ?牛 ?犛 ; 020
  70    ?告 ?口 ?凵 ?吅 ?哭 ?走 ?止 ?癶 ?步 ?此 ; 030
  71    ?正 ?是 ?辵 ?彳 ?廴 ?㢟 ?行 ?齒 ?牙 ?足 ; 040
  72    ?疋 ?品 ?龠 ?冊 ?㗊 ?舌 ?干 ?𧮫 ?只 ?㕯 ; 050
  73    ?句 ?丩 ?古 ?十 ?卅 ?言 ?誩 ?音 ?䇂 ?丵 ; 060
  74    ?菐 ?𠬞 ?𠬜 ?共 ?異 ?舁 ?𦥑 ?䢅 ?爨 ?革 ; 070
  75    ?鬲 ?䰜 ?爪 ?𠃨 ?鬥 ?又 ?𠂇 ?史 ?支 ?𦘒 ; 080
  76    ?聿 ?畫 ?隶 ?臤 ?臣 ?殳 ?殺 ?𠘧 ?寸 ?皮 ; 090
  77    ?㼱 ?攴 ?敎 ?卜 ?用 ?爻 ?㸚 ?𥄎 ?目 ?䀠 ; 100
  78    ?眉 ?盾 ?自 ?白 ?鼻 ?皕 ?習 ?羽 ?隹 ?奞 ; 110
  79    ?雈 ?𦫳 ?𥄕 ?羊 ?羴 ?瞿 ?雔 ?雥 ?鳥 ?烏 ; 120
  80    ?𠦒 ?冓 ?幺 ?𢆶 ?叀 ?玄 ?予 ?放 ?𠬪 ?𣦼 ; 130
  81    ?歺 ?死 ?冎 ?骨 ?肉 ?筋 ?刀 ?刃 ?㓞 ?丯 ; 140
  82    ?耒 ?角 ?竹 ?箕 ?丌 ?左 ?工 ?㠭 ?巫 ?甘 ; 150
  83    ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?旨 ?喜 ?壴 ; 160
  84 ;  ?旨 ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?喜 ?壴 ; 160
  85    ?鼓 ?豈 ?豆 ?豊 ?豐 ?䖒 ?虍 ?虎 ?虤 ?皿 ; 170
  86    ?𠙴 ?去 ?血 ?丶 ?丹 ?青 ?井 ?皀 ?鬯 ?食 ; 180
  87    ?亼 ?會 ?倉 ?入 ?缶 ?矢 ?高 ?冂 ?𩫏 ?京 ; 190
  88    ?亯 ?𣆪 ?畗 ?㐭 ?嗇 ?來 ?麥 ?夊 ?舛 ?䑞 ; 200
  89    ?韋 ?弟 ?夂 ?久 ?桀 ?木 ?東 ?林 ?才 ?叒 ; 210
  90    ?之 ?帀 ?出 ?𣎵 ?生 ?乇 ?𠂹 ?𠌶 ?華 ?𥝌 ; 220
  91    ?稽 ?巢 ?桼 ?束 ?㯻 ?囗 ?員 ?貝 ?邑 ?𨛜 ; 230
  92    ?日 ?旦 ?倝 ?㫃 ?冥 ?晶 ?月 ?有 ?明 ?囧 ; 240
  93    ?夕 ?多 ?毌 ?𢎘 ?𣐺 ?卣 ?齊 ?朿 ?片 ?鼎 ; 250
  94    ?克 ?彔 ?禾 ?秝 ?黍 ?香 ?米 ?毇 ?臼 ?凶 ; 260
  95    ?𣎳 ?𣏟 ?麻 ?尗 ?耑 ?韭 ?瓜 ?瓠 ?宀 ?宮 ; 270
  96    ?呂 ?穴 ?㝱 ?疒 ?冖 ?𠔼 ?冃 ?㒳 ?网 ?襾 ; 280
  97    ?巾 ?巿 ?帛 ?白 ?㡀 ?黹 ?人 ?𠤎 ?匕 ?从 ; 290
  98    ?比 ?北 ?丘 ?㐺 ?𡈼 ?重 ?臥 ?身 ?㐆 ?衣 ; 300
  99    ?裘 ?老 ?毛 ?毳 ?尸 ?尺 ?尾 ?履 ?舟 ?方 ; 310
 100    ?儿 ?兄 ?兂 ?皃 ?𠑹 ?先 ?秃 ?見 ?覞 ?欠 ; 320
 101    ?㱃 ?㳄 ?旡 ?頁 ?𦣻 ?面 ?丏 ?首 ?𥄉 ?須 ; 330
 102    ?彡 ?彣 ?文 ?髟 ?后 ?司 ?卮 ?卩 ?印 ?色 ; 340
 103    ?𠨍 ?辟 ?勹 ?包 ?茍 ?鬼 ?甶 ?厶 ?嵬 ?山 ; 350
 104    ?屾 ?屵 ?广 ?厂 ?丸 ?危 ?石 ?長 ?勿 ?冄 ; 360
 105    ?而 ?豕 ?㣇 ?彑 ?豚 ?豸 ?𤉡 ?易 ?象 ?馬 ; 370
 106    ?𢊁 ?鹿 ?麤 ?㲋 ?兔 ?萈 ?犬 ?㹜 ?鼠 ?能 ; 380
 107    ?熊 ?火 ?炎 ?黑 ?囪 ?焱 ?炙 ?赤 ?大 ?亦 ; 390
 108    ?夨 ?夭 ?交 ?尣 ?壺 ?壹 ?㚔 ?奢 ?亢 ?夲 ; 400
 109    ?夰 ?亣 ?夫 ?立 ?竝 ?囟 ?思 ?心 ?惢 ?水 ; 410
 110    ?沝 ?瀕 ?𡿨 ?巜 ?川 ?泉 ?灥 ?永 ?𠂢 ?谷 ; 420
 111    ?仌 ?雨 ?雲 ?魚 ?𩺰 ?燕 ?龍 ?飛 ?非 ?卂 ; 430
 112    ?𠃉 ?不 ?至 ?西 ?鹵 ?鹽 ?戶 ?門 ?耳 ?𦣝 ; 440
 113    ?手 ?𠦬 ?女 ?毋 ?民 ?丿 ?𠂆 ?乁 ?氏 ?氐 ; 450
 114    ?戈 ?戉 ?我 ?亅 ?琴 ?𠃊 ?亡 ?匸 ?匚 ?曲 ; 460
 115    ?甾 ?瓦 ?弓 ?弜 ?弦 ?系 ?糸 ?素 ?絲 ?率 ; 470
 116    ?虫 ?䖵 ?蟲 ?風 ?它 ?龜 ?黽 ?卵 ?二 ?土 ; 480
 117    ?垚 ?堇 ?里 ?田 ?畕 ?黃 ?男 ?力 ?劦 ?金 ; 490
 118    ?幵 ?勺 ?几 ?且 ?斤 ?斗 ?矛 ?車 ?𠂤 ?𨸏 ; 500
 119    ?𨺅 ?厽 ?四 ?宁 ?叕 ?亞 ?五 ?六 ?七 ?九 ; 510
 120    ?禸 ?嘼 ?甲 ?乙 ?丙 ?丁 ?戊 ?己 ?巴 ?庚 ; 520
 121    ?辛 ?辡 ?壬 ?癸 ?子 ?了 ?孨 ?𠫓 ?丑 ?寅 ; 530
 122    ?卯 ?辰 ?巳 ?午 ?未 ?申 ?酉 ?酋 ?戌 ?亥 ; 540
 123    ])
 124
 125 (defun shuowen-radical (number)
 126   (aref shuowen-radicals (1- number)))
 127
 128 (defvar char-db-file-coding-system 'utf-8-mcs-er)
 129
 130 (defvar char-db-ignored-attributes '(ideographic-products))
 131
 132 (defvar char-db-coded-charset-priority-list
 133   '(ascii
 134     control-1
 135     latin-iso8859-1
 136     latin-iso8859-2
 137     latin-iso8859-3
 138     latin-iso8859-4
 139     latin-iso8859-9
 140     latin-jisx0201
 141     cyrillic-iso8859-5
 142     greek-iso8859-7
 143     thai-tis620
 144     ;; =mj
 145     ;; =adobe-japan1-0
 146     ;; =adobe-japan1-1
 147     ;; =adobe-japan1-2
 148     ;; =adobe-japan1-3
 149     ;; =adobe-japan1-4
 150     ;; =adobe-japan1-5
 151     ;; =adobe-japan1-6
 152     =jis-x0208
 153     =jis-x0208@1978
 154     =jis-x0208@1983
 155     japanese-jisx0212
 156     chinese-gb2312
 157     =jis-x0208@1990
 158     chinese-cns11643-1
 159     chinese-cns11643-2
 160     chinese-cns11643-3
 161     chinese-cns11643-4
 162     chinese-cns11643-5
 163     chinese-cns11643-6
 164     chinese-cns11643-7
 165     =jis-x0213-1
 166     =jis-x0213-1@2000
 167     =jis-x0213-1@2004
 168     =jis-x0213-2
 169     korean-ksc5601
 170     chinese-isoir165
 171     katakana-jisx0201
 172     hebrew-iso8859-8
 173     chinese-gb12345
 174     latin-viscii
 175     ethiopic-ucs
 176     =big5-cdp
 177     =hanyo-denshi/ja
 178     =hanyo-denshi/jb
 179     =hanyo-denshi/jc
 180     =hanyo-denshi/jd
 181     =hanyo-denshi/ft
 182     =hanyo-denshi/ia
 183     =hanyo-denshi/ib
 184     =hanyo-denshi/hg
 185     =hanyo-denshi/jt
 186     =hanyo-denshi/ks
 187     =hanyo-denshi/tk
 188     ;; ==mj
 189     ;; ==adobe-japan1-0
 190     ;; ==adobe-japan1-1
 191     ;; ==adobe-japan1-2
 192     ;; ==adobe-japan1-3
 193     ;; ==adobe-japan1-4
 194     ;; ==adobe-japan1-5
 195     ;; ==adobe-japan1-6
 196     ==jis-x0208
 197     ==jis-x0213-1
 198     ==jis-x0213-2
 199     ==jis-x0212
 200     ==hanyo-denshi/ja
 201     ==hanyo-denshi/jb
 202     ==hanyo-denshi/jc
 203     ==hanyo-denshi/ft
 204     ==hanyo-denshi/ia
 205     ==hanyo-denshi/ib
 206     ==hanyo-denshi/hg
 207     ==hanyo-denshi/jt
 208     ==hanyo-denshi/ks
 209     ==hanyo-denshi/tk
 210     =gt
 211     =gt-k
 212     =daikanwa
 213     =daijiten
 214     =daikanwa@rev2
 215     =daikanwa@rev1
 216     =daikanwa/+p
 217     =daikanwa/ho
 218     ==daikanwa
 219     ==daijiten
 220     =cbeta
 221     ideograph-hanziku-1
 222     ideograph-hanziku-2
 223     ideograph-hanziku-3
 224     ideograph-hanziku-4
 225     ideograph-hanziku-5
 226     ideograph-hanziku-6
 227     ideograph-hanziku-7
 228     ideograph-hanziku-8
 229     ideograph-hanziku-9
 230     ideograph-hanziku-10
 231     ideograph-hanziku-11
 232     ideograph-hanziku-12
 233     ;; =>>>adobe-japan1-0
 234     ;; =>>>adobe-japan1-1
 235     ;; =>>>adobe-japan1-2
 236     ;; =>>>adobe-japan1-3
 237     ;; =>>>adobe-japan1-4
 238     ;; =>>>adobe-japan1-5
 239     ;; =>>>adobe-japan1-6
 240     ;; =>>>jis-x0208
 241     ;; =>>>jis-x0213-1
 242     ;; =>>>jis-x0213-2
 243     ;; =>>>hanyo-denshi/ja
 244     ;; =>>>hanyo-denshi/jb
 245     ;; =>>>hanyo-denshi/jc
 246     ;; =>>>hanyo-denshi/ft
 247     ;; =>>>hanyo-denshi/ib
 248     ;; =>>>hanyo-denshi/hg
 249     ;; =>>>hanyo-denshi/jt
 250     ;; =>>>hanyo-denshi/ks
 251     ;; =>>>gt
 252     =>>adobe-japan1-0
 253     =>>adobe-japan1-1
 254     =>>adobe-japan1-2
 255     =>>adobe-japan1-3
 256     =>>adobe-japan1-4
 257     =>>adobe-japan1-5
 258     =>>adobe-japan1-6
 259     =>>jis-x0208
 260     =>>jis-x0213-1
 261     =>>jis-x0213-1@2000
 262     =>>jis-x0213-1@2004
 263     =>>jis-x0213-2
 264     =>>jis-x0208@1978
 265     =>>hanyo-denshi/ft
 266     =>>hanyo-denshi/jt
 267     =>>hanyo-denshi/ks
 268     =>>gt
 269     =>>daikanwa
 270     =>>cbeta
 271     =+>jis-x0208
 272     =+>jis-x0213-1
 273     =+>jis-x0213-2
 274     =+>adobe-japan1-0
 275     =+>adobe-japan1-1
 276     =+>adobe-japan1-2
 277     =+>adobe-japan1-3
 278     =+>adobe-japan1-4
 279     =+>adobe-japan1-5
 280     =+>adobe-japan1-6
 281     =+>jis-x0208@1978
 282     =>jis-x0208
 283     =>jis-x0208@1997
 284     =>jis-x0213-1
 285     =>jis-x0213-1@2000
 286     =>jis-x0213-1@2004
 287     =>jis-x0213-2
 288     ==>ucs@bucs
 289     ==>daijiten
 290     ;; =>ucs@hanyo-denshi
 291     =>ucs@iso
 292     =>ucs@unicode
 293     =>ucs@jis
 294     =>ucs@cns
 295     =>ucs@ks
 296     =+>ucs@iso
 297     =+>ucs@unicode
 298     =+>ucs@jis
 299     =+>ucs@jis/1990
 300     =+>ucs@cns
 301     =+>ucs@ks
 302     =>>ucs@iso
 303     =>>ucs@unicode
 304     =>>ucs@jis
 305     =>>ucs@cns
 306     =>>>ucs@iso
 307     =>>>ucs@unicode
 308     ==ucs@iso
 309     ==ucs@unicode
 310     ;; ==ucs@cns
 311     ==gb2312
 312     ==ks-x1001
 313     ==cns11643-1
 314     ==cns11643-2
 315     ==cns11643-3
 316     ==cns11643-4
 317     ==cns11643-5
 318     ==cns11643-6
 319     ==cns11643-7
 320     ==gt
 321     ==jis-x0208@1990
 322     ;; ==jis-x0208@1983
 323     ==jis-x0208@1978
 324     ==gt-k
 325     =ucs@iso
 326     =ucs@unicode
 327     =ucs@cns
 328     ==big5-cdp
 329     ==cbeta
 330     =>>big5-cdp
 331     =>>gt-k
 332     =+>gt
 333     =+>big5-cdp
 334     =>big5-cdp
 335     =>big5-cdp@iwds-1
 336     =>mj
 337     =>iwds-1
 338     =big5
 339     =big5-eten
 340     =zinbun-oracle
 341     =>zinbun-oracle
 342     =ruimoku-v6
 343     =>>ruimoku-v6
 344     ==ruimoku-v6
 345     =>ruimoku-v6
 346     =jef-china3
 347     =shinjigen
 348     =>cns11643-5
 349     =>cns11643-7
 350     =>gt
 351     =>gt-k
 352     =>daikanwa
 353     =>daikanwa/ho
 354     =>cbeta
 355     =ucs-var-001
 356     =ucs-var-002
 357     =ucs-var-003
 358     =ucs-var-004
 359     =ucs-var-005
 360     =ucs-var-006
 361     =ucs-var-007
 362     =ucs-var-008
 363     =ucs-var-010
 364     =ucs-itaiji-001
 365     =ucs-itaiji-002
 366     =ucs-itaiji-003
 367     =ucs-itaiji-004
 368     =ucs-itaiji-005
 369     =ucs-itaiji-006
 370     =ucs-itaiji-007
 371     =ucs-itaiji-008
 372     =ucs-itaiji-009
 373     =ucs-itaiji-010
 374     =ucs-itaiji-011
 375     =ucs-itaiji-012
 376     =+>ucs-var-003
 377     =>ucs-itaiji-001
 378     =>ucs-itaiji-002
 379     =>ucs-itaiji-003
 380     =>ucs-itaiji-004
 381     =>ucs-itaiji-005
 382     =>ucs-itaiji-006
 383     =>ucs-itaiji-007
 384     =>ucs-itaiji-008
 385     =>ucs-itaiji-009
 386     =>ucs-itaiji-010
 387     =>ucs-itaiji-011
 388     =big5-cdp-var-001
 389     =big5-cdp-var-002
 390     =big5-cdp-var-003
 391     =big5-cdp-var-004
 392     =big5-cdp-var-005
 393     =big5-cdp-var-007
 394     =big5-cdp-var-010
 395     =big5-cdp-itaiji-001
 396     =big5-cdp-itaiji-002
 397     =big5-cdp-itaiji-003
 398     =big5-cdp-itaiji-004
 399     =big5-cdp-itaiji-005
 400     =>big5-cdp-itaiji-001
 401     =>big5-cdp-itaiji-002
 402     =>big5-cdp-itaiji-003
 403     =>big5-cdp-itaiji-004
 404     =>big5-cdp-var-001
 405     =>ucs@iwds-1
 406     =>ucs@cognate
 407     =>ucs@component
 408     =>ucs@iwds-1/normalized
 409     =>ucs-itaiji-001@iwds-1
 410     =>ucs-itaiji-002@iwds-1
 411     =>ucs-itaiji-006@iwds-1
 412     =>big5-cdp@component
 413     =>big5-cdp@cognate
 414     ==ucs@gb
 415     =ucs@gb
 416     ==ucs-var-002
 417     =ucs@JP/hanazono
 418     =koseki
 419     =shuowen-jiguge))
 420
 421
 422 ;;; @ char-db formatters
 423 ;;;
 424
 425 (defun char-db-make-char-spec (char)
 426   (let (ret char-spec)
 427     (cond ((characterp char)
 428            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 429                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 430                   (setq char-spec (list (cons '=ucs ret)))
 431                   (cond ((setq ret (get-char-attribute char 'name))
 432                          (setq char-spec (cons (cons 'name ret) char-spec))
 433                          )
 434                         ((setq ret (get-char-attribute char 'name*))
 435                          (setq char-spec (cons (cons 'name* ret) char-spec))
 436                          ))
 437                   )
 438                  ((encode-char char '=mj 'defined-only)
 439                   (setq char-spec nil)
 440                   (dolist (ccs (charset-list))
 441                     (if (and (or (eq ccs '=mj)
 442                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 443                                  (string-match "=ucs@" (symbol-name ccs))
 444                                  )
 445                              (setq ccs (charset-name ccs))
 446                              (null (assq ccs char-spec))
 447                              (setq ret (encode-char char ccs 'defined-only)))
 448                         (setq char-spec (cons (cons ccs ret) char-spec))))
 449                   )
 450                  ((encode-char char '==mj 'defined-only)
 451                   (setq char-spec nil)
 452                   (dolist (ccs (charset-list))
 453                     (if (and (or (eq ccs '==mj)
 454                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 455                                  (string-match "=ucs@" (symbol-name ccs))
 456                                  )
 457                              (setq ccs (charset-name ccs))
 458                              (null (assq ccs char-spec))
 459                              (setq ret (encode-char char ccs 'defined-only)))
 460                         (setq char-spec (cons (cons ccs ret) char-spec))))
 461                   )
 462                  ((encode-char char '=adobe-japan1 'defined-only)
 463                   (setq char-spec nil)
 464                   (dolist (ccs (charset-list))
 465                     (if (and (or (memq ccs
 466                                        '(=adobe-japan1-0
 467                                          =adobe-japan1-1
 468                                          =adobe-japan1-2
 469                                          =adobe-japan1-3
 470                                          =adobe-japan1-4
 471                                          =adobe-japan1-5
 472                                          =adobe-japan1-6
 473                                          ))
 474                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 475                                  (string-match "=ucs@" (symbol-name ccs))
 476                                  )
 477                              (setq ccs (charset-name ccs))
 478                              (null (assq ccs char-spec))
 479                              (setq ret (encode-char char ccs 'defined-only)))
 480                         (setq char-spec (cons (cons ccs ret) char-spec))))
 481                   )
 482                  ((encode-char char '==adobe-japan1 'defined-only)
 483                   (setq char-spec nil)
 484                   (dolist (ccs (charset-list))
 485                     (if (and (or (memq ccs
 486                                        '(==adobe-japan1-0
 487                                          ==adobe-japan1-1
 488                                          ==adobe-japan1-2
 489                                          ==adobe-japan1-3
 490                                          ==adobe-japan1-4
 491                                          ==adobe-japan1-5
 492                                          ==adobe-japan1-6
 493                                          ))
 494                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 495                                  (string-match "=ucs@" (symbol-name ccs))
 496                                  )
 497                              (setq ccs (charset-name ccs))
 498                              (null (assq ccs char-spec))
 499                              (setq ret (encode-char char ccs 'defined-only)))
 500                         (setq char-spec (cons (cons ccs ret) char-spec))))
 501                   )
 502                  ((setq ret
 503                         (catch 'tag
 504                           (let ((rest char-db-coded-charset-priority-list)
 505                                 ccs)
 506                             (while rest
 507                               (setq ccs (charset-name
 508                                          (find-charset (car rest))))
 509                               (if (setq ret
 510                                         (encode-char char ccs
 511                                                      'defined-only))
 512                                   (throw 'tag (cons ccs ret)))
 513                               (setq rest (cdr rest))))))
 514                   (setq char-spec (list ret))
 515                   (dolist (ccs (delq (car ret) (charset-list)))
 516                     (if (and (or (charset-iso-final-char ccs)
 517                                  (memq ccs
 518                                        '(=daikanwa
 519                                          =daikanwa@rev2
 520                                          ;; =gt-k
 521                                          =jis-x0208@1997
 522                                          ))
 523                                  (and (eq (charset-property ccs 'iso-ir) 177)
 524                                       (not (eq (charset-name ccs) '=ucs@big5)))
 525                                  ;; (string-match "=ucs@" (symbol-name ccs))
 526                                  )
 527                              (setq ccs (charset-name ccs))
 528                              (null (assq ccs char-spec))
 529                              (setq ret (encode-char char ccs 'defined-only)))
 530                         (setq char-spec (cons (cons ccs ret) char-spec))))
 531                   (if (null char-spec)
 532                       (setq char-spec (split-char char)))
 533                   (cond ((setq ret (get-char-attribute char 'name))
 534                          (setq char-spec (cons (cons 'name ret) char-spec))
 535                          )
 536                         ((setq ret (get-char-attribute char 'name*))
 537                          (setq char-spec (cons (cons 'name* ret) char-spec))
 538                          ))
 539                   )
 540                  ((setq ret (get-char-attribute
 541                              char 'ideographic-combination))
 542                   (setq char-spec
 543                         (cons (cons 'ideographic-combination ret)
 544                               char-spec))
 545                   ))
 546            char-spec)
 547           ((consp char)
 548            char))))
 549
 550 (defun char-db-insert-char-spec (char &optional readable column
 551                                       required-features)
 552   (unless column
 553     (setq column (current-column)))
 554   (let (char-spec temp-char)
 555     (setq char-spec (char-db-make-char-spec char))
 556     (unless (or (characterp char) ; char
 557                 (condition-case nil
 558                     (setq char (find-char char-spec))
 559                   (error nil)))
 560       ;; define temporary character
 561       ;;   Current implementation is dirty.
 562       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 563                                          char-spec)))
 564       (remove-char-attribute temp-char 'ideograph-daikanwa)
 565       (setq char temp-char))
 566     (insert-char-attributes char
 567                             readable
 568                             (union (mapcar #'car char-spec)
 569                                    required-features)
 570                             nil 'for-sub-node)
 571     (when temp-char
 572       ;; undefine temporary character
 573       ;;   Current implementation is dirty.
 574       (setq char-spec (char-attribute-alist temp-char))
 575       (while char-spec
 576         (remove-char-attribute temp-char (car (car char-spec)))
 577         (setq char-spec (cdr char-spec))))))
 578
 579 (defun char-db-insert-alist (alist &optional readable column)
 580   (unless column
 581     (setq column (current-column)))
 582   (let ((line-breaking
 583          (concat "\n" (make-string (1+ column) ?\ )))
 584         name value
 585         ret al ; cal
 586         key
 587         lbs cell rest separator)
 588     (insert "(")
 589     (while alist
 590       (setq name (car (car alist))
 591             value (cdr (car alist)))
 592       (cond ((eq name 'char)
 593              (insert "(char . ")
 594              (if (setq ret (condition-case nil
 595                                (find-char value)
 596                              (error nil)))
 597                  (progn
 598                    (setq al nil
 599                          ;; cal nil
 600                          )
 601                    (while value
 602                      (setq key (car (car value)))
 603                      ;; (if (find-charset key)
 604                      ;;     (setq cal (cons key cal))
 605                      (setq al (cons key al))
 606                      ;; )
 607                      (setq value (cdr value)))
 608                    (insert-char-attributes ret
 609                                            readable
 610                                            (or al 'none) ; cal
 611                                            nil 'for-sub-node))
 612                (insert (prin1-to-string value)))
 613              (insert ")")
 614              (insert line-breaking))
 615             ((consp value)
 616              (insert (format "(%-18s " name))
 617              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 618              (while (consp value)
 619                (setq cell (car value))
 620                (if (and (consp cell)
 621                         (consp (car cell))
 622                         (setq ret (condition-case nil
 623                                       (find-char cell)
 624                                     (error nil)))
 625                         )
 626                    (progn
 627                      (setq rest cell
 628                            al nil
 629                            ;; cal nil
 630                            )
 631                      (while rest
 632                        (setq key (car (car rest)))
 633                        ;; (if (find-charset key)
 634                        ;;     (setq cal (cons key cal))
 635                        (setq al (cons key al))
 636                        ;; )
 637                        (setq rest (cdr rest)))
 638                      (if separator
 639                          (insert lbs))
 640                      (insert-char-attributes ret
 641                                              readable
 642                                              al ; cal
 643                                              nil 'for-sub-node)
 644                      (setq separator lbs))
 645                  (if separator
 646                      (insert separator))
 647                  (insert (prin1-to-string cell))
 648                  (setq separator " "))
 649                (setq value (cdr value)))
 650              (insert ")")
 651              (insert line-breaking))
 652             (t
 653              (insert (format "(%-18s . %S)%s"
 654                              name value
 655                              line-breaking))))
 656       (setq alist (cdr alist))))
 657   (insert ")"))
 658
 659 (defun char-db-insert-char-reference (plist &optional readable column)
 660   (unless column
 661     (setq column (current-column)))
 662   (let ((line-breaking
 663          (concat "\n" (make-string (1+ column) ?\ )))
 664         (separator "")
 665         name value)
 666     (insert "(")
 667     (while plist
 668       (setq name (pop plist))
 669       (setq value (pop plist))
 670       (cond ((eq name :char)
 671              (insert separator)
 672              (insert ":char\t")
 673              (cond ((numberp value)
 674                     (setq value (decode-char '=ucs value)))
 675                    ;; ((consp value)
 676                    ;;  (setq value (or (find-char value)
 677                    ;;                  value)))
 678                    )
 679              (char-db-insert-char-spec value readable)
 680              (insert line-breaking)
 681              (setq separator ""))
 682             ((eq name :radical)
 683              (insert (format "%s%s\t%d ; %c%s"
 684                              separator
 685                              name value
 686                              (ideographic-radical value)
 687                              line-breaking))
 688              (setq separator ""))
 689             (t
 690              (insert (format "%s%s\t%S" separator name value))
 691              (setq separator line-breaking)))
 692       ))
 693   (insert ")"))
 694
 695 (defun char-db-decode-isolated-char (ccs code-point)
 696   (let (ret)
 697     (setq ret
 698           (cond ((eq ccs 'arabic-iso8859-6)
 699                  (decode-char ccs code-point))
 700                 ;; ((eq ccs '=gt)
 701                 ;;  (decode-builtin-char '==gt code-point))
 702                 ((and (memq ccs '(=gt-pj-1
 703                                   =gt-pj-2
 704                                   =gt-pj-3
 705                                   =gt-pj-4
 706                                   =gt-pj-5
 707                                   =gt-pj-6
 708                                   =gt-pj-7
 709                                   =gt-pj-8
 710                                   =gt-pj-9
 711                                   =gt-pj-10
 712                                   =gt-pj-11))
 713                       (setq ret (decode-char ccs code-point))
 714                       (setq ret (encode-char ret '=gt 'defined-only)))
 715                  (decode-builtin-char '=gt ret))
 716                 (t
 717                  (decode-builtin-char ccs code-point))))
 718     (cond ((null ret)
 719            (or (decode-char ccs code-point)
 720                (define-char (list (cons ccs code-point)))))
 721           ((and (<= 0 (char-int ret))
 722                 (<= (char-int ret) #x1F))
 723            (decode-char '=ucs (+ #x2400 (char-int ret))))
 724           ((= (char-int ret) #x7F)
 725            ?\u2421)
 726           (t ret))))
 727
 728 (defvar char-db-convert-obsolete-format t)
 729
 730 (defun char-db-insert-ccs-feature (name value line-breaking)
 731   (cond
 732    ((integerp value)
 733     (insert
 734      (format
 735       (cond
 736        ((memq name '(=>iwds-1
 737                      ==shinjigen
 738                      =shinjigen
 739                      =shinjigen@1ed ==shinjigen@1ed
 740                      =shinjigen@rev ==shinjigen@rev
 741                      =shinjigen/+p@rev ==shinjigen/+p@rev
 742                      ===daikanwa/ho ==daikanwa/ho
 743                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 744         "(%-18s .  %04d)\t; %c")
 745        ((eq name '=shinjigen@1ed/24pr)
 746         "(%-18s . %04d)\t; %c")
 747        ((or
 748          (memq name
 749                '(===daikanwa
 750                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 751                  =daikanwa@rev1 =daikanwa@rev2
 752                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 753                  =>>daikanwa/+p
 754                  =daikanwa/+2p =>>daikanwa/+2p
 755                  =gt ==gt ===gt
 756                  =>>gt =+>gt =>gt
 757                  =gt-k ==gt-k ===gt-k
 758                  =>>gt-k =>gt-k
 759                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 760                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 761                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 762                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 763                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 764                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 765                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 766                  =>>adobe-japan1-0 =+>adobe-japan1-0
 767                  =>>adobe-japan1-1 =+>adobe-japan1-1
 768                  =>>adobe-japan1-2 =+>adobe-japan1-2
 769                  =>>adobe-japan1-3 =+>adobe-japan1-3
 770                  =>>adobe-japan1-4 =+>adobe-japan1-4
 771                  =>>adobe-japan1-5 =+>adobe-japan1-5
 772                  =>>adobe-japan1-6 =+>adobe-japan1-6
 773                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 774                  =zinbun-oracle =>zinbun-oracle
 775                  =daijiten ==daijiten ===daijiten ==>daijiten
 776                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 777                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 778                  ===hng-khi ===hng-khm ===hng-fhs ===hng-hok
 779                  ===hng-kyd ===hng-sok
 780                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 781                  ===hng-sys ===hng-tsu ===hng-tzj
 782                  ===hng-hos ===hng-kkh ===hng-nak ===hng-jhk
 783                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 784                  ===hng-nkm ===hng-k24 ===hng-ini ===hng-nkk
 785                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 786                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 787                  ===hng-okd ===hng-wad ===hng-kmi
 788                  ===hng-zkd ===hng-doh ===hng-jyu ===hng-tzs
 789                  ===hng-sai ===hng-kad ===hng-kss
 790                  ===hng-kyo ===hng-ykk ===hng-saa
 791                  ===hng-sab ===hng-wks ===hng-wke ===hng-smk
 792                  ===hng-sgs ===hng-sts ===hng-kkd
 793                  =shuowen-jiguge ===shuowen-jiguge4 ===shuowen-jiguge5))
 794          ;; (string-match "^=adobe-" (symbol-name name))
 795          )
 796         "(%-18s . %05d)\t; %c")
 797        ((memq name '(=hanyo-denshi/ks
 798                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 799                      =>>hanyo-denshi/ks
 800                      =koseki ==koseki
 801                      =mj ==mj ===mj =>>mj =>mj
 802                      =zihai mojikyo))
 803         "(%-18s . %06d)\t; %c")
 804        ((memq name '(=hanyo-denshi/tk ==hanyo-denshi/tk))
 805         "(%-18s . %08d)\t; %c")
 806        ((>= (charset-dimension name) 2)
 807         "(%-18s . #x%04X)\t; %c")
 808        (t
 809         "(%-18s . #x%02X)\t; %c"))
 810       name
 811       (if (= (charset-iso-graphic-plane name) 1)
 812           (logior value
 813                   (cond ((= (charset-dimension name) 1)
 814                          #x80)
 815                         ((= (charset-dimension name) 2)
 816                          #x8080)
 817                         ((= (charset-dimension name) 3)
 818                          #x808080)
 819                         (t 0)))
 820         value)
 821       (char-db-decode-isolated-char name value)))
 822     (if (and (= (charset-chars name) 94)
 823              (= (charset-dimension name) 2))
 824         (insert (format " [%02d-%02d]"
 825                         (- (lsh value -8) 32)
 826                         (- (logand value 255) 32))))
 827     )
 828    (t
 829     (insert (format "(%-18s . %s)" name value))
 830     ))
 831   (insert line-breaking))
 832
 833 (defun char-db-insert-relation-feature (char name value line-breaking
 834                                              ccss readable)
 835   (insert (format "(%-18s%s " name line-breaking))
 836   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 837         separator cell sources required-features
 838         ret)
 839     (while (consp value)
 840       (setq cell (car value))
 841       (if (integerp cell)
 842           (setq cell (decode-char '=ucs cell)))
 843       (cond
 844        ((eq name '->subsumptive)
 845         (when (or (not (some (lambda (atr)
 846                                (get-char-attribute cell atr))
 847                              char-db-ignored-attributes))
 848                   (some (lambda (ccs)
 849                           (encode-char cell ccs 'defined-only))
 850                         ccss))
 851           (if separator
 852               (insert lbs))
 853           (let ((char-db-ignored-attributes
 854                  (cons '<-subsumptive
 855                        char-db-ignored-attributes)))
 856             (insert-char-attributes cell readable nil nil 'for-sub-node))
 857           (setq separator lbs))
 858         )
 859        ((characterp cell)
 860         (setq sources
 861               (get-char-attribute
 862                char (intern (format "%s*sources" name))))
 863         (setq required-features nil)
 864         (dolist (source sources)
 865           (cond
 866            ((memq source '(JP
 867                            JP/Jouyou
 868                            shinjigen shinjigen@1ed shinjigen@rev))
 869             (setq required-features
 870                   (union required-features
 871                          '(=jis-x0208
 872                            =jis-x0208@1990
 873                            =jis-x0213-1@2000
 874                            =jis-x0213-1@2004
 875                            =jis-x0213-2
 876                            =jis-x0212
 877                            =jis-x0208@1983
 878                            =jis-x0208@1978
 879                            =shinjigen
 880                            =shinjigen@1ed
 881                            =shinjigen@rev
 882                            =shinjigen/+p@rev))))
 883            ((eq source 'CN)
 884             (setq required-features
 885                   (union required-features
 886                          '(=gb2312
 887                            =gb12345
 888                            =iso-ir165)))))
 889           (cond
 890            ((find-charset (setq ret (intern (format "=%s" source))))
 891             (setq required-features
 892                   (cons ret required-features)))
 893            (t (setq required-features
 894                     (cons source required-features)))))
 895         (cond ((string-match "@JP" (symbol-name name))
 896                (setq required-features
 897                      (union required-features
 898                             '(=jis-x0208
 899                               =jis-x0208@1990
 900                               =jis-x0213-1-2000
 901                               =jis-x0213-2-2000
 902                               =jis-x0212
 903                               =jis-x0208@1983
 904                               =jis-x0208@1978))))
 905               ((string-match "@CN" (symbol-name name))
 906                (setq required-features
 907                      (union required-features
 908                             '(=gb2312
 909                               =gb12345
 910                               =iso-ir165)))))
 911         (if separator
 912             (insert lbs))
 913         (if readable
 914             (insert (format "%S" cell))
 915           (char-db-insert-char-spec cell readable
 916                                     nil
 917                                     required-features))
 918         (setq separator lbs))
 919        ((consp cell)
 920         (if separator
 921             (insert lbs))
 922         (if (consp (car cell))
 923             (char-db-insert-char-spec cell readable)
 924           (char-db-insert-char-reference cell readable))
 925         (setq separator lbs))
 926        (t
 927         (if separator
 928             (insert separator))
 929         (insert (prin1-to-string cell))
 930         (setq separator " ")))
 931       (setq value (cdr value)))
 932     (insert ")")
 933     (insert line-breaking)))
 934
 935 (defun insert-char-attributes (char &optional readable attributes column
 936                                     for-sub-node)
 937   (unless column
 938     (setq column (current-column)))
 939   (let (name value ; has-long-ccs-name
 940         rest
 941         radical strokes
 942         (line-breaking
 943          (concat "\n" (make-string (1+ column) ?\ )))
 944         lbs cell separator ret
 945         key al cal
 946         dest-ccss ; sources required-features
 947         ccss)
 948     (let (atr-d)
 949       (setq attributes
 950             (sort (if attributes
 951                       (if (consp attributes)
 952                           (progn
 953                             (dolist (name attributes)
 954                               (unless (memq name char-db-ignored-attributes)
 955                                 (if (find-charset name)
 956                                     (push name ccss))
 957                                 (push name atr-d)))
 958                             atr-d))
 959                     (dolist (name (char-attribute-list))
 960                       (unless (memq name char-db-ignored-attributes)
 961                         (if (find-charset name)
 962                             (push name ccss))
 963                         (push name atr-d)))
 964                     atr-d)
 965                   #'char-attribute-name<)))
 966     (insert "(")
 967     (when (memq '<-subsumptive attributes)
 968       (when (or readable (not for-sub-node))
 969         (when (setq value (get-char-attribute char '<-subsumptive))
 970           (char-db-insert-relation-feature char '<-subsumptive value
 971                                            line-breaking
 972                                            ccss readable)))
 973       (setq attributes (delq '<-subsumptive attributes)))
 974     (when (and (memq '<-denotational attributes)
 975                (setq value (get-char-attribute char '<-denotational)))
 976       (char-db-insert-relation-feature char '<-denotational value
 977                                        line-breaking
 978                                        ccss readable)
 979       (setq attributes (delq '<-denotational attributes)))
 980     (when (and (memq '<-denotational@component attributes)
 981                (setq value (get-char-attribute char '<-denotational@component)))
 982       (char-db-insert-relation-feature char '<-denotational@component value
 983                                        line-breaking
 984                                        ccss readable)
 985       (setq attributes (delq '<-denotational@component attributes)))
 986     (when (and (memq '<-denotational@usage attributes)
 987                (setq value (get-char-attribute char '<-denotational@usage)))
 988       (char-db-insert-relation-feature char '<-denotational@usage value
 989                                        line-breaking
 990                                        ccss readable)
 991       (setq attributes (delq '<-denotational@usage attributes)))
 992     (when (and (memq 'name attributes)
 993                (setq value (get-char-attribute char 'name)))
 994       (insert (format
 995                (if (> (+ (current-column) (length value)) 48)
 996                    "(name . %S)%s"
 997                  "(name               . %S)%s")
 998                value line-breaking))
 999       (setq attributes (delq 'name attributes))
1000       )
1001     (when (and (memq 'name* attributes)
1002                (setq value (get-char-attribute char 'name*)))
1003       (insert (format
1004                (if (> (+ (current-column) (length value)) 48)
1005                    "(name* . %S)%s"
1006                  "(name*              . %S)%s")
1007                value line-breaking))
1008       (setq attributes (delq 'name* attributes))
1009       )
1010     (when (and (memq 'script attributes)
1011                (setq value (get-char-attribute char 'script)))
1012       (insert (format "(script\t\t%s)%s"
1013                       (mapconcat (function prin1-to-string)
1014                                  value " ")
1015                       line-breaking))
1016       (setq attributes (delq 'script attributes))
1017       )
1018     (dolist (name '(=>ucs =>ucs*))
1019       (when (and (memq name attributes)
1020                  (setq value (get-char-attribute char name)))
1021         (insert (format "(%-18s . #x%04X)\t; %c%s"
1022                         name value (decode-char '=ucs value)
1023                         line-breaking))
1024         (setq attributes (delq name attributes))))
1025     (dolist (name '(=>ucs@gb =>ucs@big5))
1026       (when (and (memq name attributes)
1027                  (setq value (get-char-attribute char name)))
1028         (insert (format "(%-18s . #x%04X)\t; %c%s"
1029                         name value
1030                         (decode-char (intern
1031                                       (concat "="
1032                                               (substring
1033                                                (symbol-name name) 2)))
1034                                      value)
1035                         line-breaking))
1036         (setq attributes (delq name attributes))
1037         ))
1038     ;; (dolist (name '(=>daikanwa))
1039     ;;   (when (and (memq name attributes)
1040     ;;              (setq value (get-char-attribute char name)))
1041     ;;     (insert
1042     ;;      (if (integerp value)
1043     ;;          (format "(%-18s . %05d)\t; %c%s"
1044     ;;                  name value (decode-char '=daikanwa value)
1045     ;;                  line-breaking)
1046     ;;        (format "(%-18s %s)\t; %c%s"
1047     ;;                name
1048     ;;                (mapconcat (function prin1-to-string)
1049     ;;                           value " ")
1050     ;;                (char-representative-of-daikanwa char)
1051     ;;                line-breaking)))
1052     ;;     (setq attributes (delq name attributes))))
1053     (when (and (memq 'general-category attributes)
1054                (setq value (get-char-attribute char 'general-category)))
1055       (insert (format
1056                "(general-category\t%s) ; %s%s"
1057                (mapconcat (lambda (cell)
1058                             (format "%S" cell))
1059                           value " ")
1060                (cond ((rassoc value unidata-normative-category-alist)
1061                       "Normative Category")
1062                      ((rassoc value unidata-informative-category-alist)
1063                       "Informative Category")
1064                      (t
1065                       "Unknown Category"))
1066                line-breaking))
1067       (setq attributes (delq 'general-category attributes))
1068       )
1069     (when (and (memq 'bidi-category attributes)
1070                (setq value (get-char-attribute char 'bidi-category)))
1071       (insert (format "(bidi-category\t. %S)%s"
1072                       value
1073                       line-breaking))
1074       (setq attributes (delq 'bidi-category attributes))
1075       )
1076     (unless (or (not (memq 'mirrored attributes))
1077                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
1078                     'empty))
1079       (insert (format "(mirrored\t\t. %S)%s"
1080                       value
1081                       line-breaking))
1082       (setq attributes (delq 'mirrored attributes))
1083       )
1084     (cond
1085      ((and (memq 'decimal-digit-value attributes)
1086            (setq value (get-char-attribute char 'decimal-digit-value)))
1087       (insert (format "(decimal-digit-value . %S)%s"
1088                       value
1089                       line-breaking))
1090       (setq attributes (delq 'decimal-digit-value attributes))
1091       (when (and (memq 'digit-value attributes)
1092                  (setq value (get-char-attribute char 'digit-value)))
1093         (insert (format "(digit-value\t . %S)%s"
1094                         value
1095                         line-breaking))
1096         (setq attributes (delq 'digit-value attributes))
1097         )
1098       (when (and (memq 'numeric-value attributes)
1099                  (setq value (get-char-attribute char 'numeric-value)))
1100         (insert (format "(numeric-value\t . %S)%s"
1101                         value
1102                         line-breaking))
1103         (setq attributes (delq 'numeric-value attributes))
1104         )
1105       )
1106      (t
1107       (when (and (memq 'digit-value attributes)
1108                  (setq value (get-char-attribute char 'digit-value)))
1109         (insert (format "(digit-value\t. %S)%s"
1110                         value
1111                         line-breaking))
1112         (setq attributes (delq 'digit-value attributes))
1113         )
1114       (when (and (memq 'numeric-value attributes)
1115                  (setq value (get-char-attribute char 'numeric-value)))
1116         (insert (format "(numeric-value\t. %S)%s"
1117                         value
1118                         line-breaking))
1119         (setq attributes (delq 'numeric-value attributes))
1120         )))
1121     (when (and (memq 'iso-10646-comment attributes)
1122                (setq value (get-char-attribute char 'iso-10646-comment)))
1123       (insert (format "(iso-10646-comment\t. %S)%s"
1124                       value
1125                       line-breaking))
1126       (setq attributes (delq 'iso-10646-comment attributes))
1127       )
1128     (when (and (memq 'morohashi-daikanwa attributes)
1129                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1130       (insert (format "(morohashi-daikanwa\t%s)%s"
1131                       (mapconcat (function prin1-to-string) value " ")
1132                       line-breaking))
1133       (setq attributes (delq 'morohashi-daikanwa attributes))
1134       )
1135     (setq radical nil
1136           strokes nil)
1137     (when (and (memq 'ideographic-radical attributes)
1138                (setq value (get-char-attribute char 'ideographic-radical)))
1139       (setq radical value)
1140       (insert (format "(ideographic-radical . %S)\t; %c%s"
1141                       radical
1142                       (ideographic-radical radical)
1143                       line-breaking))
1144       (setq attributes (delq 'ideographic-radical attributes))
1145       )
1146     (when (and (memq 'shuowen-radical attributes)
1147                (setq value (get-char-attribute char 'shuowen-radical)))
1148       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1149                       value
1150                       (shuowen-radical value)
1151                       line-breaking))
1152       (setq attributes (delq 'shuowen-radical attributes))
1153       )
1154     (let (key)
1155       (dolist (domain
1156                (append
1157                 char-db-feature-domains
1158                 (let (dest domain)
1159                   (dolist (feature (char-attribute-list))
1160                     (setq feature (symbol-name feature))
1161                     (when (string-match
1162                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1163                            feature)
1164                       (setq domain (intern (match-string 2 feature)))
1165                      (unless (memq domain dest)
1166                        (setq dest (cons domain dest)))))
1167                   (sort dest #'string<))))
1168         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1169         (when (and (memq key attributes)
1170                    (setq value (get-char-attribute char key)))
1171           (setq radical value)
1172           (insert (format "(%s . %S)\t; %c%s"
1173                           key
1174                           radical
1175                           (ideographic-radical radical)
1176                           line-breaking))
1177           (setq attributes (delq key attributes))
1178           )
1179         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1180         (when (and (memq key attributes)
1181                    (setq value (get-char-attribute char key)))
1182           (setq strokes value)
1183           (insert (format "(%s . %S)%s"
1184                           key
1185                           strokes
1186                           line-breaking))
1187           (setq attributes (delq key attributes))
1188           )
1189         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1190         (when (and (memq key attributes)
1191                    (setq value (get-char-attribute char key)))
1192           (insert (format "(%s       . %S)%s"
1193                           key
1194                           value
1195                           line-breaking))
1196           (setq attributes (delq key attributes))
1197           )
1198         (dolist (feature '(ideographic-radical
1199                            ideographic-strokes
1200                            total-strokes))
1201           (setq key (intern (format "%s@%s*sources" feature domain)))
1202           (when (and (memq key attributes)
1203                      (setq value (get-char-attribute char key)))
1204             (insert (format "(%s%s" key line-breaking))
1205             (dolist (cell value)
1206               (insert (format " %s" cell)))
1207             (insert ")")
1208             (insert line-breaking)
1209             (setq attributes (delq key attributes))
1210             ))
1211         ))
1212     (when (and (memq 'ideographic-strokes attributes)
1213                (setq value (get-char-attribute char 'ideographic-strokes)))
1214       (setq strokes value)
1215       (insert (format "(ideographic-strokes . %S)%s"
1216                       strokes
1217                       line-breaking))
1218       (setq attributes (delq 'ideographic-strokes attributes))
1219       )
1220     (when (and (memq 'kangxi-radical attributes)
1221                (setq value (get-char-attribute char 'kangxi-radical)))
1222       (unless (eq value radical)
1223         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1224                         value
1225                         (ideographic-radical value)
1226                         line-breaking))
1227         (or radical
1228             (setq radical value)))
1229       (setq attributes (delq 'kangxi-radical attributes))
1230       )
1231     (when (and (memq 'kangxi-strokes attributes)
1232                (setq value (get-char-attribute char 'kangxi-strokes)))
1233       (unless (eq value strokes)
1234         (insert (format "(kangxi-strokes\t . %S)%s"
1235                         value
1236                         line-breaking))
1237         (or strokes
1238             (setq strokes value)))
1239       (setq attributes (delq 'kangxi-strokes attributes))
1240       )
1241     ;; (when (and (memq 'japanese-radical attributes)
1242     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1243     ;;   (unless (eq value radical)
1244     ;;     (insert (format "(japanese-radical\t . %S)\t; %c%s"
1245     ;;                     value
1246     ;;                     (ideographic-radical value)
1247     ;;                     line-breaking))
1248     ;;     (or radical
1249     ;;         (setq radical value)))
1250     ;;   (setq attributes (delq 'japanese-radical attributes))
1251     ;;   )
1252     (when (and (memq 'japanese-strokes attributes)
1253                (setq value (get-char-attribute char 'japanese-strokes)))
1254       (unless (eq value strokes)
1255         (insert (format "(japanese-strokes\t . %S)%s"
1256                         value
1257                         line-breaking))
1258         (or strokes
1259             (setq strokes value)))
1260       (setq attributes (delq 'japanese-strokes attributes))
1261       )
1262     (when (and (memq 'cns-radical attributes)
1263                (setq value (get-char-attribute char 'cns-radical)))
1264       (insert (format "(cns-radical\t . %S)\t; %c%s"
1265                       value
1266                       (ideographic-radical value)
1267                       line-breaking))
1268       (setq attributes (delq 'cns-radical attributes))
1269       )
1270     (when (and (memq 'cns-strokes attributes)
1271                (setq value (get-char-attribute char 'cns-strokes)))
1272       (unless (eq value strokes)
1273         (insert (format "(cns-strokes\t . %S)%s"
1274                         value
1275                         line-breaking))
1276         (or strokes
1277             (setq strokes value)))
1278       (setq attributes (delq 'cns-strokes attributes))
1279       )
1280     ;; (when (and (memq 'ideographic- attributes)
1281     ;;            (setq value (get-char-attribute char 'ideographic-)))
1282     ;;   (insert "(ideographic-       ")
1283     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1284     ;;         separator nil)
1285     ;;   (while (consp value)
1286     ;;     (setq cell (car value))
1287     ;;     (if (integerp cell)
1288     ;;         (setq cell (decode-char '=ucs cell)))
1289     ;;     (cond ((characterp cell)
1290     ;;            (if separator
1291     ;;                (insert lbs))
1292     ;;            (if readable
1293     ;;                (insert (format "%S" cell))
1294     ;;              (char-db-insert-char-spec cell readable))
1295     ;;            (setq separator lbs))
1296     ;;           ((consp cell)
1297     ;;            (if separator
1298     ;;                (insert lbs))
1299     ;;            (if (consp (car cell))
1300     ;;                (char-db-insert-char-spec cell readable)
1301     ;;              (char-db-insert-char-reference cell readable))
1302     ;;            (setq separator lbs))
1303     ;;           (t
1304     ;;            (if separator
1305     ;;                (insert separator))
1306     ;;            (insert (prin1-to-string cell))
1307     ;;            (setq separator " ")))
1308     ;;     (setq value (cdr value)))
1309     ;;   (insert ")")
1310     ;;   (insert line-breaking)
1311     ;;   (setq attributes (delq 'ideographic- attributes)))
1312     (when (and (memq 'total-strokes attributes)
1313                (setq value (get-char-attribute char 'total-strokes)))
1314       (insert (format "(total-strokes       . %S)%s"
1315                       value
1316                       line-breaking))
1317       (setq attributes (delq 'total-strokes attributes))
1318       )
1319     ;; (when (and (memq '->ideograph attributes)
1320     ;;            (setq value (get-char-attribute char '->ideograph)))
1321     ;;   (insert (format "(->ideograph\t%s)%s"
1322     ;;                   (mapconcat (lambda (code)
1323     ;;                                (cond ((symbolp code)
1324     ;;                                       (symbol-name code))
1325     ;;                                      ((integerp code)
1326     ;;                                       (format "#x%04X" code))
1327     ;;                                      (t
1328     ;;                                       (format "%s %S"
1329     ;;                                               line-breaking code))))
1330     ;;                              value " ")
1331     ;;                   line-breaking))
1332     ;;   (setq attributes (delq '->ideograph attributes))
1333     ;;   )
1334     ;; (when (and (memq '->decomposition attributes)
1335     ;;            (setq value (get-char-attribute char '->decomposition)))
1336     ;;   (insert (format "(->decomposition\t%s)%s"
1337     ;;                   (mapconcat (lambda (code)
1338     ;;                                (cond ((symbolp code)
1339     ;;                                       (symbol-name code))
1340     ;;                                      ((characterp code)
1341     ;;                                       (if readable
1342     ;;                                           (format "%S" code)
1343     ;;                                         (format "#x%04X"
1344     ;;                                                 (char-int code))
1345     ;;                                         ))
1346     ;;                                      ((integerp code)
1347     ;;                                       (format "#x%04X" code))
1348     ;;                                      (t
1349     ;;                                       (format "%s%S" line-breaking code))))
1350     ;;                              value " ")
1351     ;;                   line-breaking))
1352     ;;   (setq attributes (delq '->decomposition attributes))
1353     ;;   )
1354     (if (equal (get-char-attribute char '->titlecase)
1355                (get-char-attribute char '->uppercase))
1356         (setq attributes (delq '->titlecase attributes)))
1357     (when (and (memq '->mojikyo attributes)
1358                (setq value (get-char-attribute char '->mojikyo)))
1359       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1360                       value (decode-char 'mojikyo value)
1361                       line-breaking))
1362       (setq attributes (delq '->mojikyo attributes))
1363       )
1364     (unless readable
1365       (dolist (ignored '(composition
1366                          ->denotational <-subsumptive ->ucs-unified
1367                          ;; ->ideographic-component-forms
1368                          ))
1369         (setq attributes (delq ignored attributes))))
1370     (while attributes
1371       (setq name (car attributes))
1372       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1373                   'value-is-empty)
1374         (cond ((setq ret (find-charset name))
1375                (setq name (charset-name ret))
1376                (when (not (memq name dest-ccss))
1377                  (setq dest-ccss (cons name dest-ccss))
1378                  (char-db-insert-ccs-feature name value line-breaking))
1379                )
1380               ((string-match "^=>ucs@" (symbol-name name))
1381                (insert (format "(%-18s . #x%04X)\t; %c%s"
1382                                name value (decode-char '=ucs value)
1383                                line-breaking))
1384                )
1385               ((eq name 'jisx0208-1978/4X)
1386                (insert (format "(%-18s . #x%04X)%s"
1387                                name value
1388                                line-breaking))
1389                )
1390               ((and
1391                 (not readable)
1392                 (not (eq name '->subsumptive))
1393                 (not (eq name '->uppercase))
1394                 (not (eq name '->lowercase))
1395                 (not (eq name '->titlecase))
1396                 (not (eq name '->canonical))
1397                 (not (eq name '->Bopomofo))
1398                 (not (eq name '->mistakable))
1399                 (not (eq name '->ideographic-variants))
1400                 (null (get-char-attribute
1401                        char (intern (format "%s*sources" name))))
1402                 (not (string-match "\\*sources$" (symbol-name name)))
1403                 (null (get-char-attribute
1404                        char (intern (format "%s*note" name))))
1405                 (not (string-match "\\*note$" (symbol-name name)))
1406                 (or (eq name '<-identical)
1407                     (eq name '<-uppercase)
1408                     (eq name '<-lowercase)
1409                     (eq name '<-titlecase)
1410                     (eq name '<-canonical)
1411                     (eq name '<-ideographic-variants)
1412                     ;; (eq name '<-synonyms)
1413                     (string-match "^<-synonyms" (symbol-name name))
1414                     (eq name '<-mistakable)
1415                     (when (string-match "^->" (symbol-name name))
1416                       (cond
1417                        ((string-match "^->fullwidth" (symbol-name name))
1418                         (not (and (consp value)
1419                                   (characterp (car value))
1420                                   (encode-char
1421                                    (car value) '=ucs 'defined-only)))
1422                         )
1423                        (t)))
1424                     ))
1425                )
1426               ((or (eq name 'ideographic-structure)
1427                    (eq name 'ideographic-structure@apparent)
1428                    (eq name 'ideographic-structure@apparent/leftmost)
1429                    (eq name 'ideographic-combination)
1430                    ;; (eq name 'ideographic-)
1431                    (eq name '=decomposition)
1432                    (char-feature-base-name= '=decomposition name)
1433                    (char-feature-base-name= '=>decomposition name)
1434                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1435                    ;;               (symbol-name name))
1436                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1437                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1438                                  (symbol-name name))
1439                    )
1440                (char-db-insert-relation-feature char name value
1441                                                 line-breaking
1442                                                 ccss readable))
1443               ;; ((memq name '(ideograph=
1444               ;;               original-ideograph-of
1445               ;;               ancient-ideograph-of
1446               ;;               vulgar-ideograph-of
1447               ;;               wrong-ideograph-of
1448               ;;               ;; simplified-ideograph-of
1449               ;;               ideographic-variants
1450               ;;               ;; ideographic-different-form-of
1451               ;;               ))
1452               ;;  (insert (format "(%-18s%s " name line-breaking))
1453               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1454               ;;        separator nil)
1455               ;;  (while (consp value)
1456               ;;    (setq cell (car value))
1457               ;;    (if (and (consp cell)
1458               ;;             (consp (car cell)))
1459               ;;        (progn
1460               ;;          (if separator
1461               ;;              (insert lbs))
1462               ;;          (char-db-insert-alist cell readable)
1463               ;;          (setq separator lbs))
1464               ;;      (if separator
1465               ;;          (insert separator))
1466               ;;      (insert (prin1-to-string cell))
1467               ;;      (setq separator " "))
1468               ;;    (setq value (cdr value)))
1469               ;;  (insert ")")
1470               ;;  (insert line-breaking))
1471               ((consp value)
1472                (insert (format "(%-18s " name))
1473                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1474                      separator nil)
1475                (while (consp value)
1476                  (setq cell (car value))
1477                  (if (and (consp cell)
1478                           (consp (car cell))
1479                           (setq ret (condition-case nil
1480                                         (find-char cell)
1481                                       (error nil))))
1482                      (progn
1483                        (setq rest cell
1484                              al nil
1485                              cal nil)
1486                        (while rest
1487                          (setq key (car (car rest)))
1488                          (if (find-charset key)
1489                              (setq cal (cons key cal))
1490                            (setq al (cons key al)))
1491                          (setq rest (cdr rest)))
1492                        (if separator
1493                            (insert lbs))
1494                        (insert-char-attributes ret
1495                                                readable
1496                                                al ; cal
1497                                                nil 'for-sub-node)
1498                        (setq separator lbs))
1499                    (setq ret (prin1-to-string cell))
1500                    (if separator
1501                        (if (< (+ (current-column)
1502                                  (length ret)
1503                                  (length separator))
1504                               76)
1505                            (insert separator)
1506                          (insert lbs)))
1507                    (insert ret)
1508                    (setq separator " "))
1509                  (setq value (cdr value)))
1510                (insert ")")
1511                (insert line-breaking))
1512               (t
1513                (insert (format "(%-18s" name))
1514                (setq ret (prin1-to-string value))
1515                (unless (< (+ (current-column)
1516                              (length ret)
1517                              3)
1518                           76)
1519                  (insert line-breaking))
1520                (insert " . " ret ")" line-breaking)
1521                ;; (insert (format "(%-18s . %S)%s"
1522                ;;                 name value
1523                ;;                 line-breaking))
1524                )
1525               ))
1526       (setq attributes (cdr attributes)))
1527     (insert ")")))
1528
1529 (defun insert-char-data (char &optional readable
1530                               attributes)
1531   (save-restriction
1532     (narrow-to-region (point)(point))
1533     (insert "(define-char
1534   '")
1535     (insert-char-attributes char readable attributes)
1536     (insert ")\n")
1537     (goto-char (point-min))
1538     (while (re-search-forward "[ \t]+$" nil t)
1539       (replace-match ""))
1540     ;; from tabify.
1541     (goto-char (point-min))
1542     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1543       (let ((column (current-column))
1544             (indent-tabs-mode t))
1545         (delete-region (match-beginning 0) (point))
1546         (indent-to column)))
1547     (goto-char (point-max))
1548     ;; (tabify (point-min)(point-max))
1549     ))
1550
1551 (defun insert-char-data-with-variant (char &optional printable
1552                                            no-ucs-unified
1553                                            script excluded-script)
1554   (insert-char-data char printable)
1555   (let ((variants (char-variants char))
1556         rest
1557         variant vs ret)
1558     (setq variants (sort variants #'<))
1559     (setq rest variants)
1560     (setq variants (cons char variants))
1561     (while rest
1562       (setq variant (car rest))
1563       (unless (get-char-attribute variant '<-subsumptive)
1564         (if (and (or (null script)
1565                      (null (setq vs (get-char-attribute variant 'script)))
1566                      (memq script vs))
1567                  (or (null excluded-script)
1568                      (null (setq vs (get-char-attribute variant 'script)))
1569                      (not (memq excluded-script vs))))
1570             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1571               (insert-char-data variant printable)
1572               (if (setq ret (char-variants variant))
1573                   (while ret
1574                     (or (memq (car ret) variants)
1575                         ;; (get-char-attribute (car ret) '<-subsumptive)
1576                         (setq rest (nconc rest (list (car ret)))))
1577                     (setq ret (cdr ret)))))))
1578       (setq rest (cdr rest)))))
1579
1580 (defun insert-char-range-data (min max &optional script excluded-script)
1581   (let ((code min)
1582         char)
1583     (while (<= code max)
1584       (setq char (decode-char '=ucs code))
1585       (if (encode-char char '=ucs 'defined-only)
1586           (insert-char-data-with-variant char nil 'no-ucs-unified
1587                                          script excluded-script))
1588       (setq code (1+ code)))))
1589
1590 (defun write-char-range-data-to-file (min max file
1591                                           &optional script excluded-script)
1592   (let ((coding-system-for-write char-db-file-coding-system))
1593     (with-temp-buffer
1594       (insert (format ";; -*- coding: %s -*-\n"
1595                       char-db-file-coding-system))
1596       (insert-char-range-data min max script excluded-script)
1597       (write-region (point-min)(point-max) file))))
1598
1599 (defvar what-character-original-window-configuration)
1600
1601 ;;;###autoload
1602 (defun what-char-definition (char)
1603   (interactive (list (char-after)))
1604   (let ((buf (get-buffer-create "*Character Description*"))
1605         (the-buf (current-buffer))
1606         (win-conf (current-window-configuration)))
1607     (pop-to-buffer buf)
1608     (make-local-variable 'what-character-original-window-configuration)
1609     (setq what-character-original-window-configuration win-conf)
1610     (setq buffer-read-only nil)
1611     (erase-buffer)
1612     (condition-case err
1613         (progn
1614           (insert-char-data-with-variant char 'printable)
1615           (unless (char-attribute-alist char)
1616             (insert (format ";; = %c\n"
1617                             (let* ((rest (split-char char))
1618                                    (ccs (pop rest))
1619                                    (code (pop rest)))
1620                               (while rest
1621                                 (setq code (logior (lsh code 8)
1622                                                    (pop rest))))
1623                               (decode-char ccs code)))))
1624           ;; (char-db-update-comment)
1625           (set-buffer-modified-p nil)
1626           (view-mode the-buf (lambda (buf)
1627                                (set-window-configuration
1628                                 what-character-original-window-configuration)
1629                                ))
1630           (goto-char (point-min)))
1631       (error (progn
1632                (set-window-configuration
1633                 what-character-original-window-configuration)
1634                (signal (car err) (cdr err)))))))
1635
1636
1637 ;;; @ end
1638 ;;;
1639
1640 (provide 'char-db-util)
1641
1642 ;;; char-db-util.el ends here